Document Type

Discussion Paper

Date of Paper

Fall 10-25-2023


We conduct the first quantitative assessment of federal college subsidies during the 1930s. Overlapping generation households invest in children’s education to maximize multigenerational utility, and the government subsidizes college to maximize enrollment subject to a budget constraint and recipients satisfying ability and income qualifications. A modelling innovation assigns children educational ability through a random regression to the population mean correlated with father’s presumed ability ranking via his percentile in fathers’ earnings distribution. Simulating the theoretical model, the equilibrium that replicates actual education distributions estimates federal college subsidies increased graduation rates of the cohort of White Americans reaching college age during the 1930s by 22.12% for men and 19.16% for women; the mean ability of subsidy recipients exceeded nonsubsidized students’ mean .4 s.d. The program favored middle income groups. Most benefits accrued to high ability students with fathers in the 4th through 6th deciles of fathers’ earnings distribution. The subsidies had no effect on the graduation rates of high ability students in the bottom two deciles of fathers’ earnings. A more universal government policy that maximized stipends subject only to the budget and income criteria would have increased annual stipends by about 50 thousand while only decreasing college students’ mean ability .13 s.d. Gender biases favoring higher male graduation rates remain a puzzle.


Special thanks for the research assistance of Diya Patel.