"Public Pensions and Retirement: Evidence from the Railroad Retirement Act” (Job Market Paper)
Male labor force participation (LFP) 65+ declined by almost 50 percentage points over the 20th century and the share receiving social insurance increased from roughly zero to near universal, yet the precise role of public pensions in driving the decline is still an open question. This paper develops the earliest estimates of the effect of Defined Benefit (DB) public pension incentives on retirement timing by examining the Railroad Retirement Acts (RRA) of the 1930s, which nationalized private pensions, instituting an incentive structure comparable to Social Security. Using linked Decennial Census records between 1920-1930 and 1930-1940, I examine LFP by worker's industries a decade earlier. I estimate the reduced form effect on retirement timing using an age-specific difference in differences research design leveraging variation in industry, time, and cohort. Railroad workers differentially exited only at eligible ages, explaining roughly 12% of the previously unexplained LFP decline 65-74. I next estimate elasticities of elderly nonparticipation with respect to benefits by exploiting the switch to progressive benefits in wages. The homogeneous estimate is roughly 0.2, far lower than existing pre-1970 estimates but similar to those from Social Security today. Elasticities are heterogeneous by private wealth and follow an inverse U-shape. Responsiveness is closer to 0.1 for renters and for workers with home values above the 80th percentile, with higher values in between. These estimates suggest the 100% increase to real Social Security benefits in the 1950s explains at most 25% of the historical decline in that decade. It appears other changes, notably coverage expansions, were more important in driving historical LFP declines.
“The Expansion of Social Security and the Decline of Elderly Public Assistance” (Joint with Daniel Fetter)
The growth of social insurance and transfer programs was among the most consequential changes in the 20th century United States, and a central element of this growth was the change in the level of government that administered and funded transfers. We investigate the expansion of the nationally-administered Old Age and Survivors Insurance (OASI) program -- commonly known as Social Security -- and its role in the decline of the state-administered, means-tested Old Age Assistance (OAA) program from 1940 through 1955. In the absence of changes in state OAA policies, means-tested OAA programs would shrink as Social Security increased the resources of the elderly, meaning that the net cost of Social Security transfers and the increase in the resources of the elderly would both be smaller than the dollar value of Social Security transfers would suggest. The analysis exploits variation across states in the size of Social Security expansions due to exclusions of certain industries from coverage in the early program. Overall, OAA programs shrank in response to the growth of Social Security, but states also responded to these ``savings'' by liberalizing eligibility and payment policies. However, there was stark regional variation in states' responses. Most notably, most of the decline in OAA attributable to the expansion of Social Security was driven by Southern states.
Work in Progress:
Two-Parent Targeted Welfare and Family Structure in the Mid-20th Century: Evidence from the Implementation of AFDC-UP
This paper evaluates the effect of expanding Public Assistance to two-parent families on family structure dynamics. Under the Social Security Amendments of 1961, states were given the option to extend the Aid to Families with Dependent Children program to provide an unemployed parent segment (AFDC-UP) that covered cohabitating parents under specific eligibility criteria. To identify the effects of AFDC-UP on fertility and marital behavior, this paper leverages a difference in differences research design that exploits the staggered adoption of the program across 18 states between 1961 and 1975. The main outcomes are from newly entered data on nonmarital births by state, year, age, and race between 1952 and 1988. Semiparametric event studies show no evidence of pre-trends in nonmarital fertility up to 9 years before adoption, but persistent reductions immediately after for nonwhite women. When adjusted for changes to general fertility, nonwhite women in their 30’s experienced the largest decline in the nonmarital share of births. This paper provides evidence that the increase in the share of births outside of marriage beginning in the mid-twentieth century would have been even more pronounced in the absence of expansions in federal Public Assistance to two-parent families.
“Who Benefits from Federal Welfare Spending? Evidence from the Introduction of Progressive Cost Sharing” (Joint with Andrew Goodman-Bacon)
This paper examines whether the federal government can induce states to offer higher benefits through greater generosity in matching funds. We examine the initial switch to progressive federal matching of state public assistance expenditure in 1958 for the four federalized programs Old Age Assistance OAA, Aid to Families with Dependent Children (AFDC), Aid to the Blind (AB), and Aid to the Permanently and Totally disabled (APTD). The formula based funding on how state per capita income related to national per capita income, setting the precursor for Medicaid financing. Our empirical strategy compares higher and lower income states before and after 1958. States could respond by increasing welfare spending, which would constitute evidence of flypaper effects. They could also decrease taxes or increasing other spending. Preliminary evidence suggests progressive cost sharing was not effective in achieving more equitable benefits, but may have increased medical vendor payments, the 1950s public insurance component of welfare