The Lifetime Impacts of the New Deal's Youth Employment Program

Abstract

We study the lifetime effects of the first and largest American youth employment and training program in the United States-the Civilian Conservation Corps (CCC), 1933-1942. We match newly digitized enrollee records to census, World War II enlistment, Social Security, and death records. We find that longer service in the CCC led to improvements in height, health status, longevity, geographic mobility, and lifetime earnings but did not improve short-term labor market outcomes, including employment and wages. We address potential selection into CCC duration using several approaches, most importantly two newly developed control-function approaches that leverage unbiased estimates of the short-term effects of a randomized controlled trial of Job Corps (the modern version of the CCC). Our findings suggest that short- and medium-term evaluations of employment programs underestimate effects because they fail to capture lifetime effects and often ignore or underestimate health and longevity benefits that increase in magnitude at later ages.

Figure I

Distribution of Service Duration in the Civilian Conservation Corps Records and Jobs Corps We exclude durations greater than three years (less than 1% of the observations) in this figure. Mean duration in this sample is 9.44 months (std. dev. 7.47) for CCC and 5.8 months (std. dev. 6.6) for Jobs Corps. In Panel B, we exclude individuals who were not assigned to treatment and therefore have zero duration.

Figure II

Distribution of Reason for Discharge Values on top of the bar graph are the mean duration (in years) for each category: EOT (end of term), Emp (employment outside the program), COG (convenience of the government), UrgProp (urgent and proper call), Desert, Rej (rejected), No Rec (no record). Reasons for Jobs Corps were harmonized to match with CCC's reasons for discharge.

Figure III

Determinants of Duration Estimates and 95% confidence intervals plotted for coefficient estimates of selected variables from regressing duration on various individual, camp, and peer characteristics. Coefficients in diamonds are statistically significant at the 95% level. Mean duration for the estimation sample is 0.84 years for CCC and 0.49 years for Jobs Corps. Full results of the regression estimates are shown in Online Appendix Table III.

Figure IV

Binscatter Plots of Long-Term Outcomes Authors' computations based on death records (Panel A) and/or administrative program data matched to the Master Beneficiary Records (Panels B–D) and using the binscatter methodology of Cattaneo et al. (2024). It plots each variable controlling for birth year. We pick the number of bins (20 for Panel A, 10 for Panels B–D) and implement direct-plug-in data-driven choice of the optimal degree of polynomial and smoothness constraints for confidence band and bin means.

Figure V

Visualizing Sensitivity to Controls We follow the methodology of Chan, Gentzkow, and Yu (2022) to visualize the sensitivity of our coefficient estimates on the choice and order of covariate inclusion. We split our covariates into eight groups and run the OLS regression using every combination of covariate groups, varying the number of groups included in total. We exclude cohort (birth year) fixed effects from the eight groups and include them in all specifications. The x-axis represents the number of covariate groups to be included, with 0 only including cohort fixed effects and 8 including all covariate groups. For example, for four control groups, the number of specifications would be 8 choose 4 = 70. The y-axis shows the average coefficient estimate (solid line) and minimum and maximum estimates (dotted/dashed lines).

Figure VI

Effect of Service Duration on the Probability of Survival to Different Ages On the left y-axis, this figure reports the coefficients (and standard errors) from running linear regressions of the probability that the person survived to a given age on duration, where age ranges from age 45 to age 90. The regressions use the administrative data we collected and control for all observables at baseline (see Table III for details). On the right y-axis we plot the survival rate.