The Draft in America,
1915-1973
\label{the-draft-in-america-1915-1973}
Despite relying on military drafts throughout its history, the United
States shifted slowly away from the draft in the 1970s; for all
practical purposes, the draft ended in 1973 and since that period the
U.S. has recruited military personnel and maintained an All
Volunteer Force (AVF).11Although draft numbers were assigned
through 1975 covering men born through 1956, no men were drafted into
service after 1973. However, the Selective Service has continually
required that all men register upon turning 18; this pool exists
should a draft become necessary. See Rostker (2006) for a discussion
of the transition and a complete history of the All Volunteer Force in
the United States. The current AVF model involves selection into the
military—some who apply to enlist are deemed ineligible due to low
test scores, poor health, or other issues; at the same time, military
service is relatively rare among the most advantaged members of society.
Additionally, military service today is much more common among those
whose parents served in the military (Warrior Class, 2017).
Prior to the advent of the AVF, a draft was enacted surrounding each of
the major conflicts during the period 1915-1975 (in fact, the draft was
also used in earlier periods; see Sohani and Vafa, 2010 as well as
Barry, 2013 for more information).
At first glance, the draft appears to avoid many selection issues,
creating a situation in which enlistees could be viewed as a random
sample of the population of young males. While there is an element of
randomness in the draft, but during each era some young men were
ineligible to be drafted. 22For a discussion of the issues
surrounding selection into the military during the Vietnam era, see
Angrist (1990). Date of birth is closely linked with probability of
service, but those who were drafted were also screened on measures
related to health and cognitive function. Methodologically, this
means that some of our results may lack external validity, but by
comparing veterans to other veterans we avoid some issues of potential
endogeneity. However, it is worth noting that in different eras,
different proportions of young men served in the military. Below, we
briefly describe the draft during each major conflict relevant to our
data.
In World War II, about 16 million men served in the military; the
overall population of the US at the time was about 133 million. The WWII
military experienced extremely rapid growth at the advent of the war,
growing twenty-fold in 3 years. During this period, the majority of
young men served in the military. At the end of the war, one might
expect that all service members would have exited at around the same
time (their commitment was for the length of the war plus 6 months). In
fact, there is some variation in timing of exit. While there was
significant public pressure to release service members quickly, it was
not possible to process all service members simultaneously;
demobilization began in May 1945 and continued through 1946. During this
period, the size of the military reduced by about 87 percent. Today,
only about 10 percent of WWII veterans are living (va.gov); thus, this
group will likely not make up most of our sample, but because of
the sheer size of the military they still form a relatively large
sub-sample.
During the Korea conflict, some 6 million served; at least 1.5 million
were drafted and many of the others were draft-induced volunteers. Terms
of service were generally 24 months, allowing substantial variation of
economic conditions at exit. About 2.3 million Korea-era veterans are
alive today.
The Vietnam conflict was lengthy in nature. During that era, some 8.7
million service members served; about 3.4 million were deployed to
Southeast Asia. Draftees generally served in the Army for 24 months;
volunteers chose their branch and generally served for 3 or 4 years.
Large numbers of men were drafted in 1970, 1971, and 1972. Thus, there
is considerable variation in the timing of exit among those who were
drafted (and among draft-induced volunteers). At the end of the
conflict, the size of the military decreased dramatically. Some 7.4
million veterans from this era are alive; these veterans make up a large
proportion of today’s veteran population although some of them are too
young to be included in our sample (see next section).
Figure 1 traces out the size of the US (active component) military
endstrength, as well as the population of young men (aged 19-22) and the
ratio of young men to the endstrength. This figure demonstrated the
overall population growth during the time period, as well as the
dramatic changes in the size of the endstrength between periods of
conflict and peaceful periods. The ratio of young men to the endstrength
was highest during the period prior to WWII. Despite the relatively
small population of the country during this period, the standing
military was very small (the endstrength in 1930 was about 250,000
compared to about 12 million at the peak of WWII and roughly 1.3 million
in recent years).33The endstrength in Figure 1 includes only
those serving in the Active Components; the sizes of the Reserve
Components also have fluctuated substantially over time although these
components generally were less central to warfighting efforts in the
first half of the 20th century. During WWII, the
endstrength was substantially larger than the population of young men
(many who served were well outside the 19-22-year-old age range); during
other periods included in Figure 1, the ratio of young men to
endstrength generally exceeded 1. Immediately following WWII, the
services drew down to the point that endstrength dropped below 2
million. At this point, the population of young men was about 3 times
the size of the endstrength. In many of the years encompassing the
Korean and Vietnam wars, the endstrength remained above 2 million; the
size of the force increased during each period of conflict and then
returned to something closer to 2 million. The size of the force fell
around the advent of the AVF, and the general trend has been downward
since 1980 while the overall size of the US population has increased.
Today, the ratio of endstrength to young men is about 0.15, a ratio not
seen since prior to WWII.44Given the increasing military role
filled by women, the ratio of the total youth population to the
endstrength may be the more relevant metric; today, that ratio is
approximately 14 to 1. By these measures, the probability of serving
in the military today is markedly lower than it was during the eras
included in Figure 1. In summary, Figure 1 implies that during the
periods of conflict relevant to our analyses, a relatively high
proportion of men joined the armed forces; during WWII, military service
was in fact the norm. To help control for the differences in the veteran
population during different eras, we also include the variable that
measures the ratio of endstrength to the young male population in our
models.55We calculate this ratio by comparing the size of the
active component endstrength to the number of men aged 19-22 in the
US.
Figure : Young Men in the United States and the US Military Endstrength
[CHART]
NOTES: Authors’ tabulations based on Census data, Population
Representation in the Military Services , various years; Selected
Manpower Statistics , various years; Census data, various years.
The ratio of endstrength to population effectively describes the
proportion of young men who will serve in the military. This ratio is
driven by world events, rather than by individual decisions. Thus, it
offers a way to correct for the potential endogeneity of veteran status.
We estimate a simple logistic model that explains veteran status among
all males in our dataset as a function of this ratio, the unemployment
rate at age 19, and race/ethnicity. The ratio of endstrength to the
population of young males has a statistically signifncat and substantive
effect on the probability of being a veteran and the unemployment rate
at age 19 has a small positive effect on the probability of being a
veteran (see Table A.1, Appendix A).66Our model also includes
race/ethnicity; model results appear in Table A.1, Appendix A. Next,
we discuss our general methodology in more detail.
To summarize the key aspect of our methodology, there was certainly some
sample selection in terms of who served or did not serve in the military
during the major conflicts of the 20th century.
However, our strategy compares experiences of veterans who exited the
military during relatively prosperous time with the experiences of those
who exited during less prosperous times. Therefore, as long as the
sample selection out of the military was minimal and sample selection
into the military was somewhat consistent over the eras included in our
data, our strategy will enable us to identify the effects of economic
conditions on a specific population, veterans. Figure 1 suggests that
during WWII, a very large proportion of young men likely served in the
military; during the eras following WWII through Vietnam a relatively
constant proportion of the population appears to have served in the
military (although there was a sharp build-up at the beginning of each
conflict and a drawdown at the end). This suggests that while the
population of WWII veterans may differ from those of later eras, the
veterans of the Korean and VietNam eras may be comparable on many
measures (as noted above, these veterans should make up most of our
sample).77Future specification test: exclude WWII veterans from
models. Finally, during each major conflict, the size of the military
grew rapidly; the military tended to return rapidly to its pre-conflict
size (or to downsize further) after each conflict. While draftees could
remain in the military after the end of the conflict, the rapid
downsizing meant that most did not remain, and that former
draftees/volunteers generally had few options to extend service (or time
their exits). Nonetheless, we utilize a 2SLS strategy to correct for
this potential endogeneity.