Education, Marriage, and Social Security
by Prasanthi Ramakrishnan, Washington University in St. Louis
The optimal design of the retirement system has been a topic of discussion in the last few decades, primarily due to the current financial state of the Social Security system. However, the retirement system, set up in the year 1935, was based on the society that existed then of ‘single-earner’ couples, where there is household specialization with one spouse participating in market work and the other spouse in home production. Given the sharp decline in marriage rates as well as a rise in female labor force participation, this paper seeks to answer two main questions: (a) how does incorporating education and marriage responses to retirement policy change the optimal tax and retirement policy? (b) will the current system make the GenX (born in 1970-79) worse off than the Silent generation (born in 1940-49)?
To this end, we first document the dependence of individuals on the social security system using the CPS and Health and Retirement Survey (HRS) data, and find that elderly single men and women in the bottom 25th percentile are the most dependent on this system, deriving close to 70 percent of their total income from this. We also document trends in inequality and poverty, which show us two important aspects: first, the income and asset inequality in the old age has risen over cohorts; second, the patterns vary significantly by education and marital status, with low-educated and single individuals being the worst-off.
We then build a structural model to disentangle the role of household structure and labor force participation in old age outcomes. We develop a life-cycle model with endogenous human capital accumulation, consumption, savings, labor force participation, and marriage. The key mechanism, underlying the model, is that with the increase in returns to education (due to the rise in female labor force participation), the returns to marriage have changed. This implies that who decides to marry who is different, which has direct implications on a retirement system that provides benefits based on marital status.
Using the structural model, we will estimate welfare under the current regime of the retirement system as well as compare it with counterfactual regimes, while ensuring that the policies promote the goal of long-term solvency, such as: first, a system which is marriage neutral (no spousal or survivor benefits) and second, a system with caregiver benefits (where there is a concession given for years when the earner was ”caring” for someone – an elderly, or a child).