There is no conflict of interest.
Citation
Gelber, A. M., Jones, D., Sacks, D. W., & Song, J. (2020). The Employment Effects of the Social Security Earnings Test (No. w26696). National Bureau of Economic Research.
Highlights
- The study’s objective was to examine the impact of the Social Security Annual Earnings Test (AET) on employment.
- The study used a difference-in-differences design to compare the outcomes of individuals potentially earning above and below the AET exempt amount before and after reaching the age of 62. Using data from the Earnings Public Use File (EPUF), the authors conducted statistical models to compare differences between the treatment and comparison groups.
- The study found a significant relationship between the AET and a decrease in the likelihood of employment for Americans aged 63-64.
- This study receives a low evidence rating. This means we are not confident that the estimated effects are attributable to the AET; other factors are likely to have contributed.
Intervention Examined
Social Security Earnings Test
Features of the Intervention
The Social Security Annual Earnings Test (AET) is a policy that affects the employment decisions of Americans who are age 62 and older. This policy reduces their Social Security Old Age and Survivor Insurance (OASI) benefits, which provide annuity income to seniors and families of deceased workers, based on their income. When earnings exceed a specified amount, OASI benefits decrease by one dollar for every two dollars earned over that limit. This study investigates whether the AET impacts the employment decisions of older workers.
Features of the Study
The study used a nonexperimental design to compare the employment rate of individuals with potential earnings above and below the AET exempt amount, both before and after reaching 62 years of age. Potential earnings are defined as those an individual would earn in the absence of the incentives created by the AET. The authors used the Earnings Public Use File (EPUF) from the Social Security Administration (SSA). This dataset contains a 1 percent sample of all Social Security Numbers issued as of 2006, along with earnings histories and limited demographic information from 1951 to 2006.
The authors selected individuals born from 1931 to 1943 and examined their earnings between 1986 and 2006. The total study sample included 240,181 individuals. The treatment group consisted of individuals with earnings above the exempt amount during the base age period (ages 55-61), while the comparison group consists of individuals with earnings below the exempt amount during the same period. In the analytical sample, the average yearly employment rate was 80.5% for the treatment group and 66.6% for the comparison group. The average earnings were $30,755 for the treatment group and $18,139 for the comparison group. Over half of the analytic sample was female, with 55.9% in the treatment group and 51.7% in the comparison group.
Statistical models were used to compare outcomes between the treatment and comparison groups before and after exposure to the AET. The analysis included covariates such as sex, age, and an imputed measure of benefits representing the benefit an individual would receive at age 62 if they had no earnings at that age, based on their earnings at age 54.
Findings
Employment
- The study found that individuals aged 63-64 who were predicted to have higher earnings under the AET had lower employment rates compared to those with incomes below the threshold.
Considerations for Interpreting the Findings
The database used in the study contained only year of birth and sex. Thus, the authors were unable to include control variables for race/ethnicity in their statistical models as required by CLEAR. These preexisting differences between the groups—and not the AET policy—could explain the observed differences in outcomes. Therefore, the study is not eligible for a moderate causal evidence rating, the highest rating available for nonexperimental designs.
Causal Evidence Rating
The quality of causal evidence presented in this report is low because the authors did not include sufficient controls for pre-existing differences between the study groups in their analysis. This means we are not confident that the estimated effects are attributable to the Social Security Annual Earnings Test (AET); other factors are likely to have contributed.