📈 economics

Evaluating the Impact of Rhode Island's Self-Sustaining Reemployment Services and Eligibility Assessment (RESEA) Program on Employment Outcomes

This paper presents a large-scale randomized controlled trial demonstrating that Rhode Island's RESEA program significantly improved employment outcomes and wages for unemployed workers, particularly benefiting older and lower-income individuals, while generating a cost savings of $2.64 for every dollar spent.

Original authors: Harrison H Li, Shanna Pearson-Merkowitz, David Yokum

Published 2026-06-15

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Harrison H Li, Shanna Pearson-Merkowitz, David Yokum

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the unemployment insurance system as a giant safety net. When people lose their jobs, this net catches them and gives them money while they look for a new one. But there's a catch: if people stay on the net too long, the net gets heavy and expensive for the state to maintain, and the people on it might start to lose their "job-finding muscles."

To fix this, Rhode Island tried a new strategy called RESEA. Think of RESEA as a "check-in appointment" with a job coach. The state randomly picked about half of the people who applied for unemployment benefits and sent them a letter saying, "You must come meet with a job coach soon, or we might pause your benefits." The other half didn't get the letter and just kept doing what they were doing.

The researchers, Harrison Li, Shanna Pearson-Merkowitz, and David Yokum, ran a massive experiment (like a scientific taste test) with over 23,000 people to see if this "check-in" actually worked. Here is what they found, explained simply:

1. The Main Results: It Worked

The people who were forced to meet with a job coach did better than those who weren't.

They earned more money: On average, the "check-in" group made about $1,153 more per year than the group that didn't get the letter.
They found jobs faster: About 1.5% more people in the "check-in" group found a job within a year compared to the others.
They spent less time on unemployment: The "check-in" group stayed on unemployment benefits for nearly two weeks less than the other group.

2. The "Magic" of the Letter

Here is a surprising twist: You didn't actually have to finish the job coach meeting to get the benefit. About 31% of the people who got the letter didn't show up to the meeting. Yet, they still found jobs faster and earned more.

The Analogy: Think of the letter like a "wake-up alarm." Even if you don't get out of bed to turn it off, just hearing the alarm might make you realize, "Oh, I need to get up and start my day." The letter itself reminded people that they needed to find work, so they sped up their job search without even needing the coach's help.

3. Who Benefited the Most?

The researchers used a special computer tool (called a "causal forest," which is like a digital tree that sorts people into groups) to see who got the biggest boost.

Older workers and lower-income workers saw the biggest jumps in their wages.
Interestingly, even people who were already doing okay (higher earners, more educated) still saved time on unemployment, though their wage boost wasn't as huge.

4. The Money Math: A Great Deal for the State

The state spent money to run this program, but they saved even more.

Because people found jobs faster, the state paid out less in unemployment checks.
The researchers calculated that for every $1 the state spent on the program, they saved $2.64 in unemployment benefits. It was like buying a $1 ticket and getting $2.64 back in change.

5. The Secret Sauce: How They Did the Math

The paper also points out a "trick" the researchers used that other studies sometimes miss.

The Problem: They ran the experiment week by week. Some weeks had lots of applicants; some had few. Some weeks had more older workers; some had more young ones. If you just mixed everyone together and compared the two groups, the results could be skewed (biased) because the weeks themselves were different.
The Fix: They treated each week like a separate "mini-experiment" and then carefully combined the results. This ensured that the "week" didn't mess up the math. They argue that many past studies missed this step, which might have made their results less accurate.

Summary

In short, sending a letter to unemployed workers telling them to meet with a job coach was a win-win. It helped workers find jobs and earn more money, and it saved the state a lot of money in the process. The "scary" letter itself acted as a powerful nudge, getting people moving even if they didn't fully participate in the program.

Technical Summary: Evaluating the Impact of Rhode Island's RESEA Program

Problem and Context
Prolonged unemployment imposes significant economic, health, and wellbeing costs on individuals and the state. While Unemployment Insurance (UI) provides a critical safety net, concerns exist that it may inadvertently prolong unemployment by reducing the incentive for rapid reemployment. To address this, the U.S. federal government and most states operate Reemployment Services and Eligibility Assessment (RESEA) programs. These programs, successors to earlier Reemployment and Eligibility Assessment (REA) initiatives, mandate targeted services for UI claimants to accelerate their return to the workforce. Despite widespread adoption, there is a need for rigorous, contemporary causal evaluations of these programs under current macroeconomic and demographic conditions, particularly to determine if the RESEA model remains effective and cost-efficient.

Methodology
The authors conducted a large-scale, preregistered randomized controlled trial (RCT) in Rhode Island involving $N = 23,549$ UI claimants between February 2022 and September 2023.

Experimental Design: The study utilized a blocked randomization design where blocks corresponded to weeks. Each Wednesday, approximately 150 (later reduced to 135) newly eligible UI applicants were randomly selected for the treatment group (RESEA selection), while the remainder formed the control group. Veterans were deterministically assigned to treatment due to regulatory requirements and were excluded from the causal analysis.
Intervention: Selected claimants received a letter mandating a meeting with a job coach within 18 days. The intervention included eligibility verification, skills evaluation, job search guidance, labor market information, and referrals. Compliance was enforced through the threat of benefit suspension, though approximately 31% of the treatment group did not attend the meeting.
Outcome Measures: The study focused on three preregistered primary outcomes:
1. Annualized Wages: Sum of the two highest quarterly wages in the second through fifth quarters post-randomization, annualized.
2. Reemployment: A binary indicator of positive wages in at least two of the aforementioned quarters.
3. UI Duration: The number of weeks of UI benefits drawn within 78 weeks of the claim effective date.
  Short-term variants of these outcomes were also analyzed for the second quarter post-randomization.
Statistical Analysis:
- Primary Estimator: The authors estimated the Intention-to-Treat (ITT) effect using block size-weighted averages of difference-in-means estimators.
- Precision Gains: To improve precision without introducing bias, they employed covariate-adjusted ITT estimates using Lin's (2013) interacted linear regression method within blocks, adjusting for pre-treatment covariates (dependents, sex, age, base wage).
- Heterogeneity: Causal forests (Wager & Athey, 2018) were used to estimate Conditional Average Treatment Effects (CATE) to identify subgroups benefiting most from the program.
- Methodological Refinement: A key methodological contribution was the explicit control for the week of treatment assignment. The authors argue that failing to adjust for the block (week) in block-randomized designs can introduce confounding bias, as the proportion of treated individuals and applicant characteristics varied significantly across weeks.

Key Results
The study found that selection into the RESEA program produced statistically significant positive effects across all primary outcomes:

Wages: Selection increased annualized wages by $1,153 (95% CI: [$166, $2,141];$ p=0.022$), representing a 3.5% increase relative to the control group mean.
Reemployment: Selection increased the probability of reemployment by 1.47 percentage points (95% CI: [0.32, 2.62]; $p=0.013$ ).
UI Duration: Selection reduced the duration of UI benefits by 1.99 weeks (95% CI: [1.75, 2.24]; $p<0.001$ ), an 11.5% reduction relative to the control group mean.
Timing: The effects on wages and reemployment appeared rapidly, manifesting within two quarters of the first pay date and persisting through at least the following year. Short-term estimates were substantively similar to long-term estimates.
Heterogeneous Effects: Causal forest analysis revealed that while all participants experienced positive wage benefits, the effects were particularly large for older and lower-income workers. Conversely, higher-earning and better-educated workers showed smaller reemployment effects but still experienced significant reductions in UI duration and exhaustion rates.
Cost-Benefit Analysis: The reduction in UI benefits drawn resulted in a savings of approximately $676 per treated individual during the claim year. With a total program expenditure of $2.99 million, the state saved an estimated $7.9 million in UI payouts, yielding a return of $2.64 saved for every $1 spent.

Significance and Claims
The paper claims to provide strong causal evidence that the Rhode Island RESEA program effectively achieves its dual goals: helping workers return to employment faster and preserving state UI resources.

Policy Relevance: The findings support the continued investment in RESEA programs, demonstrating their efficacy even under conditions differing from the original studies that motivated federal adoption.
Methodological Contribution: The authors highlight a critical methodological refinement often absent in prior RCTs of job-training programs: the necessity of explicitly controlling for the week of treatment assignment (the block) in block-randomized designs. They demonstrate that ignoring the block can lead to biased estimates due to confounding by time-varying applicant characteristics. They advocate for the use of block size-weighted averages of difference-in-means or covariate-adjusted within-block estimators to ensure unbiased and precise estimates.
Robustness: The results remained statistically significant after correcting for multiple testing using Holm's method. The authors also addressed potential biases regarding out-of-state employment and self-employment, finding no evidence that RESEA selection differentially affected these unobserved employment types.

In conclusion, the study validates the RESEA model as a cost-effective intervention that accelerates reemployment and reduces reliance on unemployment benefits, with particular benefits for vulnerable subgroups such as older and lower-income workers.