Going to War and Going to College: Cohort estimates

Between cohort estimates provide a mechanism for reducing the potential upward bias attributable to the greater selectivity of veterans relative to nonveterans and offer an alternative to within-cohort comparisons. Graphically, the thought experiment is to ask how collegiate attainment changes with changes in the manpower demand in military service. Figure 3 presents these trends over a fifty-year horizon of birth cohorts for white men, using data from the 1970 Census. We consider three measures of educational attainment: high school graduation, college completion, and years of college, with the last two measures capturing trends in the postsecondary sphere. Educational attainment is shown for men and women on the left axis, while the share of men who were veterans is superimposed with the scale appearing on the right axis. If changes were purely secular – such as an increase or decrease in the demand for skilled workers – we might expect the trends for men and women to trend upwards in similar fashion. On the other hand, if the combination of military service and veteran benefits has sizable effects on educational attainment, we expect to see for men, but not for women, above trend levels of educational attainment for cohorts that would have most benefited from the G.I. Bill.
At the secondary level, which is shown in the first panel, the demands of the war had a deleterious effect on the prospects for finishing high school for men who turned 18 between 1940 and 1948. For the postsecondary attainment panels, what is most striking about these graphs is that for the three decades of birth cohorts between 1912 and 1942, the trends are quite smooth, moving upward at a steady pace, with the growth somewhat greater for men than for women. While dramatic changes in collegiate attainment mirroring changes in the manpower demands of the war are not visible, one can find patterns that are suggestive of a moderate effect of wartime service in educational attainment. Thus, for example, the growth across cohorts in the fraction of men completing college seems to have accelerated a bit for the cohorts born in the early 1920s. These are the cohorts that were both very likely to see service during the war, but would have returned to civilian life at a young enough age that it would have been plausible for them to avail themselves of the educational benefits due them as veterans.
There is also some suggestion in these graphs that the growth in college completion rates slowed a bit for cohorts that would have reached college age in the immediate post war period, then accelerated a bit for cohorts that would have turned 18 during the Korean conflict, and fell for cohorts that would have turned 18 after the conflict was over and then rose again for cohorts turning 18 during the 1960s. It is hard to know what to make of these long-term trends. The acceleration of college completion rates for cohorts born during the early 1920s could have been partially due to improvements in the economy as the country came out of the depression. Much later, the rise in college completion rates for cohorts turning 18 during the 1960s could plausibly be attributed to the exceptionally strong market for college graduates that existed at the time. Plainly, changes over such an extended period of time in the nature of G.I. benefits and selection into the military limit the usefulness of between cohort estimates over such a long time horizon.

Figure 3

Figure 3