Summary of All Sessions |
|---|
Click here for an index of all participants |
| # | Date/Time | Title/Location | Papers |
|---|---|---|---|
| 1 | June 23, 2026 8:15-8:30 | Welcome Remarks - Marcelle Chauvet (University of California-Riverside), Paulo M. M. Rodrigues (Nova School of Business and Economics) Location: Grand Auditorium | 0 |
| 2 | June 23, 2026 8:30-9:45 | JAE Lecture: Orthogonal Moments in Likelihood Models — Stéphane Bonhomme (University of Chicago) — Discussants: Ivan Fernandez-Val (Boston University) and Chris Muris (McMaster University) Location: Grand Auditorium | 0 |
| 3 | June 23, 2026 9:45-10:15 | Coffee break | 0 |
| 4 | June 23, 2026 10:15-12:20 | ASSET RETURN PREDICTABILITY Location: D-113 | 5 |
| 5 | June 23, 2026 10:15-12:00 | CAUSAL INFERENCE METHODS Location: D-106 | 4 |
| 6 | June 23, 2026 10:15-12:00 | CREDIT AND THE MACROECONOMY Location: E002 | 3 |
| 7 | June 23, 2026 10:15-12:00 | FORECASTING WITH MACHINE LEARNING Location: D-105 | 4 |
| 8 | June 23, 2026 10:15-12:20 | LABOR MARKET SORTING Location: B128 | 4 |
| 9 | June 23, 2026 10:15-12:00 | LOCAL PROJECTIONS AND SVARS 1 Location: B008 | 3 |
| 10 | June 23, 2026 10:15-12:20 | MACROECONOMIC DYNAMICS Location: D-114 | 5 |
| 11 | June 23, 2026 10:15-12:00 | NETWORKS Location: D-107 | 4 |
| 12 | June 23, 2026 10:15-12:00 | OIL AND ENERGY SHOCKS Location: D-112 | 4 |
| 13 | June 23, 2026 10:15-12:20 | POLICY INTERVENTIONS Location: B129 | 5 |
| 14 | June 23, 2026 10:15-12:00 | PRODUCTION AND MISALLOCATION Location: D-111 | 4 |
| 15 | June 23, 2026 10:15-12:00 | REGRESSION DISCONTINUITY Location: B009 | 4 |
| 16 | June 23, 2026 10:15-12:20 | RISK SPILLOVERS AND TRANSMISSION Location: D-115 | 5 |
| 17 | June 23, 2026 10:15-12:20 | TIME SERIES ESTIMATION Location: D-110 | 5 |
| 18 | June 23, 2026 12:00-13:30 | Lunch | 0 |
| 19 | June 23, 2026 13:30-15:15 | CLIMATE RISK MODELLING Location: D-113 | 3 |
| 20 | June 23, 2026 13:30-15:15 | CREDIT AND MONETARY SHOCKS Location: D-115 | 4 |
| 21 | June 23, 2026 13:30-15:15 | EVALUATION OF CREDIT POLICIES Location: B129 | 4 |
| 22 | June 23, 2026 13:30-15:15 | FIRM PRICING AND INFLATION Location: D-111 | 4 |
| 23 | June 23, 2026 13:30-15:15 | HIGH-FREQUENCY FINANCE Location: D-105 | 4 |
| 24 | June 23, 2026 13:30-15:15 | IDENTIFICATION IN MICROECONOMETRICS Location: B008 | 3 |
| 25 | June 23, 2026 13:30-15:15 | INTERGENERATIONAL MOBILITY Location: B128 | 4 |
| 26 | June 23, 2026 13:30-15:15 | LONG MEMORY AND PERSISTENCE Location: D-106 | 4 |
| 27 | June 23, 2026 13:30-15:15 | MATCHING AND SELECTION MODELS Location: D-112 | 3 |
| 28 | June 23, 2026 13:30-15:15 | MONETARY POLICY COMMUNICATION Location: D-114 | 4 |
| 29 | June 23, 2026 13:30-15:15 | PANEL DATA METHODS 3 Location: D-110 | 4 |
| 30 | June 23, 2026 13:30-15:15 | STRUCTURAL ESTIMATION Location: D-107 | 4 |
| 31 | June 23, 2026 13:30-15:15 | STRUCTURAL MACROECONOMETRICS Location: B009 | 4 |
| 32 | June 23, 2026 13:30-15:15 | WORK AND TECHNOLOGY 2 Location: E002 | 4 |
| 33 | June 23, 2026 15:15-15:45 | Coffee break | 0 |
| 34 | June 23, 2026 15:45-17:50 | ENERGY MARKETS Location: D-111 | 4 |
| 35 | June 23, 2026 15:45-17:50 | EQUITY MARKETS AND BELIEFS Location: B128 | 5 |
| 36 | June 23, 2026 15:45-17:30 | GLOBAL SHOCKS AND INFLATION Location: D-106 | 4 |
| 37 | June 23, 2026 15:45-17:30 | INFLATION EXPECTATIONS Location: D-110 | 4 |
| 38 | June 23, 2026 15:45-17:30 | LOCAL PROJECTIONS AND SVARS 2 Location: B008 | 3 |
| 39 | June 23, 2026 15:45-17:30 | MACROECONOMIC FLUCTUATIONS Location: D-113 | 4 |
| 40 | June 23, 2026 15:45-17:30 | MICROECONOMETRIC METHODS 1 Location: B009 | 4 |
| 41 | June 23, 2026 15:45-17:30 | MIGRATION AND SOCIETY Location: D-114 | 4 |
| 42 | June 23, 2026 15:45-17:30 | MIXED-FREQUENCY METHODS Location: D-105 | 4 |
| 43 | June 23, 2026 15:45-17:30 | PROGRAM EVALUATION Location: B129 | 4 |
| 44 | June 23, 2026 15:45-17:30 | SKILLS AND THE LABOR MARKET Location: D-115 | 4 |
| 45 | June 23, 2026 15:45-17:30 | SPATIAL PANEL MODELS Location: D-107 | 4 |
| 46 | June 23, 2026 15:45-17:50 | TAIL RISK IN FINANCE Location: D-112 | 5 |
| 47 | June 23, 2026 17:35-18:35 | IAAE Lecture — Using Subjective Beliefs Data for Demand and Production Estimation — Aureo de Paula (University College London) — Chair & Moderator: Silvia Goncalves (McGill University) Location: Grand Auditorium | 0 |
| 48 | June 23, 2026 18:35-20:00 | Welcome Reception | 0 |
| 49 | June 24, 2026 8:45-9:45 | IAAE Keynote — The Prestakes of Stock Market Investing — Francesco Bianchi (Johns Hopkins University) — Chair & Moderator: Francesco Ravazzolo (BI Norwegian Business School and University of Bolzano) Location: Grand Auditorium | 0 |
| 50 | June 24, 2026 9:45-10:15 | Coffee break | 0 |
| 51 | June 24, 2026 10:15-12:00 | CLIMATE AND THE MACROECONOMY 2 Location: B128 | 4 |
| 52 | June 24, 2026 10:15-12:00 | COINTEGRATION AND STRUCTURAL CHANGE Location: B009 | 4 |
| 53 | June 24, 2026 10:15-12:00 | FINANCIAL CONNECTEDNESS Location: D-115 | 4 |
| 54 | June 24, 2026 10:15-12:20 | FORECASTING FINANCIAL AND MACRO RISKS Location: D-111 | 5 |
| 55 | June 24, 2026 10:15-12:20 | INFLATION AND MACRO SHOCKS Location: D-112 | 5 |
| 56 | June 24, 2026 10:15-12:00 | JOB LOSS AND WORKERS Location: B129 | 4 |
| 57 | June 24, 2026 10:15-12:00 | LOCAL PROJECTIONS AND SVARS 3 Location: B008 | 4 |
| 58 | June 24, 2026 10:15-12:20 | MACHINE LEARNING METHODS Location: D-107 | 5 |
| 59 | June 24, 2026 10:15-12:20 | MICROECONOMETRIC METHODS 2 Location: D-110 | 5 |
| 60 | June 24, 2026 10:15-12:00 | MONETARY UNION MACROECONOMICS Location: D-113 | 4 |
| 61 | June 24, 2026 10:15-12:20 | PANEL TIME SERIES Location: D-105 | 5 |
| 62 | June 24, 2026 10:15-12:20 | SUBJECTIVE EXPECTATIONS Location: D-114 | 5 |
| 63 | June 24, 2026 10:15-12:00 | TREATMENT EFFECTS Location: D-106 | 4 |
| 64 | June 24, 2026 12:00-13:30 | Lunch | 0 |
| 65 | June 24, 2026 13:30-15:15 | ASSET PRICE BUBBLES Location: D-110 | 4 |
| 66 | June 24, 2026 13:30-15:15 | CAUSAL INFERENCE WITH PANEL DATA Location: B009 | 4 |
| 67 | June 24, 2026 13:30-15:15 | DEMAND ESTIMATION Location: D-106 | 4 |
| 68 | June 24, 2026 13:30-15:15 | ECONOMETRIC THEORY Location: D-107 | 4 |
| 69 | June 24, 2026 13:30-15:15 | FUNCTIONAL DATA FORECASTING Location: D-111 | 4 |
| 70 | June 24, 2026 13:30-15:15 | GLOBAL INFLATION Location: D-105 | 4 |
| 71 | June 24, 2026 13:30-15:15 | GOVERNMENT SPENDING AND GROWTH Location: B128 | 4 |
| 72 | June 24, 2026 13:30-15:15 | INEQUALITY AND IDENTITY Location: E002 | 3 |
| 73 | June 24, 2026 13:30-15:15 | INTERNATIONAL POLICY SPILLOVERS Location: B129 | 4 |
| 74 | June 24, 2026 13:30-15:15 | MACHINE LEARNING APPLICATIONS Location: D-115 | 3 |
| 75 | June 24, 2026 13:30-15:15 | MONETARY POLICY Location: D-112 | 4 |
| 76 | June 24, 2026 13:30-15:15 | NETWORK INTERFERENCE Location: B008 | 4 |
| 77 | June 24, 2026 13:30-15:15 | REGIME SWITCHING MODELS Location: D-114 | 4 |
| 78 | June 24, 2026 13:30-15:15 | SCORE-DRIVEN MODELS Location: D-113 | 4 |
| 79 | June 24, 2026 15:15-15:45 | Coffee break | 0 |
| 80 | June 24, 2026 15:45-17:30 | AI AND LABOR MARKETS Location: D-115 | 4 |
| 81 | June 24, 2026 15:45-17:30 | BOND MARKETS AND PORTFOLIOS Location: D-113 | 4 |
| 82 | June 24, 2026 15:45-17:30 | CHILD DEVELOPMENT AND EDUCATION Location: E002 | 4 |
| 83 | June 24, 2026 15:45-17:30 | CLIMATE AND THE MACROECONOMY 1 Location: B128 | 4 |
| 84 | June 24, 2026 15:45-17:30 | DISTRIBUTIONAL INFLATION Location: B009 | 4 |
| 85 | June 24, 2026 15:45-17:30 | HIGH-DIMENSIONAL VOLATILITY MODELS Location: D-110 | 4 |
| 86 | June 24, 2026 15:45-17:30 | IDENTIFICATION OF MACRO SHOCKS Location: D-105 | 4 |
| 87 | June 24, 2026 15:45-17:30 | INSTITUTIONAL GOVERNANCE, CONFLICT AND SOCIAL DYNAMICS Location: B129 | 4 |
| 88 | June 24, 2026 15:45-17:30 | INSTRUMENTAL VARIABLES 1 Location: B008 | 3 |
| 89 | June 24, 2026 15:45-17:30 | MACHINE LEARNING AND FACTOR MODELS Location: D-107 | 4 |
| 90 | June 24, 2026 15:45-17:30 | MONETARY POLICY TRANSMISSION 2 Location: D-114 | 4 |
| 91 | June 24, 2026 15:45-17:30 | NONLINEAR TIME SERIES MODELS Location: D-106 | 4 |
| 92 | June 24, 2026 15:45-17:30 | PANEL DATA METHODS 2 Location: D-112 | 4 |
| 93 | June 24, 2026 15:45-17:50 | PLATFORMS AND MARKET STRUCTURE Location: D-111 | 5 |
| 94 | June 24, 2026 17:30-18:00 | IAAE General Assembly. Chair: M. Chauvet (University of California Riverside) Location: Grand Auditorium | 0 |
| 95 | June 24, 2026 18:30-23:00 | Conference dinner | 0 |
| 96 | June 25, 2026 8:45-9:45 | IAAE Keynote — Data-driven Nests — Elena Manresa (Princeton University) — Chair & Moderator: Martin Weidner (Oxford University) Location: Grand Auditorium | 0 |
| 97 | June 25, 2026 9:45-10:15 | Coffee break | 0 |
| 98 | June 25, 2026 10:15-12:00 | AUCTIONS Location: D-107 | 3 |
| 99 | June 25, 2026 10:15-12:00 | DENSITY FORECASTING Location: D-105 | 4 |
| 100 | June 25, 2026 10:15-12:00 | EDUCATION AND THE LABOR MARKET Location: B128 | 3 |
| 101 | June 25, 2026 10:15-12:00 | FIRMS AND SHOCK TRANSMISSION Location: D-115 | 4 |
| 102 | June 25, 2026 10:15-12:00 | HIGH-DIMENSIONAL METHODS Location: D-106 | 4 |
| 103 | June 25, 2026 10:15-12:00 | HOUSEHOLD CONSUMPTION Location: B129 | 4 |
| 104 | June 25, 2026 10:15-12:00 | IDENTIFICATION OF TREATMENT EFFECTS Location: B008 | 4 |
| 105 | June 25, 2026 10:15-12:00 | INEQUALITY AND MOBILITY Location: D-113 | 4 |
| 106 | June 25, 2026 10:15-12:00 | INSTRUMENTAL VARIABLES 2 Location: B009 | 4 |
| 107 | June 25, 2026 10:15-12:20 | MACRO-FINANCE MODELS Location: D-112 | 5 |
| 108 | June 25, 2026 10:15-12:00 | MACROECONOMIC POLICY AND SHOCK IDENTIFICATION Location: D-108 | 4 |
| 109 | June 25, 2026 10:15-12:00 | MIGRATION AND LABOR MARKETS Location: E002 | 4 |
| 110 | June 25, 2026 10:15-12:00 | MONETARY POLICY COUNTERFACTUALS Location: D-110 | 4 |
| 111 | June 25, 2026 10:15-12:00 | SYSTEMIC RISK AND BUBBLES Location: D-111 | 4 |
| 112 | June 25, 2026 10:15-12:00 | WEATHER AND THE ECONOMY Location: D-114 | 4 |
| 113 | June 25, 2026 12:00-13:30 | Lunch | 0 |
| 114 | June 25, 2026 13:30-15:15 | AUTOREGRESSIVE MODELS Location: D-107 | 4 |
| 115 | June 25, 2026 13:30-15:15 | BANKS AND FIRM OUTCOMES Location: E002 | 4 |
| 116 | June 25, 2026 13:30-15:15 | CLIMATE POLICY Location: B128 | 4 |
| 117 | June 25, 2026 13:30-15:15 | DISTRIBUTIONAL TREATMENT EFFECTS Location: B008 | 4 |
| 118 | June 25, 2026 13:30-15:15 | EDUCATION AND HEALTH Location: B129 | 4 |
| 119 | June 25, 2026 13:30-15:15 | FORECASTING Location: D-105 | 4 |
| 120 | June 25, 2026 13:30-15:15 | GEOPOLITICAL RISK Location: D-114 | 4 |
| 121 | June 25, 2026 13:30-15:15 | INFLATION ANCHORING AND RISK PREMIA Location: D-113 | 4 |
| 122 | June 25, 2026 13:30-15:15 | INFLATION DYNAMICS Location: D-111 | 4 |
| 123 | June 25, 2026 13:30-15:15 | MONETARY POLICY TRANSMISSION 1 Location: D-110 | 4 |
| 124 | June 25, 2026 13:30-15:15 | NONPARAMETRIC INFERENCE Location: B009 | 4 |
| 125 | June 25, 2026 13:30-15:15 | PANEL DATA METHODS 1 Location: D-106 | 4 |
| 126 | June 25, 2026 13:30-15:15 | ROBUSTNESS AND REPLICATION Location: D-112 | 3 |
| 127 | June 25, 2026 13:30-15:15 | VOLATILITY AND RISK PREMIA Location: D-115 | 4 |
| 128 | June 25, 2026 15:15-15:45 | Coffee break | 0 |
| 129 | June 25, 2026 15:45-17:30 | CHILD PENALTIES AND FAMILY POLICY Location: B128 | 4 |
| 130 | June 25, 2026 15:45-17:30 | CLIMATE FINANCE Location: D-112 | 4 |
| 131 | June 25, 2026 15:45-17:30 | DISCRETE CHOICE MODELS Location: D-107 | 4 |
| 132 | June 25, 2026 15:45-17:30 | FINANCIAL TIME SERIES MODELS Location: D-111 | 4 |
| 133 | June 25, 2026 15:45-17:30 | FISCAL POLICY Location: D-115 | 4 |
| 134 | June 25, 2026 15:45-17:30 | GROWTH-AT-RISK AND NOWCASTING Location: B009 | 4 |
| 135 | June 25, 2026 15:45-17:30 | LOCAL PROJECTIONS AND SVARS 4 Location: B008 | 3 |
| 136 | June 25, 2026 15:45-17:30 | MACROECONOMETRICS Location: D-110 | 4 |
| 137 | June 25, 2026 15:45-17:30 | NATURAL DISASTERS Location: B129 | 4 |
| 138 | June 25, 2026 15:45-17:30 | PANEL AND NETWORK MODELS Location: D-106 | 4 |
| 139 | June 25, 2026 15:45-17:30 | SYNTHETIC CONTROL AND EVENT STUDIES Location: D-105 | 4 |
| 140 | June 25, 2026 15:45-17:30 | TEXT ANALYSIS IN MACROECONOMICS Location: D-114 | 4 |
| 141 | June 25, 2026 15:45-17:30 | WORK AND TECHNOLOGY 1 Location: D-113 | 3 |
141 sessions, 497 papers, and 0 presentations with no associated papers |
|---|
IAAE 2026 international Association for Applied Econometrics |
|---|
Detailed List of Sessions |
| Session 1: Welcome Remarks - Marcelle Chauvet (University of California-Riverside), Paulo M. M. Rodrigues (Nova School of Business and Economics) June 23, 2026 8:15 to 8:30 Location: Grand Auditorium |
|---|
| Session 2: JAE Lecture: Orthogonal Moments in Likelihood Models — Stéphane Bonhomme (University of Chicago) — Discussants: Ivan Fernandez-Val (Boston University) and Chris Muris (McMaster University) June 23, 2026 8:30 to 9:45 Location: Grand Auditorium |
| Session 3: Coffee break June 23, 2026 9:45 to 10:15 |
| Session 4: ASSET RETURN PREDICTABILITY June 23, 2026 10:15 to 12:20 Location: D-113 |
| Session Chair: Martin Enilov, University of Southampton |
Disagreement and Uncertainty: Enhancing Equity Premium Forecasts with Short InterestAbstractOur study demonstrates that aggregate short interest is a stronger predictor of the equity risk premium during periods of heightened financial uncertainty. Our findings strengthen the hypothesis that, during such times, disagreement among investors and downward pressure on prices allow well-informed short sellers to generate more accurate forecasts. Also, when both financial uncertainty and investor disagreement are high, its predictive ability is more pronounced. Our results offer valuable insights into enhancing equity return forecasting, an ongoing challenge that has sparked considerable debate in recent literature. |
| Presented by: Michalis Stamatogiannis, University of Liveprool |
The Risk-Based Origins of Asymmetric Information in a Government Bond MarketAbstractWe study how asymmetric information is transmitted in government bond markets using a novel high-frequency dataset covering the near-universe of dealer-to-customer transactions. We depart from the asset-specific focus of the microstructure literature by extracting traded-risk factors that capture common variation in bond order flows and returns across maturities. These factors closely align with the level, slope, and curvature components of the yield curve. Institutional order flows in the level and slope factors exhibit strong and persistent price impacts consistent with informed trading, while bond-specific order flows contain little incremental information once factor-level trading is accounted for. We document substantial heterogeneity across institutional investors. Hedge funds and asset managers display the most persistent effects, whereas banks, pension and insurance companies, and non-financial institutions exhibit weaker or no evidence of informed trading. Asymmetric information is also strongly state-dependent and peaks during periods of moderate investor disagreement about upcoming macroeconomic announcements, weakens and becomes more concentrated when disagreement is high, and largely disappears when disagreement is low. |
| Presented by: Gregory Bauer, University of Guelph |
Conditional Expected Option ReturnsAbstractThis paper extends option-implied formulas of the expected market return to obtain conditional expected option returns. Relative to existing pricing kernel specifications, we show that allowing for heterogeneous preferences towards left and right tails is key to accurately fit realized call and put option returns. Using our time series of conditional expected option returns, we find that: alphas with respect to conditional expected market returns are strongly significant, where both calls and puts have negative betas; compensation for upside, downside and variance risk is higher during economic downturns and high-volatility periods; and there is a strong factor structure in option returns, dominated by level and moneyness-slope factors. |
| Presented by: Gustavo Freire, Nova School of Business and Economics |
A Greedy Approach to High-Dimensional Asset Pricing under Weak IdentificationAbstractWe propose a new approach to factor selection that identifies factors with incremental information for explaining the cross-section of asset returns in large factor spaces. Our method combines an orthogonal greedy selection algorithm with identification-strength diagnostics to construct a hierarchical ordering of factors and quantify each factor’s unique contribution to cross-sectional return variation. The procedure (1) identifies economically important factors, (2) accounts for weak identification by explicitly distinguishing identification strength from pricing-error reduction, thereby avoiding the selection of factors that improve model fit but do not meaningfully identify risk premia; (3) establishes a transparent hierarchy of factor importance, (4) measures each factor’s incremental explanatory power independent of previously selected factors, and (5) addresses a critical question in asset pricing literature: which factors can be subsumed by others and to what extent. In simulation studies under a range of factor structures, we show that our method outperforms conventional stepwise selection procedures and LASSO-based approaches in terms of selection accuracy and stability. Empirically, applying our approach to equity, corporate bond, and options markets, we find that the set of selected factors is dense rather than sparse, highlighting the importance of incremental and jointly informative factors in explaining asset returns. |
| Presented by: lingwei kong, University of Groningen |
Enhancing cryptocurrency returns forecasting with a novel dispersion-embedded deep learning architectureAbstractCryptocurrencies have become influential financial assets, with pricing increasingly driven by investor behaviour, risk perceptions and speculative sentiment. Their growing association with energy-transition debates, ranging from energy consumption to greener consensus mechanisms, raises important economic and policy questions regarding the need for appropriate predictive models to support informed investment and risk management. Motivated by these considerations, we develop a novel dispersion-embedded deep-learning framework that integrates behavioural-finance insights with dispersion-based forecasting signals, demonstrating improved predictive performance of cryptocurrency returns and providing clearer insight into pricing dynamics within evolving energy-related financial ecosystems. Our approach integrates the Cross-Sectional Absolute Deviation (CSAD) measure of return dispersion into advanced forecasting architectures for cryptocurrency returns. The CSAD measure is constructed using daily data on 9,465 active and defunct cryptocurrencies and is employed to forecast the returns of the top 10 energy-efficient (clean) and top 10 energy-intensive (dirty) assets, allowing for an assessment of their behavioural and structural heterogeneity. Embedding CSAD within Bidirectional Long Short-Term Memory (BiLSTM) and hybrid deep learning models, including ANN, CNN, RNN, and Additive Attention variants, substantially improves prediction accuracy relative to traditional econometric and standard machine learning approaches. Our results indicate that return dispersion carries economically meaningful forecasting information, capturing behavioural patterns that enhance forecasting performance. Performance gains are found to be most pronounced for clean cryptocurrencies, suggesting higher informational efficiency and behavioural asymmetry. |
| Presented by: Martin Enilov, University of Southampton |
| Session 5: CAUSAL INFERENCE METHODS June 23, 2026 10:15 to 12:00 Location: D-106 |
| Session Chair: Julius Schäper, University of Zurich |
Testing IV Validity and LATE Interpretation using Flexible Covariate SpecificationsAbstractBuilding on the testable implications for IV validity underlying local average treatment effect (LATE) estimation, we (i) propose a simple testing procedure that may accommodate high-dimensional covariates and (ii) demonstrate that it can also detect biases arising from misspecified IV regression models. While recent research has highlighted the importance of a correct covariate specification, existing IV validity tests are not designed to capture this source of bias. Simulation studies strongly suggest that the test performs well at detecting violations of conditional independence, violations of the exclusion restriction, and biases arising from covariate misspecification. |
| Presented by: Matthias Westphal, Fernuniversität in Hagen |
Bootstrapping an Augmented LR Test for MediationAbstractStandard mediation tests in causal mediation analysis, including the Sobel and likelihood ratio (LR) tests, often exhibit low power when true effects are near zero. Bootstrap confidence-interval procedures commonly used in practice do not remedy this limitation. This problem is partly solved by augmenting the LR critical region under distributional assumptions. In this paper, we develop a grid bootstrap implementation of the augmented LR test that relaxes Gaussian assumptions and accommodates asymmetric $t$-statistics. Exploiting invariance properties, we show that bootstrap samples need only be generated at the origin, enabling efficient approximation over the parameter grid without repeated resampling. We develop a five-step adaptive algorithm to determine critical values and optimize augmentation regions. Monte Carlo simulations demonstrate accurate size control in samples as small as $n=25$, even under heavy-tailed or skewed distributions. |
| Presented by: Noud Giersbergen, University of Amsterdam |
Global Testing in Multivariate Regression Discontinuity DesignsAbstractRegression discontinuity (RD) designs with multiple running variables arise in a growing number of empirical applications, including geographic boundaries and multi-score assignment rules. Although recent methodological work has extended estimation and inference tools to multivariate settings, far less attention has been devoted to developing global testing methods that formally assess whether a discontinuity exists anywhere along a multivariate treatment boundary. Existing approaches perform well in large samples, but can exhibit severe size distortions in moderate or small samples due to the sparsity of observations near any particular boundary point. This paper introduces a complementary global testing procedure that mitigates the small-sample weaknesses of existing multivariate RD methods by integrating multivariate machine learning estimators with a distance-based aggregation strategy, yielding a test statistic that remains reliable with limited data. Simulations demonstrate that the proposed method maintains near-nominal size and strong power, including in settings where standard multivariate estimators break down. The procedure is applied to an empirical setting to demonstrate its implementation and to illustrate how it can complement existing multivariate RD estimators. |
| Presented by: Artem Samiahulin, University of Illinois at Urbana-Champaign |
Residualised Treatment Intensity and the Estimation of Average Partial EffectsAbstractThis paper introduces R-OLS, an estimator for the average partial effect (APE) of a continuous treatment variable on an outcome variable in the presence of non-linear and non-additively separable confounding of unknown form. Identification of the APE is achieved by generalising Stein's Lemma \citep{stein1981estimation}, leveraging an exogenous error component in the treatment along with a flexible functional relationship between the treatment and the confounders. The identification results for R-OLS are used to characterise the properties of Double/Debiased Machine Learning \citep{Chernozhukov_et_al_2018_DebiasedML}, specifying the conditions under which the APE is estimated consistently. A novel decomposition of the ordinary least squares estimand provides intuition for these results. Monte Carlo simulations demonstrate that the proposed estimator outperforms existing methods, delivering accurate estimates of the true APE and exhibiting robustness to moderate violations of its underlying assumptions. The methodology is further illustrated through an empirical application to \citet{fetzer2019_austerity_brexit}. |
| Presented by: Julius Schäper, University of Zurich |
| Session 6: CREDIT AND THE MACROECONOMY June 23, 2026 10:15 to 12:00 Location: E002 |
| Session Chair: Matteo Santi, Bank of Italy |
Housing Tenure, Consumption and Household Debt: Life-Cycle Dynamics during a Housing BustAbstractThe housing bust in Spain was characterized by a significant and rapid drop in home ownership among the younger cohorts, a relatively homogeneous but significant decrease in consumption, and significant movements in the rent-to-house price ratio. To uncover the causes of these movements, we solve and estimate an equilibrium life-cycle model with non-linear income dynamics, mortgages, housing, and rental markets and simulate a series of counterfactual policy changes and macroeconomic conditions observed in Spain during the period. The lion's share of the observed drop in home ownership and consumption and the housing market dynamics can be explained by more cautious credit conditions and the major shift in income dynamics observed in Spain between the boom and bust phases. |
| Presented by: Julio Galvez, CUNEF Universidad |
The Macroeconomic Costs of Asymmetric Information in BankingAbstractWe study how public signals of firm quality shape credit allocation and aggregate outcomes. Using Portuguese administrative data and three quasi-experimental information shocks - winning a public procurement contract, a grant, or a government certification - we find that firms revealed as high quality expand borrowing, pay lower interest rates, default less, and diversify their banking relationships, reducing reliance on their main lender. We show similar effects in the United States by linking federal procurement awards to FR Y-14 loan records. To interpret these findings, we develop a credit-market model with adverse selection in which disclosure lowers monitoring needs and reallocates credit toward more productive firms. Embedded in a calibrated dynamic general equilibrium framework, the model allows us to quantify the efficiency costs of informational frictions and evaluate counterfactual disclosure policies. |
| Presented by: João Quelhas, Stockholm University |
Credit supply and demand: micro shocks and macro effectsAbstractWe extract the unexpected component of responses to the euro area Bank Lending Survey to derive bank-level measures of credit supply and demand shifts (credit surprises). We then evaluate their effects on bank-specific loan volumes and interest rates using local projections, and exploit banks’ reports on the drivers of reported changes to give them an economic interpretation. We then build granular instruments for loan supply and demand shocks in the euro area market for corporate loans, and use them to identify a structural Bayesian VAR model. Our results show that contractionary credit supply shocks lead to persistent declines in lending volumes and GDP and widen lending margins, while adverse credit demand shocks mainly reduce loan volumes, with limited effects on economic activity. |
| Presented by: Matteo Santi, Bank of Italy |
| Session 7: FORECASTING WITH MACHINE LEARNING June 23, 2026 10:15 to 12:00 Location: D-105 |
| Session Chair: Jeronymo Pinto, Brazilian Ministry of Labor |
Virtue or Mirage? Complexity in Exchange Rate PredictionAbstractThis paper examines whether the “virtue of complexity” (VoC), documented in equity return prediction, extends to exchange rate forecasting. Using Ridge regressions with Random Fourier Features (Ridge–RFF), we compare predictive performance against linear regression and the random walk benchmark across three sets of fundamentals—traditional monetary, expanded monetary and non-monetary, and Taylor-rule predictors—under rolling windows of 12, 60, and 120 months. Results offer a cautionary perspective. Complexity provides modest, localized gains: in very small samples with rich predictor sets, Ridge–RFF can outperform linear regression, but never delivers systematic improvements over the random walk. As training windows expand, Ridge–RFF quickly loses ground, while linear regression often matches or surpasses it, occasionally outperforming the random walk. Market-timing analyses reinforce this fragility: complexity-based strategies yield short-sample gains but are unstable, whereas linear and random walk strategies provide more robust value. Overall, evidence points to a limited virtue of complexity in FX forecasting. |
| Presented by: Rehim Kilic, FRB |
Tip the Scales: Using Large Language Models for Sectoral Nowcasting from Business Survey NarrativesAbstractWe investigate whether large language models (LLMs) can improve real-time nowcasts of German gross value added (GVA) by quantifying qualitative narratives from the ifo Business Climate press releases. Using zero-shot prompting, we convert the monthly sectoral textual commentary into activity-sentiment scores on a fixed scale. These scores are aggregated to the quarterly frequency using publication-consistent schemes and incorporated into expanding and rolling real-time forecast exercises. The results indicate that LLM-derived sentiment contains incremental information beyond forecasts based on conventional change in the sectoral ifo Business Climate Index. Forecast gains are concentrated around periods of heightened uncertainty, whereas improvements are smaller during exceptionally stable periods and in sectors with high measurement noise and large data revisions (notably construction). The results are robust across different LLM providers, prompting configurations, aggregation schemes, and samples. Overall, systematically quantified survey narratives can enhance short-term monitoring of economic activity. The paper provides a transparent and replicable framework for integrating LLM-based text analysis into real-time forecasting, complementing traditional survey indicators. |
| Presented by: Katja Heinisch, Halle Institute for Economic Research (IWH) |
Improving Biased Forecasts in Real TimeAbstractI develop three approaches to improve forecasts of macroeconomic variables in real time, dealing with complications including data revisions and structural instability. I consider forecasts that have been found to be biased in-sample, and I illustrate the ideas with forecasts of corporate profits as a share of GDP, using the Survey of Professional Forecasters. Even when bias is clear in-sample, the time-varying nature of the bias makes it difficult to improve upon the forecasts out-of-sample. Only in forecasting the most recent vintage of the data is there a significant reduction in root-mean-squared forecast errors. |
| Presented by: Dean Croushore, University of Richmond |
Real-time Tracking of Forecasting Models under Economic Uncertainty: An Estimation-Free Decision RuleAbstractTime series forecasting in economics has no one-size-fits-all solution, as structural change and non-stationarity favor different models across time. This paper proposes the Forecasting Optimization Agent (FOA), an adaptive framework that formulates model selection as a real-time tracking problem over a fixed set of candidate forecasts. Unlike traditional ensemble methods that require re-estimation, FOA operates as an estimation-free decision rule that identifies the loss-minimizing forecast in real-time, reconciling predictive accuracy with high computational efficiency. We provide a theoretical intuition for the update rule, highlighting how its design ensures numerical stability and rapid adaptation to shifting targets, with a formal proof of convergence over a fixed target. Empirical performance is evaluated using a large set of simulated data and two competitive benchmarks: the M4 Competition and the Survey of Professional Forecasters (SPF). Accuracy is assessed through the Model Confidence Set (MCS), enabling statistically robust comparisons. The results show that FOA consistently matches or improves upon strong benchmarks, particularly during periods of structural breaks, by rapidly identifying the most appropriate model without the overhead of recursive estimation. An open-source package is available on GitHub to ensure full reproducibility. |
| Presented by: Jeronymo Pinto, Brazilian Ministry of Labor |
| Session 8: LABOR MARKET SORTING June 23, 2026 10:15 to 12:20 Location: B128 |
| Session Chair: Alexander Sarango-Iturralde, Paris 1, Panthéon-Sorbonne |
Macroeconomic shocks and the decision to leave the teaching professionAbstractWe investigate whether macroeconomic shocks affect teacher attrition. We employ Chilean data on the universe of teachers, public and private, working during 2003-2023 and exploit plausibly exogenous shocks driven by changes in world copper prices. Results show that favorable macroeconomic conditions encourage teachers to leave the profession, especially if teachers have high performance evaluations or are near retirement. Overall, 54% of the effect of copper prices on teacher attrition is driven by changes in GDP, while 40% is driven by changes in interest rates. These percentages vary by teacher's age, with financial variables being more important for teachers near retirement. |
| Presented by: Diana Alessandrini, St. Francis Xavier University |
Nonlinear Sorting, Turnover, and Heterogeneous Job TrajectoriesAbstractWe study worker--firm sorting and labor-market dynamics in the Chilean formal sector using matched employer--employee administrative data. Estimating an AKM earnings decomposition, we document that assortative matching is highly nonlinear and concentrated at the upper tail: high-type workers are disproportionately employed at high-paying firms, while allocations across most of the market are close to random. This nonlinear sorting is linked to sharply heterogeneous mobility regimes. High-type workers experience long job tenures and strong attachment to formal employment, while low-type workers cycle across short-lived jobs and non-employment. Earnings dynamics amplify these differences: high-type workers benefit from steep within-job growth and upward firm mobility, while low-type workers rely on frequent transitions to generate modest gains. We identify heterogeneous job ladders as a central mechanism driving cumulative earnings inequality. We complement this evidence with the analysis of a job ladder model with heterogeneous agents. |
| Presented by: Gabriela Contreras, Central Bank of Chile |
Labor Market Polarization under Formal–Informal Dualism: Tasks, Wages, and Dutch Disease in ColombiaAbstractWe study how tasks, formality, and sectoral reallocation jointly shaped workforce adjustment in Colombia’s urban labor market from 1984 to 2019. Using harmonized task groups mapped from CNO-70, a long imputed informality series consistent with DANE benchmarks, and within between decompositions, we document a robust hollowing out of routine work. Between 1995 and 2019, the routine share falls by about 8.6 percentage points, while manual and abstract shares rise by roughly 3.9 and 4.7 points, and close to four fifths of the routine decline reflects within industry substitution rather than shifts across industries. Adjustment is disproportionately absorbed by the formal sector, whose share increases from 55.6 percent to 63.8 percent, alongside a formal sector tilt toward high skill services. We then show that macro shocks can amplify task reallocation. Reduced form city level estimates indicate that commodity boom forces associated with Dutch Disease, in particular real exchange rate appreciation and oil rent intensity, are linked to larger increases in polarization in cities that were initially more routine intensive, with effects concentrated during the 2003 to 2014 boom window. Wages do not sustain a stable U shape; instead, wage growth rotates across decades, and residual wages point to a rising abstract premium and a late improvement of manual relative to routine. Overall, the evidence is consistent with technological adjustment within industries interacting with boom driven relative price movements that shift employment away from routine intensive tradables toward service activities combining abstract and manual tasks. |
| Presented by: Alexander Sarango-Iturralde, Paris 1, Panthéon-Sorbonne |
Who Becomes an Inventor in Italy? The Role of Firms in Talent Discovery?AbstractThis paper investigates firm heterogeneity in identifying inventor talent and rewarding inventors for patent applications. Using Social Security employment records in Italy linked to patent applications, we find substantial firm heterogeneity in the rate at which employees become inventors. Young workers are less likely to apply for their first patents at a lower-wage firm. The gap between firms disappears, however, for experienced inventors. Upon the initial patent application, young workers receive a 5-9 log-point wage increase. We build a model of employer learning and incentive contracts to explain our findings, especially why low-wage firms set a higher bonus for new inventors than high-wage firms despite similar retention rates. |
| Presented by: Sabrina Di Addario, Bank of Italy |
| Session 9: LOCAL PROJECTIONS AND SVARS 1 June 23, 2026 10:15 to 12:00 Location: B008 |
| Session Chair: Luca Neri, UCLouvain |
Targeted Local ProjectionsAbstractLocal projection (LP) and structural vector autoregression (SVAR) are commonly employed to estimate dynamic causal effects of macroeconomic policies at multiple horizons. With enough lags as controls, LP estimators have little bias but their variance can increase with the horizon due to accumulating additional shocks. Because they typically employ fewer lags or suffer from local misspecification, SVAR estimators typically incur higher bias, but their variance decreases with the horizon due to exponentiation. We propose to target the LP estimators towards their SVAR counterparts - constructed with fewer lags than LP at each horizon - to reduce their variance at the cost of incurring some bias. The resulting targeted LP estimator is a linear combination of the LP and SVAR estimators. We propose choosing this linear combination optimally to minimize the mean-squared error of the new estimator. Our simulations show that, under a locally misspecified SVAR model, targeting substantially reduces the LP variance at longer horizons while maintaining near-nominal coverage in small samples when a double bootstrap is employed. |
| Presented by: Aleksei Nemtyrev, Tilburg University |
On Noncausal Structural VARs in Macro-FinanceAbstractIn this paper, we revisit the well-known small VAR model investigated by Stock and Watson (2001) and test whether the shocks identified by the authors could also have been noncausal. We focus on the existence of noncausal components identified by the generalized covariance (GCov) estimator. Since it is observed in the literature that GCov may suffer from estimation instabilities, we first evaluate the small-sample performance of the GCov estimator within a VAR(2) framework. Next, we examine the presence of noncausal components in the Stock and Watson three-dimensional VAR model under the different Taylor rule specifications considered in their analysis. We then introduce a VARX factor-filtering approach and show that removing common macroeconomic components eliminates noncausality. Finally, we compare impulse responses from the filtered and original data to evaluate how factor filtering affects the transmission of monetary policy shocks. |
| Presented by: Lison Christiaens, Maastricht University and Liege University |
Beyond Validity: SVAR Identification Through the Proxy ZooAbstractThis paper develops a framework for robust identification in SVARs when researchers face a zoo of proxy variables. Instead of imposing exact exogeneity, we introduce generalized ranking restrictions (GRR) that bound the relative correlation of each proxy with the target and non-target shocks through a continuous proxy-quality parameter. Combining GRR with standard sign and narrative restrictions, we characterize identified sets for structural impulse responses and show how to partially identify the proxy-quality parameter using the joint information contained in the proxy zoo. We further develop sensitivity and diagnostic tools that allow researchers to assess transparently how empirical conclusions depend on proxy exogeneity assumptions and the composition of the proxy zoo. A simulation study shows that proxies constructed from sign restrictions can induce biased proxy-SVAR estimates, while our approach delivers informative and robust identified sets. An application to U.S.\ monetary policy illustrates the empirical relevance and computational tractability of the framework. |
| Presented by: Luca Neri, UCLouvain |
| Session 10: MACROECONOMIC DYNAMICS June 23, 2026 10:15 to 12:20 Location: D-114 |
| Session Chair: Stefano Fasani, Lancaster University |
Probabilistic Narrative Identification: Evidence from Tax PolicyAbstractThis paper generalizes narrative identification by interpreting documentary classifications as probabilistic information about latent exogeneity. Traditional narrative approaches treat selected policy actions as fully exogenous, implicitly imposing a zero-contamination restriction on a chosen subset of observations. Instead, I treat exogeneity as a latent binary attribute and use narrative evidence to construct a probability that a policy action belongs to the exogenous class. Identification proceeds at the level of reduced-form moments: variation in narrative support disciplines outcome–policy comovement, allowing the orthogonal-class moment to be recovered without requiring observation-level structural shocks. I implement the approach by mapping policymakers’ stated motivations into exogeneity probabilities while preserving traceability to source documents. Empirically, I first validate the measure against the benchmark classifications of Romer and Romer (2010) and Cloyne (2013). I then construct a new dataset of routine U.S. federal tax actions, expanding the available variation for narrative analysis. The resulting macroeconomic responses are closely aligned with the canonical narrative tax literature. Finally, I build a new long-run narrative dataset of French VAT changes from 1960 to 2024 and combine it with UK VAT changes from Cloyne’s dataset. The pooled UK–France evidence shows that VAT increases are contractionary and generate a short-lived increase in inflation, consistent with tax pass-through. |
| Presented by: Tom Pesso, Universitat Pompeu Fabra |
Estimating Asymmetric Hysteresis EffectsAbstractThis paper explores the potential asymmetry of hysteresis effects across two volatility regimes in the U.S. economy. We employ a bivariate VAR framework with Markov-switching volatility and measure the degree of hysteresis effects through the correlation between aggregate demand and supply shocks, following Cover et al. (2006). For identification, we impose one long-run and one short-run restriction, along with the heteroskedasticity-based approach of Rigobon (2003). Our empirical results suggest little evidence of hysteresis effects in either the high-volatility or low-volatility regime before the mid-1980s. During this period, only aggregate supply shocks had permanent effects on output, and recessions were mainly associated with transitory demand shocks. In contrast, since the mid-1980s, we find strong evidence of hysteresis effects during the high-volatility regime (recessions), whereas the evidence during the low-volatility regime (booms) is less compelling. In this later period, permanent demand shocks dominate output fluctuations during recessions, whereas aggregate supply shocks contribute substantially during expansions. |
| Presented by: Heejee Chang, University of Washington |
Inflation (de-)anchoring in the euro areaAbstractThis article measures the degree of potential de-anchoring of inflation expectations in the euro area vis-à-vis the inflation objective of the European Central Bank (ECB). The long-term mean of inflation expectations, pi*, is allowed to vary through time in a model taking market- and survey-based information into account. Survey-based information is in particular shown to play a crucial role in identifying pi*. Estimates of pi* have been close to 2% since the mid-2000s, indicating that long-term inflation expectations have overall remained well-anchored to the ECB’s inflation objective. Similarly, medium-term inflation expectations appear to have been broadly anchored, while shorter-term expectations (up to the two-year horizon) have shown tentative signs of de-anchoring. Using backcasted ILS rates, estimates of pi* are much above 2% in the early 1990s, but they converge to levels below 2% by the end of the decade when the ECB was established. |
| Presented by: Andreea Liliana Vladu, European Central Bank |
Retail inventories and inflation dynamics: The price margin channelAbstractUsing industry-level panel data and plausibly exogenous variation in supply conditions, we estimate the elasticity of retail price margins with respect to inventories along the retailer’s optimal pricing curve. We find that this elasticity is negative and statistically significant, implying that lower finished-good inventories lead to higher price margins. We assess the implications of this channel for inflation dynamics within a New Keynesian Phillips curve (NKPC) framework that links inventories to retailers’ markup behavior. Incorporating the inventory-sales ratio into the NKPC markedly improves the model’s empirical fit and helps account for two notable recent inflation episodes: the missing disinflation of 2009–2011 and the COVID-era surge. |
| Presented by: Julio Ortiz, Federal Reserve Board |
Inflation Uncertainty and Unemployment in the Long RunAbstractThis paper employs the frequency domain to examine the relationship between inflation uncertainty and unemployment. Inflation uncertainty explains 27% and 19% of the forecast error variance of unemployment in the long run for the US and UK, respectively, while its effects on other variables are limited to the short run. The analytical solution of a New-Keynesian model with downward nominal wage rigidity (DNWR) shows that higher inflation uncertainty raises long-run unemployment. With higher nominal volatility, DNWR limits downward wage adjustments, raising expected unemployment and extending the inflation-unemployment relationship beyond the short-run Phillips Curve. |
| Presented by: Stefano Fasani, Lancaster University |
| Session 11: NETWORKS June 23, 2026 10:15 to 12:00 Location: D-107 |
| Session Chair: Federico Belotti, University of Rome Tor Vergata |
Network formation with multidimensional unobserved heterogeneityAbstractWe study the formation of networks among agents whose preferences exhibit degree heterogeneity and homophily in both observed and unobserved characteristics. The presence of unobserved homophily complicates the identification of coefficients related to observed homophily. To address this challenge, we propose an identification approach that is implemented in two steps. First, we construct tetrad-level inequalities in which degree heterogeneity is canceled out. Next, the observed and unobserved homophily parameters are identified through a spectral decomposition. We propose a consistent estimator under many-node asymptotics. We show that the estimator obtains an asymptotic normal distribution and demonstrate its finite-sample performance with Monte Carlo simulations. Using a network dataset from Indian villages, we empirically investigate homophilic preferences among villagers in the formation of risk-sharing networks. Our results suggest that naive approaches to handling unobserved traits lead to significant bias, while failing to detect the underlying economic drivers of risk-sharing, such as income similarity. This highlights the critical importance of properly controlling for multidimensional unobserved heterogeneity. |
| Presented by: Suyong Song, University of Iowa |
What does outcome similarity capture? Identification and inference in pairwise regressions of peer effectsAbstractA growing number of empirical studies analyze peer effects via pairwise regressions, assessing whether the similarity of individual outcomes within a pair is higher when the two agents are connected. We provide the first analysis of identification and inference in these pairwise regressions. We find major problems with this design. We show, first, that the pairwise coefficient picks up clustering and that this bias is not solved by network randomization. Second, there is no systematic relationship between the sign and size of peer effects and the sign and size of the pairwise coefficient. Third, the pairwise parameter generally lacks a straightforward causal interpretation because of interference. Finally, we establish a central limit theorem for the pairwise OLS estimator and propose a new dyadic-robust variance |
| Presented by: Zheng Wang, New York University Abu Dhabi |
Statistical inference in large multi-way networksAbstractWe propose a new method to estimate structural parameters in multi-way networks while controlling for rich structures of fixed effects. The method is based on a series of classification tasks and is agnostic to both the number and structure of fixed effects. In contrast to full maximum likelihood approaches, our estimator does not suffer from the incidental parameter problem. For sparsely connected networks, it is also computationally faster than PPML. We provide empirical evidence that our estimator yields more reliable confidence intervals than PPML and its bias-correction strategies. These improvements hold even under model misspecification and are more pronounced in sparse settings. While PPML remains competitive in dense, low-dimensional data, our approach offers a robust alternative for multi-way models that scales efficiently with sparsity. The method is applied to study the causal effect of a policy reform on spatial accessibility to health care in France. |
| Presented by: Lucas Resende, ENSAE |
Grouped patterns of heterogeneity across time and space: A hierarchical clustering approachAbstractThis paper revisits the grouped fixed effects framework of Bonhomme and Manresa (2015), enhancing both model flexibility and computational stability. We propose performing the assignment step using a hierarchical clustering algorithm based on a convex combination of two dissimilarity matrices: one constructed from unit-specific residual trajectories over time, and the other from exogenous distances across statistical units, such as geographical distance. This design allows the method to capture complex latent structures while reducing sensitivity to random initializations, a known limitation of the k-means algorithm, and relaxing the strong group-separation assumption in Bonhomme and Manresa (2015). The estimator preserves the iterative structure of the original grouped fixed effects method but improves its robustness and interpretability. Monte Carlo simulations demonstrate that HGFE performs well even in the presence of weak group separation and moderate panel dimensions. |
| Presented by: Federico Belotti, University of Rome Tor Vergata |
| Session 12: OIL AND ENERGY SHOCKS June 23, 2026 10:15 to 12:00 Location: D-112 |
| Session Chair: Ece Fisgin, Federal Reserve Board of Governors |
The Natural Level of Oil PricesAbstractThis paper has two original contributions. First, it introduces the concept of a natural level for real oil prices, a benchmark that reflects the price consistent with long-run supply and demand fundamentals in the absence of transitory shocks. It is based on cointegration, where oil price is linked to market fundamentals to elicit a long-term price that serves as an anchor for oil price fluctuations and expectations. This framework helps distinguish between temporary dislocations and structural shifts in oil markets, a distinction particularly relevant during episodes of heightened volatility, such as the 2022 Russian invasion of Ukraine. While spot prices surged in response to geopolitical tensions, long-term prices remained relatively stable, and the deviation from the natural level gradually dissipated. Second, we chose market fundamentals series that are able to capture medium to long term variation in oil prices such as global industrial production, OPEC oil production, and non-OPEC oil production. Splitting oil production between these two set of players is critical to construct our natural level for real oil prices. Since oil prices changes are very important to determine the rate of inflation, having a natural level for real oil prices can act as a guidance for central banks to determine monetary policy in the medium to long term, where the effects of oil shocks would have dissipated. Indeed, our empirical results suggest that such transitory shocks typically dissipate within 18 to 24 months, with oil prices converging back to their natural level over this horizon. For the same reason, oil producers could also rely on a natural level of real oil prices for capital requirements. When we evaluate the predictive power of our model relative to standard benchmarks including a random walk, forward prices, and well-known statistical filters, we find a relevant improvement in forecasting accuracy. We also examine the term structure of oil futures and estimate risk premia, discussing how the natural level can shape market expectations. Al in all, the framework considered in this paper helps monetary and energy policymakers better assess whether oil-driven inflationary pressures are likely to persist or fade, depending on whether prices reflect fundamental shifts or temporary shocks. It also gives investors a clearer basis for managing exposure to oil-related assets and evaluating portfolio risks tied to commodity markets. |
| Presented by: João Issler, Getulio Vargas Foundation |
Barrels of Trouble: The change in correlation between oil and the dollarAbstractThe unconditional correlation between oil price and the dollar, historically negative, has switched sign roughly around 2020. As oil is priced in dollars, this switch amplifies the swings in oil price in domestic (non-US) currency terms. But is this switch due to an combination of transitory shocks or to a permanent structural shift? The latter is true: the response of the dollar exchange rate to oil shocks has changed sign. This is shown by leveraging on a novel weekly instrument for the identification of oil news shocks, that I use to separately identify the effects of a shock in the pre- and post-2020 periods. The paper also quantifies the importance of this mechanism for Euro Area inflation, finding that the switch will amplify inflation dynamics going forward. |
| Presented by: Riccardo Degasperi, Bank of Italy |
Power Play: How Structural Shocks Drive European Electricity MarketsAbstractThis paper investigates the identification and significance of structural shocks across European electricity markets. We analyze the response of fossil fuel prices, renewable energy generation, electricity demand, and day-ahead electricity prices to structural shocks. Focusing on Germany and Italy—two countries highly exposed to Russian gas imports but with substantial and growing renewable energy sectors— we estimate a Structural Vector Autoregression (SVAR) model. Our results provide strong empirical evidence of a merit-order effect from renewable energy sources and highlight the presence of robust cost-pass-through dynamics from fossil fuel markets onto electricity pricing. |
| Presented by: Francesco Ravazzolo, Libera Università di Bolzano |
The Design and Effect of Tariff Retaliation: Evidence from the European UnionAbstractWe show that the EU’s 2018 retaliation against US steel and aluminum tariffs targeted goods with low US import dependence and high substitutability. For the majority of tariffed goods, the US import share declined notably and remained below pre-2018 levels even after the retaliatory tariffs were lifted, reflecting asymmetric effects of tariffs on trade diversion. Moreover, although the retaliatory tariffs were instantly and fully passed through to EU importers, the retaliation did not lead to domestic price pressures as we find no evidence for inflationary effects on consumer and producer prices. |
| Presented by: Ece Fisgin, Federal Reserve Board of Governors |
| Session 13: POLICY INTERVENTIONS June 23, 2026 10:15 to 12:20 Location: B129 |
| Session Chair: Manachaya Uruyos, Chulalongkorn University |
Policy Learning with Observational Data : The Case of Hepatitis C Treatment for HIV/HCV Co-Infected PatientsAbstractDecision-makers frequently must choose a single action from a finite set of alternatives—for example, physicians selecting a treatment, investors choosing a portfolio risk level, or insurers setting premiums. To improve outcomes, policymakers often issue policy rules or guidelines to inform such choices. This paper develops a method to derive optimal policy rules from observational data under relatively weak assumptions about the underlying structure of the heterogeneous sampled population. Conditional average treatment effects are consistently estimated using a weighted K-means algorithm, assuming the outcome model is correctly specified within each homogeneous subgroup. Optimal and feasible policy rules are then implemented via a standard decision tree, allowing for both perfect and imperfect adherence to treatment. The methodology is applied to treatment options for Hepatitis C (HCV) among patients co-infected with human immunodeficiency virus (HIV), a setting in which no uniform guideline exists for modern pharmaceutical therapies. The results identify a subgroup of patients with approximately an 80% probability of spontaneous HCV clearance without treatment. Estimation results also show that reallocating treatments among treated individuals could have reduced total treatment costs by CAN$2.7–3.3 million while slightly increasing aggregate health benefits relative to the status quo. These findings demonstrate that the proposed approach can generate improved, data-driven treatment guidelines for the management of HIV/HCV co-infected patients. |
| Presented by: Raphaël Langevin, McGill University |
Employment Effects of a Statutory Minimum Wage: Evidence from a National Reform of the German Apprenticeship MarketAbstractTo enhance the attractiveness of vocational education and training and to secure an adequate supply of skilled labor, the German government introduced a statutory minimum apprenticeship wage. Since January 1, 2020, apprentices who start their training have been entitled to a minimum wage that increases annually. Using administrative register data on apprenticeship contracts, we estimate the causal effect of this legislation on apprentice employment. Exploiting regional and occupational variation in the share of apprenticeships paid at the minimum wage, we apply standard difference-indifferences, triple-difference, and synthetic difference-in-differences models. Our results indicate that the minimum apprenticeship wage increased the number of apprenticeship contracts while reducing the contract termination rate in low-wage occupations. We also find considerable heterogeneity across occupations, which may be best explained by differences in exposure to skilled labor shortages and changes in apprentices’ educational attainment. |
| Presented by: Michael Dörsam, Bundesinstitut für Berufsbildung (BIBB) |
The dark side of a windfall: unintended consequences of lottery winsAbstractEconomic windfalls, such as lottery wins, are typically viewed as beneficial. However, their unintended consequences are poorly understood. This study investigates the effect of large, random income shocks – using prizes from the Spanish Christmas Lottery – on prostitution and sex crimes. Results suggest that these financial windfalls lead to a significant increase in both offenses. The mechanisms appear to be a demand-driven rise, for prostitution, and an increase in violent sexual assaults by male perpetrators, for sex crimes. Finally, the lottery’s income shock appears to increase male sexually transmitted diseases. This research identifies a critical unintended cost of localized economic shocks, with important implications for policymakers understanding the secondary effects of income redistribution |
| Presented by: Riccardo Ciacci, Universidad Pontificia Comillas |
The Unintended Effects of Occupational Licensing: Evidence from ChildbirthAbstractOccupational licensing can expand access to services while simultaneously changing provider composition and service delivery in ways that affect quality. This paper studies the effects of licensing non-nurse midwives in the United States on service use and perinatal health outcomes. Leveraging cross-state variation in the timing of licensure, we estimate difference-in-differences models and find that licensure substantially increases out-of-hospital births, without changing pregnancy risk or reducing hospital-based obstetric supply. Despite this expansion in access, neonatal mortality improvements slow significantly following licensure. Maternal mortality exhibits differential pre-licensure trends. To interpret these findings, we develop a model of optimal license acquisition in which licensure may induce adverse selection of providers and weaken incentives for quality provision by reducing accountability in decentralized care settings. The findings indicate that licensing can alter the trajectory of intended outcomes, particularly in high-stakes environments like childbirth. |
| Presented by: Amairisa Kouki, University of Sheffield |
Labor Force Participation of Older Persons and Economic Growth: An Analytical Study on the Role of Health Expenditure and Structural Mechanisms AbstractPopulation ageing has become a global phenomenon and is widely expected to exert downward pressure on economic growth. Promoting labor force participation among older individuals—particularly when they remain healthy and productive—has been proposed as a key policy response. Using an unbalanced panel dataset of 151 countries over the period 2001–2023, this study examines the growth effects of a larger workforce aged 55–64 and 65 and above, together with the moderating role of government health expenditure and the channels through which these effects operate. The results indicate that a higher share of workers aged 55–64 is associated with lower economic growth, especially in countries that have reached aged-society status. This negative effect operates primarily through slower productivity growth and a decline in the intensive margin of labor supply. However, the adverse impact can be reversed when government health expenditure reaches at least 11 percent of GDP, suggesting a significant moderating role of health investment. In contrast, a larger share of workers aged 65 and above is associated with positive growth effects, driven mainly by improvements in labor productivity, while the moderating effect of government health expenditure is statistically insignificant. This finding may reflect self-selection; whereby healthier and more productive individuals remain in the labor force at older ages. Overall, the results underscore the importance of promoting both healthy and active ageing. Policy efforts should focus on expanding employment opportunities for older people, strengthening preventive and primary health care, and supporting complementary productivity-enhancing measures such as lifelong learning and skills upgrading. |
| Presented by: Manachaya Uruyos, Chulalongkorn University |
| Session 14: PRODUCTION AND MISALLOCATION June 23, 2026 10:15 to 12:00 Location: D-111 |
| Session Chair: Chun Pang (Alex) Chow, University of British Columbia |
Leveraging Uniformization and Sparsity for Estimation and Computation of Continuous-Time Dynamic Discrete Choice GamesAbstractContinuous-time empirical dynamic discrete choice games offer notable computational advantages over discrete-time models. This paper addresses remaining computational and econometric challenges to further improve both model solution and estimation. We establish convergence rates for value iteration and policy evaluation with fixed beliefs, and develop Newton-Kantorovich methods that exploit analytical Jacobians and sparse matrix structure. We apply uniformization both to derive a new representation of the value function that draws direct analogies to discrete-time models and to enable stable computation of the matrix exponential and its parameter derivatives for estimation with discrete-time snapshot data, a common but challenging data scenario. These methods provide a complete chain of analytical derivatives from the value function for a given equilibrium through the log-likelihood function, eliminating the need for numerical differentiation and improving finite-sample estimation accuracy and computational efficiency. Monte Carlo experiments demonstrate substantial gains in both statistical performance and computational efficiency, enabling researchers to estimate richer models of strategic interaction. While we focus on games, our methods extend to single-agent dynamic discrete choice and continuous-time Markov jump processes. |
| Presented by: Jason Blevins, The Ohio State University |
How large is the misallocation of resources in the Polish economy?AbstractThis paper quantifies the magnitude and evolution of resource misallocation in the Polish enterprise sector over 1993–2023 using a near‑census of non‑financial firms with 10+ employees. The baseline approach follows Hsieh, Klenow (2009) and shows that hypothetical output gains from eliminating within‑industry inefficiencies generate sizeable and rising gains in productivity and output, increasing from below 40% in the 1990s to roughly 70–75% in the early 2020s. Rising misallocation of resources is robust to changes in the identification scheme and to changes in the underlying model (accounting for wedges related to the use of materials or allowing for non-unit economies of scale and markups that vary across sectors). The rise in misallocation is driven primarily by services and is rather due to entry and exit, rather than within‑cohort dynamics. Firm‑level regressions show that high‑productivity firms and middle‑sized firms are disproportionately too small, that non‑exporters and more profitable firms exhibit larger wedges, and that subsidies are associated with smaller misallocation, while the same characteristics often have opposing effects on actual growth. It implies that market forces and policies do not systematically steer firms toward efficient scales. Rising misallocation can also be a factor explaining the deterioration of productivity growth in the Polish economy since the 2000s. |
| Presented by: Michał Gradzewicz, Narodowy Bank Polski & SGH Warsaw School of Economics |
Pricing and Informality: Evidence from Energy Theft in BrazilAbstractIn certain settings, goods can be consumed outside of formal markets (e.g.: theft, counterfeit, or illegal sharing of subscriptions). When the share of informality is large, firms’ pricing decisions can be substantially affected, as the extensive margin - customers migrating to informal consumption - makes demand more elastic. We study this question in the context of electricity theft in Brazil, where stolen energy can represent more than 50% of the total formal market. We use detailed micro data from a major electric utility to estimate a structural model where consumers choose if they want to be formal or informal and then, how much to consume. For identification, we leverage a natural experiment where prices increased permanently to a set of consumers. We use the model to simulate counterfactual scenarios where: (i) theft is not possible, and (ii) the firm uses different pricing strategies. We find that the presence of informality increases the elasticity of demand from 0.26 to 0.42, and reduces monopoly optimal prices by 21%. Eliminating theft altogether would allow the firm to reduce prices by 19% while keeping profits constant. We also find that price discrimination is an effective tool to reduce informality rates. |
| Presented by: Andre Trindade, Nova SBE |
Identification and Estimation of Production Function and Consumer Demand Function under Monopolistic Competition from Revenue DataAbstractWe establish nonparametric identification of production functions, total factor productivity (TFP), price markups, and firms’ output prices and quantities, as well as consumer demand, using firm-level revenue data, without observing output quantity, in a monopolistically competitive environment with a fully nonparametric demand system. This result overturns the widely held view—formalized by Bond, Hashemi, Kaplan, and Zoch (2021)— that output elasticities and markups are not nonparametrically identifiable from revenue data without quantity information. Under the additional restriction that demand satisfies the homothetic single-aggregator (HSA) structure of Matsuyama and Ushchev (2017), we further nonparametrically identify the representative consumer’s utility function from firm-level revenue data. This new identification result enables counterfactual welfare analysis without parametric assumptions on preferences. We propose a semiparametric estimator that is feasible for standard firm-level datasets under a Cobb–Douglas production specification. Monte Carlo simulations show that the estimator performs well, while treating revenue as output induces substantial bias. Applying the estimator to Chilean manufacturing data, we reject the CES specification in favour of HSA, and find that market power reduces welfare by approximately 3%–6% of industry revenue in the three largest manufacturing industries in 1996. |
| Presented by: Chun Pang (Alex) Chow, University of British Columbia |
| Session 15: REGRESSION DISCONTINUITY June 23, 2026 10:15 to 12:00 Location: B009 |
| Session Chair: Ruzica Savcic, University of Cyprus |
Inference in Regression Discontinuity Designs with Clustered DataAbstractClustered sampling is prevalent in empirical regression discontinuity (RD) designs, but it has not received much attention in the theoretical literature. In this paper, we introduce a general model-based framework for such settings and derive high-level conditions under which the standard local linear RD estimator is asymptotically normal. We verify that our high-level assumptions hold across a wide range of empirical designs, including settings of growing cluster sizes. We further show that clustered standard errors that are currently used in practice can be either inconsistent or overly conservative in finite samples. To address these issues, we propose a novel nearest-neighbor-type variance estimator and illustrate its properties in a diverse set of empirical applications. |
| Presented by: Tomasz Olma, Ludwig Maximilian University of Munich |
Sensitivity Analysis in Sharp Regression Discontinuity DesignsAbstractThis paper develops a framework for extrapolating causal effects in sharp regression discontinuity (RD) designs beyond the local neighborhood of the cutoff. We adopt a partial identification approach that yields bounds on a broad class of causal parameters, including average and distributional treatment effects. Identification relies on smoothness restrictions on the conditional expectation or distribution of potential outcomes with respect to the running variable, parameterized by sensitivity measures that capture the degree of smoothness and admit an interpretation in terms of statistical dependence. Within this framework, we derive sharp bounds on counterfactual means and distributions that can be estimated using standard nonparametric or flexible parametric methods. The approach is well suited for sensitivity analysis, providing a transparent assessment of the robustness of extrapolated RD estimates. Monte Carlo simulations demonstrate its finite-sample performance and practical relevance. |
| Presented by: Sungwon Lee, Sogang University |
Average derivative estimation under singular regressors -- A normalization approachAbstractThis paper provides the limit rates and asymptotic normality of an average derivative estimator (ADE) when the distribution of the regressors does not possess a density (is singular). Empirically relevant cases are those where some of the continuous regressors could have mass points, when there is a functional relation between the regressors, when there is a mix of discrete and continuous regressors, or the regressors have a fractal structure. In such situations the asymptotic properties of the ADE were not fully established; we demonstrate that conventional kernel-based ADEs may be inconsistent. We prove that a normalized ADE (NADE) is consistent and asymptotically normal. The application to the LaLonde data set, which exhibits mass points and is comprised of a mix of continuous, discretized, discrete and categorical variables, demonstrates the applicability and good performance of the normalized estimator. |
| Presented by: Marcia Schafgans, London School of Economics |
Dynamic Identity and Dynamic EthnosizingAbstractWhile immigrant integration is often viewed as a static outcome, identities are in fact dynamic and evolve over time. Using rich panel data from the German SOEP (2010–2019), we transform the Ethnosizer (Constant et al., 2009) from a cross-sectional measure into a dynamic tool that tracks how cultural orientations shift across periods. Methodologically, we combine dynamic panel models (System GMM) with a structural correlated random-effects logit approach, allowing us to estimate both state dependence and transition probabilities between integration states. Our results show that attachments to host and origin cultures are highly persistent once formed—yet they are not irreversible. Transitions do occur, sometimes resulting in "reverse integration," where identification with the host culture weakens rather than strengthens. These findings demonstrate that integration is not a final destination, but a path-dependent process shaped by persistence, contextual shocks, and individual experiences. We complement the empirical analysis with a theoretical model that explains the mechanisms behind identity revision and reverse integration. |
| Presented by: Ruzica Savcic, University of Cyprus |
| Session 16: RISK SPILLOVERS AND TRANSMISSION June 23, 2026 10:15 to 12:20 Location: D-115 |
| Session Chair: Yannis Tsafos, University of Glasgow |
Extreme comovements and downside/upside risk spillovers between oil and stock marketsAbstractThis paper investigates the dependence structure and extreme risk spillovers between oil and stock markets using a dependence-switching copula (DSC) model combined with a conditional Value-at-Risk (CoVaR) framework. The DSC model captures both positive and negative oil–stock dependence and allows for regime transitions, while the CoVaR framework identifies the transmission of extreme risks across the two markets. Using daily data from six major oil-exporting and oil-importing economies over the period 2000–2023, we document significant positive tail dependence between oil and stock markets in oil-exporting countries such as the United States and Canada, as well as in oil-importing/exporting regions such as European markets. In contrast, Japan—an oil-importing country—exhibits negative tail dependence, while Korea and China show no statistically significant oil–stock market dependence. We further find strong evidence of downside risk spillovers between stock and oil markets, with spillover effects operating in both directions. In contrast, upside risk spillovers are observed only from oil markets to stock markets, not in the reverse direction. Overall, downside risk spillovers dominate upside spillovers, and upside spillovers from oil to stock markets are stronger than those from stock to oil markets. Both risk levels and risk spillover effects intensify during the 2008 global financial crisis and the COVID-19 pandemic. These findings deepen our understanding of oil–stock market linkages and have important implications for risk management and policy formulation. |
| Presented by: Cathy Ning, Toronto Metropolitan University |
Switching the leverage switchAbstractThis paper introduces a new family of asymmetric stochastic volatility (SV) models that capture how both the sign and magnitude of past shocks influence future volatility. Under normality, we establish stationarity conditions, derive closed-form expressions for key moments (including variance and kurtosis), and obtain a leverage-propagation function that summarises shock transmission over time. A Monte Carlo study under Gaussian and heavy-tailed shocks shows that Bayesian MCMC provides accurate finite-sample estimates. Empirically, using daily returns on the DAX and S&P 500 indices, the Leverage Propagation SV (LPSV) model generally matches or improves upon standard asymmetric SV benchmarks in terms of in-sample fit and out-of-sample volatility forecasts, while offering a clearer description of time-varying leverage transmission. In an application to daily PM2.5 concentrations in Madrid, SV models yield broadly well-calibrated one-step-ahead exceedance probabilities, with LPSV specifications performing slightly better than a symmetric SV benchmark in mid-to-high risk bins. A counterfactual experiment further shows that extreme pollution surprises generate a persistent increase in volatility, implying high uncertainty for several subsequent days. |
| Presented by: Eva Romero, Universidad Rey Juan Carlos |
Time to hedge, not pledge on climate changeAbstractWe investigate the evolution of global financial markets’ response to climate change news over time. Using a data-driven approach to select time indicators, we model the common volatility of the worldwide oil and gas industry, and its climate change risk drivers. We identify the adoption of the Paris Agreement as a clear attention shock. Since then, climate change appears to have created considerable turbulence in the oil and gas sector. In contrast, green funds are increasingly perceived as less risky and safer investments, with important implications for the energy transition. Mitigating such non-traditional risks has thus become not only possible, but desirable. |
| Presented by: Susana Campos-Martins, Universidade Católica Portuguesa |
Coping with the Unexpected: A Forward-Looking Measure of Firm ResilienceAbstractThis paper develops a new measure of firm resilience, ReVaR, which is dynamic, forward-looking and does not rely on a specific crisis. ReVaR captures how much a firm’s conditional downside risk bounces back after an extreme loss. Using stock return data, ReVaR can be estimated at the firm-quarter level. Ex-ante ReVaR predicts post-crisis firm performance during the 2000 Internet Bubble, the 2008 GFC and the Covid-19 outbreak. Panel regressions over a long 33-year sample period reveal a strong link between resilience and innovation: firms with more R&D, more and higher-valued patents and more knowledge capital are significantly more resilient. |
| Presented by: Esther Eiling, University of Amsterdam |
Geopolitical Risk & Macroprudential PoliciesAbstractWe estimate the impact of geopolitical risk on macroprudential policy actions across a panel of 42 countries. Rising geopolitical risk leads to a statistically significant deactivation of macroprudential tightening, resulting in a less restrictive overall policy stance. A one-standard-deviation increase in GPR is associated with a reduction in tightening actions of 0.067, equivalent to approximately a 12.4% decrease relative to its standard deviation. The deactivation of macroprudential tightening is even more pronounced when geopolitical stress is preceded by a more restrictive monetary policy stance: a 50-basis-point increase in the policy rate more than doubles the baseline effect. We attribute this finding to an intertemporal policy trade-off: policy authorities prioritise short-term economic stability over medium-term systemic risk containment in response to geopolitical shocks. |
| Presented by: Yannis Tsafos, University of Glasgow |
| Session 17: TIME SERIES ESTIMATION June 23, 2026 10:15 to 12:20 Location: D-110 |
| Session Chair: Frederik Krabbe, Aarhus University |
Multivariate Local Whittle Estimation for Multivariate Functional Time SeriesAbstractIn this paper we establish consistency and asymptotic normality of the multivariate local Whittle estimator for multivariate functional curved time series. The proposed semi-parametric estimator is constructed by applying the local Whittle estimator method to periodograms formed from multivariate functional principal component scores, where each score is obtained as the inner product between the functional observation and an estimated leading multivariate eigenfunction. The methodology is examined using a Monte Carlo study based on multivariate functional long memory processes generated via a Davies-Harte spectral construction with dimension specific fractional parameters. The simulation results indicate that the proposed multivariate local Whittle estimator exhibits low bias and mean squared error in finite samples. In empirical application to high-frequency wind energy data, daily functional representations of wind speed, wind direction, and power production are constructed and analyzed using multivariate functional principal component analysis(MFPCA). The results show that the long memory behavior of the dominant latent dynamic is primarily driven by power production, as evidenced by the loading structure and the persistence properties of the leading principal component scores. |
| Presented by: Miao Yu, Leibniz University Hannover |
Local Inference and Stability Testing in Structural Time Series ModelsAbstractWe develop kernel-smoothed GMM statistics for testing parameter stability and conducting local-in-time inference in structural time-series models with potentially weak instruments. The integrated kernel statistic decomposes into a full-sample component and a stability component whose distribution does not depend on identification strength, enabling a sequential workflow: test for stability first, then proceed to full-sample or local inference. Both S-type and K-type versions are available; the latter exploits overidentifying information for additional power. When stability is rejected, pointwise confidence sets for the time-varying parameter are obtained by inverting the local kernel statistics and remain valid regardless of instrument strength. In Monte Carlo experiments calibrated to a New Keynesian Phillips Curve environment, the tests control size and the local confidence sets deliver reliable coverage. In an empirical application using shock-based instruments and SP-IV estimation, we find strong evidence of instability, a progressive flattening of the Phillips curve, and increasingly forward-looking inflation dynamics. |
| Presented by: Robin Braun, Federal Reserve Board of Governors |
Transport-based Estimation of Time-Varying Factor ModelsAbstractWe propose a novel estimator for static factor models that captures nonlinear features and time-varying loadings. The method combines entropic regularization with adversarial techniques to estimate latent structures by aligning observed data with structured representations in latent space. It reinterprets principal component analysis through the lens of information geometry and optimal transport, uncovering deeper distributional patterns and dependencies. Our simulations show improved forecasting accuracy in settings where conventional estimators are challenged by weak or nonlinear signals. An empirical application to high-dimensional exchange rate data confirms the estimator’s ability to handle complex dynamics and extract informative factors in environments where standard approaches underperform |
| Presented by: Emilio Zanetti Chini, University of Bergamo |
Barron-Loss Adaptive EstimationAbstractThe score-driven framework has gained substantial popularity in recent years, yet it relies on pre-specified scoring rules tied to assumed conditional densities, making it vulnerable to misspecification under outliers or structural breaks. We embed the flexible Barron loss within the quasi score-driven (QSD) framework, allowing the degree of robustness to be learned from the data. The resulting Barron-Loss Adaptive Estimation (BLADE) filter is shown to generate a strictly stationary, ergodic, and invertible sequence of time-varying parameters under mild regularity conditions. Within an extended quasi score-driven estimation framework, obtained by generalizing the required moment condition, the associated estimator is shown to be consistent and asymptotically normal. The Barron loss is strictly consistent for a family of functionals indexed by the shape parameter gamma, enabling smooth adaptation between classical and robust targets. We establish that the BLADE update belongs to the class of Proper and Robust Auto-regressive Derivative Adaptive (PRADA) models and are therefore expected divergence reducing. We additionally characterize conditions under which specifications with gamma values smaller than those associated with quadratic divergence achieve a larger reduction in quadratic divergence than linear updates, formalizing when robust updating improves the generalization error. Monte Carlo experiments under clean and contaminated estimation environments confirm these theoretical findings and demonstrate superior performance relative to GARCH and Beta-t GARCH models when outliers are present. |
| Presented by: Cees Diks, University of Amsterdam |
Asymptotic Properties of the Maximum Likelihood Estimator for Markov-switching Observation-driven ModelsAbstractA Markov-switching observation-driven model is a stochastic process $((S_t,Y_t))_{t \in \mathbb{Z}}$ where $(S_t)_{t \in \mathbb{Z}}$ is an unobserved Markov chain on a finite set and $(Y_t)_{t \in \mathbb{Z}}$ is an observed stochastic process such that the conditional distribution of $Y_t$ given $(Y_\tau)_{\tau \leq t-1}$ and $(S_\tau)_{\tau \leq t}$ depends on $(Y_\tau)_{\tau \leq t-1}$ and $S_t$. In this paper, we prove consistency and asymptotic normality of the maximum likelihood estimator for such model. As a special case, we also give conditions under which the maximum likelihood estimator for the widely applied Markov-switching generalised autoregressive conditional heteroscedasticity model introduced by Haas, Mittnik, and Paolella (2004b) is consistent and asymptotically normal. |
| Presented by: Frederik Krabbe, Aarhus University |
| Session 18: Lunch June 23, 2026 12:00 to 13:30 |
| Session 19: CLIMATE RISK MODELLING June 23, 2026 13:30 to 15:15 Location: D-113 |
| Session Chair: Anthoulla Phella, University of Glasgow |
Measuring Climate-Induced Systemic Volatility via GEOVOL and Bayesian Quantile and CaViaR ModelsAbstractThis paper develops an econometric framework to examine how climate-related extreme events influence systemic volatility in the European banking sector. We construct GEOVOL, a latent common volatility factor identified from cross-sectional dependence in squared standardized GARCH innovations of bank-level stock returns. To isolate climate-related volatility from general financial stress, GEOVOL is embedded in a linear state-space model and decomposed via Kalman smoothing using the CLIFS systemic stress index, producing an orthogonal component suitable for climate analysis. The impact of climate extremes is then assessed through Bayesian quantile regression, allowing covariate effects to vary across the conditional distribution of systemic volatility with full posterior inference. We further propose a stochastic CaViaR model in which the conditional upper quantile follows a latent log-linear state process with stochastic innovations, nesting the deterministic specification as a limiting case. Results show that climate extremes primarily affect the upper tail of systemic volatility. |
| Presented by: Gianmarco Vacca, Università Cattolica del Sacro Cuore |
The SPHAR-X with an application to climate riskAbstractClimate processes are intrinsically complex, characterized by joint spatio-temporal evolution on the sphere, multiscale dependence, and strong persistence. Conventional econometric analyses often collapse this complexity into global or regional averages, discarding critical spatial heterogeneity and conflating planetary dynamics with localized variation. We address this limitation by introducing the SPHAR-X model, a spherical functional autoregressive framework with exogenous input. By representing climate variables through spherical harmonics, our approach captures scale-specific dynamics via an autoregressive distributed lag (ARDL) structure. Applying this framework to ERA5 global temperature anomalies and CAMS atmospheric CO2 concentrations, we find a parsimonious and structurally consistent bridge between time-series econometrics and global climate modeling for the analysis of long-run temperature dynamics and its dependence on CO2 emission. |
| Presented by: Francesca Macchia, Luiss Guido Carli |
Taking the Highway or the Green Road? Conditional Temperature Forecasts Under Alternative SSP ScenariosAbstractIn this paper, using the Bayesian VAR framework suggested by Chan et al. (2025), we produce conditional temperature forecasts up until 2050, by exploiting both equality and inequality constraints on climate drivers like carbon dioxide or methane emissions. Engaging in a counterfactual scenario analysis by imposing a Shared Socioeconomic Pathways (SSPs) scenario of “business-as-usual”, with no mitigation and high emissions, we observe that conditional and unconditional forecasts would follow a similar path. Instead, if a high mitigation with low emissions scenario were to be followed, the conditional temperature paths would remain below the unconditional trajectory after 2040, i.e. temperatures increases can potentially slow down in a meaningful way, but the lags for changes in emissions to have an effect are quite substantial. The latter should be taken into account greatly when designing response policies to climate change. |
| Presented by: Anthoulla Phella, University of Glasgow |
| Session 20: CREDIT AND MONETARY SHOCKS June 23, 2026 13:30 to 15:15 Location: D-115 |
| Session Chair: Jakub Mućk, Narodowy Bank Polski |
Beyond Policy Rates: Macroeconomic Effects of Lender-of-Last-Resort ShocksAbstractThis paper shows that lender-of-last-resort (LOLR) shocks trigger significant and persistent macroeconomic effects. We introduce a novel identification strategy that leverages the informational content of the “haircut gap” - the difference between private market and central bank haircuts on eligible collateral - combined with narrative sign restrictions within a structural VAR framework. We show that favorable LOLR shocks improve credit market conditions, boost economic activity, raise inflation moderately, and reduce systemic financial risk, highlighting collateral policy as a powerful and distinct central banking tool. |
| Presented by: Federico Puglisi, Bank of Italy |
Identification of a Single Independent Shock in Structural VARs, with an Application to Economic UncertaintyAbstractWe establish the identification of a single shock in a structural vector autoregressive model under the assumption that this shock is independent of the remaining shocks in the system, without requiring mutual independence among the latter---in contrast to the standard assumptions in the independent component analysis literature. The shock of interest may be either non-Gaussian or Gaussian; in the latter case, identification requires that the other shocks be jointly non-Gaussian. We formally prove the global identification of the shock and the associated column of the impact multiplier matrix, and we discuss consistent parameter estimation by pseudo-maximum likelihood. A detailed Monte Carlo study illustrates the finite-sample properties of our identification and estimation procedures. Finally, we apply the method to estimate the dynamic effect of a contraction in economic activity on different types of economic uncertainty. We find that monetary policy uncertainty responds positively to output shocks and more sharply than do economic policy and fiscal policy uncertainties. |
| Presented by: Alessio Moneta, Sant'Anna School of Advanced Studies |
State-Dependent Tail Risk Amplification in Credit MarketsAbstractWe introduce a systematic tail-severity factor that measures the common intensity of extreme negative returns across firms, extracted from a large panel of daily stock returns using a Tail Index Dynamic Factor Model estimated via variational Bayes. Using state-dependent local projections over a 1986-2024 daily sample, we document that tail-severity shocks have near-zero effects on credit markets during calm periods but generate sharp nonlinear responses during financial stress. The dominant transmission channel operates through the second moment: high-yield spread volatility responds positively during stress and negatively during calm, with the differential significant under conservative uniform confidence bands at 47 out of 61 horizons. First-moment transmission is characterised not by amplification but by compression, cumulative spread repricing is weaker during stress than during calm, consistent with intermediary withdrawal, mean reversion from elevated levels, and policy interventions. Three complementary identification tests establish that the factor captures a dimension of systematic risk distinct from aggregate implied volatility: aggressive orthogonalisation with respect to VIX eliminates state dependence, VIX innovations alone cannot replicate it, while a light orthogonalisation removing only contemporaneous VIX innovations preserves most of the signal. Results are robust to alternative stress indicators, credit segments, and subsamples. |
| Presented by: Muguel Herculano, University of Glasgow |
Understanding the dynamics of export in the short run. The role of foreign and global shocks.AbstractThis paper introduces a novel Bayesian SVAR framework to identify the short-run drivers of export fluctuations across 18 EU economies, focusing on transmission through Global Value Chains (GVCs). By applying shock identification via sign restrictions within a hierarchical block exogeneity structure (domestic, foreign, and global), we offer four key insights. First, we show that while the Great Trade Collapse was primarily driven by global demand and uncertainty, the COVID-19 crisis involved a complex confluence of non-domestic demand and supply shocks. Second, foreign and global shocks are found to be the principal drivers of export synchronization among EU economies, whereas domestic shocks exhibit a significantly lower degree of co-movement. Third, tight integration within European production networks does not shield these economies from global structural shocks, which remain the primary drivers of export variance. Finally, we provide evidence that trade openness and participation in investment-specific GVCs heighten the sensitivity of domestic exports to global shocks, particularly through cost-push mechanisms in backward linkages, while shorter GVC forward linkages tend to reduce this exposure. |
| Presented by: Jakub Mućk, Narodowy Bank Polski |
| Session 21: EVALUATION OF CREDIT POLICIES June 23, 2026 13:30 to 15:15 Location: B129 |
| Session Chair: Shu Shen, University of California, Davis |
ASSESSING THE EFFECTS OF RECENT PROVISIONING RULES ON CONSUMER CREDIT ALLOCATION IN COLOMBIA AbstractColombia’s post-pandemic recovery in 2021–2022 was marked by rapid consumer credit growth, followed by deteriorating credit quality indicators amid tightening financial conditions. In January 2023, the Superintendence of Finance of Colombia (SFC) introduced higher provisioning requirements for long-term consumer loans to strengthen financial resilience. From the perspective of credit institutions (CIs), increased provisions imply higher expenses and potential profitability pressures, which could lead to adjustments in lending strategies. This study evaluates the effect of that regulatory policy on consumer credit dynamics and CI soundness. We find that this measure increased CIs’ coverage ratios, indicating that the policy achieved its intended goal, with no evidence that it significantly affected overall credit supply conditions for longer-maturity loans in terms of loan amounts, interest rates, and collateral requirements. However, we document some significant asymmetric effects across financial institutions. These findings contribute to a deeper understanding of the implications of prudential regulation for credit market behavior in emerging economies. |
| Presented by: Diego Cuesta, Central Bank of Colombia |
Credit Guarantees and Relationship Lending: Evidence from Natural DisastersAbstractEmergency, full-coverage credit guarantees are often used to sustain credit flows after adverse shocks. This paper examines how banks reallocate credit following natural disasters when such guarantees are available. We exploit staggered disaster declarations across Mexican municipalities and combine loan-level supervisory data with a difference-in-differences design. Following disasters, lending shifts toward contracts eligible for emergency guarantees, substantially increasing the share of committed amounts that are guaranteed. This reallocation is not accompanied by a broad expansion in firm-bank relationships. New relationships increase only modestly and temporarily within the eligible guaranteed segment, even though loans in this segment carry more favorable terms after disasters. Program entrants are less likely to obtain subsequent credit than other post-disaster entrants once they face standard market conditions. We further document weaker ex post performance when eligible guaranteed lending is extended through new relationships, consistent with banks using guarantee protection selectively for new borrowers with greater underlying risk. Overall, the evidence suggests that banks use full-coverage guarantees mainly to contain disaster-related risk exposure by reallocating lending toward protected contracts, rather than to generate a broad and sustained expansion in new relationships. |
| Presented by: Mariela Dal Borgo, Banco de México |
Sailing through troubled waters: Evidence from support discontinuities to firms in times of crisisAbstractWe exploit the assignment mechanism of the APOIAR Program, a targeted initiative aimed at supporting the firms most affected during the COVID-19 pandemic, to provide causal evidence on the impact of grants on firm survival and performance in times of crisis. Using sharp and fuzzy regression discontinuity designs and drawing on a combination of administrative datasets, we find that eligible firms experienced a short-term increase in profitability in 2021, with €1 of support increasing net income by €0.658. However, these effects did not persist into 2022, and we found no significant changes in turnover or cost reduction, indicating that the increase in profitability was mechanically due to the subsidy. Firms allocated part of the grant to rental payments and purchases of office supplies, including modest investments in digitalization. Our findings suggest that these funds were particularly important for ex-ante less productive, with less cash on hand, and more indebted firms. |
| Presented by: João Pereira dos Santos, ISEG Lisbon, IZA |
Dynamic Difference-in-DiscontinuitiesAbstractThe difference-in-discontinuities (diff-in-disc) design is a widely used empirical framework to address identification failures in the traditional regression discontinuity (RD) design due to a recurring confounding treatment that utilizes the same policy eligibility cutoff as the new policy of interest. This paper formalizes the repeated treatment nature of the diff-in-disc design within a general potential outcome framework. The new framework accounts for both treatment effect heterogeneity and dynamic treatment effects, including carryover effects and path-dependency in contemporaneous effects. Both standard sharp and fuzzy diff-in-disc setups are considered. We propose new identification and estimation strategies for different scenarios and study the small sample performance of proposed estimators using Monte Carlo simulations. The proposed method is applied to the seminal study of Grembi et al. (2016) on the impact of relaxing fiscal rules. The application illustrates how the proposed methods complement the existing approach. |
| Presented by: Shu Shen, University of California, Davis |
| Session 22: FIRM PRICING AND INFLATION June 23, 2026 13:30 to 15:15 Location: D-111 |
| Session Chair: Andreas Koundouros, Freie Universität Berlin |
Inflation Drivers in Firms' Words: LLM-Based Factor Analysis of Firm PricingAbstractThis paper introduces a framework for bridging microeconomic narratives and structural macro modeling in the context of firm pricing. Using Federal Reserve Beige Book data, we deploy an LLM to extract qualitative measures of price-change dynamics and firms' reported attributions of such changes to different factors at the micro level. These measures are then used to estimate a state-space representation of inflation dynamics derived from a menu-cost model featuring price stickiness and reporting frictions. Our approach enables a time-varying decomposition of inflation contributors, demonstrating the utility of micro-narrative data in identifying factors driving macroeconomic dynamics. |
| Presented by: Chenyu Hou, Simon Fraser University |
The Effects of Uncertainty on Firms’ Pricing Behavior and ActivityAbstractThis paper examines the causal effects of demand uncertainty on firms’ pricing behavior and economic activity using managers’ subjective expectations. Employing an instrumental variables approach that exploits differential industry exposure to exogenous uncertainty sources, we find that increased uncertainty causes firms to reduce prices through lower markups and decrease activity by reducing capacity utilization. To rationalize these findings, we develop a macroeconomic model where firms face capacity constraints and must commit to prices and capacity before demand uncertainty resolves. In response to increased uncertainty, the model’s putty-clay production technology generates a mechanism where capacity constraints truncate upside gains while firms bear the full downside losses, inducing firms to lower prices preemptively to minimize expected losses from excess capacity. Our calibrated model shows that a one standard deviation demand uncertainty shock reduces output by approximately 0.5 percent, with producer and consumer price inflation declining by roughly one-half and one-tenth of a percentage point, respectively. Absent markup reduction, the recessionary dynamics would be substantially more severe, as lower prices and markups dampen uncertainty’s negative effects. These findings demonstrate that idiosyncratic demand uncertainty generates disinflationary pressures through a distinct transmission mechanism—one complementing the inflationary effects of aggregate cost uncertainty emphasized in prior work—establishing demand uncertainty as an economically significant driver of business cycle fluctuations. |
| Presented by: Giuseppe Fiori, Board of Governors of the Federal Reserv |
Economic Narratives and Realities of Geopolitical RiskAbstractThis paper examines the relationship between economic narratives surrounding geopolitical events and their actual economic impacts. By employing a large-language model on a large set of newspaper articles, we identify whether the narrative of a geopolitical risk (GPR) shock is seen as acting on the supply or on the demand side. For the classification of the narrative, we equip the large-language model with a questionnaire to analyze the event's economic characterization. A Vector Autoregression model allows us to validate our GPR narrative indices by assessing whether the narratives align with economic realities. By identifying the nature of the geopolitical risk shocks in real time, central banks can more easily gauge the inherent risk to inflation and thus make better informed decisions. |
| Presented by: Yevheniia Bondarenko, Deutsche Budesbank |
Inflation Misperceptions of ConsumersAbstractInflation misperceptions of consumers complicate the conduct and communication of monetary policy. Drawing on the rich micro panel data of the ECB Consumer Expectations Survey, we empirically investigate the determinants of inflation misperceptions in a rational inattention model which we augment to account for distorting signals from salient prices. Our empirical results suggest that consumers misperceive inflation both because they are inattentive to the current inflationary environment and because they place excessive weight on food price inflation relative to headline inflation. Consumers were particularly attentive during the post-pandemic surge in inflation but became substantially less attentive as price growth began to decline. While the effect of food prices is particularly pronounced for female consumers, limited attention to inflation is associated with low financial literacy. |
| Presented by: Andreas Koundouros, Freie Universität Berlin |
| Session 23: HIGH-FREQUENCY FINANCE June 23, 2026 13:30 to 15:15 Location: D-105 |
| Session Chair: Laura Capera Romero, Vrije Universiteit Amsterdam |
A general class of model-free dense precision matrix estimatorsAbstractPrecision matrix estimation is a cornerstone concept in statistics, economics, and finance. Despite advances in recent years, estimation methods that are simultaneously (i) dense, (ii) consistent, and (iii) model-free are lacking. While each of these targets can be met separately, achieving them together is challenging.We address this gap by introducing a general class of estimators that unifies these features within a nonasymptotic framework, allowing for explicit characterization of the computational complexitysignal-to-noise ratio trade-off. Our analysis identifies three fundamental random quantitiescomplexity, signal magnitude, and method biasthat jointly determine estimation error. A particularly striking result is that ridgeless regression, a tuning-free special case within our class, exhibits the double descent phenomenon. This establishes the first formal precision matrix analogue to the well-known double descent behavior in linear regression. Our theoretical analysis is supported by a thorough empirical study of the S&P 500 index, where we observe a doubly ascending Sharpe ratio pattern, which complements the double descent phenomenon. |
| Presented by: Mehmet Caner, North Carolina State University |
On lead-lag estimation of non-synchronously observed point processesAbstractThis paper introduces a new theoretical framework for analyzing lead-lag relationships between point processes, with a special focus on applications to high-frequency financial data. In particular, we are interested in lead-lag relationships between two sequences of order arrival timestamps. The seminal work of Dobrev and Schaumburg proposed model-free measures of cross-market trading activity based on cross-counts of timestamps. While their method is known to yield reliable results, it faces limitations because its original formulation inherently relies on discrete-time observations, an issue we address in this study. Specifically, we formulate the problem of estimating lead-lag relationships in two point processes as that of estimating the shape of the cross-pair correlation function (CPCF) of a bivariate stationary point process, a quantity well-studied in the neuroscience and spatial statistics literature. Within this framework, the prevailing lead-lag time is defined as the location of the CPCF's sharpest peak. Under this interpretation, the peak location in Dobrev and Schaumburg's cross-market activity measure can be viewed as an estimator of the lead-lag time in the aforementioned sense. We further propose an alternative lead-lag time estimator based on kernel density estimation and show that it possesses desirable theoretical properties and delivers superior numerical performance. Empirical evidence from high-frequency financial data demonstrates the effectiveness of our proposed method. |
| Presented by: Takaaki Shiotani, University of Tokyo |
Informed Trading Volume and LiquidityAbstractTraditionally, total trading volume is viewed as a liquidity indicator, yet its effectiveness is often hampered by the presence of market microstructure frictions. This paper addresses this limitation by developing a stylized high-frequency information effect model to decompose trading volume into an informed signal component and an uninformed noise component. Using a non-parametric approach, the authors introduce the InfVol estimator, which identifies informed trading by selecting intraday intervals where price movements and volume innovations exceed specific data-driven thresholds. Monte Carlo simulations demonstrate that the InfVol estimator accurately recovers true informed volumes, particularly when utilizing a simple quantile-based calibration at high frequencies (e.g., 15 seconds). Empirical application to Microsoft (MSFT) stock reveals that scaling price impact by estimated informed volume produces a refined liquidity measure that co-moves more significantly with established bid-ask spread proxies than traditional measures based on total volume. This methodology provides policymakers and researchers with a robust tool to better understand price discovery and assess market vulnerabilities. |
| Presented by: Maria Ludovica Drudi, Bank of Italy |
Revisiting EWMA in High-Frequency-based Portfolio Optimization: A Comparative AssessmentAbstractThis paper compares the statistical and economic performance of state-of-the-art high-frequency (HF) based multivariate volatility models with a simpler, widely used alternative, the Exponentially Weighted Moving Average (EWMA) filter. Using over two decades of 100 U.S. stock returns (2002–2023), we assess model performance through a Global Minimum Variance portfolio optimization exercise, with and without short selling restrictions across multiple forecast horizons. We find that the EWMA model cannot consistently be outperformed by more complex HF-based volatility models at the daily and weekly forecast horizons, even delivering significant utility gains when including transaction costs due to a favorable balance between turnover and ex-post portfolio volatility. At the monthly horizon, the EWMA remains competitive towards most of its competitors. Our findings hold across alternative specifications, including different estimation window lengths, portfolio sizes and smoothing parameter values, emphasizing the continued relevance of parsimonious volatility specifications, such as the EWMA model, in realistic investment settings. |
| Presented by: Laura Capera Romero, Vrije Universiteit Amsterdam |
| Session 24: IDENTIFICATION IN MICROECONOMETRICS June 23, 2026 13:30 to 15:15 Location: B008 |
| Session Chair: Luca Barbaglia, European Commission |
Assignment at the Frontier: Identifying the Frontier Structural Function and Bounding Mean DeviationsAbstractThis paper analyzes a model in which an outcome equals a frontier function of inputs minus a nonnegative unobserved deviation. Inputs may be endogenous (statistically dependent on the deviation). If zero lies in the support of the deviation given inputs---an assumption we term assignment at the frontier---then the frontier is identified by the supremum of the outcome at those inputs, obviating the need for instrumental variables. We then consider estimation in the presence of random error that is mean-independent of inputs. Motivated by the assignment at the frontier assumption, we regularize estimation by requiring the fitted deviation's distribution to maintain a minimum probability mass in a neighborhood of zero. Finally, we derive a lower bound on the mean deviation, using only variance and skewness, that is robust to a scarcity of data near the frontier. We apply our methods to estimate a firm-level frontier production function and mean inefficiency. |
| Presented by: Dan Ben-Moshe, Ben-Gurion University of the Negev |
Learning about Corruption: A Statistical Framework for working with Audit ReportsAbstractQuantitative studies aiming to disentangle public corruption effects often emphasize the lack of objective information in this research area. The CGU Random Audits Anti-Corruption Program, based on extensive and unadvertised audits of transfers from the federal government to municipalities, emerged as a potential source to try to fill this gap. Reports generated by these audits describe corrupt and mismanagement practices in detail, but reading and coding them manually is laborious and requires specialized people to do it. We propose a statistical framework to guide the use of text data to construct objective indicators of corruption and use it in inferential models. It consists of two main steps. In the first one, we use machine learning methods for text classification to create an indicator of corruption based on irregularities from audit reports. In the second step, we use this indicator in a regression model, accounting for the measurement error carried from the first step. To validate this framework, we replicate an empirical strategy presented by Ferraz, Finan and Moreira (2012). We estimate effects of corruption in educational funds on primary school students' outcomes, between 2006 and 2015, expanding their original work. We achieved an expected accuracy of 92\% on the binary classification of irregularities, and our results endorse their findings: students in municipal schools perform significantly worse on standardized tests in municipalities where was found corruption in education. |
| Presented by: Eduardo Fonseca Mendes, Fundação Getulio Vargas |
The Combined Effect of Door-to-Door and Pay-as-You-Throw: Evidence from Emilia-RomagnaAbstractThis paper estimates the impact of door-to-door collection and of pay-as-you-throw tariffs on total waste and on the fraction of sorted waste generated in Emilia-Romagna, Italy, between 2016 and 2023. The analysis employs novel difference-indifferences techniques to assess the impacts of the two treatments, avoiding possible contamination caused by their interaction. Results are based on administrative monthly data from 279 municipalities, and show that DtD collection significantly increased the share of sorted waste and reduced total waste per capita. Pay-as-you-throw tariffs further enhanced sorting performance, with evidence of announcement effects preceding formal implementation, and evidence of preventing waste generation. Finally, a complementary analysis of waste quality data suggests that both policies improved sorting accuracy, particularly for paper and residual waste. |
| Presented by: Luca Barbaglia, European Commission |
| Session 25: INTERGENERATIONAL MOBILITY June 23, 2026 13:30 to 15:15 Location: B128 |
| Session Chair: Junwei Fan, University of Glasgow |
Intergenerational effect of initial labor market conditionsAbstractThis paper studies whether adverse labor market conditions at the time of labor market entry have intergenerational consequences for children’s human capital. Using rich Swedish administrative data that link college graduates between 1978 and 2004 to their children’s educational records, I examine how local unemployment rates at parental graduation affect children’s educational outcomes. The results show that adverse labor market entry conditions faced by fathers have persistent negative effects on children’s educational outcomes, while corresponding effects for mothers are generally small and statistically insignificant. A one–percentage-point increase in the unemployment rate at fathers’ graduation reduces sons’ Grade 9 GPA by approximately 0.09 standard deviations and lowers their probability of completing high school by about 2.5 percentage points. These effects are concentrated among sons and operate primarily through educational persistence rather than conditional high school academic performance or track choices. An analysis of potential mechanisms points to earnings losses as a central channel. Fathers who graduate during periods of high unemployment experience substantial and long-lasting midlife earnings reductions during their children’s formative years, while effects on employment, health, and family formation are limited. Taken together, the findings show that early-career labor market disadvantages can shape children’s educational opportunities across generations, even in a comparatively egalitarian welfare-state context. |
| Presented by: Qianyan Xu, Lund University |
On the Origins of Socioeconomic Inequalities: Evidence from Twin FamiliesAbstractInherited endowments and family background shape human capital, skills, and socioeconomic outcomes. A large literature using the Classic Twin Design (CTD) compares monozygotic (MZ) and dizygotic (DZ) twins to decompose outcome variance into additive genetic (A), shared environmental (C), and non-shared environmental (E) components. CTD studies typically find substantial heritability for education and economic outcomes. However, CTD inference relies on strong assumptions—random mating, no genetic dominance, equal environmental similarity across zygosity, and stability of genetic effects across contexts—violations of which can bias decompositions and complicate interpretation. This paper applies a Twin Family Design (TFD) to Danish administrative data linked to a national twin registry, connecting twins to their spouses and children. By incorporating relatives beyond the twin pair, the TFD provides additional moments that allow us to empirically assess several CTD assumptions and to connect estimates from twin studies to findings from adoption and molecular genetic research. We estimate the model via Minimum Distance, which maps parameters transparently to covariance moments and avoids relying on structural equation modeling and normality assumptions. We begin with educational attainment. A baseline CTD decomposition attributes 34% of the variance in education to additive genetic factors, 24% to shared environment, and 42% to non-shared environment. We then extend the design to assess key assumptions. First, adding spouses allows us to assess assortative mating. We estimate a spousal genetic correlation of 0.12, closely aligned with polygenic-score-based estimates reported in molecular genetic studies. Ignoring assortative mating biases shared-environment estimates upward and heritability estimates downward, because genetic similarity between spouses is misattributed to environmental transmission. Second, adding children enables a test for genetic dominance. We find no evidence of dominance for educational attainment, indicating that genetic variation is largely additive in this setting. Third, we examine gene–environment interaction by comparing cohorts differentially exposed to a major Danish school reform. Genetic variance remains stable, while shared and non-shared environmental variances are lower among reform-exposed cohorts. As a result, heritability rises mechanically when environmental variation is compressed, highlighting that heritability is context-dependent rather than an immutable trait parameter. Finally, we relax the CTD’s equal-environment assumption by allowing shared environmental similarity to differ between MZ and DZ twins. We find that MZ twins experience more similar environments than DZ twins. Once this restriction is relaxed, the implied heritability of education falls sharply—from 34% in the CTD to 15%—while the contribution of the shared environment increases to 38% and the non-shared environment accounts for the remaining 47%. This demonstrates that CTD heritability estimates are highly sensitive to assumptions about environmental similarity across twin types. We then apply the TFD to earnings, disposable income, and assets. Across these outcomes, additive genetic factors account for roughly 12–21% of cross-sectional variance, shared environmental factors for 22–29%, and non-shared factors for the remainder. Allowing environmental similarity to vary by zygosity consistently lowers heritability relative to CTD benchmarks. Beyond cross-sectional inequality, the TFD also permits a decomposition of intergenerational persistence. We estimate rank–rank IGEs of 0.24 for education, 0.18 for earnings, 0.10 for disposable income, and 0.25 for assets. Shared environmental factors account for 68% of the IGE in education, 66% in earnings, 41% in income, and 67% in assets, underscoring the central role of shared family environments in intergenerational transmission. Overall, the results show that relaxing the equal-environments assumption shifts variance decompositions toward larger environmental contributions and lower heritability, bringing twin-based estimates closer to those implied by adoption and family-based molecular genetic approaches. |
| Presented by: Konstantinos Tatsiramos, University of Luxembourg |
Mixing Backgrounds, Shaping Futures: Peer Effects of Parental Education in Military ServiceAbstractDespite universal access to grants and loans, prospective students from disadvantaged backgrounds in Denmark remain less likely to enroll in higher education. This disparity is particularly pronounced among males whose parents have not completed higher education. In this paper, we explore the role of close social networks—particularly new peer groups—in shaping educational decisions by altering perceived costs, social norms, and identity concerns. Using data on military service in Denmark, we examine whether exposure to new peers during this formative experience influences higher education enrollment among prospective first-generation college students. We hypothesize that peer quality and interactions during service can significantly impact educational choices. To isolate peer effects from selection effects, we leverage the Army’s assignment rule and within-unit variation in peer quality. Additionally, we investigate the interplay between cognitive ability, family background, and the decision to pursue higher education |
| Presented by: Stéphanie Lyk-Jensen, VIVE |
Social Comparison and Young Adult Mental Health with Endogenous Network FormationAbstractStandard models of reference-dependent well-being treat reference groups as exogenous, while evidence suggests individuals actively select into their social networks. This selection shapes both exposure to social comparisons and their mental health consequences. This raises fundamental questions: do peers’ economic outcomes affect mental health once network formation is endogenous? If so, through which mechanisms does social comparison operate? We combine individual-level survey data with detailed communication-based network measures to study the effects of peers’ income on mental health allowing for endogenous network formation. We find that exposure to higher-income peers is associated with worse mental health, particularly increased depressive symptoms, and that these effects are largely driven by endogenous network formation rather than peers’ income levels per se. Using a structural formation model, we show that relative status concerns shape network structure by inducing individuals to form ties with peers who are modestly wealthier or substantially poorer. Failing to account for these selection biases estimates of peer effects on mental health. |
| Presented by: Junwei Fan, University of Glasgow |
| Session 26: LONG MEMORY AND PERSISTENCE June 23, 2026 13:30 to 15:15 Location: D-106 |
| Session Chair: Prosper Dovonon, Concordia University |
Estimation and inference for the persistence of extremely high temperaturesAbstractWe propose a nonparametric framework for estimating the extremal index that captures the persistence of extreme observations. The framework provides unified and simple procedures for verifying the well-known local dependence condition D(d)(un), which characterizes the extremal index yet is often assessed through heuristic checks, and for selecting d (a key parameter for estimation) when the condition holds. Under a general ϕ-mixing condition, we establish the asymptotic normality of the proposed estimator and prove the consistency of both the tuning parameter selection and the verification procedure for the D(d)(un) condition. Simulation studies show improved performance relative to two commonly used methods in terms of empirical mean squared errors.We analyze summer apparent temperature data for nine European cities from 1940 to 2025. The results show strong evidence of persistence in extreme temperatures for all cities, with such extremes typically lasting at least two days. The probability of two-day extreme-temperature events is two to four times higher in the most recent three decades relative to 1940–1974. |
| Presented by: Chenhui Wang, Vrije Universiteit Amsterdam |
Semiparametric sore tests in Time Varying Long Memory seriesAbstractTraditionally, long memory models measure persistence using a single memory parameter that remains constant throughout the entire considered period. This can be an unrealistic restriction, especially if the series span a long period of time. Semiparametric score tests of the constancy of the memory parameter and other hypotheses of interest are proposed in a time-varying long memory context. The asymptotic distribution under the null hypothesis and local alternatives is obtained, and the consistency of the tests is demonstrated, with no need to restrict the short memory behaviour of the series. The performance in finite sample is analysed in a Monte Carlo exercise and its empirical implementation shown in an application to millenial scale temperature reconstructions. |
| Presented by: Josu Arteche, UPV-EHU |
Optimal Estimation in a Multicointegrated SystemAbstractOptimal estimation is explored in long run relations that are modeled within a semiparametric triangular multicointegrated system. In nonsingular cointegrated systems, where there is no multicointegration, optimal estimation is well under- stood (Phillips, 1991a). This paper establishes corresponding optimal results for singular systems, thereby accommodating a wide class of multicointegrated non- stationary time series with nonparametric transient dynamics. The optimality and sub-optimality of existing estimators are considered and new optimal estimators of both the cointegrating and multicointegrating coefficients are introduced that are based on spectral regression. |
| Presented by: Igor Kheifets, UNC Chalrotte |
Testing for contemporaneous exogeneity in linear modelsAbstractWe propose a test for contemporaneous exogenity of regressors in linear time-series models. The test takes the form of a specification test with an expanding number of moment conditions. The test does not require external instruments and is based only on information contained in the regressors. Under the null of exogeneity, the test is characterized by a standard normal limit, when the moment conditions are expanding at a particular rate relative to the sample size. We establish the consistency of the test under the alternative hypothesis when testing for exogeneity of both the full set and a subset of regressors. We illustrate in simulations the excellent empirical size and power properties of the test. |
| Presented by: Prosper Dovonon, Concordia University |
| Session 27: MATCHING AND SELECTION MODELS June 23, 2026 13:30 to 15:15 Location: D-112 |
| Session Chair: Joaquin Garcia-Cabo, Federal Reserve Board |
Multi-sided Matching with Transfers: The Economics of Labor and LoveAbstractThis paper develops an empirical framework to study the joint equilibrium of interrelated matching markets. In our model, agents from an arbitrary number of sides (e.g., men, women, firms, schools, locations) form coalitions and split a transferable surplus. The central feature of the model is cross-market surplus complementarities: the surplus created in one match can depend on partners and matches in other markets. Our methodological contribution is to extend semiparametric identification results from two-sided markets with transfers to a multi-sided environment, providing conditions for identification of the match surplus from equilibrium match patterns and observed characteristics. We apply the model to study the coevolution of the marriage and labor markets in the United Kingdom during the first two decades of the 20th century. Preliminary results suggest that employment and partnership status are mutually reinforcing in match surplus, and counterfactual simulations quantify the role that the growth in women’s education has played in the interplay between the two markets. |
| Presented by: Pauline Corblet, New York University Abu Dhabi |
Quantile Regression with Selection and EndogeneityAbstractWe develop an instrumental variables extension of the Arellano–Bonhomme quantile selection model to handle endogenous regressors. We establish identification under exclusion and rank conditions, and supply asymptotic theory with analytical standard errors. A smoothed generalised method of moments implementation stabilises computation and improves efficiency. Monte Carlo experiments demonstrate favourable finite-sample bias, coverage, and speed. An application to the UK Family Expenditure Survey shows that jointly accounting for selection and endogeneity substantially alters estimated returns to schooling, particularly for women at the high end of the wage distribution. |
| Presented by: Paul Bingley, VIVE |
Expanding the Labor Market Lens: Two New Eurozone Labor IndicatorsAbstractWe present a principal component analysis of euro area labor market conditions by combining information from 22 labor market indicators into two comprehensive series. These two novel indicators provide a systematic view of the current state and forward-looking direction of the euro-area labor market, respectively, and demonstrate superior forecasting performance compared to existing indicators. Crucially, we find significant implications for monetary policy design: a local projection analysis reveals that ECB monetary policy shocks have attenuated effects on both inflation and unemployment when the labor market forward-looking indicator is high. The dampened inflation response calls for tighter policy rate paths than a standard Taylor rule would prescribe. Finally, we show that focusing solely on the official unemployment rate may understate the actual labor market slack, and consequently, the trade-off between labor market health and inflationary dynamics. |
| Presented by: Joaquin Garcia-Cabo, Federal Reserve Board |
| Session 28: MONETARY POLICY COMMUNICATION June 23, 2026 13:30 to 15:15 Location: D-114 |
| Session Chair: Leandro M. Magnusson, University Western Australia |
Political Pressure on the Fed – Is This Time Different?AbstractWe examine how financial markets respond to political pressure on the Federal Reserve by US President Trump in his second term, in which his public calls for lower interest rates coincided with unprecedented efforts to reshape the Fed Board. Using a dataset of dozens of political pressure events in 2025, we use high-frequency asset price responses as multi-dimensional external instruments in a proxy VAR, identifying a political pressure shock on the Fed. We find that, unlike during Trump’s first term, Fed pressure does not materialize as a de facto monetary easing: although interest rates decline, equity prices tend to fall and measures of uncertainty rise. The price of gold increases particularly strongly. The identified shock can account for much of the response by financial markets to the “Liberation Day” tariff announcements in April 2025, and continues to weigh on the USD, indicating a sustained loss of confidence in the US. |
| Presented by: Ivan Frankovic, Deutsche Bundesbank |
When the Fed Reveals Its Hand: The SEP and Monetary Policy SurprisesAbstractRecent advances in high-frequency identification of monetary policy shocks reveal that measures are contaminated by information and news effects. We contribute to this literature by incorporating the intermittent release of central bank projections, i.e. the Summary of Economic Projections (SEP). We develop a theoretical framework showing that forecast releases amplify monetary policy surprises by providing additional information beyond what is conveyed through interest rate decisions alone and by anchoring expectations during non-release meetings. We confirm empirically that monetary policy surprises following SEP releases are typically 2 to 3 times larger than those without releases. To identify the information channel, we construct novel SEP surprise measures using a Bloomberg survey of market expectations about Federal Reserve projections. SEP surprises explain about 30 percent of the variation in monetary policy surprises during SEP meetings and account for essentially all of the differences between SEP and non-SEP meetings. Finally, to validate that SEP surprises contain economically meaningful information, we show that individual forecasters update their expectations of core PCE inflation in response to both common and their own idiosyncratic SEP surprises. |
| Presented by: Andrew Martinez, American University |
Money Talks: How Foreign and Domestic Monetary Policy Communications Move Financial MarketsAbstractWe provide novel insights into how foreign and domestic monetary policy communications, beyond rate announcements, affect the financial markets of open economies. We construct a high-frequency dataset that documents the impact of Federal Reserve (Fed) and Bank of Canada (BoC) rate announcements, speeches, press conferences and minutes releases to Canadian financial markets between 1997 and 2023. We find that non-rate announcements are a significant source of domestic monetary policy surprises and international spillovers. Across event types, Fed communications are particularly influential for long-term interest rates and stock futures while BoC communications matter more to short-term interest rates. Since BoC communications have little effect on U.S. interest rates, Canadian announcements have a greater impact on the CAD/USD exchange rate by inducing larger changes in the cross-country interest rate differential. |
| Presented by: Rodrigo Sekkel, Bank of Canada |
Monetary policy and stock returns beyond FOMC daysAbstractWe investigate the effect of monetary policy on stock market returns while allowing monetary policy to respond to stock prices by incorporating information beyond FOMC announcement days. Our inferential method exploits shifts in the volatilities of equity returns and interest rate shocks to disentangle both effects without imposing restrictive assumptions previously used in the literature. We find dramatic changes in the relationship between monetary policy and equity returns before, during, and after the effective lower bound (ELB) periods. An increase in stock returns triggers a rise in the policy rate only before the ELB period. Regarding the stock market response to monetary policy, the rise in the policy rate decreases stock returns only before the ELB period. However, since then, a rate hike increases stock market returns, which is explained by the effect of monetary policy on the expected equity premium. |
| Presented by: Leandro M. Magnusson, University Western Australia |
| Session 29: PANEL DATA METHODS 3 June 23, 2026 13:30 to 15:15 Location: D-110 |
| Session Chair: Chia-Min Wei, University of Wisconsin - Madison |
The Multiway Mundlak Estimator in Unbalanced PanelsAbstractThis paper demonstrates that, with multiway panel data, the common practice of augmenting linear regression models with dimension-wise averages fails to recover fixed-effects slope parameters when the panels are unbalanced. We characterize this discrepancy in terms of an omitted variable mispecification and call it the Mundlak Gap. We then propose a novel corrected Mundlak regression specification that restores the usual equivalence in full generality. We demonstrate that the least squares estimator of the resulting corrected Mundlak specification is algebraically identical to the corresponding multiway fixed effects estimator for any number of fixed effect dimensions and any unbalancedness pattern, collapsing to the standard mean-augmentation approach in balanced panels. Finally, a Gram-Schmidt algorithmic equivalence with our approach is also proposed, which decomposes the correction term into dimension-specific components, yielding interpretable auxiliary coefficients. |
| Presented by: Benjamin Harrison, EMORY UNIVERSITY |
High-Dimensional Panel Expectiles Regression: A Decomposition of the Gender Wage GapAbstractIn this paper we develop expectile panel data methods for high-dimensional fixed effects estimation in line with Guimar ̃aes and Portugal (2010), which will allow for a wide range of applications in fields such as Labour economics, Economics of Education, and Inequality. We also show how the Gelbach decomposition can be validly implemented in the context of panel expectile regressions. Using a unique Portuguese linked employer-employee dataset, we use our estimator to explore the determinants of the gender wage gap over the period 1995-2022. We find that: (i) the gender wage gap is larger in the upper tail; (ii) the difference is mainly explained both in the left and right tail by the individual unobserved heterogeneity; and (iii) assortative matching is less pronounced in the tailes. |
| Presented by: Pedro Raposo, Universidade Catolica Portuguesa |
Copula-Based Random Effects Models for General NetworksAbstractIn many contexts, individual interact with each other, creating interdependence in their final choices. Standard random effects models assume independence among individuals, which leads to inconsistent estimates of the probability of joint and conditional events. I propose a random effects estimator in which there is dependence among the unobserved heterogeneity of individuals connected in a general network. The dependence is modeled with a parametric copula, and allows to consistently estimate the probability of joint and conditional events with a statistically coherent model. The estimator is computed by maximizing a pairwise composite likelihood function, which requires bidimensional numerical integration, which works for a well-defined class of copulas. |
| Presented by: Santiago Pereda-Fernández, Universidad de Cantabria |
Panel Quantile Regression with Common ShocksAbstractThis paper develops an asymptotic and inferential theory for fixed-effects panel quantile regression (FEQR) that delivers inference robust to pervasive common shocks. Such shocks induce cross-sectional dependence that is central in many economic and financial panels but largely ignored in existing FEQR theory, which typically assumes cross-sectional independence and requires $T \gg N$. We show that the standard FEQR estimator remains asymptotically normal under the mild condition $(\log N)^2/T \to 0$, thereby accommodating empirically relevant regimes, including those with $T \ll N$. We further show that common shocks fundamentally alter the asymptotic covariance structure, rendering conventional covariance estimators inconsistent, and we propose a simple covariance estimator that remains consistent both in the presence and absence of common shocks. The proposed procedure therefore provides uniformly valid and robust inference without requiring prior knowledge of the dependence structure, substantially expanding the applicability of FEQR methods in realistic panel data settings. |
| Presented by: Chia-Min Wei, University of Wisconsin - Madison |
| Session 30: STRUCTURAL ESTIMATION June 23, 2026 13:30 to 15:15 Location: D-107 |
| Session Chair: Alejandro Sanchez Becerra, Emory University |
Bias Corrections and Diagnostics for Structural EstimationAbstractWe evaluate the bias of simulation estimators commonly used to estimate economic models in corporate finance. Our main theoretical results extend existing results on higher-order bias calculations by demonstrating how simulation methods contribute additional bias terms. We propose a bias reduction technique that is simple to compute. Monte Carlo experiments suggest our proposed procedure dramatically improves the accuracy of hypothesis tests. |
| Presented by: Samuel Engle, University of Exeter Business School |
What to Fix and What to Estimate in Structural Economic ModelsAbstractStructural models are often estimated in two steps: researchers fix (calibrate) some param- eters at prespecified values and then estimate the remaining ones. Despite how common this practice is, the split between fixed and estimated parameters is typically chosen by convention rather than by a formal econometric criterion. We propose a partition-selection method that makes this choice systematic. For each feasible partition, our procedure com- putes a sensitivity statistic measuring how much the paper’s object of interest—such as a counterfactual policy effect, welfare measure, impulse response, or treatment effect—changes when the fixed parameters are perturbed, and we select the partition that minimizes this sensitivity. We illustrate the approach in three examples spanning distinct literatures: two New Keynesian models (An and Schorfheide (2007); Nakamura and Steinsson (2018)) and a dynamic discrete choice model (Kalouptsidi et al. (2021), numerical example). We show that partition choice can substantially affect credibility: some partitions remain reliable under sizable calibration errors, while others generate large bias from small miscalibrations. The procedure is easy to implement, requires no re-estimation, and is compatible with widely used estimators including GMM, MLE, SMM, Indirect Inference (II), and Classical Minimum Distance (CMD). |
| Presented by: Joan Alegre, UC3M |
A Bias-Corrected Inference Procedure for the Random Coefficients ModelAbstractWe develop an inference procedure for the nonparametric fixed-grid estimator for random coefficient (RC) logit models. These estimators approximate the underlying distribution of RC’s via a discrete distribution with fixed support points and probability mass at every point. Although this approach is easy to implement and computationally efficient, its applied use has been limited by the lack of a valid inference procedure. Moreover, current literature faces several challenges for a valid inference procedure, such as the parameter-at-the-boundary-problem and bias. We aim to close this gap by providing an inference procedure for average functionals of the RC distribution—such as elasticities and willingness-to-pay measures—which are of central interest in applied research. In addition, we implement a bias-correction method and a valid way for bootstrapping standard errors (of the functionals) to overcome the abovementioned challenges. Although the logit model is central in this paper, the procedure can be generalized for any generalized linear RC model. |
| Presented by: Michael Pen, University of Groningen |
How to Weight in Moment Matching: An ML Approach with Applications to Earnings DynamicsAbstractFollowing the seminal paper by Altonji and Segal (1996), empirical studies commonly adopt equal or diagonal weighting in minimum distance estimation to mitigate finite-sample bias arising from sampling error in the weighting matrix. We propose a new weighting scheme that combines cross-fitting with regularized estimation of the weighting matrix, in the spirit of de-biased machine learning. We also propose a new formula for cross-fitted standard errors. We show that several canonical models in the earnings dynamics literature satisfy exact or approximate sparsity conditions that can be exploited by graphical lasso estimation of the weighting matrix. Within a many-moment asymptotic framework, we characterize the asymptotic distribution of the structural parameters. Extensive simulation studies demonstrate that our approach outperforms commonly used alternative weighting schemes. Finally, an empirical application using data from the Panel Study of Income Dynamics illustrates the practical gains of our method. |
| Presented by: Alejandro Sanchez Becerra, Emory University |
| Session 31: STRUCTURAL MACROECONOMETRICS June 23, 2026 13:30 to 15:15 Location: B009 |
| Session Chair: Håvard Hungnes, Statistics Norway |
Structural forecast analysisAbstractThis paper shows how the structural representation of a vector autoregressive model can support forecast analysis. We offer a unified framework that formalizes how the structural form of the model can help form a narrative for two key statistics in real-time VAR forecasting: the forecast errors relative to the outturn of the data, and the consequent revisions of the forecast. To illustrate the method developed, we conduct a stylised real-time exercise on the UK, focusing on the inflation surge that followed the pandemic. We show that the inflation forecast produced by a four-variable VAR model was revised upwards not only due to contractionary supply-side shocks, but also due to a mix of expansionary demand-side shocks, and a revision in the past shocks. |
| Presented by: Davide Brignone, Bank of England |
A Nonparametric Approach to Augmenting a Bayesian VAR with Nonlinear FactorsAbstractThis paper proposes a Vector Autoregression augmented with nonlinear factors that are modeled nonparametrically using regression trees. There are four main advantages of our model. First, modeling potential nonlinearities nonparametrically lessens the risk of misspecification. Second, the use of factor methods ensures that departures from linearity are modeled parsimoniously. In particular, they exhibit functional pooling where a small number of nonlinear factors are used to model common nonlinearities across variables. Third, Bayesian computation using MCMC is straightforward even in very high dimensional models, allowing for efficient, equation-by-equation estimation, thus avoiding computational bottlenecks that arise in popular alternatives such as the time varying parameter VAR. Fourth, existing methods for identifying structural economic shocks in linear factor models can be adapted for the nonlinear case in a straightforward fashion using our model. Exercises involving artificial and macroeconomic data illustrate the properties of our model and its usefulness for forecasting and structural economic analysis. |
| Presented by: Todd Clark, Johns Hopkins University |
A Structural Dynamic Factor Model of the Euro AreaAbstractThe euro area economy has a relatively short history marked by three deep crises with very different origins: external financial contagion in 2008, domestic sovereign debt crisis in 2012 and the Covid-19 pandemic in 2020. We design a Bayesian structural factor model that a) contains enough information to capture each of these heterogeneous events, b) performs well in pseudo real time forecast evaluations and c) interprets this history in terms of six basic economic shocks. |
| Presented by: Juan Figueres, European Central Bank |
Decomposing the Output Gap: Robust Multivariate Hodrick–Prescott FilteringAbstractThis paper proposes a robust multivariate Hodrick-Prescott (HP) filter for decomposing aggregate time series. While dynamic factor and unobserved components models typically rely on estimated factor loadings to extract a common cycle, our framework ensures transparency and tractability by explicitly incorporating observable, time-varying component shares into the aggregation. To empirically implement this on recent data, the filter robustly handles extreme shocks - such as the COVID-19 pandemic - by modeling them as additive outliers within the penalized least-squares objective. This adaptive weighting scheme minimizes cyclical variance, yielding a smoother and more reliable estimate of the underlying aggregate trend. Applying this framework to the expenditure components of U.S. real GDP, the decomposition reveals that private investment dominates cyclical fluctuations, while government expenditure displays a pronounced counter-cyclical pattern. By preventing structural shifts in the composition of the aggregate from being confounded with cyclical dynamics, the proposed framework offers a practical and transparent tool for real-time macroeconomic analysis. |
| Presented by: Håvard Hungnes, Statistics Norway |
| Session 32: WORK AND TECHNOLOGY 2 June 23, 2026 13:30 to 15:15 Location: E002 |
| Session Chair: Ekaterina Kazakova, HSE University |
Mapping the Dynamics of Management Styles— Evidence from German Survey DataAbstractWe study how firms adjust the bundles of management practices they adopt over time, using repeated survey data collected in Germany from 2012 to 2018. By employing unsupervised machine learning, we leverage high-dimensional data on human resource policies to describe clusters of management practices (management styles). Our results suggest that two management styles exist, one of which employs many and highly structured practices, while the other lacks these practices but retains training measures. We document sizeable differences in styles across German firms, which can (only) partially be explained by firm characteristics. Further, we show that management is highly persistent over time, in part because newly adopted practices are discontinued after a short time. We suggest miscalculations of cots-benefit trade-offs and non-fitting corporate culture as potential hindrances of adopting structured management. In light of previous findings that structured management increases firm performance, our findings have important policy implications since they show that firms which are managed in an unstructured way fail to catch up and will continue to underperform. |
| Presented by: Stefanie Wolter, Institute for Employment Reserach |
Measuring Task-Level Technological Exposure: A Language Model ApproachAbstractThis paper develops methods and tools for measuring the exposure of occupational tasks to technological substitution. Patent abstracts and task statements are matched and classified by a small, open-source language model. The model is fine-tuned and validated against a foundation AI, achieving accuracy improvements of roughly 5X over conventional ‘word embedding’ approaches. Model fine-tuning and a rules-based match threshold are critical for realizing these gains. The approach replicates stylized facts about IT exposure, but diverges sharply from survey-based measures of AI automation risk, which systematically understate exposure among high-wage occupations. A living dataset and Python package allow researchers to measure exposure across user-defined technology and task categories, with minimal time lag and at fine temporal resolution. |
| Presented by: Andre Mouton, Wake Forest University |
Heterogeneous Grouping Structures in Panel DataAbstractIn this paper we examine the existence of heterogeneity within a group, in panels with latent grouping structure. The assumption of within group homogeneity is prevalent in this literature, implying that the formation of groups alleviates cross-sectional heterogeneity, regardless of the prior knowledge of groups. While the latter hypothesis makes inference powerful, it can be often restrictive. We allow for models with richer heterogeneity that can be found both in the cross-section and within a group, without imposing the simple assumption that all groups must be heterogeneous. We further contribute to the method proposed by \cite{su2016identifying}, by showing that the model parameters can be consistently estimated and the groups, while unknown, can be identifiable in the presence of different types of heterogeneity. Within the same framework we consider the validity of assuming both cross-sectional and within group homogeneity, using testing procedures. Simulations demonstrate good finite-sample performance of the approach in both classification and estimation, while empirical applications across several datasets provide evidence of multiple clusters, as well as reject the hypothesis of within group homogeneity. |
| Presented by: Katerina Chrysikou, University of Leicester |
Learning task complexity: Role of experience in a crowdsourcing platformAbstractWorkers on online platforms often face uncertainty regarding the work time and effort to complete the tasks, which can lead to a suboptimal labor matching process, specially when employer reputation is not well established. To quantify the losses from such uncertainty, we develop an empirical model, which embeds endogenous labor supply and non-perfectly observable employer reputation and task complexity. Using unique, detailed micro-level data from a crowdsourcing platform, we show that workers initially form their expectations of task complexity by relying on observable task characteristics: several of these being anchors unrelated to the cost of effort, and others allowing workers to extract latent information from available data. Second, the worker experience alleviates uncertainty and exponentially increases the completion of attractive tasks. Third, workers value the potential employer reputation by avoiding newly posted tasks lacking detailed information (i.e., platform-computed metrics); participating in these tasks would require an extra compensation double the average microtask reward. Finally, we show that it takes 6.3 hours longer to find microworkers to complete tasks from ‘fresh’ projects, which is nearly 52.6% of the completion time of the average project. |
| Presented by: Ekaterina Kazakova, HSE University |
| Session 33: Coffee break June 23, 2026 15:15 to 15:45 |
| Session 34: ENERGY MARKETS June 23, 2026 15:45 to 17:50 Location: D-111 |
| Session Chair: Rubens Morita, University of Exeter |
Revisiting Fundamentals of the European Gas Market: The Role of Supply SubstitutionAbstractThe 2022 European energy crisis exposed the central role of supply substitution in natural gas markets, as disruptions to Russian pipeline flows were accompanied by a sharp surge in liquefied natural gas (LNG) imports. This paper revisits the fundamentals of the European natural gas market by explicitly distinguishing between pipeline gas and LNG supplies. Using a Bayesian Structural Vector Autoregression (SVAR) identified through sign and elasticity restrictions, the analysis jointly identifies the contemporaneous elasticities of natural gas prices, supply by source, inventories, and euro area industrial production. The results reveal pronounced heterogeneity across supply channels: pipeline gas is highly price inelastic, whereas LNG serves as the primary adjustment channel on the supply side. Although both supply shocks affect prices, pipeline disruptions generate sharp but short-lived inflationary effects, while LNG supply shocks exert more persistent influences on price dynamics. More broadly, the dynamics of gas prices exhibit a clear horizon-dependent structure, with short-run fluctuations driven by supply shocks and inventory behavior, and medium- to long-run movements increasingly shaped by aggregate demand forces. Counterfactual scenarios of the 2022 energy crisis quantify the stabilizing role of LNG availability and demand adjustment, highlighting how supply composition critically shapes energy price dynamics in Europe. |
| Presented by: Jorge Arenas, Universidad de Alicante |
Supply and Demand Effects in the Japanese Wholesale Electricity Market during the Global Energy CrisisAbstractWe investigate electricity price formation in Japan’s wholesale electricity market before and after the global energy crisis that began in late 2021, using the local projections method with external instruments. We find that the price response to demand shocks declines, whereas the response to supply shocks increases during the crisis period. These patterns are consistent with an upward shift in a kinked supply curve and a potential inward shift in a kinked demand curve. A complementary set-identified structural vector autoregression analysis further shows that the persistent elevation in electricity prices since September 2021 has been driven by supply-side factors, such as increases in LNG prices, while demand-side factors, including efficiency improvements, have partially offset this upward pressure. The results highlight the exposure of Japan’s electricity prices to international fuel market conditions and underscore the roles of both supply- and demand-side mechanisms in shaping price dynamics. |
| Presented by: Kazuhiko Ohashi, Hitotsubashi University |
Decomposing Carbon Prices to Analyze their Determinants: A Causal-Noncausal Application to the EU ETSAbstractThis paper studies the formation of EU allowance prices by decomposing them into contemporaneous and forward-looking components. We propose a mixed causal-noncausal autoregressive framework in which heavy-tailed innovations identify a causal component reflecting contemporaneous compliance demand and a noncausal component capturing anticipatory pricing dynamics. Using daily EUA futures data from 2013 to 2024, we show that the causal component cointegrates with short-run abatement fundamentals (most notably relative gas and coal prices) while the noncausal component exhibits faster error correction, consistent with anticipatory repositioning. Validation using weekly Commitment of Traders data indicates that forward positioning affects only the noncausal component. Specifically, hedging activity emerges as the primary driver of anticipatory dynamics, whereas speculative trading by non-compliance actors plays a significant role only during the 2018–2020 pricing regime shift. Our results help reconcile the weak role of fundamentals found in previous studies, which analyze aggregate prices without disentangling contemporaneous and anticipatory components. Diagnostic tests further show that conditional heteroskedasticity is concentrated in the noncausal component, suggesting that price volatility primarily originates from forward-looking dynamics. Overall, the findings highlight the central role of expectations in carbon price formation and have implications for the regulation and oversight of emissions trading markets. |
| Presented by: Eduardo Marques, Universite Paris Dauphine - PSL |
Fracking the Union? Oil Shocks and the Rise of U.S. Political PolarizationAbstractPolitical polarization has become an important source of macroeconomic uncertainty, yet its relationship with commodity markets remains poorly understood. This paper investigates the evolving link between U.S. partisan conflict and crude oil returns using a Markov-switching Granger causality approach that incorporates time-varying volatility and correlation, allowing for the endogenous detection of structural breaks in predictive relationships. Using monthly data from 1981 to 2025, we identify two distinct regimes. Prior to 2009, oil shocks systematically Granger-caused increases in partisan conflict, highlighting the destabilizing effects of import dependence and global supply disruptions. After the global financial crisis and the shale boom, this predictive channel largely disappeared, with oil prices no longer providing consistent information about domestic political polarization. Our findings challenge the conventional view that oil prices reliably signal U.S. political conflict, suggesting instead that their informational value depends on broader energy and political conditions. |
| Presented by: Rubens Morita, University of Exeter |
| Session 35: EQUITY MARKETS AND BELIEFS June 23, 2026 15:45 to 17:50 Location: B128 |
| Session Chair: Ming Zeng, University of Gothenburg |
Defining Current and Expected Financial Constraints using AIAbstractWe develop a novel annual measure of current and expected financial constraints for publicly listed US firms over 1993 to 2024. Applying artificial intelligence to 10-K filings' text enables more accurate and context-aware detection of financial constraints than traditional text classification techniques. Uniquely, we distinguish constraints affecting firms presently from those anticipated for the future. These constraint types are associated with distinct financial profiles and transition dynamics from which we distill three novel facts: (i) Expected constraints are seldom realized, instead, firms typically become unconstrained or postpone constraints further into the future. (ii) Firms frequently mitigate current constraints within a year, but persistence rises with severity. (iii) Firms prioritize resolving immediate over future constraints. Notably, timing-related heterogeneity impacts the practical application of the widely-used cash flow sensitivity of cash: while it identifies anticipated future financial constraints, it conflates distinct constraint types -- unconstrained and currently constrained -- and therefore fails to capture all financially constrained firms. A general implication of our work is that firms’ observable financial decisions remain informative for identifying financial constraints as liability-cashflow sensitivities distinguish unconstrained from currently binding constraints. |
| Presented by: Christoph Gortz, University of Augsburg |
Money Illusion in Earnings Growth ExpectationsAbstractWe show that inflation expectations influence beliefs about earnings growth. Using U.S. analyst forecasts matched to measures of inflation expectations, we find that higher expected inflation is followed by upward revisions in long-term growth forecasts, even though realized earnings growth does not rise commensurately. The resulting forecast errors are predictable and systematic, suggesting that analysts partially interpret nominal inflation signals as information about real earnings growth rather than fully translating nominal signals into real terms. Cross-sectional variation across analysts and firms is consistent with a belief-based distortion in growth expectations. Our findings are consistent with a growth-based form of money illusion in professional forecasts. |
| Presented by: Yeow Hwee Chua, Nanyang Technological University |
The Impact of Corporate ESG Performance on Stock Price EfficiencyAbstractThe Impact of Corporate ESG Performance on Stock Price Efficiency Brief Summary: This paper examines whether and how firms’ environmental, social, and governance (ESG) performance enhances stock pricing efficiency in China’s A-share market from 2015 to 2023. Using 17,528 firm-year observations and baseline specifications with firm and year fixed effects, we proxy efficiency by price delay (D1, D2) and stock price synchronicity (SYN). The core finding is that better ESG is associated with faster price discovery and more timely incorporation of information: a one–standard-deviation increase in ESG reduces D1 by about 8.3% of its mean and D2 by about 3.0%, while increasing SYN by roughly 0.08 units (≈0.07 of its standard deviation). These magnitudes are economically meaningful and robust across alternative controls, two-way clustered standard errors, and additional fixed-effect structures. Two mechanisms account for these gains. First, an information channel: ESG lowers information asymmetry—constructed from market microstructure and liquidity indicators—thereby accelerating the reflection of market-wide and firm-level news in prices. Mediation tests show that this channel explains a substantial share of the ESG–efficiency relation. Second, a reputation channel: higher ESG improves a composite reputation score, modestly reducing delay and raising synchronicity by facilitating investor trust and sustaining attention. A frictions test reveals that cross-agency ESG rating disagreement weakens these benefits: in high-disagreement settings, delay rises and synchronicity falls, consistent with noise impeding information aggregation. Heterogeneity analyses indicate stronger ESG effects among non-state-owned enterprises, where external monitoring and investor responsiveness are more salient. Results persist when substituting Bloomberg ESG for the main rating source, when using alternative efficiency proxies (including nonsynchronicity transforms and lag structures), and under varied winsorization and matching strategies. Overall, the evidence supports a view of ESG as information-enhancing corporate behavior: by reducing opacity and bolstering reputation, stronger ESG helps prices incorporate relevant signals more quickly—especially where ownership is private and rating signals are consistent. Key Contributions: 1.New evidence in China’s A-share market (2015–2023). Provides large-sample, firm-year panel evidence (A-shares) linking ESG to stock pricing efficiency. 2.Dual efficiency lenses used together. Combines price delay (D1/D2) and synchronicity (SYN) to capture both speed and cohesion of information incorporation, and interprets SYN within a noisy-market context. 3.Magnitude-focused results. Quantifies economically meaningful effects (e.g., ~8.3% and ~3.0% reductions in delay per 1-SD ESG; SYN +0.08), moving beyond sign/significance. 4.Mechanism decomposition. Identifies and tests two channels—information asymmetry reduction and reputation enhancement —using mediation analysis. 5.Role of rating disagreement. Shows cross-agency ESG disagreement weakens ESG’s efficiency benefits, highlighting the importance of signal consistency. 6.Ownership heterogeneity. Documents that non-state-owned enterprises. benefit more from ESG in pricing efficiency than SOEs, informing debates on governance and state influence. 7.Robustness across data and measures. Confirms results with alternative ESG source (Bloomberg) and alternative efficiency proxies, plus stringent FE/clustering choices. 8.Policy and practice implications. Suggests that improving ESG quality and harmonizing ratings can accelerate price discovery—useful for regulators, rating agencies, and issuers. Main Results: 1.ESG improves pricing efficiency. A one–standard-deviation increase in ESG lowers price delay by ~8.3% (D1) and ~3.0% (D2) and raises synchronicity by ~0.08, indicating faster incorporation of information into prices. 2.Information asymmetry: Higher ESG reduces asymmetry—measured via liquidity and microstructure proxies—which in turn speeds price discovery and increases synchronicity. 3.Reputation: ESG improves a composite reputation score; this boosts investor trust and attention, yielding modest additional reductions in delay and increases in synchronicity. 4.Rating disagreement weakens the benefits. When cross-agency ESG ratings diverge, noise rises: price delay increases and synchronicity falls, attenuating the positive impact of ESG. 5.Ownership matters. Non-state-owned enterprises experience markedly stronger efficiency gains from ESG improvements than State-Owned Enterprises, consistent with sharper market monitoring and investor responsiveness. 6.Robustness holds across data and models. The results persist with Bloomberg ESG, alternative efficiency measures, richer fixed-effect structures, and alternative standard error clustering—confirming economic and statistical significance. Conclusion: This paper examines whether firms’ ESG performance enhances stock pricing efficiency in China’s A-share market from 2015 to 2023. Using a large panel and firm- and year-fixed-effects specifications, we find that stronger ESG is associated with meaningfully faster price discovery: a one–standard-deviation increase in ESG lowers price delay by about 8.3% (D1) and 3.0% (D2) of their means and raises stock price synchronicity by roughly 0.08. These effects are economically large and statistically robust across alternative controls, clustering choices, fixed-effect structures, and data sources, including Bloomberg ESG and alternative efficiency constructs. Mechanism analyses indicate that the first channel is reduced information asymmetry: higher ESG improves transparency and market microstructure conditions, accelerating the incorporation of both market-wide and firm-specific news into prices. A second channel operates through firm reputation: better ESG elevates a composite reputation measure, modestly reinforcing investor trust and attention and yielding additional, though smaller, efficiency gains. We also document that cross-agency ESG rating disagreement materially weakens these benefits. Where disagreement is high, noise rises, delay increases, and synchronicity falls, underscoring the importance of rating consistency for the market to process ESG information effectively. Heterogeneity tests show that these effects are stronger for non-state-owned enterprises than for SOEs, consistent with sharper external monitoring, stronger investor responsiveness, and fewer non-market objectives in privately controlled firms. Together, the evidence supports a view of ESG as information-enhancing corporate behavior that helps prices align with fundamentals more quickly—especially in settings characterized by lower opacity, greater investor discipline, and clearer ESG signals. The findings carry several implications. For issuers, sustained improvements in material ESG practices can reduce information frictions and lower the cost of capital via faster price discovery. For regulators and rating agencies, harmonizing rating methodologies and improving disclosure quality can amplify ESG’s informational value by reducing disagreement-induced noise. For investors, integrating credible ESG signals may improve price efficiency and execution around news. |
| Presented by: Sinian Zheng, University College Dublin |
Analysts Are Good at Ranking StocksAbstractSell-side analysts’ forecasts of stock returns are biased and the consensus forecast is a poor cross-sectional predictor. In sharp contrast, the implicit ranking of stocks by each analyst is highly informative of subsequent returns. Long-short portfolios sorted on these rankings result in large and highly significant excess returns that cannot be explained by previous anomaly characteristics. The strong performance is most easily understood by noting the similarity between rankings and within-analyst demeaned forecasts. The latter are equivalent to removing each analyst’s fixed effect and thus controlling for unobservable analyst-specific biases. We document analogous results using analysts’ recommendations and earnings forecasts. |
| Presented by: Ming Zeng, University of Gothenburg |
VIX AMBIGUITYAbstractI propose a new measure of market-based ambiguity about volatility, VIX ambiguity, and show that it conveys information about variance premia. I find that my VIX ambiguity measure is correlated with a frequently studied vol-of-vol measure of ambiguity, VVIX, but embeds superior priced information and predicts the excess returns of S&P 500 straddles, synthetic S&P 500 variance swaps and VIX futures. VIX ambiguity has incremental information after controlling for known indicators inside and outside the VIX term structure and is economically significant. A managed S&P 500 short straddle portfolio that adjusts the exposure to variance risk based on signals from VIX ambiguity generates a significant annualized variance-risk-adjusted alpha of nearly 13%. Robustness checks alleviate concerns that my findings are driven by e.g. the crisis period, extreme observations or option market conditions. |
| Presented by: Thorsten Lehnert, University of Luxembourg |
| Session 36: GLOBAL SHOCKS AND INFLATION June 23, 2026 15:45 to 17:30 Location: D-106 |
| Session Chair: Lapo Bini, University of California, San Diego |
Global Drivers of Domestic InflationAbstractInflation data worldwide exhibit remarkable synchronization, not only in volatile food and energy components, but also in more stable core components, indicating the presence of global inflation drivers. We extract these global factors from a panel dataset spanning 20 countries and demonstrate that they explain a substantial portion of U.S. inflation fluctuations beyond what domestic factors alone can account for. Crucially, we establish that incorporating these global drivers is essential for understanding the underlying forces behind post-World War II inflation surges. Our analysis reveals that without accounting for these international components, the dominant role of supply shocks in the post-pandemic inflation surge would be significantly underestimated. |
| Presented by: Michele Modugno, Federal Reserve Board |
Networks, Real Rigidities and Monetary PolicyAbstractOver the past decades, the slope of the Phillips Curve has exhibited substantial variation across countries and sectors, with a marked flattening prior to the pandemic. This paper studies how the topology of domestic and global production networks shapes inflation dynamics. Using country--industry data on producer prices and output, we document systematic heterogeneity in Phillips-curve slopes: downstream sectors and industries with strong backward integration display flatter inflation responses, while upstream sectors and industries with strong forward linkages exhibit steeper dynamics. We rationalize these asymmetries with a network-based DSGE model where the NKPC slope reflects the interaction between real rigidities (strategic complementarities and endogenous markups), network propagation of nominal rigidities, and global slack transmission through imported intermediates. The results imply that the aggregate Phillips Curve slope is an endogenous object that shifts with production-network structure and global integration. |
| Presented by: Valerio Dionisi, Università di Milano-Bicocca |
Cracking commodity cycles: uncovering the hidden clock and decoding the driversAbstractCommodity prices exhibit pronounced cyclical behavior, but the duration and the drivers of booms and slumps remain underexplored. This study addresses these gaps in three key ways. First, it develops a commodity-specific cycle-dating algorithm that yields a comprehensive database of price cycles for 27 commodities spanning six decades. Second, the resulting data is used to investigate whether commodity price cycles are duration dependent—whether the probability of reversal increases as a phase ages. The empirical analysis reveals strong evidence of positive duration dependence, indicating the presence of a “hidden clock” mechanism in commodity price dynamics. Third, the empirical results indicate that macroeconomic and financial conditions exert a significant influence on cycle persistence. While supply-side dynamics emerge as the major drivers of phase transitions, demand-side factors also play a role: expanding trade volumes and a strengthening global financial cycle tend to extend booms and shorten slumps. |
| Presented by: Mirco Balatti, World Bank |
The Macroeconomic Effects of Global Supply Chain ShocksAbstractThis paper provides evidence that global supply chain shocks are key drivers of business cycle fluctuations, introducing a novel identification strategy based on a narrative analysis of price surcharges from the three largest container shipping companies. Negative shocks cause a persistent rise in consumer prices and a prolonged decline in economic activity. Sectoral impacts vary with exposure to global supply chains, measured by the share of inputs sourced from abroad. Spillovers extend to non-tradable sectors. These shocks accounted for up to 51% of the post-pandemic inflation. Without monetary or fiscal stimulus, recovery would have taken 18 months longer. |
| Presented by: Lapo Bini, University of California, San Diego |
| Session 37: INFLATION EXPECTATIONS June 23, 2026 15:45 to 17:30 Location: D-110 |
| Session Chair: Francesca Papagni, University of Bergamo |
When Bad News Breeds Bias: Cross-country Evidence on Inflation-as-a-Bad and Overreaction in Inflation ExpectationsAbstractUtilizing a very large household-level dataset of inflation expectations for twelve euro-area economies, we attempt to assess the formation and accuracy of inflation expectations following major disruptions of the macroeconomy which we identify during the period from 2004:1 to 2025:02. We find that the arrival of such adverse events tends to increase the degree of inaccuracy in inflation expectations. We also find that this happens because inflation expectations tend to go up in response to these adverse shocks relative to the 12-month ahead inflation realizations, which offers direct evidence of overestimation of inflation. This is consistent with overreaction of inflation expectations in response to inflationary news and with inflation-as-a-bad behavioral patterns in response to adverse non-inflationary shocks. We infer that such behavioral biases appear to have played an important role in the formation of inflation expectations in the euro-area following adverse shocks during the past two decades. |
| Presented by: Martin Geiger, Liechtenstein Institute |
Multivariate Economic Tail Risk and Scenario Analysis using the Survey of Professional ForecastersAbstractThis paper proposes forecasting joint tail risks for key macroeconomic indicators, GDP growth, inflation, and unemployment, using the US Survey of Professional Forecasters (SPF). By incorporating SPF consensus forecasts into the conditional mean of AR-GARCH-type models, the accuracy of univariate and multivariate predictive densities is significantly improved. Modeling a constant correlation matrix captures strong dependencies, particularly between GDP growth and unemployment. Using US data from 1990 to 2024, we show that the joint modeling framework enables scenario-based analysis in which predictive densities, conditioned on adverse developments in other variables, differ substantially from the baseline marginal distributions. The framework allows for a formal out-of-sample evaluation of joint predictive densities and a transparent assessment of conditional tail risks. |
| Presented by: Manuel Schick, Heidelberg University |
Beliefs and Uncertainty about InflationAbstractThis paper studies the macroeconomic effects of an inflation belief shock—an unexpected increase in household inflation expectations relative to a full-information rational forecast. We identify the shock using machine-learning methods applied to U.S. survey data and a large set of news, macroeconomic, global, and financial variables. In normal times, the shock raises inflation while reducing consumption and increasing unemployment. At the zero lower bound, it lowers real interest rates, boosts consumption, and reduces unemployment. Inflation uncertainty rises in both regimes, dampening the expansionary effects at the ZLB. A theoretical model replicates these findings, highlighting the need for monetary policy to stabilize both inflation beliefs and uncertainty about inflation. |
| Presented by: Giuseppe Pagano Giorgianni, Sapienza University of Rome |
Measuring Salience in macroeconomic time seriesAbstractThis paper proposes a statistical framework to identify and measure salience in time series data. We focus on two fundamental dimensions of salience---\emph{surprise} and \emph{prominence}---and translate them into well-defined stochastic features. Surprise arises from localized and unexpected deviations from regular behavior, while prominence is associated with persistence and long-lasting patterns generated by strong temporal dependence. We show that these dimensions affect the spectral representation of a stationary process by introducing specific features. A general spectral model based on a power transformation of the spectral density allows us to interpret the selection of the power parameter from a weighted Whittle likelihood perspective and to detect salience-driven attention and its implications on expectations. Simulation evidence illustrates how the proposed approach discriminates between non-salient dynamics and salient time series behavior. A real-data application on inflation and consumer survey expectations shows how to identify salience-induced biases in inflation forecasts. |
| Presented by: Francesca Papagni, University of Bergamo |
| Session 38: LOCAL PROJECTIONS AND SVARS 2 June 23, 2026 15:45 to 17:30 Location: B008 |
| Session Chair: Massimiliano Marcellino, Bocconi University |
More on VARs and Local Projections Equivalence: Unit Roots and Multiple InstrumentsAbstractWe show that the equivalence in population between impulse responses in Vector Autoregressions (VARs) and Local Projections (LPs) can be extended to (possibly cointegrated) unit roots with unrestricted lag structure. We also prove that structural estimation with multiple instruments for multiple endogenous regressors (LP-IV) is equivalent to a recursively block-identified Structural VAR, where the block of instruments is ordered first. Simulations and two applications illustrate our results. |
| Presented by: Alessio Volpicella, University of Pavia |
Shocks or Shifts? Identifying Macroeconomic Impulse Responses with InstrumentsAbstractThis paper documents an inconsistency in prevailing practice: instrument relevance in structural VARs is often assessed using an anchor equation’s reduced-form residual rather than the identified structural shock. As a result, the test is anchor-dependent and does not directly target the moment condition underlying identification. I propose a shock-based, anchor-independent weak instrument test based on the usual $F$-statistic—together with a bootstrap procedure that normalizes sign on the recovered shock—that yields invariant and stronger diagnostics, as well as typically tighter inference. I also show that when SVARs produce “structural’’ IRFs that are statistically indistinguishable from reduced-form dynamic multipliers, the instrument is likely weak, mis-specified, or simply superfluous. |
| Presented by: Daniele Colombo, London Business School |
High-Dimensional Nonparametric Local ProjectionsAbstractThis paper develops a flexible two-step high-dimensional nonparametric local projections estimator (HDNLP) by constructing Neyman-orthogonal pseudo-outcomes whose expectation conditional only on low-dimensional policy variables identifies the average structural function. In the first step, we construct pseudo-outcomes with high-dimensional nuisance parameters using flexible machine learning tools and cross-fitting. In the second step, we estimate the conditional expectation using lo-cal linear estimation, allowing for flexible inference on potentially nonlinear impulse responses. We establish asymptotic properties of the HDNLP estimator at standard nonparametric rates and demonstrate superior finite-sample performance relative to existing methods in simulations and in an application on the nonlinear effects of financial conditions on the economy. |
| Presented by: Massimiliano Marcellino, Bocconi University |
| Session 39: MACROECONOMIC FLUCTUATIONS June 23, 2026 15:45 to 17:30 Location: D-113 |
| Session Chair: Ricardo Nunes, University of Surrey |
The Response of Consumption to Real-Time Population: Evidence from Mobile Phone and Transaction DataAbstractThis paper provides direct measurements of the elasticity of individuals' economic transactions to real-time population (presence in a zone), by merging high-frequency mobile phone location data with card transaction records in the metropolitan area of Lyon, France’s second-largest urban area after Paris. Using a Poisson Pseudo-Maximum Likelihood estimator, we estimate the elasticity of transactions with respect to population presence at fine time intervals and a highly granular geographical division. The results provide several key findings: (1) there is substantial geographical and temporal heterogeneity in consumption responsiveness, with elasticities ranging from 1.08 during peak periods to 0.84 in peripheral areas; (2) we observe a systematic goods-services dichotomy, where food retail exhibits near-unity elasticity while experiential services show significantly lower responsiveness (0.64); and (3) there are persistent spatial frictions, with transaction flows declining by about 2% for a 1% increase in travel distance. By integrating real-time population presence and transaction data, we are able to quantify previously unobserved heterogeneities in the ways that real-time population presence translates into economic activity, and show that spatial constraints remain significant even at micro-geographic scales. |
| Presented by: John Galbraith, McGill University |
Anatomy of Inflation Expectations in Disinflation: The Role of Trust, Beliefs, and Understanding of Monetary PolicyAbstractThis paper investigates how households form and update their inflation expectations in a small open economy just after a severe inflationary shock. Drawing on a new monthly survey of households in the Czech Republic, we show that perceived and expected inflation decreased with the disinflation but did not re-anchor to the target. We show that the heterogeneity of inflation expectations is related to heterogeneity in beliefs and understanding of monetary policy and price dynamics, along with demographic factors. Moreover, the confidence in interest rate mechanisms may coexist with lower trust in the central bank, thereby pushing inflation expectations higher. Our evidence also suggests that even after decades of inflation targeting, households expect price-level reversion after a supply shock rather than low inflation consistent with the inflation target, pointing to a deeper tension between central bank communication strategies and public perceptions of what 'price stability' entails. |
| Presented by: Jaromir Baxa, UTIA AV ČR |
The Perils of Narrowing Fiscal SpaceAbstractWhen public debt is elevated, the fiscal cost of fighting inflation rises sharply, as interest rate hikes increase government interest expenditures. We formalize this mechanism in a nonlinear New Keynesian model with a state-dependent fiscal constraint on monetary policy. High debt may dampen the monetary response to inflation, generating an inflationary bias even though government debt remains fully fiscally backed. The interaction between high debt and inflationary cost-push shocks makes the fiscal limit more likely to bind, amplifying inflation. In demand-driven downturns, the fiscal constraint may become more restrictive than the zero lower bound, forcing the central bank to either print money to purchase excess debt or accept fiscal dominance. |
| Presented by: Sebastian Rast, De Nederlandsche Bank |
Interest Rate Surprises: A Tale of Two ShocksAbstractWe propose a new identification of monetary policy and information shocks. Our method exploits interest rate surprises on FOMC-announcement and macroeconomic data-release dates. Unlike approaches that impose the sign of equity comovement or rely on internal central bank forecasts, our design lets the equity response be determined by the data and is portable across settings where private forecasts are unavailable. We show that widely used FOMC-announcement rate surprises embed a sizable information component. Removing this component reveals that the contractionary effects of a positive pure monetary policy shock are stronger than the estimated effects based on FOMC-announcement surprises alone, whereas a positive information shock is expansionary with large and precisely measured effects. These conclusions are robust across VAR, near-OLS BVAR, and local-projection estimators, and across lag lengths and subsamples. Finally, we show that the information shock predicts forecast errors, consistent with market under-confidence in public signals. |
| Presented by: Ricardo Nunes, University of Surrey |
| Session 40: MICROECONOMETRIC METHODS 1 June 23, 2026 15:45 to 17:30 Location: B009 |
| Session Chair: Ana Armendariz Pacheco, Universität St Gallen |
A Debiased Difference-in-Differences Estimator for Robust Inference under Weak OverlapAbstractThis paper proposes a debiased difference-in-differences estimator that achieves valid, reliable inference under weak overlap by learning the Riesz representer adversarially. The proposed debiased estimator requires minimal assumptions to operate, avoids altoghether the problem of propensity-score inversion and is both double robust and locally robust. The framework also accommodates settings with multiple time periods and/or staggered adoption and shows strong performances under multiple data generating process with varying levels of misspecification. |
| Presented by: Piero Bertino, Universidad Carlos III de Madrid |
Two-step Parametric Estimation of Treatment Effects in the Presence of Misclassification and EndogeneityAbstractI provide a two-step parametric estimator correcting for misclassification and endogeneity biases in treatment effects. Approximate consistency is achieved applying modified maximum likelihood estimation (MMLE) to the reduced form binary discrete choice model to compute the misclassification probabilities and modified least squares (MLS) to the structural form which is augmented by a misclassification-corrected control function. The model incorporates unequal/equal misclassification probabilities for false negatives/positives and estimates misclassification rates without reliance on extraneous information or surrogate measurements. The two-step MLS (2SMLS) estimator is asymptotically normally distributed and outperforms both naive instrumental variables estimation (IV) that ignores misclassification, and OLS in terms of bias reduction. |
| Presented by: Georgios Chrysanthou, University of Dundee |
Instrumental Variable Rectified Linear Unit RegressionAbstractThis paper develops an instrumental variables method for endogeneity in rectified linear unit (ReLU) regression, where the ReLU transformation applies to the outcome variable. Building on the generalized method of moments (GMM) framework, our methodology requires only finite first moments and standard rank conditions for estimation and inference. The ReLU regression with instrumental variables yields explicit GMM solutions and avoids iterative optimization procedures. Through the Legendre-Fenchel transformation of ReLU regression outcomes, our framework enables the estimation of average quantile treatment effects between arbitrary quantiles, accommodating both continuous and non-continuous outcomes. We establish the asymptotic properties of the proposed estimators. In an empirical application examining the effect of 401(k) plans on household wealth accumulation, we find substantial heterogeneity in treatment effects across the wealth distribution, with larger impacts among households with greater financial capacity. |
| Presented by: Tatsushi Oka, Keio University |
Testing Effect Homogeneity and Confounding in High-Dimensional Experimental and Observational StudiesAbstractWe propose a framework for testing the homogeneity of conditional average treatment effects (CATEs) across multiple experimental and observational studies. Our approach leverages multiple randomized trials to assess whether treatment effects vary with unobserved heterogeneity that differs across trials: if CATEs are homogeneous, this indicates the absence of interactions between treatment and unobservables in the mean effect. Comparing CATEs between experimental and observational data further allows evaluation of potential confounding: if the estimands coincide, there is no unobserved confounding; if they differ, deviations may arise from unobserved confounding, effect heterogeneity, or both. We extend the framework to settings with alternative identification strategies, namely instrumental variable settings and panel data with parallel trends assumptions based on differences in differences, where effects are identified only locally for subpopulations such as compliers or treated units. In these contexts, testing homogeneity is useful for assessing whether local effects can be extrapolated to the total population. We suggest a test based on double machine learning that accommodates high-dimensional covariates in a data-driven way and investigate its finite-sample performance through a simulation study. Finally, we apply the test to the International Stroke Trial (IST), a large multi-country randomized controlled trial in patients with acute ischaemic stroke that evaluated whether early treatment with aspirin altered subsequent clinical outcomes. Our methodology provides a flexible tool for both validating identification assumptions and understanding the generalizability of estimated treatment effects. |
| Presented by: Ana Armendariz Pacheco, Universität St Gallen |
| Session 41: MIGRATION AND SOCIETY June 23, 2026 15:45 to 17:30 Location: D-114 |
| Session Chair: Marco Schmandt, Technical University of Berlin |
Minority Political Representation and Immigrant IntegrationAbstractThis paper investigates the impact of minority representation in English local councils on immigrants' naturalization rates between 2002 and 2020. Using a regression discontinuity design, we compare wards where minority candidates narrowly won or lost to majority candidates. Our findings indicate that the election of a minority councilor significantly increases the naturalization rate of immigrants by 2.3 percentage points, with Labour minority candidates having a stronger effect. Our findings support the role model effect over the critical mass theory, demonstrating stronger impacts in areas with lower minority population shares and less fractionalized communities. Single party majority at the council significantly enhances the ability to effect change, with Labour minority councilors prioritizing budget shifts toward education and social care. These results highlight the important role of political representation in promoting integration. |
| Presented by: Sarah Schneider-Strawczynski, University of Exeter |
Assortative Mating by Education and Race in the United States over Five DecadesAbstractWe develop a method suitable for detecting whether racial endogamy is on the rise and also whether the economic divide (i.e., the gap between individuals with different education levels and thereby with different abilities to generate income) is growing in a society. We identify these changes with the changing aggregate marital preferences over the partners’ race and education level through their effects on the share of interracial couples and the share of educationally homogamous couples. These shares are shaped not only by preferences, but also by the distributions of marriageable men and women by traits. The method proposed is designed to control for changes in the trait distributions from one generation to another. By applying the method, we find the economic divide in the US to display a U-curve pattern between 1960 and 2010 followed by its slightly negative trend between 2010 and 2015. The identified trend of racial homophily suggests that the American society has become increasingly permissive towards racial intermarriages since 1970. Finally, we refute the aggregate version of the status-caste exchange hypothesis based on the joint dynamics of the economic divide and the racial endogamy. |
| Presented by: Francisco Mendonca, International Demographic Inequality Lab |
Realization of Similar Norms in Different EnvironmentsAbstractThis paper explores the interaction between norms and labor markets. It examines childbearing patterns of a group of young female immigrants, leveraging their quasi-random allocation to the U.S. and Israel in 1989--1991. The question is whether immigrants similarly realize imported norms in both countries. Findings reveal similar childbearing profiles, consistent with origin-determined norms. However, the paths diverge in labor market engagement. Immigrants in Israel show uniform postnatal labor force participation. Conversely, immigrants in the U.S. demonstrate segregation: college-educated hard-working mothers versus low-educated less-working mothers. I provide a simple model that interprets these results as the interaction effect between norms and labor market opportunities in the host countries. |
| Presented by: Pavel Jelnov, The Max Stern Yezreel Valley College |
Random Placement but Real BiasAbstractA large literature in economics derives findings about the effect of group-level variables on individual outcomes from random placement settings. This paper shows that identification in such settings relies on additional assumptions. Based on a simple theoretical framework, we show that causal and well as reduced-form policy-interpretations are not generally valid. We provide a checklist to identify the importance of these assumptions in any given setting. We apply our framework to novel administrative data on randomly placed refugees in Germany and confirm the importance of the maintained assumptions empirically: depending on specification choices, we can even switch the signs of estimates of popular group-level variables, despite random placement. We discuss implications for the literature and alternative ``ideal experiments''. |
| Presented by: Marco Schmandt, Technical University of Berlin |
| Session 42: MIXED-FREQUENCY METHODS June 23, 2026 15:45 to 17:30 Location: D-105 |
| Session Chair: Stylianos Zlatanos, King's College London |
Mean Group and Pooled Mixed-Frequency Estimators of Responses of Low-Frequency Variables to High-Frequency ShocksAbstractThis paper proposes mean group and pooled estimators of impulse responses based on mixed-frequency auxiliary distributed lag (DL), autoregressive distributive lag (ARDL), or vector autoregressive distributed lag (VARDL) estimating equations. Our setup assumes that the data are generated by a high-frequency VAR process. While the shock of interest is directly observed at high frequency, the outcome variable is only observed as a temporally aggregated variable at a lower frequency. We derive the asymptotic distributions of the six proposed estimators. Monte Carlo experiments show that pooled estimators generally perform better than the corresponding mean group estimators for relevant sample sizes. An empirical illustration to the pass-through from daily wholesale gasoline price shocks to monthly consumer price inflation illustrates the usefulness of the proposed methods. |
| Presented by: Lutz Kilian, Federal Reserve Bank of Dallas |
False Discovery and Coverage Rate-Adjusted Inference for Impulse Response FunctionsAbstractInference for impulse response functions (IRFs) is typically reported using pointwise confidence bands, which can substantially overstate evidence when researchers examine a large family of responses (across multiple horizons, outcomes, shocks, and/or states) due to a lack of multiplicity-related error control. A growing literature proposes IRF-specific simultaneous inference procedures targeting family-wise error rate control --- such procedures often become overly conservative in practice and lose power quickly as the dimension of the response family grows, leading to uninformative empirical results. We advocate for targeting combined false discovery rate (FDR) and false coverage-statement rate (FCR) control as a natural middleground for IRF inference: this approach bounds the expected proportion of false rejections among the set of rejected null responses, and delivers appropriate post-selection confidence intervals for the selected set of impulse responses. We adapt an FDR/FCR-control procedure to VAR and LP IRF estimators that accommodates the strong cross-response dependence common in macroeconomic and financial applications through a resampling-based, stepwise algorithm. Monte Carlo experiments, together with applications using identified monetary policy, fiscal policy, and oil supply shocks, show that the proposed approach increases power and informativeness relative to simultaneous inference while limiting false inference induced by multiplicity. |
| Presented by: Giorgi Nikolaishvili, Wake Forest University |
Inference for Factor-MIDAS Regression ModelsAbstractFactor-MIDAS regression models are often used to forecast a target variable using common factors extracted from a large panel of predictors observed at higher frequencies. In the paper, we derive the asymptotic distribution of the factor-MIDAS regression estimator coefficients. We show that there exists an asymptotic bias because the factors are estimated. However, the fact that factors and their lags are aggregated in a MIDAS regression model implies that the asymptotic bias depends on both serial and cross-sectional dependence in the idiosyncratic errors of the factor model. Thus, bias correction is more complicated in this setting. Our second contribution is to propose a bias correction method based on a plug-in version of the analytical formula we derive. This bias correction can be used in conjunction with asymptotic normal critical values to produce asymptotically valid inference. Alternatively, we can use a bootstrap method, which is our third contribution. We show that correcting for bias is important in simulations and in an empirical application to forecasting quarterly U.S. real GDP growth rates using monthly factors. |
| Presented by: Yookyung Julia Koh, Tilburg University |
Markov-switching VAR models with high-dimensional transition probabilitiesAbstractThis paper develops a penalised maximum-likelihood estimator for Markov-switching vector autoregressive (VAR) models that allows transition probabilities to depend on a high-dimensional set of predictors. By applying ℓ1 (Lasso) regularisation to the transition probability coefficients, the approach performs variable selection and aids parameter estimation. We implement a modified Expectation-Maximisation (EM) algorithm that accommodates latent regimes while solving a convex, penalised multinomial logit problem for transition coefficients at each maximisation step. Monte Carlo experiments show that the estimator recovers sparse transition structures as sample size increases, tends to select parsimonious models in moderate samples, and remains effective for regime inference in high-dimensional settings. An empirical application to Growth-at-Risk with 129 macro-financial predictors on US data, demonstrates the estimator's ability to select distinct sets of predictors governing entry into and exit from growth vulnerability regimes. |
| Presented by: Stylianos Zlatanos, King's College London |
| Session 43: PROGRAM EVALUATION June 23, 2026 15:45 to 17:30 Location: B129 |
| Session Chair: Arkadiusz Szydlowski, University of Kent |
Who is Left Behind? The Take-up of Digital Programs: Evidence from BrazilAbstractWe study whether low-cost digital informational interventions can increase the take-up of fully online public programs. We embed a nationwide randomized controlled trial into the rollout of Desenrola Brasil, a digital debt relief initiative targeting indebted households in Brazil. Over 2.3 million eligible individuals were randomly assigned to receive email messages with information about how to access the program online, how to seek in-person assistance at the post office, or both. On average, the intervention had no detectable effects on platform access, debt renegotiation, or repayment. This null result is driven by extremely low exposure: only 3% of recipients opened the messages. Email opening was selective, with more vulnerable and less digitally connected individuals substantially less likely to engage. Conditional on exposure, however, providing enrollment guidance significantly increased platform access, renegotiation, and repayment. Heterogeneity analysis shows that effects are largest among socioeconomically vulnerable individuals who are not fully digitally excluded. The findings highlight both the potential and limitations of digital nudges: although they can relax frictions for reachable users, they fail to engage populations facing deeper digital barriers, constraining take-up at scale. |
| Presented by: Jessica Gagete-Miranda, Research Institute for the Evaluation of Public Policies (FBK-IRVAPP) |
Early Adopters and the Diffusion of Agroecological Knowledge in Peer GroupsAbstractPeer-to-peer knowledge diffusion is increasingly recognized as a key mechanism in various economic contexts and may play a crucial role in fostering the adoption of agroecological practices among farmers. However, the conditions for effective peer learning remain poorly understood, particularly regarding the role of the injection point -- the first individual to receive information -- within a peer group. This study examines whether the profile of the injection point affects the diffusion of agroecological knowledge. We conduct a randomized controlled trial (RCT) with roughly 850 voluntary French farmers, randomly assigned to peer groups on a digital communication platform. Over an 18-month period, only one farmer per treated group receives information on agroecology, serving as the injection point for diffusion. Treatment varies according to whether the injection point is an early adopter of agroecology or an ordinary peer. A benchmark group receives direct broadcasting of the information to all members. The experiment began in January 2025 and will run until June 2026, with data collection ongoing. |
| Presented by: Adelaide Fadhuile, Univ Grenoble Alpes |
Levelling up by levelling down: The economic and political costs of BrexitAbstractThe study uses a synthetic control method to estimate the local economic cost of Brexit. The vast majority of regions in the UK have lost as a result of Brexit. Since losses tend to be concentrated in relatively prosperous regions, Brexit has reduced regional inequalities (``levelling up``) while pushing down national output (``levelling down`` in the aggregate). Using both aggregate data from local elections and individual-level survey data from the British Election Study, we find that, politically, those areas that experienced Brexit-related output losses saw increases in support for right-wing populist parties, while the electoral fortunes of the Labour party declined. |
| Presented by: Eleonora Alabrese, University of Bath |
Immigration Policy and Remittances: Estimating Policy Effects with Interactive Fixed Effects and Compositional ChangesAbstractThis paper analyses the impact of a change in Australia’s immigration policy, introduced in the mid-1990s, on migrants’ remittance behaviour. More precisely, we compare the remittance behaviour of two cohorts who entered Australia before and after the policy change, which consists of stricter entry requirements. To address the challenge of evaluating the impact of policy change with two distinct migrant samples, we employ an interactive fixed effects model for treated potential outcomes. We propose a method to deal with compositional changes across the samples in this model under the assumption of no change in the mean of the fixed effects. The method amounts to applying the Callaway, Karami (2025) estimator on a sample matched across time periods based on observed covariates. The results show some weak evidence that those who entered under more stringent conditions - the second cohort - have a lower probability to remit but we do not find any significant differences in the level of remittances. |
| Presented by: Arkadiusz Szydlowski, University of Kent |
| Session 44: SKILLS AND THE LABOR MARKET June 23, 2026 15:45 to 17:30 Location: D-115 |
| Session Chair: Sudong Hua, Shanghai Institute for Mathematics and Interdisciplinary Sciences, Fudan University |
Skill Substitution, Expectations, and the Business CycleAbstractThis paper studies how labor market conditions around high school graduation affect postsecondary skill investments. Using administrative data on more than six million German graduates from 1995—2018, and exploiting deviations from secular state-specific trends, I document procyclical college enrollment. Cyclical increases in unemployment reduce enrollment at traditional universities and shift graduates toward vocational colleges and apprenticeships. These effects translate into educational attainment. Using large-scale survey data, I identify changes in expected returns to different degrees as the main mechanism. During recessions, graduates expect lower returns to an academic degree, while expected returns to a vocational degree are stable. |
| Presented by: Andreas Leibing, TU Dresden |
Declining Teen Employment: Causes and ConsequencesAbstractTeen employment among high school students in the United States has fallen by al- most 50 percent since the 1990s. In this paper, we study the causes and consequences of this decline by combining an empirical analysis with a quantitative general equi- librium framework. First, we provide causal evidence that attributes the majority of this decline to crowding out by adults, who were forced to compete with teenagers in low-paying service occupations due to successive recessions and job polarization. To determine the consequences of this decline, we first estimate the impact of employment hours during teenage years on lifecycle wages. Teenagers who worked more and did not attend college earn significantly higher wages and face lower unemployment later in life. We then develop a general equilibrium model in which teenagers allocate time be- tween human capital accumulation on the job or in school, and adults choose between “teen” and “non-teen” occupations. We estimate the model’s parameters via simulated method of moments targeting the causal estimates from the data. The model shows that teenagers crowded out from work suffer substantial income and welfare losses, largely due to lost on-the-job learning. As a result, some teenagers substitute toward college attendance, mitigating part of these effects. A decomposition exercise shows that minimum wage levels play a key role in transmitting the impact of increased adult competition into falling teen employment. Finally, optional vocational training policies are effective in mitigating adverse effects for teenagers who have been crowded out of the labor market. |
| Presented by: Alex Wurdinger, University of Minnesota |
Beyond technical skills. Labour market returns to skills in tertiary vocational education using an LLM approach.AbstractDespite the increasing interest in the labor market returns to tertiary vocational education (VET), the role of skills effectively acquired remains an understudied topic. In this paper, we use a unique dataset on the universe of students enrolled in tertiary vocational education in a Northern Italian region from 2015 to 2024 to investigate the returns to skills, leveraging cutting-edge LLM techniques. Analyzing program descriptions provided by each VET institution, we identify the skill content of each program by comparing its textual content with structured skill classifications, i.e. European Classification of Skills, Competences, Qualifications, and Occupations (ESCO). Our approach allows us to measure the skill content of each program and to assess the degree of alignment between training curricula and labor market demand. Using this information, we constructed a unique and novel dataset by merging the skills acquired by Tertiary VET students with the national evaluation of the programs (INDIRE) and the graduates’ employment histories derived from an administrative database (Comunicazioni Obbligatorie, Cob). The empirical analysis examines the impact of VET diplomas on labor market outcomes, including occupational outcomes and earnings. We estimate the economic returns to five different skill bundles. The findings will provide new evidence on how variation in acquired skills influences employment opportunities and career progression in the labor market. Overall, the study contributes to the literature in two ways: first, by offering the first evidence on skill-specific returns for Italy’s tertiary VET graduates; and second, by introducing an innovative LLM-based methodology for extracting and classifying skills from curricula. Our preliminary findings indicate that life skills – as well as STEM knowledge and information skills – are associated with the strongest wage and employment premia, while social and managerial skills are associated with negative returns. Moreover, program quality amplifies the returns to technical and informational skill bundles. These results shed light on the mechanisms through which VET programs generate labour-market value and inform the design of more effective, skill-oriented training policies. |
| Presented by: Federico Segato, Università Degli Studi di Bergamo |
Limits to Skill-based Countercyclical Adaptation in Business CyclesAbstractHouseholds respond to recessions by reallocating time and spending across market work, home production, and leisure. In a static framework, we show that countercyclical adjustments partly cushion shocks: skilled households substitute non-market time for market work and reduce expenditures, while unskilled households respond in the opposite direction. These adjustments are stronger across space than over time and more pronounced for home production than leisure, resulting in more volatile market-consumption and smoother non-market consumption. The mechanisms reflect the interaction of substitution and income effects and the joint role of time and money in consumption. Two limitations temper these mechanisms. Demographic heterogeneity leads some groups to exhibit procyclical adaptations that amplify shocks, and countercyclical adaptations take time to build, constrained by labor market rigidities, household coordination, and liquidity. |
| Presented by: Sudong Hua, Shanghai Institute for Mathematics and Interdisciplinary Sciences, Fudan University |
| Session 45: SPATIAL PANEL MODELS June 23, 2026 15:45 to 17:30 Location: D-107 |
| Session Chair: Jesse Chen, Columbia University |
Prediction in a Spatial Panel Data Model with Missing DataAbstractThis paper considers the generalized spatial panel data model proposed by Baltagi, Egger and Pfaffermayr (2013), which encompasses many of the spatial panel data models considered in the literature. We extend the model by allowing missing data, show the consistency of the maximum likelihood estimator (MLE) and derive the best linear unbiased predictor (BLUP) for the model. |
| Presented by: Long Liu, Florida Atlantic University |
Estimation of average partial effects in ultra-short panel data when individual-specific slopes are not identifiedAbstractWe study estimation of average partial effects in panel data models with heterogeneous and potentially correlated slope coefficients when the time dimension is ultra-short. The number of covariates may be large relative to the number of time periods, rendering conventional estimators infeasible or inconsistent. We approximate the systematic component of slope heterogeneity using a flexible high-dimensional representation and estimate the parameter of interest via sparse selection and Neyman-orthogonal moments with cross-fitting. We show consistency as the cross-sectional dimension grows with fixed time periods and illustrate good finite-sample performance in simulations. |
| Presented by: Christina Maschmann, Lund University |
Testing Spatial Interactions with Origin-side HeterogeneityAbstractThis paper proposes a Lagrange Multiplier (LM) test for spatial interactions in the presence of origin-side heterogeneous effects. We consider a spatial panel data model that allows interaction strengths to vary across units, capturing heterogeneity in how units transmit influence through a given spatial structure. Based on this specification, we test the null hypothesis of homogeneous spatial interactions. The proposed test follows an asymptotic chi-square distribution with m degrees of freedom, where m denotes the number of units allowed to exhibit heterogeneity, and its asymptotic power is derived under local alternatives. Monte Carlo simulations show that the proposed test exhibits good finite sample performance. Finally, we apply the proposed test to an empirical example of international knowledge spillovers through trade, highlighting the heterogeneous roles played by technological leaders across countries. |
| Presented by: Shi Ryoung Chang, Bank of Korea |
Fixed-T Dynamic Spatial Panel Model with Common ShocksAbstractWe consider a dynamic spatial panel model with interactive effects under large-N, fixed-T asymptotics. The specification combines contemporaneous spatial interaction, lagged dependence, observed regressors, and latent common shocks with heterogeneous loadings. The likelihood structure reflects two distinct Jacobian arguments: the cross-sectional spatial relation constitutes a high-dimensional simultaneous equations system and contributes the nontrivial spatial Jacobian, while the temporal dynamic transformation is represented by a TxT lower-triangular Jacobian with determinant equal to one. This decomposition yields a conditional Gaussian quasi-likelihood in which only the spatial transformation affects the Jacobian term. The resulting estimator does not directly estimate cross-sectional incidental parameters and thus avoids the Nickell-type bias that arises in conventional fixed-effects estimation for short dynamic panels. We derive consistency and asymptotic normality for N to infinity with fixed T, allow for conditional distribution misspecification through robust covariance formulas, and develop a tractable block coordinate algorithm exploiting the low-rank-plus-diagonal covariance structure of the transformed errors. Simulation results show satisfactory finite-sample performance. |
| Presented by: Jesse Chen, Columbia University |
| Session 46: TAIL RISK IN FINANCE June 23, 2026 15:45 to 17:50 Location: D-112 |
| Session Chair: Iason Kynigakis, University College Dublin |
Safe Distance to Systemic RiskAbstractIn this paper, we propose a new systemic risk indicator to measure the distance to the extreme losses of a financial system. Constructed from daily out-of-sample Value-at-Risk (VaR) exceptions across large U.S. financial institutions from 2000 to 2023, our indicator calculates the \hl{total} shortfall in market value during these \hl{VaR Co-E}xceptions. By applying extreme value theory (EVT) to the maximum weekly shortfalls using a half-year rolling window, we effectively model the tail risk of the financial system. Our empirical analysis demonstrates that this indicator captures accurately significant financial crises, such as the Great Financial Crisis of 2008, the sovereign debt crisis of 2010, and the COVID-19 pandemic in 2020. Through quantile regression, we show that increases in our indicator significantly predict negative shocks to industrial production growth rates. |
| Presented by: Renzhi LIU, University Paris Dauphine-PSL |
Clustering-Based Estimation of Score-Driven Models for ExtremesAbstractWe examine a clustering-based framework for improving tail-risk modeling in panels of time series by exploiting cross-sectional information. Our unsupervised machine learning method jointly estimates cluster memberships and cluster-specific parameters governing downside tail behavior via an Expectation–Maximization algorithm. As a leading example, we consider a score-driven model for time-variation in the parameters of the Generalized Pareto Distribution (GPD). We establish conditions for consistency of our estimators for the cluster assignments and the model parameters. Simulation results confirm the ability of our method to recover the correct cluster assignments and the true GPD dynamics while substantially reducing estimation error. In an empirical application to daily equity returns, our clustering-based framework consistently improves Value-at-Risk and Expected Shortfall forecasts over individual estimation. |
| Presented by: Onno Kleen, Erasmus University Rotterdam |
Tail-Risk Forecast CombinationsAbstractThis paper formulates an Early Warning System (EWS) for tail nancial risks based on real-time multi-period forecast combinations of Value-at-Risk (VaR) and Expected Short- falls (ES) of portfolio returns of non-nancial rms and banks. Forecast combinations include baseline (VaR,ES) forecasts conditional on a domestic risk factor, as well as stress (sVaR,sES) forecasts conditional on CoVaRs of the risk factor, thereby integrating stress testing into forecasting. Using monthly data of the G-7 economies for the period 1975:01-2023:04, the proposed EWS delivers significant out-of-sample tail financial risk forecasts and reliable vulnerability signals up to a 12-month forecasting horizon, with stress forecasts in the combination improving forecasting ability prior to periods of severe financial stress. |
| Presented by: Gianni De Nicolo', John Hopkins University |
Ambiguity in the Tails and the Cross-Section of Stock ReturnsAbstractWe develop an ambiguity-aware distributional framework for cross-sectional asset pricing in which investors rank assets using quantile–ambiguity scores that combine a central performance measure with penalties on adverse tail states. We prove that optimal portfolio weights are monotone in these scores, reducing portfolio choice to a transparent ranking problem. Empirically, we estimate conditional quantiles using a high-dimensional panel quantile-factor model and incorporate firm-level economic policy uncertainty extracted from corporate disclosures. Policy uncertainty acts as a bad-news characteristic: it disproportionately worsens downside quantiles relative to the median or upper tail. Portfolios sorted on quantile–ambiguity scores that include this policy uncertainty signal deliver higher Sharpe ratios, improved downside protection, and lower turnover than traditional strategies. These results show that firm-level economic uncertainty conveys meaningful information about tail risk, with important implications for ambiguity-aware portfolio management. |
| Presented by: Luiz Lima, The University of Tennessee |
Forecasting Asymmetric Betas Using Firm-Specific InformationAbstractWe demonstrate that machine learning methods provide a powerful framework for modelling conditional asymmetric risk. Using a large cross-section of US stocks and a comprehensive set of firm characteristics, we show that machine learning forecasts substantially outperform traditional benchmarks, delivering up to a 50% improvement in predictive accuracy across a wide range of asymmetric beta measures. Allowing for nonlinearities, tree-based ensembles and neural networks deliver the strongest predictive gains. Historical betas followed by trading, price and accounting variables, emerge as the most important drivers of future risk dynamics. Reconstructing the CAPM beta from semibeta forecasts based on random forests further indicates that a more granular decomposition of systematic risk yields a more accurate representation of market beta. |
| Presented by: Iason Kynigakis, University College Dublin |
| Session 47: IAAE Lecture — Using Subjective Beliefs Data for Demand and Production Estimation — Aureo de Paula (University College London) — Chair & Moderator: Silvia Goncalves (McGill University) June 23, 2026 17:35 to 18:35 Location: Grand Auditorium |
| Session 48: Welcome Reception June 23, 2026 18:35 to 20:00 |
| Session 49: IAAE Keynote — The Prestakes of Stock Market Investing — Francesco Bianchi (Johns Hopkins University) — Chair & Moderator: Francesco Ravazzolo (BI Norwegian Business School and University of Bolzano) June 24, 2026 8:45 to 9:45 Location: Grand Auditorium |
| Session 50: Coffee break June 24, 2026 9:45 to 10:15 |
| Session 51: CLIMATE AND THE MACROECONOMY 2 June 24, 2026 10:15 to 12:00 Location: B128 |
| Session Chair: Heloisa De Paula, Insper |
Building Climate Indices of Maximal Macroeconomic Relevance via the Assemblage VARAbstractWhat should a macroeconomically relevant climate index contain? The composition is inherently ambiguous, and aggregation choices affect structural inference. We introduce the Assemblage VAR, which jointly estimates the VAR and selects nonnegative aggregation weights to maximize system likelihood and coherence. We consider component-space and rank-space variants that reweight subcomponents or emphasize different parts of the distribution. Applied to disaggregated U.S. climate measures from the Actuaries Climate Index and underlying NOAA regional data, the resulting synthetic index generates economically meaningful contractionary responses of industrial production, unemployment, and housing, whereas fixed-weight indices yield weak and statistically insignificant responses. The component-space estimator emphasizes high-wind extremes; the rank-space estimator concentrates on tail realizations, consistent with threshold damages. The framework is portable beyond climate. Assembling synthetic inflation and industrial production measures sharpens monetary transmission and largely eliminates the price puzzle under recursive identification. |
| Presented by: Christophe Barrette, University Bocconi |
Global Temperature and the Global Financial CycleAbstractThis paper investigates how global temperature shocks affect the global financial cycle and US macro-financial conditions. Using a proxy-VAR that combines global and US data with exogenous temperature innovations identified from National Oceanic and Atmospheric Administration records, we show that unexpected increases in global temperature lead to a persistent contraction in world output and a synchronized tightening of the global financial cycle. US industrial production and inflation decline in parallel, indicating that climate variability can propagate through international financial linkages. The results identify global temperature shocks as a new, climate-driven source of global financial fluctuations. |
| Presented by: Paolo Gelain, Federal Reserve Bank of Cleveland |
Nonparametric Estimation and Testing of Heterogeneous Nonlinear Temperature–Yield RelationshipsAbstractThis paper proposes a mixed sieve–kernel estimator for the nonlinear crop yield response model of Schlenker and Roberts (2009). Annual crop yields are modeled as the integral of an unknown yield response function with respect to a within-season temperature density, which is estimated nonparametrically. We establish consistency and asymptotic normality of the proposed estimator in a panel data framework and develop a formal test for the equality of state-specific yield response functions. Monte Carlo simulations demonstrate good finitesample performance. An empirical application to U.S. county-level corn yields from 1950 to 2020 replicates the well-documented nonlinear temperature–yield relationship and strongly rejects yield homogeneity across states, providing clear evidence of substantial spatial heterogeneity in climate sensitivity. |
| Presented by: Chaoyi Chen, Central Bank of Hungary |
Beyond Commodity Prices: Exchange Rates and Amazon DeforestationAbstractThis paper investigates the aggregate and dynamic effects of commodity price fluctuations and exchange rate movements on deforestation in the Brazilian Amazon. We identify exogenous shocks to the prices of two key commodities, grains and cattle, as well as to the exchange rate, and use these shocks as external instruments in a Structural Vector Autoregression (SVAR) framework. We identify a new channel by which exchange rate depreciations lead to higher deforestation, operating through their impact on commodity prices in local currency terms, which underscores the importance of macroeconomic conditions, beyond global commodity markets alone, in shaping deforestation dynamics. |
| Presented by: Heloisa De Paula, Insper |
| Session 52: COINTEGRATION AND STRUCTURAL CHANGE June 24, 2026 10:15 to 12:00 Location: B009 |
| Session Chair: André Casalis, National Bank of Slovakia |
Piecewise Linear Solutions for Non-Stationary ModelsAbstractWe assess the accuracy and efficiency of time-varying linear solution methods for non-stationary rational expectations models. These methods construct a sequence of local linear approximations, each with coefficients that vary over time, based on a set of expansion points. Benchmarking against globally accurate non-linear solutions, we show, both theoretically and numerically, that their accuracy depends critically on the choice of expansion points and on agents' expectations about the future. Our results contribute to the literature on solving non-stationary stochastic models with rational expectations, spanning a wide range of sources of non-stationarity, including evolving structural parameters, changing policy regimes, and cases without a balanced growth path. |
| Presented by: Mariano Kulish, University of Sydney |
Insights from Cointegration: Gender and Age Disparities in the U.S. Labor MarketAbstractThis paper formalizes how to test important economic and statistical hypotheses for the labor market in a cointegrated framework. That formalization offers substantive gains in understanding the labor market because that market can involve multiple cointegrating relationships. Those relationships include trend-stationary gaps in labor force participation rates and in unemployment rates, added-worker and discouraged worker effects in relationships between labor force participation rates and unemployment rates, and possible stationarity of any of the variables. This paper then applies that framework to analyze the U.S. labor market, finding several such long-run relationships for data disaggregated by both age and gender, albeit with heterogeneity across age and gender. Forecasts from these models provide a benchmark from which to measure recovery from the Covid-19 pandemic. |
| Presented by: Neil Ericsson, George Washington University |
When did the Phillips Curve become flat? A time-varying estimate of structural parameters.AbstractWe provide a time-varying estimate of the parameters of the New Keynesian Phillips Curve (NKPC) by combining three recent contributions from the literature: (i) a non-parametric estimate of a vector auto-regressive model with time-varying parameters (ii) an identification of a demand shock based on the Excess Bond Premium (iii) a regression-in-impulse-response-functions approach to compute the coefficients of structural macroeconomic equations. Our methodology allows to track the evolution of the NKPC coefficients over time, with a precision which would not be achieved by resorting to rolling windows estimation or simple splits of the sample. We show that, for the US, the structural slope of the NKPC has actually decreased over time and that this decline took place relatively early, with the slope being virtually zero from 1990 onward. Furthermore, we observe a growing importance of expected inflation. Our analysis also allows to dismiss the explanation of a flatter NKPC resorting to a stronger and faster reaction over time of the Federal Reserve to demand shocks. We also show results for the Euro Area and by using a different identification strategy based on sign-restrictions. |
| Presented by: Edoardo Zanelli, Aarhus University |
Monetary Integration and Structural Change in a Small Open Economy: Evidence from SlovakiaAbstractThis paper studies how Slovakia's adoption of the euro in 2009 affected the structural macroeconomic relationships governing inflation, output, and monetary policy. We estimate a two-country structural VAR for Slovakia and the rest of the euro area, allowing for regime-specific contemporaneous interactions while keeping lag dynamics comparable across regimes. We find that euro adoption led to a pronounced flattening of the Slovak Phillips curve and a decline in the inflation sensitivity of aggregate demand, substantially reducing the inflation-output trade-off. Post-euro, Slovak supply and demand relationships become quantitatively similar to those of the euro area. These results suggest that monetary integration can induce genuine structural convergence in small open economies, reflecting changes in nominal anchoring and real economic integration beyond the mechanical loss of an independent monetary policy. |
| Presented by: André Casalis, National Bank of Slovakia |
| Session 53: FINANCIAL CONNECTEDNESS June 24, 2026 10:15 to 12:00 Location: D-115 |
| Session Chair: Dino Palazzo, |
Stablecoin ShocksAbstractWe develop novel measures of stablecoin shocks and use them to identify the causal effects of stablecoin adoption on U.S. financial markets. Combining a daily narrative dataset of stablecoin-specific news with changes in the combined market capitalization of USDC and USDT, we measure high-frequency movements in stablecoin market capitalization and implement heteroskedasticity-based identification within an event-study and SVAR-IV framework. Stablecoin demand shocks have triggered persistent declines in short-term Treasury yields, a depreciation of the U.S. dollar, and gradual spillovers into crypto and equity markets. We also document heterogeneous effects across firms: payment providers benefit from greater stablecoin adoption, whereas banks show no evidence of priced disintermediation risk. Our findings highlight stablecoin demand as a novel channel of asset-market transmission. |
| Presented by: Takaaki Sagawa, Northwestern University |
Understanding Financial Market Connectedness: Bubbles, Macroeconomic Shocks, and U.S. Asset SpilloversAbstractIn this paper, we empirically examine the drivers of financial market connectedness among asset classes in the U.S. Motivated by the systemic risk literature, we start looking at the impact of price bubbles—both positive and negative— on financial connectedness; thereafter, we analyze the role of macroeconomic shocks, particularly, a set of arguably exogenous shocks that provide causal interpretation to our results. Our results, using a quantile VAR method, indicate that financial market connectedness exhibits significant temporal variation and that price bubbles influence this dynamic. Negative bubbles in treasuries and corporate bonds and positive bubbles in high-yield corporates and the S&P 500 index affect financial connectedness. This evidence is consistent with the theoretical financial intermediation model in Acharya and Naqvi (2019). Positive and negative bubbles in the Nasdaq index also appear to impact connectedness. Macroeconomic shocks are also significant drivers of financial connectedness, as evidenced by the significant impact of monetary policy uncertainty, supply-driven oil shocks, defense-expense fiscal shocks, and investment-technology-related shocks. We find no effect of fiscal policy uncertainty or economic policy uncertainty on financial connectedness. |
| Presented by: Erwin Hansen, University of Chile |
Credit Composition, Equity Market Performance, and Crash RiskAbstractThis paper studies how household and business credit expansions differentially affect equity market performance. We show that the composition of private credit is central for asset pricing and financial stability. Expansions in household credit predict lower subsequent stock market returns and a higher probability of market crashes, whereas business credit expansions also forecast lower returns but are not associated with elevated crash risk. These patterns suggest that household credit booms are linked to mispricing and heightened downside tail risk, while business credit booms primarily reflect cyclical variation in investment opportunities and expected risk premia. We further document that these effects vary systematically across countries: the crash-risk implications of household credit are concentrated in advanced economies, while credit expansions in emerging markets primarily operate through business credit with weaker links to crash risk. Finally, we show that the relationship between credit expansions and crash risk depends on the macroprudential policy environment, with policy interventions weakening the adverse effects of household credit growth. |
| Presented by: Berrak Bahadir, Florida International University |
Market leverage and Financial SoundnessAbstractThe market value of leverage is generally lower than the corresponding book leverage, displays pronounced time variation, and spikes during periods of financial market turmoil. More importantly, and contrary to book leverage, market leverage fluctuations exhibit a declining time trend following the Global Financial Crisis. These results owe to an estimation methodology that jointly uses timely information from equity and credit markets to generate a high frequency estimate of firm-level market leverage. Using our methodology, we develop a novel measure of financial soundness for the U.S. non-financial corporate sector that we use to track economic crises in real time. |
| Presented by: Dino Palazzo, |
| Session 54: FORECASTING FINANCIAL AND MACRO RISKS June 24, 2026 10:15 to 12:20 Location: D-111 |
| Session Chair: Michael McCracken, federal reserve bank of saint louis |
Learning Exchange Rate Predictability with Machine LearningAbstractThis paper studies how exchange rate predictability can be learned from a high-dimensional information set. We find that predictability is horizon-dependent. At short horizons, forecast gains over the driftless random walk are statistically significant but modest, and predictive content is dominated by technical indicators. At long horizons, gains are stronger and more persistent, and macroeconomic fundamentals become central. Time-varying SHAP analysis makes these shifts transparent and shows that predictor importance is more unstable in the short run and smoother in the long run. Directional accuracy and certainty-equivalent returns confirm that the predictive gains are economically meaningful. These results are obtained using a machine learning forecast combination (MLFC) framework that learns by reallocating weights across heterogeneous models based on recent forecasting performance. This learning mechanism matters most when predictive content is fragmented and model performance is unstable, which is precisely the short-horizon environment emphasized in the literature. During the global financial crisis, the framework adapts by shifting weight toward models and predictors related to uncertainty and liquidity, helping to preserve out-of-sample performance under stress. The evidence suggests that exchange rates are difficult to forecast, but not uniformly unpredictable. Predictability emerges once horizon, informational structure, and adaptive aggregation are treated jointly. |
| Presented by: Biing-Shen Kuo, National Chengchi University |
The Accuracy Smoothness Dilemma in Prediction: a Novel Multivariate M-SSA Forecast ApproachAbstractForecasting presents a complex estimation challenge, as it involves balancing multiple, often conflicting, priorities and objectives. Conventional forecast optimization methods typically emphasize a single metric—such as minimizing the mean squared error (MSE)—which may neglect other crucial aspects of predictive performance. To address this limitation, the recently developed Smooth Sign Accuracy (SSA) framework extends the traditional MSE approach by simultaneously accounting for sign accuracy, MSE, and the frequency of sign changes in the predictor. This addresses a fundamental trade-off—the so-called accuracy-smoothness (AS) dilemma—in prediction. We extend this approach to the multivariate M-SSA, leveraging the original criterion to incorporate cross-sectional information across multiple time series. As a result, the M-SSA criterion enables the integration of various design objectives related to AS forecasting performance, effectively generalizing conventional MSE-based metrics. To demonstrate its practical applicability and versatility, we explore the application of the M-SSA in three primary domains: forecasting, real-time signal extraction (nowcasting), and smoothing. These case studies illustrate the framework’s capacity to adapt to different contexts while effectively managing inherent trade-offs in predictive modelling. |
| Presented by: Marc Wildi, Zurich university of applied sciences |
Pricing Macro-Panic: Forecasting the Tails of Bond Risk PremiaAbstractMoving beyond conventional mean-targeted forecasts, I predict the entire distribution of bond risk premia to capture extreme, crisis-driven tail movements. Augmenting standard predictors with an extracted upper-tail macroeconomic factor—representing a latent “macro-panic”—achieves full predictive dominance over baseline models across the quantile spectrum. Methodologically, a Quantile Random Forest (QRF) efficiently captures non-linear crisis thresholds using mean-selected predictors, offering a computationally tractable, high-performing alternative to Quantile LASSO. The resulting forecasts exhibit pronounced regime-dependent cyclicality, becoming heavily countercyclical during recessions. Crucially, this approach reveals a term-structure duality: short-term tail risks are driven by acute macro-liquidity panics (“dash for cash”), whereas long-term risks are anchored by structural safe-haven demand (“flight to safety”). Finally, I document a decay in macroeconomic predictability along the yield curve: while the QRF dominates at shorter maturities, naive empirical quantiles become highly competitive at the long end. |
| Presented by: Jiaxun Liu, Universitat Pompeu Fabra |
Macroeconomic cycles and bond return predictabilityAbstractWe study the link between the macroeconomy and expected bond returns by dissecting common macroeconomic cycles of different lengths. Two unobservable predictors generate sizeable economic value for investors: an inflation factor maximizing macroeconomic cycles of at least 8 years, and a term spread factor maximizing cycles of 1 to 3 years. The inflation factor captures the stance of monetary policy as return premia increase when the policy rule becomes “hawkish”. The term spread factor reflects investors’ perception of business-cycle risk. |
| Presented by: Stefano Soccorsi, Lancaster University Management School |
Growth-at-Risk is Investment-at-RiskAbstractWe investigate the role financial conditions play in the composition of U.S. growth-at-risk. We document that, by a wide margin, growth-at-risk is investment-at-risk. That is, if financial conditions indicate U.S. real GDP growth will be in the lower tail of its conditional distribution, we know that quantitatively, the main contributor is a decline in investment. Consumption contributes under extreme financial stress. Government spending and net exports do not play a role. We show that leverage plays a key role in determining both consumption- and investment-at-risk, which provides support to the financial accelerator mechanism proposed by Bernanke et al. (1999). |
| Presented by: Michael McCracken, federal reserve bank of saint louis |
| Session 55: INFLATION AND MACRO SHOCKS June 24, 2026 10:15 to 12:20 Location: D-112 |
| Session Chair: Rosi Chankova, Bank of England |
The Max-Share Prior for Identifying News Shocks: Evidence from Geopolitical Oil Supply DisruptionsAbstractWe propose the max-share prior as a method to easily impose max-share restrictions alongside zero, sign, and narrative restrictions for structural VAR shock identification. By combining max-share restrictions with other identifying assumptions, the maxshare prior retains the appeal of max-share restrictions for isolating news shocks with less risk of conflating multiple shocks for the sake of achieving a higher forecast error variance. By re-weighting posterior draws via importance sampling, the max-share prior is efficient to implement and scales easily. In an oil-market application, using a novel geopolitically filtered, capacity-based global disruptions series, the max-share prior improves identification of geopolitical supply news relative to current disruptions: news about future disruptions raises prices and the futures–spot spread on impact, with quantities adjusting at medium horizons—patterns consistent with precautionary demand and storage theory. |
| Presented by: Nida Cakir Melek, Federal Reserve Bank of Kansas City |
When Wording Changes What We Find: The Impact of Inflation Expectations on SpendingAbstractWe use a randomized experiment in the Bundesbank Online Panel-Households (n ≈ 3,900) to show that the estimated link between inflation expectations and household consumption flips sign depending on survey wording. This finding reconciles prior contradictory results and has direct implications for central bank survey design. Our experiment systematically varies elicitation framing of consumption question along three dimensions: the reference unit (individual vs. household), the time horizon (past one, 3, or 12 months), and the question type (attitudinal, planned, qualitative and quantitative recall-based). We find that the time horizon and question type significantly influence the estimated relationship between inflation expectations and durable consumption. While the average effect is weak, its sign and magnitude vary strongly with question design. Planned spending and attitudinal questions, such as whether it is a good time to buy, produce very similar negative associations, suggesting that respondents interpret the former as a proxy for future intentions. In contrast, quantitative recall-based questions on past spending yield a modestly positive link, especially for shorter horizons. These results highlight the critical role of survey design in shaping behavioral measurements, offering a novel explanation for mixed findings in the literature and guidance for both research and policy. |
| Presented by: Anna Mogilevskaja, University of Bonn, ECONtribute |
Long-Run Retail Price Dispersion: Dynamics, Market Structure, and Chain PricingAbstractUsing UPC-level supermarket prices for Uruguay spanning nearly sixteen years, this paper documents persistent long-run divergence in retail price dispersion within a single country. Contrary to the convergence implied by the Law of One Price, dispersion increases steadily over time. We show that this divergence is largely invisible in static analyses: once the relationship between price dispersion and market characteristics is allowed to evolve over time, the implied increase in dispersion more than doubles relative to baseline estimates. Retail price dispersion exhibits sharply different behavior within and across retail chains. Prices within chains remain tightly clustered and weakly responsive to local conditions, consistent with uniform or zone pricing. In contrast, economically meaningful divergence arises almost entirely across chains and is strongly associated with market structure, product differentiation, and competitive conditions. These findings highlight the central role of retail organization in shaping the long-run evolution of price dispersion and show that static approaches substantially understate persistent divergence in retail prices. |
| Presented by: Leandro Zipitria, FCS - UdelaR |
The Inflation Uncertainty Amplifier AbstractWe study how uncertainty shocks affect the macroeconomy across the inflation cycle using a nonlinear stochastic volatility-in-mean VAR. When inflation is high, uncertainty shocks raise inflation and depress real activity more sharply. A nonlinear New Keynesian model with second-moment shocks and trend inflation explains this via an "inflation-uncertainty amplifier": the interaction between high trend inflation and firms’ upward price bias magnifies the effect of uncertainty by increasing price dispersion. An aggressive policy response can replicate the allocation achieved under standard policy when trend inflation is low. |
| Presented by: Giovanni Pellegrino, University of Padova |
Look-Though or Look Closer? Energy Shocks and Underlying Inflation in the Euro AreaAbstractThis paper studies the transmission of exogenous energy supply shocks to underlying inflation in the euro area. Using externally identified energy supply shocks and local projection methods, it examines whether and under which conditions energy price movements propagate beyond headline inflation into core inflation. The results show that energy shocks generate sharp but short-lived increases in energy and headline inflation, while pass-through to core inflation is delayed, persistent, and concentrated in non-energy industrial goods. Crucially, this pass-through is highly state dependent. When wage growth, inflation expectations, or headline inflation are elevated, or when energy price increases are large, energy shocks transmit into core inflation faster and more strongly. In low-inflation environments, pass-through is negligible. These findings imply that linear estimates mask important nonlinearities and that the appropriateness of “looking-through” energy shocks depends critically on prevailing macroeconomic conditions. |
| Presented by: Rosi Chankova, Bank of England |
| Session 56: JOB LOSS AND WORKERS June 24, 2026 10:15 to 12:00 Location: B129 |
| Session Chair: Matteo Pazzona, Brunel University London |
Firm Shocks, Workers' Earnings and the Extensive MarginAbstractWe study how idiosyncratic firm shocks transmit to workers' earnings through both the intensive and extensive margins of employment. Using a matched employer--employee census data for Chile between 2007 and 2019, we estimate the pass-through of firm productivity and sales shocks to the wages of stayers and the displacement risk and earnings losses of leavers. For continuing workers, earnings respond modestly to firm-level shocks, revealing partial wage insurance within firms. However, adverse shocks substantially raise displacement probabilities, and displaced workers suffer persistent earnings losses of 15-20 percent. Combining both margins, we show that the overall sensitivity of expected earnings to firm shocks nearly doubles relative to estimates based on stayers alone. These effects exhibit substantial heterogeneity. Young and short-tenure workers face higher displacement probability but minimal earnings losses upon displacement, while older and long-tenure workers are less likely to be displaced but experience larger and more persistent losses if displaced. Turning to stayers, lower-skilled and lower-ranked workers within firms receive more wage insurance, while higher-skilled and higher-ranked workers receive less wage insurance, except for top earners. Hence, firms appear to insure continuing employees only partially while transferring substantial risk through separations. We conclude that a comprehensive assessment of how firms transmit risk to workers must integrate both wage and employment adjustments. |
| Presented by: Ana Sofia Leon, U. Diego Portales |
Who Suffers Most from Job Loss?AbstractIn this paper, we examine the distributional effects of job displacement resulting from firm closures in Portugal, with a focus on underlying mechanisms. Within a staggered treatment setting, our empirical strategy builds on a matched difference-in-differences event study, extending the estimator of Callaway and Sant’Anna (2021) to incorporate cohort-specific matched control groups. We find large and persistent earnings losses, concentrated among high-wage workers, primarily driven by declines in hourly wages rather than employment. Losses vary across the business cycle and are mitigated for low-type workers during COVID-19. These results highlight the importance of heterogeneity and underlying wage mechanisms in shaping post-displacement outcomes. |
| Presented by: Julian Tiedtke, Scuola Superiore Sant'Anna Pisa |
Firm-level responses to sharp changes in the access to fixed-term employment: Evidence from a large labor market reform in SpainAbstractHiring workers with fixed-term contracts (FTC) provides employers the flexibility to downsize at a reduced cost but induces turnover and potential loss of firm-specific human capital. A labor reform in 2021 in Spain aimed at curbing the large share of workers covered by FTC by limiting the possibility of hiring a worker using a fixed-term contract and expanding the use of seasonal open-ended contracts. We leverage on the heterogeneity of the intensity of use of fixed-term contracts across firms prior to the reform, the universe of employer-employee links and a quasi-Census of firm Balance Sheets between 2017 and 2023 to examine the response of hiring and firing practices and the evolution of firm-level productivity. We document six findings. Firstly, compared to moderate users of FTCs, firms using fixed-term contracts intensively reduced their use of FTCs after the 2021 reform by 17pp. Secondly, those intensive users of FT contracts exhibited employment trends similar to those of moderate users between 2017 and 2021, but reduced the total level of employment by 1pp, an impact driven by firm closures. Thirdly, intensive users of FTCs replaced fixed-term contracts by regular open-ended contracts, but not by seasonal open-ended contracts. Fourth, among intensive users of fixed-term contracts, and relative to moderate users, value added per worker increased 0.8pp after the reform. Fifth, counterfactuals based on simulations suggest that differential firm exit among intensive users of FTCs closed the productivity gap between intensive and moderate users by 0.7pp. Sixth, flows between workers in firms using fixed-term contracts intensively and other firms were moderate both before and after the reform. |
| Presented by: Cristina Barcelo, Banco de España |
Sorting into Job Loss: Heterogeneous Crime and Employment Effects of Mass LayoffsAbstractWe show that who gets displaced in mass-layoff episodes is far from random, and that this selection generates sizable heterogeneity in both post-layoff employment and criminal charges. We link Chilean administrative worker-firm records to universe-level public defender data, build quarterly panels around displacement events for private-sector men, and compare matched displaced and non-displaced workers using event-study and difference-in-differences designs with rich fixed effects. Displacement drastically reduces employment and earnings and increases criminal charges in the following quarters; on average, employment falls by approximately 18-30 percentage points, while criminal charges increase by approximately 0.3-0.4 percentage points. Crucially, magnitudes vary systematically with the intensity of the mass-layoff: effects are largest when separations occur in lower-intensity layoffs-consistent with negatively selected workers being targeted first-and attenuate as events approach true mass-layoff conditions. Selection is mostly within-firms rather than across firms. A simple labor demand model where criminal attitude signals labor productivity rationalizes these findings. Taken together, the results caution against treating mass layoffs as uniformly exogenous shocks and suggest that both research designs and policies should account for who is selected into displacement when assessing social costs and targeting support. |
| Presented by: Matteo Pazzona, Brunel University London |
| Session 57: LOCAL PROJECTIONS AND SVARS 3 June 24, 2026 10:15 to 12:00 Location: B008 |
| Session Chair: Sean McCrary, Ohio State University |
Heterogeneous Local ProjectionsAbstractIn this paper, we propose a novel methodology to estimate the dynamic effects of structural aggregate shocks on individual outcomes in the presence of a latent group structure. The methodology is based on a panel local projection estimator that allows researchers to assign individuals to the correct group and estimate group-specific dynamic responses, thus enabling researchers to estimate heterogeneous local projections. Monte Carlo simulations show that, in small samples, the methodology successfully identifies the group structure with high probability and delivers reliable inference. |
| Presented by: Yiru Wang, University of Pittsburgh |
Information matrix tests for switching regressionsAbstractThe EM principle implies the moments underlying the information matrix test for switching regressions are the expectation given the data of the moments one would test if one knew the subpopulation each observation originated from. Thus, we identify components related to conditional heteroskedasticity, conditional and unconditional skewness, and unconditional kurtosis of regression residuals within each regime. Simulations indicate analytical expressions for the asymptotic covariance matrix of those moments adjusted for sampling variability in parameter estimators provide reliable finite sample sizes and good power against various alternatives, especially combined with the parametric bootstrap. We apply the test to cross-country convergence regressions. |
| Presented by: Enrique Sentana, CEMFI |
Quantifying Deregulation and its Economic Effects: A Large Language Model ApproachAbstractWe construct a news-based index of deregulation for the United States from 1960 through 2025 using large language models (LLMs) to semantically classify newspaper articles. We distinguish articles discussing deregulation from those discussing increased regulation, assigning intensity scores that reflect both the centrality of deregulatory content and whether articles discuss advocacy, proposals, or enacted measures. Human validation confirms strong agreement between LLM and human classifications. The deregulation index captures major reform episodes including transportation and telecommunications liberalization in the 1970s--1980s, financial deregulation in the 1980s--1990s, and recent deregulatory activity. We decompose the index by sector, type of deregulation, and policy stage. We validate the news-based index against a parallel index constructed using Federal Register documents: the news-based index leads the Federal Register one by nine months, consistent with media coverage reflecting policy intentions before formal implementation. Unlike measures based on detailed statutory coding or Federal Register counts that weigh all rules equally, our approach covers the entire economy and weighs naturally by newsworthiness, capturing regulatory shifts before they materialize in law. Positive shocks to deregulation boost investment, productivity, stock prices, profits, and GDP. Industry-specific deregulation shocks boost industry-level stock returns, consistent with the evidence that a large part of deregulation measures in the United States may have favored incumbent firms rather than enabling competition. |
| Presented by: Danilo Cascaldi-Garcia, Federal Reserve Board |
Local Projections with Free-Knot SplinesAbstractThis paper introduces a nonparametric estimator for functions defined over discrete, naturally ordered supports and applies it to impulse response function estimation within the local projections (LP) framework. The estimator hypothesizes that the function admits a lower-dimensional representation by projecting it onto a basis of piecewise continuous functions. To account for uncertainty about the appropriate dimensionality in estimation and inference, the approach employs Jackknife model averaging across specifications with different dimensions, yielding the Averaged Projected Local Projections (APLP) estimator. Extensive simulations show that APLP has lower variance than the standard LP estimator while introducing little bias, and achieves similar coverage with shorter confidence intervals; the median APLP interval is about 20\% shorter. Two applications demonstrate that APLP improves interpretability by smoothing estimates across horizons and tightening confidence bands. |
| Presented by: Sean McCrary, Ohio State University |
| Session 58: MACHINE LEARNING METHODS June 24, 2026 10:15 to 12:20 Location: D-107 |
| Session Chair: Ruven Micheel, University of Münster |
Fused LASSO as Non-Crossing Quantile RegressionAbstractThis paper establishes a formal equivalence between non-crossing constraints in quantile regression and Fused LASSO regularisation with quantile-specific hyperparameters. This equivalence implies that imposing non-crossing constraints implicitly induces interquantile shrinkage, positioning the estimator of Bondell, Reich and Wang (2010) as one point on a bias-variance trade-off spectrum. We propose an adaptive framework that nests unconstrained quantile regression, non-crossing quantile regression, and composite quantile regression as special cases. Monte Carlo experiments demonstrate that cross-validation reliably selects regularisation intensities that outperform both unconstrained estimation and fully constrained alternatives. An empirical application to Fama-French factor models augmented with a market downturn indicator finds evidence of quantile-varying factor loadings, consistent with the directional predictability documented by Linton and Whang (2007). |
| Presented by: Tibor Szendrei, National Institute of Economic and Social Research |
Neural networks for nonlinear regression with serially correlated disturbances: Evidence from cloud coverAbstractWe propose a new treatment of nonlinear regression with serially correlated disturbances that incorporates autoregressive moving average structures into feedforward neural networks. The resulting model provides an alternative to modeling temporal dependence using lagged variables. In simulations, the proposed method accurately recovers regression functions of varying complexity and the underlying error dynamics across a range of time-series lengths and signal-to-noise ratios. Finite-sample properties and out-of-sample predictive performances are shown to be robust to model misspecification induced by omitted lagged variables and incorrect specification of the error dynamics. Cloud cover is an important factor in climate projections. In an empirical study of cloud cover prediction for a grid of locations within and around the Mediterranean Sea, our proposed model yields more accurate predictions than existing methods, including long short-term memory networks. Improvements are observed broadly and are particularly pronounced in mountain areas relative to linear models with serially correlated errors, consistent with the presence of stronger nonlinear effects in cloud composure in such regions. |
| Presented by: Sebastian Jensen, Aarhus University |
The Post Double LASSO for Efficiency AnalysisAbstractBig data and machine learning methods have become commonplace across economic milieus. One area that has not seen as much attention to these important topics yet is efficiency analysis. We show how the availability of big (wide) data can actually make detection of inefficiency more challenging. We then show how machine learning methods can be leveraged to adequately estimate the primitives of the frontier itself as well as inefficiency using the ‘post double LASSO’ by deriving Neyman orthogonal moment conditions for this problem and showing how they relate to the concept of moment and parameter redundancy from GMM. Finally, an application is presented to illustrate key differences of the post-double LASSO compared to other approaches. |
| Presented by: Artem Prokhorov, University of Sydney |
Bootstrapping with AI/ML-generated labelsAbstractAI/ML methods are increasingly used in economics to generate binary variables (or labels) via classification algorithms. When these generated variables are included as covariates in regressions, even small misclassification errors can induce large biases in OLS estimators and invalidate standard inference. We study whether the bootstrap can correct this bias and deliver valid inference. We first show that a seemingly natural fixed-label bootstrap, which generates data using estimated labels but relies on a corrupted version in estimation, is generally invalid unless a strong independence condition between the latent true labels and other covariates holds. We then propose a coupled-label bootstrap that jointly resamples the true and imputed labels, and show it is valid without this condition. Two finite-sample adjustments further improve coverage: a variance correction for uncertainty in estimated misclassification rates and a Hessian rotation for near-singular designs. We illustrate the methods in simulations and apply them to investigate the relationship between wages and remote work status. |
| Presented by: Benoit Perron, University of Montreal |
Deep Reinforcement Learning in Dynamic Economic Models: An Empirical StudyAbstractModern quantitative economics increasingly relies on high-dimensional models featuring non-linear dynamics and non-differentiable kinks induced by occasionally binding constraints. In such settings, conventional grid-based dynamic programming and projection methods succumb to the curse of dimensionality. Recent advances in deep learning have renewed interest in simulation-based approaches that approximate policy, value or conditional expectation functions using flexible function approximators. We study deep reinforcement learning (RL) as a numerical method for dynamic economic models, where neural-network policies are trained on simulated trajectories to maximize expected discounted utility. Despite deep RL's success in other application domains and a growing presence in economics, it remains unclear how these algorithms perform on the structural features that are typical in economic problems, for example stochastic transitions, various nonlinearities, and power-log reward functions. Therefore, we conduct a systematic study of four widely used policy-gradient methods on a staged set of canonical infinite-horizon economic models, designed to isolate how uncertainty, constraints, and dimensionality affect performance. To ensure fair and meaningful comparability, we establish a disciplined, budget-aware hyperparameter optimization and training protocol. We rigorously evaluate the algorithms along the dimensions of accuracy, stability, efficiency, and complexity using both metrics from computer science and economics. |
| Presented by: Ruven Micheel, University of Münster |
| Session 59: MICROECONOMETRIC METHODS 2 June 24, 2026 10:15 to 12:20 Location: D-110 |
| Session Chair: Sven Klaassen, Kiel University |
Partial least-squares for instrumental variable estimation with many instrumentsAbstractInstrumental variable estimation is widely used to estimate causal effects. However, this estimator is severely biased when the number of instruments increases. In order to improve the small sample properties of standard instrumental variables (IV) estimator, we use in the first stage equation a regularization technique called partial least-squares combined with cross-fitting. We show that our new estimator is consistent, asymptotically normal and attain the semiparametric efficiency bound. We propose a data-driven method for selecting the tuning parameter involved in our estimator. Simulations show that our approach reduces significantly the bias of IV estimator and permits to obtain reliable inference. We illustrate our method by an empirical application to the demand for automobiles where we augment the number of instruments by taking nonlinear transformations of the original instruments used in Berry, Levinsohn, and Pakes (1995). |
| Presented by: Marine Carrasco, University of Montreal |
Optimal Invariant Tests in an Instrumental Variables Regression With Heteroskedastic and Autocorrelated ErrorsAbstractThis paper uses model symmetries in the instrumental variable(IV) regression to derive an invariant test for the causal structural parameter. Contrary to popular belief, we show that there exist model symmetries when equation errors are heteroskedastic and autocorrelated (HAC). Our theory is consistent with existing results for the homoskedastic model(Andrews, Moreira, and Stock(2006) and Chamberlain(2007)). We use these symmetries to propose the conditional integrated likelihood(CIL) test for the causality parameter in the over-identified model. Theoretical and numerical findings show that the CIL test performs well compared to other tests in terms of power and implementation. |
| Presented by: Mahrad Sharifvaghefi, University of Pittsburgh |
Counterfactual Density Effects and the German East--West Income GapAbstractWe propose a novel framework for conducting causal inference based on counterfactual densities. While the current paradigm of causal inference is mostly focused on estimating average treatment effects (ATEs), which restricts the analysis to the first moment of the outcome variable, our density-based approach is able to detect causal effects based on general distributional characteristics. Following the Oaxaca-Blinder decomposition approach, we consider two types of counterfactual density effects that together explain observed discrepancies between the densities of the treated and control group. First, the distribution effect is the counterfactual effect of changing the conditional density of the control group to that of the treatment group, while keeping the covariates fixed at the treatment group distribution. Second, the covariate effect represents the effect of a hypothetical change in the covariate distribution. Both effects have a causal interpretation under the classical unconfoundedness and overlap assumptions. Methodologically, our approach is based on analyzing the conditional densities as elements of a Bayes Hilbert space, which preserves the non-negativity and integration-to-one constraints. We specify a flexible functional additive regression model estimating the conditional densities. We apply our method to analyze the German East--West income gap, i.e., the observed differences in wages between East Germans and West Germans. While most of the existing studies focus on the average differences and neglect other distributional characteristics, our density-based approach is suited to detect all nuances of the counterfactual distributions. |
| Presented by: Georg Keilbar, Humboldt-Universität zu Berlin |
Model Uncertainty and Measures of Inequality of OpportunityAbstractInequality of opportunity has great normative importance. This has led to a literature on measuring the part of overall inequality that is due to circumstances outside of a person's control. We contribute to such studies by evaluating the implications of uncertainty about circumstance variables and linear versus nonlinear transmission of circumstances on inequality of opportunity estimates. Applying linear Bayesian model averaging methods and three ensemble tree-based machine learning approaches to data from 31 European Union countries, we find that ignoring model uncertainty can lead to substantial overstatement of levels of inequality of opportunity. |
| Presented by: Andros Kourtellos, University of Cyprus |
Sensitivity Analysis for Treatment Effects in Difference-in-Differences Models using Riesz RepresentationAbstractDifference-in-differences (DiD) is one of the most popular approaches for empirical research in economics, political science, and beyond. Identification in these models is based on the conditional parallel trends assumption. We introduce a novel approach to sensitivity analysis for DiD models that assesses the robustness of DiD estimates to violations of this assumption due to unobservable confounders, allowing researchers to transparently assess and communicate the credibility of their causal estimation results. Our method focuses on estimation by Double Machine Learning and extends previous work on sensitivity analysis based on Riesz Representation in cross-sectional settings. We establish asymptotic bounds for point estimates and confidence intervals in the canonical setting and group-time causal parameters in settings with staggered treatment adoption. We provide extensive simulation experiments demonstrating the validity of our sensitivity approach and diagnostics and apply our approach to two empirical applications. |
| Presented by: Sven Klaassen, Kiel University |
| Session 60: MONETARY UNION MACROECONOMICS June 24, 2026 10:15 to 12:00 Location: D-113 |
| Session Chair: Jan Prüser, |
One Policy Rate, Many Stances: Evidence from the European Monetary UnionAbstractA challenge for conducting monetary policy in a currency union is the diverse economic conditions among member states. Such disparities can drive natural interest rates apart, thereby undermining the stabilizing role of a unified monetary policy. To assess the stance of monetary policy across Eurozone-19 countries, we estimate their natural rates of interest (r⇤) and inflation trends (⇡⇤) to construct a measure of the country-level neutral nominal interest rates (rstar + pistar) over 1999-2025, using a semi- structural model that jointly characterizes the trend and cyclical components of key macroeconomic variables such as output, unemployment, inflation, 10-year government bond yields, and the common policy interest rate. Our setup improves upon those in the existing literature by allowing both a short-run interest rate gap—driven by the (shadow) policy rate—and a long-run interest rate gap—driven by the country-specific 10-year government bond yields—to affect and reflect economic conditions. We also impose cointegration between the dynamics of the country-specific latent variables and common counterparts to incorporate co-movements across the euro area economies. Our results show that the stance of monetary policy is homogeneous across the countries in our sample, but that a relatively high degree of heterogeneity emerges at key historical turning points. |
| Presented by: Diego Vilan, Federal Reserve Board of Governors |
Tax and Spending Multipliers in a Monetary UnionAbstractWe examine the macroeconomic effects of federal tax and spending changes in countries of a monetary union. First, we calibrate a dynamic-stochastic general equilibrium model to the average euro area country and document that, as monetary policy is unresponsive, government investment and government consumption shocks have large multipliers and consumption tax and income tax shocks have small multipliers. Then, we test these predictions on euro area data by identifying the same four fiscal shocks in one encompassing panel structural vector autoregressive model through cross-sectional heteroskedasticity and time-fixed effects. We find that spending multipliers are large and tax multipliers small. |
| Presented by: Malte Rieth, University Halle-Wittenberg |
GLOBAL FACTORS FOR LOCAL SHOCKS IN A DATA-SCARCE ENVIRONMENT: WITH AN APPLICATION TO REGIONAL FISCAL MULTIPLIERS IN ITALYAbstractWe propose a novel econometric methodology to estimate Structural Vector Autoregressions with external instruments (proxy-SVARs or SVAR-IVs) in panel data characterized by strong cross-sectional dependence, dynamic heterogeneity, and a limited choice of external instruments for the shocks of interest. To account for cross-unit dependence, we specify for each unit of the panel a Factor-Augmented proxy-SVAR (proxy-FA-SVAR) that incorporates factors summarizing cross-sectional information from the non-policy variables of the system. The impact of the policy (target) shocks on the variables of interest is then recovered indirectly by estimating local policy reaction functions from the unit-specific proxy-FA-SVARs through a Minimum Distance approach. This is achieved using global instruments for the non-policy shocks, which are proxies common to all units in the panel. These global instruments are derived from a separate SVAR specified on factors that approximate all policy and non-policy variables in the panel. Global instruments can be further complemented with local (idiosyncratic) instruments, which are unit-specific and also constructed from SVARs estimated unit-wise. The joint use of global and local instruments makes the proxy-FA-SVARs overidentified, hence statistically testable. The proposed methodology is illustrated empirically by estimating government spending multipliers in Italy at the NUTS-2 level using annual data. Global and local instruments for the regional output shocks are derived from suitably specified Blanchard-Perotti-type SVARs for government spending and output. |
| Presented by: Marco Mazzali, Università di Bologna |
Assessing the Effects of Monetary Shocks on Macroeconomic Stars: A SMUC-IV FrameworkAbstractThis paper proposes a structural multivariate unobserved components model with external instrument (SMUC-IV) to investigate the effects of monetary policy shocks on key U.S. macroeconomic "stars"-namely, the level of potential output, the growth rate of potential output, trend inflation, and the neutral interest rate. A key feature of our approach is the use of an external instrument to identify monetary policy shocks within the multivariate unobserved components modeling framework. We develop an MCMC estimation method to facilitate posterior inference within our proposed SMUC-IV framework. In addition, we propose an marginal likelihood estimator to enable model comparison across alternative specifications. Our empirical analysis shows that contractionary monetary policy shocks have significant negative effects on the macroeconomic stars, highlighting the nonzero long-run effects of transitory monetary policy shocks. |
| Presented by: Jan Prüser, |
| Session 61: PANEL TIME SERIES June 24, 2026 10:15 to 12:20 Location: D-105 |
| Session Chair: Michael Pfaffermayr, University of Innsbruck |
Testing the null hypothesis of panel cointegration with common factorsAbstractThe paper addresses testing the null hypothesis of panel cointegration with cross-section dependence driven by unobserved common factors. The use of the continuous updated estimator proposed in Bai et al. (2009) allows consistent estimation of the cointegrating vector, the common component and the idiosyncratic component, which establishes the framework to propose a Lagrange-multiplier statistic for cointegration. The empirical relevance of the proposed methodology is illustrated through two applications. First, revisiting the Feldstein-Horioka puzzle using quarterly data for 20 OECD economies (1981-2025), we find robust evidence of panel cointegration after accounting for common factors. Second, using the Coe et al. (2009) dataset and an extended R&D panel for 24 OECD countries (1971-2023), we show that productivity, domestic R&D capital, foreign R&D spillovers, and human capital form a stable long-run relationship once unobserved global components are controlled for. Overall, the proposed framework provides a flexible and reliable approach for testing panel cointegration in environments with pervasive cross-section dependence and latent global trends |
| Presented by: Josep Lluís Carrion-i-Silvestre, University of Barcelona |
The Output Convergence Debate Revisited: Lessons from recent developments in the analysis of panel data modelsAbstractThis paper provides a critical examination of the empirical basis of the output convergence debate in the light of recent developments in the analysis of dynamic heterogeneous panels with interactive effects. It shows that popular tools such as Barro's cross-country regressions and two-way fixed effects (TWFE) estimators that assume parallel trends and homogeneous dynamics lead to substantial under-estimation of the speed of convergence and misleading inference. Instead, dynamic common correlated effects (DCCE) estimators due to Chudik and Pesaran (2015a) provide consistent estimates and valid inference that are robust to nonparallel trends and correlated heterogeneity and apply even if there are breaks, trends and/or unit roots in the latent technology factor. It also suggests a way to estimate the effect of slowly moving determinants of growth. The theoretical findings are augmented with empirical evidence using Penn World Tables data, finding little evidence of per capita output convergence across countries, very slow evidence of cross country growth convergence, and reasonably fast within country convergence. Capital accumulation is found to be the most important single determinant of cross-country differences in output while slow moving indicators such as potential for conflict and protection of property rights proved to be statistically significant determinants of the steady state levels of output per capita. We are also able to replicate a positive evidence of democratization on output, but we find that the statistical significance of this effect to fall as we allow for nonparallel trends and dynamic heterogeneity. |
| Presented by: Ronald Smith, Birkbeck, University of London |
Integrated Modified Least Squares Estimation and (Fixed-b) Inference for Systems of Cointegrating Multivariate Polynomial RegressionsAbstractWe consider integrated modified ordinary and generalized least squares estimation for systems of cointegrating multivariate polynomial regressions, i. e., systems of regressions that include deterministic variables, integrated processes and products of non-negative integer powers of these variables as regressors. The stationary errors are allowed to be correlated across equations, over time and with the regressors. The necessity to consider integrated modified generalized least squares estimation arises in case of estimation under restrictions, which in general implies that ordinary and generalized least squares estimators cease to be identical. We discuss in detail hypothesis testing for the unrestricted and restricted estimators. Furthermore, we develop asymptotically pivotal fixed-b inference, which is shown to be available only in the case of full design for up-to-the-intercept-identical hypotheses tested in all equations in systems with identical regressors in all equations. |
| Presented by: Martin Wagner, University of Klagenfurt and IHS |
Thick-Tailed Panel and Time-Series Data: Inference with Nonexistent MomentsAbstractThis paper proposes efficient estimation methods for panel data models with nonexistent moments of unobservables. The methods apply to linear as well as nonlinear panel data models, static or dynamic, additive or nonadditive in the disturbances. The problem of nonexistence of moments of the unobservables is crucial for financial econometrics as well as econometric modelling of climate change. Non-Gaussian distributions differ in fundamental ways from the Gaussian distribution: (a) Linear Functions of Non-Gaussian r.v.s typically have Di¤erent Distributions than the Original r.v.s; (b) Increasing the sample size may not lead to more precise inference; (c) with non-Gaussian r.v.s, zero correlation and statistical independence can have drastically di¤erent implications for econometric inference; (d) quasi-MLE methods with errors following identical marginal distributions can have drastically different properties depending on the temporal dependency of the unobservables; (e) in LDV models, even if the nonlinear) regression parameters were consistently estimated, non-Gaussian distributions would result in poor estimators for the marginal effects. Generalizations of Regression and GMM approaches are shown to perform poorly in the presence of non-Gaussian distributions, especially those with "thick tails", typically exhibiting serious inconsistencies. We show how Maximum Likelihood methods will instead provide the most efficient consistent, uniformly asymptotically normal (CUAN) estimators for this general class of panel data econometric models. |
| Presented by: Vassilis Hajivassiliou, London School of Economics |
Structural Panel Gravity Models: Residual Analysis and DiagnosticAbstractThis note reconsiders the estimation of structural gravity models with a staggered difference-in-difference design to evaluate the impact of preferential trade agreements of bilateral trade. Most importantly, it provides a detailed residuals analysis and a series of diagnostic tests based on them. Results indicate the presence of severe outliers, autocorrelated disturbances, pronounced heteroskedasticity and sample selection effects due to missing trade flows. However, cross-sectional dependence in the exporter and importer country dimension is weak, suggesting that clustering the standard errors by country-pair is sufficient. Lastly, there is evidence of further sources of treatment heterogeneity such as the depth of PTAs. |
| Presented by: Michael Pfaffermayr, University of Innsbruck |
| Session 62: SUBJECTIVE EXPECTATIONS June 24, 2026 10:15 to 12:20 Location: D-114 |
| Session Chair: Laura Hospido, Banco de España, CEMFI and IZA |
Five-Year Impacts of High School Financial Education on Knowledge, Preferences and Schooling DecisionsAbstractWe study the medium-run effects of a 10-hour high school financial education course delivered in Spain at age 14-15, following 1254 students five years after exposure. We examine their financial knowledge, financial inclusion, measures of patience and their educational choices. To assess whether the program improved educational decision-making, we combine elicited data on patience, expected returns to alternative schooling paths and the observed choice. Identification relies on within-school comparisons between treated 9th graders and a control group of students one grade older. Given the non-randomized design, we benchmark cohort differences to external population surveys and to in-sample estimates of groups that did not benefit from the program. We find no detectable impacts on financial knowledge, financial inclusion or schooling attainment. Our results are tentative, but suggest increased patience (2pp increase in the weight given to future consumption) and an improvement in the consistency between the chosen educational path and the perceived returns to schooling. |
| Presented by: Diego Gonzalez, AIREF |
Separating Preferences from Endogenous Effort and Cognitive Noise in Observed DecisionsAbstractWe develop a micro-founded framework to account for individuals' effort and cognitive noise which confound estimates of preferences based on observed behavior. Using a large-scale experimental dataset we find that observed decision noise responds to the costs and benefits of exerting effort on individual choice tasks as predicted by our model. We estimate that failure to properly account for decision errors due to (rational) inattention on a more complex, but commonly used, task design biases estimates of risk aversion by 50% for the median individual. Effort propensities recovered from preference elicitation tasks generalize to other settings and predict performance on an OECD-sponsored achievement test used to make international comparisons. Furthermore, accounting for endogenous effort allows us to empirically reconcile competing models of discrete choice. |
| Presented by: Tomas Jagelka, University of Bonn |
Seeing the Economy through Colored Glasses: Partisanship in Macro and (not in) Micro ExpectationsAbstractPolitical views affect households’ macroeconomic expectations, but personal economic circumstances and self-interested motives remain the dominant factors shaping their beliefs. Using an expanded dataset covering 11 U.S. Presidential elections from 1980 to 2020, we show that households' personal finance expectations exhibit significantly less partisan bias than their macroeconomic expectations, as households are more directly informed about their own situations. By linking microeconomic beliefs to corresponding macroeconomic expectations, we differentiate between partisan bias, political sentiment, and differences in belief extrapolation. An empirically estimated factor model quantifies the time-varying importance of partisanship and microeconomic disparity in driving polarized views of the macroeconomy. Finally, we show that households “cheerlead” for policies to be beneficial to the broader economy, often not because such policies are enacted by their favored winning party, but because they expect to personally gain from them. |
| Presented by: Tao Wang, Bank of Canada |
Subjective Expectations and Household Portfolio ChoicesAbstractHouseholds’ expectations, including those on their future income, are a critical driver of financial decisions. Using seventeen years of panel data linking survey measures of a wide range of subjective expectations to administrative records on household portfolios, we examine how changes in expectations influence saving and investment in risky financial assets. While saving decisions are not sensitive to expectations regarding the macro environment, they do respond to changes in micro-expectations. The median of future income and upward income risk, a longer right tail, each have positive effects on the decision to invest in stocks and bonds. A 10% increase in median future income leads to a 0.7% increase in stock ownership and a 1% increase in the value of stockholdings. The findings on stock market entry are driven by young households and the intensive margin effects are driven by older households. |
| Presented by: Ruben van den Akker, Tilburg University |
Income uncertainty, nonlinear dynamics and consumption: a subjective income expectations frameworkAbstractHousehold consumption and saving decisions depend on perceived income risk, yet macroeconomic models often discipline income processes using realized income data. We show that these two objects differ in systematic and economically meaningful ways. We develop a framework that combines subjective expectations with realized income data, while addressing non-classical elicitation error and focal-point responses, to estimate flexible income processes that capture nonlinear persistence and distributional asymmetries. Households perceive income to be substantially more persistent and less dispersed than realized data imply. Our findings support incorporating subjective probability data into structural models of income dynamics, consumption, and precautionary behavior. |
| Presented by: Julio Galvez, CUNEF Universidad |
| Session 63: TREATMENT EFFECTS June 24, 2026 10:15 to 12:00 Location: D-106 |
| Session Chair: Justin Young, Microsoft |
Quantile Individualized Average Treatment EffectsAbstractThis paper introduces the Quantile Individualized Average Treatment Effect (QIATE), a new parameter describing fine-grained causal heterogeneity. The QIATE focuses on the 'actionable' part of the causal heterogeneity, which depends on observable characteristics of individual units only. We propose two estimators: a first-stage agnostic estimator, which can be applied with any causal machine learning method, and one specifically tailored to the Modified Causal Forest. A simulation analysis indicates that both estimators are consistent and informative about treatment effect heterogeneity. We illustrate their value in an empirical analysis of heterogeneity in the effect of smoking during pregnancy on birth weight. |
| Presented by: Johanna Kutz, University of St. Gallen |
Automatic debiased machine learning and sensitivity analysis for sample selection modelsAbstractIn this paper, we extend the Riesz representation framework to causal inference under sample selection, where both treatment assignment and outcome observability are non-random. Formulating the problem in terms of a Riesz representer enables stable estimation and a transparent decomposition of omitted variable bias into three interpretable components: a data-identified scale factor, outcome confounding strength, and selection confounding strength. For estimation, we employ the ForestRiesz estimator, which accounts for selective outcome observability while avoiding the instability associated with direct propensity score inversion. We assess finite sample performance through a simulation study and show that conventional double machine learning approaches can be highly sensitive to tuning parameters due to their reliance on inverse probability weighting, whereas the ForestRiesz estimator delivers more stable performance by leveraging automatic debiased machine learning. In an empirical application to the gender wage gap in the U.S., we find that our ForestRiesz approach yields larger treatment effect estimates than a standard double machine learning approach, suggesting that ignoring sample selection leads to an underestimation of the gender wage gap. Sensitivity analysis indicates that implausibly strong unobserved confounding would be required to overturn our results. Overall, our approach provides a unified, robust, and computationally attractive framework for causal inference under sample selection. |
| Presented by: Theresa M. A. Schmitz, Heinrich Heine University Düsseldorf |
The Identification Power of Combining Experimental and Observational Data for Distributional Treatment Effect ParametersAbstractThis study investigates the identification power gained by combining experimental data, in which treatment is randomized, with observational data, in which treatment is self-selected, for distributional treatment effect (DTE) parameters. While experimental data identify average treatment effects, many DTE parameters, such as the distribution of individual treatment effects, are only partially identified. We examine whether and how combining these two data sources tightens the identified set for such parameters. For broad classes of DTE parameters, we derive nonparametric sharp bounds under the combined data and clarify the mechanism through which data combination improves identification relative to using experimental data alone. Our analysis highlights that self-selection in observational data is a key source of identification power. We establish necessary and sufficient conditions under which the combined data shrink the identified set, showing that such shrinkage generally occurs unless selection-on-observables holds in the observational data. We also propose a linear programming approach to compute sharp bounds that can incorporate additional structural restrictions, such as positive dependence between potential outcomes and the generalized Roy selection model. An empirical application using data on negative campaign advertisements in the 2008 U.S. presidential election illustrates the practical relevance of the proposed approach. |
| Presented by: Shosei Sakaguchi, The University of Tokyo |
Reevaluating Causal Estimation Methods with Data from a Product ReleaseAbstractRecent developments in causal machine learning methods have made it easier to estimate flexible relationships between confounders, treatments and outcomes, making unconfoundedness assumptions in causal analysis more palatable. How successful are these approaches in recovering ground truth baselines? In this paper we analyze a new data sample including an experimental rollout of a new feature at a large technology company and a simultaneous sample of users who endogenously opted into the feature. We find that recovering ground truth causal effects is feasible—but only with careful modeling choices. Our results build on the observational causal literature beginning with LaLonde (1986), offering best practices for more credible treatment effect estimation in modern, high-dimensional datasets. |
| Presented by: Justin Young, Microsoft |
| Session 64: Lunch June 24, 2026 12:00 to 13:30 |
| Session 65: ASSET PRICE BUBBLES June 24, 2026 13:30 to 15:15 Location: D-110 |
| Session Chair: Anthony Sanford, HEC Montréal |
An Unobserved Components Based Test for Asset Price BubblesAbstractThe general solution of the standard stock pricing equation commonly employed in the finance literature decomposes the price of an asset into the sum of a fundamental price and a bubble component that is explosive in expectation. Despite this, the extant literature on bubble detection focuses almost exclusively on modelling asset prices using a single time-varying autoregressive process, a model which is not consistent with the general solution of the stock pricing equation. We consider a different approach, based on an unobserved components time series model whose components correspond to the fundamental and bubble parts of the general solution. Based on the locally best invariant testing principle, we derive a statistic for testing the null hypothesis that no bubble component is present, against the alternative that a bubble episode occurs in a given subsample of the data. In order to take an ambivalent stance on the possible number and timing of the bubble episodes, our proposed test is based on the maximum of a doubly recursive implementation of this statistic over all possible break dates. Simulation results show that our proposed tests can be significantly more powerful than the industry standard tests developed by Phillips, Shi and Yu (2015). |
| Presented by: David Harvey, University of Nottingham |
EU ETS Market Expectations and Rational BubblesAbstractSerious concerns about a price bubble in the European Union Emissions Trading System (EU ETS) emerged during its third trading period, as allowance prices rose sharply and several studies attributed the surge to rational bubbles. We reassess this claim using an expectations-based test that exploits futures prices and thus avoids specifying a fundamental value. Importantly, we show that neglecting risk premia within this framework can generate spurious evidence of bubbles. We therefore develop a testing approach that remains valid in the presence of a dynamic risk premium and is robust when the underlying fundamental is unit root or mildly explosive. Using weekly spot and futures data from 2013 to 2023, we find that the explosive price dynamics are not attributable to rational bubbles. Instead, the evidence is consistent with shifting expectations of future allowance scarcity rather than speculative bubbles. |
| Presented by: Robinson Kruse-Becher, University of Hagen |
Beyond the Numbers: Professional Forecasters’ Narratives about Inflation and Stock Market PerformanceAbstractProfessional forecasters' heterogeneous narratives about how inflation affects stock market performance in 2023 rationalize their high disagreement regarding quantitative expectations for 12-month-ahead inflation and stock returns in December 2022. Professional forecasters causally update their return expectations in heterogeneous directions in response to information about the inflation outlook, depending on their entertained narrative. Moreover, the narratives also affect their asset allocation in a hypothetical portfolio-choice experiment. Hence, providing common signals does not necessarily lead to convergence of beliefs if agents subscribe to different narratives. |
| Presented by: Christian Conrad, Heidelberg University |
Size Distortions in Robust Estimators: Implications for Asset PricingAbstractWe evaluate the reliability of HAC estimators in asset pricing. Through simulations, we show these estimators produce inflated t-statistics and invalid inference when applied to returns with autocorrelation and heteroskedasticity. To address this, we introduce SHARFS, a simulation-based inference procedure estimating p-values using empirically calibrated null data-generating processes. Unlike traditional methods, SHARFS provides valid finite-sample inference under complex return dynamics. Applied to 212 anomalies, we find that standard estimators overstate significance: strategies deemed significant by HAC methods fail to pass our robust test. Our results challenge the credibility of conventional inference in finance and call for a shift toward simulation-based methods. |
| Presented by: Anthony Sanford, HEC Montréal |
| Session 66: CAUSAL INFERENCE WITH PANEL DATA June 24, 2026 13:30 to 15:15 Location: B009 |
| Session Chair: Alexander Newton, London School of Economics |
Double Machine Learning for Static Panel Data with Instrumental Variables: New Method and ApplicationsAbstractPanel data methods are widely used in empirical analysis to address unobserved heterogeneity, but causal inference remains challenging when treatments are endogenous and confounding variables high-dimensional and potentially nonlinear. Standard instrumental variables (IV) estimators, such as two-stage least squares (2SLS), become unreliable when instrument validity requires flexibly conditioning on many covariates with potentially non-linear effects. This paper develops a Double Machine Learning estimator for static panel models with endogenous treatments (panel IV DML), and introduces weak-identification diagnostics for it. We revisit three influential migration studies that use shift-share instruments. In these settings, instrument validity depends on rich covariate adjustment. In one application, panel IV DML strengthens the pre dictive power of the instrument and broadly confirms 2SLS results. In the other cases, flexible adjustment makes the instruments weak, leading to substantially more cautious inference than conventional 2SLS. Monte Carlo evidence supports these findings, showing that panel IV DML improves estimation accuracy under strong instruments and delivers more reliable inference under weak identification. |
| Presented by: ANNALIVIA POLSELLI, UNIVERSITY OF ESSEX |
Estimating the Intensive Margin Effect in Panel Data SettingsAbstractMany policies operate through two different channels: the extensive margin (e.g., the decision to participate) and the intensive margin (e.g., the intensity of the response among participants). This paper develops a novel identification strategy to estimate the intensive margin effect in panel data settings. I adapt the Horowitz-Manski-Lee bounds to the Changes-in-Changes framework to partially identify both the average and quantile intensive margin treatment effects. Additionally, I explore how to leverage multiple sources of sample selection to relax the monotonicity assumption in the original Horowitz-Manski-Lee bounds, which may be of independent interest. Alongside the identification strategy, I present estimators and inference results. I illustrate the relevance of the proposed methodology by analyzing a job training program in Colombia. |
| Presented by: Javier Viviens, |
Synthetic Control and Synthetic Difference-in-Differences: An Asymptotic Optimality PerspectiveAbstractWeighted panel estimators such as Synthetic Control (SC) and Synthetic Difference-in-Differences (SDID) have become central tools for program evaluation, yet their statistical accuracy remains only partially understood. We first show that the SC estimator is asymptotically optimal for estimating post-treatment effects: as the numbers of preand post-treatment periods grow, SC attains the lowest possible squared risk among all estimators based on weighted averages of control units. This result holds broadly, regardless of whether the SC estimator is unbiased, and applies when the number of control units is either fixed or divergent. It further implies that SCM cannot be outperformed by matching, inverse probability weighting, or difference-in-differences within this class of estimators. In contrast, we show that the SDID estimator does not achieve such optimality for the average post-treatment effect. Motivated by this limitation, we propose the Optimal Synthetic Difference-in-Differences (OSDID) estimator, which introduces optimal time weights to compensate for the limitations of unit weights. We establish that OSDID is asymptotically optimal: it attains the infimum of the squared risk for estimating the average post-treatment effect within the class of estimators sharing the SDID weighting structure and using the same unit weights as OSDID. |
| Presented by: Wendun Wang, Erasmus University Rotterdam |
Heterogeneous Elasticities, Aggregation and Retransformation BiasAbstractEconomists often interpret estimates from linear regressions with log dependent variables as elasticities. However, the coefficients from log-log regressions estimate the elasticity of the geometric mean of $y_i|x_i$ , not the arithmetic mean. The unbounded difference between the two is known as retransformation bias and can take either sign. We develop a specification-robust debiased estimator of the average arithmetic mean elasticity and re-estimate 50 results from top 5 papers published in 2020. We find that 19 are significantly different, with the median absolute difference being 65% of the OLS elasticity estimate. Furthermore, we show standard instrumental variables assumptions with log dependent variables do not identify the arithmetic mean elasticity. We specify a control function approach and re-estimate top 5 papers that use 2SLS with log dependent variables. Retransformation bias arises as a result of heterogeneous responses. The geometric mean elasticity corresponds to the average response. Arithmetic and geometric means are elements of the power mean family. We show power mean elasticities are sufficient statistics for a common class of decision problems. |
| Presented by: Alexander Newton, London School of Economics |
| Session 67: DEMAND ESTIMATION June 24, 2026 13:30 to 15:15 Location: D-106 |
| Session Chair: Sung-Jin Cho, Seoul National University |
The Informational Content of Consumer Choice in Differentiated Product MarketsAbstractWe study the impact of consumer inattention on market outcomes for the US ready-to-eat cereal market by estimating a discrete-type mixed logit model with heterogeneous consideration sets within and between consumer types. The full information benchmark model is statistically rejected against all limited consumer attention specifications. Under the full information assumption own-price elasticities are inflated and cross-price elasticities are an order of magnitude smaller than those of our most preferred limited consumer attention specification. Product-level markups are higher under limited attention and are estimated by all models to increase over the period from 2006 to 2020. The consideration proxy that best fits the observable data has on average six products, while there are on average 153 products in the market. While consumer behavior is best explained by limited attention, our model selection tests indicate that firms on average expect consumers to be fully informed when setting prices. |
| Presented by: Johannes Kandelhardt, Heinrich-Heine-Universität Düsseldorf |
Consumer Choice over Shopping Baskets: A Linear Demand ApproachAbstractI introduce a novel approach to modelling and estimating continuous demand systems, utilising \textit{consideration sets} to analyse differentiated products markets with very large choice sets, and where purchases over multiple goods, multiple units, and across product categories are common. I apply this approach to study intra-store competition in the Portuguese supermarket industry between 2020 and 2023. Anonymised transaction-level point-of-sale data is sufficient to estimate price elasticities across almost 30 000 goods and more than 500 product categories. The implied markups match observed price volatility, profit margin surveys, as well as reports on shifting consumer tastes during the sample period. |
| Presented by: Afonso Rodrigues, University of Oxford |
Spatial Differences in Premiums for High Quality Products: A Case Study of Organic Baby FoodAbstractHow do markups and costs explain prices of high-quality products? Why price premiums for these products are lower in wealthier or more educated areas? Using scanner data on baby food, I document that organic products generally command substantial price premiums over non-organic alternatives, but these premiums decline with county-level income, population, and education. I estimate a random coefficient nested logit demand model, and recover product-specific markups and costs across local markets. I find that organic products typically have higher costs and lower markups compared to non-organic products. For both products, markups increase and costs decrease with county-level income and education. For organic products, however, the increase in markups is less pronounced, while the decrease in costs is more pronounced, aligning with the observed spatial variation in premiums. Removing competition among organic products reduces spatial differences in organic premiums by about half, mainly due to increased markups in richer or more educated areas. Spatial differences in costs, likely tied to distribution rather than production, explain the remaining price variations. Additionally, wealthier or more educated areas benefit more from organic products due to higher consumer surplus and variable profits. These findings underscore the role of local market structures on spatial variations in price premiums, consumer welfare, and market efficiency. |
| Presented by: Seung-Hyun Hong, University of Illinois, Urbana-Champaign |
Semi-parametric instrument-free demand estimation: relaxing optimality and equilibrium assumptionsAbstractIn most markets, consumer demand results from a compound arrival/choice process: consumers arrive to a market stochastically and make independent discrete choices over which item to purchase (or not to purchase, often referred to as the “choice of the outside good”). Market demand results from an aggregation of individual consumer choices, and in general is more accurately modeled as a pricedependent probability distribution (or stochastic process) rather than as a linear demand function. We consider the problem of identifying the underlying structure of demand — consumer preferences and the distribution of arrivals — when market prices are endogeneously determined but the implied distribution of demand (including mean demand) is potentially nonlinear in prices and there are no relevant instrumental variables. In addition, demand data are truncated and censored: we do not observe the number of arriving customers or those choosing the outside good, and we only observe the minimum of demand and each hotel’s capacity. Recent studies have shown that the hypotheses of a) optimality, and b) equilibrium constitute powerful identifying restrictions that enable consistent estimation of demand in the presence of a variety of endogeneity and censoring problems. In this paper we consider whether it is possible to identify demand when assumptions a) and b) are relaxed. We introduce a conditional independence assumption that implies that after controlling for a vector of “demand shifters” x other variables affecting firm prices, z, do not also affect d emand. This implies that by controlling for x we can exploit the residual variation in firm prices due to the unobserved shocks z as “virtual price experiments” to identify the underlying structure of demand. We illustrate our approach via an empirical analysis of demand for hotels. We show that the distribution of hotel occupancies is a mixture of censored multinomial distributions that depends on prices, consumer preferences, and the distribution of arriving customers. In our empirical application, we show that the structure of demand can be estimated in a situation where there are no relevant instrumental variables without the need to impose optimality or equilibrium restrictions, with limited data. Our estimates imply stochastically shifting but downward sloping expected demand curves. Using the estimated stochastic process for demand we are able to test and reject the hypothesis that hotels are setting their prices optimally (as well as prices predicted under Bertrand-Nash equilibrium) and we strongly reject claims that revenue management systems (RMS) used by hotels to set prices are implementing “algorithmic collusion.” |
| Presented by: Sung-Jin Cho, Seoul National University |
| Session 68: ECONOMETRIC THEORY June 24, 2026 13:30 to 15:15 Location: D-107 |
| Session Chair: Mario Martinoli, Sant'Anna School of Advanced Studies |
Variance Estimation with Dependence and Heterogeneous MeansAbstractThis paper considers the problem of estimating the variance of a sum of a triangular array of random vectors with heterogeneous means. When random vectors exhibit two-way cluster dependence or weak dependence, standard variance estimators designed under homogeneous means can underestimate the true variance, which results in subsequent tests being oversized. To restore validity, this paper proposes a conservative variance estimator robust to heterogeneous means and shows its asymptotic validity. |
| Presented by: Luther Yap, National University of Singapore |
Two-way clustering with non-exchangeable dataAbstractInference procedures for dyadic data based on two-way clustering rely on the data being exchangeable and dissociated. In particular, observations must be independent if they have no index in common. In an effort to relax this we consider, instead, data where Y_{ij} and Y_{pq} can be dependent for all index pairs, with the dependence vanishing as the distance between the indices grows large. We establish limit theory for the sample mean and propose analytical and bootstrap procedures to perform inference. |
| Presented by: Koen Jochmans, Toulouse School of Economics |
Quantile optimization in semidiscrete optimal transportAbstractOptimal transport is the problem of designing a joint distribution for two random variables with fixed marginals. In virtually the entire literature on this topic, the objective is to minimize expected cost. This paper is the first to study a variant in which the goal is to minimize a quantile of the cost, rather than the mean. For the semidiscrete setting, where one distribution is continuous and the other is discrete, we derive a complete characterization of the optimal transport plan and develop simulation-based methods to efficiently compute it. One particularly novel aspect of our approach is the efficient computation of a tie-breaking rule that preserves marginal distributions. In the context of geographical partitioning problems, the optimal plan is shown to produce a novel geometric structure. |
| Presented by: Yinchu Zhu, Brandeis University |
Nonparametric Minimum-Distance Estimation of Simulation ModelsAbstractWe propose a simulated nonparametric minimum-distance estimator for the estimation of parameters of complex simulation models. To address the limitations of traditional simulation-based econometric techniques in cases where the stochastic equicontinuity condition is violated, we approximate the distance between real-world observations and data simulated from a theoretical model using a series of basis functions, allowing for the estimation of model parameters without relying on specific auxiliary models or moment selection. We study the consistency and rates of convergence of our estimator. We investigate its performance through Monte Carlo experiments and an empirical application to financial market data. |
| Presented by: Mario Martinoli, Sant'Anna School of Advanced Studies |
| Session 69: FUNCTIONAL DATA FORECASTING June 24, 2026 13:30 to 15:15 Location: D-111 |
| Session Chair: Philipp Gersing, |
Dynamic Volatility in Functional Electricity Price ForecastingAbstractThis paper develops a functional time series framework to analyze and forecast the dynamics of day-ahead electricity prices. Although prices are observed at discrete intervals, representing each day as a continuous function provides a richer characterization of intraday fluctuations. The long-range dependence is modeled through functional fractional integration, and dominant modes of variation are extracted via Functional Principal Component Analysis. The leading component scores are then framed individually using specifications that accommodate conditional heteroskedasticity, and their forecasts are combined to reconstruct predicted price curves. To quantify forecast uncertainty, a model-based Monte Carlo simulation procedure is proposed to construct prediction bands. Using data from the Iberian electricity market, the proposed functional modeling strategy shows that explicitly accounting for volatility dynamics helps to capture extreme price movements. As a result, more accurate prediction bands are obtained, as evidenced by significant reductions in the average Band Score (a proper scoring rule for interval forecasts), particularly at higher confidence levels. This leads to a more reliable assessment of tail risk. |
| Presented by: Eric Costa-Andreu, Universidad de Alicante |
Forecast Evaluation for Functional DataAbstractThis paper proposes methods for comparing the accuracy of two competing sets of functional forecasts. This is increasingly relevant as many economic and financial forecasters are interested in predicting variables which are observed as functional data objects. However, to date there have been no formal statistical tests to evaluate the relative accuracy of competing functional forecasts. We propose a suite of novel tests, building on the classic Diebold-Mariano test, to provide formal statistical guidance on forecast accuracy in the case of functional data. We derive the asymptotic properties of the tests, including their self-normalised versions, and demonstrate the validity of analytic or bootstrap-based critical values. We investigate the finite sample performance of the tests using Monte Carlo simulations. We demonstrate their practical usefulness by evaluating forecasts of the U.S. yield curve based on forward rates versus a random walk benchmark. |
| Presented by: Jack Fosten, Bayes Business School |
Forecast Combination with Random Subspaces: Accuracy Gains and InterpretabilityAbstractThis paper studies Random Subspace Regressions (RSM) as a forecast-combination strategy. We develop a finite-sample framework that makes explicit the trade-off between using more forecasts (lower combination variance) and estimating more weights (higher estimation error). Monte Carlo experiments, where individual forecasts are themselves estimated, show that RSM typically delivers lower RMSFE than equal weighting and restricted OLS. In an empirical application using real-time FRED-MD vintages to nowcast U.S. GDP growth, RSM significantly improves the forecast combination performance relative to the mean with the largest gains during the COVID-19 recession and compares favorably to Lasso, Ridge, and Random Forest benchmarks. Finally, we link RSM to Shapley value decomposition, yielding a transparent decomposition of the combined forecast into contributions of individual forecasts even in high-dimensional settings. |
| Presented by: Boris Kozyrev, Halle Institute for Economic Research (IWH) |
A Distributed Lag Approach to the Generalised Dynamic Factor ModelAbstractWe propose a new estimator for the Generalised Dynamic Factor Model (GDFM) that simplifies estimation by avoiding frequency-domain methods. Our key theoretical insight shows that the dynamic common component can be represented by a finite number of lags of contemporaneously pervasive factors under general conditions. This result reduces GDFM estimation to a simple OLS regression of observed variables on estimated factors and their lags, with factors obtained via static principal components. The approach naturally accommodates weak (non-pervasive) factors within the dynamic common space, addressing an important limitation of existing methods. We establish consistency and asymptotic normality for both the dynamic and weak common components. An application to a large European macroeconomic dataset demonstrates strong empirical performance and uncovers a sizeable weak common component - particularly in sentiment indicators and several other variables - revealing dynamics that standard methods overlook. |
| Presented by: Philipp Gersing, |
| Session 70: GLOBAL INFLATION June 24, 2026 13:30 to 15:15 Location: D-105 |
| Session Chair: Felipe Martinez, Central Bank of Chile |
Global Inflation, Regional FactorsAbstractThis paper shows that global inflation dynamics have a sizable regional component. Using a balanced panel of 61 countries that starts in 1970, we document that while the global factor, defined as the dominant principal component, explains a large portion of inflation variation in advanced economies, a model with only one principal component is less successful for developing countries. By contrast, a hierarchical dynamic factor model, which includes a global (unconstrained) factor and regional (restricted) factors, performs substantially better for emerging market and developing economies. The regional factors are linked to commodity prices and help improve the accuracy of inflation forecasts at the country level. Employing an unsupervised machine-learning technique, we show that the estimated clusters of countries, grouped according to similarities in inflation dynamics, exhibit a strong regional pattern. Our findings suggest that policymakers in developing countries should pay close attention to inflation dynamics in their neighboring countries. |
| Presented by: Hillary Stein, Federal Reserve Bank of Boston |
Unpacking Global InflationAbstractWe study the international synchronization of inflation by disaggregating inflation into goods and services components. This split is motivated by the goods and services sectors being closely aligned with tradable and nontradable inflation, respectively. We construct a dataset of goods and services inflation for 42 countries spanning 1971Q1–2023Q4 and estimate a dynamic block factor model to identify global sources of inflation. Three main findings emerge. First, the sectoral decomposition reveals multiple sources of global inflation synchronization, in contrast to the single global inflation factor typically identified in the literature. These factors differ in persistence and in their effects on goods and services inflation. Second, while global factors dominate goods inflation dynamics, they also account for a meaningful—albeit smaller—share of services inflation, indicating that even traditionally domestic inflation components exhibit global comovement. Third, the importance and composition of global factors have evolved over time and across crises. Subsample analysis suggests that their contribution has increased since the Great Moderation. These results imply that the degree of monetary policy autonomy depends on the source of global inflation: tradable and commodity shocks largely reflect external cost pressures, while synchronized nontradable inflation reflects correlated domestic demand and financial conditions across countries. |
| Presented by: Yun Young Gwak, Bank of Korea, Monash University |
Is Inflation Driven by Aggregate or Sectoral Output Gaps?AbstractWe examine whether inflation is driven by aggregate or sectoral output gaps. The aggregate output gap may not fully capture inflationary pressures because it can obscure sectoral shocks and heterogeneity in propagation to prices. We find that aggregating sectoral output gaps by weights estimated from real-time regressions produces a better fit of the Phillips curve than using the aggregate output gap. We confirm the sectorally-aggregated output gap based on these weights has significant explanatory power for inflation beyond the aggregate output gap and find it performs better in forecasting inflation, although the aggregate output gap retains its own distinct information. |
| Presented by: James Morley, University of Sydney |
In ation Heterogeneity and Differential Effects of Monetary and Oil Price ShocksAbstractThis paper examines the heterogeneous effects of monetary policy and oil price shocks on inflation across the income distribution in Chile. We find that a contractionary monetary policy shock significantly reduces inflation for all income deciles, with a larger decline among high-income households. This differential response is mainly driven by price changes in the Transport category, which accounts for a larger share of expenditures among these households. By contrast, an oil price shock significantly increases cumulative inflation. Although the initial impact is stronger for high-income households, the effect becomes larger for low- and middle-income households as the shock propagates through other sectors, driven by a comparatively stronger price response in the Food category. Moreover, we document substantial dispersion in household-level inflation rates and show that low-income households experienced higher cumulative inflation than their high-income counterparts between 2009 and 2023. |
| Presented by: Felipe Martinez, Central Bank of Chile |
| Session 71: GOVERNMENT SPENDING AND GROWTH June 24, 2026 13:30 to 15:15 Location: B128 |
| Session Chair: Bent Sorensen, University of Houston |
Interactive Climate Effects on Economic GrowthAbstractTraditional economic models of climate change impacts rely on annual mean temperatures, overlooking crucial interactions between temperature and precipitation. We develop a dynamic panel model that incorporates these interactions to better capture the complex relationship between climate variables and economic growth. Our estimation shows that both temperature and precipitation exhibit statistically significant nonlinear effects on economic growth, with precipitation significantly influencing growth through its interaction with temperature. Based on a high-emission scenario SSP585, we project climate change impacts on economic growth until 2100. We find that the interaction between temperature and precipitation mitigate the damages induced by independent effects of temperature and precipitation and dampen the benefits from climate change. On a global level, economic production is projected to decline by approximately 61% by 2100 due to climate change, with interactive effects alleviating around 22% of the damages. |
| Presented by: Menghan Yuan, Nord University |
Cascading Transformation: Estimating the Effects of Green Government Spending through the Production NetworkAbstractA novel dataset is constructed to examine the dynamic employment effects of green and brown public procurement across the U.S. production network. Using detailed contract-level data, I estimate both direct and indirect impacts via local projections. A one standard-deviation green spending shock crowds in employment significantly more on impact than a brown spending shock. However, these direct effects quickly converge to zero. Employment in upstream sectors increase several quarters after the shock suggesting a substantial employment shift between recipient and supplying sectors. Downstream effects are small. Possible mechanisms lie in the duration of contracts, the type of contracts and the backward linkages of the sectors green and brown spending target. These findings highlight the value of a dynamic network perspective on heterogeneous government spending objectives and suggest that green procurement can produce significantly different effects than brown spending along the supply chain, underscoring the economic potential of green fiscal policies that simultaneously take sectoral interdependencies and potential worker transitions into consideration. |
| Presented by: Jan-Erik Thie, University of Potsdam, Macroeconomic Policy Institute |
Employer-to-employer Mobility and Wages in Europe and the United StatesAbstractI produce novel evidence on worker reallocation across employers and between employment and unemployment for several European countries over the past two decades. I construct a dataset of monthly transition rates by developing a novel approach to measure them using cross-sectional data from the European Union Labor Force Survey. Transition rates exhibit similar cyclical patterns across countries, but their levels are persistently different. I compute an indicator of the pace of worker reallocation up the job ladder, and find that it varies substantially across countries, is procyclical, and exhibits a systematic positive relationship with wage inflation. |
| Presented by: Daniel Borowczyk-Martins, Copenhagen Business School |
What Determines U.S. School Capital Spending? Long-Run Target and Short-Run Adjustment PathsAbstractWe examine determinants of long-run school capital spending in the United States. We find that in the long run---identified from co-integration analysis---the elasticity (of school capital spending per student) with respect to income per capita is above unity, implying that a district that is more than twice as wealthy as another, has twice as much school capital (per student). The elasticity with respect to enrollment is slightly negative and positive of similar magnitude with respect to population. We find (using error correction models) that adjustment to income stretches over decades while adjustment to enrollment and population takes place within five years. A similar analysis for current spending reveals very similar results, except adjustment to income shocks is much faster. We do not find positive impacts of capital spending shocks on income, population, or enrollment and we therefore interpret the results as casual impacts on capital and current spending. |
| Presented by: Bent Sorensen, University of Houston |
| Session 72: INEQUALITY AND IDENTITY June 24, 2026 13:30 to 15:15 Location: E002 |
| Session Chair: Sofiana Sinani, CERGE-EI |
Not by Bread Alone: Rethinking Labor-Market Adjustment after Economic ShocksAbstractEconomic shocks are typically evaluated through their effects on wages and employment. Yet such shocks not only redistribute income but also reorder occupational hierarchies in ways that measures of labor income and employment cannot capture. We argue that occupational social status – the rank implied by one’s occupation – therefore constitutes a distinct and consequential margin of labor-market adjustment. East Germany’s post-reunification economic transformation offers a unique setting to study the importance of social status. Despite sharp efficiency gains and rising wages, shifting comparative advantages moved East German workers from cognitive into manual occupations, thereby lowering aggregate occupational status. A model of occupational choice illustrates how this decoupling of wages and status can arise and how welfare assessments change once status enters the utility function. Empirically, we exploit quasi-experimental variation in firm liquidations using data on 6,608 firms linked to administrative labor-market histories for more than 200,000 workers. Applying the recentered instrumental-variable design of Borusyak and Hull (2023), we show that regions with higher- than-expected liquidation exposure experience faster worker reallocation, greater sectoral mobility, and smaller declines in occupational social status. Finally, we use voting behavior as a measure of revealed satisfaction with the economic transformation. Counties with larger status declines exhibit persistently higher support for far-right parties despite rising wages. This indicates that status trajectories shape labor-market experiences and well-being in ways that measures of labor-income and employment alone fail to capture. |
| Presented by: David Wittekopf, European University Institute |
Unequal Survival: The Effect of Mortality Decline on Income InequalityAbstractThis study investigates the impact of the substantial mortality declines observed globally since 1960 on income inequality within countries. Exploiting a novel source of exogenous variation in mortality, we find that lower mortality has significantly increased inequality. We identify two primary mechanisms driving this result. First, mortality declines have primarily benefited low-income individuals, increasing their relative share of the population without corresponding gains in productivity. This shift has directly contributed to rising inequality. Second, because health gains are marginal, some individuals survive in a state of limited functional independence and require ongoing care. This care is predominantly provided informally, reducing labor market engagement and indirectly affecting earnings among caregivers. Together, these mechanisms offer new insights into the structural drivers of inequality both within and across countries. |
| Presented by: Saeed Khodaverdian, University of Hamburg |
Faith, Interrupted: Identity and Behavior After Forced AtheismAbstractGovernments have long intervened in national identity formation, yet it remains unclear whether such efforts truly reshape values or primarily induce compliance. This paper studies Albania’s 1967 Cultural Revolution, when the Communist regime criminalized religion nationwide and promoted a secular “pure Albanian” identity. We exploit this reform as a nationwide shock to religiosity and examine its effects on public and private expressions of identity, values, and socioeconomic behavior. Combining administrative and survey data and applying a residualized event-study design, we show that religious names declined by 50–150%, while secular Albanian names increased by 50%. These shifts were largest in more illiterate districts, consistent with stronger exposure to state control, yet substantial resistance persisted: religious minorities were more likely to continue religious naming practices. Survey evidence reveals large declines in religious participation and upbringing among those born after the reform, but no change in belief in God or broader political values. This pattern suggests that the reform affected observable behavior and identity signaling, but no deeper ideological shifts. |
| Presented by: Sofiana Sinani, CERGE-EI |
| Session 73: INTERNATIONAL POLICY SPILLOVERS June 24, 2026 13:30 to 15:15 Location: B129 |
| Session Chair: Laura Kuitunen, European University Institute |
International spillovers of fiscal news shocksAbstractThis paper investigates the international transmission of U.S. fiscal news shocks, emphasizing the importance of the sentiment channel for the global economy. We identify these shocks using federal government spending forecasts from the Survey of Professional Forecasters. Employing the local projection method, we find that anticipated increases in U.S. spending raise U.S. activity and improve sentiment and financial conditions, while the U.S. dollar appreciates and the trade balance deteriorates. Positive U.S. fiscal news also boosts sentiment and eases financial conditions abroad, stimulating demand and output growth. However,in a broad sample, exchange-rate movements do not translate into higher net exports because stronger domestic demand raises imports; in contrast, for countries highly exposed to U.S. trade, the trade channel becomes significant while the financial channel weakens, and the sentiment channel remains important. Furthermore, spillover effects – like their domestic counterparts – are much stronger during U.S. recessions than during expansions. Finally, we show that spending announcements are substantially noisy. Using two-country DSGE model with imperfect information we show that removing this noise would amplify domestic and foreign effects by about 40 percent. |
| Presented by: Grzegorz Wesołowski, University of Warsaw |
Spillovers of U.S. Large-Scale Asset Purchases: The Role of External Corporate Bond CreditAbstractThis paper evaluates the effects of U.S. Large-Scale Asset Purchases (LSAPs) on emerging economies (EMEs), focusing on external corporate bond credit as a key transmission channel. To estimate these effects, I use a panel Bayesian VAR model with U.S. LSAP shocks identified by Swanson(2021), defining credit as the share of long-term, USD-denominated bonds in total external credit to the non-financial corporate (NFC) sector in EMEs. The main findings are: (i) a contractionary LSAP shock significantly reduces this ratio, coupled with financial and real economic strains in EMEs, and heightens global risk aversion; (ii) LSAP shocks explain approximately 40% of the variation in global risk aversion and 33% of the fluctuation in external corporate bond credit in EMEs; (iii) the external corporate bond credit ratio is a primary propagation mechanism for LSAP shock to real variables. When firms have limited access to financing alternatives beyond long-term, USD-denominated bonds, higher relative financing costs lead to a steeper decline and slower recovery in GDP and investment in EMEs; and (iv) small open advanced economies (SOAEs) with developed financial markets exhibit milder responses to LSAP shocks, with external corporate bond credit playing a negligible role. |
| Presented by: Francisca Torrealba, Banco de México |
Household Macroprudential Policies, Corporate Credit, and Resilience to Macro ShocksAbstractThis paper studies the effects of macroprudential policies on the resilience of the real sector to macroeconomic shocks. We use credit registry data from Turkey, exploiting (i) the adoption of a new macroprudential policy regime starting in 2011 that targeted household credit and (ii) the subsequent stress episode triggered by the sharp currency depreciation in 2018. Banks with higher exposure to macroprudential policies reduce household lending and reallocate credit toward firms, with positive effects on firm investment and employment. This increase in corporate lending is directed toward safer borrowers and enhances future resilience to macroeconomic shocks. Consistent with this mechanism, firms and provinces served by banks that engage in more prudent corporate lending following the 2011 reforms are relatively less affected during the 2018 currency depreciation, exhibiting smaller declines in investment and growth. Overall, our results suggest that macroprudential policies can generate positive spillovers across asset classes and strengthen real-sector resilience to macroeconomic shocks. |
| Presented by: Inci Gumus, Sabanci University |
The International Transmission of the Fed Information Effects:Evidence from EuropeAbstractWhen the Fed signals a booming US economy, part of Europe catches a cold - part of it thrives. This paper asks how an upswing in the US business cycle transmits to the Euro Area through trade. I study the question empirically through panel local projections that exploit EA industries' asymmetric input-output linkages with the US economy. These linkages prove decisive for the international propagation of US upturns, often obscured in aggregate responses. The analysis reveals asymmetric spillovers to the Euro Area from a US demand shock that can be interpreted as macroeconomic news released by the Federal Reserve. The good news about the US economy foretell a two-speed economy for Europe: industries that ultimately export to the United States move in step with the US business cycle, while those that import from the US contract as adverse global price and exchange rate movements pass the US demand shock downstream as a cost-push shock. Expenditure-switching effects contribute but cannot fully explain the observed pattern in US import demand. In contrast, monetary tightening by the Fed exerts contractionary effects on Euro Area aggregate activity with weaker dependence on trade linkages than for the Fed’s information effects. The empirical evidence motivate a theoretical two-country model that replicates the qualitative patterns observed in the data. |
| Presented by: Laura Kuitunen, European University Institute |
| Session 74: MACHINE LEARNING APPLICATIONS June 24, 2026 13:30 to 15:15 Location: D-115 |
| Session Chair: Sukanya Mukherjee, Institute of Economics Research in Halle (Saale) |
Mind the Gap: Gender Differences in Inflation ExpectationsAbstractA persistent gender gap in inflation expectations has been documented across countries and over time, with women systematically reporting higher expected inflation than men, yet its underlying causes remain under debate. This paper studies the gender gap using large-scale survey data from the ECB Consumer Expectations Survey and a double machine learning framework that allows for rich heterogeneity and valid inference in high-dimensional settings. The results show that the gender gap is not constant but varies systematically across individuals. Objective characteristics such as education and financial literacy play a limited role, whereas subjective factors, such as forecast confidence and uncertainty, as well as perceived economic vulnerability, emerge as key drivers of the gender gap. Therefore, differences in belief formation, rather than differences in information or exposure to price signals alone, are central to understanding the gender gap in inflation expectations. |
| Presented by: Winnie Coleman, Freie Universität Berlin |
Sample Selection in Unconditional Quantile ModelsAbstractWe propose the construction of a consistent estimator that addresses the problem of sample selection in unconditional quantile models. The proposed approach is based on three steps: (i) estimation of a control function using a logistic distribution regression; (ii) construction of a counterfactual distribution of the latent dependent variable conditional on the previously estimated control function; (iii) application of the recentered influence function (RIF) on the estimated counterfactual distribution and, finally, we run an ordinary least square regression. |
| Presented by: Stefanie Sunao, Insper |
Agree to Disagree: Robust Yield Prediction for Indian Farms using a weighted Multi-MLMs, Multi-Fold Framework, based on NSSO Survey DataAbstractThis paper proposes a multiple model framework that aggregates the predictions from Random Forest, XGBoost and Support Vector Regression, over repeated cross-validation, to produce robust farm-level estimates. The final ensemble point estimator achieves prediction accuracy that is in the upper-bound of candidate model performance, with $R^2=0.54$ and $R^2=0.59$ for median rice and wheat farms. Empirically, between-model divergence dominates within-model overfitting, so that high-performers are exponentially prioritized and the fold-level prediction variance is homoscedastic across candidate models. Additionally, quantile based conditional prediction intervals are used to identify farms with high tail-risk, and feature rankings highlight the relative importance of climate conditions, input costs and farm management practices. Together, this generates a farm-profile with the potential to inform agricultural policy and risk management. |
| Presented by: Sukanya Mukherjee, Institute of Economics Research in Halle (Saale) |
| Session 75: MONETARY POLICY June 24, 2026 13:30 to 15:15 Location: D-112 |
| Session Chair: Frederik Kurcz, DIW Berlin |
Rules vs. Discretion: Decoding FOMC Policy DeliberationsAbstractThis study provides evidence on the usage and preferences of Federal Reserves Federal Open Market Committee (FOMC) regarding the balance between rules and discretion in policy decisions. Analyzing FOMC transcripts over 40 years, we find that while Discretion has been a consistent feature in the language of the FOMC, the use of the language of Rules surged notably in the mid-1990s, aligning with theoretical advancements in monetary policy. We identify that a rise in Discretion terminology occurs during economic downturns and periods of heightened uncertainty. In contrast, a rise in the language of Rules is supported by higher references to terms such as credibility and commitment, and is more prevalent among hawkish FOMC members. Our findings link the increased use of the language of Rules (Discretion) to tighter (easier) monetary policy, revealing a significant role of this debate in shaping policy outcomes, in particular periods. |
| Presented by: Klodiana Istrefi, European Central Bank |
Do Monetary Policy Shocks Affect the Neutral Rate of Interest?AbstractWe develop a Trend-Cycle Bayesian VAR that jointly estimates the real neutral rate of interest, $r_t^*$, and identifies monetary policy shocks. A key innovation is that the framework allows cyclical shocks, most notably monetary policy shocks, to affect the trend component of macroeconomic variables, providing a new way to assess whether transitory disturbances have persistent effects. Using external instruments, we find that contractionary monetary policy shocks reduce $r_t^*$ and lower trend GDP growth, while the model's estimates of $r_t^*$ remain consistent with standard benchmark measures. We then quantify the contribution of monetary policy shocks to the secular decline in $r_t^*$. Although these shocks at times generate sizable movements in $r_t^*$, their contribution to the long-run decline is modest, and their net effect on $r_t^*$ since the early 1990s is slightly positive. We complement these findings with cross-country evidence from other advanced economies, pointing to similar effects. |
| Presented by: Luis Uzeda, Bank of Canada |
Predictable Forecast Errors in Full-Information Rational Expectations Models with Regime ShiftsAbstractThis paper shows that regime shifts in Full-Information Rational Expectations (FIRE) models generate predictable, regime-dependent forecast errors. The key mechanism is that under FIRE, agents form expectations as a weighted average of regime-conditional forecasts, so that ex-post forecast errors depend systematically on the realized sequence of regimes relative to agents' expectations---even in relatively large samples. Hence, forecast error predictability alone is neither sufficient to reject FIRE nor informative about alternative expectations theories. We propose a regime-robust test of FIRE for models with regime shifts. Applying the test to an estimated New Keynesian model with regime shifts in monetary policy, we find that the baseline test rejects FIRE for the 1970s and the post-2020 period but not for intermediate subperiods. The rejection for the 1970s disappears when regime transition probabilities are re-estimated for each subperiod using observed SPF inflation expectations, which allows for changes in perceived monetary policy credibility. The interpretation is that agents who rationally anticipated an eventual shift to hawkish policy systematically under-predicted inflation during a dovish regime that turned out to be more persistent than expected. The post-2020 rejection persists, providing empirical motivation to consider richer regime structures or alternative theories of expectations formation for this episode. |
| Presented by: André Kurmann, Drexel University |
Quantifying the Fiscal Channel of Monetary PolicyAbstractIn business-cycle models the effects of monetary policy depend on the fiscal reaction to interest rate changes. This paper investigates the fiscal reaction by presenting new evidence on the effects of U.S. monetary policy on fiscal policy instruments. Subsequently, it estimates a Heterogeneous Agent New Keynesian model with flexible fiscal feedback rules to match and interpret the empirical results. I find that U.S. fiscal policy responds to monetary-induced output contractions with debt-financed, countercyclical tax and transfer policies, amid a gradual decline in spending to accommodate the debt increase. The model implies that monetary policy unopposed by a business-cycle stabilization motive of fiscal policy would be roughly one-third more contractionary. As a result, the fiscal channel renders the effects of monetary policy state-dependent on the fiscal capacity for stabilization policy. |
| Presented by: Frederik Kurcz, DIW Berlin |
| Session 76: NETWORK INTERFERENCE June 24, 2026 13:30 to 15:15 Location: B008 |
| Session Chair: Duong Trinh, University of Graz |
Mediated InterferenceAbstractI develop identification and estimation theory for identifying causal effects in network experiments where treatment induces network changes. Under a network formation model—where post-treatment connections depend on pre-treatment links, treatments, and independent shocks—I decompose the total effect into direct, mediated indirect, interaction, and arm selection components. I show identification of these effects is possible under a strong exogeneity assumption: that unobservables affecting network formation are independent of unobservables affecting outcomes. I propose doubly-robust estimators that achieve $\sqrt{n}$-consistency without cross-fitting by exploiting neighborhood stability—a condition the structural model implies automatically—and develop network HAC variance estimators for valid inference. Applications to a field experiment on tax compliance in Austria demonstrate that the treatment effects mediated through networks can be substantially smaller than naive regression estimates of peer influence suggest. |
| Presented by: Meng Hsuan Hsieh, University of Michigan |
Fixed-Population Causal Inference for Models of EquilibriumAbstractIn contrast to problems of interference in (exogenous) treatments, models of interference in unit-specific (endogenous) outcomes do not usually produce a reduced-form representation where outcomes depend on other units' treatment status only at a short network distance, or only through a known exposure mapping. This remains true if the structural mechanism depends on outcomes of peers only at a short network distance, or through a known exposure mapping. In this paper, we first define causal estimands that are identified and estimable from a single experiment on the network under minimal assumptions on the structure of interference, and which represent average partial causal responses which generally vary with other global features of the realized assignment. Under a fixed-population, design-based approach, we show unbiasedness and consistency for inverse-probability weighting (IPW) estimators for those causal parameters from a randomized experiment on a single network. We also analyze more closely the case of marginal interventions in a model of equilibrium with smooth response functions where we can recover LATE-type weighted averages of derivatives of those response functions. Under additional structural assumptions, these ``agnostic" causal estimands can be combined to recover model parameters, but also retain their less restrictive causal interpretation. |
| Presented by: Konrad Menzel, New York University |
Regression Discontinuity Designs Under InterferenceAbstractWe extend the continuity-based framework to Regression Discontinuity Designs (RDDs) to identify and estimate causal effects under interference when units are connected through a network. Assignment to an "effective treatment," combining the individual treatment and a summary of neighbors' treatments, is determined by the unit's score and those of interfering units, yielding a multiscore RDD with complex, multidimensional boundaries. We characterize these boundaries and derive assumptions to identify boundary causal effects. We develop a distance-based nonparametric estimator and establish its asymptotic properties under restrictions on the network degree distribution. We show that while direct effects converge at the standard rate, the rate for indirect effects depends on the number of scores fixed at the cutoff. Finally, we propose a variance estimator accounting for network correlation and apply our method to PROGRESA data to estimate the direct and indirect effects of cash transfers on school attendance. |
| Presented by: Tiziano Arduini, University of Rome Tor Vergata |
Heterogeneous Peer Effects with Endogenous Network FormationAbstractThis paper introduces a new econometric framework for modeling social interactions with heterogeneous responses to peers while addressing the endogenous formation of links. Our proposed Selection-corrected Heterogeneous Spatial Autoregressive (SCHSAR) model overcomes the limitations inherent in the spatial autoregressive (SAR) specification by achieving these dual objectives. This unified framework jointly models link formation and outcome determination. We incorporate a finite mixture structure to capture rich heterogeneity in peer effects and explicitly account for unobserved individual-specific factors driving both network formation and outcome equations, thereby addressing the network endogeneity for credible estimation of heterogeneous spillover effects. Standard likelihood-based methods are not well suited for the two-stage problem; therefore, for estimation and inference, we propose a fully Bayesian data augmentation approach to handle computational challenges and complex latent structure. We present a simulation study that validates our proposed approach. Our empirical application to an innovation network among U.S. firms reveals significant positive, yet heterogeneous, peer effects on corporate R&D investments, after accounting for endogenous network formation. The findings highlight different firm behaviors and uncover notable transmitters and absorbers in response to exogenous R&D policy shocks. This framework enables quantification of firm-level direct and spillover effects, providing valuable insights for evidence-based and targeted policy design. |
| Presented by: Duong Trinh, University of Graz |
| Session 77: REGIME SWITCHING MODELS June 24, 2026 13:30 to 15:15 Location: D-114 |
| Session Chair: Juan Reyes, King's College London |
When Does Sentiment Predict Volatility?AbstractCan media sentiment improve volatility forecasts? We study this question using daily realised volatility for DJIA stocks and media-based sentiment measures from the Thomson Reuters MarketPsych Indices. Augmenting standard HAR models with sentiment yields only modest average improvements, consistent with prior literature. However, this average masks strong state dependence. We show that sentiment becomes materially more informative in the upper tail of the volatility distribution and, more importantly, during periods when volatility persistence weakens. Using a fractional integration framework to measure persistence, we find that sentiment-based predictors deliver their largest gains precisely when lagged volatility becomes less informative. Out-of-sample, these gains are concentrated in regimes characterised jointly by high volatility and low persistence. Our findings suggest that media sentiment does not provide a generic forecasting enhancement, but instead supplies forward-looking narrative information when standard price-based volatility models are least reliable. |
| Presented by: Frances Liu, University of Technology Sydney |
Estimation of the Heuristic Switching Model for the learning-to-forecast experimentsAbstractWe estimate a hidden Markov model of switching between multiple forecasting heuristics using data from “learning-to-forecast” experiments. In these experiments, participants predict the evolution of an endogenous financial asset price. The realized price is determined by a function of the average forecast across participants, with the functional form varying across experimental treatments. Because individuals may condition their forecasts on past prices, expectations and realizations are linked through a feedback loop. We adopt a Bayesian approach to estimate a heuristic-switching model in which individual point forecasts are generated by latent behavioral rules, such as adaptive and trend-following heuristics. These heuristics are unobserved, and individuals are allowed to switch between them over time. Switching probabilities depend on the past relative performance of the forecasting rules. The key observed variable is the individual numerical forecast in each experimental round. Our estimation procedure jointly identifies the structural parameters of the forecasting heuristics and the parameters governing rule selection. In addition, we recover the sequence of hidden states corresponding to the heuristics employed by each participant. The results reveal treatment-dependent heterogeneity in heuristic prevalence: different functional forms of price determination lead to distinct patterns of rule usage and switching behavior. These findings provide new evidence on expectation formation and adaptive learning in environments with endogenous feedback. |
| Presented by: Valentyn Panchenko, University of New South Wales |
Multi-regime endogenous switching model: the individual economic return of internal migration in ItalyAbstractCompared to the past, internal migration in Italy is characterized by flows of more qualified and educated migrants, with increased opportunities to choose between multiple destination areas. This involves a polychotomous sorting process in migration decision that makes more difficult to correct the estimation of the economic returns of migration for the endogeneity of the migration’s choice. In this study, we leverage on the proposal of a new maximum likelihood estimator to identify the parameters of a three-regime endogenous switching model, in which each regime corresponds to a specific migratory choice. We find that, as a result of the decision-making process, the economic return of internal migrants in Italy is higher if they move to medium-urbanized areas than to highly urbanized ones. |
| Presented by: Laura Magazzini, Scuola Superiore Sant'Anna Pisa |
The Macroeconomic Effects of AI UncertaintyAbstractThis paper examines the macroeconomic implications of uncertainty shocks related to artificial intelligence (AI). I construct a novel text-based AI Uncertainty (AIU) Index from newspaper coverage that displays sharp increases around notable AI developments. The index also demonstrates limited correlation with established measures of economic uncertainty. Using SVAR-IV with an instrument that orthogonalises first- and second-moment coverage of AI, I find that positive AI uncertainty shocks generate significant contractionary effects on equity prices, hours worked, and wages, with smaller and less persistent effects on employment and output. Industry-level estimates, in turn, highlight heterogeneous adjustments along both the labour quantity and price margins. These findings indicate that AI uncertainty is a distinct source of economic fluctuations, with a response pattern that differs from that of conventional uncertainty shocks. |
| Presented by: Juan Reyes, King's College London |
| Session 78: SCORE-DRIVEN MODELS June 24, 2026 13:30 to 15:15 Location: D-113 |
| Session Chair: Dennis Umlandt, University of Innsbruck |
Self-driving neural networks for yield curve modelingAbstractWe propose a factor model with time-varying loadings for term structure modeling and forecasting. While maintaining the interpretation of the factors as level, slope, and curvature through explicit identification restrictions, we allow the loadings to take flexible shapes by specifying them as neural networks that evolve over time using a ``self-driving'' updating scheme based on past forecast errors, with gradient scaling to improve robustness. Using an empirically calibrated simulation study and an application to U.S. Treasury yields across 24 maturities, we show that flexible and dynamic factor loadings improve forecasting performance relative to standard benchmarks, including Nelson–Siegel models and the random walk. The gains are strongest at medium maturities and shorter forecast horizons, highlighting the importance of capturing curvature dynamics. In-sample results further illustrate how time-varying loadings provide insight into changes in yield curve shape beyond traditional parametric specifications. |
| Presented by: Sicco Kooiker, Vrije Universiteit Amsterdam |
From Rotational to Scalar Invariance: Enhancing Identifiability in Score-Driven Factor ModelsAbstractWe show that, for a certain class of scaling matrices including the inverse square-root of the conditional Fisher Information, score-driven factor models are identifiable up to a multiplicative scalar constant under very mild restrictions. This result has no analogue in parameter-driven models, as it exploits the different structure of the score-driven factor dynamics. Consequently, score-driven factor models overcome the issue of rotational invariance that typically affects dynamic factor models, thereby enhancing the economic and financial interpretability of the estimated factors. Our restrictions are order-invariant and can be generalized to score-driven factor models with dynamic loadings and nonlinear factor models. We test extensively the identification strategy using simulated and real data. The empirical analysis on financial and macroeconomic data reveals a substantial increase of log likelihood ratios and significantly improved out-of-sample forecast performance when switching from the classical restrictions adopted in the literature to our more flexible specifications. |
| Presented by: Emilija Dzuverovic, Ca’ Foscari University of Venice |
An Extended Score-Driven Dynamic Factor Model: Recovering Composite Indicators from the PandemicAbstractWe propose an extended score-driven (ESD) dynamic factor model (DFM) that accommodates non-Gaussian innovations, nonlinear factor dynamics, and time-varying volatility. The main novelty of our model is a state equation that includes both lagged and contemporaneous scores, implying that factors are not predetermined. We show that this novel model nests both the classic (parameter-driven) state-space DFM as well as the more recent score-driven DFM, bridging the gap between these two model classes. Empirically, our ESD-DFM proves useful for working with COVID-19–era observations, which have posed substantial challenges for macroeconomic modeling. For instance, while the Federal Reserve Bank of Philadelphia suspended publication of its leading index due to pandemic-related anomalies, our model remains robust to such extreme observations and enables reliable computation of the index. We further apply the ESD-DFM to The Conference Board’s (TCB's) Coincident and Leading Economic Indices (CEI and LEI). When indices are constructed from the estimated factors, the unprecedented divergence between TCB's CEI and LEI observed during the post-pandemic period disappears: although the reconstructed LEI declines in 2022, it resumes an upward trajectory from late 2023 through mid-2025. |
| Presented by: Evgenii Vladimirov, Erasmus University Rotterdam |
An Observation-Driven Framework for Dynamic Reduced-Rank RegressionAbstractReduced-rank structures arise naturally in many economic and financial models, yet empirical implementations typically treat them as static. This paper introduces the Generalized Autoregressive Reduced-Rank Regression (GARRR) framework, which allows reduced-rank coefficient matrices to evolve dynamically according to score-driven updating rules while preserving parsimony and interpretability in a likelihood-based setting. We establish theoretical properties of the model, including stationarity, invertibility, and asymptotic normality of maximum likelihood estimators under both statistical and economically motivated identification schemes. An application to asset pricing based on the Arbitrage Pricing Theory (APT) shows substantial time variation in reduced-rank pricing relations and demonstrates that allowing factor loadings to vary over time leads to sizable improvements in pricing performance relative to static specifications. |
| Presented by: Dennis Umlandt, University of Innsbruck |
| Session 79: Coffee break June 24, 2026 15:15 to 15:45 |
| Session 80: AI AND LABOR MARKETS June 24, 2026 15:45 to 17:30 Location: D-115 |
| Session Chair: Anel Kaliyeva, Université Côte d'Azur |
Information Frictions in Student Loan Markets: Evidence from ChatBot Experiments at ScaleAbstractPublic student loan programs often feature complex eligibility rules, so students may need to apply to learn whether they qualify. When applying is costly, eligibility uncertainty can deter even qualified students. We study this friction in a randomized trial with 100 thousand Colombian high-school seniors offered an automated WhatsApp chatbot providing student-loan information. Treatment arms isolate generic information, personalized eligibility guidance, and personalized repayment projections. Personalized eligibility guidance increases loan applications by 38% and raises tertiary enrollment by 8% among students who engage with the chatbot, while generic information does not have enrollment effects. Adding repayment projections increases application activity but does not increase enrollment. Approval rates conditional on applying are unchanged, indicating that the intervention operates through the extensive margin of application rather than lender-side screening. The results highlight eligibility uncertainty as a first-order barrier to credit access and higher education. |
| Presented by: Christian Posso, Banco de la República |
Labor Supply Response to Income Tax Information: Divergence between Stated and Revealed PreferencesAbstractComplex tax incentives, such as means-tested tax transfers, are known to distort labor supply decisions. This study conducts a randomized experiment to examine whether providing information about income taxation induces individuals to change their labor supply. The results show that tax information provision increases stated labor supply by an average of 1.2%, and raises the probability of planning to earn above the threshold by 4.3%. However, this increase in stated intentions does not translate into actual labor supply, as revealed by the end-of-year follow-up survey. The findings suggest that while information can correct misconceptions and shift intentions, psychological frictions may limit the effectiveness of such interventions in changing actual behavior. |
| Presented by: Jun Takahashi, Yokohama City University |
Measuring Firm Engagement with Employee Well-Being and Mental Health using Corporate DisclosuresAbstractWe develop a scalable, replicable measure of firms’ engagement with employee well-being from corporate disclosures. Using the universe of Danish firms’ annual reports linked to population-wide matched employer–employee administrative data, we construct a text-based indicator that captures a latent dimension of workplace environment. Our measure varies substantially within industries and is largely orthogonal to standard proxies for firm structure and workforce composition. We further show that higher engagement is systematically associated with greater mental-health-related health-care utilization and health-related work absence among employees. Our approach illustrates how corporate text can recover economically meaningful firm-level soft information and is readily portable to other countries, corpora, and thematic domains. |
| Presented by: Riccardo Di Francesco, University of Southern Denmark |
Adoption of AI and latent demand for ICT skillsAbstractThe objective of this paper is to quantify the adjustments required for agents to achieve market participation. To this end, we develop a statistical framework in which participation is governed by agent-specific latent cutoffs drawn from a parametric distribution. The framework is implemented in three stages. First, candidate threshold distributions are estimated and selected via maximum likelihood. Second, the parameters of the distribution are allowed to depend on observable characteristics, thereby yielding agent-specific expected thresholds. Finally, the estimated threshold function is inverted, using a total differential, to quantify the covariate adjustments required to cross the participation cutoff. The framework is applied to firm-level AI adoption in France. The results indicate that most firms operate below their estimated workforce threshold for adoption. Translating these adoption gaps into ICT labor adjustments, the implied demand for ICT engineers is found to be highly sensitive to complementary investments in digital skills and cloud infrastructure. Under a medium-term scenario, the implied demand amounts to approximately 25,000 ICT engineers. |
| Presented by: Anel Kaliyeva, Université Côte d'Azur |
| Session 81: BOND MARKETS AND PORTFOLIOS June 24, 2026 15:45 to 17:30 Location: D-113 |
| Session Chair: Juri Marcucci, Bank of Italy |
Clearing Markets and Client Clearing ServicesAbstractThe global shift toward central clearing accelerated after the 2007–09 financial crisis. While prior research has focused primarily on clearing member intermediaries, this paper examines client clearing, which now accounts for the majority of risk managed in centrally cleared markets. Using confidential transaction-level data from the credit default swap market, we show that client clearing enhances netting efficiency for member dealers and generates pricing advantages for clients. Adoption of clearing leads clients to expand their dealer networks and reduce counterparty concentration, thereby improving market access and competition. Clients, who rely on clearing member firms to utilize central counterparties, turn to members that are less risky and with whom they have established trading relationships to facilitate access. Offering these services has spillover benefits for member firms’ dealer arms, strengthening client retention and pricing power. Clients’ dependence on clearing members creates operational fragilities under stress, especially for those with limited member relationships. Our findings offer new insights into the economics of client clearing and are particularly relevant in light of recent clearing mandates, most notably in U.S. Treasury markets. |
| Presented by: Robin Lumsdaine, Kogod School of Business, American University |
Extracting Long-Term Market Expectations from Government Bond YieldsAbstractStandard affine term structure models imply little time variation in long-horizon risk-free forward rates, as short-rate expectations converge to a sample-dependent mean. We propose a model with a market-based, time-varying endpoint, in which OIS forward rates act as an unspanned factor anchoring long-horizon expectations. We impose that beyond roughly 10 years the risk-free exposure to this proxy exceeds that to the first yield factor. Applied to US Treasuries, German Bunds, and UK Gilts, the model yields economically meaningful risk-free rates and term premia out to 30 years. We demonstrate the usefulness of our proposed term structure model by using it to forecast debt-servicing costs. First, the excess forward premium is removed from forward rates, sharply reducing forecast bias for future yields. Then, using March 2026 estimates from the model, it is shown that adjusting current forward rates lowers projected debt servicing costs. This is especially the case for governments that issue a large proportion of debt at short maturities and where yield curves are steeply upward sloping. |
| Presented by: Ana Beatriz Galvao, Bloomberg Economics, U of Warwick |
Risk-Budgeted Mean-Variance PortfoliosAbstractWe introduce the Risk-Budgeted Mean-Variance (RBMV) portfolio, a novel framework that connects the classical Markowitz mean-variance problem and the risk budgeting approach. By modifying the risk budgeting optimization problem to include constraints on expected returns and volatility , RBMV offers a disciplined way to manage the trade-off between risk concentration and return maximization. The investor gains a lever to adjust how close the portfolio sits to either framework, depending on her preferences. We show that the optimization problem that defines the RBMV portfolio is convex, efficiently computable, and typically delivers competitive returns with reduced risk concentration in the context of long-only portfolios. We illustrate our method- ology using daily equity returns from the U.S. and show that our methodology efficiently controls the volatility of returns while also delivering Sharpe ratios that are consistently higher than the traditional mean-variance approach. |
| Presented by: Raul Riva, Brazilian School of Economics and Finance |
A European Safe Asset? Not Without the InvestorsAbstractWe study bonds issued by the European Union (EU) as joint and several liabilities of its member countries and show that they pay higher interest rates than comparably safe and large sovereign issuers. The spread reflects their greater sensitivity to adverse market shocks, which becomes particularly pronounced during periods of monetary tightening. Using novel data, we document that EU bonds have a small investor base because they are excluded from major fixed-income indices due to their lack of formal sovereign status. This exclusion lowers expected prices during crises, making EU bonds unattractive to investors with liquidity needs, such as mutual funds and foreign central banks. Expectations of state-contingent purchases by the European Central Bank (ECB) can substantially compress this premium even when not directed at EU bonds. A demand-based asset pricing framework suggests that the spread would be negligible if the EU were recognized as a fully sovereign issuer and a new safe asset would arise. |
| Presented by: Juri Marcucci, Bank of Italy |
| Session 82: CHILD DEVELOPMENT AND EDUCATION June 24, 2026 15:45 to 17:30 Location: E002 |
| Session Chair: Henrike Alm, Justus-Liebig-Universität Gießen |
Conduct and Consequences: Behavioral Rank and Academic OutcomesAbstractWe document a previously unknown mechanism in the education production function: Ordinal behavioral rank exerts a lasting causal influence on student behavior and subsequent educational outcomes. Leveraging variation in student disruptiveness distributions across more than 250 high-school classrooms, we estimate the effect of a student’s disruptiveness rank in high school, holding absolute disruptiveness constant. A higher disruptive rank sharply increases later disruptive behavior, reduces academic achievement, and lowers university admission rates---even though high-school teachers and peers have no information about students' prior disruptive rank. These findings reveal that rank-based identity formation and behavioral expectations persist across educational transitions, generating long-term impacts on both discipline and human capital. |
| Presented by: Tommaso Sartori, Monash University |
Subjective Beliefs and Anchoring Bias in Childhood: Experimental Evidence from Flemish ClassroomsAbstractThis paper examines the impact of anchoring, a cognitive bias by which individuals rely too heavily on an initial piece of information when making judgements, on primary school students’ expectations in Flanders, Belgium. Using a randomized controlled trial, students exposed to a low, unrelated numerical prompt (the anchor), believed that they could recall, on average, two fewer words out of a total of 30 than their peers in the control group, corresponding to a decline of approximately 7%. This effect was particularly pronounced among students with high socioeconomic status, low academic achievement, negative perceptions of school, as well as amongst high achieving students. The paper proposes a theoretical model is proposed, whereby anchoring operates through students’ subjective beliefs about the likelihood of being correct, conditional on exposure to an anchor, or incorrect in their predictions. Once this belief-based mechanism is accounted for, the anchoring effect disappears. Understanding how and when such cognitive biases emerge is crucial for the design of effective educational interventions, especially in early stages of learning when decision-making habits are still forming. Overall, the findings highlight the relevance of behavioural insights for education policy and point to the potential of tailored nudges to improve outcomes in school settings. |
| Presented by: Diogo Vieira Nunes da Conceição, KU Leuven |
From Home to School: The Changing Dynamics of Skill FormationAbstractThis paper investigates the joint roles of home investments and school quality in children’s cognitive and socio-emotional development during middle childhood (ages 5–11). Using longitudinal data from the UK Millennium Cohort Study linked to the National Pupil Database, we estimate a multistage latent-factor production function. The model allows both home and school inputs to be endogenous and permits their productivities to vary with age. We find that cognitive skill formation transitions from being primarily home-driven at ages 5–7 to more school-driven at ages 7–11: the marginal productivity of parental investments declines with age, while school quality becomes a significant input. Socio-emotional skills are highly persistent and comparatively less responsive to either input. Counterfactual simulations show that early increases in disadvantaged families’ investments yield the largest reductions in cognitive gaps, whereas equalizing school quality generates smaller gains once inequality has accumulated. |
| Presented by: Qianyao Ye, Xiamen University |
Beyond Cigarettes: The Impact of School Smoking Bans on Smoking Behavior, Perception, and Substance Use SpilloverAbstractThis paper examines the broader effects of school smoking bans on adolescent substance use and perceptions. Exploiting staggered variation in the introduction of school smoking bans across federal states in Germany, we estimate causal effects using a difference-in-differences design, and we show that school smoking bans reduce smoking and also lower alcohol and cannabis consumption. The spillover effects are strongest in social settings involving friends, highlighting the role of peer environments. We find no evidence that school smoking bans change overall attitudes toward smoking or alcohol, although smoking perceptions become more positive among remaining smokers. This pattern suggests that the school smoking ban works mainly by limiting opportunities for substance use and reshaping peer interactions, rather than by changing health related beliefs. |
| Presented by: Henrike Alm, Justus-Liebig-Universität Gießen |
| Session 83: CLIMATE AND THE MACROECONOMY 1 June 24, 2026 15:45 to 17:30 Location: B128 |
| Session Chair: Alejandro Puerta-Cuartas, UC3M |
Climate Growth-at-RiskAbstractWe study how temperature shocks reshape the entire conditional distribution of GDP per capita. Using a global panel of countries, we separately identify local and global temperature shocks and trace their dynamic effects with panel quantile local projections. Local and global warming reduce output for up to a decade, with global shocks generating substantially larger losses. The central finding is that both shocks primarily erode upside potential: the 90th percentile of future growth declines more than the median, while the left tail moves comparatively little. This pattern implies that climate risk operates less through heightened downside risk and more through a reduced likelihood and magnitude of high-growth outcomes. We link these distributional results to persistent declines in capital accumulation and productivity, together with trade contractions under global shocks, and to sustained shortfalls in sectoral value added. Local warming is most damaging in low-income and hot economies; global shocks also weaken upside potential in cooler climates. |
| Presented by: Damiano Di Francesco, Scuola Superiore Sant'Anna Pisa |
Measuring climate costsAbstractWe study how climate shocks, by impacting firm performance, affect the macroeconomy. Using firm-level data from two small open economies exposed to significant physical risks (Portugal and Ireland), alongside granular ERA5 climate data, and a dynamic general equilibrium model with heterogeneous firms, we provide evidence that physical climate shocks affect firms through three distinct channels: a reduction in their productivity, an acceleration of capital depreciation, and the destruction of human capital. Destructive shocks in both economies depress sales and capital immediately, while employment adjusts more gradually. Using the model, we quantify the cost of climate shocks in consumption-equivalent welfare terms. |
| Presented by: Laszlo Tetenyi, Banco de Portugal |
Temperature shocks, inflation and households’ inflation expectations in the euro areaAbstractGlobal warming is intensifying. The year 2024 marked the warmest year in over a century and 2025 was the third hottest year. These developments have renewed interest in understanding the macroeconomic consequences of climate change for prices, expectations and economic activity. We empirically show that unexpected positive deviations of local temperatures from historical means lead to statistically significant and persistent increases in aggregate inflation, driven primarily by the energy component. Consistent with this pattern, households revise their inflation expectations upwards, especially short-term ones, while medium-term expectations respond less, suggesting stronger anchoring. At the same time, following a positive temperature anomaly, households’ expectations about economic activity worsen, suggesting that households expect such events to act like adverse supply-side shocks. Importantly, we uncover substantial heterogeneity: lower-income and less-educated households’ expectations exhibit significantly stronger responses. |
| Presented by: Guido Bulligan, Banca d'Italia |
An Inferential Framework for Climate-Induced Poverty VulnerabilityAbstractWith the increase in the frequency and severity of climatic shocks, quantifying their impact on vulnerability to poverty has gained significant attention. This paper formalizes a simulation approach to climate vulnerability for the design of targeted, place-based public policies to mitigate climate risk. We propose using SHapley Additive exPlanation values to characterize the most vulnerable to climate shocks and estimate the heterogeneous impact of specific climate shocks on poverty vulnerability. We illustrate our approach empirically by considering Ecuador, a biodiverse country with high exposure to climate risk and vulnerability. Our analysis reveals that the vulnerable are mostly individuals informally employed in the primary sector and living in rural areas located in the Amazonian Region, which motivates the implementation of a targeted place-based formalization policy. While household characteristics account for most of the cross-sectional variation in vulnerability, climate shocks act as triggering events that can push marginally vulnerable households into poverty, with effects that are highly localized and heterogeneous across space. |
| Presented by: Alejandro Puerta-Cuartas, UC3M |
| Session 84: DISTRIBUTIONAL INFLATION June 24, 2026 15:45 to 17:30 Location: B009 |
| Session Chair: Marco Del Negro, Federal Reserve Bank of New York |
Ethical Indices for Inflation and PricesAbstractDifferential impact of price movements on different income groups is examined. Different demographic groups face different commodity bundles and choice sets. We examine construction of group specific price indices, and the overall aggregation of these indices. The latter exercise faces the ethical welfare-theoretic aggregation challenge as inequality measures. We expand on this and propose transparent ”optimal” solutions, as Ethical Price Indices (EPI). Construction and dynamic movement of the corresponding ”real incomes”, and inflation rates is also analyzed. |
| Presented by: Esfandiar Maasoumi, Emory |
Global Inflation SpilloversAbstractThis paper develops a multi-country open economy model with heterogeneous price-setting behavior and incomplete exchange rate pass-through to provide theoretical foundations for analyzing global inflation spillovers. The model predicts that inflation spillovers between country pairs follow a gravity structure: they decrease with geographic distance and increase with bilateral trade intensity and source-country economic size. Using the Diebold-Yilmaz methodology, I analyze inflation spillovers across a large set of countries from 1970 to 2023. The recent global inflation surge significantly increased spillovers, especially among developed economies and from developed to developing countries. The inflation spillover index has increased alongside international trade and now exceeds levels obtained in the late 1970s and early 1980s. Gravity equation regressions reveal that inflation spillovers between pairs of countries decrease as geographic distance increases, as predicted by theory. Moreover, inflation spillovers are more sensitive to the economic size of the source country than the target country, especially since the 1990s. |
| Presented by: Kamil Yilmaz, Koc University |
Money Supply, Pure Monetary Policy and Fiscal Shocks in Inflation Dynamics: A Regime-Switching AnalysisAbstractThis paper investigates the drivers of U.S. inflation by estimating a threshold VAR model with regime-dependent dynamics defined by money supply growth and the fiscal balance. The framework distinguishes between four policy regimes—No Dominance, Monetary Dominance, Fiscal Dominance, and Joint Dominance—and identifies five structural shocks: aggregate supply, money supply, fiscal, monetary policy, and residual demand (non-policy). Results show that money supply shocks are the most persistent drivers of inflation when money growth is high, particularly when combined with large deficits. Excluding M2 growth from the model leads to inflation persistence being absorbed by other demand-side shocks, suggesting that money supply dynamics are often an omitted but influential factor in inflation models. Historical decompositions indicate that while inflation in the 1970s was driven by monetary expansions and aggregate supply shocks, the post-2020 surge was primarily fueled by demandside shocks—including fiscal stimulus and delayed policy tightening. These findings underscore the importance of accounting for monetary-fiscal regime interactions and explicitly modeling money supply dynamics to accurately characterize inflation persistence. |
| Presented by: Clemente Pinilla-Torremocha, Bank of England, and European Research University |
Tradeoffs for the poor, divine coincidence for the richAbstractWe use an estimated medium-scale HANK model to investigate how the tradeoff between stabilizing inflation and consumption volatility varies for households with different levels of wealth. Consumption for the rich is mostly affected by demand shocks via their exposure to highly procyclical profits---for them, stabilizing consumption and inflation coincide. The poor are more vulnerable to supply shocks, hence aggressively stabilizing inflation is costly in terms of their consumption volatility. While they dislike inflation because it erodes real wages, they are hurt even more by an aggressive monetary policy response to inflation, which reduces real wages further while increasing unemployment. |
| Presented by: Marco Del Negro, Federal Reserve Bank of New York |
| Session 85: HIGH-DIMENSIONAL VOLATILITY MODELS June 24, 2026 15:45 to 17:30 Location: D-110 |
| Session Chair: Carsten Chong, HKUST |
The Vector Conditional Autoregressive Wishart Model for Multivariate Stock Market VolatilityAbstractThe multivariate GARCH approach of Bollerslev, Engle and Wooldridge (1988), which in its vectorized formulation is known as the VEC model, is the most exible multivariate time series model for conditional covariance matrices. However, estimation of VEC models based on daily returns of risky assets is computationally challenging so that their unrestricted estimation is feasible only for settings with a quite small number of assets. In this paper we suggest a novel VEC Conditional Autoregressive Wishart (VEC-CAW) model which is based on daily realized covariance matrices computed from high-frequency intraday returns. We analyze properties of this VEC-CAW and focus on its estimation. In order to make maximum likelihood (ML) estimation of the realized VEC-CAW model computationally feasible, we derive the analytical expression for the gradient of the log-likelihood under the conditional Wishart assumption for realized covariance matrices. Constrained optimization guaranteeing stationarity and positive definiteness proceeds using Bregman divergences. Doing so we successfully conduct unrestricted ML estimation of the VEC-CAW model with reasonable computation time. Then finite sample properties of the ML estimator are investigated in a Monte Carlo study with up to 10 assets. Further, we estimate an unrestricted VEC-CAW in an empirical illustration. |
| Presented by: Jan Vogler, Ruhr University Bochum |
Practical estimation methods for high-dimensional multivariate stochastic volatility modelsAbstractWe propose computationally inexpensive and efficient estimators for multivariate stochastic volatility (MSV) models with cross-dependence, Granger causality, and higher-order persistence in latent volatilities. The proposed class of estimators is based on a few moment equations derived from the VARMA representations of MSV models. Except for cross-dependence parameters, closed-form expressions for the other parameters are derived where no numerical optimization procedure or choice of initial parameter values is required. To increase the stability and efficiency of volatility persistence parameter estimates, we suggest shrinkage-type VARMA estimators where averaging or matrixvariate regression (MVR) is employed. We derive the asymptotic distribution of these estimators. Due to their computational simplicity, VARMA estimators allow one to make reliable – even exact – simulation-based inferences by applying Monte Carlo test techniques. In empirically realistic setups, simulation results show that the proposed shrinkage estimator based on MVR is superior to Bayesian and QML estimators in terms of bias and root mean square error. We examine the precision of the shrinkage estimator using large-scale simulated data where models up to 1,500 dimensions and 4,503,000 parameters are fitted and studied. The proposed estimators are applied to stock return data, and the effectiveness of the proposed estimators is assessed in two ways. First, we show the usefulness of the proposed models and methods in estimating high-frequency returns with many assets and observations. Second, in the context of dynamic minimum variance portfolio strategy, we find unrestricted higher-order MSV models outperform existing alternatives, including multivariate GARCH-type models. |
| Presented by: Md Nazmul Ahsan, Canada Mortgage and Housing Corporation (CMHC) |
Regularized Wishart Stochastic Volatility for High-Dimensional Covariance ForecastingAbstractWe extend Uhlig's (1994,1997) Wishart stochastic volatility (WSV) model by introducing a regularized state transition for the precision matrix that shrinks the covariance forecasts toward a prior reference matrix. This regularization ensures stationarity of the return process and stabilizes the eigenvalues of the covariance forecasts, while preserving closed-form expressions for filtering, prediction, and likelihood evaluation. We provide conditions for stationarity and for the existence of second- and fourth-order moments, show that the model admits a multivariate GARCH representation, and derive bounds that illustrate how regularization prevents degeneracy and excessive dispersion in the eigenvalues of the covariance matrix forecasts. To account for regime shifts in the correlation structure, we further embed a time-varying directional forgetting scheme into the regularized model, allowing the forgetting rate to differ over time and across directions in the return space. In a high-dimensional application with up to 1,000 assets, the regularized models deliver significantly more accurate covariance forecasts than the baseline WSV and perform well compared to several benchmark models. |
| Presented by: Julian Spies, University of Cologne |
Realized Variance DisagreementAbstractWe propose realized variance disagreement as a nonparametric measure of segmentation between equity and options markets. Defined as the aggregate difference between diffusive variance components of high-frequency asset returns and their risk-neutral conditional expectations recovered from options expiring at the end of the trading day, this statistic captures realized violations of the no-arbitrage condition that diffusive variance coincides under the physical and risk-neutral probability measures. We derive a central limit theorem that enables feasible inference on integrated variance disagreement between equity and options markets. We further develop market integration tests that can detect spot and integrated variance disagreement. Empirically, we find evidence for episodes of disagreement between equity and options markets. These periods exhibit mild persistence and differ in magnitude across assets. |
| Presented by: Carsten Chong, HKUST |
| Session 86: IDENTIFICATION OF MACRO SHOCKS June 24, 2026 15:45 to 17:30 Location: D-105 |
| Session Chair: Jorge de la Cal Medina, University of Amsterdam, Tinbergen Institute |
Estimating Macroeconomic Effects of Government Spending ShocksAbstractBuilding on the identification strategy of Ramey (2011), this paper proposes a novel approach to identifying fiscal policy shocks. Specifically, we orthogonalize the forecast error of U.S. government spending from the Survey of Professional Forecasters (SPF) with respect to a broad set of forecasts of other macroeconomic variables, as well as a collection of additional structural macroeconomic shocks. We first employ the identified shock in a standard SVAR-IV framework, distinguishing also between federal and state–local government spending, and find fiscal multipliers that differ from those obtained using the raw forecast error. Finally, exploiting Bayesian estimation techniques, we estimate an SVAR-IV model with a large information set. |
| Presented by: Giacomo Porcellotti, Università di Torino |
The pass-through of sectoral cost shocks and the aggregate price volatilityAbstractThe transmission of input cost shocks to prices has been the subject of intense debate. This article highlights the key role of corporate attention in this mechanism. We develop a model of rational inattention where firms face costs to process information about their economic environment. In a low-volatility environment, firms are rationally inattentive to idiosyncratic cost shocks, resulting in price inertia and negligible pass-through. However, we show that heightened aggregate price volatility acts as a "wake-up call": the need to track aggregate uncertainty reduces the marginal cost of processing firm-specific information. This informational spillover leads to a faster and more complete adjustment of prices to idiosyncratic input costs. We empirically test this prediction on U.S. manufacturing data using a panel local projection approach and Granular Instrumental Variables (GIV) to identify sectoral-level supply shocks. We find that the pass-through of these shocks is virtually zero in stable regimes but becomes significant and amplified during periods of high aggregate volatility, consistent with our theoretical framework. |
| Presented by: Lucia Veraldi, Paris Dauphine PSL |
Temperature Distributional Shocks: Identification and Macroeconomic EffectsAbstractThis paper introduces a methodology to identify temperature distributional shocks, i.e., shocks capturing shifts not only in the average but also across different quantiles of the temperature process. Using data for the globe and a panel of 21 economies, we consistently recover three types of local and global temperature distributional shocks and estimate their macroeconomic impacts on output and total factor productivity growth. The first type of shock captures a classic distributional shift and yields economic responses consistent with studies that assume the average temperature as a sufficient statistic for climate change. Our key contribution is to uncover two additional types of shocks reshaping the temperature distribution: (i) a variability- shock that moves mid-lower and mid-upper quantiles in opposite directions, and (ii) an extremes-shock that shifts the tails relative to the center. These variability and extremes shocks induce macroeconomic responses not documented in standard average-based studies. Our results reveal the value of modeling changes in the whole temperature distribution for climate-macro analysis and carry important implications for social cost of carbon estimation and climate-related risk assessment. |
| Presented by: Maria Dolores Gadea, University of Zaragoza |
Optimal Decision Rules for Impulse Response Matching under Weak IdentificationAbstractDSGE models with different structural parameters can be observationally equivalent for the moments used in limited-information estimation. This is consequential for policy: parameter values that generate the same moments can imply different welfare rankings of policy rules. We study the choice of an interest-rate rule from a statistical decision-theoretic perspective. We apply a quasi-Bayes decision rule that minimizes posterior expected loss. The rule averages welfare over the structural values the moments leave plausible. We illustrate the construction in a Monte Carlo design and in a medium-scale New-Keynesian DSGE model estimated by impulse-response matching. |
| Presented by: Jorge de la Cal Medina, University of Amsterdam, Tinbergen Institute |
| Session 87: INSTITUTIONAL GOVERNANCE, CONFLICT AND SOCIAL DYNAMICS June 24, 2026 15:45 to 17:30 Location: B129 |
| Session Chair: Liang Zhong, The University of Hong Kong |
Breaking the Glass Ceiling: How Female Mayors Affect Women's Political Advancement: Evidence from Spanish Municipal ElectionsAbstractThis paper investigates whether female executive leadership generates spillover effects on other women's political careers. Using a fuzzy regression discontinuity design exploiting close mixed-gender elections in Spanish municipalities (2011-2023), we find that female mayors improve other women's positions on party lists. Female mayors reduce normalized list rank by 0.021 points and increase access to first position by 5.7 percentage points. Effects differ by candidate tenure: new female candidates experience large improvements in list placement, while incumbent women benefit primarily through enhanced electoral success. The overall probability of election to council increases by 4.2 percentage points, and gender-specific analysis confirms these effects are exclusive to women. The share of women on party lists is unchanged, indicating that female mayors operate through repositioning within lists rather than expanding female candidacies. These results are more consistent with signaling to party gatekeepers than with network or role model mechanisms. |
| Presented by: David Mesa-Ruiz, University of Edinburgh |
List Experiment, Lying Costs and Rational Misreporting of Sexual Violence in BrazilAbstractWe contribute to the literature on the measurement of Intimate Partner Violence (IPV) by addressing validity challenges due to its notorious underreporting, social desirability bias, and incomplete knowledge about victims’ information-transmission mechanism. First, we challenge the List Experiment’s (LE) assumption of "no liars" with a new "partial liars" assumption within a structural sender-receiver game model, capable of generating rational misreporting equilibria. Second, our empirical evidence is obtained from a LE conducted in the context of a randomized response information game, carried out on a large (> 10, 000) representative longitudinal household survey in Brazil (PCSVDF-Mulher). Third, estimation has taken a non-conventional route by interpreting the likelihood function from the LE’s treatment group as a convolution, then using the estimates from the controls’ sample as a first step to back up a maximum likelihood two-step estimation procedure of the sensitive answer parameters. Fourth, we explore the advantages of combining a LE with a set of direct CTS-type IPV questions to shed light on the prevalence of non-partner sexual violence (NPSV) determinants. Our Sender-Receiver List Experiment Participation Game estimated not only the true (unobserved) prevalence of NPSV (≈ 0.10), a value greater than twice the difference-in-means figure (≈ 0.04); but also a quite elusive parameter, its probability of misreporting, conditional on being a victim, i.e., false negatives (≈ 0.56). A parabola-type effect emerged from the estimated age versus race profile of NPSV prevalence, where non-whites depict rates approximately twice those of whites, although both groups achieve their minimum prevalence (around 0.07 and 0.03, respectively) at 35 years of age. Our results have the potential to qualify IPV measurement to help the public policies of Brazil and other developing countries’ such as China, India, Indonesia, Nigeria, Pakistan, and Russia, to name a few, overcome one of the largest human rights and economic development scourges in the world. |
| Presented by: Jose Carvalho, Universidade Federal do Ceara |
Aid, Interrupted: Conflict Dynamics Following the USAID Suspension in AfricaAbstractForeign aid withdrawal may put the political stability of aid-dependent states at risk, yet short-term conflict responses to donor-side aid disruptions remain poorly understood. We exploit the abrupt and unanticipated suspension of the United States Agency for International Development (USAID) in January 2025 as an exogenous shock to examine its effects on conflict dynamics in Africa. Using monthly conflict data for 44 Sub-Saharan African countries, we apply an event-study design to trace conflict responses in highly aid-dependent countries relative to less aid-dependent countries following the suspension. We find a general increase in conflict, with a swift 12 percent rise in armed conflict between organized groups and a delayed 10--14 percent increase in militia-perpetrated violence against civilians. We show that these effects are primarily driven by countries with limited state capacity and high political corruption, pointing to the role of institutions in addressing socio-political tensions arising from aid disruptions. |
| Presented by: David Ubilava, University of Sydney |
Unconditional Randomization Tests for InterferenceAbstractResearchers are often interested in the existence and extent of interference between units when conducting causal inference or designing policy. However, testing for interference presents significant econometric challenges, particularly due to complex clustering patterns and dependencies that can invalidate standard methods. This paper introduces the pairwise imputation-based randomization test (PIRT), a general and robust framework for assessing the existence and extent of interference in experimental settings. PIRT employs unconditional randomization testing and pairwise comparisons, enabling straightforward implementation and ensuring finite-sample validity under minimal assumptions about network structure. The method’s practical value is demonstrated through an application to a large-scale policing experiment in Bogot´a, Colombia (Blattman et al., 2021), which evaluates the effects of hotspot policing on crime at the streetsegment level. The analysis reveals that increased police patrolling in hotspots significantly displaces violent crime, but not property crime. Simulations calibrated to this context further underscore the power and robustness of PIRT. |
| Presented by: Liang Zhong, The University of Hong Kong |
| Session 88: INSTRUMENTAL VARIABLES 1 June 24, 2026 15:45 to 17:30 Location: B008 |
| Session Chair: Yaroslav Korobka, CERGE-EI |
Many Instruments Estimation and Inference under Clustered DependenceAbstractThe literature on many weak instruments in a heteroskedastic environment under data independence is largely developed. When data dependence, in particular clustering, is present, it poses difficulties in making correct and convenient inferences about structural parameters. We show that clustering either deems the conventional jackknife instrumental variables estimation inconsistent or makes its inferences distorted. We suggest an alternative approach to the estimation of and making inferences about structural parameters, which is computationally attractive and allows general structures of intra-cluster correlations, presence of many instruments and possibly weak identification. We use the natural extension of jackknifing, the leave-cluster-out methodology, applied to the instrument projection matrix, which allows one to dispose of the cross-cluster dependencies in the influence function of the structural parameter estimate. We further weigh the observations by inverse cluster sizes to flexibly adjust for cluster size heterogeneity, which relaxes the usual requirements on the maximal cluster size growth and facilitates derivation of asymptotic properties. We set out a formal asymptotic framework to analyze the proposed weighted leave-cluster-out instrumental variables (WLCOIV) estimator, with an increasing number of clusters, possibly increasing cluster sizes, and presence of many possibly weak instruments. We prove a central limit theorem for the influence function embedded in the WLCOIV estimator under both strong and weak identification, and show consistency of the associated WLCOIV variance estimator. Finally, we run a small simulation experiment and illustrate with the celebrated Angrist and Krueger (1991) dataset, comparing the WLCOIV estimator to other estimators. |
| Presented by: Stanislav Anatolyev, CERGE-EI and New Economic School |
Power Bounds and Efficient Loss for Asymptotically Optimal Tests in IV RegressionAbstractConditional tests based on the Anderson–Rubin (AR) and Lagrange multiplier (LM) statistics satisfy a minimax, invariant criterion under homoskedastic errors but discard essential information in overidentified IV models with heteroskedastic or autocorrelated errors. The conditional likelihood ratio (CLR) test instead uses information beyond the AR and LM statistics. Among known tests similar under weak identification and efficient under standard asymptotics, CLR is the only one that avoids information loss induced by restrictive invariance criteria. We derive a useful power bound for CLR to show its power–size gap converges to one when distinguishing the null from the alternative is trivial (total variation distance near one). In contrast, the information loss of AR–LM based tests can generate severe power failures: LM and conditional quasi-LR tests can even be inconsistent, with power arbitrarily close to size even under strong identification. We revisit Yogo (2004) to illustrate that these impossibility designs arise in practice. |
| Presented by: Marcelo Moreira, FGV |
Generalized AKM: theory and evidenceAbstractThis paper introduces an estimator for quadratic forms based on the linear parameters of a semi-parametric model. The leading example is the workhorse model of wage determination by Abowd, Kramarz, and Margolis (1999, AKM): our estimator targets standard variance components while allowing for a nonparametric treatment of both worker- and firm-level observable characteristics. We propose a bias-corrected estimator robust to heteroskedasticity that controls for approximating functions of the covariates. We show that this estimator is asymptotically unbiased and consistent when the number of linear parameters (e.g. the AKM fixed effects) is proportional to the sample size. In particular, consistency hinges on a strengthened smoothness condition on the nonparametric component’s functional class. In an empirical application, we show that adding a rich set of controls to the standard AKM model yields implausibly large firm effects. Our method addresses this issue, yielding estimates of variance components that are more robust relative to conventional approaches. Confounding—not functional-form choice—drives the standard model’s instability. |
| Presented by: Yaroslav Korobka, CERGE-EI |
| Session 89: MACHINE LEARNING AND FACTOR MODELS June 24, 2026 15:45 to 17:30 Location: D-107 |
| Session Chair: Anna Sznajderska, SGH Warsaw School of Economics |
Quantile-Covariance Three-Pass Regression FilterAbstractWe propose a factor model for quantile regression using quantile-covariance(qcov), called the Quantile-Covariance Three-Pass Regression Filter (Qcov3PRF). This method estimates the supervised factors from a set of predictors to forecast the conditional quantile of a target. Our approach differs from the Partial Quantile Regression (PQR) as Qcov3PRF successfully allows the estimation of more than one relevant factor by virtue of using qcov. By estimating the true number of relevant factors, Qcov3PRF forecasts are consistent and asymptotically normal when both time and cross sectional dimensions become large. Simulations confirm these asymptotic results, showing Qcov3PRF exhibits good finite sample properties. Empirical applications to forecasting Growth-at-Risk highlight merits of Qcov3PRF over PQR. |
| Presented by: Pedro Isaac Chavez Lopez, Bank of Mexico |
Unfolding Regional Business Cycles: Factor Models for Three-Way State-Level TensorsAbstractThis paper develops a new approach to characterize regional business cycle dynamics using high-dimensional state-level data. Regional cycles are estimated using three-way tensor decompositions subject to orthogonal time components and non-negative factor loadings. The constraints act as inductive biases that resolve scale, sign, and rotation indeterminacy and produce what is known in the machine learning literature as a “parts-based” representation of the data. A model with four factors captures over 80% of the variation in the data while reducing its dimensionality by more than 90%. The estimated regional cycles correspond to distinct economic forces and align closely with well documented regional specializations. These patterns emerge directly from the data without the use of covariates or geographic clusters. |
| Presented by: Sebastian Fossati, University of Alberta |
Explainable Machine Learning for U.S. Recessions Nowcasting: The Shapley ApproachAbstractMuch of the recession forecasting literature is focused on producing a model for predicting current and future economic states. Nevertheless, the question of why a model predicts a given class (recession or expansion), that is, which features are mostly responsible for the classification outcome, remains unaddressed. The main goal of our paper is to fill this gap. To address this, we apply Shapley values (Lundberg and Lee 2017) to provide model transparency. Using high-quality macroeconomic data from multiple U.S. government agencies, we use competing linear and non-linear models. We then compute Shapley values across 154 feature candidates to nowcast US recessions. Our findings offer novel insights into feature importance and model decision-making, which allows rationalizing model predictions. This is useful, for example, in the context of monetary policy, where the central banker needs, not only a good prediction of whether or not we are in a recession, but also a sense of why that prediction was made. |
| Presented by: Jerome Lahaye, Fordham University |
Data leakage in time-series forecasting: Lessons from exchange rate predictionAbstractThis article highlights the problem of data leakage in studies using economic models in exchange rate forecasting competitions, which has become increasingly challenging with the rise of machine learning techniques. We point to four types of data leakage, which are most relevant in these forecasting contests. Next, we construct a rich panel dataset of monthly observations for G10 currencies against the U.S. dollar over the period 1990–2025 and use them to develop XGBoost models linking exchange rates with macroeconomic fundamentals. Our results show that, when using the same model, forecasting performance relative to the random walk benchmark crucially depends on how information leakage is controlled for, as the XGBoost model can either under- or outperform the random walk benchmark. The article contributes to the literature by offering guidelines for machine learning practitioners on the proper design of time-series forecasting competitions, but at the same time presents a new evidence on the ability of XGBoost framework to solve the exchange rate disconnect puzzle. |
| Presented by: Anna Sznajderska, SGH Warsaw School of Economics |
| Session 90: MONETARY POLICY TRANSMISSION 2 June 24, 2026 15:45 to 17:30 Location: D-114 |
| Session Chair: Carl-Wolfram Horn, Frankfurt School of Finance & Management |
The role of the housing market in monetary policy transmissionAbstractWe estimate how house prices, consumption and housing credit in the euro area respond to contractionary monetary policy shocks using Jordà (2005) local projection method and monetary policy shocks from Altavilla et al. (2019). We find that consumption, GDP and inflation fall over the short to medium term and rebound later, while house prices decline markedly in the first year and recover only slowly thereafter. We also find that the main margin of adjustment is the price of credit, as mortgage rates rise sharply on impact and revert to pre-shock levels over time, while credit quantities adjust more modestly. The effects are stronger and more persistent in periphery economies and in countries with a higher prevalence of variable-rate mortgages, highlighting the housing channel as an important source of asymmetric monetary transmission within the euro area. |
| Presented by: Joana Sousa-Leite, Banco de Portugal |
Conventional monetary policy across the wealth distribution: the Maltese caseAbstractThis paper studies how conventional monetary policy affects household wealth in Malta across the wealth distribution. We combine euro-area macroeconomic responses to a monetary policy shock with household-level microdata on assets, liabilities, income, and consumption from the 2023 Maltese Household Finance and Consumption Survey to quantify distributional wealth effects. The impact of the shock depends both on the size of households’ asset and liability holdings and on the composition of their portfolios across wealth deciles. Housing and the underlying mortgage debt play a major role for the transmission of a contractionary policy. Households at the bottom of the wealth distribution appear to be the least affected, whilst middle deciles are hit the hardest. Finally, wealthiest households experience more moderate losses. Additional results on income sources and consumption-bundle prices also reveal heterogeneous effects. The former are characterised by their reliance on self-employment and entrepreneurial activities, while the latter depend on the consumption allocation for each decile. |
| Presented by: Germano Ruisi, Central Bank of Malta |
Uncertainty, Bank Lending Standards, and the Transmission of Monetary PolicyAbstractThis paper analyses how uncertainty affects the transmission of monetary policy. Using state‑dependent local projections we first show that the impact of monetary policy shocks on corporate lending is markedly weaker when uncertainty is high. We then develop a quantitative macro‑banking model to rationalize these findings. In the model, higher uncertainty steepens the loan‑supply curve due to higher expected default, thereby dampening the effects of monetary policy on credit and investment. Finally, we use the euro area credit register to validate that banks steepen their loan supply curve for more uncertain borrowers. Consistent with the model mechanism, low capital banks facing high uncertainty display a higher sensitivity of lending rates and a lower sensitivity of lending volumes. |
| Presented by: Luis Herrera Bravo, Banco de España |
Bond Market Fragmentation and Monetary Policy in the Euro AreaAbstractWe examine the state-dependent transmission of monetary policy and its cross-country heterogeneity in the Euro Area under varying levels of sovereign bond market fragmentation. We find that contractionary monetary policy significantly dampens real activity and prices in times of low financial market fragmentation. It has only muted effects in highly fragmented markets. The results are robust to controlling for the state-dependent effect of monetary policy across the business cycle. We find no significant evidence that the transmission of monetary policy becomes more heterogeneous across countries in high- compared to low-fragmentation regimes. |
| Presented by: Carl-Wolfram Horn, Frankfurt School of Finance & Management |
| Session 91: NONLINEAR TIME SERIES MODELS June 24, 2026 15:45 to 17:30 Location: D-106 |
| Session Chair: Savi Virolainen, University of Helsinki |
The Global Carbon Budget as a cointegrated systemAbstractThe Global Carbon Budget, maintained by the Global Carbon Project, summarizes Earth’s global carbon cycle through four annual time series beginning in 1959: atmospheric CO2 concentrations, anthropogenic CO2 emissions, and CO2 uptake by land and ocean. We analyze these four time series as a multivariate (cointegrated) system. Statistical tests show that the four time series are cointegrated with rank three and identify anthropogenic CO2 emissions as the single stochastic trend driving the nonstationary dynamics of the system. The three cointegrated relations correspond to the physical relations that the sinks are linearly related to atmospheric concentrations and that the change in concentrations equals emissions minus the combined uptake by land and ocean. Furthermore, likelihood ratio tests show that a parametrically restricted error-correction model that embodies these physical relations and accounts for the El Niño/Southern Oscillation cannot be rejected on the data. The model can be used for both in-sample and out-of-sample analysis. In an application of the latter, we demonstrate that projections based on this model, using Shared Socioeconomic Pathways scenarios, yield results consistent with established climate science. |
| Presented by: Morten Nielsen, Aarhus University |
Estimating large-scale nonlinear macroeconomic models using the ensemble transform Kalman filterAbstractThis paper shows how to estimate large-scale nonlinear Dynamic Stochastic General Equilibrium (DSGE) models using the Ensemble Transform Kalman Filter (ETKF). Estimating large-scale DSGE models with many state variables, such as multi-sector and multi-country models or those with numerous shocks and real and nominal frictions, presents significant computational challenges. Failing to account for nonlinearities leads to inconsistent parameter estimates. However, standard nonlinear filtering methods become infeasible in high-dimensional settings due to the large number of particles required in particle filtering or the computational burden of tensor product- based discretization methods. This paper demonstrates that the ETKF, combined with data augmentation, provides a computationally efficient and accurate alternative. We illustrate this through a simulation study estimating a multi-country DSGE model and an empirical application that quantifies the importance of nonlinearities in estimating the role of sectoral shocks in driving aggregate fluctuations within a multi-sector model. For standard estimates of elasticities of substitution between goods in both demand and production, we find that sectoral shocks contribute approximately 36% more to aggregate fluctuations in the nonlinear economy compared to the linear economy. |
| Presented by: Otilia Boldea, Tilburg University |
Structural Analysis in Nonlinear Vector Autoregressions using Bayesian Additive Regression TreesAbstractThis paper integrates existing research on Bayesian Additive Regression Tree (BART) models with structural analysis of time series data with potential non-linearities. To estimate the model, we use a multivariate extension of BART based on a seemingly unrelated regression (SUR) framework, called Seemingly Unrelated BART (SUBART), which we adapt to suit the properties of macroeconomic time series. The multivariate extension enables us to incorporate various (structural) identification methods beyond the order-dependent recursive Cholesky decomposition. To induce sparsity in the model, we incorporate a Dirichlet prior over the splitting variables, which effectively shrinks the predictor space by selecting only the most relevant (lagged) variables for building the trees. We evaluate the performance of our approach in a simulation study for both linear and non-linear processes highlighting its flexibility with respect to the complexity of the data. The usefulness of the model is illustrated in an empirical application on monetary policy shocks. |
| Presented by: Benedikt Schwab, Universität Konstanz |
Identification by non-Gaussianity in structural smooth transition vector autoregressive modelsAbstractWe show that structural smooth transition vector autoregressive models are statistically identified if, conditionally on past information, the shocks are mutually independent and at most one of them is Gaussian. This extends a known identification result for linear structural vector autoregressions to a nonlinear specification featuring a time-varying impact matrix. In particular, we provide explicit identification results for logistic, threshold, and nonrandom exogenous transition weights. We also propose an estimation method, show how a blended identification strategy can be adopted to address weak identification, and, in the special case of independent and identically distributed shocks, establish a sufficient condition for ergodic stationarity. The introduced methods are implemented in the accompanying R package sstvars. Our empirical application finds that a positive climate policy uncertainty shock reduces production and raises inflation under both low and high economic policy uncertainty, but its effects, particularly on inflation, are stronger during the latter. |
| Presented by: Savi Virolainen, University of Helsinki |
| Session 92: PANEL DATA METHODS 2 June 24, 2026 15:45 to 17:30 Location: D-112 |
| Session Chair: Hyunseok Jung, University of Arkansas |
Inference after discretizing time-varying unobserved heterogeneityAbstractApproximating time-varying unobserved heterogeneity by discrete types has become increasingly popular in economics. Yet, provably valid post-clustering inference for target parameters in models that do not impose an exact group structure is still lacking. This paper fills this gap in the leading case of a linear panel data model with nonseparable two-way unobserved heterogeneity. Building on insights from the double machine learning literature, we propose a simple inference procedure based on a bias-reducing moment. Asymptotic theory and simulations suggest excellent performance. In the application on fiscal policy we revisit, the novel approach yields conclusions in line with economic theory. |
| Presented by: Jad Beyhum, KU Leuven |
Robust Inference Methods for Latent Group Panel Models under Possible Group Non-SeparationAbstractThis paper presents robust inference methods for general linear hypotheses in linear panel data models with latent group structure in the coefficients. We employ a selective conditional inference approach, deriving the conditional distribution of coefficient estimates given the group structure estimated from the data. Our procedure provides valid inference under possible violations of group separation, where distributional properties of group-specific coefficients remain unestablished. Furthermore, even when group separation does hold, our method demonstrates superior finite-sample properties compared to traditional asymptotic approaches. This improvement stems from our procedure's ability to account for statistical uncertainty in the estimation of group structure. We demonstrate the effectiveness of our approach through Monte Carlo simulations and apply the methods to two datasets on: (i) the relationship between income and democracy, and (ii) the cyclicality of firm-level R&D investment. |
| Presented by: Oguzhan Akgun, University of Burgundy |
Semiparametric Panel Data Models with Interactive Fixed EffectsAbstractThis paper develops methods for estimation and inference in semiparametric panel data models with both correlated random effects and interactive fixed effects. Building on the Mundlak specification to control for unobserved heterogeneity in both the cross-sectional and time dimensions, our estimation procedure identifies the nonparametric function, finite-dimensional parameters associated with observed time-invariant and cross-section-invariant regressors, and unobserved interactive effects. We develop a two-stage estimation approach: the first stage jointly estimates the parametric components and interactive effects, while the second stage recovers the nonparametric function using Robinson-style estimation. We establish the asymptotic properties of our proposed estimators under large $N$ asymptotics with fixed $T$. Monte Carlo simulations support the asymptotic developments. We illustrate the practical utility of our approach via an empirical application examining the relationship between firm innovation and market competition. |
| Presented by: Emma Henry, University of Alabama |
Discrete-time Hazard Models for Non-repeated Events: a Generalized Correlated Random Effects ApproachAbstractThis paper proposes a generalized Correlated Random Effects (CRE) approach for discrete-time hazard models with non-repeated events. Individual heterogeneity is specified as a sum of some arbitrary function of the means of exogenous covariates in the model and a random error, which is more general than the conventional CRE approach that assumes a linear function for the dependence structure. We estimate the model using a matching estimator that compares the outcomes of individuals with similar covariate values. We study the asymptotic properties of the estimator and discuss several extensions that relax some of the assumptions in the model. Our Monte Carlo simulations show that our estimator outperforms the conventional CRE model under more general specification for individual heterogeneity. We apply our estimator to data on truck driver turnover to analyze how different types of shocks (e.g., equipment, pay, scheduling) affect the employment duration of truck drivers while accounting for individual heterogeneity that may be correlated with the covariates in the model. |
| Presented by: Hyunseok Jung, University of Arkansas |
| Session 93: PLATFORMS AND MARKET STRUCTURE June 24, 2026 15:45 to 17:50 Location: D-111 |
| Session Chair: Pierre Dubois, Toulouse School of Economics |
Technology Equalizers: How Digital Platforms Level the Playing Field for Small FirmsAbstractWe study how platform technologies differentially impact the productivity of smaller firms. Platforms reduce a broad range of costs of coordinating and motivating transactions: the cost of processing information, of matching and reaching customers, of informational asymmetries, and of completing and enforcing contracts. Our theory suggests that, by reducing transaction costs, platforms allow firms to replace high fixed cost, low variable cost technologies (insourcing) by low fixed cost, high variable cost technologies (outsourcing). Using data from the European Investment Bank Investment Survey (EIBIS), we find that digital platform adoption is associated with increases in labor productivity of 3.4% to 5.1% on average, with smaller firms experiencing a larger increase of 6.5% to 10.5%. Our results are robust to controlling for endogeneity by using instrumental variables (US adoption rates in the same sector and regional internet speed) in a three stage estimation procedure (since adoption is a binary variable). We conclude that while the large physical and human capital requirements of IT investments favor large firms and lead to industry concentration, digital platforms can level the playing field by facilitating the access of smaller firms to these same technologies through the market. |
| Presented by: Christoph Weiss, European Investment Bank |
How Do Suppliers Choose in a Platform Market? A Case Study of the Game IndustryAbstractThis paper studies developers’ platform choice, with a focus on exclusivity, in the sixth-generation video game console market. We estimate a supply-side structural model in which third-party developers choose among single-platform exclusivity and simultaneous multi-platform release, accounting for platform-specific development costs, cross-platform synergies, and exclusivity incentives. We use a revenue-based approach that recovers these primitives without estimating a full demand system, relying on observed revenues and a selection-corrected revenue model to discipline payoffs across platform choices. In a static setting, we find substantial asymmetries in development costs and synergies across platforms, as well as sizable benefits associated with permanent exclusivity. Extending the model to allow developers to endogenously choose launch timing substantially reduces the estimated exclusivity benefits—particularly for platforms with smaller installed bases—indicating that part of what appears as an exclusivity premium in static models reflects the option value of waiting rather than direct platform compensation. |
| Presented by: Arifah Hasanbasri, University of Pittsburgh |
An Empirical Analysis of Merger EfficienciesAbstractWe develop an econometric method to study merger efficiencies. Classification techniques are employed first to determine the sign of the merger’s effect on output levels in specific markets. These classifications are then combined with familiar oligopoly theory results to yield bounds on marginal cost savings. Applying this framework to the 2013 merger of US Airways and American Airlines we find that the merger led to output expansions in more than half of the markets where the sign of the output effect could be determined, and in at least 44% of the total number of markets that were directly affected by the merger and where the market structure was otherwise stable. Pro-competitive effects were more prevalent in larger markets and, to some extent, in markets that serve the merging carriers’ hubs. Averaging across the markets experiencing output expansions, the lower bound on the marginal cost reduction was slightly above 2 USD, capturing 0.8% of the market price. The analysis provides insights regarding the nature and magnitude of merger efficiencies. |
| Presented by: Alon Eizenberg, Hebrew University Jerusalem |
Pharmaceutical Advertising in Dynamic EquilibriumAbstractDirect-to-consumer advertising (DTCA) of prescription drugs may expand treatment access but risks promoting overuse and business-stealing without welfare gains. Among developed nations, only the United States and New Zealand permit DTCA, whereas detailing - promotion aimed at prescribers - is widely practiced. This paper analyze the impact of DTCA on profits by modeling a counterfactual environment which bans DTCA. This is done through the lens of a dynamic equilibrium framework that we provide for adapting Experience Based Equilibrium \citep{FershtmanPakes2012} for empirical work. EBE incorporates constraints on the cognitive abilities of decision makers and mitigates the computational concerns of researchers. Using data from four therapeutic markets we first validate the EBE's ability to replicate observed advertising patterns, then simulate counterfactual DTCA bans. Both the data and our empirical work indicate that DTCA and detailing are strong complements, and our results illuminate the need to account for this when evaluating the ban. The ban leads firms to lower DTCA and has a negative effect on profits in all markets, but the magnitude of the effect varies; from under 5% in the market for Ulcer to 27.5% for Asthma medications. |
| Presented by: Pierre Dubois, Toulouse School of Economics |
Pumped Hydro Energy Storage Facility: Optimal OperationAbstractHydropower is a renewable and flexible energy source that provides essential storage capacity and enhances grid stability. Among storage technologies, pumped hydro energy storage (PHES) remains the most cost-effective solution for long-duration energy storage and plays a key role in power systems with increasing penetration of variable renewable energy. As electricity prices become more relevant under the energy transition, understanding the optimal operation and valuation of PHES assets is increasingly important from a financial perspective. This paper develops a market-based framework that models a PHES facility as a profit-maximizing asset operating in liberalized electricity markets. Using a stochastic optimal control approach calibrated with real-life technical and operational parameters of the La Muela pumped storage plant and observed electricity prices from the Spanish wholesale market, the model derives an economically intuitive trigger (switching) price governing optimal pumping and generation decisions while accounting for reservoir water inventory dynamics and electricity price uncertainty. The results show that inventory dynamics and electricity price seasonality are central to PHES valuation and optimal operation. Optimal strategies are implemented at a monthly frequency, with peak profitability achieved in September at an optimal trigger price of approximately €60/MWh, yielding total profits exceeding €175 million per year. These findings highlight the importance of price-based operational rules and intertemporal inventory management for maximizing the economic value of PHES facilities. |
| Presented by: Isabel Figuerola Ferretti, ICADE |
| Session 94: IAAE General Assembly. Chair: M. Chauvet (University of California Riverside) June 24, 2026 17:30 to 18:00 Location: Grand Auditorium |
| Session 95: Conference dinner June 24, 2026 18:30 to 23:00 |
| Session 96: IAAE Keynote — Data-driven Nests — Elena Manresa (Princeton University) — Chair & Moderator: Martin Weidner (Oxford University) June 25, 2026 8:45 to 9:45 Location: Grand Auditorium |
| Session 97: Coffee break June 25, 2026 9:45 to 10:15 |
| Session 98: AUCTIONS June 25, 2026 10:15 to 12:00 Location: D-107 |
| Session Chair: Qidi Hu, Shanghai Institute for Mathematics and Interdisciplinary Sciences, Fudan University |
Auction Identification with Unobserved Rejected OffersAbstractAbstract We study identification of buyer and seller primitives in second-price auctions when sellers may reject the highest bid but only successful transactions are observed. In our framework a sale occurs only if the top bidder value exceeds a seller-chosen threshold that is unobserved by the econometrician. We first show that point identification using variation in the number of bidders is often achievable but may fail: in the two-point case the model has set identified data regions in which price distributions can be rationalised by two pairs of value and threshold distributions. For general supports we reduce identification to a fixed point in the ratio of sale probabilities across bidder counts; uniqueness yields point identification. We then show that upper-tail support separation between the buyer value and threshold distributions is a sufficient condition for point identification. A conduct assumption maps thresholds into seller values; absent conduct we derive bounds based on assumptions similar to Haile and Tamer (2003). We quantify biases from ignoring rejected offers, at the population level, for evaluation of counterfactual mergers and entry/exit episodes, and in Monte Carlo experiments, for the distributions themselves. The Monte Carlo experiments validate our approach. |
| Presented by: David Genesove, Hebrew University of Jerusalem |
Revenue equivalence and auction formats in the real world: Evidence from millions of transactions in fish marketsAbstractThis paper tests the empirical relevance of Vickrey’s revenue equivalence theorem using rich field data from first-sale fish markets. The analysis is based on nine million transactions observed over eleven years in five geographically proximate French fish markets. One market operates under a purely descending auction format, while the remaining four operate under a hybrid descending-then-ascending auction format. I compare fish prices across auction mechanisms using both parametric linear regressions and nonparametric exact matching on discrete covariates to account for differences in product availability across fish markets. The results show that fish prices are, on average, higher under the descending auction format, by between 4% and 8% when pooling all species, with substantial heterogeneity across species. Counterfactual simulations further indicate that switching to a descending auction format would increase revenues in three of the four markets currently operating under the descending-then-ascending format. |
| Presented by: François-Charles Wolff, Nantes University |
Position Auctions with Organic SearchAbstractRecent regulatory actions, such as the FTC v. Amazon antitrust case, have raised concerns about the impact of sponsored advertisements on consumer welfare in online search platforms. While theoretical models of position auctions typically predict that sellers are ranked by consumer-match/seller-quality in equilibrium, these models often abstract from the coexistence of sponsored and organic listings. We develop a model in which sellers can appear in both sponsored and organic positions and examine how this affects equilibrium outcomes and consumer welfare. Our model captures a key tradeoff: high-quality sellers value the visibility from sponsored placement but also expect to appear prominently in organic rankings. As a result, under certain conditions, lower-quality sellers may outbid them to obtain the sponsored position - lending some support to the FTC’s concern. However, we show that this outcome only arises when all sellers are relatively high-quality, which limits potential consumer harm. |
| Presented by: Qidi Hu, Shanghai Institute for Mathematics and Interdisciplinary Sciences, Fudan University |
| Session 99: DENSITY FORECASTING June 25, 2026 10:15 to 12:00 Location: D-105 |
| Session Chair: Vítor Gentini, Fundação Getulio Vargas |
Stochastic Volatility-in-mean VARs with Time-Varying SkewnessAbstractThis paper introduces a Bayesian vector autoregression (BVAR) with stochastic volatility-in-mean and time-varying skewness. Unlike previous approaches, the proposed model allows both volatility and skewness to directly affect macroeconomic variables. We provide a Gibbs sampling algorithm for posterior inference and apply the model to quarterly data for the US and the UK. Empirical results show that skewness shocks have economically significant effects on output, inflation and spreads, often exceeding the impact of volatility shocks. In a pseudo-real-time forecasting exercise, the proposed model outperforms existing alternatives in many cases. Moreover, the model produces sharper measures of tail risk, revealing that standard stochastic volatility models tend to overstate uncertainty. These findings highlight the importance of incorporating time-varying skewness for capturing macro-financial risks and improving forecast performance. |
| Presented by: Ana Skoblar, European Central Bank |
A New Model of Trend Inflation Using Disaggregates, Survey Expectations, and UncertaintyAbstractThis paper develops a new empirical model that estimates trend inflation by combining modeling features that have advanced the literature on trend inflation over the past two decades. These features include incorporating information about long-term inflation expectations from surveys in a flexible way, modeling aggregate inflation via sectoral data (goods and services), allowing for stochastic volatility (SV) in the shocks to the trend and transitory components of inflation, allowing for a time-varying price Phillips curve, and allowing for time-varying uncertainty effects on the level of inflation. We estimate the model using state-of-the-art Bayesian methods. We document the competitive properties of the new model compared to variants that include only a subset of the above features. The new model provides a more interpretable historical decomposition of inflation data than the models it extends. The decomposition suggests that uncertainty effects play a greater role than cyclical effects in explaining inflation fluctuations. |
| Presented by: Saeed Zaman, Federal Reserve Bank of Cleveland |
Tail-Aware Density Forecasting of Locally Explosive Time Series: A Neural Network ApproachAbstractThis paper proposes a Mixture Density Network specifically designed for forecasting time series that exhibit locally explosive behavior. By incorporating skewed t-distributions as mixture components, our approach offers enhanced flexibility in capturing the skewed, heavy-tailed, and potentially multimodal nature of predictive densities associated with bubble dynamics modeled by mixed causal-noncausal ARMA processes. In addition, we implement an adaptive weighting scheme that emphasizes tail observations during training and hence leads to accurate density estimation in the extreme regions most relevant for financial applications. Equally important, once trained, the MDN produces near-instantaneous density forecasts. Through extensive Monte Carlo simulations and two empirical applications, on the natural gas price and inflation, we show that the proposed MDN-based framework delivers superior forecasting performance relative to existing approaches. |
| Presented by: Julien Peignon, University Paris-Dauphine |
NEWS IV: A MODEL WITH NEWS AND IMPLIED VOLATILITY FOR ENHANCED VOLATILITY PREDICTIONAbstractThis research explores how news and implied volatility (IV) influence volatility predictions for the Ibovespa index ETF and five widely traded Brazilian stocks. We proposed a model which integrates IV and LDA-derived news topics and employed random forest to tackle nonlinearities and high dimensionality. Our results indicate that a model relying only on external variables (IV and news topics) significantly improves forecast accuracy for horizons beyond one day compared to the autoregressive models, even when these accounting for asymmetries and discontinuities. Our analysis of variable importance demonstrates that news topics are crucial for long-term forecasts, while IV significantly influences short-term predictions. This dissertation contributes to the economic literature by underlining the importance of textual data analysis and IV in volatility forecasting. |
| Presented by: Vítor Gentini, Fundação Getulio Vargas |
| Session 100: EDUCATION AND THE LABOR MARKET June 25, 2026 10:15 to 12:00 Location: B128 |
| Session Chair: Costanza Marconi, University of Bergamo |
Business Cycles and the Gender Gap in Educational InvestmentAbstractWe examine how economic conditions shape gender differences in educational investments. Using Swedish register data and within-family variation in siblings’ exposure to local unemployment rates prior to graduating high school, we study university enrollment and major choice over the business cycle. We find that men do not increase their overall university attendance during recessions but are more likely to shift towards fields associated with lower unemployment risks. In contrast, women respond to recessions by increasing both university attendance and completion rates. However, they do not shift toward fields with lower unemployment risks or higher earnings, although many women are academically prepared for demanding degrees. In the long run, exposure to recessions during the final high school years widens the gender gap in lifetime earnings. These findings suggest that men and women use different strategies to cope with recessions, with potentially long-term economic consequences. |
| Presented by: Hugo Morgado Azevedo, Lund University |
Firms’ Skill Demand and Worker Sorting: Education vs. ExperienceAbstractWe study how firms trade off formal education and relevant work experience when hiring workers. To do so, we embed a survey experiment in a large-scale employer survey and link responses to matched employer–employee registry data for more than 10{,}000 Norwegian firms. The experiment allows us to classify firms into types: experience lovers, who consistently prefer more experienced candidates; experience haters, who consistently prefer less experienced candidates. These types are similar on observable characteristics, but map into substantial differences in labor market outcomes. Compared with experience-loving firms, experience-hating firms grow faster, employ workers with flatter wage-tenure profiles, and display weaker positive sorting. |
| Presented by: Eyo Herstad, Erasmus University Rotterdam |
When Research Meets Teaching: Evidence from Student-Level DataAbstractThis paper studies whether the alignment between instructors’ research agendas and the content of the courses they teach affects student outcomes in higher education. Using administrative data from a large Italian university linked to publication records, we construct a novel text-based measure of research–teaching alignment that captures topic similarity between course syllabi and abstracts of instructors’ recent publications. Our empirical strategy exploits within-student and within-instructor variation to address selection concerns. We find that greater research–teaching similarity is associated with higher student performance, with stronger effects in advanced courses and among less-prepared students. Moreover, alignment positively affects graduation rates and final scores, as well as employment probability and wages one year after graduation. We provide suggestive evidence that these effects operate through increased exposure to frontier knowledge and methodologies and through higher perceived teaching quality, as students report greater satisfaction when courses are more closely aligned with instructors’ research. |
| Presented by: Costanza Marconi, University of Bergamo |
| Session 101: FIRMS AND SHOCK TRANSMISSION June 25, 2026 10:15 to 12:00 Location: D-115 |
| Session Chair: Juan Urquiza, Pontificia Universidad Catolica de Chile |
The Role of Firm Heterogeneity for the Transmission of Aggregate ShocksAbstractWe study whether firm-level heterogeneity helps explain U.S. macroeconomic fluctuations in response to aggregate shocks. Using quarterly Compustat and CRSP data from 1985 to 2024, we construct two revenue-based statistics inspired by Melitz (2003) model: the average firm and the marginal near-default firm, which summarize key features of the firm distribution under endogenous selection. We augment a Bayesian VAR with these statistics and compare it to a standard aggregate VAR and to a functional-VAR that incorporates the full cross-sectional distribution of firm revenues. We find that firm-level heterogeneity contains information not captured by aggregate variables. Including the two statistics allows the VAR to closely match the impulse responses obtained using the full distribution and improves out-of-sample forecast accuracy. These results show that the two statistics are sufficient to summarize the cross-sectional information relevant for macroeconomic transmission. |
| Presented by: Lorenza Rossi, Lancaster University |
Heterogenous Effects of FX Intervention Shocks on FirmsAbstractWe analyze the impact of FX intervention shocks on a panel of roughly 100,000 Japanese firms over up to 30 years. A purchase of 11 billion U.S. Dollar induces a Yen-depreciation by 1.1%; this increases firms’ revenues by 0.9 percentage points, profitability by 0.1 percentage points, and employment slightly, whereas leverage decreases. These effects hold for most industries, however, with marked differences across the 18 considered NACE-2 industries. Effects are heterogeneous regarding firm size and export orientation; larger export-oriented firms take advantage of higher revenues and profitability, while smaller domestically oriented firms create jobs and decrease leverage. |
| Presented by: Karoline Offen, Martin-Luther-Universität Halle-Wittenberg |
Monetary Policy and Financial Uncertainty: LP-IV versus Proxy-SVAR with High-Dimensional DataAbstractThis paper estimates how exogenous monetary policy tightenings affect financial uncertainty and compares two leading external-instrument estimators: Proxy-SVAR and Local Projection-IV (LP-IV). Using U.S. data and two benchmark instruments---Romer and Romer narrative shocks (quarterly, 1969--2007) and Miranda-Agrippino and Ricco high-frequency surprises (monthly, 1990--2016)---we find that a 25bp contractionary policy shock increases the VIX with a hump-shaped response that peaks after a few months/quarters and then gradually reverts. Quantitatively, LP-IV responses are systematically larger than Proxy-SVAR responses (approximately 2 to 3 times), with the largest gap in the monthly high-frequency setting (VIX peak 7.45 vs.\ 2.62). A factor-augmented VAR using principal components from the 122-series FRED-MD panel does not close this gap, while adding the same factors as controls in LP-IV reduces the discrepancy. This pattern is consistent with attenuation from VAR propagation restrictions even after enriching the information set. Decomposing the high-frequency instrument into pure policy and information-effect components shows that pure tightening raises uncertainty more strongly, while the information channel dampens the VIX response and helps interpret the price puzzle. Finally, we propose a smoothed high-dimensional LP-IV estimator that preserves the consistency of the LP while delivering smooth impulse responses. |
| Presented by: Alessia Paccagnini, University College Dublin |
U.S. Fiscal Spillovers to Emerging EconomiesAbstractThis paper studies the international spillovers of U.S. fiscal policy volatility shocks. We estimate U.S. fiscal reaction functions with time-varying volatility, and recover innovations to both the level and the volatility of several fiscal policy instruments. Embedding these shocks in a Panel VAR for emerging market economies (EMEs), we find that a fiscal volatility shock (rivaling the 2011 debt-ceiling episode) reduces output by about 0.4 percent and investment by nearly 0.7 percent after one year. These effects persist for two years and propagate primarily through global financial conditions: the U.S. BAA corporate spread rises significantly, tightening borrowing costs in EMEs. Fiscal volatility shocks account for 6 percent of output fluctuations in EMEs, comparable to U.S. monetary policy shocks. The impact varies systematically with country characteristics: effects are lower in countries with inflation-targeting frameworks, fiscal rules, and lower trade openness. |
| Presented by: Juan Urquiza, Pontificia Universidad Catolica de Chile |
| Session 102: HIGH-DIMENSIONAL METHODS June 25, 2026 10:15 to 12:00 Location: D-106 |
| Session Chair: Javier Fuertes Pina, Universidad Carlos III de Madrid |
Forecasting Inflation with the Iterative fused LASSO: the benefits of regularized regression methodsAbstractIn this paper we evaluate a high-dimensional time series modeling framework that jointly addresses three central challenges: structural break detection, outlier handling, and variable selection. We formulate the problem as a regularized least squares estimator that combines adaptive LASSO sparsity with fused LASSO penalties to capture piecewise-constant coefficient dynamics. Our main contribution is an efficient algorithm that scales to high-dimensional settings within practical runtimes. Besides the framework itself, we illustrate the approach using one empirical application to U.S. monthly inflation, covering out-of-sample forecasting. We benchmark our forecasts against Random Forest, a competitive nonlinear learning method that has demonstrated strong and consistent performance across a wide range of prior studies. Overall, the proposed method delivers improved forecast performance while preserving interpretability through a fundamentally linear regression structure, fitted via nonlinear optimization and augmented with data-driven regime and outlier adjustments. These results indicate that the methodology provides a practical and robust option for high-dimensional macroeconomic estimation and forecasting. |
| Presented by: Angelo Sérgio Pereira, Pontifícia Universidade Católica do Rio de Janeiro |
Broken Adaptive Ridge in Time Series: the TS-BARAbstractWe extend the Broken Adaptive Ridge (BAR) estimator to high- and ultra–high-dimensional vector autoregressions (VARs). BAR implements an adaptive L2 penalty via iteratively reweighted ridge, strongly shrinking small coefficients while leaving large (autoregressive) effects almost unbiased. Under physical dependence and row-wise weak sparsity, we derive non-asymptotic prediction and estimation error bounds allowing sub-exponential dimensional growth under a restricted eigenvalue and beta–min condition, but without any "irrepresentable" assumption. Monte Carlo experiments for a wide range of high-dimensional VAR designs show that BAR generally outperforms Lasso, adaptive Lasso, and SCAD in terms of root mean square forecast error. An empirical application to U.S. inflation forecasting confirm that BAR delivers robust improvements in out-of-sample performance over standard penalised VAR approaches. |
| Presented by: Gianluca Cubadda, Tor Vergata University of Rome |
High-dimensional regression model estimation using regularization by thresholdingAbstractWe propose a regularized OLS-type estimator for variable selection and es- timation in high-dimensional linear regression models. Our approach builds on recent advances in sparse covariance and precision matrix estimation, em- ploying a thresholding-based regularization scheme that jointly incorporates all available regressors. This allows for a theoretically grounded estimation framework that extends beyond existing procedures, and accommodates more general assumptions on the regressors, including dependent stochastic pro- cesses and fat tails. We establish asymptotic validity of the estimator and de- velop an enhanced procedure based on a Bonferroni-type testing rule, which improves the identification of true regressors. We assess the finite sample per- formance of our approach by an extensive set of Monte Carlo experiments, also comparing it with other alternatives proposed in the literature. Finally, we illustrate the practical relevance of the methodology in an empirical macroe- conomic forecasting application. |
| Presented by: Yiannis Dendramis, Athens University of Economics and Business |
Nonparametric Identification and Convergence Rates for Elasticities under EndogenityAbstractElasticities are fundamental objects in empirical economics, yet their nonparametric identification under endogenous regressors remains largely unexplored. This paper develops a general framework for identifying conditional average elasticities using a control-function approach, without imposing parametric restrictions on the structural outcome equation. We propose an estimation procedure based on a sparse multi-output Deep Neural Network suited to nonlinear and high-dimensional demand systems. Extending existing results to the multi-output setting, we derive an $L^2$ convergence rate of $O_{\mathbb{p}}(m^{-1/8})$ for the elasticity functional. Monte Carlo simulations show that the estimator recovers elasticity matrices accurately across a range of data-generating processes, including log-log, random utility models, and nonlinear generic specifications, particularly in large samples. The results provide a flexible and theoretically grounded approach to elasticity estimation under endogeneity. |
| Presented by: Javier Fuertes Pina, Universidad Carlos III de Madrid |
| Session 103: HOUSEHOLD CONSUMPTION June 25, 2026 10:15 to 12:00 Location: B129 |
| Session Chair: Keshav Sureka, ETH Zurich |
Heterogeneous Earnings RiskAbstractWhile recent progress has improved our understanding of heterogeneity in earnings dynamics, heterogeneity in earnings risk remains largely unexplored. This paper introduces a framework to jointly estimate earnings dynamics and risk by combining monthly earnings data from the Survey of Income and Program Participation (SIPP) with four-month-ahead earnings expectations from the Survey of Consumer Expectations (SCE). We find that both realized earnings dynamics and individuals’ ability to predict future earnings vary systematically across the earnings distribution. As a result, earnings risk inferred solely from earnings dynamics differs substantially from risk that incorporates individual expectations. Low-income individuals experience more volatile earnings, but their expectations are also more predictive. Accounting for expectations reduces predicted earnings risk by 75% on average over a four-month period, with the largest reductions in the lower tail. Thus, models based only on realized earnings dynamics overstate household earnings risk, with implications for welfare and consumption–saving behavior. |
| Presented by: Eva Janssens, University of Michigan |
Can Digital Payment Stimulate Household Consumption Growth Persistently? An Inconsistent Time Preference PerspectiveAbstractThis study examines whether digital payments can sustainably stimulate household consumption by integrating behavioral economics with intertemporal choice. We develop a multi-period self-control model with quasi-hyperbolic discounting, in which frictionless digital transactions reduce the “pain of paying” and heighten present bias. The framework predicts a front-loaded consumption response after adoption, followed by a gradual attenuation of consumption growth as households learn (imperfect) self-control and tighten liquidity. We test these predictions using unbalanced household panel data from the 2013–2019 China Household Finance Survey (CHFS), exploiting within-household variation with household fixed effects and city×year fixed effects. Consistent with the model, digital payment adoption is associated with higher consumption levels: in the benchmark specification, first-year adopters exhibit about a 10% increase in total consumption and longer-term users about a 13–14% increase. The growth effects, however, are short-lived: consumption growth rises significantly in the adoption year (around 6–7 percentage points) but becomes statistically weak thereafter. Category-level results show that the level effects are disproportionately concentrated in hedonic spending (roughly 26%) relative to essential “life” consumption (about 7%), and the long-run growth deceleration is especially pronounced for hedonic categories. Mechanism evidence is consistent with behavioral channels: the response varies systematically with proxies for present bias and credit-card usage, and households appear to adjust by curbing consumption-related borrowing and re-optimizing liquid asset holdings. Heterogeneity further indicates stronger effects in rural areas (suggesting opportunity-set expansion), among households with weaker self-control capacity (lower education), and with more favorable income expectations (e.g., improved health). Instrumental-variable estimates using city-level adoption intensity corroborate the main patterns and reinforce concerns about the sustainability of payment-driven consumption growth. |
| Presented by: Shuoxun Zhang, Sichuan University |
Energy Consumption and Inequality in the U.S.: Who are the Energy Burdened?AbstractUsing a comprehensive definition of energy consumption that includes both residential energy and gasoline for transport, we identify 15% if households in the PSID as energy burdened (EB). Logit analysis reveals that being nonwhite, single with dependents, receiving public assistance, lacking post-secondary education, and unemployment significantly increase the probability of being EB. We document four key empirical facts: (1) EB status exhibits substantial persistence, with 49% of households remaining burdened 2 years later; (2) EB households have significantly higher marginal propensities to consume (19.4 vs 8.8 cents per dollar) and marginal propensities to consume energy (4.6 vs 0.9 cents per dollar); (3) EB households experience lower expected energy consumption growth despite higher expected income growth; and (4) EB households face more volatile income but similar consumption volatility after controlling for household fixed effects. From 1999 to 2019, energy consumption inequality increased 80% substantially exceeding income inequality growth (29%). Residential energy inequality rose and then decline, while transport energy inequality increased steadily by 40%. These findings provide guidance for theoretical models, suggesting the need for non-homothetic preferences, occasionally binding credit constraints, precautionary savings motives, and persistence mechanisms. |
| Presented by: Cristina Fuentes-Albero, Board of Governors of the Federal Reserv |
Negotiating clean energy: Women’s Bargaining Power and the impact of energy transition subsidyAbstractAccess to clean cooking fuel remains a persistent challenge for billions globally, with women bearing a disproportionate share of the associated health burden. This paper provides an empirical analysis of India’s Pradhan Mantri Ujjwala Yojana (PMUY)—the world’s largest initiative to promote LPG adoption among poor households. Using a difference-in-difference approach with propensity score matching on nationally representative survey data, we quantify the impact of PMUY on clean fuel uptake. The study also examines the central role of women’s bargaining power within households and the presence of cheap alternative fuels as key factors that modulate the effectiveness of the program. We find a statistically significant increase in LPG adoption among eligible households, with program effects amplified in those households where women have greater agency. These results underscore the importance of targeting intra-household dynamics and local fuel availability in designing and implementing successful clean cooking interventions. |
| Presented by: Keshav Sureka, ETH Zurich |
| Session 104: IDENTIFICATION OF TREATMENT EFFECTS June 25, 2026 10:15 to 12:00 Location: B008 |
| Session Chair: Michela Bia, LISER and University of Luxembourg |
Non-parametric Identification of Dynamic Treatment Effects: a Graphical Approach to CausalityAbstractThis paper proposes a unified framework to study different types of dynamic treatment effects (DTE), extending the dynamic potential outcome (DPO) framework developed by \cite{torgovitsky19}. The DPO framework is used to study the direct causal effect of the immediate preceding past on the present in his paper. In this paper, I first show that DTE such as the indirect causal effect of the distant past on the present can be non-parametrically identified in a DPO framework which does not necessarily assume that the distant past enters the function of the present potential response. I then develop an algorithm leveraging a novel graphical representation of the DPO model. This graph-based algorithm provides a computationally tractable means of scaling the DPO framework beyond binary variables to multi-valued variables, circumventing the curse of dimensionality inherent in dynamic treatment effect studies. |
| Presented by: Chenyue LIU, University of Toronto |
Testing Exclusion and Shape Restrictions in Potential Outcomes ModelsAbstractExclusion and shape restrictions play a central role in defining causal effects and interpreting estimates in potential outcomes models. To date, the testable implications of such restrictions have been studied on a case-by-case basis in a limited set of models. In this paper, we develop a general framework for characterizing sharp testable implications of general support restrictions on the potential response functions, based on a novel graph-based representation of the model. The framework provides a unified and constructive method for deriving all observable implications of the modeling assumptions. We illustrate the approach in several popular settings, including instrumental variables, treatment selection, mediation, and interference. As an empirical application, we revisit the US Lung Health Study and test for the presence of spillovers between spouses, specification of exposure maps, and persistence of treatment effects over time. |
| Presented by: Kirill Ponomarev, University of Chicago |
Estimating Causal Effects of Discrete and Continuous Treatments with Binary InstrumentsAbstractWe propose an instrumental variable framework for identifying and estimating causal effects of discrete and continuous treatments with binary instruments. The basis of our approach is a local copula representation of the joint distribution of the potential outcomes and unobservables determining treatment assignment. This representation allows us to introduce an identifying assumption, so-called \emph{copula invariance}, that restricts the local dependence of the copula with respect to the treatment propensity. We show that copula invariance identifies treatment effects for the entire population and other subpopulations such as the treated. The identification results are constructive and lead to practical estimation and inference procedures based on distribution regression. An application to estimating the effect of sleep on well-being uncovers interesting patterns of heterogeneity. |
| Presented by: Ivan Fernandez-Val, Boston University |
Harnessing Genetic Variants for Local Average Treatment Effect EstimationAbstractWhen multiple instruments are available, conventional instrumental variable estimators aggregate across potentially heterogeneous margins of compliance and may yield effects that lack a clear economic interpretation. The problem is compounded when some instruments violate the exclusion restriction, as is common in certain empirical contexts such as those using genetic variants as instrumental variables. We propose a clustering-based plurality framework for instrumental variable estimation that jointly addresses instrument heterogeneity and invalid instruments. Rather than imposing a single common causal parameter, our approach groups instruments according to similarity in first-stage and applies a plurality rule on subgroups with similar reduced-form to identify locally valid subsets. This yields a collection of margin-specific local average treatment effects instead of a single pooled estimate. We extend plurality-based identification to settings with non-mutually exclusive instruments, such as Mendelian Randomization designs where all individuals are exposed to all genetic variants. We illustrate the method in a two-sample Mendelian Randomization analysis of the causal effect of educational attainment on smoking participation. Our results confirm a negative causal effect of education on smoking that remains robust under pleiotropy-robust estimators, while revealing substantial heterogeneity across instrument-defined margins that is masked by pooled IV approaches. The framework provides a unified way to interpret and validate high-dimensional instruments in the presence of both treatment effect heterogeneity and potential violations of exclusion. |
| Presented by: Michela Bia, LISER and University of Luxembourg |
| Session 105: INEQUALITY AND MOBILITY June 25, 2026 10:15 to 12:00 Location: D-113 |
| Session Chair: Tomeu López-Nieto Veitch, University of Bologna |
Bargaining Power Metrics from a Distributional PerspectiveAbstractMeasuring intra-household bargaining power—a key topic in household and labor economics—is commonly done by aggregating survey responses on decision-making in areas such as expenditures, time use, and women’s labor market participation into a composite index. We identify limitations of existing indices and propose a distribution-based alternatives that capture the informational content of response patterns inspired by entropy logic, emphasizing deviations from common behaviors to reflect stronger or weaker individual agency relative to social norms. Our approach further improves the information content of the index by incorporating decision-making reports from both spouses and using the full range of the decision question scale, unlike previous studies relying on female responses and collapsing the scale into a simple decision dichotomy. Beyond outlining the conceptual foundations of the new approach, we use quantile regressions to show that the distribution-based indices consistently reflect known determinants of bargaining power and capture shifts in bargaining positions more reliably than existing benchmarks. |
| Presented by: Natalia Radchenko, American University, Washington DC |
Marital Sorting, Household Inequality and SelectionAbstractUsing CPS data for 1976 to 2022 we explore how wage inequality has evolved for married couples with both spouses working full time full year, and its impact on household income inequality. We also investigate how marriage sorting patterns have changed over this period. To determine the factors driving income inequality we estimate a model explaining the joint distribution of couples' wages which accounts for the spouses' employment decisions. This requires that we introduce and estimate a bivariate regression model with a bivariate selection rule. We find that income inequality has increased for these households and increased assortative matching of wages has exacerbated the inequality resulting from individual wage growth. We find that positive sorting partially reflects the correlation across unobservables influencing the wages of both members of the marriage. We decompose the changes in sorting patterns over the 47 years comprising our sample into structural, composition and selection effects and find that the increase in positive sorting primarily reflects the increased skill premia for both observed and unobserved characteristics. |
| Presented by: Aico van Vuuren, University of Groningen |
A Trajectories-Based Approach to Measuring Intergenerational MobilityAbstractThis paper develops an approach to intergenerational mobility in which the trajectories of parental incomes during childhood and adolescence are the conditioning objects for characterizing dependence across generations. We use functional regression methods to produce an intergenerational elasticity curve that measures how marginal changes in income at each age affect expected offspring permanent income. Using the PSID, estimates of this curve exhibit near monotonicity with respect to age, so that parental incomes in middle childhood and adolescence have larger marginal effects than incomes in early childhood. When interactions are allowed to occur between incomes at different ages, we find a complex pattern of substitutability between incomes at ages that are close in time versus complementarity between parental incomes for ages early childhood and adolescence. Qualitatively similar results hold for offspring education while we do not find evidence of age-specific effects for occupation. We conclude that important information about the links between parental incomes and children exists beyond the scalar characterization of parental permanent income. |
| Presented by: Yoosoon Chang, Indiana University |
Genetically Informed Estimation of Marginal Returns to EducationAbstractReturns to education have long been a central focus of social science research due to their implications for productivity, social mobility, and inequality. A persistent challenge in this literature is the inability to fully account for heterogeneity in skills and initial endowments, which may confound estimates of causal returns. This paper addresses this limitation by incorporating genetic information—specifically, the Educational Attainment Polygenic Index (EA PGI)—to study the role of genetic predisposition in shaping the returns to education. We develop a novel extension of the Marginal Treatment Effects (MTE) framework that introduces a categorical moderator, building on a monotone “warp” mapping that aligns group-specific resistance ranks to a common pooled scale. This allows for valid comparisons of marginal returns across genetic groups with different selection patterns. Exploiting the mid-20th-century expansion of university access in the United Kingdom, we compute the reduction in distance to university as an instrument for college and find a significant negative gene-environment interaction. Causal returns to college are substantially higher for individuals with low genetic predisposition to education compared to those with high predisposition. Using a Shapley-based decomposition, we show that this gap is driven almost entirely by heterogeneity in returns (GxE) rather than by differences in sorting into college (rGE). These findings suggest that higher education acts as an equalizer, playing a compensatory role that mitigates genetic inequality. |
| Presented by: Tomeu López-Nieto Veitch, University of Bologna |
| Session 106: INSTRUMENTAL VARIABLES 2 June 25, 2026 10:15 to 12:00 Location: B009 |
| Session Chair: Deniz Ozabaci, University of New Hampshire |
Instrumental variable estimation via a continuum of instruments with an application to estimating the elasticity of intertemporal substitution in consumptionAbstractThis study proposes new instrumental variable (IV) estimators for linear models utilizing a continuum of instruments. The effectiveness of the new estimation method is attributed to the unique weighting function employed in the minimum distance objective functions. The proposed estimators enjoy analytical formulas and are nuisance-parameterfree, avoiding the choice of an arbitrary number of moments or a bandwidth as in previous literature. They are robust to weak instruments and heteroskedasticity of unknown form. Moreover, they are robust to the high dimensionality of excluded exogenous variables. Further, inference drawn from these estimators is also straightforward. Comprehensive Monte Carlo simulations confirm that the proposed estimators exhibit excellent finite-sample properties and outperform alternative estimators over a wide range of cases. The new estimation procedure is then applied to gauge the elasticity of intertemporal substitution (EIS) in consumption, a parameter of central importance in both macroeconomics and finance. For quarterly data of the U.S. from Q4 1955 to Q1 2018, the EIS estimates obtained through our approach exceed one and are statistically significant. These findings persist across model transformations, distinct sets of IVs, various data structures, and different data ranges. |
| Presented by: Carlos Velasco, Universidad Carlos III de Madrid |
Econometric Inference Using Hausman InstrumentsAbstractWe examine econometric inferential issues with Hausman instruments. The instrumental variable (IV) estimator based on Hausman instrument has a built-in correlation across observations, which may render the textbook-style standard error invalid. We develop a standard error that is robust to these problems. Clustered standard error is not always valid, but it can be a good pragmatic compromise to deal with the interlinkage problem if Hausman instrument is to be used in econometric models in the tradition of Berry, Levinsohn, and Pakes (1995). Additionally, we find that the Hausman IV is in fact equivalent to the judge IV proposed by Kling (2006), which broadens the implication of our results beyond the industrial organization literature. |
| Presented by: Ruoyao Shi, University of California Riverside |
Two-Sample IV: Efficient Two-Step Estimation and Tests for Overidentification and Weak-InstrumentsAbstractTwo-sample IV is a popular estimation method when the outcome and treatment variables are available in different samples, whereas instruments are available in both samples. The standard estimator is two-sample two-stage least squares esti mator, which is efficient under homoskedasticity and homogeneity of the samples. We develop a robust two-step procedure for efficient estimation under general het eroskedasticity and heterogeneity of the samples, and propose a related two-sample Hansen overidentification test. A key feature of our approach is that only summary statistics form the linear regressions of the reduced form and first-stage in the two samples are needed. These are the six objects of the estimated coefficient vectors, and the homoskedastic and heteroskedasticity robust estimated variance matrices. We further show that the first-stage F-statistic in the treatment sample can be used as a test for weak instruments in the standard way under homoskedasticity and homogeneity, with the relative bias here a proportional bias. We propose an extension of the effective F-statistic of Montiel Olea and Pflueger (2013) for the heteroskedastic case, following the generalization in Windmeijer (2025). We illus trate the estimators and tests in an application studying the effect of education on voting behavior from Marshall (2019), with cluster robust inference. |
| Presented by: Fatima Kasenally, University of Oxford |
Nonparametric Sample Selection Models with Endogenous RegressorsAbstractFollowing a triangular set up similar to Newey et. al. (1999), this paper presents estimators for nonparametric sample selection models with endogenous regressors, under different additivity and error separability constraints. We propose estimators for nonparametric sample selection model with nonseparable errors while allowing for endogenous regressors, as well as for its nonparametric and additive nonparametric counterparts. We achieve identification combining the work by Lewbel (2012), and Lewbel et. al. (2023), as well as Jun et. al. (2016). Hence, we can achieve identification with minimal to no exclusion restrictions. We show that estimators we propose are consistent and asymptotically normal. Using Monte Carlo simulations, we also present the finite sample properties of the estimators, which support our asymptotic findings. Our result show that, under the full additivity constraint, the estimators are oracle efficient. As for models with nonseparable error components, we show that the efficiency is lowest among the other models we consider in this paper, as expected. Finally, we apply our estimators to an empirical problem and consider how maternal employment and child care affect children’s cognitive ability development. |
| Presented by: Deniz Ozabaci, University of New Hampshire |
| Session 107: MACRO-FINANCE MODELS June 25, 2026 10:15 to 12:20 Location: D-112 |
| Session Chair: Daniel Gründler, |
Adversarial Density Forecast for Macro-Financial RisksAbstractForecasting macroeconomic risks, and especially tail risks, requires capturing complex distributional dynamics but is constrained by the limited sample sizes typical of macro-financial time series. While parametric benchmarks like skew-t distributions offer stability at the cost of shape rigidity, flexible nonparametric methods often suffer from finite-sample overfitting and tail miscalibration. This paper proposes a Penalized Conditional Wasserstein Generative Adversarial Networks (PcWGAN) for density forecasting. We formalize the estimator as a penalized sieve minimizer of the conditional Wasserstein-1 distance, augmented by moment-quantile penalties. Theoretically, we establish that this penalization acts as a Tikhonov regularizer that restricts the estimator to a “moment tube” with reduced metric entropy, thereby tightening finite-sample oracle risk bounds without altering the asymptotic consistency target. To address the slow convergence rates typical of nonparametric sieves, we implement a second-stage anchored monotone rescaling with interleaved cross-fitting. We show that this semiparametric calibration improves tail reliability without materially distorting the global shape learned in the first stage. Monte Carlo experiments indicate that penalization stabilizes adversarial learning in small samples, and that calibration delivers additional gains in tail matching relative to the uncalibrated distribution; under bimodal misspecification it also performs well against parametric benchmarks. In an empirical application to U.S. Outlook-at-Risk, the proposed approach combines flexible distributional learning with improved tail reliability required for policy analysis, supporting risk measurement in macro-financial settings. |
| Presented by: Tae-Hwy Lee, University of California Riverside |
When Long Run Trends are Unknown: Bond Pricing ImplicationsAbstractThis paper assesses the informativeness of the Treasury yield curve about the long-run real interest rate, r-star, when bond investors are uncertain about its value. We propose a macro-finance model where inflation, growth, and the monetary policy rate are driven by a combination of persistent trends and transitory cycles. Investors only observe the aggregate macroeconomic variables but infers trends and cycles to price bonds. In spite of imperfect information, our model preserves the simplicity of standard affine term structure models. Our estimation reveals wide investors uncertainty about r-star that does not disappear over time, and an increasing r-star trend before the Volcker era, largely contrasting with perfect information estimates. Because investors confuse trends with cycles, the yield curve can under or overreact to structural monetary policy shocks. |
| Presented by: Borel Ahonon, McGill University |
When The Treasury Does Monetary PolicyAbstractDebt management decisions have macroeconomic effects comparable to monetary policy. Using high-frequency movements in interest rate futures around U.S. Treasury issuance announcements, we identify a Treasury policy shock—an unanticipated change in public debt supply across maturities. A shock that raises the five-year Treasury yield transmits strongly to corporate borrowing rates, tightens credit conditions, and lowers industrial production significantly. These effects closely mirror those of a conventional monetary policy tightening. While long-term treasury yields increase significantly, the shock has minimal effects on short-term interest rates. We show that this pattern reflects the Federal Reserve’s sterilization of short-term Treasury issuance, while issuance at longer maturities is only partially offset. |
| Presented by: Kevin Pallara, Bank of Italy |
Sampling Frequency and Parameter Information Under Temporal Aggregation in Continuous-Time Macroeconomic ModelsAbstractWhenever a macroeconomic model is taken to data at a chosen sampling frequency, a set of design questions arises: which parameters benefit from more frequent observation, which from wider spacing, and how does the observation structure—especially temporal aggregation of flow variables—reshape these tradeoffs? We develop a framework for answering such questions that pairs exact finite-sample information calculations with two diagnostic decompositions that make the mechanisms transparent. A frequency-domain decomposition attributes each parameter’s information to bands of the spectrum, while a sensitivity–collinearity decomposition of the Cramér–Rao bound separates intrinsic signal strength from uncertainty created by confounding with nuisance parameters. Together, these diagnostics produce parameter-by-parameter frequency requirements and a confounding diagnosis that characterize properties of the model rather than merely features of a particular dataset. In particular, temporal aggregation interacts with sampling frequency to generate parameter confounding in ways that depend on model structure, and mixed-frequency sampling can alleviate this confounding through nuisance-parameter relief. We validate the approach with extensive Monte Carlo experiments for CARMA and DSGE models calibrated to a long-run U.S. GDP series; the information-based predictions closely track simulation-based precision across sampling designs. The methodology is computationally inexpensive, applies broadly to linear Gaussian models in state-space form - including those with temporally aggregated flow variables - and provides a practical tool for researchers designing estimation strategies and diagnosing identification bottlenecks in continuous-time macroeconomic applications. |
| Presented by: Nikolay Iskrev, Bank of Portugal |
Endogenous TVPVARsAbstractI propose a new time-varying parameter VAR (TVPVAR) model in which the time variations in the autoregressive coefficients and the components of the covariance matrix are endogenous with respect to structural shocks. In the first empirical application, I find that the endogenous response of the model parameters reinforced the recent increase in the pass-through of oil price shocks and their contribution to U.S. inflation during this period. In a second exercise, I introduce a new algorithm for structural scenario analysis and show that exogenous as well as policy induced uncertainty play important roles for euro area monetary transmission. |
| Presented by: Daniel Gründler, |
| Session 108: MACROECONOMIC POLICY AND SHOCK IDENTIFICATION June 25, 2026 10:15 to 12:00 Location: D-108 |
| Session Chair: Win Monroe, Copenhagen Business School |
A Public-Private Partnership? Central Bank Funding and Credit SupplyAbstractWe exploit the surprise announcement and subsequent amendment of a central bank funding scheme to test how public liquidity provision affects credit market outcomes. Contrary to the notion that public liquidity is primarily a substitute for private liquidity, banks that are more exposed to stress in private wholesale funding markets use less central bank funding. We rationalise this pattern by establishing an “equilibrium channel” of public liquidity. The mere availability of central bank funding reduces the cost of private wholesale funding. This stimulates lending by banks exposed to wholesale funding, regardless of whether they actually use the central bank funding. Using a surprise amendment to the design of the scheme, we show that the “strings attached” to central bank funding help to explain why it is an imperfect substitute for private funding. |
| Presented by: Win Monroe, Copenhagen Business School |
The Deposit Franchise and the Risk-Taking Channel of Monetary PolicyAbstractWe develop a tractable model in which a bank’s deposit franchise shapes its risk-taking response to monetary policy. Banks with weaker pass-through to deposit rates (lower deposit betas) see larger profit gains when rates rise and therefore reduce risk-taking more after contractionary shocks. We test this channel using the Federal Reserve's confidential loan-level data, interacting high-frequency monetary policy surprises with pre-determined banks’ deposit betas, in regressions saturated with bank and borrower-time fixed effects. We find that low-deposit-beta banks reduce risk-taking significantly more following monetary tightening, confirming that the deposit franchise plays a crucial role in the interaction of monetary policy and financial stability. In a horse race against bank capital-based explanations of risk-taking (e.g., search-for-yield), our deposit-franchise mechanism retains independent explanatory power. |
| Presented by: Ricardo Duque Gabriel, Federal Reserve Board of Governors |
The Uncertain Exorbitant Privilege and Duty AbstractThis paper examines how shocks to exchange rate and macroeconomic uncertainty affect the United States’ net foreign asset (NFA) position. I employ a structural VAR that combines external instruments, narrative (event) inequalities, and shock-dependent restrictions to address endogeneity in uncertainty measures and their links to net portfolio flows. I find that both exchange rate and macroeconomic uncertainty shocks reduce the United States’ NFA deficit, with macroeconomic uncertainty exerting a larger and more persistent effect than exchange rate uncertainty. The NFA improvement is accompanied by greater macroeconomic fluctuations in the short run but dampened volatility at longer horizons. In addition, macroeconomic uncertainty raises exchange rate volatility, whereas higher exchange rate uncertainty tends to dampen macroeconomic uncertainty. These results are consistent with the convenience-yield and dominant-currency pricing literatures, heightened volatility rebalances foreign demand toward United States safe assets and transmits through global terms of trade and financing conditions. |
| Presented by: Julian Fernandez, Pontifica Universidad Javeriana |
Fiscal Announcements and Output: The Role of Debt DynamicsAbstractUsing data for 16 OECD countries (1981--2011), we examine the short-run effects of fiscal consolidation announcements on GDP growth. We distinguish between expenditure cuts and tax increases and assess how shocks transmit through changes in the debt-to-GDP ratio, separating direct and indirect effects. Identification relies on a panel model with country random effects and year fixed effects, exploiting within-country variation in narrative fiscal announcements to identify the debt-transmission channel. We find no significant direct effect of announcements on output. However, spending cuts produce negative indirect effects through the debt ratio, leading to overall contractionary outcomes, while tax increases have negligible total effects. Simulation-based sensitivity analysis, following \citet{imai2010general} and \citet{ricciardi2019bayesian}, confirms the robustness of the indirect debt channel. |
| Presented by: Giovanni Trovato, University of Rome |
| Session 109: MIGRATION AND LABOR MARKETS June 25, 2026 10:15 to 12:00 Location: E002 |
| Session Chair: Sabrina Di Addario, Bank of Italy |
The Gender Wage Gap in a Highly Regulated MarketAbstractThis paper investigates the gender wage gap in a highly regulated labor market, focusing on public school teachers in Mexico and drawing on rich administrative payroll data covering eight years and more than 850,000 teachers. We estimate a 2% gender wage gap for equal work, based on comparisons of earnings in identical teaching positions within the same school. Gender differences in advancement through the horizontal promotion system explain about half of the gender wage gap for equal work, while the remaining portion is attributable to pay disparities among older teachers. We identify larger gaps—around 5%—in total earnings, driven by differences in multiple job holding and higher earnings in secondary positions. Motherhood reduces female teachers’ earnings through declines in wage rates and employment intensity, though these effects are smaller than those documented for the general population. We identify one channel underlying the wage decline: women’s increased preference for working in localities with greater amenities. |
| Presented by: Paola Bordon, Universidad de Chile |
The Economic Value of EU Citizenship: Evidence from the 2004 Enlargement and the German Labor MarketAbstractImmigrants from non-EU countries face considerable barriers in the German labor market. This paper develops a wage posting model to illustrate the underlying mechanisms and provides reduced-form evidence of the causal effects of EU citizenship on labor market outcomes. I exploit the 2004 enlargement of the European Union, which granted EU status to immigrants from ten Eastern European countries residing in Germany. Using a difference-in-differences framework as well as an event-study and rich administrative data from the Sample of Integrated Labour Market Biographies (SIAB), I compare the labor market trajectories of this group before and after enlargement with those of non-EU immigrants who did not benefit from such a change in legal status. The results show that EU citizenship increases wages by 3.6 percent. Unemployment rates increase by 0.7 percentage points. The effects are persistent over time. |
| Presented by: Niklas Isaak, RWI - Leibniz Institute for Economic Research |
The role of indebtedness in migration decisionsAbstractIn this paper, we estimate the role of indebtedness in the individuals' migration decisions. To this end, we merge three unique datasets. The first is a matched employer-employee dataset covering all Hungarian workers. The second contains information on mortgage loans for all Hungarian citizens, and the third one provides data on employment history of Hungarian workers in Austria. We exploit two exogenous shocks. The first is the opening of the Austrian labour market to Hungarian workers in 2011, and the second is the sudden depreciation of the Hungarian forint against the euro and Swiss franc during the great financial crisis. We find that holding a previously contracted foreign-currency (FX) mortgage loan increases the probability of working in Austria. If an individual had taken out FX mortgage loan before working in Austria, this increases the duration of employment there by 10%. The effect is increasing in the size of the loan principal. |
| Presented by: Lajos Szabo, Magyar Nemzeti Bank |
Capital Across Borders, Jobs at Home: The FDI-Unemployment Nexus in the OECDAbstractThis paper examines the FDI-unemployment nexus in OECD countries from both theoretical and empirical perspectives. We first build a search and matching model which accounts for inward and outward FDI capital stocks and identifies key channels through which FDI affects unemployment. In a subsequent empirical panel ARDL estimation, we show that both inward and outward FDI can reduce unemployment conditional on technological and institutional factors. Inward FDI is most unemployment-reducing in less innovative and less technologically advanced countries, while for outward FDI this is the case in technologically more advanced countries with sufficient absorptive capacity and stronger bargaining institutions. For inward (outward) FDI, the technology diffusion (reverse spillover and head-office) channel dominates these long run effects. Our findings imply that policies which strengthen absorptive capacity, diffusion, and domestic linkages can make FDI more employment-friendly, whereas in advanced economies the composition and integration of FDI may matter more than broad FDI-attraction alone. |
| Presented by: Kaan Celebi, University of Technology Chemnitz |
| Session 110: MONETARY POLICY COUNTERFACTUALS June 25, 2026 10:15 to 12:00 Location: D-110 |
| Session Chair: Elena Afanasyeva, Federal Reserve Board |
Two sides of the same coin: expected reaction functions and perceived monetary policy shocksAbstractTo identify the causal effect of monetary policy, researchers often use either high-frequency or narrative monetary policy shock measures as instruments. By their nature, these instruments tend towards being either weak or endogenous, respectively. This paper proposes a third approach. First, we develop a framework that specifies the links between true and perceived FOMC reaction functions. Second, we use this framework to develop an econometric model to estimate the perceived FOMC reaction functions implied by monetary policy expectations. This allows us to extract both expected reaction function coefficient estimates and two perceived monetary policy expectations shock measures, which we show to be both relevant and exogenous instruments. Finally, using our two instruments, we demonstrate that the impulse response function estimates are consistent with New Keynesian theory. |
| Presented by: Christopher Sutherland, Bank of Canada |
A New Interpretation of the Volcker Disinflation, Money Growth Targeting, and the Monetarist ExperimentAbstractTo combat high US inflation, Federal Reserve Chairman Paul Volcker implemented what was termed the Monetarist Experiment from October of 1979 to September of 1982. While the intended disinflation was achieved, it was accompanied by large volatility in macroeconomic aggregates, including a double dip recession. This paper shows that the monetary policy regime of this period resulted in self-fulfilling expectations that contributed to this volatility. New theoretical results show that the Federal Reserve’s policy of targeting the growth rate of simple-sum money in a high inflation environment can yield indeterminacy. In contrast, we show that targeting Divisia measures of money growth are much less susceptible to indeterminacy. Our analysis also helps to explain the difficulty the Federal Reserve had in hitting its money growth targets. Ultimately, our findings call into question some common wisdom derived from the Monetarist Experiment about the usefulness of monetary aggregates for policy. |
| Presented by: John Keating, University of Kansas |
Inference for Macroeconomic Policy CounterfactualsAbstractRecent innovations (McKay and Wolf (2023), Barnichon and Mesters (2023)) have generated considerable interest in policy counterfactuals based on macroeconometric tools. Computation of these counterfactuals constitutes a “regression in impulse response space” that resembles an instrumental variables regression. We use this platform to develop a suite of tools for frequentist inference about policy counterfactuals, including a bias-based test for weak identification, a test for whether a counterfactual can be implemented in the data, and methods for inference on the counterfactual responses under both weak and strong identification. To confront the challenge of weak identification, we provide a disciplined approach to select horizons that contain as much information as possible for the counterfactual of interest. These tools provide insights to help applied researchers design better specifications for counterfactual analysis. Only after applying the lessons we draw from our theoretical development can the empirical findings of McKay and Wolf (2023) be confirmed using inference methods robust to weak identification. |
| Presented by: Utso Pal Mustafi, Frankfurt School of Finance & Management |
News, Noise, and UncertaintyAbstractWe employ a parsimonious optimal filtering framework with time-varying volatility and investigate how financial uncertainty fluctuations affect the transmission of beliefs to the broader economy under imperfect information. In a model with noisy signals, changes in uncertainty shift the signal-to-noise ratio and hence the sensitivity of agents’ expectations to new information. When embedded into a canonical macroeconomic model with financial frictions, this framework yields state-dependent effects of volatility on shock amplification. When fundamental volatility shifts from median to high levels (e.g., during a financial crisis), the impact of noise shocks on investment more than doubles. When volatility is driven by pure noise, expectation shocks are dampened. In an estimation on actual corporate spreads and their survey-based expectations, we find that countercyclical fundamental volatility is supported by the data. |
| Presented by: Elena Afanasyeva, Federal Reserve Board |
| Session 111: SYSTEMIC RISK AND BUBBLES June 25, 2026 10:15 to 12:00 Location: D-111 |
| Session Chair: Anton Skrobotov, HSE University |
Conditional Method Confidence SetAbstractThis paper proposes a Conditional Method Confidence Set (CMCS) which allows to select the best subset of forecasting methods with equal predictive ability conditional on a specific economic regime. The test resembles the Model Confidence Set by Hansen et al. (2011) and is adapted for conditional forecast evaluation. We show the asymptotic validity of the proposed test and illustrate its properties in a simulation study. The proposed testing procedure is particularly suitable for stress-testing of financial risk models required by the regulators. We showcase the empirical relevance of the CMCS using the stress-testing scenario of Expected Shortfall. The empirical evidence suggests that the proposed CMCS procedure can be used as a robust tool for forecast evaluation of market risk models for different economic regimes. |
| Presented by: Lukas Bauer, University of Freiburg |
A Flexible Model of Tail DependenceAbstractTail dependence exhibits a complex structure, with both its strength and shape varying significantly across space and time. In particular, the dependence in one region of a joint distribution can differ markedly from that in an adjacent region. To address this, we propose a unique measure of \textit{asymmetric quantile interdependence} (AQI), derived from three simple axioms. AQI is applicable in higher dimensions and accommodates tail events at varying levels of severity. In the bivariate case, it nests the classic tail dependence coefficient, while in more general (multivariate) settings, it effectively captures the asymmetric nature of tail dependence in a simple and flexible manner. Simulation studies and empirical applications confirm the robustness of AQI and underscore its practical relevance. From an operations perspective, AQI provides a decision-oriented tool enabling planners to set thresholds, reserves and hedges directly from the statistic without fitting a full parametric model. |
| Presented by: Evarist Stoja, University of Bristol |
High-Dimensional Tests for Systemic Risk MeasuresAbstractMultivariate systemic risk measures like Conditional Value-at-Risk (CoVaR) or Marginal Expected Shortfall (MES) have become important tools in applied macro and financial economics to quantify interconnected risk. This paper proposes a flexible information efficiency test that allows to scrutinize the predictive power of different factors from the information set (e.g., climate risks) for the risk prediction. The test can be carried out simultaneously across multiple horizons and quantile levels controlling the type I error at the nominal level. It is implemented as a `many moment equality' test allowing the number quantiles, horizons, and covariates to grow with the sample size. Asymptotically valid critical values are derived via a small-large Block Multiplier Bootstrap. |
| Presented by: Daniel Gutknecht, Goethe University Frankfurt |
Confidence Sets for the Emergence, Collapse, and Recovery Dates of a BubbleAbstractWe propose constructing confidence sets for the emergence, collapse, and recovery dates of a bubble by inverting tests for the location of the break date. We examine both likelihood ratio-type tests and the Elliott-M\"uller-type (2007) tests for detecting break locations. The limiting distributions of these tests are derived under the null hypothesis, and their asymptotic consistency under the alternative is established. Finite-sample properties are evaluated through Monte Carlo simulations. The results indicate that combining different types of tests effectively controls the empirical coverage rate while maintaining a reasonably small length of the confidence set. |
| Presented by: Anton Skrobotov, HSE University |
| Session 112: WEATHER AND THE ECONOMY June 25, 2026 10:15 to 12:00 Location: D-114 |
| Session Chair: Marco Tibullo, Queen Mary University of London |
Saving for sunny days: The impact of climate (change) on consumer prices in the euro areaAbstractClimate (change) affects the prices of goods and services in different countries or regions differently. Simply relying on aggregate measures or summary statistics, such as the impact of average country temperature changes on HICP headline inflation, conceals a large heterogeneity across (sub-)sectors of the economy. Additionally, the impact of a weather anomaly on consumer prices depends not only on its sign and magnitude, but also on its location and the size of the area affected by the shock. This is especially true for larger countries or regions with diverse climate zones, since the geographical distribution of climatic effects plays a role in shaping economic outcomes. Using time series data of geolocations, we demonstrate that relying solely on country averages fails to adequately capture and explain the influence of weather on consumer prices in the euro area. We conclude that the information content hidden in rich and complex surface data can provide valuable insights into the role of weather and climate variables for price stability, and more generally may help to inform economic policy. |
| Presented by: Nazarii Salish, Universidad Carlos III de Madrid |
Estimating National Weather Effects from the Ground UpAbstractUnderstanding the effects of weather on macroeconomic data is critically important, but it is hampered by limited time series observations. Utilizing geographically granular panel data leverages greater observations but introduces a ``missing intercept'' problem: ``global'' (e.g., nationwide spillovers and GE) effects are absorbed by time fixed effects. Standard solutions are infeasible when the number of global regressors is large. To overcome these problems and estimate granular, global, and total weather effects, we implement a two-step approach utilizing machine learning techniques. We apply this approach to estimate weather effects on U.S. monthly employment growth, obtaining several novel findings: (1) weather, and especially its lags, has substantial explanatory power for local employment growth, (2) shocks to both granular and global weather have significant immediate impacts on a broad set of macroeconomic outcomes, (3) responses to granular shocks are short-lived while those to global shocks are more persistent, (4) favorable weather shocks are often more impactful than unfavorable shocks, and (5) responses of most macroeconomic outcomes to weather shocks have been stable over time but the consumption response has fallen. |
| Presented by: Daniel Wilson, Federal Reserve Bank of San Francisco |
Bivariate Weather Extremes and the US EconomyAbstractWe construct new proxies of extreme weather for macroeconomic analysis using the bivariate time-varying distribution of daily temperatures and precipitation in the US. We show that our indicators are robust to binning bias and uncover important seasonal variation in the tails of the bivariate distribution. Using panel local projections, we estimate season‑ and sector‑specific impulse response functions for US GDP and find strong heterogeneity in economic responses, with the most significant effects occurring in summer. In particular, extreme precipitation leads to sizable economic contractions when it coincides with extremely high temperatures, whereas effects outside summer are more limited and sector‑specific. Overall, the results show that considering temperature and precipitation jointly is essential for accurately assessing the economic impacts of weather extremes. |
| Presented by: Konstantin Boss, |
When Weather Hits the Tails: Sectoral Risk in Germany AbstractThis paper studies how innovations in temperature, precipitation, and sunshine anomalies reshape tail risk dynamics in German sectoral production. Using Quantile Local Projections, we estimate quantile-specific impulse responses of sectoral growth to these weather shocks, allowing the responses to vary across seasons and during episodes of extreme weather. We document that weather shocks affect sectoral downside and upside risks, with large heterogeneity in the direction, magnitude and persistence across industries, seasons and climate variables. In particular, mild temperature and sunshine innovations are often associated to a reduction in downside risk by lifting up the left tail; whereas extreme anomalies can reverse these gains and amplify vulnerability. Precipitation shocks have been found to have limited impact on sectoral growth, mostly negative, affecting the majority of sectors during autumn. |
| Presented by: Marco Tibullo, Queen Mary University of London |
| Session 113: Lunch June 25, 2026 12:00 to 13:30 |
| Session 114: AUTOREGRESSIVE MODELS June 25, 2026 13:30 to 15:15 Location: D-107 |
| Session Chair: Ramon Punder, University of Amsterdam |
Autoregressive Models with Non-Causal ARCH VolatilityAbstractThis paper introduces a novel non-causal (forward-looking) ARCH specification in which conditional heteroskedasticity depends on leads of the process. When observed in calendar time, this time inversion allows large past shocks to affect the entire conditional distribution rather than only its scale, as in standard ARCH models. The resulting forward-looking dynamics can generate symmetric bimodal predictive densities, providing a new interpretation of economic uncertainty. We establish the stochastic properties of autoregressive processes with errors following the proposed non-causal ARCH specification and derive sufficient conditions for consistency and asymptotic normality of an Approximate Quasi–Maximum Likelihood Estimator (AQMLE). A kernel-based estimator of the marginal error distribution and the predictive density is also developed. Simulation results demonstrate good finite-sample performance. An empirical application to monthly CPI data shows that the proposed model captures distributional dynamics more effectively than traditional approaches, particularly during periods of elevated uncertainty and volatility. |
| Presented by: Gabriele Mingoli, Aarhus University |
Uniform inference with general autoregressive processesAbstractA unified theory of estimation and inference is developed for an autoregressive process with root in (-∞,∞) that includes the stationary, local-to-unity, explosive and all intermediate regions. The discontinuity of the limit distribution of the t-statistic outside the stationary region and its dependence on the distribution of the innovations in the explosive regions are addressed simultaneously. A novel estimation procedure, based on a data-driven combination of a near-stationary and a mildly explosive artificially constructed instrument, delivers mixed-Gaussian limit theory and gives rise to an asymptotically standard normal t-statistic across all autoregressive regions. The resulting hypothesis tests and confidence intervals are shown to have correct asymptotic size (uniformly over the space of autoregressive parameters and the space of innovation distribution functions) in autoregressive, predictive regression and local projection models, thereby establishing a general and unified framework for inference with autoregressive processes. Extensive Monte Carlo simulation shows that the proposed methodology exhibits very good finite sample properties over the entire autoregressive parameter space and compares favourably to existing methods within their parametric (-1,1] validity range. We demonstrate how our procedure can be used to construct valid confidence intervals in standard epidemiological models as well as to test in real-time for speculative bubbles in the price of the Magnificent Seven tech stocks. |
| Presented by: Katerina Petrova, |
Two Gaussians, Too Many: A bootstrap-based approach to assess identifiability in non-Gaussian structural vector autoregressionsAbstractWe propose a bootstrap-based approach to evaluate the asymptotic validity of Independent Component Analysis (ICA) identification and inference in structural vector autoregressions (SVARs). ICA-based identification requires that out of n mutually independent structural shocks, at most one is Gaussian. The diagnostic evaluates this condition by measuring the divergence between the conditional bootstrap distribution of a maximum likelihood estimator of the structural impact matrix and its asymptotic benchmark under valid identification. We establish that the bootstrap distribution of the non-Gaussian maximum likelihood estimator, under the validity of identification conditions, is asymptotically normal. This simplifies the diagnostic to a test of normality of the bootstrap replications of the estimator. Crucially, under the null of valid identification, the diagnostic induces no pre-testing bias as bootstrap replications and sample size diverge jointly (at an appropriate rate). It ensures the test statistic is asymptotically independent of the data. Monte Carlo simulations with Normal-Inverse Gaussian (NIG) structural shocks demonstrate that the diagnostic attains near-exact nominal size under valid identification conditions and exhibits substantial power against identification failure caused by the presence of multiple Gaussian structural shocks. Based on the estimates of a SVAR model in the macroeconomic and financial uncertainty literature, we demonstrate its potential as a practical, robust tool for validating ICA-based identification without any pre-testing bias. |
| Presented by: Paritosh Junare, Università di Bologna |
Proper and Robust Autoregressive Derivative Adaptive ModelsAbstractThis paper introduces the class of Proper and Robust Autoregressive Derivative Adaptive (PRADA) models, extending score-driven updates beyond the logarithmic scoring rule to all strictly proper and locally proper scoring rules and strictly consistent scoring functions. PRADA updates reduce an expected local divergence measure under misspecification and thereby generalize the information-theoretic foundation of score-driven models beyond the Kullback-Leibler divergence. They are interpreted as the online analogues of M-estimators, and are linked to online Z-estimation through strict identification functions. When derived from scoring functions or identification functions, PRADA updates operate directly on elicitable functionals of the postulated conditional distribution, such as conditional means, quantiles or risk measures, and therefore do not require a parametric model. The results provide general conditions under which updates are guaranteed to reduce their corresponding divergence, establish robustness through bounded and censored updates, and encompass many existing score-driven inspired models as special cases. |
| Presented by: Ramon Punder, University of Amsterdam |
| Session 115: BANKS AND FIRM OUTCOMES June 25, 2026 13:30 to 15:15 Location: E002 |
| Session Chair: Robert Petrunia, Lakehead Universtiy |
Fintech Adoption and the Branch Network of Retail Banks in China: are branches closing?AbstractWe provide new evidence on the effect on bank branching after the rise of fintech in China. After a long period of significant growth, the branch network peaked in 2018 and remained moderately stable. Market-oriented financial reforms led to a significant increase in bank competition and expansion of the branch network. However, the adoption of fintech is a new competitive force which affects negatively the optimal number of branches. If customers access financial products are moved to digital platforms, the role of geography in commercial banking is considerably diminished. However, geographic variables, such as distance of firms and depositors to their nearest branch, are still relevant, related to the use of in-person banking services. Our results are robust to the use of cross-city and cross-bank regressions as well as the use of instrumental variables. Our low estimate of the distance to travel cost elasticity is 0.55, which suggests an increase of the size of the branch network would not provide banks with much gain of deposits and loans to face competition from fintech companies. |
| Presented by: Paulo Regis, University of Southampton |
Disasters and Innovation: Firm-Level Evidence from ChinaAbstractDo natural disasters destroy or boost innovation? Using a novel dataset linking geocoded disaster records to patent filings and firm financials for Chinese manufacturing firms (1998–2007), we estimate staggered difference-in-differences models that reveal a 14–19 percent increase in patenting following disaster exposure. This aggregate effect masks stark heterogeneity: large firms and those with strong cash positions drive the response, while state-owned enterprises and financially constrained firms show no adaptation. Internal finance emerges as the binding constraint—high-cashflow firms increase patenting by 26–30 percent, while external credit channels remain underutilized. Regional financial development amplifies effects, with firms in high-credit provinces showing responses nearly double the baseline. Environmental regulation does not impede adaptation; if anything, policy zones are associated with stronger innovation responses. These findings suggest that disasters can spark technological upgrading, but the institutional context matters. |
| Presented by: Alessandro Giacardi, Sapienza University of Rome |
Branches and roots: banking presence and deforestation in Amazonian indigenous landsAbstractIndigenous lands in the Brazilian Amazon are environmental and socio-ethnic sanctuaries that are increasingly threatened by illegal deforestation. Merging satellite imagery with branch-level changes in local banking presence, we use local-projection instrumental variable estimates to show that every new bank branch lowers cumulative forest loss inside these territories by roughly 1.7 square kilometers, for at least five years. To explain this finding, we use the same instrument strategy and a wide range of administrative datasets to estimate the causal mediation informed by a structural model of choice between honest and capital-intensive criminal activity. The dominant causal pathway is that bank presence strengthens social capital, which in turn lowers deforestation in indigenous land. The evidence does not support alternative causal channels: bank presence improves local economic conditions or heavy machinery acquisition, but these in turn do not influence the dependent variable. |
| Presented by: Jean de Oliveira, Goethe University Frankfurt |
Ownership Networks, Financing and Firm GrowthAbstractThis paper extends the literature on firm dynamics by incorporating ownership networks and financing into the analysis of firm growth. Using administrative data covering the universe of privately owned Ecuadorian manufacturing firms from 2000 to 2019, we construct firm-level measures of co-ownership networks and assess their impact on growth outcomes. Employing a quantile fixed-effects dynamic panel regression framework, we identify heterogeneous effects of firm age across the conditional distribution of growth. Both leverage and ownership network variables exhibit statistically significant effects on firm growth, which demonstrates their influence on firm dynamics through channels such as resource access and inter-firm connections. For younger firms, we find no significant relationship between age and growth, which suggests ownership networks and financing conditions are more relevant drivers in a firm's early years. These findings show inter-firm linkages, particularly ownership networks, capture additional aspects of firm dynamics not explored in previous research. |
| Presented by: Robert Petrunia, Lakehead Universtiy |
| Session 116: CLIMATE POLICY June 25, 2026 13:30 to 15:15 Location: B128 |
| Session Chair: Anna Matzner, Vienna Institute for International Economic Studies |
Intermittency and the potential of wind energy for CO2 abatementAbstractThe potential of wind energy for CO2 emissions abatement is hampered by intermittency, given the limited storability of power. Realized volatility of wind speed comes through as a particularly informative intermittency measure in an analysis of the dynamic relation between emissions, net electricity import, and wind energy in Denmark. As the system variables are strongly persistent and move together, we conduct a (fractional) cointegration analysis, extended to accommodate covariates, including intermittency, climate, and demand variables. Marginal emissions avoided (MEA) are estimated at 0.53 tonnes per MWh of wind generation and significant, even when accounting for intermittency. |
| Presented by: Bent Jesper Christensen, Aarhus University |
Why Methane Matters: A Dual Approach to Policy Impact and Abatement Cost Modeling in the U.S. Oil and Gas SectorAbstractThis paper examines the impact of regulatory announcements on methane emissions and evaluates the economic potential for methane abatement in US oil and gas basins. We first conduct an event study analysis of an adopted but not yet implemented US regulation that would impose substantial fees on methane super-emitters. The results suggest that the implicit tax signal embedded in the regulation announcement far exceeds the marginal abatement cost for a substantial share of installations, although this estimate reflects only a single average point on the cost curve. To capture the full distribution of costs, we then construct marginal abatement cost curves for the Permian, Appalachian and Anadarko U.S basins. This expanded analysis reveals sizeable “missed-money” opportunities: approximately 8% of methane emissions could be abated at zero net cost and nearly 50% could be avoided for less than 1.5 USD/t CO2e. These cost estimates are significantly lower than recent estimates, revealing considerable untapped potential for cost-effective methane mitigation. |
| Presented by: Kévin Vandermarlière, |
ENERGY-SAVING TECHNOLOGY SHOCKS, EMISSIONS, AND THE MACROECONOMYAbstractWe use restrictions derived from frontier models of directed technical change to identify an energysaving technology shock in a Bayesian structural VAR of the U.S. economy. This shock is associated with a persistent reduction of the carbon intensity of output. It also leads to a delayed but strong increase of GDP which gives rise to substantial additional fossil fuel consumption and new emissions. As a result, per capita emissions fully rebound after an initial decline. These effects can largely be attributed to a substitution of fossil fuel end-use by electricity, much of which has historically been generated using fossil fuels. |
| Presented by: Soroosh Soofi Siavash, Bank of Lithuania |
Distributional and macroeconomic implications of carbon taxation: Regional variation in revenue recyclingAbstractThe distributional effects of carbon taxation depend to a large extent on the accompanying revenue-recycling mechanism, which determines how the generated tax revenue is redistributed among households. This paper examines the impact of regionally differentiated revenue recycling that is based on variation in local infrastructure and compares it to alternative recycling approaches. First, I use household consumption data to estimate energy spending, emissions, and the resulting carbon tax burden across households. I then compare this burden to revenue-recycling transfers and identify the key factors driving the differences across households. Second, I develop a quantitative framework to assess both the macroeconomic and distributional implications of carbon taxation under various recycling schemes. The results indicate that while carbon taxes tend to create horizontal inequalities - disadvantaging rural households - revenue recycling can significantly mitigate these effects. The specific form of recycling appears less critical. Macroeconomic outcomes remain similar across recycling mechanisms but are more pronounced when no recycling is implemented. |
| Presented by: Anna Matzner, Vienna Institute for International Economic Studies |
| Session 117: DISTRIBUTIONAL TREATMENT EFFECTS June 25, 2026 13:30 to 15:15 Location: B008 |
| Session Chair: Bets Ruscoe, Monash University |
Influence of Treatment Duration and Group Size on Average Treatment EffectsAbstractWhen heterogeneous treatment effects are corrected with treatment dummies, conventional two-way fixed effects regressions encounter issues not due to negative weights, but because of the correlation between regression coefficients and regressors. We break down the heterogeneous treatment effects into three components: the duration of treatments, the group size of treated units, and pure idiosyncratic errors. As the duration of treatment increases, the treatment effects may also improve. Similarly, when a larger number of units are treated simultaneously, the impact of the treatment may increase as well. Existing estimators struggle to separately identify or estimate these two effects. We propose a new estimation method to address this issue. With a large sample, our proposed estimators demonstrate favorable asymptotic properties, even in the presence of serial dependence errors. We illustrate how to apply our estimator using the empirical example of the Mexico Medicare program. |
| Presented by: Donggyu Sul, University of Texas at Dallas |
Distributional treatment effect with latent rank invarianceAbstractTreatment effect heterogeneity is of great concern when evaluating policy impacts, such as assessing the proportion of individuals who are better off under the treatment. However, existing analyses have mostly been limited to summary measures such as an average treatment effect, due to the fundamental limitation that we cannot simultaneously observe both a treated potential outcome and an untreated potential outcome for a given unit. In this paper, I propose a conditional independence framework to circumvent this limitation and estimate moment-identified distributional treatment effect (DTE) parameters, such as marginal distribution of treatment effect. The key identifying assumption is that the two potential outcomes are conditionally independent given a latent variable, which is informed by two proxy variables. Interpreting this latent variable as underlying individual-level heterogeneity, I motivate the identifying assumption as ‘latent rank invariance.’ In implementation, I assume a finite support on the latent variable and propose an estimation strategy based on nonnegative matrix factorization and plug-in GMM. Using Neyman orthogonality, I establish asymptotic normality of the estimator, enabling inference for DTE parameters. |
| Presented by: Myungkou Shin, University of Surrey |
Prognostic Variables Estimation of Treatment Effect Distributions: Revisiting the National JTPA StudyAbstractExperimental studies of job training programs often indicate modest average treatment effects in earnings, yet they may generate sizable welfare gains for many workers that are offset by losses for others. This paper introduces a new distributional regression framework that identifies the treatment effect distribution by leveraging variation in prognostic variables, baseline characteristics that predict untreated outcomes without directly affecting treatment effects, to uncover heterogeneity in program returns. I develop a two-step estimator for the distribution and establish its large-sample properties. Applying the method to the National Job Training Partnership Act (JTPA) Study with remote past earnings as prognostic variables, I uncover substantial, previously undocumented offsetting heterogeneity: over 10% of participants gain over $10,000 cumulatively over the 30-month follow-up, despite modest mean effects. I propose an R^2-like measure that decomposes the variance of treatment effect and reveal that less than 20% of the variance can be explained by conventional conditional average treatment effects, demonstrating how standard mean impact estimates substantially understate the effectiveness of public job training programs. |
| Presented by: Young Ahn, University of Pennsylvania |
Demographic Parity Gaps and Treatment RulesAbstractA growing literature studies welfare-maximizing allocation of beneficial interventions under capacity constraints, while a parallel literature develops metrics and constraints for fairness in algorithmic decision-making. It is not known how correcting for endogenous treatment affects the fairness properties of welfare-based allocation rules. Using the Job Training Partnership Act (JTPA) data and building on Kitagawa and Tetenov (2018) and Abadie et al. (2002), we use plug-in allocation rules that rank individuals by predicted gains and treat only the top-k fraction. We compare OLS plug-in rules to IV plug-in rules that use assignment as an instrument for endogenous take-up. We evaluate each rule both excluding and including sensitive attributes (gender and ethnicity) as covariates. Welfare is measured using the Kitagawa-Tetenov empirical welfare criterion, and fairness is assessed via proportional-parity (Saleiro et al., 2018) and total-variation (TV) similarity. We find that capacity constraints can generate sizable demographic distortions even when unconstrained policies appear approximately fair. Correcting for endogeneity via IV generally moves allocations toward proportional representation relative to OLS, especially for gender and at tighter treatment caps, while also improving welfare. Including sensitive attributes as covariates does not improve proportional fairness in this setting and increase deviations from parity at low treatment caps. Overall, the results suggest that addressing endogeneity can mitigate - rather than exacerbate - fairness efficiency trade-offs in constrained allocation problems. |
| Presented by: Bets Ruscoe, Monash University |
| Session 118: EDUCATION AND HEALTH June 25, 2026 13:30 to 15:15 Location: B129 |
| Session Chair: Andrea Pop-Catalisan, Paris School of Economics |
From Mothers to Children: Intergenerational Returns to EducationAbstractThis study examines the intergenerational effects of maternal education on early childhood development using the 1997 Turkish Compulsory Schooling Law in a regression discontinuity framework. Drawing on the 2018 Turkey Demographic and Health Survey, it evaluates how policy-induced increases in maternal schooling affect children’s sociocognitive and physical development. Results show significant improvements in children’s ability to follow directions, interact with peers, and in physical outcomes, including higher birth weight, reduced stunting, and better anthropometric measures. Benefits vary by maternal background: sociocognitive gains are concentrated among children of rural-origin mothers, while physical improvements are stronger for children of urban-origin mothers. Effects are amplified when maternal grandmothers are educated, highlighting intergenerational complementarities. The mechanisms appear to operate primarily through improvements in maternal health behaviors and knowledge—such as earlier and higher-quality prenatal care and greater exposure to informational resources—and through enhanced parenting practices, particularly substantial reductions in child neglect and more attentive supervision. |
| Presented by: Betul Turkum, Aix-Marseille School of Economics |
Maternal Exposure to Terrorism and Child Skills DevelopmentAbstractThis paper examines the intergenerational effects of maternal exposure to terrorism on early childhood skill development. Using data from the 2018 Turkish Demographic and Health Survey linked to detailed records of terrorist incidents, I measure mothers' exposure to conflict-related fatalities in their birth cities during early schooling years. I employ a two-stage difference-in-differences estimator that exploits spatial and cohort-level variation in exposure. The results show that maternal exposure to terrorism significantly reduces children's socio-emotional and physical development, while having no detectable effects on literacy and numeracy. Further analysis suggests that these effects operate through reduced parental investments, lower maternal education and lower wealth. Several robustness checks confirm the the findings. |
| Presented by: Sonkurt Sen, University of Bonn |
Delaying Cancer: The Effect of Education on the Age at Cancer DiagnosisAbstractThis paper investigates whether educational attainment affects the timing of cancer onset. We exploit Austria’s 1962 school reform – which increased compulsory schooling from eight to nine years – to estimate the effect of additional education on the age at first cancer diagnosis. The reform increased the likelihood that students entered the school-based track, which is associated with white-collar employment and later entry into the workforce, rather than the vocational track. Using a regression discontinuity design, accelerated failure time models, and newly linked administrative data covering schooling and cancer diagnoses for the full population of Austria, we find that education beyond the vocational training delays cancer diagnoses among men, particularly for cancers linked to smoking and diet. We find no effect for women or for cancers less clearly related to behaviour. These findings suggest that education influences not only overall health outcomes but also the timing of serious health events – an underexplored mechanism that may help explain persistent inequalities in healthy life expectancy. |
| Presented by: Yuliya Kulikova, OIST (Japan)/ IIASA (Austria) |
Empty Nest Syndrome: Parents’ Labor Supply and Well-BeingAbstractWhen children leave the parental home, parents may gain additional time to allocate to either leisure or employment. This paper examines how the transition to an empty nest—defined as the point when all children have moved out of the parental household—affects parents’ labour supply and well-being. Focusing on mothers and fathers separately, I employ an event-study approach using data from the Survey of Health, Ageing, and Retirement in Europe (SHARE). The results show that mothers who are already working tend to increase their weekly hours following the empty-nest transition, while there is little change in fathers’ labour supply. However, there is no evidence of mothers re-entering the labour market, nor are there significant effects on depressive outcomes for either parent. |
| Presented by: Andrea Pop-Catalisan, Paris School of Economics |
| Session 119: FORECASTING June 25, 2026 13:30 to 15:15 Location: D-105 |
| Session Chair: Davud Rostam-Afschar, University of Mannheim |
Forecasting GDP using a Zoo of risk and uncertainty indicatorsAbstractAgainst a backdrop of heightened geopolitical and economic uncertainty, timely monitoring of economic activity is crucial for policy design. This paper examines whether freely available medium- and high-frequency risk and uncertainty indicators can forecast monthly GDP fluctuations in the US and the Eurozone. Using indicators spanning February 2007--August 2025, we construct a ``synthetic'' GDP measure as a weighted combination of the indicators, with weights estimated using regularization methods (LASSO and ridge regression). The synthetic GDP is informative for one-step-ahead monthly forecasting and provides a transparent decomposition of the risk factors shaping GDP dynamics over time. For the US, systemic/financial risk---and more recently policy and socio-environmental uncertainty---emerges as the most informative predictor of GDP fluctuations. For the Eurozone, global business-cycle indicators and economic sentiment dominate, with additional influence from US systemic/financial risk. In both regions, indicators linked to expectations and financial/political uncertainty have gained importance relative to real-economy indicators. Finally, we produce multi-step GDP forecasts through end-2025 by projecting the risk indicators using ARIMA models and Fourier series. |
| Presented by: Jan Ditzen, Free University of Bozen-Bolzano |
Comparing forecast performance on large panel data with unknown clustering structureAbstractWe introduce a novel Diebold-Mariano type test for evaluating the equal predictive accuracy of forecast models in large panel data settings. Our framework accommodates forecast errors that display substantial heterogeneity with unknown and clustered dependence structures in the cross-sectional dimension, as well as serial correlation over time. A key advantage of our approach is that it does not require prior knowledge of the number or composition of clusters and allows for overlapping or non-independent clusters, making it particularly well suited for complex data environments such as financial forecasting. To illustrate its practical relevance, we apply the test to intraday sovereign credit default swap (CDS) spread forecasts. We find that thresholding becomes particularly important when average score differences are small. Our results highlight the importance of robust testing procedures in forecasting applications with complex and unknown dependence structures. |
| Presented by: Lotta Rüter, Karlsruhe Institute of Technology |
Forecasting private sector credit in the EUAbstractThis paper proposes a methodology to forecast household and corporate credit growth for the 27 EU Member States at the country level. Credit growth is measured as a smoothed flow-to-lagged stock ratio and displays cross-country heterogeneity. We document key features of credit dynamics across countries and sectors and describe data transformations aimed at ensuring comparability across forecasting exercises. Building on this, we evaluate a range of forecasting models, including univariate and multivariate, linear and non-linear specifications, estimated using rolling and expanding windows. The framework delivers both unconditional forecasts and forecasts conditional on external indicators, and allows for a systematic assessment of the contribution of macro- and financial indicators to forecast accuracy. Forecasts are generated at multiple horizons and evaluated using a common set of metrics. The results provide evidence on the relative performance of alternative models and indicators across countries, sectors, and forecast horizons. |
| Presented by: Mirjam Salish, European Commission and Oesterreichische Nationalbank |
Do Professional Forecasters Care About Their Peers' Forecasts?AbstractWe investigate whether professional forecasters respond to information about their peers' forecasts. To assess this, we design and implement a randomized controlled trial among professional forecasters. Forecasters update their reported beliefs considerably when being informed about their relative position in the distribution of all forecasts in the panel. We obtain this insight from standard Bayesian updating regressions, which suggest forecasters who receive our information treatment put significantly less weight on their prior forecasts. Accounting for treatment intensity, we find that updating is stronger for forecasters who face a larger surprise. This causal evidence suggests that professional forecasts may partly reflect distributional considerations beyond pure conditional expectations, in line with strategic motives or intrinsic preferences about forecasters' relative positions in the distribution. |
| Presented by: Davud Rostam-Afschar, University of Mannheim |
| Session 120: GEOPOLITICAL RISK June 25, 2026 13:30 to 15:15 Location: D-114 |
| Session Chair: Adriana Cornea-Madeira, ISEG University of Lisbon |
Fedspeak, LLM-Derived Signals, and High-Frequency TradingAbstractFinancial markets are often assumed to incorporate Federal Reserve policy decisions instantaneously. Yet it remains unclear whether markets fully absorb policyrelevant information at the moment of release, or whether information continues to be revealed as the policy narrative unfolds across sequential communications. I construct a novel measure of policy communication shocks, the Communication Divergence Signal (CDS), which captures semantic, tonal, and framing shifts across sequential FOMC communications. Using large language models and high-frequency eventstudy methods, I show that communication divergence generates significant and persistent movements in intraday asset prices and abnormal trading volume. Abnormal trading volume is measured relative to non-announcement benchmarks at the same intraday horizon, isolating excess trading activity associated with information arrival. I further document systematic responses in daily open interest, consistent with belief updating and position reallocation rather than transitory liquidity provision. These findings highlight the role of communication sequencing and framing in shaping price discovery and trading dynamics across financial markets. |
| Presented by: Aihui Wei, CUNY Graduate Center |
Tariff Escalation and the Dollar: Convenience Yields and Exchange Rate Adjustment AbstractStandard open-economy models often predict that unilateral tariffs appreciate the domestic currency. During the 2024–2025 U.S. tariff escalation, however, the dollar depreciated. We study exchange-rate and financial-market responses to U.S. tariff announcements across the 2018–2019 trade conflict and the renewed escalation in 2024–2025 using an announcement-based dataset and high-frequency asset prices. We document three main findings. First, exchange rates adjust gradually rather than jumping on impact. Local projections reveal persistent post-announcement drifts over subsequent days and weeks in both episodes. Second, in 2024–2025 — but not in 2018–2019 — CIP deviations for long-maturity G10 government bonds compress following U.S. tariff announcements, consistent with a decline in the relative convenience yield on U.S. Treasuries in our framework. CIP deviations for the Chinese renminbi move in the opposite direction, rising sharply — a cross-currency asymmetry that is difficult to reconcile within a single-channel account. Third, financial-market responses are nonlinear in tariff magnitude. Announcements at or below the median tariff change generate economically modest responses, whereas announcements in the upper decile — characteristic of 2024–2025 — trigger substantially larger exchange-rate and bond-market movements. To interpret these patterns, we develop a stylized settlement-network framework in which tariff news reduces expected participation in dollar-based settlement. Because global dollar usage underpins the convenience yield on U.S. safe assets, shifts in perceived network participation can lower the convenience yield on U.S. Treasuries and thereby transmit to exchange rates. Anchored to the bond-pricing framework of Du, Im, and Schreger (2018), the framework shows how targeted and weakly credible tariff actions can leave safe-asset demand largely unaffected, while broad and credible measures generate larger and more persistent financial adjustment — including dollar depreciation rather than the appreciation standard theory predicts. |
| Presented by: Stefan Laseen, Sveriges Riksbank |
Geopolitical hybrid threatsAbstractHybrid threats encompass a range of hostile activities aimed at achieving geopolitical objectives, including cyberattacks, infrastructure sabotage, espionage, economic and political coercion, disinformation, and other forms of hybrid warfare. We construct a Geopolitical Hybrid Threat (GHT) index based on newspaper coverage of these activities. The index surges from the mid-2010s, surpassing the high levels observed during the late Cold War. Using a narrative instrument in a vector autoregression, we find that hybrid threat shocks suppress aggregate demand by raising uncertainty, reducing confidence, and tightening financial conditions, while triggering policy responses in the form of increased defense spending and monetary easing. |
| Presented by: Norbert Metiu, Deutsche Bundesbank |
Drivers of the Extreme Tail Behavior of Banks’ Risk SentimentAbstractThe March 2023 banking crisis underscored the vital importance of monitoring extreme shifts in investor sentiment as early indicators of financial instability. Through a newly constructed bank-specific investor sentiment tracker (BRST), we introduce methodologies for analyzing the key drivers behind extreme sentiment episodes affecting major European and US banks. Our results demonstrate that sentiment dynamics follow a dual-layer structure: immediate sentiment changes stem primarily from bank-specific momentum, while long-term sentiment shifts increasingly reflect broader market-wide pressures. These techniques provide policymakers, supervisory authorities, and market participants with enhanced capabilities for timely risk assessment and proactive intervention. |
| Presented by: Adriana Cornea-Madeira, ISEG University of Lisbon |
| Session 121: INFLATION ANCHORING AND RISK PREMIA June 25, 2026 13:30 to 15:15 Location: D-113 |
| Session Chair: Athanassios Stavrakoudis, University of Ioannina |
Measuring Macroeconomic Stars with Scarring EffectsAbstractPotential output and the natural rate of unemployment are commonly estimated through trend-cycle decompositions, where they are identified as underlying trends reflecting slow-moving supply factors. In this paper, we extend this framework to accommodate the possibility that cyclical disturbances affect trends endogenously through “scarring” effects. Our results show that a significant share of business-cycle fluctuations is explained by these endogenous movements in the trends. Accordingly, the estimated cycle—relieved of the burden of explaining the persistence of real variables— tracks inflation developments more closely than in specifications without scarring, including through a steeper Phillips curve. This shift reduces the model’s reliance on cost-push factors. The tighter connection between inflation and the cycle also implies that the model relies less on cyclical innovations and more on the supply-driven components of the trends to account for fluctuations in real activity, especially during episodes when inflation and output diverge or are only weakly correlated. This rebalancing of volatility from cost-push shocks toward supply factors carries important implications for the conduct of monetary policy. |
| Presented by: Manuel González-Astudillo, Board of Governors of the Federal Reserv |
Inflation Narratives and Risk PremiaAbstractTheory suggests that inflation risk premia are generally positive when investors believe supply shocks will outweigh demand shocks, and negative otherwise. We measure these beliefs using demand and supply narratives derived from inflation news through Causality Extraction, a tool that identifies causal relationships between inflation and its causes. We extract Euro Area demand and supply narratives from the Financial Times and Reuters inflation news, and United States demand and supply narratives from the Wall Street Journal's inflation news. Our primary narrative variable, NetDemand, is the difference in the number of articles attributing inflation to demand versus supply factors. Consistent with theory, for both the Euro Area and United States, inflation risk premia are inversely related to NetDemand across maturities. NetDemand’s explanatory power remains strong even after controlling for the composite PMI and VIX, strengthens with risk aversion in the US and with inflation volatility in the EA, and is not subsumed by the demand and supply contributions to inflation, views of professional forecasters, and narratives obtained from LLMs. |
| Presented by: Giovanni Trebbi, European Central Bank |
Permanent and Transitory Shocks to the Expected Inflation Term StructureAbstractWhat drives the persistence of inflation expectations across horizons? This paper studies the long-run dynamics of inflation expectations by treating the entire expected inflation term structure as a single curve-valued time series. We document persistent low-frequency movements across maturities that are inconsistent with stationarity at the functional level. Motivated by this evidence, we model the term structure as a non-stationary functional autoregressive process and decompose innovations into permanent and transitory components using long-run restrictions in the spirit of Blanchard and Quah, adapted to a functional setting. Empirically, we find that a single common stochastic trend drives long-run movements in inflation expectations, while transitory shocks generate short-lived deviations. While permanent and transitory shocks have similar short-run effects on inflation and real activity when normalized by their impact on inflation, monetary policy responds more strongly and persistently to permanent shifts in expectations. Correlations with external measures of expectation and macroeconomic shocks provide additional interpretive evidence on the economic content of the decomposition. |
| Presented by: Fabio Gomez-Rodriguez, Lehigh University |
Price Connectedness in the Extra Virgin Olive Oil Retail Market in Greece: A Quantile Regression ApproachAbstractThis study investigates price connectedness in the extra virgin olive oil market across four major Greek retail chains. Using a quantile regression approach with block-bootstrap inference, the analysis explores how inter-brand price relationships vary across the conditional price distribution. The research utilizes daily supermarket-level price data for extra virgin olive oil collected between March 20, 2023, and November 4, 2025. The results reveal substantial heterogeneity in competitive intensity both within and between retailers: some chains exhibit strong price co-movements, while others display asymmetric or weaker responses. These patterns indicate differentiated pricing strategies and varying degrees of market integration. Inter- and intra-retailer dynamics highlight the importance of retailer and brand-specific environments in shaping strategic interactions. |
| Presented by: Athanassios Stavrakoudis, University of Ioannina |
| Session 122: INFLATION DYNAMICS June 25, 2026 13:30 to 15:15 Location: D-111 |
| Session Chair: Alejandro Gonzalez, Washington University in St. Louis |
Inflation FactorsAbstractThis paper develops an econometric framework for identifying latent factors that provide real-time estimates of supply and demand conditions shaping goods- and services-related price pressures in the U.S. economy. The factors are estimated using category-specific personal consumption expenditures (PCE) data on prices and quantities, using a sign-restricted dynamic factor model that imposes theoretical predictions of the effects of fluctuations in supply and demand on prices and associated quantities through factor loadings. The resulting estimates are used to decompose total PCE inflation into contributions from common factors—goods demand, goods supply, services demand, services supply, and inflation expectations—and category-specific idiosyncratic components. Validation exercises demonstrate that the estimated factors provide an informative and coherent narrative of inflation dynamics over time and can be effectively used for forecasting and policy analysis. |
| Presented by: Viacheslav Sheremirov, Federal Reserve Bank of Boston |
Unobserved Components with Stochastic Volatility and Volatility Responsiveness (UCSVVR)AbstractThis paper develops an unobserved components model in which observation-equation volatility feeds directly into the trend equation, establishing a direct link between inflation uncertainty and the evolution of the permanent component. The proposed Unobserved Component Model with Stochastic Volatility and Volatility Responsiveness (UCSVVR) is estimated using Markov chain Monte Carlo (MCMC) methods. The baseline model is extended to allow time variation in the volatility responsiveness parameter, accommodating the possibility that the link between inflation uncertainty and trend dynamics has itself evolved. We apply the UCSVVR to U.S. inflation data from 1959 onward, covering both headline and core measures of CPI and PCE inflation. The data provide strong evidence of a negative volatility responsiveness channel, in which episodes of elevated uncertainty exert downward pressure on trend inflation. Furthermore, we evaluate the model performance in an out-of-sample exercise and find that it outperforms standard benchmark models, with gains that increase at longer forecast horizons. |
| Presented by: Gabriel Arce-Alfaro, Central Bank of Ireland |
An International Factor Analysis of Factor Shares, Price Rigidities and the Inflation-Output Trade-OffAbstractWe find that inflation is statistically significant for explaining movements in the income shares of labor, capital and profits, even after controlling for other variables that might generate these co-movements, such as changes in the degree of competition or unionization. These controls are generated through factor analysis, which explains the covariances of the observed variables in terms of the underlying unobservables. Accounting for the observed co-movement between inflation and the income shares without nominal rigidities is difficult, since the income shares are not likely to impact inflation or monetary policy, and are independent of most variables and shocks, including those to productivity. Hence, the relationship is evidence of the relevance of nominal rigidities at the aggregate level. |
| Presented by: Christian Jensen, University of South Carolina |
Bargaining Shocks and the Macroeconomy: A Narrative Approach AbstractWhat are the effects of increases in workers’ bargaining power on output and the labor share? This paper uses a narrative approach to answer this question. I draw on historical and institutional evidence to track how the stance of the U.S. executive branch toward the capital-labor conflict has evolved over time. Using this narrative information, together with a minimal amount of economic theory, I estimate a structural vector autoregression to assess the effects of bargaining shocks on the economy. I find that a typical increase in workers’ bargaining power lowers output by 0.4–0.8 percent in the long run, while its effects on the labor share are negligible or negative. Bargaining shocks explain about one-third of output fluctuations at all horizons but account for little of the variance in unemployment. Absent bargaining shocks, the recent decline in the labor share would have been 20 percent more pronounced. Demand shocks account for most of the labor share’s recent decline. |
| Presented by: Alejandro Gonzalez, Washington University in St. Louis |
| Session 123: MONETARY POLICY TRANSMISSION 1 June 25, 2026 13:30 to 15:15 Location: D-110 |
| Session Chair: Miguel Boucinha, European Central Bank |
Is monetary policy transmission heterogenous across euro area countries and time? A reassessmentAbstractThis paper assesses the degree of heterogeneity in monetary policy transmission across euro area countries for the period 2000–2025. Using monthly local projections, we estimate the effects of monetary policy on a broad set of transmission variables at both the euro area and national levels. We find that the responses of mortgage and labor market variables display significant asymmetries. In contrast, the responses of output and inflation remain notably homogeneous. Second, our results suggest that key structural differences, including household indebtedness, debt maturity, interest rate rigidity and sectoral composition, shape the national responses to a common monetary policy shock. However, taken altogether, these differences tend to offset one another, resulting in a broadly uniform transmission of monetary policy across euro area countries. Third, using a rolling-window FAVAR estimation, we show that time-varying heterogeneity in monetary policy transmission is characterized by temporary and crisis-driven divergences that consistently revert to a low baseline level, reflecting the monetary policy cycle and the strength of monetary policy interventions rather than deeper structural economic divergences across euro area countries. |
| Presented by: Thaïs MASSEI, Paris School of Economics / Banque de France |
Asymmetric Monetary Policy Transmission Across Euro Area Manufacturing IndustriesAbstractWe study the asymmetric transmission of monetary policy shocks on euro-area manufacturing output and producer prices using monthly data from 2003 to 2024. Employing externally identified monetary policy shocks and nonlinear local projections, we show that contractionary policy actions generate large and persistent declines in output and prices, while expansionary shocks have weaker effects. These asymmetries are diffuse across manufacturing industries and quantitatively important, with the contribution of monetary policy to forecast error variance driven primarily by contractionary shocks. We further show that the effects of monetary tightenings are amplified in industries characterized by greater financial fragility and stronger exposure to interest-sensitive demand. |
| Presented by: Marco Flaccadoro, Banca d'Italia |
Estimating the Heterogeneous GDP Effects of Euro Membership Across Different Regions and Macroeconomic Regimes: A Bayesian Structural Time Series ApproachAbstractUsing a modern Bayesian structural time series causal inference method this paper empirically estimates the GDP per capita effects of euro area membership across the initial twelve euro area members from 1999 to 2019. The focus is on detecting effect heterogeneity across regions and macroeconomic environments. Results show that the regional effects of euro membership are highly regime‑dependent and spatially heterogeneous, shifting from modestly positive in the pre‑crisis period to sharply negative during the sovereign‑debt crisis, especially in Southern Europe. Structural gains remain concentrated in Germany and Austria throughout, while the post‑crisis recovery brings broadly positive incremental benefits from continued membership even for regions with negative long‑run counterfactuals. These patterns reveal a monetary union with persistent long‑run asymmetries but improved short‑run stabilization over time. |
| Presented by: Mara Kritzinger, |
Stablecoins and Monetary Policy Transmission*AbstractThis paper studies the effects of stablecoin adoption on bank intermediation and the transmission of monetary policy. Using evidence from the rapid expansion of stablecoins combined with confidential granular data on euro area banks and their individual borrowers, we document three main findings. First, stablecoin adoption induces a deposit-substitution mechanism, whereby funds shift from retail bank deposits to digital assets. This reallocation increases banks’ reliance on wholesale funding and can ultimately constrain their intermediation capacity. Second, we show that stablecoins alter the pass-through of policy rates to bank funding costs and lending conditions and potentially weaken the predictability of policy actions. These effects are nonlinear and depend critically on the scale of stablecoin adoption, their design features, and their regulatory treatment. Third, we document a potential risk associated with the growing prevalence of foreign-currency-denominated stablecoins. Their diffusion is likely to increase banks’ reliance on foreign-currency wholesale funding. We show that banks with greater exposure to this source of funding exhibit a weaker loan-supply response to domestic monetary policy shocks, indicating a weakening of monetary policy transmission and a potential erosion of monetary sovereignty. |
| Presented by: Miguel Boucinha, European Central Bank |
| Session 124: NONPARAMETRIC INFERENCE June 25, 2026 13:30 to 15:15 Location: B009 |
| Session Chair: Joachim Freyberger, University of Bonn |
Algorithmic Covariates and Selective InferenceAbstractEmpirical economic studies increasingly rely on covariates generated by algorithms with tuned parameters. If the tuning process uses estimates of the linear projection coefficients of interest, standard inference breaks down: Wald confidence intervals systematically undercover these coefficients. I study tune-robust confidence intervals and show how they restore correct large-sample coverage. An application revisiting the association between local electoral outcomes and national government transfers illustrates the practical relevance of these results. |
| Presented by: David Pacini, University of Bristol |
A Kernelization-Based Approach to Nonparametric Binary Choice ModelsAbstractWe propose a new estimator for nonparametric binary choice models that does not impose a parametric structure on either the systematic function of covariates or the distribution of the error term. A key advantage of our approach is its computational scalability in the number of covariates. For instance, even when assuming a normal error distribution as in probit models, commonly used sieves for approximating an unknown function of covariates can lead to a large-dimensional optimization problem when the number of covariates is moderate. Our approach, motivated by kernel methods in machine learning, views certain reproducing kernel Hilbert spaces as special sieve spaces, coupled with spectral cut-off regularization for dimension reduction. We establish the consistency of the proposed estimator and asymptotic normality of the plug-in estimator for weighted average partial derivatives. Simulation studies show that, compared to parametric estimation methods, the proposed method effectively improves finite sample performance in cases of misspecification, and has a rather mild efficiency loss if the model is correctly specified. Using administrative data on the grant decisions of US asylum applications to immigration courts, along with nine case-day variables on weather and pollution, we re-examine the effect of outdoor temperature on court judges' ``mood'', and thus, their grant decisions. |
| Presented by: Guo Yan, University of Melbourne |
Worst-Case Regression with Interval-Observed and Missing CovariatesAbstractResearchers often confront regressions where key covariates are missing or only interval-observed. Common fixes, e.g. imputation or auxiliary modeling assumptions, resolve ambiguity at the expense of credibility. Instead, we derive the sharp identified set, the smallest parameter set consistent with both the data and the maintained regression model. A computationally tractable (Hausdorff consistent) estimator for this set is provided with associated asymptotic (uniformly) valid confidence regions for the true parameter. Technically, this is achieved by extending a previous random set framework for finitely many moments to infinite-dimensional Polish (separable and complete) spaces. |
| Presented by: Martin Fankhauser, Bocconi University |
Minimax optimal estimation under shape restrictionsAbstractEconomic theory often implies shape restrictions, such as monotonicity or inequality constraints, that can help improve the precision of estimators. A common way to incorporate such restrictions is through projection-based estimation, which naturally arises in constrained linear, IV, and maximum likelihood models. These estimators require choosing a projection direction, but this choice is usually implicit, its consequences are not well understood, and we show that it can substantially affect the performance of estimators. We first establish that with a single constraint, projecting along the inverse covariance matrix $\Sigma^{-1}$ of the unconstrained estimator is uniformly optimal for all boundary values of the true parameter. In more general settings with multiple potentially binding constraints, the optimal direction depends on the unknown true parameter. To address this challenge, we develop a minimax framework for selecting projection directions. We show that when at most two constraints can bind, $\Sigma^{-1}$ remains minimax optimal, and we provide sufficient conditions under which this result extends to arbitrarily many constraints. We further propose a test for these conditions and study the resulting estimator in simulations and in an empirical application to gasoline demand. Across these exercises, $\Sigma^{-1}$ emerges as a robust and practically attractive projection rule for constrained estimation. |
| Presented by: Joachim Freyberger, University of Bonn |
| Session 125: PANEL DATA METHODS 1 June 25, 2026 13:30 to 15:15 Location: D-106 |
| Session Chair: Wei Miao, KU Leuven |
Analysis of Multiple Long-Run Relations in Panel Data ModelsAbstractThe literature on panel cointegration is extensive but does not cover data sets where the cross section dimension, n, is much larger than the time series dimension T. This paper proposes a novel methodology that filters out the short run dynamics using sub-sample time averages as deviations from their full-sample counterpart, and estimates the number of long-run relations and their coefficients using eigenvalues and eigenvectors of the pooled covariance matrix of these sub-sample deviations. We refer to this procedure as pooled minimum eigenvalue (PME). We show that PME estimator is consistent and asymptotically normal as n and T →∞ jointly, such that T≈n^{d}, with d>0 for consistency and d>1/2 for asymptotic normality. Extensive Monte Carlo studies show that the number of long-run relations can be estimated with high precision, and the PME estimators have good size and power properties. The utility of our approach is illustrated by micro and macro applications using Compustat and Penn World Tables. |
| Presented by: Alexander Chudik, Federal Reserve Bank of Dallas |
Likelihood and Order Selection for Dynamic Panels with Interactive Effects and ARMA ErrorsAbstractWe study dynamic panel data models with short time dimension T and large cross-sectional dimension N in which outcomes follow an ARMA(p,q) process with interactive effects, extending the workhorse AR(1) dynamic specification. The model implies a parametric structure for the TxT covariance matrix of the outcome that captures general dynamic dependence through an ARMA process, common shocks through low-rank factor effects, and time-series heteroskedasticity in the idiosyncratic innovations. Exploiting these covariance restrictions, we develop estimation and inference based on Gaussian quasi-maximum likelihood and a second-moment GMM approach. Because identification and estimation are driven by the covariance restrictions rather than within transformation, the framework avoids incidental-parameter (Nickell) bias when T is fixed. Under regularity conditions, we establish consistency and asymptotic normality as N grows to infinity with fixed T, and provide feasible variance estimation. We also propose information criteria that consistently select the number of factors r and jointly recover the true orders (p,q,r). |
| Presented by: Pablo Mones, Columbia University |
Nonlinear Panel Data Models with Robust Correlated Random EffecctsAbstractWhile linear panel data models with correlated random effects maintain consistency even under misspecification of the individual effect, this robustness property fails in nonlinear settings. We address this limitation by replacing the traditional parametric specification of individual effects with a nonparametric function, preserving consistency of the structural parameters when the relationship between individual effects and covariates is unknown. We employ a profile least squares style approach, obtaining the finite-dimensional parameters through optimization of the profiled objective function. The estimator achieves parametric rates of convergence for the structural parameters while accommodating nonparametric estimation of the nuisance function. Simulations confirm the theoretical results, showing consistency while existing methods are biased and inconsistent when the individual effect specification is misspecified. We illustrate the practical relevance of our approach via a study of the public capital productivity puzzle whereby we show proper modeling of the production function and individual effects leads to positive and significant returns to public capital. |
| Presented by: Daniel Henderson, University of Alabama |
Inference in high-dimensional short panel data after discretizing unobserved heterogeneityAbstractApproximating time-varying unobserved heterogeneity using discrete types has become increasingly popular in economics. Recently, \citet{beyhum2024inference} develop valid post-clustering inference procedures for target parameters in low-dimensional panel data models with a large time dimension $T$. However, their approach is not applicable to panels with a small $T$, a setting common in many empirical applications. Motivated by the growing availability of high-dimensional data, we ask whether rich covariate information can compensate for a short time dimension. We answer this question affirmatively. Specifically, we study high-dimensional fixed-$T$ panel data models with nonseparable two-way unobserved heterogeneity and show that sufficiently informative covariates can effectively substitute for long time-series variation in conducting valid inference. Drawing on insights from the double machine learning literature, we propose a two-step inference procedure based on bias-reducing moments. We establish its asymptotic validity and demonstrate through simulations that it achieves excellent finite-sample performance. |
| Presented by: Wei Miao, KU Leuven |
| Session 126: ROBUSTNESS AND REPLICATION June 25, 2026 13:30 to 15:15 Location: D-112 |
| Session Chair: Tomas Jagelka, University of Bonn |
Counting Defiers: A Design-Based Model of an Experiment Can Reveal Evidence Beyond the Average EffectAbstractWe estimate the numbers of always takers, compliers, defiers, and never takers in the sample of people in an experiment, rather than a hypothetical population from which it was drawn, using structure from the randomization design. Our data include only a binary intervention and outcome. We develop a visualization to show that samples with defiers can sometimes generate the data in more ways than samples without defiers, yielding a higher design-based likelihood. We propose a maximum likelihood decision rule that can harness this evidence, which is not captured by standard hypothesis tests, and we provide optimality conditions. We illustrate the output of our decision rule for all possible data in samples of 50 and 200 with half in intervention, demonstrating a pattern in when the MLE includes defiers despite a positive average effect. We provide insights into effect heterogeneity in two published experiments with interventions that could plausibly backfire for some people despite statistically significant positive average effects on takeup of desirable health behaviors. In both, our 95% credible sets include the estimated Frechet bounds, demonstrating that evidence is weak. Yet, our MLE includes no defiers in one; in the other, the MLE includes a count of defiers equal to the estimated upper Frechet bound, over 18% of the sample. The MLE can support a monotonicity assumption or a specific alternative as a step toward improving the average effect of future interventions by targeting them away from noncompliers. Our dbmle package, compatible with Python and Stata, implements our statistics. |
| Presented by: Neil Christy, University of Michigan |
Testing Most Influential SetsAbstractSmall influential data subsets can dramatically impact model conclusions, with a few data points overturning key findings. While recent work identifies these most influential sets, there is no formal way to tell when maximum influence is excessive rather than expected under natural random sampling variation. We address this gap by developing a principled framework for most influential sets. Focusing on linear least-squares, we derive a convenient exact influence formula and identify the extreme value distributions of maximal influence – the heavy-tailed Fréchet for constant‑size sets and heavy tailed data, and the well-behaved Gumbel for growing sets or light tails. This allows us to conduct rigorous hypothesis tests for excessive influence. We demonstrate through applications across economics, biology, and machine learning benchmarks, resolving contested findings and replacing ad‑hoc heuristics with rigorous inference. |
| Presented by: Nikolas Kuschnig, Monash University |
Does p-Hacking Mitigate or Amplify Publication Bias?AbstractThis paper studies the effects of p-hacking on the bias of published estimates when papers with statistically significant results are selectively published. We show that fast p-hacking -- actions that lead to large changes in p-values -- always increases the amplification bias from selective publication. On the other hand, slow p-hacking -- actions that lead to small changes in p-values -- increases amplification bias when selection is weak, but mitigates it when selection is strong. Thus, the effect of p-hacking depends on the selection environment and is an empirical question. We show that the true distribution of effects, as well as the probability of fast and slow p-hacking, are identified under a normality assumption. |
| Presented by: Yong Cai, University of Wisconsin-Madison |
| Session 127: VOLATILITY AND RISK PREMIA June 25, 2026 13:30 to 15:15 Location: D-115 |
| Session Chair: Thorsten Lehnert, University of Luxembourg |
Intraday dynamics of behavioral heterogeneity in stock pricesAbstractWe estimate using high frequency data a stock price model in which investors switch between market efficiency and momentum beliefs based on their previous forecast performance. We find that typically momentum beliefs are only a small fraction but at times dominate the market. We find that on most days the fraction of momentum beliefs increases realized volatility but does not affect stock prices. On days of large stock price losses (gains), the fraction of momentum beliefs has a negative (positive) impact on stock prices and increases realized volatility more than (the same as) in other days. |
| Presented by: Joao Madeira, Iscte |
Interest Rate Volatility and the Equity Term StructureAbstractThis paper studies the role of interest rates and interest-rate volatility in the pricing of equities, with a particular focus on dividend strips. Since 2022, discount- rate risk has re-emerged as a central driver of equity valuations, making the term structure of equity cash flows a natural laboratory to study rate-related uncertainty. We develop an affine term structure model of dividend strip yields estimated on both U.S. and European markets, allowing us to decompose equity valuations across maturities into cash-flow and discount-rate components. We find that interest-rate volatility is negatively priced and statistically significant in both regions, indicating that rate-uncertainty shocks behave as bad-state risk factors. Moreover, the variance decomposition of risk premia shows that rate volatility explains a sizable share of time variation in expected returns, particularly in Europe. These results highlight the importance of discount-rate uncertainty as a structural component of the modern equity term structure. |
| Presented by: Alexandre Remiat, Kanopy-AM |
Global implied volatility and variance risk premiumAbstractMotivated by the global nature of financial risk, we propose a simple measure of global implied volatility (GIV ) and show that it better describes international financial market dynamics than the VIX. Predictive evidence on both realized and implied volatilities around the world leads to the conclusion of truly global risk dynamics, not merely US spillovers. A direct global measure of variance risk premium based on GIV explains expected excess returns in various international equity markets outperforming a previously proposed measure based on cross-country aggregation of local variance risk premia. We show that this is due to the international price of jump risk which is better captured by our global measure of variance risk premium. Foreign exchange markets confirm our findings. Currency portfolios based on our global variance risk premium generate significant value in excess of those based on aggregated local variance risk premia and the US variance risk premium. |
| Presented by: Katerina Tsakou, Swansea University |
Semiparametric Estimation of Probability Weighting Functions Implicit in Option PricesAbstractThis paper develops a semiparametric estimation method that jointly identifies the probability weighting and utility functions implicit in option prices. Our econometric method obtains objective return distributions by transforming the options’ implied risk-neutral distributions according to the posited rank-dependent utility model. We nonparametrically estimate the probability weighting function using the kernel density of suitable utility-adjusted probability integral transforms. The parameters of the utility function are estimated by maximizing the resulting profile likelihood. We establish the asymptotic properties of our estimation procedure, and demonstrate its good finite sample performance in Monte Carlo simulations. Empirical results based on S&P 500 index option prices and returns over the period 1996–2023 reveal the relevance of probability weighting, in particular at the monthly horizon where the weighting function is inverse S-shaped. These results are robust to various specifications of the utility function. |
| Presented by: Jeroen Dalderop, University of Amsterdam |
| Session 128: Coffee break June 25, 2026 15:15 to 15:45 |
| Session 129: CHILD PENALTIES AND FAMILY POLICY June 25, 2026 15:45 to 17:30 Location: B128 |
| Session Chair: Yannay Shanan, Bar Ilan University |
The Bias in Pseudo-PanelsAbstractThe pseudo-panel approach is widely adopted to estimate child penalties when panel data are unavailable and researchers must instead rely on repeated cross-sectional data. This paper demonstrates that estimates based on pseudo-panel data are systematically biased for both child penalties and pre-trends. We derive a closed-form expression for the asymptotic bias and show that the pseudo-panel approach effectively imputes missing values in repeated cross-sectional datasets. However, this interpolation fails to preserve the full correlation between event time and the outcome, inducing an omitted variable bias for negative event times. Consequently, pre-trend estimates are mechanically attenuated toward zero and are therefore unsuitable for testing the parallel trends assumption. Moreover, child penalty estimates inherit a bias equal to linear combinations of the pre-trend effects. We validate these theoretical results using observational data. |
| Presented by: Timo Haller, Berlin School of Economics |
Identification of Child PenaltiesAbstractA growing body of research estimates child penalties, the gender gap in the effect of parenthood on earnings, using event studies that normalize treatment effects by counterfactual earnings. I formalize the identification framework underlying this approach and show that it does not identify its target estimand when the parallel trends assumption \emph{in levels} is violated. Insights from human capital theory suggest such violations are likely: higher-ability individuals tend to delay childbirth and have steeper earnings growth, a mechanism that causes conventional estimates to understate child penalties for early-treated parents. I propose instead to target the effect of parenthood on the gender earnings ratio and show that this parameter is identified without additional assumptions. Using Israeli administrative data, a bias-bounding exercise for the conventional estimator suggests substantial understatement (25–50\%) of child penalties for early-treated parents, while later-treated groups exhibit smaller biases. I also document that estimates of the new parameter are roughly 30\% smaller in absolute value than the conventional estimates for later-treated parents. |
| Presented by: Dor Leventer, Tel Aviv University |
Spillover Effects of Public Childcare Expansion in Regional Labor MarketsAbstractThis paper examines the spillover effects of expansions in publicly subsidized childcare for children under the age of three. Using German administrative data, I exploit two policy reforms that generated cross-county variation in the increase of childcare coverage rates. I implement a Difference-in-Differences design combined with matching to identify causal effects. The results show a statistically significant and economically meaningful increase in employment probabilities. Counties that experienced a sharp rise in childcare coverage saw a reduction in unemployment rates that was nearly six percent larger compared to control counties. Excluding mothers and childcare workers, who are directly affected by the expansion, only slightly reduces the estimated effect. This finding suggests the presence of substantial labor market spillover effects beyond the directly treated groups. |
| Presented by: Nils Wehrenberg, Friedrich-Alexander-University Erlangen-Nuremberg |
Adjustments to Reduced Cash Transfers: Religious Safety Nets and Children’s Long-Term OutcomesAbstractThis paper examines how families adjust to changes in unconditional cash transfers, and how these adjustments affect children’s long-term outcomes. In 2003, Israel reformed its child allowance program, significantly reducing unconditional cash benefits for large families. Using a sharp date-of-birth cutoff introduced by the reform, we show that Arab families responded by reducing completed fertility and increasing paternal employment. Consequently, we find little evidence that the decline in transfers negatively affected the education or labor outcomes of Arab children. In contrast, Jewish families substituted for the loss in government benefits by enrolling their school-aged children in ultra-Orthodox religious schools, without changing their fertility or labor supply. These schools act as informal safety-nets by providing valuable services unavailable in mainstream public schools but focus primarily on religious studies over secular subjects. In the long run, this substitution between formal and informal safety nets resulted in lower educational attainment among Jewish students and may have steered them toward a more religious lifestyle. Our results highlight the importance of existing support structures in determining the effects of policy changes, particularly in contexts where religious and public welfare systems compete. |
| Presented by: Yannay Shanan, Bar Ilan University |
| Session 130: CLIMATE FINANCE June 25, 2026 15:45 to 17:30 Location: D-112 |
| Session Chair: Jelena Zivanovic, European Stability Mechanism |
Funding the fittest? Pricing of climate transition risk in the corporate bond marketAbstractWe study whether climate transition risk affects the cost of debt and how corporate bond investors value green innovation. Using confidential bond holdings and global firm data, we find that bond investors condition climate risk pricing on firms' transition efforts. We show that the carbon premium is on average 15 percent lower for emission-intensive firms that engage in green innovation. Institutional investors, particularly mutual funds, exhibit higher demand for bonds issued by such transitioning firms. Our findings are consistent with a risk-based pricing channel and highlight the role of risk-bearing investors in allocating capital to firms central to the transition. |
| Presented by: Maurice Bun, De Nederlandsche Bank |
The Effects of Climate Change and Climate Policy on Credit RiskAbstractThis study examines how climate-related physical and transition risks affect credit risk. We develop a modular framework based on a threshold model for credit migrations, linking latent credit factors to key economic indicators to measure credit risk in bond markets. Using historical rating migrations, we estimate the model parameters and apply the framework to U.S. data to assess the implications of alternative climate policies. These policies generate projected paths for the credit cycle that differ markedly in direction, magnitude, and volatility. These differences translate into substantial variation in both expected losses and tail risks for diversified bond portfolios. Notably, high-quality bond cohorts are most sensitive to policy choices. Comparisons with a continuation of current policies show that orderly transitions (characterized by reduced physical damages and increased transition costs) entail higher initial expenses but deliver net savings by 2050. In contrast, disorderly transitions result in steep cost increases after 2030 and overall higher costs by 2050. |
| Presented by: Erik Kole, Erasmus University Rotterdam |
House on Fire: Climate Risk, Mortgages, and Monetary PolicyAbstractWe study whether exposure to climate risk affects banks' mortgage-lending decisions. Using detailed wildfire hazard and occurrence data in Portugal, we find evidence that banks charge higher risk premiums for mortgages originating in at-risk but not directly affected areas. This result is established by exploiting the variation in mortgage pricing within very granular bins of mortgages and debtors with similar characteristics, which are differentially exposed to wildfire risk. We investigate whether monetary policy affects the pricing of climate risk. We find that when monetary policy tightens, the climate risk premium increases. |
| Presented by: Diana Bonfim, European Central Bank and Católica Lisb |
Georisk and systemic risk in Europe's financial system -- Unpacking transmission channelsAbstractAnalyzing a sample of European financial institutions from 2002 to 2025, we show that geopolitical risk (GPR) significantly heightens systemic risk of European banks and non-banks. We measure systemic risk as the time-varying tail dependency between the whole financial system and a particular financial institution (dynamic $\Delta$CoVaR), relying on Bayesian methods. Banks exhibit stronger responses to GPR shocks than non‑banks. Transmission depends on economic and institutional conditions: global and financial uncertainty amplify the effects of GPR shocks, while trade integration dampens them. Balance‑sheet characteristics also matter—reliance on short‑term funding and illiquid assets amplifies shock propagation, whereas higher intangible asset intensity attenuates short‑run spillovers. In contrast, we find no robust evidence that macroprudential regulation mitigates the transmission of geopolitical shocks. These results indicate that effects on the riskiness of the financial system are driven primarily by balance‑sheet fragilities and amplified by heightened uncertainty. |
| Presented by: Jelena Zivanovic, European Stability Mechanism |
| Session 131: DISCRETE CHOICE MODELS June 25, 2026 15:45 to 17:30 Location: D-107 |
| Session Chair: Irene Botosaru, McMaster University |
Model Selection Tests for Incomplete ModelsAbstractThis paper expands the scope of likelihood-based model selection tests to a broad class of discrete choice models. A notable feature is that each of the competing models can make either a complete or incomplete prediction. We provide a novel cross-fitted likelihood-ratio statistic for such settings, which can be compared to a normal critical value. The proposed test does not require any information on how an outcome is chosen when multiple solutions are predicted. This allows the practitioner to compare, for example, a model that predicts a unique equilibrium to another model that allows for multiple equilibria. We examine the finite-sample properties of the test and provide guidance on the choice of tuning parameters through Monte Carlo experiments. |
| Presented by: Yan Liu, Kyoto University |
A Ranking Representation of Optimal Sequential SearchAbstractSequential search models provide a powerful framework for analyzing consumer search decisions using action sequence data. Existing applications, however, typically rely on optimal rules under which subsequent decisions depend on unobserved information revealed in earlier steps. Implementing such models therefore often requires restrictive specifications and simulation-intensive procedures, increasing computational burden and limiting empirical applicability. This paper introduces a new representation of the optimal solution for a broad class of sequential search processes. We show that the outcome of an optimal sequential search process admits an equivalent representation as a partial ranking of all feasible actions. This representation enables sequential search models to be implemented through their ranking equivalents with substantial simplification. For the Weitzman-style benchmark, we develop a rank-based GHK simulator that reduces simulation requirements while improving accuracy, computational efficiency, and ease of implementation. The ranking representation further extends to a wide range of settings, including environments with partially observed action sequences and multi-stage information-acquisition processes, such as product discovery, which can be accommodated within a unified empirical approach. Overall, our results improve both the tractability and the empirical applicability of sequential search models. |
| Presented by: Tinghan Zhang, Tilburg University |
Sequential Estimation of Dynamic Discrete Choice Models with Unobserved HeterogeneityAbstractEstimating dynamic discrete choice models with unobserved heterogeneity is computationally costly because it requires repeatedly solving fixed-point equations for all unobserved types. We develop the EM-NPL($q$) framework that combines the Expectation-Maximization (EM) algorithm with an inner fixed-point solve truncated to $q$ iterations. For the workhorse class of linear-in-parameters models, we establish a truncation-invariance result: for any $q \geq 1$, EM-NPL($q$) is numerically identical to the EM-NPL estimator that solves the inner fixed-point problem to convergence. Therefore, the choice of $q$ affects computation but not statistical properties. We also establish consistency, asymptotic normality of our estimator, and local convergence of the EM-NPL($q$) algorithm. In Monte Carlo simulations, EM-NPL($q$) reduces runtime by at least 20\% and can be 3--5 times faster. In an application to cola demand, we show that ignoring unobserved heterogeneity understates long-run own-price elasticities by up to 40\%, short-run elasticities by up to 85\%, and compensating variation from a soda tax by up to 90\%. |
| Presented by: Ertian Chen, University College London |
Correlated Random Coefficient Distributions in Panel ModelsAbstractWe consider a static linear panel model with both correlated random coefficients, which may depend on the observable regressors, and uncorrelated random coefficients, which are independent of them. Given short panels, we derive sufficient assumptions for the identification of the distribution of the correlated coefficients, allowing for irregular designs and weak restrictions on the serial structure of unobservables. Identification is achieved via a two-step strategy. In the first step, the distribution of the uncorrelated coefficients is identified separately from the correlated ones. This is accomplished differently depending on the model's dimensions. In the \emph{irregular} design, where the number of time periods equals the number of correlated coefficients ($T=p$), a stayers-based argument is used, leveraging observations where the determinant of a key covariate matrix equals zero. In the \emph{regular} design ($T>p$), an exact algebraic annihilation is achieved for the entire sample by projecting the model onto a subspace orthogonal to the covariates. In the second step, the distribution of the correlated coefficients is recovered via deconvolution. |
| Presented by: Irene Botosaru, McMaster University |
| Session 132: FINANCIAL TIME SERIES MODELS June 25, 2026 15:45 to 17:30 Location: D-111 |
| Session Chair: Qichen Zhou, Nanjing University |
Linear-Rational Term Structure Model with Risk Premium FactorsAbstractThis paper extends the Linear Rational Square-Root model of Filipovic, Larsson and Trolle (2017) to include an unspanned risk premium factor and presents a methodology for fitting the model to zero-coupon bonds. We propose simulated moment matching in order to identify volatility and the market price of risk dynamics, which are only weakly identifiable under (quasi) maximum likelihood estimation. The risk premium factor shows lower forecast errors and better simulation properties than the constant-drift adjustment used for the market price of risk by Filipovic et al (2017). Unspanned stochastic volatility factors and the risk premium factor are both essential components of the model, as the model that accommodates both of them yields the lowest estimation errors and matches the first and second moments of both yields and yield changes best. |
| Presented by: Michel van der Wel, Erasmus University Rotterdam |
Tempered Particle Smoothing and LearningAbstractSequential Monte Carlo methods are a well established tool to learn time-varying and static parameters in non-linear and non-Gaussian state-space models for example for forecasting or risk analysis. While particle filters approximate filtering distributions online, Gibbs sampling and particle learning requires smoothing distributions that condition on full sample information. Yet, standard particle smoothers often suffer from particle degeneracy in high-dimensional settings or in the presence of outliers. This paper introduces a tempered particle smoother that combines the tempered particle filter of Herbst & Schorfheide (2019) with the Forward Filtering Backward Smoothing algorithm of Godsill et al. (2004). We establish mean squared error convergence and demonstrate the method’s superior performance and robustness through data-based simulations and a particle learning application for time-varying stock market volatility. The proposed approach improves stability and accuracy, enhancing the use of particle methods for complex dynamic models in macroeconomics and finance. |
| Presented by: Elias Wolf, European Stability Mechanism |
Robust Tests for Directional PredictabilityAbstractThis paper develops robust quantile-based tests for detecting directional predictability within and across time series. The approach combines nonparametric sieve quantile approximation with self-normalisation to address nonlinearities commonly observed in conditional quantile structures, particularly those induced by heteroscedasticity in economic and financial data such as equity returns. These features often render standard quantilogram-based tests invalid or severely size-distorted. By replacing linear quantile regression with sieve-based estimation of the conditional quantile function, the proposed tests achieve greater robustness to misspecification. Self-normalisation further enables inference using standard critical values without explicit long-run variance estimation. The tests are founded on the martingale-difference property of quantile-score residuals, for which asymptotic theory is established. Monte Carlo simulations demonstrate favourable finite-sample performance. An empirical application to the left tail of risk-adjusted equity-return distributions of selected US financial institutions shows that heightened directional predictability and systemic risk detected by tests based on linear quantile regression during periods of financial stress largely reflect rises in conditional volatility during these times. |
| Presented by: Natalia Bailey, Monash University |
Emergency Lending, Macroeconomic Conditions, and Bank Loan Spreads: Evidence from the Main Street Lending ProgramAbstractInter-bank lending decrease during crises, which makes firms financing even more difficult, especially for SMEs, and leads to the phenomenon of ’liquidity hoarding’. To help firms get liquidity supply and recover from the crisis, the FED promotes the Main street Lending Program(MSLP). However, the actual participation rate of banks is lower than expected. We then explore why they are reluctant to lend and how the Fed can improve the efficiency. We use machine-learning algorithms to explore the most important contributors in forecasting bank loan spread and find that the servicing fee rate and MSELF, a facility enables banks to expand existing loans, contributes most in the forecasting. Furthermore, we find that these policy channels are state dependent. Policymakers can improve the efficiency of the policy by changing these channels and work as the ”forward guidance” to change banks’ lending behavior. These findings provide crucial insights into the design of future monetary policies during crises. |
| Presented by: Qichen Zhou, Nanjing University |
| Session 133: FISCAL POLICY June 25, 2026 15:45 to 17:30 Location: D-115 |
| Session Chair: Sebastian Hienzsch, University of Göttingen |
Fiscal Expansions and Their Financing: A Bayesian Local Projection AnalysisAbstractIn this paper, we empirically quantify the macroeconomic effects of an increase in government expenditure in two major European economies, France and Italy, and compare them with those observed in the US. To this end, we estimate a Bayesian Local Projection (BLP) model in the spirit of Miranda-Agrippino and Ricco (2021) and Ferreira et al. (2025) over the sample period 1961–2023. Our results indicate that government spending shocks are highly persistent and generate positive effects on GDP in all countries considered. The expansion in expenditure is primarily financed through higher fiscal deficits. The estimated fiscal multipliers exceed one only in the case of France. These results are particularly relevant in light of the recent initiatives by European countries to increase military spending. |
| Presented by: Marco Lorusso, University of Perugia |
Fiscal Multipliers and Political FragmentationAbstractThis paper provides novel empirical evidence on how political fragmentation shapes the fiscal transmission mechanism. Using data from 16 OECD countries (1978--2019) and narrative accounts to identify exogenous fiscal interventions, we show that when political fragmentation is high, the fiscal GDP multiplier is significantly lower. The multiplier is above unity and relatively stable over time when fragmentation is low, but generally well below unity when fragmentation is high. We show that interventions are comparable across states and argue that a conditional confidence channel helps explain our findings: only in low-fragmentation periods do fiscal interventions boost household and business confidence, translating into stronger consumption and investment responses. |
| Presented by: Ricardo Duque Gabriel, Federal Reserve Board |
Austerity and Productivity: Evidence from euro area RegionsAbstractThis paper provides novel empirical evidence on the causal link between fiscal austerity and productivity using granular data covering more than 120 regions from ten euro area countries for the period 1999-2019. We construct a new measure of regional total factor productivity (TFP) that accounts for capital and labor utilization. We identify exogenous reductions in regional public spending via a Bartik-type instrument that combines regional sensitivities to changes in national government expenditures with narrative national consolidation episodes. While lowering GDP, investment, and employment, fiscal consolidations trigger a significant and persistent increase in utilization-adjusted TFP. Our findings are consistent with a cleansing mechanism: austerity accelerates the exit of marginal producers, increasing average sector productivity, particularly in the sectors more exposed to the fiscal intervention. Austerity-driven recessions amplify the productivity gains of economic downturns considerably. |
| Presented by: Ana Sofia Pessoa, International Monetary Fund |
Euro Area Output Gaps and the Transmission of Common ShocksAbstractThe euro area output gap plays a key part in the determination of the ECB’s monetary policy stance. The actual transmission of common shocks that hit the euro area work through the member economies with possibly large and significant spillovers across the euro area. We adopt a multicycle version of the Beveridge-Nelson decomposition that allows to jointly estimate the output gaps of the nine largest euro area economies. Taking full account of the interlinkages across the euro area, we study the dynamics and transmission of a US financial shock. Distinguishing between the domestic and spillover effects of the common shocks, we can study how the aggregate behavior of the euro area output gap is explained by the effects of shocks on its components, i.e. the members’ output gaps. |
| Presented by: Sebastian Hienzsch, University of Göttingen |
| Session 134: GROWTH-AT-RISK AND NOWCASTING June 25, 2026 15:45 to 17:30 Location: B009 |
| Session Chair: Angela Torres Noblejas, University of Alicante |
House-Price-at-Risk in a Small Open Economy: Forecasting Tail Risks with Domestic and Global IndicatorsAbstractWe develop a House-Price-at-Risk framework to forecast the conditional distribution of real house price growth in Slovakia, a small open economy influenced by euro area and global conditions. We model downside and upside risks using quantile regression, allowing predictors to have heterogeneous effects across quantiles. To handle a large information set with mixed-frequency predictors (monthly and quarterly), we compare three parsimonious dimension-reduction approaches: principal components, partial quantile regression, and a mixed-frequency dynamic factor model. We find that macroeconomic predictors improve distributional forecasts of real house price growth relative to a quantile autoregressive benchmark, mixed-frequency information is particularly valuable at short horizons, and international predictors mainly add value at longer horizons when relying on lower-frequency information. |
| Presented by: Laura Coroneo, University of York |
Entropic Tilting of Forecasts to SPF Histograms: Analytics & ApplicationsAbstractThis paper develops a new, direct approach to entropic tilting of model-based predictive distributions to match histogram forecasts provided in the U.S. Survey of Professional Forecasters (SPF). We focus on tilting to histogram probabilities directly, rather than to moments of fitted distributions. We reformulate the single-histogram tilting problem and derive a novel analytic characterization for the multiple-histogram case, with iterative solutions via Iterative Proportional Fitting. Application to quarterly real-time forecasts of major macroeconomic aggregates from a Bayesian vector autoregression with time-varying volatility shows that tilting to SPF histograms significantly improves on the model's baseline forecasts, particularly during periods around the Great Recession and the COVID-19 pandemic. |
| Presented by: Elmar Mertens, European Central Bank |
Nowcasting the Income DistributionAbstractThis paper proposes a transparent and operational framework to nowcast the income distribution in real time. We combine decile means of the annual household income distribution with a broad panel of quarterly macroeconomic indicators in a mixed-frequency dynamic factor model (DFM) and implement a disciplined pseudo-real-time evaluation from 1990 to 2024. Factor models that exploit high-frequency macro information systematically improve nowcast accuracy relative to univariate ARIMA benchmarks across the distribution, with the largest gains in the middle. Cross-decile dependence is informative but not the dominant source of gains. Combining variable selection and distribution-structured factors delivers additional benefits at the tails, where predictability is intrinsically weaker. The framework is intentionally portable: it relies on off-the-shelf state-space tools, handles ragged edges and mixed frequencies, and produces updatable signals suitable for institutional monitoring. It complements microsimulation and functional approaches by supplying evaluable decile-level priors when microdata are delayed, and it can be embedded in broader pipelines that require timely distributional inputs. |
| Presented by: Nina Brehl, DIW Berlin |
A Chicken-and-Egg Problem: Profits and Inflation after COVID-19AbstractIn 2022, the contribution of unit profits to domestic price pressures in the Euro area more than doubled relative to its historical average. This increase was uneven across industries and particularly pronounced in upstream sectors. While profits have attracted growing attention as a potential driver of inflation, the existing literature has not examined their role within a fully structural macroeconomic framework alongside a broad set of competing price pressures. In this paper, I propose a unified Bayesian VAR framework to disentangle how profits have affected inflation in the Euro area, focusing on the period after the pandemic. The results show that while demand and energy prices initially drove inflation, profits in upstream companies amplified price pressures, accounting for 12.2% of the increase in headline inflation above the 2% target. In the labor-capital distribution debate, profits contribute to price pressures about three times more than wages. Moreover, firms more than offset by a factor of seven the losses they incurred during the pandemic. Overall, the findings suggest that increasing concentration in key sectors poses a material risk to price stability and warrants close monitoring. |
| Presented by: Angela Torres Noblejas, University of Alicante |
| Session 135: LOCAL PROJECTIONS AND SVARS 4 June 25, 2026 15:45 to 17:30 Location: B008 |
| Session Chair: Zhan Shi, University College London |
Wild inference for wild SVARs with application to heteroscedasticity-based IVAbstractStructural vector autoregressions are used to compute impulse response functions (IRF) for persistent data. Existing multiple-parameter inference requires cumbersome pretesting for unit roots, cointegration, and trends with subsequent stationarization. To avoid pretesting, we propose a novel \emph{dependent wild bootstrap} procedure for simultaneous inference on IRF using local projections (LP) estimated in levels in possibly \emph{nonstationary} and \emph{heteroscedastic} SVARs. The bootstrap also allows efficient smoothing of LP estimates. We study IRF to US monetary policy identified using FOMC meetings count as an instrument for heteroscedasticity of monetary shocks. We validate our method using DSGE model simulations and alternative SVAR methods. |
| Presented by: Madina Karamysheva, NRU Higher School of Economics |
Estimator Averaging of Local Projection and VAR Impulse ResponseAbstractLocal projections (LP) and vector autoregressions (VAR) are standard tools for impulse response analysis and exhibit a well-known finite-sample bias–variance trade-off: LPs have relatively low-bias but high-variance at longer horizons, whereas VARs are more stable but can be biased under misspecification. Recent work combines LP and VAR via model averaging, choosing weights to maximize in-sample fit rather than to minimize the risk of the structural impulse responses. We instead propose estimator averaging: for each horizon, we select LP–VAR weights to minimize the mean squared error of the combined impulse response. We derive oracle weights, develop feasible AR-sieve-bootstrap implementations, and compare them with an $R^2$-based model-averaging benchmark. We establish consistency and the limiting distribution of the feasible estimator under a short-memory linear data generating process. Monte Carlo results show sizable risk reductions relative to LP and VAR alone, and in an application revisiting Bauer and Swanson (2023) on high-frequency IV monetary policy shocks, estimator averaging delivers stable and economically intuitive IRFs for yields, activity, prices, and credit spreads. |
| Presented by: Balázs Vonnák, Magyar Nemzeti Bank |
A More Balanced Approach: Factor-Augmented Local ProjectionAbstractI propose Factor-augmented Local Projection (FA-LP), which incorporates a factor model into Local Projection (LP) to mitigate concerns about LP’s efficiency limitations, particularly at long horizons. Beyond the familiar informational gains, factor models can purify measurement error in macroeconomic observables. Relative to conventional LP, FA-LP offers three potential advantages: (1) bias attenuation under the recursive identification scheme when measurement error in the policy variable is limited; (2) efficiency improvements; and (3) stronger identification when using external instruments (IVs). Monte Carlo simulations confirm these advantages and show the gains are robust across horizons and identification schemes. An empirical application examining the effects of monetary policy shocks shows that FA-LP delivers more precise inference and improves the effective strength of IVs. Importantly, these advantages cannot be replicated by simply expanding the set of covariates in LP, which instead introduces additional noise and worsens estimation performance. |
| Presented by: Zhan Shi, University College London |
| Session 136: MACROECONOMETRICS June 25, 2026 15:45 to 17:30 Location: D-110 |
| Session Chair: Andrea Renzetti, Bank of England; Bocconi University |
Double Descent in Time SeriesAbstractDouble descent —a spike in test error at the interpolation threshold followed by improvement in the overparameterized regime— is well understood for ordinary linear regression through singular-value geometry of the training design matrix. This note explains how the same mechanism applies to autoregressive (AR) and vector autoregressive (VAR) forecasting, focusing only on facts that follow from standard linear algebra and time-series definitions. We write AR/VAR as a linear regression with a structured design matrix, identify the interpolation threshold in terms of T and (m, n), and show how the same “critical term” that drives divergence in linear regression appears for AR/VAR with the time-series design matrix. We also give conservative, verifiable bounds on the maximal lag order p for invertible OLS and discuss how minimum-norm solutions operate when overparameterized. |
| Presented by: Pablo Guerron, Boston College |
Narrative Sign Restrictions in a Daily Vector AutoregressionAbstractThis paper introduces a daily vector autoregression (VAR) constructed from financial variables and less frequently observed macro variables as a useful tool to identify structural shocks by narrative sign restrictions. Estimation is carried out using filtering methods to account for the mixed frequency of the data and pooling of the VAR coefficients in the time domain to efficiently condition on past data in previous months. The approach is illustrated by identifying sentiment shocks from unexplained daily variation in the U.S. stock market, where the responses to a negative sentiment shock are very similar to those for a positive uncertainty shock. |
| Presented by: Laust Særkjær, Banque de France |
Are Terms of Trade shocks always significant on GDP? It depends on expectationsAbstractThere is ample empirical evidence indicating that the terms of trade are an important driver of economic growth in emerging economies like Peru. In particular, given that these economies’ exports depend on commodity prices, it is clear that there is a statistically significant relationship between an increase in export prices and GDP growth in subsequent periods. However, in the period following 2021, in the case of Peru, a high level of political uncertainty was observed. This increased uncertainty was reflected in a drop in business confidence and private investment overall. At the same time, there was a significant increase in commodity prices, particularly copper, but this has not been reflected in higher economic growth this time around. Everything seems to indicate a non-linearity between the increase in the terms of trade and GDP growth. In particular, if business confidence is very low (in the pessimistic range), an increase in the terms of trade is less likely to be reflected in higher economic growth. In this article, we explore this type of non-linearity using an empirical framework spanning the period 2002–2025. Thus, we estimate a Bayesian vector autorregressive model with thresholds and mean-stochastic volatility (TBVAR-SV) for the Peruvian economy. We include the stochastic volatility in mean component to control for outliers, such as the COVID-19 pandemic. We use business confidence as a threshold variable and identify terms-of-trade shocks using traditional zero restrictions. We explore the dynamic effect of these shocks on GDP growth for both optimistic and pessimistic regimes, also using a relevant set of macroeconomic variables for Peru. We find that terms-of-trade shocks are statistically significant for GDP only in the case of the optimistic regime, even for larger shocks. Ultimately, the empirical results validate our previous hypothesis. |
| Presented by: Fernando Perez Forero, BCRP |
Theory-based priors for the output gapAbstractEstimating potential output and the output gap requires identifying restrictions on the dynamics of trend and cycle. We develop a general econometric framework that specifies theory-based restrictions directly over the joint distribution of the latent trajectory, rather than through a particular state evolution equation. The approach nests standard unobserved-components models, while allowing external quantitative or qualitative information to be incorporated in a transparent and precision-weighted manner. Simulation evidence suggests that our prior can recover potentially asymmetric cyclical evolutions without having to estimate potentially challenging non-linear models. An empirical application to U.S. GDP shows that the Covid pandemic was not associated with a drop in the potential level of output, but the Great Financial Crisis was. |
| Presented by: Andrea Renzetti, Bank of England; Bocconi University |
| Session 137: NATURAL DISASTERS June 25, 2026 15:45 to 17:30 Location: B129 |
| Session Chair: Asli Leblebicioglu, Baruch College, CUNY |
The mitigating effect of social capital on post-disaster unemploymentAbstractEvidence shows that after climate disasters economic conditions often deteriorate. With climate change increasing the severity and frequency of disasters, there is growing interest in building disaster-resilient communities. Despite this, little is known regarding the complex short-run dynamics of recovery trajectories or the potential mitigating capacity of social capital. One dimension of social capital which has received little attention despite being potentially crucial for post-disaster economic recovery is economic connectedness or the extent to which low-SES individuals are friends with high-SES individuals. This study investigates the economic impact of Hurricane Ian- the third costliest hurricane in history- in the two years post-disaster using monthly unemployment data combined with a novel measure of economic connectedness based on billions of Facebook friendships. We establish oscillating unemployment impacts across the following three distinct phases: 1) the initial shock phases (first 1-3 months, unemployment increases by 0.172 percentage points), 2) the medium-term recovery phase (next 4-9 months, unemployment returns to baseline), and 3) the longer-term rebound phase (final 13-24 months, unemployment increases by 0.377 percentage points). We show that economic connectedness has no mitigating effect in the initial shock phase or medium-term recovery phase. However, a one-standard deviation increase in the extent to which low-SES individuals are connected with high-SES individuals reduces the effect of the hurricane on unemployment in the longer-term rebounds phase by 0.220 percentage points. Compared to other measures of social capital and disaster resilience, this mitigating effect is unique to this measure of economic connectedness. Furthermore, our analysis reveals that our results are driven both by having chances to interact with high-SES individuals and befriending them conditional on interaction. Finally, we rule out potential mechanisms such as informal financial lending and influencing of norms, behaviours, and expectations. Thus, our results highlight the potential of cross-class connections to support post-disaster job-search efforts of low-SES populations. This is most likely by facilitating access to resources such as information regarding job listings, career advice, and influential referrals. Overall, economic connectedness can be a useful tool for building disaster resilient communities and potentially reducing income gaps in post-disaster settings. |
| Presented by: Lihini de Silva, Monash University |
Natural Disasters and Fiscal SheltersAbstractUsing a novel dataset on U.S. natural disasters and high frequency measures of economic activity, we evaluate the effectiveness of federal disaster assistance. Exploiting quasi-random variation in whether aid from the Federal Emergency Management Agency is granted or denied, we compare otherwise similar events. States receiving aid recover within 20 weeks, while denied states face deeper and more persistent contractions. Recovery is stronger when aid is timely, generous, and includes direct transfers. Pre disaster mitigation lowers future disaster frequency and costs, while stronger fiscal capacity enhances resilience by enabling governments to sustain post disaster recovery. |
| Presented by: Álvaro Fernández-Gallardo Romero, Bank of Spain |
Physical climate risk, credit risk and lending activityAbstractWe study how physical climate risk shapes credit risk, lending conditions, and banks’ risk-taking by combining high-resolution Copernicus geospatial data with loan-level observations from AnaCredit. Exploiting three major European floods (2022–2024) as natural experiments, we implement a spatial regression discontinuity design comparing firms located just inside versus just outside flood boundaries (within 300–500 meters). Flood exposure triggers pronounced dynamics: lending volumes increase by about 5% immediately after floods, reflecting liquidity needs, but decline by roughly 6% in the subsequent quarter as investment demand contracts. Interest rates follow a similar pattern, while default rates rise by 0.7 percentage points—about one-third of the baseline. To isolate transmission mechanisms, we exploit multiple lending relationships and absorb firm–time fixed effects, decomposing the estimated response into demand and supply components. The results indicate that demand factors dominate, with limited evidence of supply-side constraints. Nonetheless, relationship strength amplifies adjustments: main banks extend roughly 10 percentage points more credit to affected borrowers but impose tighter collateral terms. The findings shed light on how physical climate shocks propagate through credit markets and inform climate stress-testing and financial stability analysis. |
| Presented by: Davor Djekic, European Central Bank |
The Catalytic Impact of the World Bank's Climate Finance in Emerging MarketsAbstractThis paper analyzes whether World Bank mitigation finance catalyzes additional green investment by public and private actors beyond the projects it directly finances. We assemble a novel subnational dataset combining project-level World Bank mitigation lending with comprehensive data on local green investment across five major emerging economies—Brazil, China, India, Indonesia, and Mexico—over 2010–2021. Exploiting within subnational location variation, we estimate the catalytic impact of World Bank finance along both the extensive margin (the number of green projects) and the intensive margin (average project size). We find that a $100 million increase in World Bank mitigation finance in a locality raises the number of green investment projects by approximately 8%. The effect holds for both public and private investors and is stronger in locations with greater financial development, better infrastructure, and higher human development. We also find that World Bank mitigation finance increases average project size, although at a diminishing rate. At the sample mean, a $100 million increase in World Bank mitigation finance is associated with roughly $600 million in additional green investment within a location per year. |
| Presented by: Asli Leblebicioglu, Baruch College, CUNY |
| Session 138: PANEL AND NETWORK MODELS June 25, 2026 15:45 to 17:30 Location: D-106 |
| Session Chair: Chris Muris, McMaster University |
Triadic Network FormationAbstractWe study estimation and inference for triadic link formation with dyad-level fixed effects in a nonlinear binary choice logit framework. Dyad-level effects provide a richer and more realistic representation of heterogeneity across pairs of dimensions (e.g. importer–exporter, importer–product, exporter–product), yet their sheer number creates a severe incidental parameter problem. We propose a novel “hexad logit” estimator and establish its consistency and asymptotic normality. Identification is achieved through a conditional likelihood approach that eliminates the fixed effects by conditioning on sufficient statistics, in the form of hexads—wirings that involve two nodes from each part of the network. Our central finding is that dyad-level heterogeneity fundamentally changes how information accumulates. Unlike under node-level heterogeneity, where informative wirings automatically grow with link formation, under dyad-level heterogeneity the network may generate infinitely many links yet asymptotically zero informative wirings. We derive explicit sparsity thresholds that determine when consistency holds and when asymptotic normality is attainable. These results have important practical implications, as they reveal that there is a limit to how granular or disaggregate a dataset one can employ under dyad-level heterogeneity. |
| Presented by: Cavit Pakel, University of Oxford |
(Debiased) Inference for Fixed Effects Estimators with Three-Dimensional Panel and Network DataAbstractInference for fixed effects estimators of linear and nonlinear panel models is often unreliable due to Nickell- and/or incidental parameter biases. This article develops new inferential theory for (non)linear fixed effects M-Estimators with data featuring a three-dimensional panel structure, such as sender × receiver × time. Our theory accommodates bipartite, directed, and undirected network panel data, integrates distinct specifications for additively separable unobserved effects with different layers of variation, and allows for weakly exogenous regressors. Our analysis reveals that the asymptotic properties of fixed effects estimators with three-dimensional panel data can deviate substantially from those with two-dimensional panel data. While for some specifications the estimator turns out to be asymptotically unbiased, in other specifications, it suffers from a particularly severe inference problem, characterized by a degenerate asymptotic distribution and complex bias structures. We address this atypical inference problem, by deriving explicit expressions to debias the fixed effects estimators. |
| Presented by: Daniel Czarnowske, |
Endogeneity in Nonparametric Panel Data Estimation: Application to Conditional Production FrontiersAbstractThis paper develops a nonparametric panel data framework for production frontier estimation in the presence of endogenous environmental variables. In frontier models, endogeneity does not only bias conditional mean relationships but may distort identification of the production boundary itself through violations of boundary independence. We address this problem by proposing a fully nonparametric control-function approach that restores identification of the conditional frontier under endogeneity. Exploiting the time structure of panel data, we show that lagged outcomes and controls can be used to construct internal instruments within a nonseparable framework. This allows inputs and output to be “purged” of endogenous external factors prior to frontier estimation. The resulting transformed variables define a production technology that is orthogonal to endogenous controls, ensuring consistent estimation of both full and robust order-m (quantile) frontiers. We establish consistency and asymptotic normality of the proposed estimators. An empirical application to Italian listed firms (2014–2023) examines the role of sustainability and digital finance in shaping productivity. Sustainability is associated with a systematic outward shift of the production frontier across the firm size distribution. In contrast, digital finance exhibits heterogeneous effects: it shifts the frontier downward for smaller firms but upward for larger firms, indicating size-dependent complementarities. These results highlight the importance of accounting for endogeneity when evaluating technological shifts and efficiency. |
| Presented by: Camilla Mastromarco, University of Calabria |
An Adversarial Approach to IdentificationAbstractWe introduce a new framework for characterizing identified sets of structural and counterfactual parameters in econometric models. By reformulating the identification problem as a set membership question, we leverage the separating hyperplane theorem in the space of ob- served probability measures to characterize the identified set through the zeros of a discrep- ancy function with an adversarial game interpretation. The set can be a singleton, resulting in point identification. A feature of many econometric models, with or without distributional assumptions on the error terms, is that the probability measure of observed variables can be expressed as a linear transformation of the probability measure of latent variables. This structure provides a unifying framework and facilitates computation and inference via linear programming. We demonstrate the versatility of our approach by applying it to nonlinear panel models with fixed effects, with parametric and nonparametric error distributions, and across various exogeneity restrictions, including strict and sequential. |
| Presented by: Chris Muris, McMaster University |
| Session 139: SYNTHETIC CONTROL AND EVENT STUDIES June 25, 2026 15:45 to 17:30 Location: D-105 |
| Session Chair: Timo Schenk, Erasmus University Rotterdam |
Forecasting Synthethic ControlAbstractWe introduce a novel forecasting method employing global deep learning models for estimating the causal effects of interventions across multiple units, incorporating counterfactual and synthetic control for policy evaluation in shared markets. This approach addresses potential spillover effects and leverages time series data for identification. We redefine causal effect estimation as predicting outcomes without intervention, first estimating counterfactual outcomes using high-dimensional time series data. This process utilizes cross-correlation in time series, employing an autoregressive recurrent neural network with parameter sharing. The second stage estimates and tests the average treatment effect on the target variable for statistical significance. Demonstrated through simulations and empirical studies, our method uniquely estimates effects using pre-treatment data in scenarios where traditional control unit assumptions fail. An empirical example estimates the impact of promotional deals on US grocery store sales, showcasing the method's applicability and contribution to existing literature. |
| Presented by: Klaus Ackermann, Monash University |
Out-of-sample gravity predictions and trade policy counterfactualsAbstractGravity equations are often used to evaluate the effect of trade policies, such as regional trade agreements. We argue that their suitability for this purpose crucially depends on their out-of-sample predictive power. We propose a methodology to evaluate the performance of out-of-sample predictions obtained with gravity equations and with machine learning methods. We find that the 3-way gravity model is difficult to beat when the purpose is to evaluate policy interventions, further cementing this model’s place as the predominant tool for applied trade policy analysis. However, when the goal is to predict individual flows, machine learning methods can be preferable. |
| Presented by: Nicolas Apfel, University of Innsbruck |
Difference-in-differences for mediation analysis using double machine learningAbstractWe propose a difference-in-differences (DiD) framework with mediation for possibly multivalued discrete or continuous treatments and mediators, aimed at identifying the direct effect of the treatment on the outcome (net of effects operating through the mediator), the indirect effect via the mediator, and the joint effects of treatment and mediator, consistent with the framework of dynamic treatment effects. Identification relies on a conditional parallel trends assumption imposed on the mean potential outcome across treatment and mediator states, or (depending on the causal parameter) additionally on the mean potential outcomes and potential mediator distributions across treatment states. We propose ATET estimators for repeated cross sections and panel data within the double/debiased machine learning framework, which allows for data-driven control of covariates, and we establish their asymptotic normality under standard regularity conditions. We investigate the finite-sample performance of the proposed methods in a simulation study and illustrate our approach in an empirical application to the US National Longitudinal Survey of Youth, estimating the direct effect of health care coverage on general health as well as the indirect effect operating through routine checkups. |
| Presented by: Sarina Oberhänsli, University of Fribourg |
Inference in Event Studies with Approximately Parallel TrendsAbstractExamining pre-trend violations is a common approach to validating the parallel trends assumption necessary for difference-in-difference designs. When parallel trends are rejected by the data, however, applied researchers commonly continue to interpret treatment effect estimates as long as the observed pre-trend violations are deemed sufficiently small. To rationalize this behavior, we recast difference-in-differences designs as relying solely on an approximate version of parallel trends, which allows parallel trends to fail in some realizations of the data. Our reformulation delivers new inference procedures that account for uncertainty about possible deviations from parallel trends. |
| Presented by: Timo Schenk, Erasmus University Rotterdam |
| Session 140: TEXT ANALYSIS IN MACROECONOMICS June 25, 2026 15:45 to 17:30 Location: D-114 |
| Session Chair: Jiaming Mao, Xiamen University |
What's the story? The media channel of monetary policy transmission to the publicAbstractThe media play a crucial intermediary role in transmitting central bank messages. This paper investigates how media coverage of European Central Bank (ECB) communication influences consumer inflation expectations, with a particular focus on what monetary policy related topics matter most to consumers. I identify seven key topics: interest rates, inflation, economic growth, purchase programme, uncertainty, fiscal policy, and financial markets. I use Latent Semantic Indexing with factor rotation to measure focus on those topics in media coverage of the ECB in leading daily economic outlets in the euro area. Employing an event-study approach, I isolate shifts in media topic focus and assess their impact on expectations through local projections. The findings reveal that media coverage significantly affects consumer inflation expectations, with discussions on inflation and economic growth raising expectations, while coverage on financial markets dampens them. Responses to the interest rate and purchase programme topics are in line with the central bank information effect. Rather than the central bank's own messaging, I find that the responses are mainly driven by the media's framing of ECB communication. Although media generally reinforce ECB messaging, the exception is the fiscal (policy) topic, where consumer expectations move in opposing directions depending on the source. These findings offer a new perspective on the role of media in the transmission of central bank communication on public expectations. |
| Presented by: Laura Pagenhardt, |
Bubbles Talk: Narrative Augmented Bubble PredictionAbstractFinancial bubble theories emphasize the importance of behavioral mechanisms centered around investor beliefs, which can be potentially gleaned from prevailing narratives, that reflect investors' psychological states and link them to economic events. By summarizing market narratives into meaningful and economically relevant features, guided by bubble theories, we offer a novel approach to bubble prediction. We then test whether the variation of narratives and bubble measures are related on a predictive basis, as bubble theories imply. Our findings reveal that most of our narrative features exhibit statistically significant predictive power for bubble measures, and that the narrative-augmented models outperform non-augmented benchmarks in out-of-sample tests. These results offer new insights into the understanding of bubbles and lay the foundation for using narratives to develop early warning systems (EWS) for bubble formation and deflation, and for investigating the causal relationship between narratives and economic events. |
| Presented by: Yuting Chen, University College Cork |
Forecasting the macroeconomy with corporate disclosures and language modelsAbstractThis paper examines the predictive power of corporate disclosures in macroeconomic forecasting. Leveraging advanced language models, I analyze the prevalence and tonal polarities of 31 accounting topics across 890,277 10-K and 10-Q filings. Firm disclosures exhibit superior performance in long-horizon forecasts, with their predictive strength primarily driven by discussions on corporate income taxes, revenue performance, mergers and acquisitions, and repurchase agreement activities. These topics transmit information to macroeconomic outcomes through firms’ capital expenditures and dividend policies. Moreover, firm disclosures provide particularly strong forecasts of GDP and Investment during periods of economic downturn, outperforming the FRED-MD dataset. |
| Presented by: Tri Phan, University of Basel |
Structural RegularizationAbstractWe propose a method for modeling data by using structural models based on economic theory as regularizers for statistical models. We show that even if a structural model is misspecified, as long as it is informative about the data-generating mechanism, our method can outperform both the (misspecified) structural model and unregularized statistical models. The method permits a Bayesian interpretation of theory as prior knowledge and can be used both for statistical prediction and causal inference. It contributes to transfer learning by showing how incorporating theory into statistical modeling can significantly improve out-of-domain predictions and offers a way to synthesize reduced-form and structural approaches to causal inference. Simulation experiments demonstrate the potential of our method in various settings, including first-price auctions, dynamic models of entry and exit, and demand estimation with instrumental variables. Finally we apply our method to empirical analysis of risk preferences. Our method has potential applications not only in economics, but in other scientific disciplines whose theoretical models offer important insight but are subject to significant misspecification concerns. |
| Presented by: Jiaming Mao, Xiamen University |
| Session 141: WORK AND TECHNOLOGY 1 June 25, 2026 15:45 to 17:30 Location: D-113 |
| Session Chair: Daiji Kawaguchi, University of Tokyo |
Human–Machine Gap in Work Allocation: A Revealed Preference ApproachAbstractWe study how decision-makers allocate work between humans and machines when both inputs are technologically equivalent under controlled price variation. Using an online experiment on Prolific, in which decisions have real monetary consequences, we elicit revealed preferences over human–machine work allocation and structurally estimate behavioral models. We identify and estimate a positive "human premium", defined as the systematic tendency to allocate more work to a human worker than to a machine at identical relative prices. Our results indicate a high degree of internal consistency in individual choices, meaning that participants behave in accordance with well-defined preferences. Structural estimates reveal that participants do not fully maximize their own profits and do not perceive human and machine inputs as perfect substitutes. To address heterogeneity, we also examine the relationship between the individual human premium and individual characteristics collected in the second part of the experiment. Overall, these findings suggest that intrinsic preferences for human involvement in production may represent an important behavioral constraint on the diffusion of automation technologies in the workplace. |
| Presented by: Mikhail Anufriev, University of Technology Sydney |
The Earnings Premium for Long Hours: A Directed Search ApproachAbstractThere are substantial differences across occupations in the earnings premium for working long weekly hours, causing gender-biased sorting and inequality in the labor market. To explain why these premiums emerge, previous work emphasizes the role of firm-side production technologies. However, hours and wages are equilibrium outcomes, which also depend on worker-side preferences and the labor market structure. To study the importance of these forces, we build a tractable directed search and matching framework with two-sided multidimensional heterogeneity, where wages and hours are jointly endogenous. We estimate the model using data representative for the upper end United States labor market. We find that productivity differences are important in shaping the large earnings premiums in Business and Law occupations. Yet to a large extent, the earnings premium simply compensates workers for the additional disutility associated with long working hours. In tighter markets, such as Tech, the added bargaining power of workers inflates the premium. Counterfactual simulations show that removing firms’ incentives to reward long hours reduces the gender gap in earnings marginally, while equalizing the worker-side costs of working long hours almost halves the gap. |
| Presented by: Sebastiaan Maes, University of Antwerp |
Can Capping Overtime Improve Worker Welfare? Evidence from Japan’s 2019 Labor ReformAbstractWe study Japan’s 2019 overtime cap—360 hours annually, or approximately 47 hours per week—using panel data from 2015 to 2023. Workers who previously exceeded the threshold significantly reduced their working hours, with little loss in annual earnings. The reform improved self-reported health and work-life balance, while job satisfaction and skill development opportunities remained unaffected. Firms appear to have responded by reorganizing tasks rather than increasing work intensity. These findings suggest that capping overtime can enhance worker welfare without adverse labor market effects in contexts where long working hours are widespread. |
| Presented by: Daiji Kawaguchi, University of Tokyo |
| # | Participant | Roles in Conference |
|---|---|---|
| 1 | Ackermann, Klaus | P139 |
| 2 | Afanasyeva, Elena | P110 C110 |
| 3 | Ahn, Young | P117 |
| 4 | Ahonon, Borel | P107 |
| 5 | Ahsan, Md Nazmul | P85 |
| 6 | Akgun, Oguzhan | P92 |
| 7 | Alabrese, Eleonora | P43 |
| 8 | Alegre, Joan | P30 |
| 9 | Alessandrini, Diana | P8 |
| 10 | Alm, Henrike | P82 C82 |
| 11 | Anatolyev, Stanislav | P88 |
| 12 | Anufriev, Mikhail | P141 |
| 13 | Apfel, Nicolas | P139 |
| 14 | Arce-Alfaro, Gabriel | P122 |
| 15 | Arduini, Tiziano | P76 |
| 16 | Arenas, Jorge | P34 |
| 17 | Armendariz Pacheco, Ana | P40 C40 |
| 18 | Arteche, Josu | P26 |
| 19 | Bahadir, Berrak | P53 |
| 20 | Bailey, Natalia | P132 |
| 21 | Balatti, Mirco | P36 |
| 22 | Barbaglia, Luca | P24 C24 |
| 23 | Barcelo, Cristina | P56 |
| 24 | Barrette, Christophe | P51 |
| 25 | Bauer, Gregory | P4 |
| 26 | Bauer, Lukas | P111 |
| 27 | Baxa, Jaromir | P39 |
| 28 | Belotti, Federico | P11 C11 |
| 29 | Ben-Moshe, Dan | P24 |
| 30 | Bertino, Piero | P40 |
| 31 | Beyhum, Jad | P92 |
| 32 | Bia, Michela | P104 C104 |
| 33 | Bingley, Paul | P27 |
| 34 | Bini, Lapo | P36 C36 |
| 35 | Blevins, Jason | P14 |
| 36 | Boldea, Otilia | P91 |
| 37 | Bondarenko, Yevheniia | P22 |
| 38 | Bonfim, Diana | P130 |
| 39 | Bordon, Paola | P109 |
| 40 | Borowczyk-Martins, Daniel | P71 |
| 41 | Boss, Konstantin | P112 |
| 42 | Botosaru, Irene | P131 C131 |
| 43 | Boucinha, Miguel | P123 C123 |
| 44 | Braun, Robin | P17 |
| 45 | Brehl, Nina | P134 |
| 46 | Brignone, Davide | P31 |
| 47 | Bulligan, Guido | P83 |
| 48 | Bun, Maurice | P130 |
| 49 | Cai, Yong | P126 |
| 50 | Cakir Melek, Nida | P55 |
| 51 | Campos-Martins, Susana | P16 |
| 52 | Caner, Mehmet | P23 |
| 53 | Capera Romero, Laura | P23 C23 |
| 54 | Carrasco, Marine | P59 |
| 55 | Carrion-i-Silvestre, Josep Lluís | P61 |
| 56 | Carvalho, Jose | P87 |
| 57 | Casalis, André | P52 C52 |
| 58 | Cascaldi-Garcia, Danilo | P57 |
| 59 | Celebi, Kaan | P109 |
| 60 | Chang, Yoosoon | P105 |
| 61 | Chang, Heejee | P10 |
| 62 | Chang, Shi Ryoung | P45 |
| 63 | Chankova, Rosi | P55 C55 |
| 64 | Chavez Lopez, Pedro Isaac | P89 |
| 65 | Chen, Chaoyi | P51 |
| 66 | Chen, Ertian | P131 |
| 67 | Chen, Jesse | P45 C45 |
| 68 | Chen, Yuting | P140 |
| 69 | Cho, Sung-Jin | P67 C67 |
| 70 | Chong, Carsten | P85 C85 |
| 71 | Chow, Chun Pang (Alex) | P14 C14 |
| 72 | Christensen, Bent Jesper | P116 |
| 73 | Christiaens, Lison | P9 |
| 74 | Christy, Neil | P126 |
| 75 | Chrysanthou, Georgios | P40 |
| 76 | Chrysikou, Katerina | P32 |
| 77 | Chua, Yeow Hwee | P35 |
| 78 | Chudik, Alexander | P125 |
| 79 | Ciacci, Riccardo | P13 |
| 80 | Clark, Todd | P31 |
| 81 | Coleman, Winnie | P74 |
| 82 | Colombo, Daniele | P38 |
| 83 | Conrad, Christian | P65 |
| 84 | Contreras, Gabriela | P8 |
| 85 | Corblet, Pauline | P27 |
| 86 | Cornea-Madeira, Adriana | P120 C120 |
| 87 | Coroneo, Laura | P134 |
| 88 | Costa-Andreu, Eric | P69 |
| 89 | Croushore, Dean | P7 |
| 90 | Cubadda, Gianluca | P102 |
| 91 | Cuesta, Diego | P21 |
| 92 | Czarnowske, Daniel | P138 |
| 93 | Dal Borgo, Mariela | P21 |
| 94 | Dalderop, Jeroen | P127 |
| 95 | Dörsam, Michael | P13 |
| 96 | de la Cal Medina, Jorge | P86 C86 |
| 97 | De Nicolo', Gianni | P46 |
| 98 | de Oliveira, Jean | P115 |
| 99 | De Paula, Heloisa | P51 C51 |
| 100 | de Silva, Lihini | P137 |
| 101 | Degasperi, Riccardo | P12 |
| 102 | Del Negro, Marco | P84 C84 |
| 103 | Dendramis, Yiannis | P102 |
| 104 | Di Addario, Sabrina | P8 C109 |
| 105 | Di Francesco, Damiano | P83 |
| 106 | Di Francesco, Riccardo | P80 |
| 107 | Diks, Cees | P17 |
| 108 | Dionisi, Valerio | P36 |
| 109 | Ditzen, Jan | P119 |
| 110 | Djekic, Davor | P137 |
| 111 | Dovonon, Prosper | P26 C26 |
| 112 | Drudi, Maria Ludovica | P23 |
| 113 | Dubois, Pierre | P93 C93 |
| 114 | Dzuverovic, Emilija | P78 |
| 115 | Eiling, Esther | P16 |
| 116 | Eizenberg, Alon | P93 |
| 117 | Engle, Samuel | P30 |
| 118 | Enilov, Martin | P4 C4 |
| 119 | Ericsson, Neil | P52 |
| 120 | Fadhuile, Adelaide | P43 |
| 121 | Fan, Junwei | P25 C25 |
| 122 | Fankhauser, Martin | P124 |
| 123 | Fasani, Stefano | P10 C10 |
| 124 | Fernandez, Julian | P108 |
| 125 | Fernandez-Val, Ivan | P104 |
| 126 | Fernández-Gallardo Romero, Álvaro | P137 |
| 127 | Figueres, Juan | P31 |
| 128 | Figuerola Ferretti, Isabel | P93 |
| 129 | Fiori, Giuseppe | P22 |
| 130 | Fisgin, Ece | P12 C12 |
| 131 | Flaccadoro, Marco | P123 |
| 132 | Fonseca Mendes, Eduardo | P24 |
| 133 | Fossati, Sebastian | P89 |
| 134 | Fosten, Jack | P69 |
| 135 | Frankovic, Ivan | P28 |
| 136 | Freire, Gustavo | P4 |
| 137 | Freyberger, Joachim | P124 C124 |
| 138 | Fuentes-Albero, Cristina | P103 |
| 139 | Fuertes Pina, Javier | P102 C102 |
| 140 | Gabriel, Ricardo Duque | P108 |
| 141 | Gabriel, Ricardo Duque | P133 |
| 142 | Gadea, Maria Dolores | P86 |
| 143 | Gagete-Miranda, Jessica | P43 |
| 144 | Galbraith, John | P39 |
| 145 | Galvao, Ana Beatriz | P81 |
| 146 | Galvez, Julio | P6 P62 |
| 147 | Garcia-Cabo, Joaquin | P27 C27 |
| 148 | Geiger, Martin | P37 |
| 149 | Gelain, Paolo | P51 |
| 150 | Genesove, David | P98 |
| 151 | Gentini, Vítor | P99 C99 |
| 152 | Gersing, Philipp | P69 C69 |
| 153 | Giacardi, Alessandro | P115 |
| 154 | Giersbergen, Noud | P5 |
| 155 | Gomez-Rodriguez, Fabio | P121 |
| 156 | Gonzalez, Alejandro | P122 C122 |
| 157 | Gonzalez, Diego | P62 |
| 158 | González-Astudillo, Manuel | P121 |
| 159 | Gortz, Christoph | P35 |
| 160 | Gradzewicz, Michał | P14 |
| 161 | Gründler, Daniel | P107 C107 |
| 162 | Guerron, Pablo | P136 |
| 163 | Gumus, Inci | P73 |
| 164 | Gutknecht, Daniel | P111 |
| 165 | Gwak, Yun Young | P70 |
| 166 | Hajivassiliou, Vassilis | P61 |
| 167 | Haller, Timo | P129 |
| 168 | Hansen, Erwin | P53 |
| 169 | Harrison, Benjamin | P29 |
| 170 | Harvey, David | P65 |
| 171 | Hasanbasri, Arifah | P93 |
| 172 | Heinisch, Katja | P7 |
| 173 | Henderson, Daniel | P125 |
| 174 | Henry, Emma | P92 |
| 175 | Herculano, Muguel | P20 |
| 176 | Herrera Bravo, Luis | P90 |
| 177 | Herstad, Eyo | P100 |
| 178 | Hienzsch, Sebastian | P133 C133 |
| 179 | Hong, Seung-Hyun | P67 |
| 180 | Horn, Carl-Wolfram | P90 C90 |
| 181 | Hospido, Laura | C62 |
| 182 | Hou, Chenyu | P22 |
| 183 | Hsieh, Meng Hsuan | P76 |
| 184 | Hu, Qidi | P98 C98 |
| 185 | Hua, Sudong | P44 C44 |
| 186 | Hungnes, Håvard | P31 C31 |
| 187 | Isaak, Niklas | P109 |
| 188 | Iskrev, Nikolay | P107 |
| 189 | Issler, João | P12 |
| 190 | Istrefi, Klodiana | P75 |
| 191 | Jagelka, Tomas | P62 C126 |
| 192 | Janssens, Eva | P103 |
| 193 | Jelnov, Pavel | P41 |
| 194 | Jensen, Sebastian | P58 |
| 195 | Jensen, Christian | P122 |
| 196 | Jochmans, Koen | P68 |
| 197 | Junare, Paritosh | P114 |
| 198 | Jung, Hyunseok | P92 C92 |
| 199 | Kaliyeva, Anel | P80 C80 |
| 200 | Kandelhardt, Johannes | P67 |
| 201 | Karamysheva, Madina | P135 |
| 202 | Kasenally, Fatima | P106 |
| 203 | Kawaguchi, Daiji | P141 C141 |
| 204 | Kazakova, Ekaterina | P32 C32 |
| 205 | Keating, John | P110 |
| 206 | Keilbar, Georg | P59 |
| 207 | Kheifets, Igor | P26 |
| 208 | Khodaverdian, Saeed | P72 |
| 209 | Kilian, Lutz | P42 |
| 210 | Kilic, Rehim | P7 |
| 211 | Klaassen, Sven | P59 C59 |
| 212 | Kleen, Onno | P46 |
| 213 | Koh, Yookyung Julia | P42 |
| 214 | Kole, Erik | P130 |
| 215 | kong, lingwei | P4 |
| 216 | Kooiker, Sicco | P78 |
| 217 | Korobka, Yaroslav | P88 C88 |
| 218 | Kouki, Amairisa | P13 |
| 219 | Koundouros, Andreas | P22 C22 |
| 220 | Kourtellos, Andros | P59 |
| 221 | Kozyrev, Boris | P69 |
| 222 | Krabbe, Frederik | P17 C17 |
| 223 | Kritzinger, Mara | P123 |
| 224 | Kruse-Becher, Robinson | P65 |
| 225 | Kuitunen, Laura | P73 C73 |
| 226 | Kulikova, Yuliya | P118 |
| 227 | Kulish, Mariano | P52 |
| 228 | Kuo, Biing-Shen | P54 |
| 229 | Kurcz, Frederik | P75 C75 |
| 230 | Kurmann, André | P75 |
| 231 | Kuschnig, Nikolas | P126 |
| 232 | Kutz, Johanna | P63 |
| 233 | Kynigakis, Iason | P46 C46 |
| 234 | Lahaye, Jerome | P89 |
| 235 | Langevin, Raphaël | P13 |
| 236 | Laseen, Stefan | P120 |
| 237 | López-Nieto Veitch, Tomeu | P105 C105 |
| 238 | Leblebicioglu, Asli | P137 C137 |
| 239 | Lee, Sungwon | P15 |
| 240 | Lee, Tae-Hwy | P107 |
| 241 | Lehnert, Thorsten | P35 C127 |
| 242 | Leibing, Andreas | P44 |
| 243 | Leon, Ana Sofia | P56 |
| 244 | Leventer, Dor | P129 |
| 245 | Lima, Luiz | P46 |
| 246 | Liu, Frances | P77 |
| 247 | Liu, Long | P45 |
| 248 | LIU, Chenyue | P104 |
| 249 | Liu, Yan | P131 |
| 250 | Liu, Jiaxun | P54 |
| 251 | LIU, Renzhi | P46 |
| 252 | Lorusso, Marco | P133 |
| 253 | Lumsdaine, Robin | P81 |
| 254 | Lyk-Jensen, Stéphanie | P25 |
| 255 | M. Magnusson, Leandro | P28 C28 |
| 256 | Maasoumi, Esfandiar | P84 |
| 257 | Macchia, Francesca | P19 |
| 258 | Madeira, Joao | P127 |
| 259 | Maes, Sebastiaan | P141 |
| 260 | Magazzini, Laura | P77 |
| 261 | Mao, Jiaming | P140 C140 |
| 262 | Marcellino, Massimiliano | P38 C38 |
| 263 | Marconi, Costanza | P100 C100 |
| 264 | Marcucci, Juri | P81 C81 |
| 265 | Marques, Eduardo | P34 |
| 266 | Martinez, Andrew | P28 |
| 267 | Martinez, Felipe | P70 C70 |
| 268 | Martinoli, Mario | P68 C68 |
| 269 | Maschmann, Christina | P45 |
| 270 | MASSEI, Thaïs | P123 |
| 271 | Mastromarco, Camilla | P138 |
| 272 | Matzner, Anna | P116 C116 |
| 273 | Mazzali, Marco | P60 |
| 274 | McCracken, Michael | P54 C54 |
| 275 | McCrary, Sean | P57 C57 |
| 276 | Mendonca, Francisco | P41 |
| 277 | Menzel, Konrad | P76 |
| 278 | Mertens, Elmar | P134 |
| 279 | Mesa-Ruiz, David | P87 |
| 280 | Metiu, Norbert | P120 |
| 281 | Miao, Wei | P125 C125 |
| 282 | Micheel, Ruven | P58 C58 |
| 283 | Mingoli, Gabriele | P114 |
| 284 | Modugno, Michele | P36 |
| 285 | Mogilevskaja, Anna | P55 |
| 286 | Mones, Pablo | P125 |
| 287 | Moneta, Alessio | P20 |
| 288 | Monroe, Win | P108 C108 |
| 289 | Moreira, Marcelo | P88 |
| 290 | Morgado Azevedo, Hugo | P100 |
| 291 | Morita, Rubens | P34 C34 |
| 292 | Morley, James | P70 |
| 293 | Mouton, Andre | P32 |
| 294 | Mućk, Jakub | P20 C20 |
| 295 | Mukherjee, Sukanya | P74 C74 |
| 296 | Muris, Chris | P138 C138 |
| 297 | Nemtyrev, Aleksei | P9 |
| 298 | Neri, Luca | P9 C9 |
| 299 | Newton, Alexander | P66 C66 |
| 300 | Nielsen, Morten | P91 |
| 301 | Nikolaishvili, Giorgi | P42 |
| 302 | Ning, Cathy | P16 |
| 303 | Nunes, Ricardo | P39 C39 |
| 304 | Oberhänsli, Sarina | P139 |
| 305 | Offen, Karoline | P101 |
| 306 | Ohashi, Kazuhiko | P34 |
| 307 | Oka, Tatsushi | P40 |
| 308 | Olma, Tomasz | P15 |
| 309 | Ortiz, Julio | P10 |
| 310 | Ozabaci, Deniz | P106 C106 |
| 311 | Paccagnini, Alessia | P101 |
| 312 | Pacini, David | P124 |
| 313 | Pagano Giorgianni, Giuseppe | P37 |
| 314 | Pagenhardt, Laura | P140 |
| 315 | Pakel, Cavit | P138 |
| 316 | Pal Mustafi, Utso | P110 |
| 317 | Palazzo, Dino | P53 C53 |
| 318 | Pallara, Kevin | P107 |
| 319 | Panchenko, Valentyn | P77 |
| 320 | Papagni, Francesca | P37 C37 |
| 321 | Pazzona, Matteo | P56 C56 |
| 322 | Peignon, Julien | P99 |
| 323 | Pellegrino, Giovanni | P55 |
| 324 | Pen, Michael | P30 |
| 325 | Pereda-Fernández, Santiago | P29 |
| 326 | Pereira, Angelo Sérgio | P102 |
| 327 | Pereira dos Santos, João | P21 |
| 328 | Perez Forero, Fernando | P136 |
| 329 | Perron, Benoit | P58 |
| 330 | Pesso, Tom | P10 |
| 331 | Pessoa, Ana Sofia | P133 |
| 332 | Petrova, Katerina | P114 |
| 333 | Petrunia, Robert | P115 C115 |
| 334 | Pfaffermayr, Michael | P61 C61 |
| 335 | Phan, Tri | P140 |
| 336 | Phella, Anthoulla | P19 C19 |
| 337 | Pinilla-Torremocha, Clemente | P84 |
| 338 | Pinto, Jeronymo | P7 C7 |
| 339 | POLSELLI, ANNALIVIA | P66 |
| 340 | Ponomarev, Kirill | P104 |
| 341 | Pop-Catalisan, Andrea | P118 C118 |
| 342 | Porcellotti, Giacomo | P86 |
| 343 | Posso, Christian | P80 |
| 344 | Prüser, Jan | P60 C60 |
| 345 | Prokhorov, Artem | P58 |
| 346 | Puerta-Cuartas, Alejandro | P83 C83 |
| 347 | Puglisi, Federico | P20 |
| 348 | Punder, Ramon | P114 C114 |
| 349 | Quelhas, João | P6 |
| 350 | Radchenko, Natalia | P105 |
| 351 | Raposo, Pedro | P29 |
| 352 | Rast, Sebastian | P39 |
| 353 | Ravazzolo, Francesco | P12 |
| 354 | Rüter, Lotta | P119 |
| 355 | Regis, Paulo | P115 |
| 356 | Remiat, Alexandre | P127 |
| 357 | Renzetti, Andrea | P136 C136 |
| 358 | Resende, Lucas | P11 |
| 359 | Reyes, Juan | P77 C77 |
| 360 | Rieth, Malte | P60 |
| 361 | Riva, Raul | P81 |
| 362 | Rodrigues, Afonso | P67 |
| 363 | Romero, Eva | P16 |
| 364 | Rossi, Lorenza | P101 |
| 365 | Rostam-Afschar, Davud | P119 C119 |
| 366 | Ruisi, Germano | P90 |
| 367 | Ruscoe, Bets | P117 C117 |
| 368 | Sagawa, Takaaki | P53 |
| 369 | Sakaguchi, Shosei | P63 |
| 370 | Salish, Nazarii | P112 |
| 371 | Salish, Mirjam | P119 |
| 372 | Samiahulin, Artem | P5 |
| 373 | Sanchez Becerra, Alejandro | P30 C30 |
| 374 | Sanford, Anthony | P65 C65 |
| 375 | Santi, Matteo | P6 C6 |
| 376 | Sarango-Iturralde, Alexander | P8 C8 |
| 377 | Sartori, Tommaso | P82 |
| 378 | Savcic, Ruzica | P15 C15 |
| 379 | Særkjær, Laust | P136 |
| 380 | Schafgans, Marcia | P15 |
| 381 | Schäper, Julius | P5 C5 |
| 382 | Schenk, Timo | P139 C139 |
| 383 | Schick, Manuel | P37 |
| 384 | Schmandt, Marco | P41 C41 |
| 385 | Schmitz, Theresa M. A. | P63 |
| 386 | Schneider-Strawczynski, Sarah | P41 |
| 387 | Schwab, Benedikt | P91 |
| 388 | Segato, Federico | P44 |
| 389 | Sekkel, Rodrigo | P28 |
| 390 | Sen, Sonkurt | P118 |
| 391 | Sentana, Enrique | P57 |
| 392 | Shanan, Yannay | P129 C129 |
| 393 | Sharifvaghefi, Mahrad | P59 |
| 394 | Shen, Shu | P21 C21 |
| 395 | Sheremirov, Viacheslav | P122 |
| 396 | Shi, Zhan | P135 C135 |
| 397 | Shi, Ruoyao | P106 |
| 398 | Shin, Myungkou | P117 |
| 399 | Shiotani, Takaaki | P23 |
| 400 | Sinani, Sofiana | P72 C72 |
| 401 | Skoblar, Ana | P99 |
| 402 | Skrobotov, Anton | P111 C111 |
| 403 | Smith, Ronald | P61 |
| 404 | Soccorsi, Stefano | P54 |
| 405 | Song, Suyong | P11 |
| 406 | Soofi Siavash, Soroosh | P116 |
| 407 | Sorensen, Bent | P71 C71 |
| 408 | Sousa-Leite, Joana | P90 |
| 409 | Spies, Julian | P85 |
| 410 | Stamatogiannis, Michalis | P4 |
| 411 | Stavrakoudis, Athanassios | P121 C121 |
| 412 | Stein, Hillary | P70 |
| 413 | Stoja, Evarist | P111 |
| 414 | Sul, Donggyu | P117 |
| 415 | Sunao, Stefanie | P74 |
| 416 | Sureka, Keshav | P103 C103 |
| 417 | Sutherland, Christopher | P110 |
| 418 | Szabo, Lajos | P109 |
| 419 | Szendrei, Tibor | P58 |
| 420 | Sznajderska, Anna | P89 C89 |
| 421 | Szydlowski, Arkadiusz | P43 C43 |
| 422 | Takahashi, Jun | P80 |
| 423 | Tatsiramos, Konstantinos | P25 |
| 424 | Tetenyi, Laszlo | P83 |
| 425 | Thie, Jan-Erik | P71 |
| 426 | Tibullo, Marco | P112 C112 |
| 427 | Tiedtke, Julian | P56 |
| 428 | Torrealba, Francisca | P73 |
| 429 | Torres Noblejas, Angela | P134 C134 |
| 430 | Trebbi, Giovanni | P121 |
| 431 | Trindade, Andre | P14 |
| 432 | Trinh, Duong | P76 C76 |
| 433 | Trovato, Giovanni | P108 |
| 434 | Tsafos, Yannis | P16 C16 |
| 435 | Tsakou, Katerina | P127 |
| 436 | Turkum, Betul | P118 |
| 437 | Ubilava, David | P87 |
| 438 | Umlandt, Dennis | P78 C78 |
| 439 | Urquiza, Juan | P101 C101 |
| 440 | Uruyos, Manachaya | P13 C13 |
| 441 | Uzeda, Luis | P75 |
| 442 | Vacca, Gianmarco | P19 |
| 443 | van den Akker, Ruben | P62 |
| 444 | van der Wel, Michel | P132 |
| 445 | van Vuuren, Aico | P105 |
| 446 | Vandermarlière, Kévin | P116 |
| 447 | Velasco, Carlos | P106 |
| 448 | Veraldi, Lucia | P86 |
| 449 | Vieira Nunes da Conceição, Diogo | P82 |
| 450 | Vilan, Diego | P60 |
| 451 | Virolainen, Savi | P91 C91 |
| 452 | Viviens, Javier | P66 |
| 453 | Vladimirov, Evgenii | P78 |
| 454 | Vladu, Andreea Liliana | P10 |
| 455 | Vogler, Jan | P85 |
| 456 | Volpicella, Alessio | P38 |
| 457 | Vonnák, Balázs | P135 |
| 458 | Wagner, Martin | P61 |
| 459 | Wang, Tao | P62 |
| 460 | Wang, Wendun | P66 |
| 461 | Wang, Zheng | P11 |
| 462 | Wang, Yiru | P57 |
| 463 | Wang, Chenhui | P26 |
| 464 | Wehrenberg, Nils | P129 |
| 465 | Wei, Aihui | P120 |
| 466 | Wei, Chia-Min | P29 C29 |
| 467 | Weiss, Christoph | P93 |
| 468 | Wesołowski, Grzegorz | P73 |
| 469 | Westphal, Matthias | P5 |
| 470 | Wildi, Marc | P54 |
| 471 | Wilson, Daniel | P112 |
| 472 | Wittekopf, David | P72 |
| 473 | Wolf, Elias | P132 |
| 474 | Wolff, François-Charles | P98 |
| 475 | Wolter, Stefanie | P32 |
| 476 | Wurdinger, Alex | P44 |
| 477 | Xu, Qianyan | P25 |
| 478 | Yan, Guo | P124 |
| 479 | Yap, Luther | P68 |
| 480 | Ye, Qianyao | P82 |
| 481 | Yilmaz, Kamil | P84 |
| 482 | Young, Justin | P63 C63 |
| 483 | Yu, Miao | P17 |
| 484 | Yuan, Menghan | P71 |
| 485 | Zaman, Saeed | P99 |
| 486 | Zanelli, Edoardo | P52 |
| 487 | Zanetti Chini, Emilio | P17 |
| 488 | Zeng, Ming | P35 C35 |
| 489 | Zhang, Shuoxun | P103 |
| 490 | Zhang, Tinghan | P131 |
| 491 | Zheng, Sinian | P35 |
| 492 | Zhong, Liang | P87 C87 |
| 493 | Zhou, Qichen | P132 C132 |
| 494 | Zhu, Yinchu | P68 |
| 495 | Zipitria, Leandro | P55 |
| 496 | Zivanovic, Jelena | P130 C130 |
| 497 | Zlatanos, Stylianos | P42 C42 |
This program was last updated on 2026-06-23 17:38:16 EDT