There is no conflict of interest.
Citation
Dahl, G., & Knepper, M. M. (2022). Unemployment Insurance, Starting Salaries, and Jobs. National Bureau of Economic Research, Working Paper 30152. https://www.nber.org/papers/w30152. [Comparison between 6 moderate unemployment insurance reform states and 20 Southern and Midwestern comparison states]
Highlights
- The study's objective was to examine the impact of moderate unemployment insurance reforms on public benefits receipt, employment, and earnings and wages. The authors investigated similar research questions for two other contrasts, the profiles of which are available using the study search.
- The study used a difference-in-differences design to study the impact of unemployment insurance reforms on unemployment insurance recipiency rates, unemployment rates, employment, starting salaries, and posted salaries. The authors used a variety of data sources, including BLS Local Area Unemployment Statistics, Department of Labor administrative data, the Current Population Survey, an administrative dataset from the Equal Employment Opportunity Commission, a proprietary dataset from Glassdoor, and a dataset from Burning Glass Technologies. For the aggregate state-level analysis, the authors compared the unemployment insurance benefits recipiency rates and unemployment rates of 6 moderate reform states and 20 Southern and Midwestern states before and after the unemployment insurance reforms. For the establishment-level analysis, the authors compared log employment, starting salaries, and posted salaries at a moderate reform state establishment and same-firm establishment in another state before and after the unemployment insurance reforms.
- For the aggregate state-level analysis, the authors found negative statistically significant relationships for unemployment insurance recipiency rates and unemployment rates between the treatment and comparison groups. For the establishment-level analysis, the authors found a positive statistically significant relationship for employment but a negative statistically significant relationship for starting salaries and posted salaries between the treatment and comparison groups.
- The study receives a low evidence rating. This means we are not confident that the estimated effects are attributable to the unemployment insurance reforms in six moderate reform states; other factors are likely to have contributed.
Intervention Examined
Unemployment Insurance (UI)
Features of the Intervention
Unemployment insurance provides temporary income for eligible laid-off workers. During the long recovery from the Great Recession, several states made record numbers of unemployment insurance payouts. As a result, states depleted their unemployment insurance funds and passed legislation to make cuts.
Six states passed moderate unemployment insurance reforms in the 2010s that cut the maximum benefits duration from 26 to 20 weeks. These states were Florida, Georgia, Kansas, Michigan, Missouri, and South Carolina. These moderate reforms ended up reducing the maximum value of benefits by 23%.
Features of the Study
The study used a difference-in-differences design to study the impact of moderate reform states’ unemployment insurance reforms on unemployment insurance recipiency rates, unemployment rates, employment, starting salaries, and posted salaries between 2008 and 2018. The authors conducted an aggregate state-level analysis and an establishment-level analysis. For the aggregate state-level analysis, the sample consisted of the 6 moderate reform states and the 20 Southern and Midwestern comparison states. For the establishment-level analysis, the sample was restricted to multi-state firms that operated in one of the moderate reform states and at least one other state. The authors had a study sample of nearly 1.5 million establishment-year observations for the employment outcome, 1 million establishment-year observations for the starting salaries outcome, and nearly 1.4 million establishment-year observations for the posted salaries outcome.
For the aggregate state-level analysis, the treatment condition included the 6 moderate reform states, and the comparison condition included 20 Southern and Midwestern states that did not implement a moderate unemployment insurance reform. For the establishment-level analysis, the treatment condition was an establishment located in one of the six moderate reform states, and the comparison condition was a same-firm establishment located outside of the moderate reform states.
The authors used a variety of data sources. For the aggregate state-level analysis, they used the BLS Local Area Unemployment Statistics from July 2010-December 2018, Department of Labor administrative data, and the Current Population Survey. The authors compared unemployment insurance recipiency rates and unemployment rates of the six moderate reform states and 20 Southern and Midwestern comparison states before and after the unemployment insurance reforms.
For the establishment-level analysis, the authors used an administrative dataset from 2008-2015 from the Equal Employment Opportunity Commission and the Basic Monthly Current Population Survey. They also used a proprietary dataset from Glassdoor that includes self-reported salary, company, and workplace location data from 2008-2016 along with data on posted wages from online job ads from Burning Glass Technologies from 2010-2017. The authors compared log employment, starting salaries, and posted salaries at a moderate reform state establishment and same-firm establishment in a comparison state before and after the unemployment insurance reforms.
Findings
Public benefits receipt
- The authors found a negative statistically significant relationship in unemployment insurance recipiency rates between the 6 moderate reform states and the 20 comparison states.
Employment
- The authors found a negative statistically significant relationship in unemployment rates between the 6 moderate reform states and the 20 comparison states. The authors found a positive statistically significant relationship in employment at establishments in moderate reform states compared to same-firm establishments in the comparison states.
Earnings and wages
- The authors found a negative statistically significant relationship in starting salaries at establishments in 6 moderate reform states compared to same-firm establishments in the 20 comparison states. The authors found a negative statistically significant relationship in posted salaries at establishments in moderate reform states compared to same-firm establishments in the comparison states.
Considerations for Interpreting the Findings
The authors did not include necessary data to calculate group-level attrition. The composition of the treatment and comparison groups may have significantly changed over time. It is possible that there could be other factors driving these findings.
Causal Evidence Rating
The study receives a low evidence rating. This means we are not confident that the estimated effects are attributable to the moderate unemployment insurance reforms; other factors are likely to have contributed.