Going to War and Going to College: Effects of World War II

Using a dummy variable indicating birth after the third quarter of 1927 (these men would have turned 18 after V-J day) as an instrument for the fraction serving variable underscores the fact that the sharp break in military manpower demand occurring in the fall of 1945 identifies our results. The fact that the OLS (Panel A) and IV (Panel B) point estimates are virtually identical make it quite clear that what identifies these models is the comparison of outcomes for men who were likely to begin service before and after V-J day. The inclusion of covariates for quarter of birth, shown in columns (2) and (4) and other even-numbered columns in this table, allow for cross-cohort differences in educational attainment by quarter of birth. The inclusion of such effects does not move the point estimate for the 1928 interval. Similarly, we have included a quadratic trend (in addition to the simple linear trend) and found that our results are not sensitive to this specification choice. Moreover, adding (or subtracting) an additional age cohort around the 1923 start point makes little difference in the estimated effects, as there is very little variation in the share serving and the educational outcomes among these men.
Estimates extending the period of analysis through the 1929, 1930, 1931 and 1932 are shown in columns (5)-(12). Particularly in the OLS estimates, there is a noticeable drop in the magnitude of the World War II effect with the addition of the 1929 birth cohort, as the effect on the years of college completed drops from .15 years to .11 years. Estimates that use the extended time horizon – effectively comparing outcomes in cohorts with a positive level of World War II participation with those with no participation – are conceptually closest to the desired treatment effects. However, the longer time series estimates also introduce an increased opportunity of specification error.
In attempting to measure the effects of World War II and the G.I. Bill on the educational attainment of returning veterans, it is important to specify the control group – and the expected experiences of the control group – carefully. To answer the question of the whether the combination of World War II service and the G.I. Bill increased educational attainment, we need to compare the veterans to the control of “no service, no benefits.” If participation in the Korean conflict had little effect on educational attainment for the cohorts born prior to 1930 that were at risk for service in World War II, the simple comparison of World War II veterans to non-World War II veterans will accomplish this objective. However, those not serving in World War II were at increased risk for participation in the Korean conflict and service in Korea brought both the hardship of the potential interruption of education and the benefit of the G.I. Bill for veterans of the Korean conflict. For men born in 1927 and 1928 and starting service in 1950 (when they were between ages 22-23), most would have finished their secondary education. We would expect the Korean War experience to have fewer adverse effects on educational attainment, particularly at the secondary level, than service in World War II. At the same time, these men were eligible for generous educational subsidies and many, did, in fact use them (see Appendix Table 1). Thus, it seems likely that the Korean War service had a positive effect on post secondary attainment for these cohorts. Beyond these direct effects of the Korean conflict on the educational attainment of the cohorts of men at risk of service during the conflict, the Korean conflict may have had “spillover effects” inducing some to obtain extra education as they used the availability of educational deferments to avoid service.