Skip to main content

Social security and retirement timing: Evidence from a national sample of teachers (Morrill & Westall, 2019)

Review Guidelines

Absence of conflict of interest.


Morrill, M. S., & Westall, J. (2019). Social security and retirement timing: Evidence from a national sample of teachers. Journal of Pension Economics and Finance, 18, 549–564.


  • The study’s objective was to examine the impact of Social Security coverage on public school teachers’ retirement decisions.  

  • The authors conducted a nonexperimental comparison group analysis to compare retirement rates for teachers who were covered by Social Security with those for teachers who were not covered by Social Security. The authors used American Community Survey (ACS) data from 2010 to 2016 and data from the Center for Retirement Research, Public Pensions Database (PPD) describing state-level Social Security and pension plan information.  

  • The study found that teachers covered by Social Security were more likely to retire starting at age 62, when retirement eligibility under Social Security begins. Social Security coverage was also associated with lower labor force participation rates for teachers ages 64 and older.  

  • This study receives a moderate evidence rating. This means we are somewhat confident that the estimated effects are attributable to Social Security coverage and not to other factors.  

Intervention Examined

Section 218 of the Social Security Act

Features of the Intervention

The Social Security Act of 1935 only provided Social Security coverage to private sector workers. In 1950, Section 218 of the Social Security Act extended retirement benefits to public sector workers in states that opted into the system. Subsequent changes to the Social Security Act mandated that coverage be expanded to any state or local public sector employees not included in another pension system. About 60 percent of public school teachers are covered by Social Security. 

Features of the Study

The authors used a nonexperimental comparison group design to compare the retirement rates of teachers with and without Social Security coverage. The authors used individual-level data from the ACS from 2010 to 2016 and data on teacher pension plans and Social Security coverage from the PPD. The sample consisted of 88,788 teachers ages 50 to 70. The average age of teachers in the sample was 57, and most teachers in the sample were female (77 percent) and had an advanced degree (60 percent). Eighty-seven percent of the sample identified as White, 8 percent as African American, and 4 percent as another race. Six percent of the sample identified as Hispanic or Latino.  

The study included teachers from 47 states and the District of Columbia. The authors excluded three states (Georgia, Missouri, and Rhode Island) because school districts in those states decided whether to provide Social Security coverage. Thirty-five states offered Social Security coverage to teachers, and 12 states plus the District of Columbia did not.  


  • Teachers ages 62 to 70 in states that provided Social Security coverage had a 3.7 percentage point higher probability of retirement from any position than teachers in states that did not provide Social Security coverage. The authors analyzed the impact of Social Security coverage at each age individually, from 62 to 70, and found a higher probability of retirement specifically at ages 62, 64, and 66 for teachers in states that provided Social Security coverage. 

  • Starting at age 64, teachers in states that provided Social Security coverage had a lower probability of labor force participation, suggesting that Social Security coverage was associated with a higher probability of retirement. This lower labor force participation was true for teachers at ages 64, 66, 67, and 70.  

Considerations for Interpreting the Findings

The study authors estimated multiple related impacts on outcomes related to employment. Performing multiple statistical tests on related outcomes makes it more likely that some impacts will be found statistically significant purely by chance and not because they reflect program effectiveness. The authors did not perform statistical adjustments to account for the multiple tests, so the number of statistically significant findings is likely to be overstated. 

Causal Evidence Rating

This study receives a moderate evidence rating. This means we are somewhat confident that the estimated effects are attributable to Social Security coverage of public school teachers and not to other factors. 

Reviewed by CLEAR

May 2021

Topic Area