Working Paper: NBER ID: w26407
Authors: James J. Feigenbaum; Hui Ren Tan
Abstract: What was the return to education in the United States at mid-century? In 1940, the correlation between years of schooling and earnings was relatively low, less than it had been in 1915 or than it would be in later decades. In this paper, we estimate the causal return to schooling in 1940, constructing a large linked sample of twin brothers to account for differences in unobserved ability and family background. Though imperfect, the twins identification strategy allows us to compare the return to education to recent studies implemented similarly. We find that the return to education was relatively low in 1940, with each additional year of schooling increasing labor earnings by approximately 4%. Returns to education were evident both within and across occupations and were higher for sons born to lower SES families.
Keywords: Return to Education; Twins; Causal Inference; Labor Earnings
JEL Codes: J2; J3; N3; N32
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| additional year of schooling (I23) | labor earnings (J31) |
| twins identification strategy (J79) | control for unobserved ability and family background (C29) |
| return to education in 1940 (I26) | lower than returns in later decades (N22) |
| education (I29) | various economic outcomes (P42) |
| twins identification strategy (J79) | mitigate selection bias (C52) |