Society of Government Economists 2026 Annual Conference

April 17, 2026
Time	Location	Event
08:30 to 09:00	Grossman Hall	SGE Membership meeting, Grossman Hall
09:00 to 10:00	Grossman Hall	Plenary Session: The Future of Economic Research (and Data), Grossman Hall
10:15 to 12:00	see below	Concurrent Sessions 1
12:00 to 13:00		Lunch
13:15 to 15:00		Concurrent Sessions 2
15:15 to 17:00		Concurrent Sessions 3
17:00 to 18:30		Poster Session & Networking Reception

Summary of All Sessions
Click here for an index of all participants

#	Date/Time	Title/Location	Papers
1	April 17, 2026 10:15-12:00	From Farm to Fork to Policy: New Perspectives on Food and the Economy	3
2	April 17, 2026 10:15-12:00	From Streets to Statistics: Measuring Crime, Displacement, and Violence	4
3	April 17, 2026 10:15-12:00	How Households Cope: Consumption, Delinquency, and Income Support	3
4	April 17, 2026 10:15-12:00	Small Firms, Big Picture: GDP, Advisory Services, and Corporate Responsibility	3
5	April 17, 2026 10:15-12:00	The Cost of Care: Medicaid, Savings, and Preventable Spending	4
6	April 17, 2026 10:15-12:00	Who Works, Who Earns, Who Searches: New Measures of Labor Supply and Compensation	4
7	April 17, 2026 13:15-15:00	Access, Equity, and Reproductive Health: Evidence from Policy and Practice	3
8	April 17, 2026 13:15-15:00	Credit, Bubbles, and Climate: Financial Markets Under Pressure	4
9	April 17, 2026 13:15-15:00	Measuring What Matters: Data, Methods, and the Productivity Debate	4
10	April 17, 2026 13:15-15:00	Measuring What We Mean: Definitions, Data, and Economic Well-Being	4
11	April 17, 2026 13:15-15:00	The AI Disruption: Measuring Exposure, Work, and Market Response	4
12	April 17, 2026 13:15-15:00	Transshipment, Evasion, and Trade Policy	4
13	April 17, 2026 15:15-17:00	Capital, Capacity, and Credibility: Macroeconomic Dynamics in Theory and Data	3
14	April 17, 2026 15:15-17:00	Housing Costs and Household Strain: Insurance, Mortgages, and Rent	4
15	April 17, 2026 15:15-17:00	Innovation, Risk, and Reward: Firms, Patents, and the AI Frontier	4
16	April 17, 2026 15:15-17:00	Lowering the Transition: Storage, Efficiency, and the Value of Energy Assets	3
17	April 17, 2026 15:15-17:00	Prices in Practice: Measurement Challenges Across Markets and Households	4
18	April 17, 2026 15:15-17:00	The Utility of State Wage Data	4
19	April 17, 2026 17:00-18:30	Poster Session & Networking Reception	17

 

19 sessions, 83 papers, and 0 presentations with no associated papers

Society of Government Economists 2026 Annual Conference
Detailed List of Sessions

 

Session 1: From Farm to Fork to Policy: New Perspectives on Food and the Economy April 17, 2026 10:15 to 12:00
Session Chair: Cristina Miller, USDA Rural Development
Missing the Lunch Rush: Downtown Workforce Dynamics and the Food-Away-From-Home Economy Abstract The food-away-from-home (FAFH) sector—particularly in downtown areas—is closely linked to the presence of a daytime workforce. This study uses the COVID-19 pandemic as a natural experiment to examine how reductions in workplace activity reshape the FAFH economy. Drawing on anonymized, high-frequency mobility data from 2018 to 2022, we track weekly foot traffic to restaurants and other FAFH establishments across major U.S. downtowns. We define downtowns using the National Establishment Time Series, identifying census blocks with dense office agglomerations, and construct indices of workplace presence to measure daytime worker density before, during, and after pandemic disruptions. Using a difference-in-differences framework, we estimate the causal effects of reduced workplace presence on FAFH visits, comparing downtown areas with high concentrations of office employment to non-downtown areas with lower exposure to remote work. Findings provide new evidence on how workplace dynamics drive urban food demand and how persistent shifts toward remote and hybrid work may alter the structure of the FAFH sector. This work is relevant for policymakers and industry, as FAFH accounts for a growing share of U.S. food expenditures and plays a central role in employment, food access, and agricultural demand. By linking workplace-driven changes in FAFH consumption to the agricultural supply chain, the study informs strategies for strengthening food system resilience and supporting both producers and downstream markets. Results also shed light on how evolving food purchasing behaviors influence diet quality, access to prepared foods in urban areas, and the stability of markets dependent on foodservice demand. Together, these insights advance understanding of the long-term implications of workforce shifts for the U.S. food economy.
By Eliana Zeballos; USDA - ERS
Presented by: Eliana Zeballos, USDA - ERS
Discussant:   Sophie Collyer, Center on Poverty and Social Policy, Columbia University
Macro Consequences of Diet Quality and Health Abstract Health spending constituted the largest share (27%) of the federal budget amounting to $1.7 Trillion, 90% of which are used to treat chronic diseases. In 2024, 60-70% of the population has at least one chronic disease and this prevalence is expected to increase in the coming years causing health spending to rise unsustainably. I establish three key empirical results: better diets are correlated with less chronic diseases; better diets are costlier; chronic diseases impact earnings. At first, including these results as features into a simplified two-period model, I identify a structural wedge between private and social optimal choices of diet quality. Households under-invest in diet quality because they do not internalize the fiscal externalities of public health costs and lost tax revenue. I then evaluate three policy changes related to food subsidies: item-specific restrictions, price incentives, and in-kind transfers. I demonstrate analytically that restrictions and price subsidies are largely ineffective due to the fungibility of subsidies for infra-marginal recipients and the price inelasticity of diet quality respectively. In-kind transfers of high quality diets as in the recent "Food-Is-Medicine" movement correct for the externalities as it directly results in socially optimal diets. Next, I expand the simplified model into a full general equilibrium Aiyagari model with heterogeneity in assets, productivity, and health, calibrated to US data. I find that replacing existing benefits with high quality diet food bundles for the bottom 13–19.5% of the income distribution reduces the prevalence of chronic diseases by 6.6–10 percentage points and lowers aggregate medical spending by 1.6–2.4% of GDP. Importantly, expanding coverage to the 19.5\% percentile yields a pareto improvement, generating welfare gains for both recipients (+29.8%) and non-recipients (+6.6%) thereby resolving the double-paying inefficiency where public funds finance both non-optimal diets and the resulting medical care.
By Jonas Ho; University of California - Irvine
Presented by: Jonas Ho, University of California - Irvine
Discussant:   Eliana Zeballos, USDA - ERS
Predicting Rates of Food Insecurity in the United States in the Absence of Official Data Collection Abstract As of 2024, more than 47 million people in the United States (14.4%) lived in food-insecure households. In the coming years, however, we will not know whether the national prevalence of food insecurity has risen, fallen, or remained stable, as the USDA recently announced the permanent suspension of food security data collection on the Current Population Survey (CPS). The elimination of the CPS Food Security Supplement (FSS) leaves a critical gap in the national data on economic well-being. This paper presents a model that addresses this gap by predicting national food insecurity rates in the absence of official USDA data. The model draws on established correlates of food insecurity – national rates of poverty and unemployment, and food-specific inflation – to estimate food insecurity rates for all individuals, adults, children, and households. The predicted rates align closely with official food insecurity rates between 2010 and 2024, with a typical difference of 0.3 percentage points. Sensitivity tests show that the preferred model specification outperforms alternatives. The paper further demonstrates how this method can be used to assess changes in food insecurity associated with recent federal policy changes, including funding cuts to the Supplemental Nutrition Assistance Program (SNAP). While continuing to measure food insecurity using the method employed by the USDA since 1995 is the only way to guarantee consistent data on this critical indicator, the model presented here may prove useful in estimating food insecurity in future years when this USDA data is unavailable.
By Sophie Collyer; Center on Poverty and Social Policy, Columbia University
Presented by: Sophie Collyer, Center on Poverty and Social Policy, Columbia University
Discussant:   Jonas Ho, University of California - Irvine
Session 2: Small Firms, Big Picture: GDP, Advisory Services, and Corporate Responsibility April 17, 2026 10:15 to 12:00
Session Chair: Sarah Atkinson, USDA
Small Business GDP: 2007-2022 Abstract Gross Domestic Product (GDP) is a key measure of a country or region’s economic health. Areas with expanding economies based on real GDP growth relative to population can provide higher standards of living for their populations, while those that do not will tend to see declining standards of living. While real GDP is a convenient single measure for analyzing an economy, it also lacks nuance, so the Bureau of Economic Analysis (BEA) also puts out a host of companion measures as well. While the numbers provided by the BEA can document changes between industries, they cannot see changes within industries. Since firms of all different sizes, from independent contractors to multinational corporations, contribute to the economy, understanding which types of firms are driving industry growth can help provide further information to policy discussion surrounding GDP. To fill that gap this issue brief measures the economic contribution small businesses (employing fewer than 500 employees) made to the United States economy between 2007 and 2022, by calculating small business GDP (SGDP), using several publicly available government datasets. In 2022, Small business contributed $8.5 trillion of the $21.0 trillion private sector non-farm GDP, or 40.6 percent. Over the last sixteen years the small business share of GDP has decline by 5.6 percentage points, because large businesses have on average outgrown small businesses. Additionally, SGDP is disaggregated into the contributions of nonemployers, employers with between 1 and 99 employees, and employers with between 100 and 499 employees. This breakdown reveals that more than 100 percent of the decline in SGDP is from the declining share of GDP generated by small businesses with between 1 and 99 employees. Small businesses between 100 and 499 employees slightly increased share compared with 2007, while those with no employees slightly decreased share.
By Robert Press; Small Business Administration
Presented by: Robert Press, Small Business Administration
Discussant:   Dina El Mahdy, Morgan State University
The Impact of Publicly Funded Small Business Advisory Services: Firm Take-Up and Performance in the United States Abstract This paper studies the impact of geographic proximity to and utilization of publicly funded advisory services offered to US small businesses on firm take-up and performance. We leverage a novel administrative dataset from the Northern California Small Business Development Center (SBDC) Network covering all firm-center interactions from 2006–23. To address endogeneity in firm engagement with centers, we exploit exogenous variation in center-firm geographic proximity generated by center closures and openings. We instrument for paired center-firm consulting time with changes in distance resulting from these organizational shifts. A one standard deviation reduction in distance between a firm and corresponding center (20 miles) increases average annual consulting time by 0.15 hours; each additional consulting hour raises average annual firm revenue and employment by 5.1% and 1.7%, respectively. Back-of-the-envelope calculations suggest these advisory services are cost-effective. This study provides novel causal evidence on take-up and effectiveness of small business advisory services in the US using quasi-experimental variation in geographic proximity. Our findings highlight the importance of both physical distance and localized expertise in shaping small business outcomes, which may be particularly important in under-served industries and geographies.
By Scott Kaplan; US Naval Academy    Ryan Raimondi; US Marine Corps
Presented by: Scott Kaplan, US Naval Academy
Discussant:   Robert Press, Small Business Administration
Measuring the Local and Global Impact of Corporate Social Responsibility using Artificial Intelligence Abstract The accounting community spends more than $1 trillion a year summarizing complex financial data into easy-to-use measures (Soloveichik 2023; Ewans et al. 2021). These financial accounting measures are useful to researchers, policymakers, workers, businesses, and potential investors because they allow those stakeholders to make decisions based on the financial aspects of a project. If similar easy-to-use measures were available for corporate social responsibility (CSR), then those stakeholders could also make decisions based on the CSR aspects of a project. This paper outlines a method which uses artificial intelligence (AI) to extract and standardize dispersed CSR information at the firm-donation-year level. Thanks to the economies of scale associated with AI, it is feasible for a small group of researchers to create a very large database of CSR activity. This paper then presents results from a preliminary CSR database that was constructed using the methods outlined. The preliminary database has already identified more than $6 billion of employee volunteer time that is not tracked in conventional financial data. Future research will use the same methods to construct a complete CSR database that will be based on public filings, CSR reports, news media, and social platforms for U.S. firms from 2000 forward.
[slides]
By Dina El Mahdy; Morgan State University    Denis Gracanin; Virginia Tech    Rachel Soloveichik; Bureau of Economic Analysis    Majid behravan; Morgan State University    Sezer Dumen; Virginia Tech
Presented by: Dina El Mahdy, Morgan State University
Discussant:   Scott Kaplan, US Naval Academy
Session 3: How Households Cope: Consumption, Delinquency, and Income Support April 17, 2026 10:15 to 12:00
Session Chair: Scott Wentland, Bureau of Economic Analysis
Which Debts Get Paid Last? Abstract This paper exploits the scheduled expiration dates of property tax breaks to study how homeowners’ credit consumption and debt payment behavior respond to anticipated expense shocks. When tax breaks end, credit card balance, home equity credit line balance, new debt inquiries, and mortgage balance do not change, implying that credit consumption is smooth. However, homeowners become persistently delinquent on their student loan payments, which suggests that the marginal utility of consumption and from staying current on other debt (e.g., mortgage, auto, and credit card) is larger than the disutility from becoming delinquent on student loan payments. When budget constraints bind, student loans lie at the bottom of the debt repayment preference ranking. The results also show that expense shocks matter for student loan repayment, even for homeowners.
By Natee Amornsiripanitch; Federal Reserve Bank of Philadelphia    Siddhartha Biswas; Federal Reserve Bank of Philadelphia    Adam Scavette; Federal Reserve Bank of Philadelphia    David Wylie; Federal Reserve Bank of Philadelphia
Presented by: Adam Scavette, Federal Reserve Bank of Philadelphia
Discussant:   Ignacio Gonzalez, American University
Consumption Responses to a Major Minimum Wage Increase: Evidence from Spain Abstract This paper examines the effects of minimum wage increases on household consumption, focusing on Spain’s 2019 reform, which raised the wage floor by an unprecedented 22.3% in a low-inflation environment. Using high-frequency confidential transaction data from point-of-sale devices and credit card payments, we exploit geographic variation in exposure to the reform to identify its effects. We find that the reform increased household consumption by 4.5\%, with the largest gains concentrated in nonessential categories such as electronics, leisure, and spending at restaurants and hotels. We complement this analysis with household-level evidence from the Spanish Household Budget Survey. Consistent with the transaction data, affected households increased their consumption substantially—by 10.2%—and exhibited a marginal propensity to consume of 0.8, with spending gains again concentrated in nonessential categories. Firm-level sectoral evidence further suggests that the costs of the reform were primarily absorbed through the exit of less productive firms rather than through broad employment or price adjustments.
By Ignacio Gonzalez; American University    Hector Sala; Universitat Autònoma de Barcelona    Pedro Trivin; University of Milan
Presented by: Ignacio Gonzalez, American University
Discussant:   Adam Scavette, Federal Reserve Bank of Philadelphia
Towards County Personal Consumption Expenditures Abstract This project introduces a new framework for producing county-level Personal Consumption Expenditures (PCE) statistics. The paper outlines the data sources and methodological innovations used to construct these statistics, emphasizing the integration of diverse county-level datasets. Key inputs include administrative data on receipts and wages, county-level expenditures on medical care, housing cost data from the American Community Survey, and other relevant local economic indicators. A central methodological contribution is the incorporation of a residency adjustment based on inter-county spending flows. This adjustment leverages transaction-level data—primarily from card purchases—where both the purchaser’s and merchant’s locations are known, allowing for a more accurate attribution of consumption to the county of residence rather than the point of sale. In addition to describing the construction of the statistics, we present a set of exploratory "research" estimates to illustrate the utility and potential insights of the new series. These preliminary results highlight geographic variation in consumption patterns and demonstrate the feasibility of producing consistent, granular PCE estimates at the county level.
By Mahsa Agha Gholizadeh; Bureau of Economic Analysis
Presented by: Mahsa Agha Gholizadeh, Bureau of Economic Analysis
Discussant:   Scott Wentland, Bureau of Economic Analysis
Session 4: From Streets to Statistics: Measuring Crime, Displacement, and Violence April 17, 2026 10:15 to 12:00
Session Chair: David Johnson, International Association for Research in Income and Wealth
A Picture Worth a Thousand Words: The Effect of Murals on Crime Abstract Research Summary: Drawing on the tenets of situational crime prevention, place-based interventions including street lighting and transforming vacant lots into parks have proven not only effective in countering crime but also relatively inexpensive. The installation of large exterior murals fits well within this framework. In Philadelphia, the staggered implementation of the Mural Arts program’s murals lends itself to a difference-in-differences quasi-experimental design that exploits spatial and temporal variation to test the effect of street art installation on crime. Using this method, I uncover significant crime decreases on street segments that receive a mural in the two years after painting begins, with the largest significant effects found for disorder crimes (up to 50%) and lesser but still significant effects identified for violent and property crime. While these impacts are large in terms of percentage, the absolute fall in crime on treated blocks is small. I also detect relatively large crime displacement to adjoining blocks. The bulk of the crime decrease is found for blocks with repeated murals and when using increased time horizons, pointing to longer-term neighborhood transitions that improve with maintenance. Policy Implications: The findings of this study support funding for arts programs to address local crime complaints in addition to the current goals of public art investment. Still, displacement of crime to adjoining blocks and the modest absolute level of crime reduction emphasize the need for a more wholesale blight remediation strategy that extends beyond the block level. Future research should explore the aspects of larger areas that contribute to crime reduction and displacement, examine whether there is heterogeneity in effect based on the location and content of artworks, and utilize placement randomization to fully rule out other concurrent neighborhood interventions.
[slides]
By Maya Moritz; University of Pennsylvania
Presented by: Maya Moritz, University of Pennsylvania
Discussant:   Michael Mueller-Smith, University of Michigan
Job Displacement, Crime, and the Value of Unemployment Insurance in the US Abstract This extended abstract includes placeholders, but no results or discussion thereof, for results that may only be released pending Census disclosure review (anticipated date: Late February 2026). This paper measures the impact of job loss and unemployment insurance on criminal behavior and social welfare. Comparing a sample of workers impacted by mass layoff events across the United States to those who were not laid off, we show that those who experience a lay off earned [EARNING RESULT] and were [OFFENDING RESULT]. We study how changes to unemployment compensation design by changing the replacement rate, duration of benefits, improving likelihood of successfully filing by working at an employer with higher successful UI filing can [MITIGATE/EXACERBATE] these results. Using a simulated instrument approach we show that those laid off who would receive a one-standard-deviation (X percentage point) higher replacement rate [EARNING RESULT] and offend [OFFENDING RESULT]. Altering the duration by 10 weeks [RESULTS]. Moving to an employer with a one-standard deviation higher rate of UI filing post layoff [RESULTS]. Interpreting these results through a model we consider the relative tradeoffs across a social planner considering altering the replacement rate, the duration, or filing success in the presence of an externality and [CONCLUSION].
By Michael Mueller-Smith; University of Michigan
Presented by: Michael Mueller-Smith, University of Michigan
Discussant:   Maya Moritz, University of Pennsylvania
Estimating the Impact of the Age of Criminal Majority: Decomposing Multiple Treatments in a Regression Discontinuity Framework Abstract This paper studies the impact of adult prosecution on recidivism and employment trajectories for first-time felony youth criminal defendants. We use extensive linked Criminal Justice Administrative Record System (CJARS) and socio-economic data from Wayne County, Michigan (Detroit). Using the discrete age of majority rule, and a regression discontinuity design, we find that adult prosecution reduces future criminal charges over 5 years by 0.48 felony cases (20%) while also worsening labor market outcomes: 7.6 fewer employers (19%) and $613 less earnings (21%) per year. We develop a novel econometric framework that combines standard regression discontinuity methods with predictive machine learning models to identify mechanism-specific treatment effects that underpin the overall impact of adult prosecution. We leverage these estimates to consider four policy counterfactuals: (1) raising the age of majority, (2) increasing adult dismissals to match the juvenile disposition rates, (3) eliminating adult incarceration, and (4) expanding juvenile record sealing opportunities to those prosecuted in the adult system. All four scenarios generate positive returns for government budgets. After accounting for increases in recidivism generated by many of these policies and the corresponding victim costs borne by society, we find positive social returns for expanding the reach of juvenile record sealing and increasingly dismissing marginal adult charges, while raising the age of majority breaks even. Eliminating prison for first time adult felony defendants increases net social costs. Any opinions and conclusions expressed herein are those of the authors and do not reflect the views of the U.S. Census Bureau. The U.S. Census Bureau reviewed this data product for unauthorized disclosure of confidential information and approved the disclosure avoidance practices applied to this release (DMS number: 7512453, DRB approval number: #CBDRB-FY22-291).
By Ben Pyle; Boston University School of Law    Michael Mueller-Smith; University of Michigan    Caroline Walker; U.S. Census Bureau
Presented by: Ben Pyle, Boston University School of Law
Discussant:   Keith Finlay, U.S. Census Bureau
Measuring the Scope and Long-Term Impacts of Intimate Partner Violence from Linked Administrative and Census Records Abstract Intimate partner violence (IPV) is a common and costly public health and economic challenge, yet the impacts of IPV are often difficult to estimate due to limitations in measuring IPV experiences with pre- and post-exposure contexts. This paper provides a comprehensive analysis of the scope and long-term outcomes associated with IPV by integrating administrative and census data. The study relies on links between IPV-related criminal court records and 2010 Decennial Census household rosters to identify potential victims. By tracking these individuals longitudinally through tax records, public program participation data, and other data, the research measures the long-lasting impacts of domestic abuse. This data-driven approach provides population-based estimates on how IPV disrupts the social, economic, and criminal trajectories of adults in the United States.
By Shawn Ratcliff; U.S. Census Bureau    Keith Finlay; U.S. Census Bureau    Jordan Papp; University of Michigan    Cora Peterson; Centers for Disease Control and Prevention
Presented by: Keith Finlay, U.S. Census Bureau
Discussant:   Ben Pyle, Boston University School of Law
Session 5: The Cost of Care: Medicaid, Savings, and Preventable Spending April 17, 2026 10:15 to 12:00
Session Chair: Linden McBride, South Carolina Department of Employment and Workforce
Bootstrapped Standard Errors for Monthly and Quarterly Injury Rates from the Survey of Occupational Illnesses and Injuries (SOII) Abstract This project develops a procedure to construct work-related incidence rates along with accompanying standard errors at monthly and quarterly frequencies for private industry from 2009-2020 using data from the Survey of Occupational Illnesses and Injuries (SOII) and the Current Population Survey (CPS). The publication of higher frequency series would facilitate a better understanding of the within-year seasonal fluctuations of nonfatal workplace injuries, potentially allowing for the crafting of season-specific policy by BLS stakeholders. In addition to total monthly and quarterly incidence rates for cases involving days away from work (DAFW), the procedure can also produce rates by broad (2-digit) occupation and industry, various demographics (i.e. sex and race), nature of injury, as well as seasonally adjusted rates. While rates for both states and finer occupation & industry aggregations are also possible, data thinness raises concerns over the viability of creating counts and rates at the monthly and quarterly frequency for finer aggregations and for smaller states.
By Benjamin Raymond; Bureau of Labor Statistics
Presented by: Benjamin Raymond, Bureau of Labor Statistics
Discussant:   Farah Farahati, Senior Medical Economics Advisor
Medicaid Expansion and Financial Behavior Abstract We examine the effect of Medicaid expansion on financial behavior and financial literacy. Medicaid expansion was a key component of the 2010 Affordable Care Act’s health insurance reforms that significantly broadened coverage, particularly for low-income adults under the age of 65. Its rollout was staggered: while 6 states expanded before 2014 via waivers, 25 states and the District of Columbia initially opted in during 2014, with a total of 41 states having expanded by mid-2025 (Kaiser Family Foundation [2025]). We use data from the National Financial Capability Study (NFCS), a large-scale survey conducted every three years from 2009 to 2024. The survey collects rich information on financial capability indicators, healthcare access, financial resources, financial experiences and attitudes, as well as socioeconomic and demographic characteristics. Since its inception in 2009, the NFCS has surveyed a nationally representative sample of over 25,000 American adults in each wave, including 500 respondents from each of the 50 states and the District of Columbia. We supplement these data by appending state-level characteristics and policies from UKCPR. Given the staggered rollout of Medicaid expansion, we estimate dynamic treatment effects using the approach developed by Callaway and Sant’Anna [2021], which allows for treatment effect heterogeneity across groups and over time. While a large literature examines Medicaid’s impact on health access and health outcomes (see Sun [2025] and citations within), its direct and indirect effects on broader financial health metrics, such as savings behavior, medical debt, and reliance on fringe financial services, remain underexplored. Notable exceptions include the work of Das [2025], Baten et al. [2024] and Fitzpatrick and Fitzpatrick [2021]). The existing literature has laid valuable groundwork; however, many contributions rely on fewer years of data, approaches suited to other aims, or a limited outcome scope. We build on this work with more recent data, methods targeted to our question, and an expanded set of outcomes. Our research explores key indicators of individual financial well-being, including health insurance status, the presence of an emergency fund, medical debt, and reliance on fringe financial services such as rent-to-own furniture, pawn shops, or payday loans. The analysis focuses on low-educated adults under age 65, a demographic particularly sensitive to Medicaid expansion’s effects. We find as others do that Medicaid expansion increases the probability of having health insurance and decreases the probability of having medical debt as seen in Table 1. Next, we will consider access to financial products such as bank accounts and credit cards. We also examine the use of fringe lending products such as payday loans, pawn shops, and rent-to-own . Finally, we inquire as to whether Medicaid expansion increases objective and subjective financial literacy measures as developed by Lusardi and Mitchell [2023].
By Susan Averett; Lafayette College    Julie Smith; Lafayette College
Presented by: Julie Smith, Lafayette College
Discussant:   Adam Bloomfield, Federal Deposit Insurance Corporation
Playing Catch-up with Health Savings Abstract We use comprehensive tax data to study how saving behavior responds to the Health Savings Account (HSA) “catch-up” contribution provision, which raises HSA contribution limits for individuals aged 55 and older. Using a regression discontinuity design, we find a sharp increase in contributions among those previously near the limit and smaller increases among unconstrained savers. Induced contributions are not immediately withdrawn and do not appear to crowd out retirement savings. Responses are strongest among payroll contributors and long-term savers. However, married couples do not appear to coordinate their HSA behavior to take advantage of the complex spousal rules governing catch-up contributions. Our findings highlight how tax incentives shape HSA saving and suggest that tax-advantaged account design meaningfully affects household financial behavior.
By Jacob Berman; U.S. Department of the Treasury    Adam Bloomfield; Federal Deposit Insurance Corporation    Sita Slavov; George Mason University
Presented by: Adam Bloomfield, Federal Deposit Insurance Corporation
Discussant:   Julie Smith, Lafayette College
Preventable Public Spending on Late-Stage Kidney Disease Abstract Chronic kidney disease (CKD) affects more than 37 million Americans and is a major driver of public spending in Medicaid and Medicare. Early detection costs roughly $40 per person, yet treatment commonly begins only after progression to end-stage renal disease, where care exceeds $90,000 annually per patient. This paper tests whether public insurance design generates preventable high-cost treatment. We study how Medicaid financing structure and diagnostic access influence treatment intensity and government expenditures. Using national datasets (NHANES, USRDS, CMS Part D, SAHIE, ACS), we construct county-level measures of disease detection, prescribing intensity, and socioeconomic barriers across more than 2,500 U.S. counties. Variation in Medicaid expansion status and local diagnostic infrastructure provides policy-relevant differences in access to early detection. We then map observed treatment patterns into fiscal outcomes using a public-payer budget model over a ten-year horizon. Counties with weaker diagnostic access exhibit systematically higher late-stage treatment intensity. Expansion states show higher linkage to care and lower expected long-run treatment costs relative to non-expansion states. Screening adults aged 55+ is cost-saving, reducing spending by $556,600 per 1,000 individuals over ten years. A national implementation scenario (65 million adults aged 55+, 60% uptake) yields approximately $16 billion in public savings. These findings imply that a substantial share of prevention failure arises from program design and financing rather than from clinical uncertainty alone. Aligning coverage and payment with early detection and improving diagnostic access could reduce long-run federal and state expenditures. Using CKD as an example provides evidence that insurance and reimbursement design influence treatment timing and long-run public spending. The policy implication is that prevention failure arises from program design Aligning coverage with early detection and improving diagnostic access could reduce long-run federal and state expenditures. CKD provides evidence that insurance design affects treatment timing and long-run public spending.
By Farah Farahati; Senior Medical Economics Advisor    Evelyn Rizzo Evelyn Rizzo; Mobilityheor.com
Presented by: Farah Farahati, Senior Medical Economics Advisor
Discussant:   Benjamin Raymond, Bureau of Labor Statistics
Session 6: Who Works, Who Earns, Who Searches: New Measures of Labor Supply and Compensation April 17, 2026 10:15 to 12:00
Session Chair: Erin George, Women's Bureau
Analyzing the Gender Wage Gap in the United States: 2022-2024 Abstract Discrimination in the labor market is a key focus of attention among policy researchers. Specifically, there is a long history of research on the gender wage gap – that is, understanding the relationship between gender and earnings and why women’s earnings tend to be lower than those of men at the median (see, most notably, Blau and Kahn, 2017). In this context, researchers typically look to explain to what extent the wage gap is due to discrimination versus other observable factors. Each year, the U.S. Census Bureau’s Income in the United States report highlights the female-to-male earnings ratio among full-time, year-round workers (Kollar and Scherer, 2025). Over the past 60 years, this ratio has broadly increased, from women earning roughly 60 cents for each dollar earned by men in 1960 to roughly 80 cents for each dollar earned by men in recent years. However, in the past two report releases (i.e., from 2022 to 2023 and also from 2023 to 2024), the Census Bureau has reported declines in this ratio – most recently, from 82.7 in 2023 to 80.9 in 2024. While the Census Bureau has offered speculative explanations for this decline in public settings, little work has been undertaken to more fully understand the reasons underpinning this decline. Insofar as policymakers view pay equality as a desirable outcome, it is essential to understand the drivers of the earnings gap in order to pull the appropriate policy lever(s) in response. The various analyses of the gender wage gap produced over decades have been conducted using a variety of datasets, but largely use data from prior to the COVID-19 pandemic. In this research, I update these findings using data from after the COVID-19 pandemic, which precipitated a variety of shocks in the labor market that could contribute to the decline in the female-male earnings ratio noted above. Specifically, I hope to identify the factors that lead to the gender wage gap and how the relative importance of these characteristics may have changed in recent years by answering the following questions: 1)What was the relationship between sex and earnings in 2022, 2023, and 2024? 2)To what extent is that relationship attributable to discrimination versus other observable factors? 3)Which elements of the sex-earnings relationship most explain the decline in the female-male earnings ratio observed over the years in question? I rely on the 2023 to 2025 Current Population Survey Annual Social and Economic Supplements (CPS ASEC – reference years 2022 to 2024), using OLS and quantile regression and Oaxaca-Blinder decomposition to offer a better understanding of the dynamics underpinning the sex-earnings relationship during the post-COVID period.
By Zachary Scherer
Presented by: Zachary Scherer,
Discussant:   Bob LaJeunesse, Bureau of Labor Statistics
Marriage Market Competition and Labor Supply Abstract This paper examines how marriage market prospects influence decisions related to labor supply. Using U.S. microdata, we construct a novel and granular measure of marriage-market tightness. To address endogeneity from demographic mobility or labor-market shocks, we implement a Bartik-style instrument, approximating local population changes with national demographic shifts. We show that competition has sizable effects across gender and education groups. For single men, greater competition decreases non-participation and hours for those without a bachelor's degree, and increases income and hours for those with a bachelor's degree. For single women without a bachelor's degree, greater competition increases non-participation.
[slides]
By Joseph Pickens; United States Naval Academy    Demid Getik; Durham University
Presented by: Joseph Pickens, United States Naval Academy
Discussant:   Zachary Scherer,
Measuring Active Job Search: Experimental Evidence from Screener and Probe Techniques in a Large Online Survey Abstract Employment status classification is core to many major U.S. household surveys, including the Current Population Survey (CPS) – the official source of U.S. unemployment statistics – as well as the American Community Survey (ACS) and the Survey of Income and Program Participation (SIPP). If not currently employed, one must typically be “actively” searching for work to count as unemployed and thus part of the labor force. However, the distinction between active and passive job search may be less apparent to respondents than the distinction between working and not working, and measurement of active search varies considerably across surveys. In this study, we randomly vary the questions used to measure job search in a large online survey, and find differences in active search rates (and thereby in unemployment rates) by approach. The CPS first asks respondents if they did “anything” to look for work (yes/no “screener”). If yes, a follow up question probes their search activities to determine if they searched actively or only passively (activity “probe”). In contrast, both the ACS and the SIPP use a consolidated, screener-only approach that directly asks respondents whether they “actively” looked for work. Historically, these surveys have produced differing estimates of employment status, and previous work finds that question design affects the size of these differences. We experimentally evaluate the CPS and ACS/SIPP approaches against two alternative methods: an ACS/SIPP screener that includes a definition of “actively looking for work”, and a straight-to-probe approach with no screener. We randomize respondents across four distinct search question sequences. Critically, each respondent is probed on any search activities regardless of their screener response, with a mark-all-that-apply list of active and passive search methods. This lets us measure the frequency of inconsistent responses between screeners and probes, i.e.: respondents who report an active search activity but answered “no” to a screener, or who say “yes” to [actively] searching but do not report any [active] search methods. While the vast majority of respondents provide consistent responses, we find that our CPS-style question sequence does produce “false negatives” (respondents who actively searched responding “no” to the screener). The ACS/SIPP style screener, on the other hand, results in both false negatives and false positives, which may help explain why ACS unemployment estimates tend to be larger than CPS rates. Simultaneously, we find that more non-employed respondents would report actively searching on an ACS/SIPP style screener if also given examples of active versus passive activities, suggesting a degree of respondent confusion regarding what constitutes active job search. Of the alternatives we test, an ACS/SIPP style screener with respondent instructions yields the most comparable responses to a CPS-style question sequence.
By Alex Opanasets; US Census Bureau    Matthew Pesner; US Census Bureau
Presented by: Alex Opanasets, US Census Bureau
Discussant:   Joseph Pickens, United States Naval Academy
Measuring the Nonemployee Workforce Abstract Labor market analysts widely observe a shift in the nature of work in the U.S. economy. In recent decades, increasingly more people have earned money through some type of self-employment arrangement or non-wage work. Consequently, there is a strong need for data on the movement of workers into and out of self-employment and business ownership. Yet few government statistics are published about this segment of workers and none document the dynamics of self-employment work patterns. Quantifying transitions is essential to understanding the 21st century labor market and the long-term career trajectories of its participants. To fill this gap, the Census Bureau is developing data infrastructure based primarily on IRS business tax returns to expand public-facing measures of self-employment and business ownership, which fall outside of traditional wage and salary employment. These new statistics will be reported separately by demographics and other characteristics of the business owners or the nonemployee workforce. In this paper we summarize progress made in developing these new statistics and then discuss ongoing efforts to incorporate a new data source– IRS 1099-NEC, Nonemployee Compensation filings. We examine the challenges of integrating these data and share preliminary statistics on what we learn about non-traditional workers from this new source. The Census Bureau currently publishes Nonemployer Statistics by Demographics (NES-D) Data which provide tabulated information on the population of businesses without employees. New research (Business Owners and the Self-Employed: Thirty-Three Million (and Counting!) | NBER) has produced new longitudinal linkages for nonemployer businesses, and a database of jobs that tracks individuals as they move in and out of self-employment / business ownership work arrangements. More recently, IRS shared information with the Census Bureau derived from filings of IRS Form 1099-NEC, Nonemployee Compensation (hereafter, “1099-NEC”). These forms document payments made by businesses to people who perform services but who are not employees. Traditionally, these people have been labeled as independent contractors. The 1099-NEC serves the same purpose for the contractor as the W-2 does for the employee – it reports payments to the IRS which the contractor is expected to report on their personal income tax forms. These 1099-NEC records will improve Census measurement in three ways. First, they provide additional information for a subset of known self-employed workers and the jobs they hold. Second, they expand the set of known self-employment workers and jobs, and third, they provide nonemployee workforce and payment measures for businesses. We show summary statistics, broken out by demographic traits (sex, race, age, foreign-born status and veteran status), detailing how many people receive 1099-NECs and how many match to other self-employment tax filings such as Form 1040 Schedule C. We summarize how many jobs are represented by these 1099-NEC filings, dividing this population into categories based on whether a person has multiple 1099-NEC filings and/or multiple Schedule C filings. We count how many nonemployer businesses receive 1099-NECs and group them according to how many they receive. Finally, we count issuing businesses and classify them by whether or not they have wage-and-salary employees (versus only 1099-NEC workers).
By Christopher Goetz; U.S. Census Bureau    Henry Hyatt; U.S. Census Bureau    Nicholas LaBerge    Adela Luque; U.S. Census Bureau    Vitaliy Novik; Census Bureau    Kristin Sandusky; U.S.Census Bureau    Martha Stinson; U.S. Census Bureau
Presented by: Kristin Sandusky, U.S.Census Bureau
Discussant:   Alex Opanasets, US Census Bureau
Session 7: The AI Disruption: Measuring Exposure, Work, and Market Response April 17, 2026 13:15 to 15:00
Session Chair: Sabrina Pabilonia, Bureau of Labor Statistics
AI-Enhanced Mapping of End-Use Concordance for Trade Analysis Abstract This technical note introduces a systematic framework for classifying traded goods by their economic end use—consumer, intermediate, and capital goods. The approach adopts an economic-function perspective consistent with the United National Broad Economic Categories, emphasizing how goods are primarily utilized once traded. Classification integrates Harmonized System (HS) product descriptions with sectoral evidence from FAO, USDA, OECD, UNIDO, and industry market reports, applying consistent rules of thumb to address mixed-use products. For goods serving multiple functions, proportional allocations are provided to reflect dominant uses, while ancillary uses below a 10 percent threshold are excluded. The resulting End Use Concordance Database enhances the consistency of trade analysis, global value-chain mapping, and input–output modeling, and is designed to be transparent, empirically grounded, and adaptable across countries and datasets. Large language models (LLMs) were trained and employed in the classification process; however, all HS code assignments were subject to manual verification.
By Erika Bethmann; USITC
Presented by: Erika Bethmann, USITC
Discussant:   Susan Fleck, Bureau of Labor Statistics
Generative AI and labor market dynamics: An evaluation with the Quarterly Workforce Indicators Abstract Labor market exposure to generative artificial intelligence is typically assessed at the worker occupation level. This is because generative AI has the ability to perform many tasks that were previously performed using labor inputs. For the same reason, most recent assessments of AI's impacts have focused on occupation-level measures. Recent evidence using private administrative payroll data suggests that labor market impacts of generative AI may be disproportionately observed among early-career workers in AI-exposed occupations, and that this may be the result of reduced hiring of new workers (e.g. Brynjolfsson et al. 2025). However, publicly available sources of information to study this topic are limited. Survey-based datasets typically lack the sample size or longitudinal detail needed to assess detailed labor market dynamics, while administrative sources of data lack detailed information on occupation, are inaccessible to most researchers, or both. This paper assesses the impacts of generative AI availability on hiring, separation behavior, and other measures of labor market dynamics. It also provides a demonstration of the extent to which publicly available data can be pressed to continually assess the labor market impacts of new technologies. To do this, it leverages newly released detailed industry tabulations provided by the Census Bureau's Quarterly Workforce Indicators (QWI). These new QWI measures provide detailed information about labor market dynamics at the 5- and 6-digit NAICS industry level, including hiring and separation behavior among young workers. By crosswalking occupations to industry using public-use ACS microdata, we are able to map external measures of potential AI exposure to detailed industry, state, and age group level trends. These measures are based on a near-universe of private sector flows, providing a comprehensive measure of how AI exposure has predicted recent changes in hiring, separations, employment, and earnings. Although the primary aim of this project is to assess the impacts of generative AI, a major goal of this project is to provide a framework for how public data users can expand the scope of customized metrics they track, without needing to resort to restricted-use data. This project relies solely on published data tabulations and public-use microdata that are accessible to anyone, as well as extant measures of occupational AI exposure. A secondary goal of this project is to demonstrate and assess the feasibility of occupation-to-industry crosswalking at the detailed level. Preliminary results using ACS suggest that the availability of more detailed industry information in QWI may improve the ability to assess impacts. Although the observed variation in AI exposure is attenuated by any level of industry crosswalking, it is substantially less attenuated when this crosswalking can be performed at the most detailed industry level.
By Lee Tucker; U.S. Census Bureau
Presented by: Lee Tucker, U.S. Census Bureau
Discussant:   Michael Dalton, BLS
AI Disruption Meets Regulatory Capture: Distinct Time Trends in the KSA Similarity Trap Abstract Purpose: As artificial intelligence reshapes labor markets, this study investigates how knowledge, skills, and abilities similarity between occupations creates a "Similarity Trap" that hinders upskilling, with effects moderated by state occupational licensing stringency through public choice capture mechanisms. It challenges assumptions that higher skill alignment reliably supports upward mobility while testing licensing's amplifying role. Design/methodology/approach: Current Population Survey panels (2010-2025) link occupational transitions to knowledge, skill, and ability similarity. Network analysis visualizes occupational clusters and mobility flows. Occupational regulation index captures state licensing stringency. Event study designs estimate domain-specific, time-varying effects on upward mobility probabilities, interacting licensing regimes with similarity measures and artificial intelligence exposure quartiles. Findings: Higher overall knowledge, skills, and abilities similarity persistently reduces upward mobility likelihood, confirming the Similarity Trap channels workers laterally or down skilled. Critically, domains diverge over time: knowledge similarity effects prove volatile across technological cycles; skill similarity exhibits strengthening negative divergence as artificial intelligence devalues routine alignment; ability similarity converges with weakening negative impact, signaling foundational capacities' growing universality. State licensing stringency amplifies these traps, particularly skill lock-in. Originality/value: Disaggregates similarity trap into time-varying knowledge, skill, and ability components while quantifying public choice licensing capture. Documents diverging human capital returns under artificial intelligence disruption—skill penalties worsen, ability premiums emerge—challenging static mismatch models and revealing regulatory amplifiers. Practical implications: Prioritize ability-focused upskilling and skill rebundling over knowledge deepening. Target licensing deregulation in high-skill-similarity occupations where artificial intelligence and regulation compound mobility barriers. Modular learning pathways support cycle-sensitive workforce transitions.
By Ting Zhang; University of Baltimore    Laurie Schintler; George Mason University
Presented by: Ting Zhang, University of Baltimore
Discussant:   Joseph Pickens, United States Naval Academy
AI Exposure and the Future of Work Abstract How will AI reshape the nature of work in the U.S. economy and in particular, which occupations will be most impacted by this transformative technology? Understanding the potential impact of AI on occupations will be critical to anticipating how it will influence the future employer demands for skills and workforce development and training needs. This article seeks to identify which occupations are currently most affected by AI, how these occupations vary by educational attainment, and what models of occupational projections suggest about their future employment trends. We employ three approaches to assessing which occupations are most impacted by AI technologies: First, we extend an approach originally adopted by the Pew Research Center, using the prevalence of 16 specific work activities out of 41 from the O*NET Generalized Work Activities (GWA) Survey to measure occupational exposure to AI. Like Pew, our method uses O*NET survey data of occupational incumbents on the relative importance of 16 high-exposure tasks to divide occupations into 4 quartiles based on little, some, moderate, and high exposure to AI. We extend this analysis to include self-reported data on both the relative importance and intensity of each task to create a modified index of AI-exposure. We then examine how the resulting groupings correspond to long-term projected occupational changes from the Bureau of Labor Statistics (BLS), in particular examining how groups of occupations with varying degrees of AI exposure differ in terms of net employment change by educational attainment. Second, we complement this occupation-level analysis with a detailed, task-based assessment using ONET Detailed Work Activities (DWAs). DWAs provide fine-grained descriptions of the specific activities that underlie generalized work activities and occupations. For each DWA, we evaluate exposure along two distinct but related dimensions. We assess exposure to LLM-enabled technologies, classifying whether currently available large language models -- used directly or embedded within software applications -- can perform core cognitive components of the activity, distinguishing between no meaningful exposure, partial exposure, and high exposure. We then assess exposure to broader digital automation, encompassing widely deployed enterprise software, analytics systems, workflow automation, and algorithmic decision-support tools that may materially restructure how work is performed, even when LLMs are not central. These two assessments are conducted independently to distinguish tasks primarily affected by generative AI from those shaped by longer-standing forms of digital automation. We aggregate DWA-level classifications to the occupational level using ONET mappings, allowing us to characterize how AI and digital technologies affect different components of work within occupations and to clarify whether observed exposure reflects substitution, augmentation, or limited change in task execution. Finally, we conduct a meta-analysis of studies that have identified which occupations are most likely to be impacted by AI. We review the major approaches in the literature that are used to make those determinations and compare our results above to the ones from this body of literature.
By Erik Vasilauskas; W.E. Upjohn Institute for Employment Research    Michael Horrigan; W.E. Upjohn Institute for Employment Research
Presented by: Erik Vasilauskas, W.E. Upjohn Institute for Employment Research
Discussant:   Leland Crane, Federal Reserve Board
Session 8: Credit, Bubbles, and Climate: Financial Markets Under Pressure April 17, 2026 13:15 to 15:00
Session Chair: Mahsa Agha Gholizadeh, Bureau of Economic Analysis
Designing a Better Model to Predict Loan Loss Given Delinquency Abstract Under the USDA FSA guaranteed farm loan program, the lender is guaranteed up to 95% of loan principal in the case of borrower default. When efforts to assist the borrower in resuming timely payments fail, a loan enters default. If the proceeds of collateral sold are not sufficient to cover principal due on the loan a loss claim occurs (Settlage et al., 2000). While the economic impact of loss claims has been relatively small in recent years, ranging from between $57 million in 2004 to $25 million in 2022, or between 0.2 to 0.6 percent of total loan volume, this does impose an economic burden borne by the taxpayer. Consequences for the borrower include loss of collateral and future farm credit, difficulty obtaining nonfarm credit, and the potential for judgments issued against future earnings. Three models are used to identify key factors associated with the likelihood of a borrower incurring a loss claim on a guaranteed loan, given initial loan delinquency. The first two are traditional models, a logistic regression incorporating borrower level fixed effects and cox proportional hazard model. A random forest machine learning method is also used. The model results are compared and evaluated based upon consistency with accepted academic theory, their ability to reveal new information, and their out-of-sample predictive accuracy. Similar models and evaluation techniques were used to study farm loan repayment by Ifft et al. (2018), Dixon et al. (2008), Chen et al. (2021) and Long et al.(2016). USDA FSA guaranteed loan program administrative data is used. The data covers a subset of borrowers having an outstanding FSA guaranteed loan in 2007. The final dataset provides a full history of all loans received by these borrowers between 1994-2023 including loan terms, borrower information, delinquency history, loss claim history, and ultimate loan outcome. Initial model results indicate that factors associated with a higher probability of a loss claim include a higher interest rate, larger unpaid principal amount, or loans for the purpose of covering operating or living expenses compared to livestock, machinery or real estate purchases. Factors associated with lower loss claim probability or full repayment include Farm Credit Association versus a commercial bank or other lender, a higher percentage of loan guaranteed by FSA, lines of credit, and being male or married. These findings will be of interest to policy makers, loan officers, and financial economists. They provide an indication of which loans are most likely to go to default given initial delinquency and hence enable better monitoring of borrowers and facilitate the construction of more effective and rapid intervention strategies. These results are also of interest to economic researchers, as they provide greater insights into factors associated with loan default and ways in which proportional hazard and machine learning models can be employed in future studies.
[slides]
By Sarah Atkinson; USDA
Presented by: Sarah Atkinson, USDA
Discussant:   Trevor Bakker, U.S. Census Bureau
Impacts of Carbon Disclosures for Investors on French Manufacturing Firms' Trade, Financial and Environmental Performance Abstract This paper studies how investor-level climate disclosure regulation transmits to real firm outcomes through financing channels. Using administrative tax filings, customs records, and OECD TeCO2 data, we examine France’s Article 173, a pioneering disclosure rule novel for its investor focus and its coverage of emissions embodied in imports. Exploiting cross-firm variation in reliance on bond financing, we implement a difference-in-differences design and find that a 10-percentage-point increase in exposure reduces imported carbon by 5.84%, but also leads to contractions in firm size and import activity. Effects are concentrated among financially constrained firms, consistent with heightened investor scrutiny tightening access to external finance. We further document that broad declines in carbon intensities among historically high-emission origins limit firms’ ability to green supply chains. The results reveal trade-offs between environmental objectives and financial frictions, highlighting the need for complementary policies when relying on investor-focused disclosure rules to support the low-carbon transition.
[slides]
By Thomas Rowley; Università Bocconi
Presented by: Thomas Rowley, Università Bocconi
Discussant:   Qingyu Liu, The George Washington University
A Search Model of Financial Bubbles Abstract Financial bubbles pose significant risks to macroeconomic stability, characterized by explosive asset prices and abnormal trading volumes. This paper endogenously generates both features within a rational framework by developing a search-and-matching model. We provide a micro-foundation for the "ride-the-tide" behavior of speculators, driven by the interaction of heterogeneous agents and the rational expectation of a future collapse risk. A key contribution of this paper is its empirical tractability for policy analysis. Unlike many theoretical bubble models, our framework allows for the structural estimation of unobservable fundamental values and the identification of a bubble's potential turning point using standard market data. We demonstrate the model's application using data from the cryptocurrency market. This framework offers a novel tool for policymakers to monitor asset market exuberance and assess financial stability risks in real-time.
By Qingyu Liu; The George Washington University
Presented by: Qingyu Liu, The George Washington University
Discussant:   Thomas Rowley, Università Bocconi
Economics and Norms of Credit Card Debt Abstract Using linked credit bureau and Census Bureau data, we show that revolving credit card debt is highly persistent: around 75% of credit cardholders with and 70% without it will continue revolving or not, respectively, after eight years. Furthermore, revolving is largely unresponsive to income growth over this time, and initial revolving negatively predicts future income growth in the cross-section. Standard rational or behavioral models of consumption smoothing cannot generate these facts in the absence of heterogeneous consumers. We propose decision rule heterogeneity such that consumers’ propensity to borrow and income growth are correlated negatively, strongly enough across types to overcome consumption smoothing within each type. Our model is sufficient to fit all the motivating facts. We then study the empirical sources of this heterogeneity and show that much of it is shaped by parental behavior and childhood environment. Using variation in the age of children at parental separation, we document a causal effect of exposure to parental revolving on children’s revolving in adulthood. Specific parental repayment behaviors, such as overpaying installment debt while revolving credit card debt, predict the same behaviors in children. Among U.S.-born children of foreign-born migrants, revolving patterns strongly correlate with cultural attitudes in parents’ countries of origin, even after controlling for income. Among foreign-born migrant children from countries with high rates of adoption to the U.S., revolving in adulthood is strongly predicted by whether parents revolve, not by whether parents are U.S.-born (versus foreign-born) or its interaction with whether parents revolve, limiting the scope for genetic influences on revolving. Our findings suggest that children learn debt repayment norms, highlighting how socially transmitted behaviors shape borrowing outcomes.
By Trevor Bakker; U.S. Census Bureau    Dominic Russel; Harvard    Claire Shi; Harvard
Presented by: Trevor Bakker, U.S. Census Bureau
Discussant:   Sarah Atkinson, USDA
Session 9: Measuring What We Mean: Definitions, Data, and Economic Well-Being April 17, 2026 13:15 to 15:00
Session Chair: John Creamer, US Census Bureau
National Experimental Well-Being Statistics (NEWS) Combining Survey and Administrative Data to Improve Income and Poverty Statistics Abstract The National Experimental Well-being Statistics (NEWS) project aims to produce the best possible estimates of income and poverty given all available survey, decennial census, administrative, and third-party data. We estimate improved income and poverty statistics by addressing bias from unit non-response, missing information and measurement error. In addition to pre-tax money income, which release 1 focused on, this release greatly expands the income definitions our data cover by additionally creating measures of 1) disposable income, 2) a resource measure inclusive of non-health means-tested in-kind benefits, and 3) the income concept used for estimating the Supplemental Poverty Measure (SPM). In addition, this release implements several improvements, most importantly: 1) we estimated federal and state taxes and credits, 2) we integrated additional administrative data on means-tested program benefits, 3) we updated our model for combining survey and administrative earnings to better estimate the unobserved true earnings distribution given the earnings reported across data sources and the other observable characteristics of individuals. We estimate income and poverty from 2016 to 2021 and find that standard survey-based income and poverty estimates are biased, and these biases vary by subgroup and over time. This can cause estimates to miss important facts about how economic well-being is changing over time and why. Existing estimates underestimate income across the distribution, by 7 to 18 percent at the 25th percentile, depending on the year and measure of income used. Likewise, poverty estimates are overstated by 1 to 3.5 percentage points, again depending on the year and measure. To give another example, the NEWS and survey estimate of child poverty differ over time due to changes in survey nonresponse bias and sources of income, particularly with the large expansion of unemployment insurance during the pandemic. Child poverty was 1.5pp lower in NEWS than the survey in 2018, they were not statistically different in 2019 or 2021, and the NEWS estimate was 3.8pp lower than the survey in 2020 (according to the Supplemental Poverty Measure).
By Adam Bee; US Census Bureau    John Creamer; US Census Bureau    Joshua Mitchell; U.S. Census Bureau    Nikolas Mittag; CERGE-EI    Elizabeth Pelletier; University of Washington    Jonathan Rothbaum; The George Washington University    Carl Sanders; Indiana University Bloomington    Lawrence Schmidt; Massachusetts Institute of Technology    Matthew Unrath; University of Southern California
Presented by: Jonathan Rothbaum, The George Washington University
Discussant:   Jeffrey Thompson, Federal Reserve Bank of Boston
The Devil's in the Definitions: How Income and Poverty Measurement Decisions Shape Our Understanding of Economic Well-Being, 2004-2024 Abstract Accurate measurement of economic well-being and inequality in the United States is of interest to both researchers and policymakers. The U.S. Census Bureau contributes to this discourse by producing annual estimates of median household income and poverty. This paper adds to the existing work at the Census Bureau by exploring how using different income definitions, poverty thresholds, and inflation adjustments affect measurement of economic wellbeing over the past two decades. We produce a historical series (2004-2024) of two new measures of income: market and disposable income. We present these new measures alongside the two measures currently used at Census, money income and post-tax income, and discuss differences in levels and changes in the four measures over time. In addition, the paper discusses poverty trends using the newly presented measures. We also explore the implications of the choice of inflator, supporting existing evidence that shows that chained price inflators show larger increases in economic well-being. Taken together, these new estimates speak to how economic well-being has changed in the U.S. over the past two decades and highlights the importance of measurement decisions in understanding these trends. This work is consistent with recommendations from a recent National Academies report on improving estimates of income, consumption, and wealth in the United States (National Academies of Science, Engineering, and Medicine 2024) as it adds supplementary estimates of post-tax and transfer income to currently produced statistics.
By Elizabeth Pelletier; U.S. Census Bureau    John Creamer; US Census Bureau
Presented by: Elizabeth Pelletier, U.S. Census Bureau
Discussant:   Sanders Korenman, Baruch College, CUNY
The Seasons They Are A-Changin’: A Century of Definitions and a Way Forward Abstract What is seasonality? To date, there is not yet a consensus definition, as over a dozen unique definitions of seasonality can be found in the economics and statistics literature. This lack of consistency complicates the identification of, and adjustment for, seasonal effects in economic time series. In this paper, we propose a new definition of seasonality as well as how to detect and adjust for it, ultimately aiming to build consensus around a more unified approach. We first review the literature and identify common themes across definitions with an eye towards formulating a foundational definition – expressed in both plain language and mathematical terms – that can be used by academics, data providers, policymakers, and practitioners alike. We then propose to classify a series as seasonal if, among all its possible repeating cycles, the ones associated with the fundamental periodicity are demonstrably larger than the others. In other words, we define seasonality as a measure of relative peak dominance in the spectral density function, following Granger (1978). Next, we develop a new methodology for detecting seasonality and performing seasonal adjustment that follows directly from our definition and leverages properties of the sample periodogram, the empirical counterpart to the spectral density function. We illustrate the performance of our test statistic through simulation studies and find that it is well-sized and has good power. Lastly, we introduce Stochastic Spectral Imputation (SSI), an adjustment procedure that outperforms standard methods currently used by national statistical offices when applied to prominent macroeconomic time series.
By Carter Bryson; Bureau of Economic Analysis    Gary Cornwall; Bureau of Economic Analysis
Presented by: Gary Cornwall, Bureau of Economic Analysis
Discussant:   Jonathan Rothbaum, The George Washington University
Rural Economic Resilience: Measuring Resilience and Trying to Understand Its Influences Abstract There is an extensive body of research exploring the economic consequences of the declines in manufacturing employment experienced by “legacy cities” and “left behind places,” and a related literature that attempts to measure the economic “resilience” of communities and the factors associated with resilience. This paper contributes to the existing literature by 1) developing a measure of resilience that separately reflects the influence of shifts in demand for non-traded and traded goods, and 2) developing a database of over 2,000 unique, time-varying local-area traits that can be used to evaluate the traits associated with economic resilience. Using this new measure and the local-area data, we seek to identify communities most at risk of economic decline and policy variables that are most successful at protecting regions from economic shocks. We are particularly interested in the resilience of rural communities. Are there traits of rural communities that can help us identify places that are more at-risk of extensive harm from trade shocks or recessions? Are there local characteristics that make the economies in some places more resilient to trade shocks and recessions?
By Jeffrey Thompson; Federal Reserve Bank of Boston    Melissa Gentry; USF
Presented by: Jeffrey Thompson, Federal Reserve Bank of Boston
Discussant:   Gary Cornwall, Bureau of Economic Analysis
Session 10: Measuring What Matters: Data, Methods, and the Productivity Debate April 17, 2026 13:15 to 15:00
Session Chair: Breno Braga, Urban Institute
Manufacturing Dispersion: How Data Cleaning Choices Affect Measured Misallocation and Productivity Growth in the Annual Survey of Manufactures Abstract Measurement of dispersion of productivity levels and productivity growth rates across businesses is a key input for answering a variety of important economic questions, such as understanding the allocation of economic inputs across businesses and over time. While item nonresponse is a readily quantifiable issue, we show there is also misreporting by respondents in the Annual Survey of Manufactures (ASM). Aware of these measurement issues, the Census Bureau edits and imputes survey responses before tabulation and dissemination. However, edit and imputation methods that are suitable for publishing aggregate totals may not be suitable for estimating other measures from the microdata. We show that the methods used dramatically affect estimates of productivity dispersion, allocative efficiency, and aggregate productivity growth. Using a Bayesian approach for editing and imputation, we model the joint distributions of all variables needed to estimate these measures, and we quantify the degree of uncertainty in the estimates due to imputations for faulty or missing data.
[slides]
By Hang Kim; University of Cincinnati    Martin Rotemberg; Economics    Thomas White; Us Census Bureau
Presented by: Thomas White, Us Census Bureau
Discussant:   Peter Meyer, U.S. Bureau of Labor Statistics
Decoding the Productivity Puzzle: A New Perspective on the Relationship between Remote Work and Productivity Abstract The relationship between remote work and productivity remains contentious, with empirical studies yielding conflicting results. This paper resolves this "productivity puzzle" by shifting analytical focus from remote work adoption—which is subject to selection bias—to the inherent ability to work remotely, measured through task-based occupational characteristics. This distinction is critical because less productive workers systematically self-select into remote positions, obscuring the true causal relationship between remote capability and productive performance. Using a comprehensive panel dataset spanning 156 four-digit NAICS industries from 2003 to 2022, I construct industry-level measures of remote work ability by classifying occupations based on the importance of email, phone, and memo usage in their task requirements, then aggregating these classifications through employment shares. This approach leverages the long historical availability of task-based measures, allowing the exploration of productivity dynamics over nearly two decades. I combine these remote ability measures with Bureau of Labor Statistics data on labor productivity and multifactor productivity to estimate dynamic effects. One of the main central findings challenges conventional expectations of immediate productivity effects. Productivity gains from expanded remote work ability materialize gradually rather than instantaneously. Labor productivity improvements peak approximately three years after increases in remote ability, while multifactor productivity effects reach their maximum around the six-year mark. This temporal pattern helps explain why previous studies—particularly those examining sudden pandemic-induced shifts to remote work—have produced inconsistent results. Short-run disruptions and adjustment costs associated with rapid transitions likely mask substantial longer-run productivity benefits that emerge only after organizations adapt. To address endogeneity concerns, I implement an instrumental variable strategy exploiting variation in global imports of computer and communication equipment by non-U.S. countries. This approach isolates changes in U.S. remote work ability driven by worldwide technology diffusion patterns plausibly exogenous to U.S.-specific productivity dynamics. The instrumental variable estimates reveal substantially larger effects than ordinary least squares specifications—approximately an order of magnitude larger—suggesting that conventional approaches severely underestimate productivity benefits of remote capability, likely due to measurement error and omitted variables bias. Quantitatively, a one standard deviation increase in remote ability—corresponding to approximately a 20 percentage point expansion in remote-ready employment share—is associated with labor productivity gains ranging from 20 percent of a standard deviation in baseline specifications to more than two standard deviations in the instrumental variable specification. The findings carry particular relevance for post-pandemic productivity dynamics. Between 2019 and 2022, remote work ability increased by approximately 5 percentage points. The instrumental variable estimates map this expansion to an expected increase in labor productivity of roughly 8 index points over three years. Observed data indicate productivity has risen by approximately 5 index points from 2019 to 2022, suggesting substantial predicted gains have materialized. However, given the significant lags identified, additional productivity improvements attributable to pandemic-era increases in remote capability are likely to continue emerging through the mid-2020s. This perspective reconciles seemingly paradoxical observations of workplace disruption alongside robust productivity growth and suggests current productivity data may understate long-run benefits of remote work adoption.
[slides]
By Maria Tito; Federal Reserve Board of Governors
Presented by: Maria Tito, Federal Reserve Board of Governors
Discussant:   Jill Redmond, U.S. Bureau of Labor Statistics
Total Factor Productivity and Remote Work: Are the Benefits Behind Us? Abstract Remote work surged during the pandemic. We examine the relationship between the rise in remote work and productivity growth among 61 detailed industries over the longer term (2019–2024), where we define remote work based on those primarily working from home as reported in the American Community Survey. We also examine the relationship for the most recent changes in remote work arrangements and productivity. We find a positive relationship between remote work and total factor productivity (TFP) over the long run. We also examine other key factors such as unit labor costs, unit services costs, complementary software investments, and office building costs. In the recent short run (2023–2024), decreases in remote work are associated with increases in labor productivity, but not TFP. We also find that remote work does not result in real hourly compensation growth, despite the productivity gains.
[slides]
By Jill Redmond; U.S. Bureau of Labor Statistics    Sabrina Pabilonia; Bureau of Labor Statistics
Presented by: Jill Redmond, U.S. Bureau of Labor Statistics
Discussant:   Maria Tito, Federal Reserve Board of Governors
Comparing early estimates of US manufacturing output Abstract Measures of labor productivity can be based on different concepts of output. This paper examines two quarterly output measures for the US manufacturing sector. Value-added output is a measure of the value of what was produced minus the costs of all intermediate inputs (materials, energy, and services) used in production. Sectoral output is a measure of the value of what was produced minus the value of intermediates purchased from within the US manufacturing sector. We discuss justifications for using these or other output measures when constructing manufacturing productivity statistics. One issue is which data are available quickly. I test regression models of quarterly growth rates of these output measures for the 2005–2024 period in order to understand their relationships with underlying economic activity. The predictors, which come mainly from the Census Bureau Economic Indicators program, are available shortly after each quarter. Exports, residential construction activity, inventory change, and new business startup rates are good predictors. Durable goods output growth is well predicted by these variables, but nondurable goods much less so.
By Peter Meyer; U.S. Bureau of Labor Statistics
Presented by: Peter Meyer, U.S. Bureau of Labor Statistics
Discussant:   Thomas White, Us Census Bureau
Session 11: Transshipment, Evasion, and Trade Policy April 17, 2026 13:15 to 15:00
Session Chairs:
Ward Reesman, U.S. International Trade Commission
Jacob Howard, MITRE
Heterogeneous effects of rules of origin Abstract (This paper belongs to a proposed session entitled "Transshipment, Evasion, and Trade Policy") This paper studies the heterogeneous impact of rules of origin in trade agreements on trade in intermediate goods. We extend the analysis of Conconi et al. (2018) to estimate the effects of rules of origin in NAFTA on Mexico's imports of intermediate goods, focusing on industry-specific and exporter-specific margins. We find the textile industry is an important driver of the overall estimated negative impact of NAFTA rules of origin on intermediates goods trade from Mexico's non-FTA partners relative to NAFTA partners, with varying estimates across the other industries. For exporters, estimates suggest that developing economies are generally more impacted than developed economies, except when China is excluded from the sample. Many of the most negatively impacted exporters by NAFTA rules of origin are located in Asia, a region associated higher levels of intermediate goods trade.
By Ross Jestrab; U.S. International Trade Commission    Ward Reesman; U.S. International Trade Commission
Presented by: Ward Reesman, U.S. International Trade Commission
Discussant:   Paul Phillips, U.S. International Trade Commission
Tariffs in Translation: Linking Economic Incentives to Linguistic Shifts in Customs Declarations Abstract National-security tariffs create incentives for U.S. importers to strategically reclassify goods into lower-duty categories by adjusting product descriptions and declared tariff lines. This paper develops a systematic empirical framework to identify and quantify how industry-specific tariff shocks alter the content and specificity of shipment descriptions in U.S. customs data, with an emphasis on deliberate misclassification behavior. Moving beyond approaches that rely solely on surface linguistic regularities, we introduce a theory-informed measure of descriptive ambiguity that links linguistic variation to economic incentives, enabling causal inference on how tariff changes affect the precision, substitutability, and classification-relevance of stated product attributes. We then discuss the implications of our research for import elasticity estimation and import enforcement monitoring.
By Jacob Howard; MITRE    Meenu Ravi; MITRE Corporation
Presented by: Jacob Howard, MITRE
Discussant:   Eric Neuyou, USITC
Measuring and Understanding Unexplained Trade: A Gravity Approach Abstract While countries can use transshipments to illegally circumvent tariffs levied on their exports, such illegal transshipments are difficult to measure or calculate. In this paper, we estimate gravity regressions for fourteen aggregated manufacturing sectors in the 1995-2022 period, and use the residuals from those regressions as `unexplained trade' suggestive of the presence of transshipments. We find that the median unexplained trade value among all countries and sectors is negative, and has been decreasing between 1995 and 2022. Findings further suggest that China uses Vietnam and Cambodia rather than Mexico and Canada to re-route exports to the United States, and that Russia re-routes trade through its geographic neighbors. We then estimate the relationship between countries' geopolitical characteristics and the value of unexplained trade they conduct, finding that a pair of countries is expected to have a substantially lower level of unexplained trade if one country, but not the other, is under international sanctions.
[slides]
By Saad Ahmad; University of Arkansas    Paul Phillips; U.S. International Trade Commission
Presented by: Paul Phillips, U.S. International Trade Commission
Discussant:   Ward Reesman, U.S. International Trade Commission
Applications of Network Analysis for Trade Policy Abstract (This paper belongs to a proposed session entitled "Transshipment, Evasion, and Trade Policy") Network analysis comprises tools that depict the structure of the trade relationships among entities (e.g. firms, industries, countries) and determine their importance as a buyer or a supplier of goods. It also describes their trade channels through upstream and downstream linkages as well as direct and indirect linkages. Traditional trade analysis frameworks, such as the gravity model and computer generalized equilibrium often focuses on country-pair or industry-pair interactions. The rise of global value chains and complex regional and geopolitical agreements have created an interconnected trade network, where a policy change in one country can trigger systemic ripple effects across the world. This paper uses a livestock value chain application to demonstrate the utility of network properties in analyzing and quantifying the linkages among industries. This pa- per’s theoretical framework is built on input-output linkages and their role in the transmission of shocks. The analytical framework can be customized to study the diffusion of various polices and shocks such as export restrictions, imposition of tariffs, financial crises, change to foreign direct investments, geopolitical alignments, and supply chain disruptions.
By Eric Neuyou; USITC
Presented by: Eric Neuyou, USITC
Discussant:   Jacob Howard, MITRE
Session 12: Access, Equity, and Reproductive Health: Evidence from Policy and Practice April 17, 2026 13:15 to 15:00
Session Chair: Laura Crispin, Saint Joseph's University
The Eyes of Texas are Upon OB/GYNs: Physician Migration and Crowdsourced Enforcement of Abortion Regulations Abstract A civil liability enforcement mechanism (CLEM) seeks to circumvent constitutionality by replacing criminal prosecution with private civil suits. Unlike criminal enforcement, which targets only direct perpetrators, civil liability under CLEM expands the pool of potentially liable actors and lowers the evidentiary burden, imposing higher expected costs on a broader set of people. Texas Senate Bill 8 (SB-8, 2021) was the first law to apply CLEM to abortion regulation: any private citizen could sue anyone who "aids or abets" an abortion after six weeks of gestation for minimum damages of $10,000, with no cap on total liability and no recovery of legal fees for prevailing defendants. This crowdsourced enforcement architecture was deliberately designed to survive pre-enforcement judicial review by eliminating any state official as a named defendant. Drawing on Becker's (1968) rational-choice framework, CLEM raises expected operating costs for a wider class of reproductive health physicians than criminal enforcement because both the probability of being sued and the set of potentially liable physicians are larger under civil enforcement. These features make SB-8 an important case for understanding how enforcement design, not just the underlying prohibition, shapes labor market responses. We examine the effect of SB-8's civil liability enforcement mechanism on the labor market for reproductive health physicians in Texas. We exploit SB-8's May 2021 signing as a natural experiment, focusing on the thirteen-month window before the Dobbs v. Jackson Women's Health Organization decision. We use two complementary data sources: a Medicare administrative provider panel spanning 2007 to 2022, which tracks physician locations and practice status, and Wagescape job posting data covering approximately 1.3 million physician postings from 2016 to 2022, including posted salaries for roughly half the sample. We estimate triple-difference models that compare changes in outcomes for reproductive health physicians relative to other physicians within Texas, against the same comparison in other states. We find that reproductive health physicians did not leave Texas at differential rates following SB-8. Neither posted salaries nor job posting volume for reproductive healthcare providers changed relative to comparison groups. Heterogeneity analyses by splitting the Wagescape sample into urban and rural subsamples similarly yield null results. The null findings are consistent with the high fixed costs of interstate practice relocation, estimated at $150,000 to $250,000 in licensing, credentialing, and malpractice insurance costs, which may make short-run migration unresponsive even to substantial increases in expected civil liability. Our results stand in contrast to prominent news media accounts claiming a large physician migration response following SB-8, and complement a rapidly emerging literature documenting that criminal abortion restrictions reduce OB/GYN supply. We contribute to this literature by showing that CLEM, despite its theoretically broader reach, had no discernible increased effect on physician location or labor market conditions above and beyond the effect of the shifting national abortion policy landscape. These findings have implications for understanding how enforcement mechanism design interacts with physician labor market frictions, and for forecasting the labor market consequences of civil liability enforcement should it be applied to other domains of reproductive or healthcare policy.
By Martin Andersen; University of North Carolina at Greensbo    Kaden Grace; University of Tennessee
Presented by: Kaden Grace, University of Tennessee
Discussant:   Jiayuan Wang, University of Washington
Fertility responses to local taxation Abstract This study asks how fertility is impacted when local governments leverage property taxation to fund public services. In the United States, property taxation is the primary revenue instrument for local governments, and strengthening the local welfare state relies heavily on property tax financing. In recent years, progressive local governments have increasingly used property tax levies to fund public investments in children. While locally funded investments in children are often motivated by concerns about inequality, their effectiveness depends not only on the benefits they provide but also on how their financing interacts with household behavior. The fiscal effects of a policy are part and parcel of its public impact: whereas there has been much discussion regarding how public provision of childhood services can equalize opportunities among children, far less attention has been paid to how the property tax increases essential to financing these services affect family formation and fertility. Lower-income households face higher effective property-tax burdens relative to income, and part of the tax burden is consistently passed on to renters through higher housing costs (ITEP 2024). If low-SES groups respond by reducing or delaying childbearing at higher rates, the beneficiaries of property-tax-funded programs may increasingly skew toward older, wealthier, or home-owning families. I apply a sharp regression discontinuity (RD) design exploiting close school tax referenda in Ohio (2010–2018) to study the effects of property taxation on births, using the American Community Survey (ACS) data to construct aggregate and subgroup fertility outcome measures at the school-district level. I choose Ohio due to unique institutional features favorable to the study purpose: Ohio has a plethora of over 2,700 school tax referenda during the study period, 87 percent of which were property tax ballots; and House Bill 920, which caps real revenue growth from existing levies through automatic millage rollbacks, creating a tight mapping between referendum outcomes and changes in effective property tax liabilities, and thus enabling elasticity estimates with respect to marginal property tax liability, apart from estimates of effects of levy passage. As a first step, I estimate effects of levy passage, using Calonico et al. (2014)’s rdrobust model in Stata which automatically selects optimal bandwidth and applies kernel weighting. Density and balance tests show no evidence of manipulation or discontinuities of age composition of the populations around the referendum cutoff, supporting the validity of the RD design. I find passage of a new school-district property tax levy reduces the total fertility rate by 0.25 births in the district in the five-year period post-referendum, a substantial (12%) reduction compared to the study-period average TFR at 2.03 and comparable in size to the negative effect on TFR of the 2008 financial crisis shown in Comolli (2017). Group fertility analyses reveal that fertility reductions are concentrated among younger women (ages 20–34) and those whose highest education attainment is high school or below, and no significant effect is found among older or college-or-above-educated mothers. Findings suggest heterogeneous fertility responses to property taxation along socioeconomic lines and potential need for compensatory policies.
By Jiayuan Wang; University of Washington
Presented by: Jiayuan Wang, University of Washington
Discussant:   Gabriel Cruz, University of Maryland, College Park
Preventing Unplanned Pregnancies: Impacts on Educational Outcomes and the Next Generation Abstract This paper provides evidence on the consequences of preventing an unplanned pregnancy---for the mother and for the next generation---using a contraceptive access shock in Chile. In September 2015, Chile eliminated the prescription requirement for emergency contraception, sharply increasing access in a context where abortion was completely banned. I combine individual-level administrative records on births, student outcomes, and pharmacy sales to exploit variation in pharmacy proximity, combined with the sharp temporal change in conceptions around the reform, to identify causal effects. The access shock reduced birth rates, especially among adolescents, and improved their educational outcomes. Children conceived after the shock have higher standardized GPA scores and attendance rates in elementary school than those conceived just before. The mechanism analysis shows that parental characteristics shift at the reform cutoff, indicating changes in the composition of who becomes a parent. However, within-family estimates across siblings conceived on either side of the reform remain positive, suggesting that birth planning itself---independent of selection into parenthood---improves children's outcomes. These findings provide suggestive evidence that greater access to family planning generates intergenerational benefits by improving the conditions into which children are born.
By Gabriel Cruz; University of Maryland, College Park
Presented by: Gabriel Cruz, University of Maryland, College Park
Discussant:   Kaden Grace, University of Tennessee
Session 13: Lowering the Transition: Storage, Efficiency, and the Value of Energy Assets April 17, 2026 15:15 to 17:00
Session Chair: Cristina Miller, USDA Rural Development
How Battery Storage Reduces the Cost of Grid Stability under Transmission Constraints Abstract Electricity systems with growing shares of wind and solar generation face a key challenge: maintaining system stability when supply and demand fluctuate rapidly. To address this, system operators procure short-notice “stability services” that correct imbalances within seconds or minutes. These services are essential for reliability but can become extremely expensive when supply must be sourced locally due to transmission constraints, creating opportunities for market power. This paper studies how large battery storage affects the cost of these stability services using evidence from Australia’s National Electricity Market. The analysis focuses on the 100-megawatt Hornsdale Power Reserve, one of the world’s first utility-scale batteries, which began operating in South Australia in late 2017. Rather than asking whether battery entry lowers prices on average, the paper examines when and under what conditions batteries have the largest effects. Using high-frequency market data, the paper distinguishes between two operational states of the grid. In separated intervals, transmission constraints isolate South Australia from the rest of the market, forcing the system operator to rely on local providers to maintain stability. In connected intervals, regions are fully integrated and stability services can be sourced nationally. To capture battery participation more precisely, the analysis measures the amount of battery capacity offered at competitively low prices, rather than relying on a simple before-and-after comparison. The results show that battery storage has strong state-dependent effects. When South Australia is isolated and local market power is most likely to arise, additional competitively priced battery supply leads to large reductions in the prices paid for stability services. Even when the grid is fully connected, battery participation still produces smaller but statistically meaningful price reductions at the national level, indicating broader spillover effects. These findings have clear policy implications. They suggest that the value of battery storage is concentrated in periods of system stress, when reliability risks and costs are highest. Evaluations based solely on average price changes may therefore understate the public benefits of storage investment. More broadly, the paper highlights the importance of accounting for transmission constraints and institutional market design when assessing new energy technologies and their role in supporting reliable and cost-effective electricity systems.
By Dereck Yang; University of Maryland
Presented by: Dereck Yang, University of Maryland
Discussant:   Scott Wentland, Bureau of Economic Analysis
Behavioral Barriers and Cost-Effective Design in Residential Energy Retrofit Programs: A Serbian Case Study Abstract Governments increasingly rely on residential energy retrofit programs to reduce emissions, improve energy security, and lower household energy costs. However, the cost-effectiveness of these programs depends not only on engineering estimates but also on household behavior, program design, and participation constraints. This paper evaluates a national residential retrofit initiative in Serbia, with a focus on behavioral barriers to participation in multi-unit apartment buildings. In 2021, the Government of Serbia introduced subsidies for insulation, window replacement, and heating system upgrades. While the program generated interest among single-family homeowners, participation in multi-unit apartment buildings remained extremely low, with only about 3.5 percent of buildings reaching agreement to retrofit. To understand the causes of low participation, we conducted structured interviews with four stakeholder groups: homeowner association managers, apartment residents, district heating managers, and municipal authorities. The interviews indicated that information frictions, coordination problems, and transaction costs were key barriers to collective decision-making. Based on these findings, we designed a lab-in-the-field experiment to test how different information structures and delivery methods affect collective investment decisions. The experiment substantially reduced information and transaction costs and allowed participants to engage directly with retrofit decision scenarios. The study revealed that when households “own” the information—meaning they actively process and internalize the decision inputs—they are more inclined to invest in energy efficiency. In the experimental setting, approximately 75 percent of multi-unit building residents reached agreement to retrofit, compared to the observed 3.5 percent participation rate in the field. The Ministry of Mining and Energy used these findings to refine program design and communication strategies. The results suggest that relatively low-cost behavioral and administrative interventions—particularly those that reduce information and coordination costs—can substantially improve the effectiveness of subsidy programs. These findings have implications for the design of energy efficiency and decarbonization programs in both emerging and advanced economies.
By Weiwei Tasch; George Mason University
Presented by: Weiwei Tasch, George Mason University
Discussant:   Dereck Yang, University of Maryland
Valuation of Subsoil Energy Assets: Pilot Estimates for the U.S. National Balance Sheet Abstract Energy plays a critical role in the U.S. economy. While the U.S. Government produces high quality estimates of physical quantities of subsoil assets like oil and natural gas reserves, the national economic accounts do not currently report a detailed accounting of natural resources on the national balance sheet (or Integrated Macroeconomic Accounts). This paper develops new estimates of the value of U.S. subsoil energy assets, focusing on oil and natural gas as test cases for improving the national accounts. We estimate the value of these assets as the net present value of natural resource rents for the oil and gas industries. These natural resource rents are commonly estimated as a residual, after subtracting intermediate input, labor, and estimated capital costs from total revenue. We develop a modified version of the residual method that uses the return on corporate bonds as a proxy for the rate of return to capital and adjusts depreciation costs using the corporate tax rate. One novelty of this approach is that it adjusts for inconsistencies that arise when using other formulations of the residual method, such as those that assume a constant rate of return to capital over time. Under the most reasonable assumptions, we estimate the total value of U.S. oil and gas subsoil assets to be worth about $2.3 trillion. Further vetting and refining of the methods, assumptions, and data sources for these estimates will serve as an initial step toward incorporating subsoil assets on the national balance sheet.
By Matthew Chambers; U.S. Bureau of Economic Analysis    Steven Anderson; USGS    Julie Hass; U.S. Bureau of Economic Analysis    Melissa Lynes; EIA    Ian Mead; EIA    Scott Wentland; Bureau of Economic Analysis
Presented by: Scott Wentland, Bureau of Economic Analysis
Discussant:   Weiwei Tasch, George Mason University
Session 14: Housing Costs and Household Strain: Insurance, Mortgages, and Rent April 17, 2026 15:15 to 17:00
Session Chair: Sisi Zhang, Federal Reserve Bank of Philadelphia
The Role of Residual Policies in Homeowners Insurance Market Abstract Rising climate risk, insurer withdrawals, non-renewals, and premium spikes have drawn increasing attention to state-created residual insurance plans—the insurers of last resort designed to provide basic coverage when options in the voluntary market are unavailable. This paper provides the first loan-level comprehensive analysis of how residual insurance policies affect homeowners. We examine the following research questions: (1) Who relies on these insurers of last resort? (2) How do state residual operations, depopulation incentives, and regulation on private insurers shape the participation of the residual market? (3) What factors drive entry into and exit from the residual market? (4) When borrowers transition from the residual market back to the voluntary market, do they face higher premiums and/or sort into lower-rated insurers?
By Nam Nguyen; Stanford University    Sisi Zhang; Federal Reserve Bank of Philadelphia
Presented by: Sisi Zhang, Federal Reserve Bank of Philadelphia
Discussant:   Laurence Bristow, Bank Policy Institute
Price Regulation in Consumer Financial Markets: The Case of Mortgage Escrow Abstract We use the 2018-2024 Home Mortgage Disclosure Act data in a difference-indifference framework to show that regulating the interest rate lenders pay on mortgage escrow funds does not benefit the average consumer. Lenders offset much of the lost revenue by increasing up-front origination fees. They increase origination fees the most for low-income borrowers while origination fees for high-income borrowers remain unchanged, suggesting a regressive cross-subsidy. The regulation also lowers the likelihood applications are originated, an effect arising mostly from lenders denying applications. Again, low-income applicants are more affected. Unlike for pricing, effects on origination indicate a market distortion signaling total welfare losses stemming from this binding price floor.
By Laurence Bristow; Bank Policy Institute    Daniel Grodzicki; None
Presented by: Laurence Bristow, Bank Policy Institute
Discussant:   Sisi Zhang, Federal Reserve Bank of Philadelphia
Rent Burden and Household Well-Being in the SIPP Abstract Since approximately 2020, rising housing costs in the U.S. have increased the salience of housing affordability for policymakers, researchers, and the public. One increasingly popular indicator of housing affordability is the share of the population facing a ‘housing cost burden,’ i.e. paying over 30% of their income for their rent or mortgage. The Survey of Income and Program Participation (SIPP) includes detailed data on income, earnings, housing, and utility costs over a moderately long time period (2013 through 2023) alongside measures of program participation and household well-being. The availability of these data provides a compelling opportunity to estimate and analyze the extent of housing cost burdens and how they have changed over time. In this paper, I present a new set of SIPP summary estimates of national and subnational housing cost burden prevalence. I summarize these estimates by demographic characteristics, economic characteristics (e.g. earnings, income, employment), household well-being characteristics (e.g. food security and reported non-payment of rent or mortgage), and housing tenure. These estimates provide new context for discussions of housing affordability, especially regarding variation in rent burden dynamics across the income distribution. SIPP data also allow me to investigate the relationship between the social safety net and self-reported difficulties paying for housing. Preliminary findings provide some evidence of limited change over time in the prevalence of housing cost burdens, but meaningful variation in the time trends across the income distribution as more high income households became cost burdened over time. Similar shifts also occur across demographic categories and measures of well-being. Beyond these new population statistics, I present models predicting either individual rent burden or the likelihood of non-payment of rent or mortgage conditioning primarily on economic variables while controlling for other demographic factors. I also fit models of state-level variation in rent burden trends for states with sufficient sample size, to showcase geographic differences in these trends over time. Taken together, my results provide a novel snapshot of the evolution of housing affordability in recent years and showcase variability in the regional and demographic context of these estimates.
By Tim Smith; US Census Bureau
Presented by: Tim Smith, US Census Bureau
Discussant:   Hugh Montag, Bureau of Labor Statistics
Nonresponse Imputations and Related Measurement Issues in the CPI for Shelter Abstract Shelter is the largest component of U.S. CPI, so the accuracy of shelter inflation is critical for the accuracy of overall inflation. Nonresponse in the BLS Housing Survey, which underpins CPI shelter measurement, has increased over time and now represents roughly 40 percent of observations. Missing rents are imputed using a class-mean approach based on a limited set of covariates. We evaluate this method and find it likely biases measured shelter inflation, and therefore overall CPI inflation, upward. We propose an alternative imputation strategy that incorporates structure type and tenure length, both of which are correlated with nonresponse and rent growth.
By Hugh Montag; Bureau of Labor Statistics
Presented by: Hugh Montag, Bureau of Labor Statistics
Discussant:   Tim Smith, US Census Bureau
Session 15: Innovation, Risk, and Reward: Firms, Patents, and the AI Frontier April 17, 2026 15:15 to 17:00
Session Chair: Susan Fleck, Bureau of Labor Statistics
Environmental Innovation and Firm Strategy: Preliminary Findings from the 2023 Annual Business Survey Abstract The longer society waits to reduce carbon emissions on an eventual path to net zero, the more critical environmental innovation becomes to averting a catastrophic rise in global temperatures above 2°C. Innovation surveys such as the EU Community Innovation Survey (CIS) and the US Annual Business Survey (ABS) now include questions on environmental innovation related to reducing a firm’s carbon footprint. However, as with all self-reported innovation questions that comport with the Oslo Manual (OECD/Eurostat 2018), the requirements for reporting are only that the innovation is available to users and that it comprises a substantive change from what was done before. Neither the success nor the type of environmental innovations are investigated in the innovation surveys making the findings unpersuasive due to potential social desirability bias (Tourangeau, et al. 2000) and of limited usefulness for informing environmental innovation policy related to the energy transition. The environmental innovation questions in the CIS since 2012 have generated a sizable amount of research to date related to factors associated with such innovation, and outcomes related to employment and productivity growth (Kemp, et al. 2023; Horbach, et al. 2013; Horbach and Rennings 2013; Ghisetti, et al. 2015). However, the CIS to date has not investigated decarbonization strategies explicitly so is silent on how environmental innovations are thought to be improving carbon performance. The 2023 Annual Business Survey (ABS) and its Sustainability and Climate Impact module provide a unique opportunity to assess the methods being used to increase carbon performance for manufacturing firms that comprise the bulk of business emissions. Questions to be investigated include: • Are affirmative environmental innovation responses pertaining to reduction in carbon emissions associated with reported decarbonization strategies? • Are affirmative environmental innovation responses pertaining to reduction in carbon emissions associated with reported decarbonization actions? • Are there firms with aggressive decarbonization strategies or action that do not report environmental innovations? Alternatively, are there firms reporting environmental innovation with respect to carbon emissions that do not report any decarbonization strategies or actions? This preliminary analysis will inform follow-on analysis linking environmental innovation and decarbonization initiatives to observed carbon performance available in the Manufacturing Energy Consumption Survey (MECS) collected for the same reference year. These data are not currently available but should be available in the Federal Statistical Research Data Center system sometime in the Spring of 2026. The follow-on research will inform the construct validity of the environmental innovation measure as well as assessing the value of collecting data on specific decarbonization initiatives.
By Timothy Wojan; Massive Data Institute    Zheng Tian; Pennsylvania State University    Minsu Kim; Pennsylvania State University    Jason Brown; Federal Reserve Bank of Kansas City    Stephan Goetz; Pennsylvania State University
Presented by: Timothy Wojan, Massive Data Institute
How Does Patent Protection Affect Venture Capital Investment? Abstract Patent protection is the primary legal mechanism governing how inventors appropriate returns from innovation. By defining the scope and strength of the rights it confers, the patent system can shape how investors allocate capital to new ventures. Yet the direction of this influence remains contested. One view posits that patent protection enables startups to use patents to attract capital by signaling innovation quality and protecting their inventions. A second view argues that patent protection enables incumbents to use patents strategically to erect legal barriers that raise the costs of market entry for startups, thereby reducing investor interest in new entrants. A third view suggests that patent protection has limited influence on startup funding because companies frequently rely on alternative appropriability mechanisms such as lead time and secrecy. The venture capital (VC) industry provides a natural setting to study the relationship between patent protection and startup funding. As professional intermediaries allocating capital on behalf of limited partners, venture capitalists are highly responsive to shifts in legal and institutional environments. Moreover, VC-backed startups disproportionately drive innovation and account for about half of recent IPOs. VC firms’ response to changes in patent protection is therefore informative about the broader economic consequences of changes in the patent system. To identify the effect of patent protection on VC investment, we exploit the U.S. Supreme Court’s Bilski v. Kappos, 561 U.S. 593 (2010), decision as a quasi-natural experiment. The ruling unexpectedly narrowed the scope of patent-eligible subject matter under Section 101 of the Patent Act by reaffirming the exclusion of abstract ideas and declining to endorse the Federal Circuit’s machine-or-transformation test as the exclusive eligibility standard. Although the Court did not articulate a comprehensive replacement test, the decision raised rejection risk for certain types of inventions, particularly in software- and business-method-related technologies. Importantly, Bilski applied immediately to both new patent applications and the existing stock of patents, shifting expectations about the scope of patent protection and the returns to innovation. We find that industries more exposed to Bilski experienced significantly larger increases in VC investment in the years following the decision. These effects are economically meaningful and are driven almost entirely by increases in initial funding rather than follow-on financing. The results are robust to alternative specifications and are not explained by pre-existing trends, changes in entrepreneurial entry, or differential exposure to the Great Recession. Investor responses to Bilski also vary systematically with startups’ reliance on patents. Greater exposure is associated with increased VC investment in startups without patents but reduced investment in startups holding pre-Bilski patents, for both initial and follow-on rounds. This pattern is consistent with our conceptual framework, in which weaker patent protection reduces incumbent deterrence while simultaneously lowering the private value of patents for patent-dependent firms. Taken together, the evidence reveals a more nuanced link between patent rights and entrepreneurship than is commonly assumed: under some conditions, weaker patent protection can redirect VC capital toward less patent-intensive startups by lowering barriers to VC-backed entry.
[slides]
By Elif N. Guler; Indiana University Bloomington Kelley School of Business    Alexander Montag; University of Warwick    Michael Woeppel; Indiana University
Presented by: Elif N. Guler, Indiana University Bloomington Kelley School of Business
Discussant:   Timothy Wojan, Massive Data Institute
AI and Coder Employment: Compiling the Evidence Abstract The rapid improvement of large language models (LLMs) has raised questions about if and when LLMs will automate jobs. In this paper we evaluate whether LLMs have had any discernible impact on the aggregate labor market so far. We focus on occupations that are computer programming-intensive, motivated by recently-disclosed data that shows coding is one of the most LLM-exposed tasks. Linking O*NET to CPS we find that aggregate employment of coders has decelerated sharply since the introduction of ChatGPT. While show that much of this deceleration is attributable to the exposure of coders to slowing industries. The slowing growth of coder-intensive industries appears to be in line with long-run trends. However, even controlling for industry-level factors there is a substantial post-ChatGPT deceleration consistent with an occupation-specific “LLM shock”.
By Leland Crane; Federal Reserve Board    Paul Soto; Federal Reserve Board
Presented by: Leland Crane, Federal Reserve Board
Discussant:   Breno Braga, Urban Institute
The Fog of Innovation: Does Uncertainty from R&D Cloud the Horizon? Abstract Not all uncertainty shocks are alike. We isolate measures of uncertainty stemming from R&D activity (the "fog of innovation") and show they can have starkly different macroeconomic effects compared to measures of aggregate uncertainty. We build these measures from micro-foundations using analyst forecasts. First, we find that firm-level R&D is related to increased uncertainty at forecast horizons of up to several years. We then construct measures of aggregate and R&D-driven uncertainty, finding that, while aggregate uncertainty has a negative impact on industrial production consistent with a negative demand shock, measures of R&D-based uncertainty have contrasting effects. In particular, while a measure of R&D-based uncertainty based on analyst disagreement acts like a negative demand shock, a measure based on forecast errors has a positive impact on industrial production, consistent with a positive supply shock. Our aggregate and R&D-based uncertainty is thus useful for forecasting both microeconomic and macroeconomic outcomes at different horizons.
By Matin Safshekan; George Washington University    Roberto Samaniego; George Washington University
Presented by: Matin Safshekan, George Washington University
Discussant:   leo sveikauskas, BLS
Session 16: The Utility of State Wage Data April 17, 2026 15:15 to 17:00
Session Chair: Danielle Sandler, U.S. Census Bureau
Business Formations of College Graduates Abstract There is an extensive literature that has examined the economic returns to college for people who attend and complete their schooling. However, this research has focused almost exclusively on the wage and salary outcomes of graduates, because these outcomes are the easiest to measure, particularly with the availability of administrative records on earnings and graduates. One unexamined outcome is the business formation of college graduates. This margin is particularly important for two reasons. First, measuring business formation captures economic outcomes not typically measured in the literature, which heavily relies on Unemployment Insurance wage record data, and the majority of business founders are not able to collect unemployment insurance. Second, to the extent that these businesses become successful and hire employees, the social benefits of colleges are underestimated if we focus solely on wage and salary jobs. In this paper, we link administrative records on college graduates from 13 states to the universe of applications for an Employer Identification Number (EIN) with the IRS (Form SS-4). This linkage allows us to identify business formation of post-secondary graduates at a number of degree levels. Furthermore, we link these EINs to the Longitudinal Business Database to identify which of these businesses became employer businesses within two years of the EIN application. This novel linkage allows us to document a number of important facts about the business formation of graduates. First, we show that there is large variation in the rates of business formation by degree level and degree field, and we show that a number of these businesses are high quality. We document the extent to which graduates form businesses in similar industries as their employment. Importantly, we also show that graduates are more likely to form businesses in the state of their institution. Finally, we show that most businesses are formed in mid-career. We believe these descriptive statistics are the beginning of a rich vein of research on the broader economic impacts of college graduates, and that documenting their size and existence is an important first step.
By Andrew Foote; US Census Bureau    Cody Orr; U.S. Census Bureau
Presented by: Cody Orr, U.S. Census Bureau
Discussant:   Ishrat Alam, South Carolina Department of Employment and Workforce
Unemployment Insurance Employment Outcomes (UIEO): The Composition and Re-employment Dynamics of UI Claimants Abstract The COVID-19 Recession saw unemployment insurance (UI) programs take on unprecedented roles in the stabilization of the U.S. labor market, motivating renewed interest in understanding the effects of UI policies. However, analyzing UI policy and the measurement of associated labor market outcomes is fraught with measurement issues and data limitations. In this paper, we use newly assembled administrative data from several U.S. states that link high-frequency UI claims to longitudinal earnings and employment records from the Longitudinal Employer-Household Dynamics (LEHD) program. These data give us the most comprehensive picture to date of UI claimants in the U.S. and allow us to characterize their demographic, job, and employer characteristics and track their labor market outcomes. We document the composition of UI claimants, their average benefits and claim duration, and their associated employment outcomes such as re-employment timing, earnings dynamics, and their mobility across industries, employers, and states. We link weekly initial and continuing UI claims from six states in 2020 to quarterly earnings in LEHD. As LEHD is derived from UI earnings records, we have complete coverage of UI-eligible employment and earnings histories for the claimants within our sample regardless of their state of employment. The weekly frequency of paid claims enables us to measure unemployment spells with more precise timing than typical quarterly or annual administrative earnings data. We also know for these claims that the worker is unemployed rather than out of the labor force, and that the worker was subject to a layoff. This detail allows us to observe displaced workers and their unemployment spells with unprecedented accuracy and also track them longitudinally for significant spans of time. We highlight the utility of this data for research on UI as well as its potential for the production of public-use statistics on the UI population. We document substantial heterogeneity in earnings losses and gains across claimant subgroups, shaped by differences in unemployment spell duration, re-employment type, and geographic and industry mobility. We highlight that variation in earnings outcomes is sensitive to several measures of mobility, including recall, re-employment to a new industry, or to an employer in a new state. Mobility across employers, industry, and state of employment are associated with significant dispersion in outcomes relative to workers who did not move. Earnings gains and losses from displacement can be substantial and have a strong industry and unemployment duration component. Our measures of UI collectors and their characteristics contribute to a growing literature measuring the UI system, its coverage, and heterogeneity in participation. Our results provide detail on the heterogeneity of claimants, with the ability to also track and measure their re-employment outcomes. We bring new administrative data with precise measurement of the timing, duration, and collection of UI benefits along with the infrastructure to measure workers' outcomes. We also contribute to a large literature measuring the cost of job loss. Our results provide precision in identifying involuntary joblessness and sufficient detail to better understand the heterogeneity in transitions and costs from job loss.
By Lawrence Warren; US Census    David Wasser; U.S. Census Bureau    Caelan Wilkie-Rogers; U.S. Census Bureau
Presented by: David Wasser, U.S. Census Bureau
Discussant:   Linden McBride, South Carolina Department of Employment and Workforce
From CIPs to SOCs: Mapping the Path from Education to Employment using State Education and Wage Records Abstract To what extent do individuals enter the occupation for which their education has prepared them? Is the path from educational credential to occupation clearer in some career domains than others? Using education records provided by the South Carolina Commission on Higher Education matched to wage records from the South Carolina Department of Employment and Workforce, we trace 2023 graduates from their degree of study, using both degree level and Classification of Institutional Program (CIP) codes, to their post-graduation occupations in 2024 and 2025, using Standard Occupation Classification (SOC) codes. We find that over 70 percent of these workers are employed in occupations outside of the National Center for Education Statistics (NCES) CIP to SOC crosswalk. To better understand this finding, we explore patterns in underemployment relative to education, overemployment relative to education, and patterns within and across career domains as defined by the National Career Clusters Framework.
By Linden McBride; South Carolina Department of Employment and Workforce
Presented by: Linden McBride, South Carolina Department of Employment and Workforce
Discussant:   David Wasser, U.S. Census Bureau
Estimating Coverage of Postsecondary Graduates in Linked Education and Unemployment Wage Records: Evidence from South Carolina Abstract Unemployment insurance (UI) wage data linked with educational data are widely used to study workforce outcomes among postsecondary graduates. However, few studies have assessed the extent to which UI wage data capture the underlying population of graduates. In this study, we examine coverage of postsecondary graduates in South Carolina using 2009-2023 college completion data from the South Carolina Commission on Higher Education (CHE) matched to UI wage data from the South Carolina Department of Employment and Workforce. First, we benchmark CHE completion data against Integrated Postsecondary Education Data System (IPEDS) completion data for South Carolina institutions. Then, we estimate coverage rates defined as the proportion of graduates observed in UI wage records 1, 3, 5, and 10 years after graduation. We assess coverage across degree levels and Classification of Instructional Program (CIP) codes. By assessing variation in coverage of postsecondary graduates by degree level, program, and time since graduation, this study informs the scope of linked education and UI wage records for measuring workforce outcomes in South Carolina.
By Ishrat Alam; South Carolina Department of Employment and Workforce
Presented by: Ishrat Alam, South Carolina Department of Employment and Workforce
Discussant:   Cody Orr, U.S. Census Bureau
Session 17: Capital, Capacity, and Credibility: Macroeconomic Dynamics in Theory and Data April 17, 2026 15:15 to 17:00
Session Chair: Gary Cornwall, Bureau of Economic Analysis
Disaggregate Capital Accounts: Accounting for States' Human Capital Abstract Human capital is recognized as a main driver of economic growth but is not measured like traditional capital--equipment, structures and intellectual property products-- in the national accounts. This study estimates human capital stocks and investment between 2008 and 2023 at the state level. I measure human capital as the present value of the population's future earnings. I build "bottom up" estimates of human capital by state by combining CPS data with small area estimation techniques. Results allow me to rank and compare states' human capital investment and growth in the 2008-2023 period. I report results with and without accounting for state-specific purchasing power parities. Preliminary results show that higher income areas—Washington, DC, Massachusetts, New Jersey--have higher estimated levels of human capital due to high expected future earnings paths. After accounting for state purchasing power parities, Washington DC, Utah, and North Dakota have the highest average per-capita human capital in 2023, reflecting a good balance between expected lifetime earnings and purchasing power parity of the geography. States with low initial wages during the Great Recession around the starting period of 2008 have the highest price-adjusted growth in per-capita human capital, ranging between 3.3 and 3.5 percent a year.
By Justine Mallatt; Bureau of Economic Analysis
Presented by: Justine Mallatt, Bureau of Economic Analysis
Discussant:   Yu Sugisaki, Boston College
Capacity Utilization and Inflation Dynamics in New Keynesian Models Abstract We incorporate labor hoarding into a New Keynesian model with endogenous capacity utilization. Firms choose capacity under demand uncertainty and meet realized demand by first adjusting labor intensity before expanding capacity. This asymmetry links productivity and inflation to the capacity utilization rate and generates three state-dependent implications that are difficult to jointly generate in standard New Keynesian models. First, following an expansionary demand shock, labor productivity rises procyclically as firms intensify the use of existing idle capacity before investing in new capacity. Second, the labor share can respond countercyclically when productivity gains—driven by higher utilization—outpace wage growth. Third, inflation exhibits a hump-shaped response, but may increase sharply at high utilization rates as markups rise and productivity effects fade. We characterize the conditions under which these dynamics arise and show, using Bayesian IRF matching, that the model delivers impulse responses consistent with the data. Our results highlight the central role of capacity constraints in shaping the determinants, dynamics, and distributional consequences of inflation.
By Ignacio Gonzalez; American University    Vasudeva Ramaswamy; American University
Presented by: Vasudeva Ramaswamy, American University
Discussant:   Justine Mallatt, Bureau of Economic Analysis
Minimum Wages and Optimal Monetary Policy Abstract Central banks increasingly emphasize that minimum-wage hikes can act as macroeconomic cost-push shocks with broader real effects, while business leaders caution that rapid increases in the minimum wage may threaten firms’ viability. Motivated by these concerns, I study the macroeconomic consequences of faster minimum-wage growth and characterize the optimal monetary policy response in a dynamic stochastic general equilibrium (DSGE) model with endogenous firm entry and exit. Specifically, I extend an existing New Keynesian model with firm dynamics (e.g., Colciago and Silvestrini, 2022) by introducing a statutory minimum wage that enters firms’ marginal costs. The extension has two key features. First, the economy is populated by minimum-wage households that behave as hand-to-mouth consumers, alongside high-wage households that follow the permanent income hypothesis. Second, aggregate labor input is modeled as a CES composite of labor supplied by minimum-wage and high-wage workers, implying imperfect substitution between the two types. I calibrate the model to the Japanese economy, which provides a useful laboratory given its relatively large low-wage sector and the prospect of sizable policy-driven minimum-wage increases. My main findings are as follows. First, when calibrated to the Japanese economy, the model predicts that a one-percentage-point increase in real minimum-wage growth raises the aggregate price level by about 12 bps within a year, reduces output and employment by 0.6% and 0.7%, and lowers the mass of operating firms by 0.7%. Second, the model’s impulse responses align qualitatively with empirical patterns in Japanese regional data: two-way fixed-effects local projections reveal that minimum-wage accelerations are followed by higher prices and weaker real activity, alongside a deterioration in firm dynamics. Third, relative to a Taylor-rule benchmark, the optimal commitment policy is often more contractionary under preference weights on real-activity stabilization that are commonly used in quantitative policy analyses.
[slides]
By Yu Sugisaki; Boston College
Presented by: Yu Sugisaki, Boston College
Discussant:   Vasudeva Ramaswamy, American University
Session 18: Prices in Practice: Measurement Challenges Across Markets and Households April 17, 2026 15:15 to 17:00
Session Chair: Erick Sager, Federal Reserve Board
Hedonic price indexes under static pricing: an application to PPI microprocessors Abstract Price change for microprocessors largely coincides with product turnover. This static pricing challenges some price index methods and makes accounting for quality change paramount in designing price indexes. We evaluate the performance of several hedonic methods of quality adjustment under static pricing. We find the relative performance of these methods depends on sample size. For the small product samples feasible for microprocessors, the low variance of time-dummy hedonics gives them an advantage over less simple specifications, but with the potential downside of being more biased.
By Steven Sawyer; Bureau of Labor Statistics    Brian Adams; University of Akron
Presented by: Steven Sawyer, Bureau of Labor Statistics
Discussant:   Ana Aizcorbe, Bureau of Economic Analysis, retired
When Do Chained Deflators for IT Goods Overstate Quality Change? Abstract This paper argues that, in general, chained price indexes for IT goods will show gypically not account for quality change properly, a problem we call commingling. We compare a chained index to the weighted time product dummy index (WTPD), an index that we argue does not commingle, and show that the lifecycle patterns typically seen in prices and expenditure shares for these goods can push the chained Törnquist index to fall faster than the WTPD index (and, hence, overstate quality change). We also show an exception where the chained index will not commingle and illustrate these points using price indexes reported in the literature.
By Ana Aizcorbe; Bureau of Economic Analysis, retired
Presented by: Ana Aizcorbe, Bureau of Economic Analysis, retired
Discussant:   Steven Sawyer, Bureau of Labor Statistics
Intergenerational Heterogeneity in Inflation: A Migration Story Abstract We document a previously overlooked source of bias in the measurement of inflation arising from the interaction between the Consumer Price Index’s (CPI) two-step aggregation procedure and internal migration across geographic areas. While the Chained CPI (C-CPI-U) corrects for substitution across goods by updating expenditure weights over time, both the CPI and the C-CPI-U maintain fixed geographic weights. As a result, neither index accounts for shifts in population and expenditure across locations with different price levels. We develop a tractable model of consumption and location choice to quantify the resulting geographic aggregation bias. In the model, households choose both what to consume and where to live, and migration responds endogenously to differences in real wages across areas. Because the CPI aggregates item–area prices using reference-period expenditure weights, changes in the spatial distribution of the population generate a wedge between measured inflation and the true cost of living when households relocate toward areas with different price levels. We derive a first-order approximation showing that this bias depends on the divergence between reference-period and contemporaneous item–area expenditure shares, which in turn reflects migration patterns. We apply the framework to housing price variation and observed U.S. migration patterns from 1980 to 2019 to quantify the magnitude of the bias. When households systematically relocate across areas with different housing costs, measured inflation can be understated or overstated relative to an index that chains across geographic areas. Our results imply that failing to account for migration may distort intergenerational comparisons of real income and wealth.
By Michael Navarrete; Federal Reserve Bank of Atlanta
Presented by: Michael Navarrete, Federal Reserve Bank of Atlanta
Discussant:   Mark Kutzbach, Federal Deposit Insurance Corporation
Bank Competition for Neighborhood Deposits Abstract Spatial markets are typically measured using administrative boundaries chosen for convenience rather than grounded in economic behavior. We estimate tract-level banking concentration using a structural model of household deposit allocation, yielding heterogeneous and overlapping catchment areas. This approach reveals that 89% of variation in local competitive conditions occurs across neighborhoods within MSAs rather than between MSAs. Monte Carlo merger simulations reveal that coarse market definitions can mask localized competitive harm: 2.6% of simulated mergers pass MSA-level screening thresholds while substantially increasing concentration in ten or more neighborhoods. These false negatives arise systematically when merging firms serve similar customers but operate spatially segmented networks. For understanding competition in any spatially differentiated market, knowing where a household lives within a market matters more than knowing which market it lives in.
By Mark Kutzbach; Federal Deposit Insurance Corporation    Gary Wagner; University of Louisiana at Lafayette    Christopher Watson; Federal Deposit Insurance Corporation
Presented by: Mark Kutzbach, Federal Deposit Insurance Corporation
Discussant:   Michael Navarrete, Federal Reserve Bank of Atlanta
Session 19: Poster Session & Networking Reception April 17, 2026 17:00 to 18:30
Session Chair: David Johnson, International Association for Research in Income and Wealth
The Disappearing Farm: What Declines in Farm Numbers Reveal About Agricultural Stress Abstract Across the U.S., farmers have been facing one of the largest suicide crisis to date. Due to compounding external and internal factors that influence both their mental health and farms. This study aims to examine the parallels between the 1980s Farm Crisis and the current mental health challenges among farmers in the U.S. today using online news to survey the emerging issue. Data was collected from both national and local news outlets published between April 2025 and October 2025, using target keywords related to the topics of farmer mental health, economic hardship, and agricultural policy. The USDA Census of Agricultural data (2017,2022) was also used to put into context the geographic patterns of these issues. Findings show that external stressors, financial stress/stability, and isolation, are strong determinants of farmer distress and are persistent and remain under-addressed. This study highlights the value of media-based surveillance in detecting emerging public health issues and underscores the urgent need for comprehensive support systems for U.S. farmers
By Bethel Amare
Presented by: Bethel Amare,
US-China Trade War: Modeling Subsidy and Tariff efficiencies on Trade Revenue Abstract Abstract This paper presents theories and specific models to illuminate the effect and efficiency of export subsidy and protective tariff policy on trade revenue. Escalating protectionism and trade wars between the US and China over the last three decades with overwhelming data provide insights for this paper (K. Keyimu 2011) to build theoretical models. The paper employs elasticity approaches to derive two specific models evaluating efficiency of export subsidy and tariff policy on trade revenue. The function and viability of the World Trade Organization (WTO) has been threatened by economic skirmishes between the US and China and resulting tit-for-tat trade wars. These result in ripple effects across the global trading system, and the US trade wars already went global. Consequently, the WTO has not been able to control market distortions as it was intended. Based on this fact, the paper examines ongoing trade wars and presents two models by which efficiencies of export subsidies and tariff policies are evaluated. A generic question such as “Do Trump tariffs always gain revenue?” opens a door for the research. The paper uses the partial differential of trade revenue with respect to changes of export subsidies and protective tariffs as retaliatory tools to examine who gains and who loses from tit-for-tat trade wars. The paper derives two functions, named as “kappa” and “lambada.” They are multivariable functional relationships of market forces which represent the price elasticity of demand and the price elasticity of supply. Therefore, the two functions can also be called “subsidy and tariff policy efficiency indicators.” Consequently, the research answers “How do commercial policies such as tariffs and subsidies affect revenue as a function of their respective elasticities?” Under guidance of this research question two significant findings are presented. Findings 1)Efficiency of Subsidies on Revenue: The research indicates that for inelastic products, an increase in subsidies leads to a decrease in trade revenue. Conversely, for elastic products, subsidies can positively affect trade revenue. 2)Efficiency of Tariffs on Revenue: It is found that for inelastic products, increasing tariffs results in higher trade revenue, while for elastic products, the opposite is true, leading to a decrease in tariff revenue.
By Keyimu Kalibinuer; University of Maryland Baltimore County
Presented by: Keyimu Kalibinuer, University of Maryland Baltimore County
Understanding the Racial Income Gap: Structural Barriers, Criminal History, and Discrimination Abstract This thesis examines the relationship between race and income and identifies the factors that help explain the observed wage gaps. Specifically, it tests whether racial differences in earnings persist after accounting for relevant demographic and socioeconomic variables such as level of education, geographic location (rural or urban), parents' highest level of education, and work experience level. The purpose of this analysis is to see how much of the relationship between race and income can be explained by the aforementioned control variables and how much of it is due to unmeasurable factors such as racial discrimination and individual skill set. The analysis will have a specific focus on the interaction between race and incarceration, assessing how criminal legal system involvement shapes economic opportunity. Furthermore, unlike most studies about the race-income gap, this analysis will utilize an interaction variable between race and incarceration to see if there is a compounding effect on income. This is a particularly important aspect to evaluate because it is commonly known that the criminal legal system disproportionately penalizes Black individuals compared to their white counterparts. Using multivariate regression models with logged wages as the dependent variable, the study incorporates race, incarceration, and a wide array of demographic controls associated with both income and race. Initial results indicate a statistically significant negative relationship between income and both race and incarceration; this thesis will further evaluate the extent to which these relationships are explained by observed demographic factors versus residual disparities consistent with systemic discrimination and if there is a “double jeopardy” effect between incarceration and race on annual income.
By Ameenah Habib; Georgetown University
Presented by: Ameenah Habib, Georgetown University
AI and industry productivity growth Abstract Industries are differentially exposed to opportunity or disruption from new AI technologies. We use measures of this AI-association (or alternatively AI-exposure, or AI-intensity) drawn from the literature on AI-exposure of occupations. We convert AI-exposure data on occupations to measures for industries using standard occupation-by-industry matrices. We test whether AI-association can account for an acceleration of productivity growth in industries from 2021 to 2024 relative to the prior decade. AI exposure does not guarantee adoption of new technology. We also look for evidence that the exposed industries invest in computer hardware, software, R&D, specialized staff, or purchasing services from data centers. Productivity growth can be associated with these investments, and they can also affect a total factor productivity residual. Systematic relationships between AI-association of an industry, and accelerated productivity growth in that industry indicates that some novel form of productive input (perhaps “AI capital”) is having an effect.
By Peter Meyer; U.S. Bureau of Labor Statistics    Sabrina Pabilonia; US Bureau of Labor Statistics
Presented by: Sabrina Pabilonia, US Bureau of Labor Statistics
Consequences of eviction for parenting and non-parenting college students Abstract Amidst rising and increasingly unaffordable rents, 7.6 million people are threatened with eviction each year across the United States—and eviction rates are twice as high for renters with children. One important and neglected population who may experience unique levels of housing insecurity is college students, especially given that one in five college students are parents. In this study, we link 11.9 million student records to eviction filings from housing courts, demographic characteristics reported in decennial census and survey data, incomes reported on tax returns by students and their parents, and dates of birth and death from the Social Security Administration. Parenting students are more likely than non-parenting students to identify as female (62.81% vs. 55.94%) and Black (19.66% vs. 14.30%), be over 30 years old (42.73% vs. 20.25%), and have parents with lower household incomes ($100,000 vs. $140,000). Parenting students threatened with eviction (i.e., had an eviction filed against them) are much more likely than non-threatened parenting students to identify as female (81.18% vs. 62.81%) and Black (56.84% vs. 19.66%). In models adjusted for individual and institutional characteristics, we find that being threatened with an eviction was significantly associated with reduced likelihood of degree completion, reduced post-enrollment income, reduced likelihood of being married post-enrollment, and increased post-enrollment mortality. Among parenting students, 38.38% (95% confidence interval (CI): 32.50-44.26%) of non-threatened students completed a bachelor's degree compared to just 15.36% (CI: 11.61-19.11%) of students threatened with eviction. Our findings highlight the long-term economic and health impacts of housing insecurity during college, especially for parenting students. Housing stability for parenting students may have substantial multigenerational benefits for economic mobility and population health.
By Nick Graetz; Princeton University    Adam Chapnik; Princeton University    Danielle Sandler; U.S. Census Bureau    Sonya Porter; US Census Bureau
Presented by: Danielle Sandler, U.S. Census Bureau
Pressure to Spend: Transportation Project Selection Under ARRA Abstract Federal policy outcomes often depend on the bureaucrats who implement them. This study examines the behavior of local transportation officials charged with spending a large federal stimulus that granted wide discretion but little time for decision-making. The American Recovery and Reinvestment Act of 2009 (ARRA) unexpectedly allocated \$7.3 billion in formula-based transit capital grants to localities. The funds required no local match and allowed broad project-selection freedom but had to be obligated within a year. I use a Large Language Model to process unstructured text descriptions of ARRA awards and extract multidimensional characteristics of 3,763 projects funded through 954 grants. For each project, I code the activity performed, the stated justification, and the anticipated effects. The analysis shows that local officials prioritized (i) bus fleet rehabilitation and replacement, (ii) operating and passenger facilities repair, and (iii) software and communication systems improvement. Rather than expanding service (e.g., adding new routes or buying additional buses), these choices focused on improving the quality of existing operations. Project justifications and predicted effects similarly emphasize reliability, comfort, and long-term system performance. Across multiple definitions, no more than 40\% of projects and funds were directed at service expansions. When given flexible capital funding under time pressure, local bureaucrats appear to prioritize deferred maintenance and operational bottlenecks rather than expanding service. This allocation is consistent with under-maintenance under normal funding arrangements and implies that large stimulus packages alone are unlikely to substantially increase transit provision in the US without first addressing these constraints.
By Arseniy Braslavskiy; University of Maryland
Presented by: Arseniy Braslavskiy, University of Maryland
Are we there yet? An AI roadmap using productivity data Abstract The use of artificial intelligence (AI) is on the rise across our economy. The Bureau of Labor Statistics’ Productivity Program has a wealth of information that can be used to infer the impact of AI across various productivity measures. We use a roadmap that traces AI across industries that produce AI software, house data, supply the equipment to create AI, and arrive at industries that consume AI in production. We examine total factor productivity, asset data including investment, and additional measures such as labor productivity and energy input to make the journey along the industries where artificial intelligence and its effects overlap. We can trace the flow of innovation from the capital assets like software and research and development in the industry where software is created, to the data processing industry, passing through manufacturing industries, with the help of services provided by other industries. Through this roadmap we gather evidence of where artificial intelligence starts, what variables to pay attention to, and what industries play a role.
By Lamae Maharaj; U.S. Bureau of Labor Statistics
Presented by: Lamae Maharaj, U.S. Bureau of Labor Statistics
Resilience in African Agricultural Systems: Adapting to a Changing Climate Abstract Global climate change presents a significant challenge to the world today. Since 1850, the combined temperature of land and ocean has been increasing at 0.06 ◦C on average per decade, with the rate of warming more than three times faster than that since 1982, at approximately 0.20 ◦C per decade (IPCC, 2023). Agriculture, which is the foundation of food security, is highly vulnerable to these changes in climate. In Africa, rising temperatures, altered precipitation patterns, and increased wind variability are disrupting crop growth cycles, intensifying extreme weather events, and influencing pest and disease prevalence (Lobell et al., 2011; Sultan & Gaetani, 2016). These dynamics threaten crop yields and quality, with significant implications for food security and rural livelihoods (FAO, 2021). This article synthesizes recent research on climate impacts on African agricultural systems, examines feedback mechanisms between farming practices and climate dynamics (Campbell et al., 2016), and outlines adaptive strategies to enhance resilience (Thornton et al., 2018). This paper analyzes how changing climate patterns affect crop yields, food security, and rural livelihoods in Sub-Saharan Africa, particularly for staple crops like maize, sorghum, and cassava.
By Yaya Sissoko; Indiana University of Pennsylvania    Brian Sloboda; University of Maryland, Global Campus
Presented by: Brian Sloboda, University of Maryland, Global Campus
Employment Recoveries and Hiring Costs in the Survey of Income and Program Participation Abstract Empirical evidence suggests that after a recession the rate of recovery in the labor market is proportional to the severity of unemployment that precedes it, irrespective of other aspects of the recession. This proportionality can only be partially explained by a standard Diamond, Mortensen, and Pissarides (DMP) matching models of the labor market. This suggests other feedback mechanisms linking high unemployment levels to lower hiring rates beyond congestion caused by the high number of job seekers relative to available work in a post-recession labor market. This paper develops and estimates DMP model that includes a flexible model of hiring costs to firms. The model is estimated using data from the Survey of Income and Program Participation (SIPP) that is extended to include industry level characteristics from the Job Opening and Labor Turnover Survey (JOLTS) and the Economic Census from 2000-2022. The inclusion of the JOLTS also allows for a comparison between hiring rates and other labor market variables between household and business level surveys. The performance of hiring cost models based on unemployment duration, firm recruitment and productivity, and potential match quality are compared on their ability to match employment growth measured in the SIPP following the 2001, 2008, and 2020 recessions in the United States.
By Patrick Burke; US Census Bureau
Presented by: Patrick Burke, US Census Bureau
Banks as Firms: The Macroeconomics of Financial Firm Dynamics Abstract This paper studies the firm dynamics of banks and their role in shaping aggregate and regional business cycles. I develop a framework in which banks are heterogeneous firms that endogenously enter and exit oligopolistic regional loan markets, competing on screening ability, funding costs, loan appeal, and regulatory costs. The model delivers sharp predictions linking loan rates, market shares, and entry decisions to bank fundamentals and local frictions. Guided by theory, I estimate bank-level fundamentals and region-level loan market frictions using a structurally identified dynamic state-space model. Screening ability emerges as the primary driver of bank growth and market share, while loan market frictions explain over half of aggregate loan spreads. Embedding the estimated bank dynamics in a quantitative multi-region New Keynesian DSGE model, I show that banking shocks account for roughly one-third of output fluctuations and generate substantial regional heterogeneity. Bank dynamics drive the uneven amplification and spatial propagation of shocks through novel credit elasticity and banking network channels, leading to the asymmetric regional transmission of national monetary policy.
By Tyler Pike; University of Maryland
Presented by: Tyler Pike, University of Maryland
Pay Tables and Productivity: Compressed Wages and Match Quality Abstract Many labor markets are characterized by heterogeneous workers competing for jobs which differ in difficulty or required skills. Comparative advantage theories argue wage differences across jobs act as signals for the supply of skilled labor in a competitive market, facilitating the sorting of specialized workers to more complex jobs. In the United States' public sector, however, federal and state laws often constrain the amount wages are allowed to differ by job. For example, members of the military receive the same basic pay across all job types, given equal rank and years of service. Counterintuitively, the empirical evidence of the quality of match between workers and jobs show public sector workers, on average, have a higher match quality than private sector workers--despite compressed wages. This paper examines the effects of wage compression on the evolution of match quality throughout a worker's career. It shows that wages in the private sector may be "noisy" measures of match quality for workers, leading to lower quality matches than in compressed wage environments.
By Kathryn McGinnis; Cornell University
Presented by: Kathryn McGinnis, Cornell University
Trust the police: The impact of a procedurally just policy on racial disparities and traffic stops Abstract Law enforcement agencies seeking to increase public trust may choose to incorporate tenets of procedural justice theory when making policy decisions. Allowing community members to feel heard and respected in the policy-making process by soliciting and incorporating their experiences and feedback can increase citizens’ feelings of trust and legitimacy and, in turn, improve police-community relations. In September 2022, in response to concerns raised by community groups, the Mecklenburg County (NC) Sheriff’s Office decided to end the use of regulatory traffic stops, or stops made for non-moving violations. The policy change was explicitly intended to decrease racial disparities and increase traffic safety. This study assesses the impact of that policy change on racial disparities among stopped drivers, measured by disparity index, the ratio of the proportion of Black stops to the proportion of the Black population, and disparity ratio, the ratio of the per capita stop rate for Black individuals to the per capita stop rate for white individuals; and on traffic safety, measured as the number of serious injury or fatal crashes. An interrupted time series analysis of traffic stops made by the sheriff’s office between September 2021 and September 2023 found an immediate, negative policy effect on racial disparities, but no evidence of a sustained effect, and no statistically significant effect on traffic safety. Despite the policy’s mixed success, a citizen-involved policy-making process still contributes to improved public perceptions of the agency.
By Moriah Sharpe; American University
Presented by: Moriah Sharpe, American University
Understanding Severe Psychological Distress Among Older Veterans Abstract Older veterans represent a population that experienced unique hardships such as difficult homecomings, very few mental health resources, and difficulty resuming civilian life. However, there is limited understanding surrounding severe psychological distress (SPD) among older veterans. What is the geographic distribution and patterns of severe psychological distress (SPD) among older veterans? The study uses data from the 2021 National Health Interview Survey and Census ACS to gather data about the regional and county variation of older veterans, compare K6 responses of veterans aged 65 and above and veterans aged 65 and above with severe psychological distress, use adjusted odds ratios to compare the descriptive characteristics of older veterans and older veterans with SPD, and compare the prevalence of SPD between younger and older veterans. My findings suggest that there are geographical and demographic patterns among older veterans with severe psychological distress. Policymakers should develop programs that target older veterans in the South and West regions and high-prevalence counties across the country, identify younger veterans at risk for mental health struggles, and address combat trauma and struggles with reintegration into civilian life.
By Tiffany Nguyen; University of Maryland, Baltimore County
Presented by: Tiffany Nguyen, University of Maryland, Baltimore County
The People Behind the Enterprises: Linking Employer Businesses, Owners and their Characteristics Using Administrative Records Abstract In response to declining response rates, starting in 2020 the Census Bureau began providing nonemployer demographics not through a survey, but a program that leverages existing administrative and census records to identify the business owner universe and their characteristics: the annual Nonemployer Statistics by Demographics series (NES-D). Following the successful transition from surveying nonemployers to the adrec-based NES-D, a related initiative for employers, known as the Employer Characteristics Project (EC), has been underway. While the methodology underlying EC is related to that of NES-D, employer businesses introduce novel challenges due to their more complex organizational structures. In this presentation we will cover our methodology, challenges, and comparisons of administrative record based demographics to those from surveys. Beyond unburdening surveys and lowering costs, we expect the Employer Characteristics data to enable substantial advances in business research. The linkage between businesses and business owners will allow researchers to examine the outcomes of owners of exiting businesses, the role of owner characteristic in firm ex ante heterogeneity, and more.
By Vitaliy Novik; Census Bureau
Presented by: Vitaliy Novik, Census Bureau
Short Run Trade Adjustment to a Temporary 145 Percent U.S. Tariff on China Abstract How quickly can extreme tariffs reduce targeted imports and induce short-run sourcing away from a systemically important supplier? This paper studies the 2025 episode in which the U.S. temporarily raised tariffs on the majority of Chinese imports to about 145 percent, before a ninety day truce following the Geneva Accord reverted the schedule to roughly 30 percent. Near-prohibitive rates paired with an announced rollback generate a sharp, time-bounded shock, enabling estimation of adjustment and persistence within months. I use monthly U.S. import data at the HS10 level and the U.S. Customs and Border Protection product coverage list to build an HS10–country–month panel from January 2023 to August 2025, which contains a total of 7.62 million observations. Identification uses a triple-difference design comparing (i) covered versus uncovered HS10 codes, (ii) China versus a broad set of low-tariff non-China exporters not subject to the China-specific escalation, and (iii) the initial truce period June–August 2025 relative to the pre-period. Specifications include product×month and product×origin fixed effects, and the estimates capture short-run reduced-form persistence that appears in the post-roll back months. The first contribution is product-level evidence on intensive-margin adjustment when tariffs become extreme yet are widely understood to be temporary. In the truce months, treated Chinese shipments contract sharply but do not collapse. Import quantities fall by about 26.7 percent and import values fall by about 27.8 percent, while CIF unit values decline by only about 1.7 percent on average. Unit values reflect composition and quality mix as well as pricing, but this pattern is consistent with limited exporter price compression and with most near-term adjustment operating through volumes rather than large unit-value changes. Dynamic event-study estimates show stable pre-period dynamics, a clear break at the escalation, and an incomplete rebound that leaves treated Chinese flows below their earlier trajectory by August 2025. The second contribution quantifies how much extensive-margin reallocation is feasible over the same short horizon. Focusing on products where total U.S. imports are relatively stable, covered HS10 codes experience a clear increase in supplier variety relative to uncovered codes: the number of active non-China origin countries rises by about 3.1, from a pre-period mean of roughly 12. However, this diversification is not broad-based. Month-to-month entry and exit do not surge, and origin-level activity regressions show no uniform rise in the likelihood that a given non-China country begins supplying a covered HS10 product. Instead, variety expands through a limited set of plausible product–origin matches, with sourcing shifts concentrating in established supplier hubs rather than diffusing across the full set of potential exporters. The paper shows that an extreme but temporary tariff can quickly reduce direct exposure to a targeted supplier, and that depressed covered-China flows persist into the first months after the rollback, while rapid, broad-based diversification remains limited in the short run and operates mainly through established alternative supplier relationships rather than economy-wide reshoring or a broad wave of new exporter entry.
By Tianhao Wu; Harvard University
Presented by: Tianhao Wu, Harvard University
Treatment Delays for Black and White Heroin Users Abstract Heroin use has significant adverse effects on individual users, their families, and society in the form of lost productivity, excessive health care spending, increased criminal activity, and premature deaths due to overdose. Treatment can mitigate such costs, yet heroin users are known to exhibit substantial delays between first use of the drug and first treatment. Roughly half of the more than 460,000 heroin users examined in this study had delays in excess of ten years and the Black-White difference in average treatment delays is even larger. Treatment delays form a continuum, ranging from very little delay (months) to many years. Where on that continuum do Black and White heroin users fall and what economic and demographic factors affect those outcomes? More importantly, are there any levers for policymakers to change those outcomes? This study provides answers to those questions through a rigorous empirical analysis of a large national database of heroin users who entered formal treatment. A generalized ordered logit model of treatment delay is estimated that leads to prediction probabilities of Black-White locations on the treatment-delay continuum. A detailed postestimation predictive analysis shows that if rules/regulations were changed to expand the number of programs that accept Medicaid payments, treatment delays could be shortened, more so for Blacks than Whites.
By James Duggan
Presented by: James Duggan,
Determinants of Abortion Intention Among Women Aged 17-27 in Cameroon Abstract This study investigates the determinants of abortion intention among women aged 17-27 in Cameroon, using data from a survey conducted between 2022 and 2024 across all ten regions. The final sample comprises 4,730 valid responses collected via online and in-person questionnaires. We constructed an abortion intention score by aggregating normalized items into a continuous measure (0-1), later dichotomized for econometric modeling. Guided by Ajzen’s Theory of Planned Behaviour and Becker’s economic framework, we estimated multilevel logistic models to capture individual and contextual effects. Results show that religious affiliation and perceptions significantly deter abortion intention, whereas financial difficulty, poverty, and education cost pressure increase it. Social pressures-including fear of stigma and potential family rejection-are strong positive correlates, and age slightly reduces intention. Regional heterogeneity is notable (ICC 0.16-0.18). Impact analyses reveal robust effects of social pressure (EB 0.17; PSM 0.21) and sizable, method sensitive effects of financial constraint (EB 0.08; PSM 0.48). The study concludes that abortion intention reflects an interplay of moral norms, economic precarity, and social sanctions, suggesting that targeted financial support, comprehensive school based sexual education, and anti stigma initiatives could reduce unsafe outcomes.
By Aristide Merlin Ngono; Université de Dschang
Presented by: Aristide Merlin Ngono, Université de Dschang

 

This program was last updated on 2026-04-14 15:03:21 EDT