Unemployment Insurance Take-Up Evaluation

The intervention consists of two waves of letters to be sent to all workers in the treatment group corresponding to two quarters of data the Washington Employment Security Department (ESD) receives on potential job losses experienced in Washington state.

The first wave of letters will be mailed on May 21, 2024, to monetarily eligible workers who likely separated from their jobs in 2024Q1. The second wave of letters will be mailed in mid-August 2024 to monetarily eligible workers who likely separated from their jobs in 2024Q2. Mailing addresses were procured from ESD's internal records on past UI recipients and a data sharing agreement with the Washington State Department of Licensing.

All letters will inform workers that if they have lost their jobs through no fault of their own, they may qualify to receive UI. These letters explain UI's function, eligibility requirements, and provide information on how to apply. In the first wave, additional treatment arms vary messages on top of this basic information to provide messages targeting stigma, misperceptions about job search, and salience of benefit levels. Messages will be cross-randomized.

Results on applications and receipt in the first wave will inform which version(s) of the letter are employed in the second wave to be mailed in August 2024.

2024-05-21

2024-08-30

Primary outcomes of interest from the first stage are:
UI application rate [Timeframe: 2,4,6 months after initial mailing]
UI recipiency rate [Timeframe: 2,4,6 months after initial mailing]
Baseline characteristics of applicants (demographics such as age, gender, veteran status, disability status) [Timeframe: 2,4,6 months after initial mailing]
Baseline characteristics of recipients [Timeframe: 2,4,6 months after initial mailing]

The primary outcomes of interest from the second stage concern labor supply. These outcomes will be measured by the number of weeks on UI (proxied by number of weekly payments), non-employment duration, re-employment wages, and occupation (to the extent available -- see explanation)- and industry-switching. We will follow these second stage outcomes over the following two years after the intervention.

We will investigate whether the treatment has heterogeneous effects according to workers' previous experience with the UI system, economic position (hourly wage/hours worked/earnings in previous job, industry, occupation, firm size, firm layoff rate), and geography.

The Washington State legislature began requiring the reporting of occupation SOC codes starting in 2024Q1 (see background information: https://esd.wa.gov/employer-taxes/about-soc) but reporting of such codes may be incomplete. Also, some voluntary reporting of SOC codes began earlier for some firms. At the time of pre-registration, the availability of high-quality SOC codes for the time covered by the trial is uncertain. Whether we measure the effects on occupation-switching will depend largely on the quality and availability of data.

Number of responses to outreach letters (i.e., phone calls to the number and visits to redirect website link listed on the outreach letter) [Timeframe: 2, 4, 6 months after initial mailing]

Individuals were selected from the set of UI-covered workers in the state of Washington who had stable employment at a single firm for two consecutive quarters and then experienced a sharp contraction in the number of hours worked in the most recent quarter (i.e., 2024Q1, for the first wave of the trial). Workers must be monetarily eligible (worked at least 680 hours in their base year or alternate base year). Workers must not have already claimed UI or Paid Family and Medical Leave benefits around the same time as this hours drop. The sample excludes workers who experienced an hours contraction in the following industries: Mining, Quarrying, and Oil and Gas Extraction, Information, Finance and Insurance, Management of Companies and Enterprises, and Educational Services. Such industries were excluded either because they had relatively high wages or because most of their separations were judged not to be UI-eligible.

From this list, workers were randomly assigned to control groups or one of various cross-randomized treatment arms (including a generic information group).

Not available

Treatment was randomly assigned using a computer (R) within strata constructed based on whether a worker had ever received UI in Washington state before.

Individual-level.

No

N/A

In the first wave, there are 29,220 workers. In the second wave, we anticipate there will be approximately 30,000 workers. However, this will not be known until 2024Q2 tax records are reported in August 2024.

In Wave 1:
4,000 control workers;
3,153 treatment workers receiving generic letters;
3,153 treatment workers receiving letters with benefit ("benefit") salience;
3,153 treatment workers receiving letters with de-stigmatizing ("stigma") message;
3,153 treatment workers receiving letters with job search duration ("duration") information;
3,153 treatment workers receiving letters with duration and stigma;
3,153 treatment workers receiving letters with duration and benefit;
3,153 treatment workers receiving letters with stigma and benefit;
3,149 treatment workers receiving letters with duration and stigma and benefit

The sample size by treatment arm for the second wave will be informed by the results of the first wave.

Preliminary power calculations suggest that the minimum detectable effect size on recipiency is roughly 0.7 percentage point change (35%) when comparing the treated group (receiving any letter) to the status quo control, assuming conventional 80% power and 95% confidence intervals. We assume 2% of the control group receives UI absent any intervention (standard deviation = .14) between the time of intervention and the end-line. This effect size is smaller than effects seen in the literature for similar outreach interventions for public benefits.