article 3 week 1

Overview

• New claimants for Unemployment Insurance (UI) were randomly assigned to experiments designed to expedite their return to work
• The first experiment offered a $500 bonus to claimants finding employment within 11 weeks which reduced insured unemployment duration by over a week
• The second experiment provided the $500 bonus to the subsequent employer of the claimant which led to reduced insured unemployment duration for white women by around one week

Experiment Details

• The Illinois Department of Employment Security conducted two controlled social experiments to test the effectiveness of cash bonuses in reducing insured unemployment duration from mid-1984 to mid-1985
• In the Claimant Bonus Experiment, new UI claimants were informed they would qualify for a $500 cash bonus if they secured a job (30+ hours/week) within 11 weeks and retained it for four months, incentivizing intense job search and quicker re-employment
• The Employer Bonus Experiment, new UI claimants were informed their next employer would be eligible for a $500 bonus if the claimant found employment within 11 weeks and remained employed for four months which aimed to provide a wage-bill/training subsidy, potentially decreasing insured unemployment duration

Motivation and Design

• The Illinois UI experiments stemmed from a decade of UI system criticism and economists' struggles to discern UI recipients' behavior versus what it would be without the system
• UI benefits may prolong joblessness by subsidizing additional job searches or nonmarket time consumption
• Illinois experiments explored whether bonuses to UI beneficiaries/employers would reduce unemployment relative to a randomly selected control group, providing strong evidence against UI being solely a benign income transfer
• Both treatments involved a $500 bonus, reflecting a balance between budget constraints (maximum $750,000 bonus payments) and generating responses (about 5% of annual payments)
• Claimants had to secure 30+ hours/week jobs within 11 weeks, retaining them for at least four months

Eligibility and Site Selection

• Individuals had to file initial UI claims between July 29 and November 17, 1984
• Individuals had to be eligible for 26 weeks of UI benefits
• Individuals had to register with 1 of 22 Job Service offices in northern and central Illinois
• Individuals had to be 20-55 years old
• Claimants were assigned to a control group, Claimant Experiment treatment group, or Employer Experiment treatment group via simple random assignment based on the last two digits of their Social Security number, ensuring equal probability of assignment
• Limiting to 22 Job Service offices lowered monitoring and communication costs

Operational Execution

• The experimental design aimed for simplicity, yet it introduced complexities within the existing agency
• Creating specialized tools to supervise & monitor enrolled claimants was essential
• Claimants were administered monetarily eligible surveys for the 20-54 age group
• Records to track claimant treatment assignment and experiment experience were kept
• Willing claimants were required to sign an agreement to participate
• Willing claimants received packets of informative materials by mail once their nonmonetary eligibility for benefits was assessed

Operational Tracking (cont.)

• Claimant Experiment participants finding jobs within 11 weeks submitted a Notice of Hire
• Employer Experiment participants who found jobs within 11 weeks gave a Notice of Hire to the employee for their employer to submit
• Vouchers would be delivered and then submitted back for $500 payment
• Job Service office logs recorded transactions
• Baseline surveys, office logs, and administrative databases were all analyzed

Data Analysis

• Benefits Information System: records claim dates, amounts, and benefit timing
• Wage Records data base: each claimant's quarterly earnings in UI-covered employment from Q3 1983-Q3 1985
• The access to the data bases protected against nonresponse bias

Experimental Enrollment Statistical Validity

• Approximately 4,000 eligible UI claimants were assigned to each experimental category
• Procedures ensured each sample could be treated as a random sample from the population of eligible claimants aged 20-54
• There were important differences in claimants' willingness to participate: 84% in the Claimant Experiment vs 65% in the Employer Experiment

Experimental Use Statistical Validity

• There were significant differences in actual program use
• 14% of the Claimant Experiment group received a bonus
• Only 3% of the Employer Experiment group led to a bonus payment

Key Differences & Complications

• Understanding and participation of both worker and employer was required in the Employer Experiment where as in the Claimant experiment was more straightforward
• Limited use suggested a limited scope for UI benefit reduction in the Employer Experiment
• Randomness was supported because none of the differences in proportions between any pair of groups is statistically different from zero at conventional significance levels

Claimant Incentives

• Workers were assigned to the Claimant Experiemtn and could lower their reservation wage or search more intensely
• Workers were free to substitute income for leisure

Employer Incentives

• Employers who participated had incentive to participate
• Enrollees would receive a subsidy rather than a bonus
• Enrollees had increased probability of being hired, but it does not necessitate increasing total employment

Earnings Impact

• Shortened search time may result in unfavorable matching of workers and jobs
• The first job accepted by a Claimant Experiment participant may have led a lowered labor market efficiency

Experimental Sub-Groups

• Certain groups of workers are possible to have responded more strongly than others to the experimental incentives
• White women unambiguously reduced UI benefit payments and weeks of insured unemployment

Race & Sex Statistical Variance

• The effects of Claimant incentive varied by race with whites nearly 3x more likely to receive benefits than blacks.
• Whites' employers received Employer incentives 10x more often than blacks' employers
• Disparities in bonuses are attributed to eligibility and take-up rates differences
• Higher claimant experiment bonus was more significant than the corresponding rate for the employer experiment
• Benefit and take ups rates varied little by sex or experiment when compared to races

Benefit-Cost Ratios

• The overall Claimant Experiment benefit-cost ration was at 2.32
• The overall Employer Experiment benefit-cost ration was at 4.29
• The benefit-cost rate for white women in the Employer experiment was at 7.07

Important Caveats & Factors for Further Research

• Full-Scale effects on the experiment could have significant displacement effects
• 100% takeup rate would drive rates of reduction of UI benefits greatly

Conclusions

• $500 Claimant Experiment bonus reduced state benefits by $158 and unemployment by one week.
• White women who were assigned to the Employer treatment group received $164 less in UI benefits
• Experimental groups saw experience reduced unemployment compared to control groups
• Experimental groups were not worse off from assignment to the experiments.

Additional Considerations

• Further research recommended to address whether a full-scale program modeled to reduce effects by the Claimant/Employer experiments would result in improvements

Model and Experimental Advantages

• Administrative Data avoids measuring attrition surveys like Ashenfelter and Plant

Alternate Research

• Measuring responses and variations with size of the bonus size
• Understanding long range implications for bonus and unemployment, and employee takeup rates
• Analysis to better compare the costs between treatment and social groups.
• Additional experiments design to efficiently reduce unemployment while avoiding harm to unemployed individuals.