How Does Social Security Affect Saving? Evidence from Eligibility Changes for State and Local Government Employees

by Clara M. Zverina, Harvard University

My paper will investigate the effect of Social Security coverage on private saving. While this question has received much attention in cross-sectional, time-series, and cross-national studies in the past [e.g., Feldstein and Pellechio (1979, Feldstein (1996), Modigliani and Sterling (1983)], none of the existing estimates of the crowd-out effect of Social Security on private saving is well-identified. My project aims to address this short-coming through a quasi-experimental research design, using exogenous variation in Social Security eligibility of some state and local government employees between 1954 and 1991. While state and local government employees were originally excluded from Social Security, public employers were allowed to voluntarily elect Social Security coverage for their employees through Section 218 Agreements with the Social Security Administration, starting in 1954. In 1990, coverage of all state and local government employees by Social Security or an equivalent state pension plan was federally mandated, and resulted in an additional 3.7 million public employees joining the Social Security system. While most states had state pension plans before they opted into Social Security, they often excluded significant numbers of public employees. My study thus focuses on those state and local government employees who were excluded from pre-existing state pension plans, and then became eligible for Social Security either through a Section 218 Agreement of their employer, or in 1991 when coverage became mandatory. I plan to use asset data from the Survey of Income and Program Participation (SIPP) matched to the Social Security Administration’s Summary Earnings Records (SER). The SER data will allow me to determine exactly when an individual started paying Social Security contributions, and thus became eligible for Social Security benefits. In a difference-in-difference design, I can then compare asset changes of those newly eligible for Social Security to asset changes over the same period of state and local government employees with similar characteristics who were previously covered either by Social Security or a state pension plan. Due to the fact that states and municipalities entered into and left Section 218 Agreements idiosyncratically throughout the period from 1954 to 1990, my research design uses multiple eligibility changes to estimate the crowd-out effect of Social Security on private saving.

Back to 2011 Dissertation Fellowship Recipients