Authors: Sergio Pinto, Alaa Abdelfattah, Marshall Steinbaum, Justin C Wiltshire

Abstract: We leverage consumer credit panel data covering 2011–2019 to examine how hospital consolidation impacts individual medical debt. In our preferred event study specifications, exposure to hospital mergers causes significant increases in the likelihood of having medical debt in collections, the flow of new debt generated over the previous twelve months, and the total outstanding stock. These effects amplify over time: five years after treatment, the effects on our outcomes of interest range from 11% to 20%. We also find evidence that additional balances are carried on credit cards. In a placebo test, we find no effect of merger exposure on analogous outcomes for non-medical debt in collections. Using comprehensive county-level healthcare spending and utilization data from the Institute for Health Metrics and Evaluation’s Disease Expenditure Project (ultimately sourced to insurance claims), we investigate the mechanisms underlying these effects. We identify three complementary channels: hospital mergers raise the price of commercially insured care, expand the set of healthcare encounters for which patients bear out-of-pocket financial responsibility, and shift the local payer mix away from Medicare and toward commercial insurance. Together, these channels increase both the cost and the breadth of patients’ financial exposure to healthcare.

Seminar Notes

Venue

UC Davis Economics Alumni Conference 2025

Objective

To understand if hospital mergers affect the medical debt of individual in the same commuting zone

Importance

Hospitals often claim M&As are good for patients and employees.

Literature disagrees - higher prices, lower wages/employment

This paper looks at medical debt

Background

A lot of increase in concentration (HHI) in hospitals between 2001-2019

Medical debt began falling after 2014 ACA expansions

Data & Key Variables

UC Consumer Credit Panel - 2% nationally representative, 100% Californians

Hospital mergers - American Hospital Association/Zach Cooper. 2011-2019. Change in HHI.

138 treated CZs with a merger that increases HHI

Methodology

Event study/Diff n Diff Sun & Abraham estimator

Results

Increase in share of population with medical debt and an increase in the size of medical debt balance with a concentration-increasing merger

Key Table/Figure

Authors: Richard De Thorpe

Abstract: How has remote work reshaped residential sorting and housing demand, and what are the implications for state and local governments? To estimate causal effects, I propose a novel instrument for remote work that exploits quasi-random variation in the timing and size of office lease expirations, captured through a Bartik-style exposure measure at the residential block level. Expirations allow tenant firms to reduce office space and switch employees to remote work, generating strong first-stage effects. Remote work causes modest increases in housing and property tax expenditures in exchange for space, homeownership, and public schools, but not other neighborhood characteristics. It significantly increases migration, particularly out of cities and states that levy income taxes. At the neighborhood level, higher 2020 remote work shares cause subsequent residential turnover, demographic clustering, and property tax revenue windfalls. Taken together, the results indicate that remote work induces migration consistent with Tiebout sorting, and accounts for 10% of migration since 2020. Residential choices and tax bases now depend less on employment proximity and more on affordability and tax-benefit linkage.

Reading Notes

Objective

To understand the effects of remote work on residential sorting and how that affects housing demand and local public finance

Importance

Difficult to measure effect of remote work on residential choice - self-selection, reverse causality

Background

Commuting costs have traditionally anchored households to workplaces

Office leases typically last 3-15 years and carry significant costs for early termination

Data & Key Variables

ACS 2005-2023 - Remote work status, migration, residential location

Public use LODES 2004-2022 - Residential-to-Work block

CompStak - 1.2 million commercial leases, 673,000 office leases. 20% of U.S. office real-estate stock. Used to measure annual lease expirations at the block level

Methodology

Bartik-style instrument using square footage of leases expiring in work blocks employing residents of t-1 residence block

Instrumented dynamic diff-in-diff with neighborhoods with above average shares of remote work in 2020 as the treatment group and neighborhoods with below average shares as controls.

Results

Remote work likelihood, using the office lease expiration of local work locations as an instrument, increases migration by 11 percentage points.

It also decreases the probability of living in jurisdictions with income tax, though increases property tax. It also decreases private school enrollment and increases public school enrollment.

Raises houses expenses modestly, increases space, and decreases likelihood of renting

Key Table/Figure

Authors: Karen Benjamin Guzzo, Gabrielle Juteau, Anna Belykh, Wendy D. Manning

Citation: Guzzo, Karen Benjamin and Juteau, Gabrielle and Belykh, Anna and Manning, Wendy D. (2026). The Reliability of Repeated Measures of Retrospective Parity and Birth Dates in U.S. Survey Data: A Research Note. Demography, 12772821.

Abstract: In a longitudinal survey, most people consistently report parity and birth date information. One in 10 respondents, and one in five parents, report parity inconsistently across waves. Among those who reported children at an earlier wave, 11% report different dates of first birth. Inconsistency in parity and first birth dates was higher among males and less advantaged people. For demographers interested in the context of childbearing—and the prevalence, predictors, and consequences of childbearing in particular circumstances—longitudinal surveys collecting retrospective fertility histories are the primary source of data. Despite earlier work on the subject, the reliability of such data remains unclear, presenting a potential threat for drawing conclusions. Using Waves IV and V of the National Longitudinal Study of Adolescent to Adult Health (Add Health), supplemented by a smaller, population-based dataset for comparison, we investigate interwave consistency of individuals’ retrospective reports of parity (benchmarked to Wave IV) and date of first birth. In Add Health, about 10% of all respondents, and about 20% of parents, report inconsistently about the number of children born as of Wave IV, depending on whether the information is drawn from Wave IV or V. Among those with children born prior to Wave IV, about 11% report different dates of first birth; inconsistency heaps around ±1 month and ±12 months. Inconsistency in reporting parity and birth dates is more common among males, less socioeconomically advantaged individuals, and racialized minorities. The findings suggest that existing data collection approaches do not perform equally well across subpopulations, with implications for the study of key demographic events.

Reading Notes

Objective

To understand the inter-wave consistency of individuals’ reports of parity (number of children) and date of first birth

Importance

A non-trivial share of survey respondents reporting fertility information inconsistently is problematic for research that relies on survey reported fertility outcomes

Background

Under reporting fertility is a known problem in surveys and reporting full fertility histories at multiple points in a longitudinal survey is potentially a solution

Data & Key Variables

Add Health - 20,745 individuals who were in grades 7-12 in the 1994-1995 school year.

Full birth histories were collected in Waves III-V, though the question wording in III was confusing, so paper compares waves IV and V

Measure parity as of Wave IV using data on birthdates from both waves.

Waves IV (2008) and V (2016-2018) were around 9 years apart

Methodology

Consistency measure: 1) parity in Wave IV is consistent with parity in Wave V, 2) parity is lower in Wave IV, 3) parity is higher in Wave IV

First birth consistency measured as exact date, and difference in months

Demographic variables from Wave V: age, sex, race/ethnicity, education, household income, employment status, Economic hardship = history of eviction or falling behind on bills, incarceration history, relationship stability/number of partners, relationship status, child death

Results

Around 1 in 10 respondents have inconsistent parity across waves, with equal shares “finding” or “losing” a birth. This share is higher, 1 in 5, for those with at least one child.

Of those with children, 1 in 10 report a different date of birth of their first born child between waves.

Inconsistency is higher among men than women, among non-white respondents compared with white respondents, and among the economically disadvantaged

Key Table/Figure

Authors: Aria Golestani, Taylor Mackay, Matthew B Ross

Abstract: Affordable housing development is a central policy tool for addressing housing insecurity. Local opposition to new projects often centers on concerns about public safety. We study the effects of new affordable housing on nearby crime in Los Angeles from 2005 to 2023 using administrative records on housing-site openings and permits linked to geocoded public safety data. Exploiting temporal and geographic variation in site openings, we find that any pooled effects are highly localized: 911 calls increase by roughly 20 percent in the immediate vicinity of new sites, but these effects attenuate quickly with distance and become difficult to detect at broader spatial scales. These average effects, however, mask substantial heterogeneity across housing types. Supportive housing developments are associated with much larger and more persistent increases in nearby crime, with the strongest effects concentrated near project sites but extending over a broader surrounding areas. By contrast, effects for non-supportive affordable housing are modest and generally similar to those of market-rate multifamily housing. The results suggest that the local public safety effects of affordable housing are not homogeneous across project types and concerns about public safety are most salient for supportive housing developments. These findings provide new and nuanced evidence that informs a contentious policy debate in which local public safety concerns have often constrained efforts to expand affordable housing and where those concerns are not typically incorporated into siting decisions.

Seminar Notes

Objective

To examine the effect of affordable housing construction on public safety in Los Angeles

Importance

Opposition to affordable housing usually cites potential externalities: traffic, parking, crime/public safety

This paper adds to the prior work by examining very local outcomes

Background

Median home prices in the US have more than doubled in the past 20 years. 50% of households are cost burdened

Affordable housing = housing that is built (or rehabbed) for the purpose of renting at below-market rates to lower-income renters

Data & Key Variables

Affordable housing site openings 2005-2023 - City of LA Housing Department (LAHD)

Local public funding - address, site name, # units, year occupancy began

Geolocated, incident-level data on reported crimes, arrests, and 911 calls from LAPD

311 calls, traffic collisions, and home sales

Occupancy start dates: Property HHH supportive housing loan records, permits issued by LA Department of Building & Safety

Methodology

Event study using staggered timing of housing site openings. Borusyak et al (2024) imputation approach

Treated = within 0.25 files of housing site

Donut control between 0.25-0.5 mile

Results

Localized increase in property crimes of 6% over 5 years post site opening.

No effect in broader treated area, no change in violent crime or other outcomes

Reported crimes and 911 calls increase, arrests do not change

Key Table/Figure

Authors: Mark Edward Votruba, Jeffrey R. Kling

Citation: Votruba, Mark Edward and Kling, Jeffrey R. (2009). Effects of neighborhood characteristics on the mortality of black male youth: Evidence from Gautreaux, Chicago. Social Science & Medicine, 68(5), 814-823.

Abstract: We estimate the effect of neighborhood characteristics on the mortality of poor black male youth in families relocated through the Gautreaux Assisted Housing Program, a residential mobility program implemented in Chicago, USA in 1976. Within our sample (N=2850), 52 post-placement deaths were observed, the majority of which (30) were homicides. All-cause and homicide mortality rates were substantially lower among those relocating to Census tracts with higher fractions of residents with college degrees, which suggests that relocating to more-advantaged neighborhoods can ameliorate the mortality risks faced by this population. The estimated effect declines over the post-placement period, a result consistent with evidence that Gautreaux families routinely relocated following their initial placement. A causal interpretation of these findings is undermined somewhat by evidence of neighborhood selection, though the mortality effect estimate is very robust to inclusion of covariates predictive of placement tract characteristics. Mortality effect estimates relating to Census tract measures of socioeconomic deprivation other than education were weaker in magnitude and generally insignificant, suggesting that neighborhood levels of human capital more strongly affect the mortality risks faced by this population than racial composition or neighborhood poverty.

Reading Notes

Objective

To estimate the effect of neighborhood characteristics on mortality using variation from the Gautreaux Assisted Housing Program

Background

The Gautreaux program operated from 1976 to 1998. Much of the literature is plagued by attrition bias.

From 1976-1980 the consent decree required HUD to move families to neighborhoods with <30% Black. This was relaxed in 1981.

The Gautreaux program was administered by the Leadership Council for Metropolitan Open Communities.

Families started choosing own units in 1990, but were assigned randomly before that

Data & Key Variables

Gautreaux participant data from HUD 1976-1994. Date of intake and placement, intake address and placement address, identifying information

Tract-level characteristics from 1990 Census

National Death Index & State Vital Statistics

Methodology

Cox proportional hazard model

Each Cox regression includes a single placement neighborhood characteristic as covariate of interest

Results

Moving to a more-advantaged neighborhood decreases the mortality risk faced by black youth, particularly the risk of homicide

Percent of residents with a college degree in tract is most important predictor of reduced mortality

Key Table/Figure

Authors: Linh T Tô

Abstract: This paper evaluates the hypothesis that, in setting wages, firms respond to costly signals by workers when such costs are informative of their values to the firms. For workers who become mothers, uncertainty about their future values can influence firms’ decisions to distribute career and promotion opportunities. Consequently, workers may forgo paid parental leave even when there is no human capital depreciation associated with taking leave. I build a signaling model with a continuous choice of leave period when such choice is restricted due to the maximum allowed paid leave duration. Using administrative data from Denmark and a parental leave policy extending the maximum allowed duration of parental leave, I show how a leave extension affects wages, hours, and promotion opportunities for workers whose signaling ability changes with the extension. In contrast to human capital theory, an individual can take longer leave but gain in wages when the larger choice set allows more workers to signal their type. The paper provides evidence of the labor market consequences of parental leave-taking due to signaling and the importance of asymmetric information in shaping parental leave choice.

Reading Notes

Objective

To estimate the role taking parental leave has in signaling productivity to the employer

Importance

Parental leave taking may serve as a signal of worker productivity to the employer.

Denmark has a long history of parental leave, with several changes in duration, and also public provision of high-quality childcare

Background

1984 reform increased parental leave from 14 weeks to 24 weeks.

In 1994 an additional 52 weeks of child care leave was added, which could be taken until child is 9.

In 2002 child care leave was discontinued and 32 weeks of sharable parental leave was added

Data & Key Variables

Danish Social Statistics Register (SHSS) - days taken for parental leave between 1984-2007

DREAM database provides weekly parental and childcare leave 1992-2012.

IDA provides information about employment in November of each year 1980-2012

Hours worked data available back to 1964

Methodology

Use parental leave reforms in 1984 and 2002 to test implications of signaling model by comparing women who pool at the maximum leave duration before and after reform.

Match mothers across treatments by pre-birth characteristics to estimate counterfactual leave taking

Results

Women who take longer leaves experience greater reductions in labor supply and hourly wages conditional on working.

Even women for whom the parental leave duration was not expected to be a binding constraint take more leave after reform.

Key Table/Figure

Authors: Steve Coffey, Gabriel Ehrlich, John C. Haltiwanger, Ron S. Jarmin, David Johnson, Tod Johnson, Matthew D. Shapiro

Abstract: This paper presents early results from the RESET Demonstration Project, an initiative that constructs price and sales indices using item-level transactions data. The project’s overarching goal is to demonstrate the feasibility and advantages of leveraging modern transactions data for economic measurement by pointing toward a future in which official economic statistics are rebuilt on 21st-century data and technology. The paper’s results introduce new inflation measures for Food at Home and evaluates them on two fronts: how they compare to current official statistics and what they offer as novel measures that capture more accurately the dynamics of consumer choice. The paper also addresses the challenges of producing reliable and credible official statistics from item-level transactions data and reports findings on recent Food at Home inflation trends.

Seminar Notes

Venue

NBER CRIW SI 2026

Objective

To use transactions-level data to measure prices and quantities in order to measure inflation and GDP

Importance

Ultimately want to try to replace traditional measures (surveys, enumeration) with national and sub-national indices of prices and quantities from transactions data

Background

Circana has near-universe of transactions for 2/3 of consumer goods. Monthly releases. As timely as current sources.

This paper focuses on prices of food

Compare to CPI - items selected periodically, sales-weighted probability sample. Weights lagged and from separate source

Data & Key Variables

Circana data - January 2020-November 2025

Food at home - item (UPC/SKU) by week CPI data for same period/category

Methodology

Calculate Price Bilateral Geometric Indices - rolling 12 months weights.

Trailing Share Laspeyeres two stage

Decomposition of differences between measured prices and weights

Results

Series follows the CPI food at home prices and inflation rates well

Adjusting for chain drift important

Some gaps in 2020/2021 and 2025 - RESET series lower.

Differences being driven more by measured prices than the weights. Currently digging into specific products such as eggs and tomatoes. Substitution effects

Key Table/Figure

Authors: John Van Reenen, Robin Burgess, Ignacio Banares-Sanchez, Pol Simpson, Yifan Wang, David Laszlo

Abstract: Do industrial policies that promote clean energy offer a “ray of hope”, increasing a country’s growth and welfare, whilst simultaneously reducing carbon emissions? We study the impact of Chinese solar subsidies whose implementation by city-regions went alongside massive expansion of the sector and a dramatic fall in global solar prices. We construct new city and firm panel data on solar policies, patenting and output. Using synthetic-difference-in-differences 2004-2020, we find production and innovation subsidies were more effective than demand-side (installation) subsidies in generating large and persistent increases in local innovation, net entry, production and exports. Demand policies did, however, reduce local pollution. To examine aggregate effects, we build and structurally estimate a quantitative spatial model with endogenous innovation and heterogeneous productivity across firms and cities, which accounts for business stealing and knowledge spillovers. Counterfactual analysis shows that: (i) local effects remain substantial at the macro level explaining 40%-50% of the aggregate changes in solar innovation, prices and revenues; (ii) social benefits to Chinese citizens exceed subsidy costs by 65% (and double this when environmental benefits are included); and (iii) although all subsidy types increase welfare, innovation subsidies are the most cost-effective.

Seminar Notes

Venue

NBER CRIW SI 2026

Objective

To understand how important Chinese policy to support solar was for the rise of the solar industry and whether those policies improved welfare for Chinese citizens

Importance

73% of global greenhouse gases are from the energy sector. Decarbonizing requires addressing energy sector use of fossil fuels

Background

Huge fall in cost per watt of energy in solar power. Now around $0.20 per watt - competitive with fossil fuels

China moved from almost no share of solar PV in 2000 to 80% today

Data & Key Variables

Policy text from PKULaw Database - solar subsidies with implementation date by type: 1) production, 2) innovation, 3) demand/installation

Microdata on universe of solar panel manufactures in China

Methodology

Synthetic diff-in-diff using staggered introduction of policies by city

Model - heterogeneous manufacturing firms under monopolistic competition. Firm decisions: entry, exit, production, and technology upgrading. Grid planner builds clean (solar) or dirty (coal) power plants

Results

Innovation and production subsidies generate large city-wide increases in solar industry.

Weaker impacts from demand/installation subsidies, but these have a large impact on pollution reduction

Cross-city effects dominated by positive spillovers in innovation and learning

Net gain for Chinese citizens of $1.65 per $1 of subsidy (and larger if social cost of carbon included)

Key Table/Figure

Authors: Renato Faccini, Seho Kim, Javier Miranda

Abstract: We develop a theory in which a lower cost of job loss reduces the compensating wage premium workers require to join risky young firms. By lowering labor costs relative to safer firms, this encourages entrants to choose high upside experimentation, raising aggregate productivity. Using Danish matched employer–employee data and regional labor-market variation, we show that higher job-finding rates are associated with lower wage differentials between experimenting and non-experimenting young firms, both across firms and within firms hiring across multiple areas. A randomized survey experiment supports the worker-side mechanism: worsening unemployment prospects increases the wage premium workers require to accept employment at higher-failure-risk young firms.

Seminar Notes

Venue

NBER CRIW SI 2026

Objective

To understand why some economies produce more breakthrough innovations by exploring worker-side risk

Importance

Workers decide whether to join a young experimental firm or a safe firm. Experimental firm may fail. Choice depends on cost of unemployment.

Experimentation is not just about brave entrepreneurs, but about employees willing to risk working for those entrepreneurs

Background

High-growth young firms disproportionately contribute to employment and productivity group, but also volatile

Data & Key Variables

Danish employer-employee matched data

Variation in job finding rate across Danish commuting zones

Methodology

Calibrated model - workers are not just inputs, but risk sharing partners. Shared risk is priced in to wages

Use variation in job finding rate across commuting zones to estimate wage premia for young firms by unemployment risk

Vignette experiment - elicit wage premia necessary to take job at riskier firm

Results

Calibrated model - increases in job finding rate reduce the wage premia risk firms must pay. Increase innovation

Experimental firms in commuting zones with higher job finding rates pay lower premiums

Vignette experiment - If you double the expected unemployment duration, workers demand a 2.03% higher wage

Key Table/Figure

Authors: Seula Kim, Michael A Navarrete

Abstract: We study spatial heterogeneity in inflation across regions with different income levels and the role of retailer market structure. Using NielsenIQ Retail Scanner data, we document that, from 2006 to 2020, poorer MSAs experienced annual food inflation 0.46 percentage points higher than richer MSAs, implying an 8.8 percentage point cumulative gap. Poorer areas also have fewer products, fewer retailers, and higher market concentration. To identify causal effects, we exploit the 2014–2015 avian influenza outbreak and show that higher local retail concentration amplifies the pass-through of cost shocks. We develop a heterogeneous-firm model with customer capital accumulation that rationalizes these patterns.

Seminar Notes

Venue

NBER CRIW SI 2026

Objective

To understand whether inflation varies over geography/income and how that relates to market structure

Importance

Inflation typically measured at national level, which may mask important geographical heterogeneity

Most salient for food inflation - high expenditure share for low-income households

Background

Consumption baskets vary across high and low income households

Bird flu outbreak - late 2014, substantial price spike in 2015

Data & Key Variables

NielsenIQ data 2006Q1-2020Q3

Weekly UPC-level data on retail prices, quantities, store locations. 100+ retail chains, 400,000+ stores

21 PCE food categories at MSA level

Large retailers = top decile by national sales

Methodology

Price index - geometric Laspeyers index. Lagged expenditure share as weights

Model: Oligopolistic competition and customer base accumulation

Triple diff in diff using bird flu as exogenous cost shock to retailers. Variation based on MSA exposure to price shock, interacted with market concentration

Results

Food inflation rates vary across regions. Larger share of large retailers and more market concentration in poorer regions.

Food prices have been growing faster in poorer areas - national index is closest to richest decile, so does not describe food inflation for poorer regions well

Market concentration correlated with higher food inflation.

More pass through of bird flu price shock in regions with more market concentration.

Key Table/Figure

Authors: Bruce D. Meyer, Dan T. Rosenbaum

Citation: Meyer, Bruce D. and Rosenbaum, Dan T. (2001). Welfare, the Earned Income Tax Credit, and the Labor Supply of Single Mothers. The Quarterly Journal of Economics, 116(3), 1063-1114.

Abstract: During 1984–1996, welfare and tax policy were changed to encourage work by single mothers. The Earned Income Tax Credit was expanded, welfare benefits were cut, welfare time limits were added, and welfare cases were terminated. Medicaid for the working poor was expanded, as were training programs and child care. During this same time period there were unprecedented increases in the employment and hours of single mothers. We show that a large share of the increase in work by single mothers can be attributed to the EITC and other tax changes, with smaller shares for welfare benefit cuts, welfare waivers, training programs and child care programs.

Reading Notes

Objective

To explore whether or not tax and transfer programs increased the incentives for single mothers to work

Importance

The changes in these programs are plausibly exogenous and can be used to calculate welfare effects of tax and transfer programs, policy changes make understand cash assistance programs more important, and there is little work on EITC, Medicaid, etc for single mothers

Data & Key Variables

CPS ORG files and March CPS for 1984-1996 - children/no children, income taxes and EITC, AFDC and food stamps, Medicaid, Waivers, Training/Childcare, Labor market participation

Methodology

Diff in diff comparing women without children to women with children

Also structurally-based probit model that does not depend on women without children as a comparison group

Results

EITC and other tax changes account for 60%+ of weekly and annual employment of single mothers relative to single women without children

Key Table/Figure

Authors: Ann Bartel, Maya Rossin-Slater, Christopher Ruhm, Meredith Slopen, Jane Waldfogel

Citation: Bartel, Ann and Rossin-Slater, Maya and Ruhm, Christopher and Slopen, Meredith and Waldfogel, Jane (2025). The impact of paid family leave on employers: evidence from New York. Community, Work & Family, 28(2), 153-171.

Abstract: To study the impacts of New York’s 2018 Paid Family Leave (PFL) policy on employer outcomes, we designed and fielded a survey of small firms in New York and a control state, Pennsylvania, which does not have a PFL policy. We match each NY firm to a comparable PA firm and use difference-in-differences models to analyze within-match-pair changes in outcomes. Contrary to common concerns about the burdens of PFL on employers, we find no evidence that PFL had any adverse impacts on employer ratings of employee performance or their ease of handling long employee absences. Instead, we find suggestive evidence of an improvement in employers’ ratings of employee commitment and cooperation, concentrated in the first policy year. We also observe an increase in employers’ ratings of the ease of handling employee absences in the first policy year. Lastly, we find a rise in the incidence of employee leave-taking in the second policy year, driven by the smallest firms in our study.

Reading Notes

Objective

To understand the impacts of a paid family leave policy on employers.

Importance

One of the key arguments against paid family leave policies in the U.S. is its cost to employers, but we have little evidence about those costs

Background

New York’s Family Leave Policy took effect in January 2018.

Workers could claim up to 8 weeks of leave with 50% wage replacement rate (with max set at fraction of state average wage).

Amount of leave and wage replacement increased over time

Data & Key Variables

Survey to firms with 10-99 employees, fielded 2016-2019 in New York and Pennsylvania.

Sample of 4,573 firms participating in at least one year

Methodology

Difference-in-differences with matched New York and Pennsylvania firms.

Matched by industry sector, metro/non-metro, firm size, county unemployment rate, and county average wage

Results

Employee performance is consistently high and shows no post-PFL change.

But ease of dealing with long absences increases, particularly in the first year after implementation.

Rise in leave taking in 2nd year after policy change

Key Table/Figure

Authors: Ben Sprung-Keyser, Nathaniel Hendren, Sonya R Porter

Abstract: We examine the geographic incidence of local labor market growth across locations of childhood residence. We ask: when wages grow in a given US labor market, do the benefits flow to individuals growing up in nearby or distant locations? We begin by constructing new statistics on migration rates across labor markets between childhood and young adulthood. This migration matrix shows 80% of young adults migrate less than 100 miles from where they grew up. 90% migrate less than 500 miles. Migration distances are shorter for Black and Hispanic individuals and for those from low-income families. These migration patterns provide information on the first order geographic incidence of local wage growth. Next, we explore the responsiveness of location choices to economic shocks. Using geographic variation induced by the recovery from the Great Recession, we estimate the elasticity of migration with respect to increases in local labor market wage growth. We develop and implement a novel test for validating whether our identifying wage variation is driven by changes in labor market opportunities rather than changes in worker composition due to sorting. We find that higher wages lead to increased in-migration, decreased out-migration and a partial capitalization of wage increases into local prices. Our results imply that for a 2 rank point increase in annual wages (approximately $1600) in a given commuting zone (CZ), approximately 99% of wage gains flow to those who would have resided in the CZ in the absence of the wage change. The geographically concentrated nature of most migration and the small magnitude of these migration elasticities suggest that the incidence of labor market conditions across childhood residences is highly local. For many individuals, the “radius of economic opportunity” is quite narrow.

Reading Notes

Objective

To understand who benefits from local economic growth by constructing new statistics on migration rates across labor markets between childhood and young adulthood

Importance

Builds on previous literature by focusing on young adult migration.

Develops a new test of exogeneity for a given source of wage variation using geographic variation in wages

Background

Application of new source of wage variation: Heterogeneous recovery from the Great Recession. Comparing those who were young adults in 2010 versus those in 2017.

Leave one out at the origin (age 16) location level.

Data & Key Variables

U.S. born children in 1978-1992 birth cohorts (primary analysis 1984-1992)

Numident linked to parent 1040 records at age 14-18 (targeting age 16)

Residential History File (IRS 1040, W2, 1099 + HUD) for location at age 26

Race and education from Decennial Census & ACS

Income at young adulthood from W2 (wage income) and 1040 (family income)

Methodology

Using an overidentification condition from migration matrix. An increase in wages in location d will affect wages of people in location o in proportion to the probability of moving from o to d.

Compare observed wages relative to predicted wages where predicted wages are based on a weighted average of pre-period probability of migrating to a given destination.

Results

28% live in the same tract at 26 that they lived in at age 16. 55% live within 10 miles, 77% live within 100 miles, 90% live within 500 miles

Migration rates are shorter for Black (average 133 miles vs 194 miles for whites) and Hispanic youth and low parental income households

People are differently exposed to changes in local labor market conditions based on where they grew up and their race and income

Key Table/Figure

Authors: Catalina Amuedo-Dorantes, José Bucheli

Abstract: Access to the labor market is crucial for the economic integration of asylum seekers. This study estimates the causal impact of work authorization timing on the labor market outcomes of likely asylum seekers. We link USCIS administrative records with American Community Survey microdata and use congestion-driven variation in processing times within an instrumental variable framework. Faster authorization boosts labor force participation, employment, and earnings, while effects on hours worked and occupational choices are modest. The impacts are concentrated within the first decade after arrival and diminish over time, indicating that processing delays slow integration but do not permanently hinder it.

Seminar Notes

Venue

ACLEC 2025

Objective

To understand the effect of work permit processing time on the labor market outcomes of asylum seekers

Importance

Rising asylum claims, particularly defensive asylum claims.

Work authorization is critical for integration, but asylum seekers face delays in ability to apply for work permits. Minimum 150 + 180 days. But due to backlog, can be more than 12 months.

Background

USCIS randomly routes work permits to 5 processing centers when local processing center is busy

Data & Key Variables

EAD application data 2012-2021 Gender, PUMA, Determination year, origin-level processing time

ACS 2012-2021 data on non-citizens likely to be asylum-seekers

Methodology

Use variation in origin, year, US location in EAD approval time

Outcomes: labor force participation, annual income, hours, occupation

Instrument with work permits from non-asylum-seekers - a measure of system congestion

Results

One additional week of waiting decreases employment by 2.8 percentage points and annual income by $2,300

Key Table/Figure

Authors: Christian Zimmermann, Xiaodong Liu, Michael D. König, Chih-Sheng Hsieh

Abstract: We study the impact of research collaborations in coauthorship networks on research output and how optimal funding can maximize it. Through the links in the collaboration network, researchers create spillovers not only to their direct coauthors but also to researchers indirectly linked to them. We characterize the equilibrium when agents collaborate in multiple and possibly overlapping projects. We bring our model to the data by analyzing the coauthorship network of economists registered in the RePEc Author Service. We rank the authors and research institutions according to their contribution to the aggregate research output and thus provide a novel ranking measure that explicitly takes into account the spillover effect generated in the coauthorship network. Moreover, we analyze funding instruments for individual researchers as well as research institutions and compare them with the economics funding program of the National Science Foundation. Our results indicate that, because current funding schemes do not take into account the availability of coauthorship network data, they are ill-designed to take advantage of the spillover effects generated in scientific knowledge production networks.

Reading Notes

Objective

To measure the research output of authors in RePEc and create a model of research productivity, co-author networks, and funding decisions

Importance

They create a model of the network of economists and use that to understand the dynamics of this network.

They also simulate some changes to the network to predict what that would do to the overall research productivity of economists

Background

Collaboration between economists has grown over the last decade. In 1996 50% of all articles were co-authored and now the number is over 75%

Data & Key Variables

RePEc author data and publication data: 8447 papers and 3610 authors for the years 2010 to 2012

Methodology

Theoretical model with a Nash equilibrium

Bayesian network estimation.

Descriptive network graphs

Results

The number of lifetime citations significantly predicts research output. Holding citations constant, the effect of experience (time since Ph.D) is negative.

Men produce more and NBER affiliation and Ivy League education both are positively correlated with higher research output.

Co-authors tend to match with co-authors of the same ethnicity, same gender, and same affiliation.

Being co-authors in the past and sharing common co-authors are both associated with higher probabilities of co-authorship. Being in a PhD advisor-advisee relationship is also correlated with co-authorship.

Key Table/Figure

Authors: Olalekan Bello, Claudia Macaluso, Abdoulaye Ndiaye

Abstract: Employer market power is typically measured and analyzed under the assumption of fixed labor market boundaries, often approximated by geographic units such as commuting zones or counties. We show that this practice misses an important margin: workers broaden the geographic scope of their job search when their home market is concentrated or cyclically weak. Using over 63 million online applications to U.S. hourly jobs, we document two stylized facts. First, the spatial breadth of search is heterogeneous across workers: about one in three searches in a market either smaller or larger than a typical commuting zone. Second, search breadth is also cyclical: when local labor markets deteriorate, an increasing share of workers searches farther from home. To interpret these facts, we propose a search-and-matching model that endogenizes labor market boundaries by combining granular firms with multi-market search. Relative to a fixed-market benchmark, the model has a set of distinct predictions: first, workers in concentrated markets reallocate search effort away from them; second, re-application to large firms is dampened; third, within-firm wages fall with distance to market boundaries, and, fourth, productivity pass-through is attenuated where cross-market search is more costly. Each prediction holds in the data. Effective labor market monopsony exposure is thus jointly determined by local market structure and workers’ search-effort reallocation.

Seminar Notes

Venue

CES Seminar 2025

Objective

To understand how workers trade off skill differences and geographical proximity in job search

Importance

Restricting labor markets by geography and occupation distorts our models if workers are heterogeneous in their flexibility across these factors in their job search.

Matters for quantitative spatial models, measures of monopsony & market power, and effect of place-based policies

Background

Migration is costly, so for average job seeker job search is local, but average worker may not be the marginal worker

Data & Key Variables

Snagajob - large job search and application website. Hourly, low and middle skilled jobs

Daily Jan 2, 2020 - June 10, 2025.

Location (zipcode) and job title (SOC 8-digit)

Applications from workers to jobs. 26 million job postings, 12 million applicants

Methodology

Skill difference - Euclidian distance between job j and mean job an individual applies to, using O*NET skill importance values

Results

Average geographic difference and average skill distance are positively correlated within searcher.

Tradeoff between and geographical proximity in application volume, time in job search spell, and labor market tightness

Key Table/Figure

Authors: William W. Olney, Owen Thompson

Citation: Olney, William W. and Thompson, Owen (2026). The Determinants of Declining Internal Migration. Journal of Urban Economics, 153, 103856.

Abstract: Internal migration in the United States has declined substantially over the past several decades, which has important implications for individual welfare and macroeconomic adjustment. This paper studies the determinants of internal migration and how they have changed over time. We use administrative data from the IRS covering the universe of bilateral moves between every Commuting Zone (CZ) in the country over a 23 year period. This data is linked to information on local wage levels and home prices, and we estimate bilateral migration determinants in rich regression specifications that contain CZ-pair fixed effects. Consistent with theoretical predictions, we find that migration is decreasing with origin wages and destination home prices, and is increasing with destination wages and origin home prices. We then examine the contributions of earnings and home prices to the noted overall decline in internal migration. These analyses show that wages on their own would have led to an increase in migration rates, primarily because migrants are increasingly responsive to high earnings levels in potential destination CZs. However, these wage effects have been more than offset by housing related factors, which have increasingly impeded internal mobility. In particular, migration has become much less responsive to housing prices in the origin CZ, such that many households that would have left in response to high home prices several decades ago now choose to stay.

Reading Notes

Objective

To understand how the decline in migration relates to wages and house prices in origin and destination locations

Background

Migration rates in America have declined over the last several decades, which doesn’t seem to be related to changing population characteristics

Data & Key Variables

IRS bilateral migration flows data 1991-2013 at county level, aggregated to CZ (confidentiality protections changed in 2014 from 10 per cell to 20 per cell)

SOI income file: mean household labor market earnings by CZ

County-level house price index from Borgin et al (FHFA series)

SEER demographics

Distance calculated based on most populous county in CZ

Methodology

Rosen-Robak model of migration choice

Poisson Pseudo-Maximum Likelihood with CZ fixed effects, estimated with both gaps between origin and destination wages and house prices, and with each included separately

Results

People respond to both wage and house price gaps when making migration decisions, but the response is asymmetric

Wages alone would have increased migration because migration decisions have become more sensitive to wage differences.

But migration decisions have become less sensitive to house prices, particularly in origin location, leading to overall decline.

Comments

It seems like you could make an argument for causality going the other direction here - migration has decreased, so the gap between origin and destinations wages and house prices is growing

Key Table/Figure

Authors: Gary Cornwall, Scott Wentland

Abstract: This paper reconsiders the difference-in-differences (DiD) research design for panel data, particularly when serial correlation stems from first-order model misspecification (i.e., dependence in yt rather than exclusively in εt). When time-series issues like this are overlooked, the traditional parallel trends assumption is insufficient. In fact, for most panel applications (T > 2 periods), DiD designs will misidentify and inflate a time-invariant treatment effect. To correct this, we show that DiD assumptions should be modified for dynamic panels and how explicitly accounting for temporal dependence in the design can recover the true, dynamically-robust effect. We then evaluate a simple modification to DiD designs through Monte Carlo simulations and then explore its implications with empirical examples. Two examples leverage a policy shock used in recent literature to reevaluate the impact of household credit constraints on outcomes like state-level GDP growth and labor market participation. When we implement the proposed modification, which can be as simple as incorporating a lagged outcome and group interaction into a DiD model, the results effectively decompose the identified structural effect from the inflated cumulative effect, yielding a more generalizable estimand for most panel applications. Further, accounting for autocorrelation in the outcome may help practitioners avoid mistakenly discarding a DiD design when pre-trends are, in fact, conditionally parallel. Finally, we find synthetic DiD and synthetic control methods do not remedy this particular issue, as similar modifications (e.g., pre-whitening) are needed to address temporal dependence.

Seminar Notes

Venue

BEA Seminar 2025 & ASSA 2026

Objective

To explore the assumptions of diff-in-diff from the time series perspective

Importance

Most people just cluster their SE and don’t think about serial correlation in either standard errors or outcomes

Estimated treatment effect could be orders of magnitude mismeasured.

Estimand is a function of the autocovariance

Background

DiD is most popular of the quasi-experimental methods. Simple, accessible, adaptable

Canonical example- Card & Krueger (1994) used only 2 time periods

Literature focuses on Gap DiD [between treated and control] not Trend DiD [Between pre and post]

Data & Key Variables

Replication using LFPR & GDP - Kumar & Liang (2019 & 2024)

Methodology

Econometric theory, simulation

Estimating equation including lagged dependent variable, on its own and interacted with time dummies GMM

Results

Serial correlation is not necessarily confined to standard errors.

Should incorporate a lagged dependent variable in most DiD designs

Key Table/Figure

Authors: Yana Gallen, Melanie Wasserman

Abstract: This paper estimates gender differences in access to informal information regarding the labor market. We conduct a large-scale field experiment in which real college students seek information from 10,000 working professionals about various career paths, and we randomize whether a professional receives a message from a male or a female student. We focus the experimental design and analysis on two career attributes that prior research has shown to differentially affect the labor market choices of women: the extent to which a career accommodates work/life balance and has a competitive culture. When students ask broadly for information about a career, we find that female students receive substantially more information on work/life balance relative to male students. This gender difference persists when students disclose that they are concerned about work/life balance. In contrast, professionals mention workplace culture to male and female students at similar rates. After the study, female students are more dissuaded from their preferred career path than male students, and this difference is in part explained by professionals’ greater emphasis on work/life balance when responding to female students. Finally, we elicit students’ preferences for professionals and find that gender differences in information provision would remain if students contacted their most preferred professionals.

Reading Notes

Objective

To estimate gender differences in access to informal information on the labor market with a large-scale field experiment

Importance

Gender differences in access to information could lead to occupational segregation by gender

Background

Literature explores information provision, but limited research on information-seeking behaviors.

This paper also expands use of correspondence studies to using real people, which allows for agent selection, since people don’t choose who to email (or which job to apply for, or any other myriad things studied with correspondence studies) randomly

Data & Key Variables

Similar to a traditional correspondence study, but uses real students.

Limit amount of personal information available on platform and recruit a relatively homogenous population (similar majors & years in college)

Methodology

Feb 2020-June 2020 recruited 100 undergraduates to send messages to 10,000 professionals.

Students interested in 4 career paths: finance, management consulting, data science, and law.

Each student given a list of 100 professionals to contact, and were provided text for the initial message. Students sent any responses to the research team.

Results

Female students receive more information about work/life balance than male students.

The majority of the mentions of work/life balance are negative.

Student preferences for advisers exacerbates the gendered effects, rather than mitigates them.

Key Table/Figure