Skip to main content

How did the elimination of the US earnings test above the normal retirement age affect labour supply expectations? (Michaud & Van Soest 2008)

Absence of conflict of interest.

Citation

Michaud, P., & Van Soest, A. (2008). How did the elimination of the US earnings test above the normal retirement age affect labour supply expectations? Fiscal Studies, 29(2), 197-231. doi:10.1111/j.1475-5890.2008.00073.x

Highlights

  • The authors used a nonexperimental approach and the data from Health and Retirement Study data from 1996 to 2002 and Social Security earnings history to estimate the impact.
  • The study examined the impact of repealing the earnings test from Social Security recipients between full retirement age and age 70 on older workers’ subjective probability of retirement.
  • The authors used a nonexperimental approach and the data from Health and Retirement Study data from 1996 to 2002 and Social Security earnings history to estimate the impact.
  • The study found that after the repeal of the earnings test, the subjective probability of working of full-time past age 65 for older male workers who were likely to have most of their projected Social Security benefits reduced under the earnings test has increased their expected probability of working relative to those who were not likely to be affected. •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 repeal of the earnings test, but other factors might also have contributed.

Intervention Examined

Social Security Earnings Test

Features of the Study

In 2000, Congress repealed the earnings test on workers ages 65 to 69, which previously reduced Social Security benefits if earnings exceeded certain thresholds. The study used more than 4,000 observations of male workers ages 51 to 61 from the Health and Retirement Study matched to Social Security records. The authors used a difference-in-difference regression study design to compare the outcomes of those with pre-repeal earnings high enough to be impacted by the repeal (defined as surpassing 80 percent of the exempt amount) to those who were unlikely to be affected. The outcomes were self-reported probabilities of full-time employment after age 62 and after age 65, and whether the probability was greater than zero at each age.

Findings

Employment

  • The study found that, after the repeal, people who were subject to losing 50 to 99 percent of their projected Social Security benefits at the normal retirement age significantly increased their subjective probability of working by 4 percentage points more than those who were not likely to be impacted by the changes of the policy. The study also found that the repeal was associated with a 10 percentage point increase in worker’s self-reported probability of working past age 65 for the group losing 1 to 49 percent of their projected Social Security benefits compared with those who were not likely to be impacted by the policy change.

Considerations for Interpreting the Findings

The authors created the treatment groups according to earnings before the policy change and did not account for anticipated changes in hours or wages. This likely underestimates the effects on the subjective probability of working full-time past the normal retirement age. In addition, the authors included those on the margin in the treatment group rather than the control group, which could also underestimate the effects of the policy change. They conducted a sensitivity test by including projected earnings as a control variable, which did not change the findings, though Social Security earnings are top-coded at $90,000, so the effects on high earners might not be fully captured.

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 repeal of the earnings test, but other factors might also have contributed

Reviewed by CLEAR

September 2019

Topic Area