Absence of conflict of interest.
Citation
Behaghel, L., & Blau, D. M. (2010) Framing Social Security Reform: Behavioral responses to changes in the full retirement age. (IZA Discussion Paper No. 5310). Bonn, Germany: IZA.
Highlights
- The study examined the impact of changes to the Full Retirement Age (FRA) on labor force exit, Old Age and Survivor’s Insurance (OASI) claims, and retirement.
- The study uses a statistical model and the data from the Health and Retirement Study (HRS) to compare outcomes of people in birth cohorts who were affected by the change in FRA with those of a comparison group of people who were not.
- The study found that the change in the FRA was associated with significant increases in the likelihood of claiming OASI and exiting the labor force at the new, later full retirement age.
- The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the change in FRA, but other factors might also have contributed.
Intervention Examined
Changes to the Full Retirement Age (FRA)
Features of the Intervention
In 1983, a Social Security reform law increased the age at which people can retire with full OASI benefits, referred to as the FRA. At that time, the FRA was 65 years old, and the law increased the FRA for cohorts born from 1938 to 1943 by 2 months for each successive birth cohort. Those born between 1938 and 1943 reached retirement age between 2004 and 2009.
Features of the Study
The study used a statistical model (differences-in-differences model) to examine the effects of the change in FRA on several outcomes related to employment (exit from the labor force, retirement, and OASI claims). For the HRS analysis, the intervention group was defined as people in birth cohorts 1938 to 1941 (those who were subjected to the change in FRA), and the comparison group was defined as people in birth cohorts 1932 to 1937 (those who were not subject to the change in FRA).
Using data from the HRS, the authors compared outcomes at ages 62 to 66 of people in birth cohorts who were affected by the change (1938 to 1941) with those in birth cohorts who were not (1932 to 1937). The HRS is a survey of households with people older than age 50 and their spouses conducted every other year. The sample included roughly 2,900 people ages 64 to 66. There are about 1,125 people in the intervention group and 1,750 in the comparison group.
The analysis controls for birth cohort, age, individual background characteristics, and the removal of the earnings test for individuals who reached their FRA, enacted in 2000.
Findings
Employment
- The FRA increase was associated with a significant increase in hazard rate of labor force exit (the likelihood of exiting the labor force conditional on not having exited already) at the new, later FRA by 1 percentage point. In other words, those that were subjected to the new FRA were more likely to exit the labor force than those that were not subject to the new FRA.
- The study found that the FRA increase was not significantly associated with a change in the likelihood of retirement at the FRA.
- The study found that the FRA increase significantly increased the hazard rate of OASI claims at the new, later FRA by 13.6 percentage points.
Considerations for Interpreting the Findings
The authors showed that the trends in benefit claims and retirement were similar between the groups before the change in FRA. The trends in hazard rate of labor force exit seemed to vary more, which might be attributable to the fact that labor force exits were measured monthly and the cell size was small.
The OASI claim outcome is based on data indicating when the individual first collected any Social Security benefit. Thus, people who claim other Social Security benefits (such as disability or dependent benefits) before age 62 cannot be included in the analysis of OASI claims.
Causal Evidence Rating
The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the change in FRA, but other factors might also have contributed.