This website uses cookies.

By clicking the "Accept" button or continuing to browse our site, you agree to first-party and session-only cookies being stored on your device to enhance site navigation and analyze site performance and traffic. For more information on our use of cookies, please see our Privacy Policy.

  1. Home
  2. ASSA Annual Meeting
  3. 2026
  4. Program
  5. Collaborative State and Local Policy Research
« Back to Results

Collaborative State and Local Policy Research

Paper Session

Monday, Jan. 5, 2026 1:00 PM - 3:00 PM (EST)

Philadelphia Convention Center, 203-B
Hosted By: American Economic Association
  • Chair: Valentin Bolotnyy, Stanford University

Reducing Default Eviction Judgments: An Information Experiment

John Eric Humphries
,
Yale University
Winnie van Dijk
,
Yale University
Shannon Bradford
,
Yale University
Göksu Zeybek
,
Yale University

Abstract

Civil courts play an essential role in adjudicating disputes and protecting property rights, yet many defendants fail to participate in proceedings, leading to default judgments in favor of plaintiffs. We study how information provision affects civil court outcomes through a randomized field experiment in Virginia eviction cases, conducted in partnership with a local legal aid provider. We find that sending defendants information about hearing dates and legal aid availability substantially reduces default judgments and increases dismissals. However, we find only modest increases in legal representation, indicating that increased court attendance, rather than legal assistance, drives the results. These findings indicate that small and low-cost procedural changes can meaningfully lower default judgments in civil court.

Individual and Social Effects of Shelter for People Experiencing Homelessness: Evidence From Los Angeles County's Winter Shelters Program

Derek Christopher
,
Stanford University
Olivia Martin
,
Stanford University
Mark Duggan
,
Stanford University

Abstract

We exploit variation from shocks to shelter availability from Los Angeles County's winter shelters program to study the effects of shelter provision and sheltered (versus unsheltered) homelessness. We leverage records of homeless services to generate daily, site-level counts of shelter beds and occupants from 2014 to 2019. We pair shelter counts with block-level crime data and facility-level data on ER visits to assess the impacts of shelter on crime and health outcomes. Further, using enrollment-level homeless services data, we evaluate the effect of shelter on returns to homeless services, observed exits from homelessness, and observed mortality. While preliminary estimates suggest that street outreach may be just as effective as shelter in reducing an individual's likelihood of still being homeless 6-18 months in the future, our findings indicate that increased investment in shelter provision may be optimal given its ability to mitigate both private and social costs of homelessness.

The Labor Market Returns to Customized Job Training

Natalie Millar
,
Stanford University

Abstract

Economic theory dating back to Becker (1962) predicts that employers and workers should share the cost of job training in specific, non-transferable skills, leaving workers or the government to fund general skills training. Customized job training (CJT) programs, which exist in most U.S. states, defy this logic by using public subsidies to teach workers a range of skills, including those that are firm-specific. Are governments mistakenly subsidizing training that firms would pay for on their own, or does CJT generate benefits that justify public investment? I answer these questions using unique hand-collected data from Tennessee on firms' grant applications and trainees' enrollment linked with rich administrative data on education, earnings, and public assistance. I exploit the fact that there is quasi-random rationing among equally comparable firms and equally eligible prospective trainees. The estimates show that enrolling in a CJT program, typically lasting about four months, increases earnings by 3% per quarter over five years, comparable to the return from one additional year of work experience. To explain these findings, I classify the skills taught in each program by mapping program descriptions in firms' grant applications to O*NET detailed work activities. Despite the fact that CJT programs primarily produce transferable skills across industries and occupations, benefits to the government through higher individual income tax revenues more than offset training costs, yielding an exceptionally high marginal value of public funds relative to other job training programs.

How Retrainable Are AI-Exposed Workers?

Ben Hyman
,
Federal Reserve Bank of New York
Karen Ni
,
Harvard University
Laura Pilossoph
,
Duke University

Abstract

We document the extent to which AI-exposed workers can successfully retrain for AI-intensive work. We assemble a new job training dataset spanning over 2.9 million unique job training participation spells from all U.S. Workforce Investment Act (WIA) programs between 2012 – 2023. Linking occupational measures of AI exposure with earnings records observed both before and after training, we compare earnings and occupational transitions of job trainees to observationally similar AI-exposed workers who only received job search assistance services through the Wagner-Peyser program. We then construct an AI retrainability index (AIR) at the occupation level that weights transitions into more AI-intensive occupations by individual earnings returns from training. This allows us to disentangle whether AI retrainability is driven by AI skill transitions versus general returns to training.

Discussant(s)
Vincent Reina
,
University of Pennsylvania
Noah Boden-Gologorsky
,
Stanford University
Jonathan Roth
,
Brown University
Daniel Rock
,
University of Pennsylvania
JEL Classifications
  • H7 - State and Local Government; Intergovernmental Relations
  • I0 - General