THE DEMOGRAPHIC LEGACY OF THE VIETNAM WAR
Martha J. Bailey and Eric Chyn
May 12, 2017
Abstract:
This paper quantifies the impact of the Vietnam War on fertility rates in the late 1960s and early 1970s. The
availability of the hardship deferment (III-A) for paternity created a powerful incentive to father a child for
men wishing to dodge the draft. By 1969, over 4 million hardship deferments had been granted—more than
twice as many as education deferments (II-S). New time series and estimates from the 1969 draft lottery
suggest that draft risk significantly increased fertility rates, especially among childless women likely to be
partnered with draft-eligible men. Fertility rates plummeted after President Nixon’s elimination of the
“paternity deferment” in 1970—a policy change that had particularly large effects in states like California
and New York where abortion was legalized the same year. These results suggest considerable caution in
attributing declines in fertility rates in the early 1970s as well as improving the circumstances and behaviors
of the “marginal child” to the legalization of abortion.
Contact Information:
Bailey: Department of Economics, University of Michigan, 611 Tappan Street, Ann Arbor, Michigan
48109; [email protected]; Website: www-personal.umich.edu/~baileymj. Chyn: Department of
Economics, University of Virginia; [email protected]; Website: ericchyn.com.
Acknowledgements:
During work on this project, Chyn was supported by the NICHD (T32 HD0007339) as a UM Population
Studies Center Trainee. We gratefully acknowledge the use of the services and facilities of the Population
Studies Center at the University of Michigan (funded by NICHD Center Grant R24 HD041028). We are
also grateful for helpful comments from Emily Beam and John DiNardo. Exceptional research assistance
was provided by Anna Erickson, Rachel Bacille, and Annie Wentz.
The Vietnam War had a direct impact on the nearly 9 million men who served in the military, but
its effects rippled through the civilian population as well. Dodging the draft to avoid serving in an unpopular
war became a pre-occupation of many young men. Many enrolled in college to obtain student deferments
(II-S) (Card and Lemieux 2001) and some even took more extreme measures such as committing felonies
(Kuziemko 2010). Fathering a child presented another less studied strategy to dodge the draft. After
President Kennedy’s Executive Order 11098 on March 14, 1963, paternity qualified men for a “hardship
deferment” (III-A) to allow his family to avoid the “undue hardship” imposed by the loss of financial
support (Tatum and Tuchinsky 1970).1 Hardship deferments were extremely popular, with over 4 million
men holding them in 1969—more than twice the number with education deferments (Figure 1).
This paper presents new evidence that the availability of hardship deferments led to large increases
in fertility rates during the Vietnam era. Time series show a pronounced fertility notch in the late 1960s
driven by elevated numbers of first births (hardship deferments only required men to have one child) among
women in their early twenties (women likely to be partnered with draft-eligible men). Following President
Nixon’s Executive Order eliminating paternity as grounds for hardship in April 1970, fertility rates
plummeted—especially first births among women ages 20 to 24.
Following Angrist (1990) we use the 1969 Vietnam draft lottery to draw inferences about the causal
effect of hardship deferments on childbearing. This lottery randomly assigned draft numbers using men’s
birth dates, which allows us to quantify the effect of draft risk on childbearing nine to 11 months after the
draft lottery. Using the 1973 and 1976 National Surveys of Family Growth (NSFG), we find that women
partnered with men randomly assigned a high risk numbers in the 1969 lottery were over40 percent more
likely to give birth nine to 11 months later than women whose husbands had the lowest risk of induction.
This effect is driven by childless women whose likelihood of giving birth nine to 11 months later was
elevated by almost 90 percent.
1
Men receiving hardship deferments were classified as III-A. The requirement for this deferment was that the
registrant demonstrate a “bona fide” relationship with his child. Men receiving deferments based on college enrollment
were classified as II-S.
Demographic Legacy of the Vietnam War
1
This response was driven by men at high risk with enough time to have their classifications shifted
to III-A—roughly those in the middle third of the draft number distribution.
Consistent with the
interpretation that these fertility responses were driven by the lure of deferments, we find no such patterns
by lottery number for women with children—husbands would have already been eligible for hardship
deferments in 1969. We also find no evidence of fertility responses to draft risk in the 1970, 1971, and 1972
lotteries which took place after Nixon had eliminated paternity as a grounds for new hardship deferments
in April 1970.
A final set of results examines how these results affect the standard interpretation of the post-1970
fertility decline, which until now has focused on the role of abortion legalization. If draft-dodging had
similar effects on fertility rates in states that repealed abortion restrictions in 1970 relative to those that
legalized abortion with Roe v. Wade in 1973, these results should matter little for this literature. However,
if draft-dodging had differential effects on fertility rates in early repeal states, standard research designs
may misstate the effects of abortion legalization on fertility rates and the outcomes of the “marginal child.”
Using Gallup polls from the period, we find that anti-war sentiment was largest in states that
repealed abortion restrictions in 1970—states such as California, New York, and Washington. Similarly,
our analysis of the 1969 draft lottery shows that the childbearing response to randomly assigned draft risk
was significantly higher in states that repealed repeal abortion in 1970. That is, fertility rates appear were
significantly more elevated in states such as California, New York, and Washington where anti-war
sentiment was higher. The implication is that Nixon’s elimination of the “paternity deferment” in April
1970 would have led to a larger fertility declines in those states after 1970.
How much these estimates revise standard claims about the role of abortion legalization depends
upon the external validity of the local average treatment effect (LATE) of draft risk from the 1969 lottery.
We use two complementary approaches to generate such counterfactuals: we directly extrapolate from the
lottery-based LATE and also use a polynomial extrapolation that uses only pre-1965 and post-1973 fertility
rates to infer trends in the absence of the Vietnam War. Using either approach, we find that differential pretrends in fertility rates in states legalizing abortion in 1970 versus those legalizing with Roe v. Wade
Demographic Legacy of the Vietnam War
2
significantly affect inferences. Both methods result in a much smaller, statistically insignificant effects of
abortion legalization on fertility rates from 1971 to 1973. Adjusting these differential pre-trends, however,
also suggests that Roe v. Wade may have led to an overlooked decrease in fertility rates of around 2 percent.
The imprecision of these estimates caution against strong conclusions, but this suggestive evidence helps
reconcile an apparent puzzle, namely that the 1970 legalization of abortion in a handful of states led to 4 to
5 percent declines in birth rates, whereas Roe v. Wade just three years later did not.
We conclude that there is considerable ambiguity in the interpretation of the causes of fertility
declines in the early 1970s and the reasons for changes in the circumstances and behaviors of the “marginal
child” (Gruber, Levine, and Staiger 1999, Charles and Stephens 2006, Rotz 2012). The temporal proximity
of abortion legalization and the elimination of the paternity deferment in 1970 make disentangling these
two policy changes challenging—if not impossible, especially for the state of New York. Although it is
likely that abortion legalization in 1970 and the elimination of the paternity deferment had effects, the
relative magnitudes of these effects is unclear. Moreover, many of the “marginal children” considered in
the abortion literature may have been conceived under great duress to fathers (and mothers) desperately
seeking to avoid the draft. To the extent that these circumstances impacted children, the demographic legacy
of the Vietnam War may extend to the present.
I.
SHORT HISTORY OF VIETNAM ERA SELECTIVE SERVICE POLICY CHANGES
In the early 1960s, nearly all 18 to 26-year-old American male citizens and most noncitizens living
in the U.S. and its territories were required to register for the draft. Presidential Proclamation 2799 (July
20, 1948) required men to register at their nearest draft board within 5 days of their 18th birthday (Tatum
and Tuchinsky 1970).2 Following registration, the U.S. Selective Service classified registrants as available
for civilian (I-A-O) or military service (I-A), deferred, or ineligible for service. Men who enrolled in college
could apply for II-S deferments, and men who could prove a “bona fide” relationship with their children
2
The law now requires registration within 30 days of one’s 18th birthday.
Demographic Legacy of the Vietnam War
3
could apply for III-A “hardship” deferments.3 After receiving III-A deferments, men would not be drafted
so long as they maintained this relationship. All men classified as I-A were considered available for military
service, and local draft boards were instructed to prioritize calls within the I-A class as follows (from highest
to lowest priority): delinquents, volunteers, single men, and married men, with the oldest in each category
to be selected first.4
In 1965, the rapid escalation in the scale of the Vietnam War significantly changed the likelihood
that I-As would be called. President Johnson supplemented the 23,000 military advisers already on the
ground by 1965 with the first brigade of combat troops in March and rapidly increased ground force levels
throughout 1965 (VanDeMark 1991, page 92). Figure 2 shows that mobilization resulted in sharp increases
in inductions. In response, many more men applied for deferments, especially for education and paternity.
By 1969, over 4 million men held III-A deferments—more than twice the number holding II-S education
deferments.
As popular support for the Vietnam War erupted into protests and civil unrest, newly elected
President Nixon followed through on his campaign promises to increase the transparency of the draft,
equalize the risk of induction across race and class, and limit the duration of each man’s eligibility for the
draft. His first major reform, announced on May 13, 1969, asked Congress to change the order of calls from
youngest to oldest and limit eligibility for draft to one year. He also proposed a lottery which would
distribute the risk of induction equally “by lot” among all who were eligible in a particular year—though
he did not announce when the lottery would be held. Both of these changes were signed into law in the
Selective Service Amendment Act of November 26, 1969. The first draft lottery, held on December 1, 1969,
was used to call men for examination in 1970.
3
President Kennedy expanded the grounds for hardship to include paternity in Executive Order 11098 in 1963. Men
holding these deferments were often called “Kennedy fathers.”
4
The first four categories included men ages 19-25 only. There were two additional priority categories within the I-A
and I-A-O classification after married men: men ages 26, youngest called first, and men between age 18 and six months
and age 19, oldest called first (Selective Service System 1960).
Demographic Legacy of the Vietnam War
4
The lottery specified that all I-A classified men who were 19 to 25 years old (born from 1944 to
1950) would be called first, in order of their birthdates and then permutations of their last name, first and
middle names. Capsules with slips of paper with numbers for each of the 366 days of the year (the extra
day is February 29th) were drawn sequentially from a glass jar. The first date drawn, September 14, assigned
non-deferred men born from 1944 to 1950 on that date to be called first. The second date drawn, April 24,
assigned non-deferred men born in the same cohorts to be called second. This process of drawing was used
to assign induction priority for each birth day of the year. Because of the birth days were inadequately
mixed, subsequent analysis revealed that later months tended to have higher numbers (i.e. December
birthdays were less likely to be called than January birthdays) (Starr 1997). The 1969 lottery did not
supersede existing deferments.5
Nixon’s second initiative was announced on April 23, 1970, in Executive Order 11527 in his
Special Message to Congress on Draft Reform (Anderson and Tollison 1991, Nixon 1970). He directed that
no future deferments would be granted on the basis of employment (occupation, agriculture) or paternity
as a grounds for hardship, except in cases where a local draft board determines that “extreme hardship”
would result. However, all those holding occupational or hardship deferments as well as any who would
have been granted deferments from pending applications were grandfathered in. His April 23 speech also
requested that Congress restore Presidential discretionary authority on the deferment of students seeking
BAs, but his Executive Order eliminating student deferments was not issued until later in 1970. Subsequent
draft lotteries were conducted on July 1, 1970 for those born in 1951 and on August 5, 1971 for those born
in 1952. Lotteries were also conducted after 1971 but never used to call anyone to service.
5
The Selective Service Amendment Act of 1969 (Public Law 91-124) gives the sequence of induction as follows: (1)
men 19-25 (either unmarried or married after August 26, 1965) in order of their random selection; (2) men 19 to 25,
married before August 26, 1965, in the order of their random selection.
Demographic Legacy of the Vietnam War
5
THE EFFECTS OF THE VIETNAM WAR ON ADULT OUTCOMES
II.
A large literature considers the effects of the Vietnam War on veterans.6 But the literature most
closely related to this study examines how the Vietnam War affected the adult outcomes of non-veterans,
especially men who dodged the draft. Card and Lemieux (2001) find that the availability of II-S deferments
increased college enrollment during the Vietnam War, and Kuziemko (2010) finds that many lower-SES
men (presumably those who could not obtain a II-S) committed crimes to avoid serving in the war. She
finds that black and low-SES men with “bad” lottery numbers (i.e. numbers with high risk of induction)
had higher rates of self-reported delinquent behavior and were overrepresented in Georgia prisons.
Given these extreme efforts to dodge the draft, it is not implausible that many would have fathered
children to obtain a hardship deferment. The literature contains very little evidence on this topic. The one
source we have found is Kutinova’s (2009) differences-in-differences analysis which shows that that
fertility rates increased in 1966 after President Johnson eliminated lower draft prioritization for married,
childless men, as married men shifted to using hardship deferments (III-A). Working against this increase
in fertility rates, Bitler and Schmidt (2012) show that the reduction in marriageable men due to increased
inductions (and perhaps incarcerations) depressed fertility rates in the late 1960s.
Our analysis incorporates elements from each of these studies. We first provide time series evidence
that increases in fertility rates corresponded to mobilization during the Vietnam War and fell with the
elimination of the paternity deferment in 1970. Next, we use Angrist’s (1990) lottery-based design to
estimate the effects of husbands’ draft risk on women’s childbearing 9 to 11 months after the lottery.
6
A number of prominent studies use the Vietnam-era lottery to examine the impact of the Vietnam War on veterans’
post-war earnings, disability, and labor-force outcomes (Angrist 1990, Angrist and Chen 2008, Angrist, Chen, and
Frandsen 2009). Subsequent studies extend Angrist’s draft-lottery instrumental variables research design to examine
the effects of Vietnam-era military service on other outcomes among veterans. Conley and Heerwig (2011) find no
effect of veteran status on housing tenure and mixed evidence of effects on residential stability and extended family
living arrangements. Eisenberg and Rowe (2009) show that Vietnam-era service is strongly associated with increased
smoking in the shorter-term but that these effects fade with age. Lindo and Stoecker (2012) find a positive effect of
Vietnam-era service on violent crime and a negative effect on non-violent crime. Neither Conley and Heerwig (2011)
nor Angrist and Chen (2008) find evidence of an impact of Vietnam-era military service on veteran’s mortality.
Demographic Legacy of the Vietnam War
6
TIME SERIES EVIDENCE THAT DEFERMENTS FOR PATERNITY INCREASED FERTILITY RATES
III.
Time series evidence suggests an important role for the draft in increasing fertility rates. From 1950
to 1980, Figure 3A shows the smooth evolution of the general fertility rate (solid line). Vietnam-era
mobilization corresponds to the fertility notch of the late 1960s. After 1965, the rapid post-baby boom
decline in the total fertility rate slowed and fertility rates even increased briefly between 1968 and 1970.
Consistent with III-A deferments elevating birth rates, Figure 3B shows that much of the increase in
childbearing was due to elevated first birth rates. Figure 3C indicates this was especially the case among
women who would have been the likely mates of draft-eligible men: women ages 20 to 24 and, to a lesser
degree, women ages 25 to 29.7 Reversing a slight downward trend, first births among women ages 20 to 24
stopped falling in 1965 and increased slightly from 1965 to 1970. A modest elevation in first births among
women 15 to 19 and 25 to 29 years old is also evident in this period, where first births appear slightly above
trend. After the elimination of paternity as grounds for a III-A deferment in April 1970, the general fertility,
fist birth rates, and first birth among 20 to 24 year old women dropped sharply.
A comparison of the U.S. and Canadian fertility time series provides additional evidence on the
link between draft avoidance and childbearing. From 1925 to 2011, Figure 4A shows that U.S. and
Canadian total fertility rates were different in levels but experienced similar trends—this is true both before
1940, during World War II, and during the baby boom and early 1960s. Canada did not participate in the
Vietnam War, so a departure in these common trends from 1965 to 1970 would be expected and consistent
with U.S. draft avoidance increasing fertility rates. Accordingly, Figure 4B shows that the Vietnam era
represents a departure from the U.S.-Canadian congruence. As the decline in the U.S. total fertility rate
slowed and then reversed between 1965 and 1970, the Canadian total fertility rate continued to fall.
Moreover, Canadian birth rates exhibit little evidence of the fertility notch seen in the U.S. among women
ages 20 to 24 (figure 4B).
7
We use women’s ages in this analysis, because the age of the father is not available for all births or in all years in
Vital Statistics.
Demographic Legacy of the Vietnam War
7
But, these dramatic changes in fertility rates occurred at a time of tremendous social and
demographic change. Some of the many potential confounders include the uneven diffusion of the birth
control pill between 1960 and 1970 and the legalization of abortion in five states in 1970 and in the
remainder in 1973 with Roe v. Wade.8 A number of papers highlight how increasing access to the birth
control pill (Guldi 2005, Bailey 2006, 2010, 2012, 2013) and the legalization of abortion (Levine et al.
1996, Levine, Trainor, and Zimmerman 1996, Ananat, Gruber, and Levine 2007, Rotz 2012, Joyce, Tan,
and Zhang 2013, Myers forthcoming) in the 1960s and 1970s are associated with significant drops in birth
rates.9 Importantly, abortion (up to the 24th week) was legalized in New York on April 11, 1970, less than
two weeks after Nixon’s elimination of paternity as grounds for deferment on April 23, 1970. This timing
makes it difficult to separate the effects of abortion legalization and the Vietnam draft using standard
approaches (e.g., time series, regression discontinuity, or differences-in-differences analysis), especially for
the state of New York. We, therefore, use the 1969 draft lottery to understand the causal effect of draft risk
on childbearing.
IV.
EVIDENCE THAT PATERNITY DEFERMENTS INCREASED CHILDBEARING FROM THE 1969 LOTTERY
Our research design exploits the random assignment of induction risk in the 1969 draft lottery to
test the causal (although partial equilibrium) effects of hardship deferments on childbearing. If men respond
to greater draft risk by trying to father children, we would expect those randomly assigned to higher draft
risk to be more likely to father children nine months later—conditional upon men being able to obtain a
deferment for paternity before being drafted.
8
See a summary of these state-level legal changes in Bailey et al. (2011) and Guldi (2011).
Joyce, Tan, and Zhang (2013) argues that response to New York’s legalization of abortion affected birth rates in
states within driving distance, with closer states being more strongly affected.
9
Demographic Legacy of the Vietnam War
8
A. Hypothesized Effects of Randomly Assigned Draft Risk on Childbearing
Figure 5A plots the unconstrained response pattern of childbirth as a solid line (labeled “if everyone
has time to conceive”). It shows that men with the lowest two thirds to three quarters of the draft lottery
numbers would have faced the largest incentives to father a child to dodge the draft.
There were, however, constraints on men’s ability to respond. Men with the lowest draft numbers
– those with the highest risk of being inducted on January 1– would have been unlikely to avoid service by
fathering a child. This is because a pregnancy would need to be verified by a physician, the man’s case
would need to have been reopened, documents reviewed by the draft board, and reclassified before he was
inducted. Tatum and Tuchinsky (1970) note that “when an unborn child is the only basis of the III-A
deferment, you must send the board the doctor’s letter before and induction order is issued or you may not
receive the deferment [REG 1622.30(c)(3)]. Despite this regulation, it may be possible to show that you
didn’t learn of the pregnancy until after the induction order was issued. If so, learning that the child is
expected is ‘a change in the registrant’s status resulting from circumstances over which the registrant had
no control’ [REG 1625.2], and the local board may reopen the classification and cancel the induction order.”
The exact timing of classification and induction certainly varied across draft boards, but men with numbers
in the lower third of the distribution likely would have been inducted before February.10
The consequence of this constraint should be a hump-shaped pattern of responses. Men with very
low lottery numbers would not have been unlikely to get a deferment by fathering a child, which is depicted
as a dashed line in Figure 5A.11 However, men middle-range draft numbers (e.g., 121-240) would have had
10
This timing is difficult to document directly, but it was typically very soon after the lottery. Tracy (2005) notes, “A
man with draft a number of 131 was ordered to report for his physical exam on February 18, 1970, and subsequently
classified as 1-A” (p. 45).
11
The best possible scenario for conceiving in time to obtain a III-A deferment would be for a woman to ovulate on
or just after December 1—the day of the draft lottery. The woman’s pregnancy could be verified by a physician around
mid- to late December. Then, paperwork could be submitted to be reviewed by the draft board. Given the holiday
work break and time required for processing and consideration, even this most ideal set of circumstances may not
have been fast enough to obtain a III-A reclassification before January 1. Moreover, later ovulation, delayed diagnosis
by a physician, and any processing delays by a draft board would make it near impossible to obtain a III-A deferment
before receiving their selective service induction notice.
Demographic Legacy of the Vietnam War
9
both the incentive to use fatherhood to obtain a III-A deferment and more time to secure one.12 However,
these men who have the highest draft numbers would have little incentive to father a child if they were
certain they would not be inducted. While the highest number called in the 1969 draft lottery was 195, men
with numbers above this still may have responded to the lottery because there was uncertainty in early 1970
about which numbers would be called.
B. Testing for the Hump-Shaped Relationship between Draft Lottery Number and Childbearing
We test for this hypothesized fertility pattern using the pooled 1973 and 1976 NSFG which contains
the exact date of birth for the first husbands of 5,869 women married to men subject to the 1969 to 1972
draft lotteries (Smock, Granda, and Hoelter 2013).13 Based upon their husband’s birth date, 4,574 were
subject to the 1969 lottery and 1,122 were subject to the later lotteries.
The NSFG conveniently ameliorates several potential sources of bias. First, the NSFG surveys
women about their childbearing and husbands, which avoids the problem of missing responses from men
who deployed to Vietnam (Kuziemko 2010). Second, the NSFG asks women about all of their husbands
which means we can still examine the relationship of her past childbearing to an earlier husband, even if
she is no longer married to him due to a divorce or his death in Vietnam.
The NSFG also has two limitations: it only asks women about the birthday(s) of her husband(s)—
not the fathers of her non-marital children, which means we miss fertility responses of non-married
individuals. This was likely rare—being married was not required for a III-A deferment, but it certainly
helped make the case to the draft board that a man’s absence would impose financial hardship on the woman
and child.14 A second limitation is that the NSFG asks women to recall her husband’s birthday. To the
extent that women misreport, our key independent variable (lottery number) will be measured with error.
12
If this group was called beginning in February, they could father a child anytime between December 1 and January
30. Those with higher numbers may have even had until mid- to late-February to father a child.
13
Results that use current husband to assign draft lottery numbers provide similar results. We cannot use unmarried
women in the analysis, because they were not asked for their partner’s birthdays.
14
It also seems unlikely that a woman would help a man with whom she was not previously partnered try to dodge
the draft. Conceiving a child with an unknown man—especially one who may be inducted anyway—would have been
a tremendous sacrifice for a stranger.
Demographic Legacy of the Vietnam War
10
We use the following linear probability model to examine couples’ responses to randomly assigned
draft risk,
∑
(1)
,
where Y is equal to 1 if the woman had a birth in September, October or November 1970 and Draft Risk
captures risk of induction for different groups of lottery numbers, k, defined subsequently.15 X is a column
vector containing a husband’s month of birth fixed effects (to account for non-random assignment of birth
months) and a survey year dummy variable for 1976 in every specification. These covariates account for
the greater likelihood of lower draft numbers for fathers born in November and December and also for
survey effects. In additional specifications, we also include covariates as a test for random assignment and
increase precision.16 Following Solon, Haider, and Wooldridge (2015), we present weighted results in the
main analysis and unweighted results in the appendix. Heteroskedasticity robust, Huber-White standard
errors are reported.
The point estimates of interest,
, summarize the increased likelihood of having a child nine to 11
months after the draft lottery (a conception within three calendar months of December 1, 1969) relative to
the omitted category of women married to men with a high induction risk and no time to respond. Of
course, any childbearing response to the lottery occurred in the context of multiple options for dodging the
draft. Men with education deferments (II-S), for instance, would have had no incentive to respond by having
a child. Because we do not observe which men held other deferments, our estimates represent intention-totreat effects across men with and without other dodging options. This is a feature of the 1969 lottery that
makes these estimates easier to generalize, because eligible recruits would have had similar deferment
options for dodging the draft in the absence of the lottery.
15
To the extent that December 1969 conceptions resulted in births before September 1970, our analysis will understate
the effects of draft risk on childbearing responses to the draft.
16
However, it turns out that covariates do no increase precision.
Demographic Legacy of the Vietnam War
11
C. Results: Fertility Responses to the 1969 Draft Lottery
Figure 5B summarizes the results of a specification that uses 10 equally sized grouping of draft
lottery numbers (k=10), where the lowest draft lottery number group is the omitted group.
This
specification allows a flexible characterization of nonlinear responses by draft number. As expected, the
fertility responses were small and statistically indistinguishable from zero for women married to men with
very low draft numbers (38 to 109) and the highest draft numbers (297 to 366). Women married to men at
high risk of induction with slightly more time to respond (lottery numbers 110 to 296) exhibit a marked
increase in childbearing. For example, men with draft lottery numbers 110 to 146 (likely called in early to
mid-February) were around 68 percent more likely to have fathered a child nine to 11 months after the draft
lottery relative to the highest risk group.17 Because many couples would have decided to try to conceive a
child before learning that 195 would be the highest number called, fertility rates are similarly elevated for
men above this threshold (lottery numbers 185 to 296), even though they were not, in fact, called.
Table 1A summarizes these estimates for three evenly-sized draft risk groups (k=3): 0-122
(omitted), 123-244, and 245-366. Having a husband with moderate draft risk (defined as lottery numbers
123-244) raises the likelihood of giving birth nine to 11 months after the lottery by about 1.6 percentage
points. This represents a 40 percent (0.016/0.040) increase relative to the mean for women married to men
in the highest draft risk category (numbers 1 to 122). In addition, Table 1 demonstrates that these estimates
are robust to the addition of a range of covariates. Consistent with random assignment, including dummy
variables for the husband’s month of birth and the survey year (column 2) and adding demographic controls
for education, race, mother’s birth cohort and state fixed effects (columns 3 and 4) affect the magnitudes
negligibly. Omitting the NSFG post-stratification weights (which tends to increase the weight of minorities
because they were oversampled) has small effects on the estimated magnitudes (Appendix Table A1).
17
Men with lottery numbers between 110 and 146 were 2.7 percentage points more likely to father a child nine to 11
months after the lottery relative to a baseline of 4 percentage points for men in the lowest draft lottery group (numbers
0 to 37).
Demographic Legacy of the Vietnam War
12
If the hypothesized mechanism for these effects is draft dodging, we anticipate the childbearing
response should be largest among women who were childless before the 1969 lottery. This is because
husbands of women with children would have been already been eligible for the hardship deferment without
fathering another child (although an additional child may have helped convince a draft board). Consistent
with draft-dodging being the motivation for childbearing, Figure 5C shows that the hump-shaped
childbearing response to draft risk is driven by childless women—precisely the group that could alter their
husband’s draft risk the most by conceiving. Table 1B shows that first births increased by more than 85
percent (0.029/0.033) for men with draft numbers between 123 and 244 relative to those in the omitted
category. Moreover, panel D shows the likelihood of a higher-order birth was not significantly elevated for
men with draft numbers between 123 and 244, a result summarized in Table 1C.
An alternative approach to testing whether the hump-shaped response to the draft lottery was due
to draft dodging is to test for hump-shaped responses in subsequent lotteries. Because Nixon eliminated
hardship as a grounds for deferment on April 23, 1970, the women married to men subject to the 1970,
1971, and 1972 draft lotteries had no incentive to conceive to help their husbands avoid service.18 If draft
dodging was the motivation for the childbearing response, draft lottery numbers for women with partners
subject to the 1970, 1971 and 1972 lotteries should not be related to childbirth 9 to 11 months later.
We test this formally by examining the childbearing responses of women married to men subject
to the 1970 to 1972 draft lotteries. Table 2 presents the results which are generally small and statistically
insignificant. Owing to small sample sizes, the effects in Table 2 are not statistically different from those
in Table 1. Small sample sizes reflect the fact that the 1970 to 1972 draft lotteries included only one birth
cohort each (rather than the seven birth cohorts subjected to the 1969 lottery, which limits precision. It is,
however, reassuring that they provide no evidence of elevated childbearing after the paternity deferment is
eliminated.
18
These lotteries occurred on July 1, 1970, August 5, 1971 and February 2, 1972 and included only 19 year olds (the
birth cohorts of 1951, 1952 and 1953, respectively). We omit examination of higher order births as in Table 1 as these
occur so infrequently for this group.
Demographic Legacy of the Vietnam War
13
In summary, childbearing responses to the 1969 draft lottery appear among women married to men
with a high risk of being inducted and enough time to obtain a hardship deferment for fatherhood. In
addition, they are largest for childless wives of men who would not have qualified for a hardship deferment
without fathering a child. We find no evidence that men subject to the 1970 to 1972 draft lotteries—lotteries
occurring after Nixon eliminated the paternity deferment—also exhibited a hump-shaped fertility response.
The implication of these findings is that fertility rates in 1970 were significantly elevated by the Vietnam
War and the availability of hardship deferments for paternity.
D. A National Counterfactual Fertility Rate for 1970
Counterfactual estimates illuminate the magnitude of the implied effects in terms of the national
fertility rates. Our approach takes the LATE for the middle-risk group of lottery numbers and scales it to
generate a national figure in the following manner,
ω
1,000.
(2)
If everyone in the U.S. had been at risk of the draft, the national effect of the lottery might be inferred
directly using the estimates of
from table 1. However,
is only estimated for the population subject to
the draft and in the 1973 and 1976 NSFG sample frame. Assuming the effects of the 1969 lottery were
zero for individuals with husbands not subject to the 1969 lottery and not in the NSFG sampling frame, we
scale the lottery estimate by the share of women in the 1973 and 1976 NSFG samples who were married to
someone in the 1969 lottery, , and the share of the childbearing population in the U.S. represented in the
NSFG samples, ω. Fertility rates are, by convention, scaled by 1000, which motivates our scaling in (2).
To generate the counterfactual fertility rate had the 1969 lottery not been in place, we set t=1970. We obtain
parameters for these estimates from the NSFG and 1970 census (Ruggles and al. 2015). The 1973 and 1976
NSFGs suggest that 25 percent of women of childbearing age would have been married to a first husband
at risk for the draft in 1969. The 1970 census indicates that 71 percent of the population of 15 to 44 year
old women in the U.S. should be represented in the NSFG sampling frame. We, therefore, compute that
roughly 3 fewer births per 1,000 women would have occurred in 1970 (0.016 from Table 1A, column 4 x
0.25 x 0.71 x 1,000) had the Vietnam lottery not existed. The counterfactual national general fertility rate
Demographic Legacy of the Vietnam War
14
in 1970 is, therefore, 85 (=88 3). This counterfactual is depicted in Figure 6A as the blue dot labelled
“Lottery counterfactual.”
This counterfactual assumes that all men subject to the draft would have responded to the Vietnam
draft in the same manner as men with the 123-244 lottery numbers. This makes sense if all men subject to
the draft before 1969 and their partners had had time to respond, and there would have been no “safe”
number to allow them to avoid the draft. Moreover, all individuals in our lottery sample as in the population
at large would have had access to other deferments. Keep in mind that this counterfactual ignores any
response to draft risk that occurred before the 1969 lottery’s effects in September to November of 1970 and
also assumes a zero response to draft risk for cohorts too young to be subject to the 1969 draft (but expecting
to be subject to the 1970 draft lottery).19 Finally, the momentum of deciding to have a child with a partner
may be difficult to reverse in some cases, we expect that the availability of deferments for paternity may
have also affected fertility rates into 1971.
The external validity of the lottery LATE is more difficult to assess. There are well known
limitations in extrapolating LATEs to the population (DiNardo and Lee 2011), and there are multiple
reasons this may apply in our context. More risk-averse couples may have sought deferments for paternity
before the draft lottery, which is consistent with the 10-percent increase in the first birth rate between 1965
and 1969. These individuals would not have needed to respond by conceiving, because they were eligible
for a hardship deferment. By this logic, responses to the 1969 lottery may have been smaller than the
population responses. Response to the lottery could be smaller than in the population at large due to
19
These effects are likely large. For instance, men anticipating that they could be subject to a lottery (based on Nixon’s
May 1969 announcement of his plans) could have begun fathering children before the official November 1969 change
in legislation. Younger men expecting to be subject to the lottery in 1970 could have begun fathering children before
Nixon’s April 23 elimination of paternity as a grounds for deferment, which would have elevated birth rates for all
men—not just those with lottery numbers. Conceptions before Nixon’s April 23 elimination of paternity as grounds
for deferment would have been born as late January 1971.This calculations assumes a 28-day menstrual cycle and that
women beginning their last period on April 9 would have conceived on around April 23. Assuming 40 weeks of
gestation from the first day of the last period, these babies would have been due around January 14, 1971, and could
have arrived in early or late January, due to variation in gestational length across women.
Demographic Legacy of the Vietnam War
15
selection (the category still childless in 1969 is less prepared or least motivated to dodge) or larger due to
the salience of risk (deployment was imminent).
The fact that our lottery-based counterfactual is much smaller than the likely Vietnam War induced
fertility notch suggests our lottery-based counterfactual estimate is conservative. Eliminating the notch
entirely (i.e. obtaining an estimate roughly in line with the linearly interpolated estimate) would require an
effect over 3.25 times as large as the lottery-based counterfactual.
Figure 6B plots the counterfactual for first births.20 As before, we obtain parameters for these
estimates from the NSFG and Census. The 1973 and 1976 NSFGs suggest that 15.9 percent of women of
in their sampling frame would have been childless and married to a first husband at risk for the draft in
1969. Only 28 percent of women in the NSFG were childless, and the 1970 census indicates that only 71
percent of U.S. women of childbearing ages 15 to 44 were sampled in the NSFG. We, therefore, compute
that roughly 1 fewer births per 1,000 women would have occurred in 1970 (0.029 from Table 1B, column
4 x 0.138 x .28 x 0.71 x 1,000) had the 1969 Vietnam lottery not taken place. This implies a counterfactual
1970 first birth rate of 32.7 (=33.6 0.9), 3 percent lower than occurred with the 1969 lottery. This number
is depicted in Figure 6B as the point labelled as “Lottery counterfactual.” Visually, Figure 6B shows that
the first birth rate notch is only modestly reduced in the lottery counterfactual, which underscores the
likelihood that is conservative. Eliminating the notch entirely (i.e. obtaining an estimate roughly in line
with the linearly interpolated estimate) would require an effect over 6.5 times as large as for first births.
V.
IMPLICATIONS FOR THE LITERATURE ON THE IMPACT OF ABORTION LEGALIZATION
Given the potential magnitude of the childbearing response to the Vietnam War, a final analysis
considers how these findings affect the literature’s conclusions about the fertility effects of abortion
20
We estimate
ω 1,000. is only estimated for the population of nulliparous women married
to first husbands subject to the draft and in the 1973 and 1976 NSFG sample frame. Assuming the effects of the 1969
lottery were zero for individuals with husbands not subject to the 1969 lottery and not in the NSFG sampling frame,
we scale the lottery estimate by the share of women in the 1973 and 1976 NSFG samples who were married to someone
in the 1969 lottery and childless, , and the share of the childless childbearing population in the U.S. represented in
the NSFG samples of childless women, ω.
Demographic Legacy of the Vietnam War
16
legalization. The evidence quantifying the role of abortion comes primarily from a differences-indifferences research design that compares the decline in fertility rates in states that legalized abortion in
1970 (Alaska, California, Hawaii, New York, and Washington) to the decline in fertility rates in states that
legalized abortion in 1973 with Roe v. Wade.21 This research design was pioneered by Levine et al. (1999)
and Gruber et al. (1999), who estimated the following linear regression model:
,
,
,
,
,
,
,
,
,
,
(3)
where Repeal is a dummy variable for the five states repealing bans on abortion in 1970; Reform is a dummy
variable for states that reformed abortion restrictions prior to Roe v. Wade (Arkansas, Colorado, Delaware,
Florida, Georgia, Kansas, Maryland, New Mexico, North Carolina, Oregon, South Carolina, Virginia); and
,
, and
are a dummy variables for years 1971-73, 1974-76, and 1977-80, respectively.
Other covariates include the unemployment rate, percent of population that is nonwhite, and per capita
income (
,
) as well as state (
) and year ( ) fixed effects and state-specific linear trends (
).
Using this specification, Levine et al. (1999) find that abortion legalization caused aggregate
fertility rates to fall by 4 to 5 percent, drops driven by large shifts in teen and non-marital childbearing.22
Angrist and Evans (1999) build on this identification using years of a cohort’s exposure to liberalized
abortion and finds that legalized abortion reduced teen birth rates by 2.2 percent among whites and 10
percent among blacks. They also find that this policy change led to an 18 percent decline in non-marital
childbearing among black teens (born 1949-1954) though none among whites. Myers (forthcoming) builds
on the Angrist and Evan’s exposure specification using the June Current Population Surveys to show that
abortion access for young women led to a 34 percent reduction in first births.23
21
California did not explicitly repeal bans on abortion as in other states. The September 5, 1969, the California
Supreme Court ruled in People v. Belous effectively overturned California’s 1850 statutory ban on abortion.
22
In specifications that account for distance from early legalizing states to capture inter-state commuting spillovers as
more individuals drove to those states (and, thus, account for contamination in the comparison group), their estimates
double to around 8 percent.
23
Myers’ (forthcoming) focus is on legal consent to abortion for young women. Her treatment of interest is the
interaction of abortion legalization and laws allowing young, unmarried women to consent to abortion. Because
Demographic Legacy of the Vietnam War
17
A key identifying assumption in each of these studies is that the legalization of abortion was the
only reason that fertility rates fell more quickly in Alaska, California, Hawaii, New York, and Washington
after 1970. However, if the fertility response to the Vietnam War raised fertility rates relatively more in
states that legalized abortion in 1970, a sharper post-1970 decline in fertility rates in these states could also
reflect the decline in draft-induced childbearing. In contrast, if the fertility response to the Vietnam War
raised fertility rates relatively less in states that legalized abortion in 1970, a sharper post-1970 decline in
fertility rates in these states would understate the role of abortion in driving fertility decline. In short,
elevated fertility rates to dodge the Vietnam draft may be an important omitted variable.
A. Heterogeneity across States in the Hardship Deferment’s Effect on Childbearing
To test the potential role of the Vietnam draft and the hardship deferment as an omitted variable,
we examine whether childbearing responses nine to 11 months after the draft lottery were significantly
higher for men at moderate induction risk in states that repealed or reformed abortion by interacting these
variables with the draft risk indicators in equation (1),
∑
.
(4)
The dependent variable is childbearing nine to 11 months after the 1969 draft lottery and the covariates
remain as defined in equations (1) and (3). Following Levine et al. (1999) and Gruber et al. (1999), we
allow the effects of the draft on childbearing to vary in early repeal states, Repeal, and abortion reform
states, Reform.24 The coefficients on DraftRisk,
, capture the role of the draft in raising birth rates
Alaska, California, Hawaii, New York, and Washington had passed consent laws prior to 1971, the change in early
consent to abortion in these states is coincident with legalization in 1970. Other states in her analysis code legal
consent for young women ages 18 to 20 to abortion in 1971 (District of Columbia) and 1974 (Massachusetts, Missouri,
North Dakota, South Carolina, Table 1).
24
The NSFG asks respondents the state where they lived from ages 6 to 18. We use this information to assign each
individual the legal status of abortion in early adulthood. The fact that state is not birth state or state of residence at
the time of the survey conveniently reduces measurement error in each woman’s access to legal abortion around the
time of the 1969 draft lottery. It is also the state in which her husband likely would have been registered for the draft.
Demographic Legacy of the Vietnam War
18
nationally. Of interest are the coefficients on the interaction of DraftRisk and Repeal, , which capture the
differential response to the risk of induction in states that effectively repealed abortion bans in 1970 and
the potential role of the draft as an omitted variable.
Table 3 presents the results. Column 1 reprints findings from Table 1 (panel B, column 4) showing
the relationship between having first birth nine to 11 months after the 1969 draft lottery and draft number
categories. Column 2 adds to this specification the second and third components of equation 3—the
interactions of draft number categories with dummies for states that repealed abortion bans or liberalized
access (reform states) after 1970. These estimates show that the national childbearing response to the
Vietnam draft appears driven by very large responses in the states repealing abortion bans in 1970.
Consistent with fertility rates in repeal states being lower than in control states throughout the 1960s and
1970s, the point estimates show that individuals with the lowest draft numbers in early repeal states had
significantly lower birth rates in the latter part of 1970. However, husbands with high induction risk and
more time to respond (draft numbers 123-244) in early repeal states were a whopping 8 percentage points
(s.e. 3.6) more likely to become fathers nine to 11 months after the 1969 lottery. Controlling for state fixed
effects (column 3) changes this estimate negligibly. This is over 8 to 9 times as large as the corresponding
number for men in control states, statistically insignificant estimates of 0.9 to 1 percentage point (s.e. 1.5).
Although 9 births per 1,000 women is large in magnitude relative to the 1970 birth rate of 90 births per
1,000, the NSFG sample sizes are too small to distinguish effects of this size from zero. Interestingly, there
are economically significant childbearing responses among men in repeal states in the highest draft number
group, but they are not precisely estimated and are statistically indistinguishable from estimates in control
states. As expected, estimates among new mothers are also much larger than among those with one or more
children (column 4). Pooling both groups of women (column 5), the overall response in states repealing
abortion restrictions before 1970 is 6.2 births per 1,000 women.
In contrast, we find less evidence of large fertility responses to the draft lottery for women in states
where abortion restrictions were reformed (but not repealed) relative to control states—a result robust to
the inclusion of state fixed effects (column 3). The estimated magnitudes in repeal states are sizable and
Demographic Legacy of the Vietnam War
19
also larger than estimates in control states, though statistically insignificant (columns 2-3). Taken literally,
these figures suggest that the childbearing response to the 1969 lottery was significantly higher in states
legalizing abortion in 1970 relative to states legalizing abortion in 1973 with Roe v. Wade.
This evidence is also consistent with Gallup polls of the era measuring significantly higher antiVietnam sentiment in states that repealed abortion before 1973. In January of 1973, for instance,
respondents in states that repealed abortion before Roe v. Wade were 11 percentage points (s.e. 0.039) more
likely to say that they thought the U.S. had made a mistake in sending troops to fight in Vietnam—a finding
that is robust to the inclusion of age, sex, race, and education dummies.25 In the same survey, 13 percentage
points more respondents (s.e. 0.042) disapproved of the way President Nixon was handling the situation in
Vietnam, 10 percentage points fewer respondents (s.e. 0.034) believed the Nixon Administration was
telling the public all they should know about the Vietnam War, and 8 percentage points more respondents
(s.e. 0.042) opposed resuming bombing of Hanoi and Haiphong if North Vietnam refused to agree to
“reasonable peace terms”—again, all findings robust to the inclusion of age, sex, race, and education
dummies.
In summary, significantly more anti-war sentiment and draft-dodging through fatherhood in repeal
states implies that the decline in childbearing in these states should be significantly larger after 1970 after
Nixon eliminated paternity as grounds for hardship. This implies that the coincidence of the repeal of
abortion bans and Nixon’s elimination of the hardship deferment may have led previous studies to overstate
the role of abortion legalization in reducing fertility rates.
B. Counterfactual Fertility Rates in Early Repeal, Reform, and Control States
We construct counterfactual fertility rates by state groups using the same approach as in section
IV.D in equation (2) for each state group g (
,
,
and ω ) where g represents Repeal, Reform, or
Control states in t=1970. Parameters from the 1970 Census and NSFG together with information from table
25
Gallup Poll 862 in January 12-15 of 1973 asked 1,549 respondents, “In view of the developments since we entered
the fighting in Vietnam, do you think the U.S. made a mistake sending troops to fight in Vietnam?”
Demographic Legacy of the Vietnam War
20
3, column 6, we calculate that the lottery counterfactual fertility rate for 1970 in repeal was 72—a reduction
from the observed fertility rates of 15 percent.26 Counterfactuals for both control and repeal states are plotted
in Figure 7A. The lottery estimate completely eliminates the fertility notch in repeal states–it is virtually
identical to the estimate obtained by linearly interpolating between 1966 and 1973. As implied by Table 3,
the counterfactual adjustment has virtually no effect on the fertility notch in control states, which reflects
that the point estimates for this group of states are small in magnitude.
Taken literally, this implies either that (1) there was very little fertility response to the Vietnam
War in control states, (2) that local draft boards in control states were unlikely to entertain a last minute
request for a change of status after the 1969 lottery, or (3) that individuals in control states had children in
advance of the 1969 lottery to defer service (just not in response to their lottery number).27 The first seems
unlikely because control states have a visible fertility notch as well. The second and third explanations
would lead to a fertility notch—though one that is not as large in 1970.
As an alternative approach to estimating the effect of the draft dodging (not the draft lottery) on
births uses non-linear, time series interpolation between 1965 and 1972. We implement this by regressing
the general fertility rate on a set of state group fixed effects, g, interacted with a sixth-order polynomial
(sextic) in year of observation, t, or ∑6
0
. We use this flexible polynomial to fit
the evolving GFR in the 1955-1965 and 1972-1980 periods for control, reform and repeal state groups.
Then we use this model to predict counterfactual fertility rates for the Vietnam War period, 1966-1971. The
resulting counterfactual estimates are plotted in Figure 7B. Notably, this counterfactual is similar to the
lottery-based counterfactual for early repeal states, predicting a GFR in 1970 of 71.8 (versus 72.6 for the
26
The 1973 and 1976 NSFGs show that 25, 26.2, and 28.9 percent of this sample in repeal, reform, and control states,
respectively, would have been married to a first husband at risk for the draft in 1969. The 1970 census indicates that
70.6, 73, 69.7 percent of women in repeal, reform, and control states, respectively, would have represented in the 1973
and 1976 samples.
27
Another explanation relates to the fact that that the 1973 and 1976 are not designed to be representative at the state
level and may provide noisy estimates of sub-national means. We cannot address this directly. If this were the case,
we would expect the estimates for smaller state groups (repeal and reform) to be noisier than for the largest group of
states (control group). This is inconsistent with the null results in the control group states being spurious.
Demographic Legacy of the Vietnam War
21
lottery-based estimate)—a reduction of 16 percent from the observed rate of 85.4. The time series
counterfactual, however, implies a much larger reduction in the GFR in control states by 1970 of 79.6—a
reduction of 12 percent from the observed GFR of 90.5.
Figure 7B also reveals that the different evolution of the GFR before 1966 implies a violation of
the parallel trends assumption in equation (3). This is consistent with findings that childbearing responses
to the 1969 lottery was larger in early repeal states. That the parallel trends assumption appeared to have be
met in previous studies reflects the fact that the GFR was disproportionately elevated in early repeal states
between 1965 and 1972.
We also perform similar exercises examining counterfactual firth birth rates in repeal and control
states in figure 8. Panel A uses the estimated effect of draft risk on first births from column 4 of Table 3 to
estimate a lottery counterfactual for 1970. Similar to our findings for the general fertility rate, the results
show repeal state first birth rates would have been substantially lower in 1970 in absence of the of Vietnam
era paternity deferments while there is a much smaller reduction in control states. Panel B reaches a similar
conclusion based on a counterfactual using time series interpolation.
VI.
SEPARATING THE VIETNAM DRAFT FROM ABORTION LEGALIZATION
A final analysis uses these counterfactual fertility rates to re-examine the causal effect of abortion
legalization on birth rates. Using the standard differences-in-differences specification as in equation (3),
table 5 shows how the point estimates on the interaction of Repeal and dummy variables for year groupings
changes. Column 1 reprint the Gruber, Levine, and Staiger (1999) results.28 Columns 2 closely replicate
these estimates using our data. Columns 3 shows that using only 1970 forward sample has little impact on
these results. That is, using only 1970 to estimate the pre-differences in fertility rates between repeal and
control states has very little effect on the estimates.
28
Note that Gruber et al. (1999) omit the inclusion of the terms for reform states so that their specification differs from
Levine et al. (1996). Our analysis follows the specification used in Gruber (1999).
Demographic Legacy of the Vietnam War
22
Column 4 replaces the 1970 fertility rate with the lottery counterfactual for each state for the 19701980 sample.29 Consistent with panel A of figure 7, the lottery counterfactual completely eliminates the
notch and also the sharper drop in fertility rates in early repeal states relative to control states. The
statistically significant positive estimate of abortion repeal in 1970 (0.087, s.e., 0.0142) indicates that the
gap in fertility rates between early repeal and control states in fertility rates became more positive (as the
negative gap fell), which is consistent with fertility rates falling more rapidly in control states from 1971 to
1973 than in repeal states in the counterfactual-based data. Consistent with the results in panel B of figure
7, the polynomial counterfactual reduces fertility rates more in control states as compared to the lottery
counterfactual. This counterfactual suggests that the early legalization of abortion may have reduced
fertility rates by around 1.8 percent (s.e., 0.003).
This first set of findings suggests that there is no firm way to choose a “correct counterfactual” is
best. However, these findings suggest there is considerable uncertainty about the role of abortion
legalization. The most optimistic picture offered by these estimates is that the legalization of abortion may
for at most one third of Levine et al.’s (1999) estimates, but that evidence is based only on polynomial
extrapolation and is imprecise. The big picture here is that the availability of the paternity deferment may
have had a significant effect on birth rates as well as inferences about the effects of abortion.
A second finding is that fertility rates in controls fell much more rapidly after Roe v. Wade relative
to repeal states. The lottery counterfactual the 1971 birth rate counterfactual suggests a significant
narrowing of the gap in fertility rates between these states of 2.7 log points after 1974 to 1975 relative to
1971 to 1973. This change in the gap is consistent with Roe v. Wade having a significant negative effect
on birth rates in control states, which seems to have persisted to 1980.
29
Specifically, we subtract the estimated effect of the lottery in repeal states (equal to 0.0616*0.289*0.697*1000)
from the observed GFR in each of the repeal states (Alaska, California, Hawaii, New York, and Washington). We do
a similar adjustment for control and repeal states using their respective point estimates obtained from Table 3 and
appropriate scaling terms.
Demographic Legacy of the Vietnam War
23
VII.
THE DEMOGRAPHIC LEGACY OF THE VIETNAM WAR
In the late 1960s, dodging the Vietnam draft was a preoccupation for many young men—driving
many to desperate measures to avoid serving in an unpopular war. This paper quantifies the importance of
paternity—one important and understudied form of draft avoidance: paternity. More than twice as many
hardship deferments for paternity were issued in 1970 than education deferments. Strong incentives to
become a father resulted in a striking fertility notch of the late 1960s that evaporated shortly after Nixon
eliminated paternity as a means to defer military service in April of 1970.
We provide novel evidence to that the Vietnam draft and the availability of the hardship deferment
raised birth rates. First, the fertility notch was concentrated among women very likely to be partnered with
draft eligible men. Time-series evidence shows the fertility notch was driven by first births to 20 to 24year-old women. Even though Canada’s birth rates tended to follow the U.S. birth rates closely, there was
no fertility notch in Canada in the late 1960s, where the Vietnam draft did not apply. Second, the 1969
Vietnam draft lottery shows a robust, hump-shaped response by men at high risk for induction with time to
respond. Women married to men in this group were nearly 85 percent more likely to have a child 9 to 11
months after the December 1969 lottery drawing. Counterfactual fertility estimates scaling these lottery
estimates suggest that the general fertility rates and first birth rates would have been 13 percent lower in
the absence of the Vietnam War.
A second set of findings suggest a revision to the conventional wisdom about the role of abortion
legalization in the U.S. The repeal and reform of abortion legalization may have had significant impacts on
many outcomes. But the temporal proximity of these changes to Nixon’s elimination of paternity as grounds
for hardship deferment in April of 1970 and the disproportionately larger fertility responses to the Vietnam
War in states that legalized abortion before Roe v. Wade complicates identification. Our analysis of
heterogeneity finds that childbearing responses to randomly assigned draft numbers in states repealing
abortion bans in 1970 were eight to nine times larger than in states legalizing abortion in 1973 with Roe v.
Wade. This may be not be surprising given that New York and California, two of the largest state
populations in the early repeal group, were also known hotbeds of anti-war sentiment. Accordingly, Gallup
Demographic Legacy of the Vietnam War
24
polls of the era also suggest significantly more anti-war sentiment in these states. Interestingly, our
counterfactual analyses suggest that the availability of the hardship deferment for paternity may have led
researchers to overlook the role of Roe v. Wade, which may play a larger role in reducing American fertility
rates and altering the living circumstances of children than previously known.
Demographic Legacy of the Vietnam War
25
VIII.
REFERENCES
People v. Belous. 1969. 71 Cal. 2d 954.
Roe v. Wade. 1973. 410 U.S. 113.
Selective Service Amendment Act of 1969. Pub. Law 91-124.
Ananat, Elizabeth Oltmans, Jonathan Gruber, and Phillip Levine. 2007. "Abortion Legalization and LifeCycle Fertility." Journal of Human Resources 42 (2):375-397.
Anderson, Gary M., and Robert D. Tollison. 1991. "Congressional Influence and Patterns of New Deal
Spending,1933-1939." Journal of Law and Economics 34 (1):161-175.
Angrist, Joshua D. 1990. "Lifetime Earnings and the Vietnam Era Draft Lottery: Evidence from Social
Security Administrative Records." American Economic Review 80 (3):313-36.
Angrist, Joshua D., and Stacey Chen. 2008. "Long-Term Economic Consequences of Vietnam-Era
Conscription: Schooling, Experience and Earnings." IZA Discussion Papers 3628.
Angrist, Joshua D., Stacey H. Chen, and Brigham R. Frandsen. 2009. "Did Vietnam Veterans Get Sicker in
the 1990s? The Complicated Effects of Military Service on Self-Reported Health." NBER Working
Paper 14781.
Angrist, Joshua D., and William N. Evans. 1999. "Schooling and Labor Market Consequences of the 1970
State Abortion Reforms." In Research in Labor Economics, edited by Solomon Polachek, 75-113.
Emerald Group Publishing.
Bailey, Martha J. 2006. "More Power to the Pill: The Impact of Contraceptive Freedom on Women's
Lifecycle Labor Supply." Quarterly Journal of Economics 121 (1):289-320.
Bailey, Martha J. 2010. "'Momma’s Got the Pill': How Anthony Comstock and Griswold v. Connecticut
Shaped U.S. Childbearing." American Economic Review 100 (1):98-129.
Bailey, Martha J. 2012. "Reexamining the Impact of U.S. Family Planning Programs on Fertility: Evidence
from the War on Poverty and the Early Years of Title X." American Economic Journal: Applied
Economics 4 (2):62-97.
Bailey, Martha J. 2013. "Fifty Years of U.S. Family Planning: New Evidence on the Long-Run Effects of
Increasing Access to Contraception." Brookings Papers on Economic Activity Spring:341-409.
Bailey, Martha J., Melanie Guldi, Allison Davido, and Erin Buzuvis. 2011. "Laws and Policies Governing
Contraceptive Access, 1960–1980." University of Michigan Working Paper. Accessed 8/1/2011.
http://www-personal.umich.edu/~baileymj/ELA_laws.pdf.
Bitler, Marianne P., and Lucie Schmidt. 2012. "Birth Rates and the Vietnam Draft." American Economic
Review 102 (3):566-69.
Card, David, and Thomas Lemieux. 2001. "Going to College to Avoid the Draft: The Unintended Legacy
of the Vietnam War." American Economic Review 91 (2):97-102.
Charles,
Kerwin
Kofi,
and
Melvin
Stephens,
Jr.
2006.
"Abortion Legalization
and Adolescent Substance Use." Journal of Law and Economics 49 (2):481-505.
Conley, Dalton, and Jennifer A. Heerwig. 2011. "The War at Home: Effects of Vietnam-Era Military
Service on Post-War Household Stability." NBER Working Paper 16671.
DiNardo, John E., and David S. Lee. 2011. "Program Evaluation and Research Designs." In Handbook of
Labor Economics, edited by D. Card and O. Ashenfelter. New Holland: Elsevier.
Eisenberg, Daniel, and Brian Rowe. 2009. "The Effects of Smoking in Young Adulthood on Smoking and
Health Later in Life: Evidence based on the Vietnam Era Draft Lottery." Forum for Health
Economics and Policy 12 (2):4.
Gruber, Jonathan, Phillip Levine, and Douglas Staiger. 1999. "Abortion Legalization and Child Living
Circumstances: Who Is the "Marginal Child"?" Quarterly Journal of Economics 114 (1):263-291.
Guldi, Melanie. 2005. "Abortion or The Pill - Which Matters More? The Impact of Access on the Birth
Rates of Young Women." Accessed. http://harris.princeton.edu/seminars/pdfs/Ananat.pdf.
Guldi, Melanie. 2011. "A Survey of the Literature on Early Legal Access to the Birth Control Pill and
its Influence on Young Women’s Fertility, Education, Career,and Labor Supply." In Research
Demographic Legacy of the Vietnam War
26
Handbook on the Economics of Family Law, edited by Lloyd R. Cohen and Joshua D. Wright, 271291. Northampton, MA: Edward Elgar.
Joyce, T., R. Tan, and Y. Zhang. 2013. "Abortion before & after Roe." Journal of Health Economics.
Kutinova, Andrea. 2009. "Paternity Deferments And The Timing Of Births: U.S. Natality During The
Vietnam War." Economic Inquiry 47 (2):351-365.
Kuziemko, Ilyana. 2010. "Did the Vietnam Draft Increase Human Capital Dispersion? Draft-Avoidance
Behavior by Race and Class." Princeton University Working Paper. Accessed.
Levine, Phillip B., Douglas Staiger, Thomas J. Kane, and David J. Zimmerman. 1999. "Roe v. Wade and
American Fertility." American Journal of Public Health 89 (2):199-204.
Levine, Phillip B., Amy B. Trainor, and David J. Zimmerman. 1996. "The Effect of Medicaid Abortion
Funding Restrictions on Abortions, Pregnancies, and Births." Journal of Health Economics 15
(5):555-578.
Levine, Phillip, Douglas Staiger, Thomas Kane, and David Zimmerman. 1996. "Roe v. Wade and American
Fertility." NBER Working Paper 5615.
Lindo, Jason M., and Charles F. Stoecker. 2012. "Drawn into Violence: Evidence on 'What Makes a
Criminal' from the Vietnam Draft Lotteries." NBER Working Paper 17818.
Myers, Caitlin. forthcoming. "The Power of Abortion Policy: Re-Examining the Effects of Young Women's
Access to Reproductive Control." Journal of Political Economy. Accessed February 3, 2017.
http://community.middlebury.edu/~cmyers/Power_JPE.pdf.
Nixon, Richard. 1970. "Executive Order 11527—Amending the Selective Service Regulations." Santa
Barbara,
CA:
The
American
Presidency
Project.
Retrieved
from:
http://www.presidency.ucsb.edu/ws/?pid=60484.
Rotz, Dana. 2012. "The Impact of Legal Abortion on the Wage Distribution: Evidence from the 1970 New
York Abortion Reform." SSRN Working Paper. Accessed. http://dx.doi.org/10.2139/ssrn.2181061.
Ruggles, S., and et al. 2015. Integrated Public Use Microdata Series (Version 6.0) [Machine-Readable
database]. Retrieved from:
Selective Service System. 1960. "Annual Report of the Director of the Selective Service." Washington:
Government Printing Office. Retrieved from: https://catalog.hathitrust.org/Record/000506708.
Smock, Pamela J., Peter Granda, and Lynette Hoelter. 2013. Integrated Fertility Survey Series, Release 7,
1955-2002
[United
States].
ICPSR26344-v7.
Retrieved
from:
http://doi.org/10.3886/ICPSR26344.v7
Solon, Gary, Steven J. Haider, and Jeffrey M. Wooldridge. 2015. "What are We Weighting For?" Journal
of Human Resources 50 (2):301-316.
Starr, Nordton. 1997. "Nonrandom Risk: The 1970 Draft Lottery." Journal of Statistics Education 5 (2).
Tatum, Arlo, and Joseph S. Tuchinsky. 1970. Guide to the Draft. Boston: Beacon Press.
Tracy, James, ed. 2005. The Military Draft Handbook: A Brief History And Practical Advice for the Curious
And Concerned. San Francisco, CA: Manic D Press, Inc.
VanDeMark, Brian. 1991. Into the Quagmire: Lyndon Johnson and the Escalation of the Vietnam War.
New York: Oxford University Press.
Demographic Legacy of the Vietnam War
27
Figure 1. Hardship and Education Deferments, 1960 to 1973
4,500,000
III-A Deferments (paternity)
4,000,000
3,500,000
April 1970: Nixon
eliminates paternity
as grounds for III-A
II-S Deferments (student)
3,000,000
2,500,000
2,000,000
1,500,000
1,000,000
500,000
0
1958
1960
1962
1964
1966
1968
1970
1972
1974
Notes: Total inductions by fiscal year exclude U.S. territories. Fiscal years run from July of the previous year to June
of the current year (i.e. 1961 data reflects inductions from July 1, 1960-June 30, 1961). Source: Annual data from
annual and semi-annual Selective Service reports, 1960-1973.
Figure 2. Inductions, 1960 to 1973
335742
335913
296503
259490
201874
151252
152075
144533
99843
86894
58421
70305
34532
24774
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
Notes: See figure 1.
Demographic Legacy of the Vietnam War
28
Figure 3. U.S. Fertility Rates, 1950 to 1980
A. U.S. General Fertility Rate
130
120
1965-70:
Vietnam
War
escalates
110
100
April 1970: Nixon
eliminates paternity
deferment
90
80
70
60
50
1950
1955
1960
1965
1970
1975
1980
B. First Birth Rate (First Births per 1000 Women 15 to 44 in Population)
36
Fertility notch
consists
largely of first
births
34
32
30
1965-70:
Vietnam
War
escalates
28
26
1950
1955
1960
1965
1970
Demographic Legacy of the Vietnam War
1975
29
1980
C. Age-Group-Specific First Birth Rates (First Births per 1000 Women in Age Group)
100
Fertility notch
driven by first
births to women
ages 20-24
90
80
70
60
50
40
30
20
10
0
1950
1955
1960
15-19
1965
20-24
1970
25-29
1975
30-34
Note: The general fertility rate is measures as births per 1,000 women ages 15 to 44. The first birth rate is computed
by dividing the number of first births by the number of women ages 15 to 44 (panel B) and the number of women in
the indicated age group (panel C) and multiplying by 1,000. Sources: 1950 to 1967 Vital Statistics Volumes (Bailey
2010) and 1968 to 1980 Natality Files (US DHHS 2002).
Demographic Legacy of the Vietnam War
30
1980
Figure 4. The U.S. Fertility Notch: A Comparison to Canada, 1920-2010
A. Total Fertility Rates in Canada and the U.S.
4000
4500
3500
4000
3500
3000
3000
2500
2500
2000
2000
1500
1500
1000
1000
500
0
1925
500
US Total Fertility Rates
Canadian Total Fertility Rate (right axis)
1935
1945
1955
1965
1975
1985
1995
2005
0
2015
B. Birth Rates among Women Ages 20 to 24, 1960-2010
290
240
April 1970: Nixon eliminates
deferments for paternity
190
U.S. birth rates deviate
from Canadian birth rates
140
90
40
1960
1970
1980
US 20-24 years
1990
2000
2010
CA 20-24 years
Sources: In panel A, Canadian fertility rates are from Milan (2013) and U.S. fertility rates are from
http://www.cdc.gov/nchs/data/statab/t001x01.pdf (NCHS 2013). Panel B uses Canadian fertility rates from Milan
(2013), and U.S. fertility rates are from the 1960 to 1967 Vital Statistics Volumes (Bailey 2010) and 1968 to 1980
Natality Files (US DHHS 2002).
Demographic Legacy of the Vietnam War
31
Figure 5. The Effect of the 1969 Draft Lottery on Childbearing 9 to 11 Months Later
A. Hypothesized Childbearing Response
B. All Births
Demographic Legacy of the Vietnam War
32
C. First Births
D. Higher Order Births
Notes: This figure plots point estimates from a linear probability regression model where the dependent variable is an
indicator for having a birth 9 to 11 months after the 1969 draft lottery. We plot the coefficients for 10 equal-sized
groups of draft lottery numbers. The omitted group indicator is the group containing the first 37 lottery numbers. We
also include controls for the husband’s birth month and controls for the respondent’s age, education and race in the
results. Source: 1973 and 1976 NSFG (Smock et al. 2012).
Demographic Legacy of the Vietnam War
33
Figure 6. Counterfactual Fertility Rates, 1970
A. General Fertility Rate
120
110
100
Lottery
counterfactual
90
80
70
60
1950
1955
1960
1965
1970
1975
1980
B. First Birth Rate
35
Lottery
counterfactual
34
33
32
31
30
29
28
27
26
25
1950
1955
1960
1965
1970
Demographic Legacy of the Vietnam War
1975
34
1980
Figure 7. Counterfactual General Fertility Rates in 1970, by the Legality of Abortion
A. Counterfactual Based on the 1969 Lottery Estimates
130
Control GFR
120
Repeal GFR
110
Lottery counterfactual in control states
100
90
80
Lottery counterfactual in repeal states
70
60
1950
1955
1960
1965
1970
1975
1980
1985
B. Counterfactual Based on Polynomial Fit
130
Control GFR
120
Repeal GFR
110
Control polynomial
counterfactual
Repeal polynomial
counterfactual
100
90
80
70
60
1950
1955
1960
1965
1970
1975
1980
1985
Notes: The lottery-based counterfactuals in panel A are constructed by subtracting from the observed GFR the quantity
0.0616*0.289*0.697*1000 in repeal states and 0.00185*0.25*0.706*1000 in control states. The polynomial
counterfactuals in panel B are constructed by fitting a sextic interacted with state groups for the 1955-1965 and 19721980 periods and predicting fertility rates in the Vietnam era. See text for details.
Demographic Legacy of the Vietnam War
35
Figure 8. Counterfactual First Birth Rates in 1970, by the Legal Status of Abortion
A. Counterfactual Based on the 1969 Lottery Estimates
38
Control FBR
36
Lottery counterfactual
in control states
Repeal FBR
34
32
30
28
26
24
22
1950
Lottery counterfactual in repeal states
1955
1960
1965
1970
1975
1980
1985
1980
1985
B. Counterfactual Based on Polynomial Fit
38
36
34
32
30
28
26
24
22
1950
Control FBR
Repeal FBR
Control polynomial counterfactual
Repeal polynomial counterfactual
1955
1960
1965
1970
1975
Notes: The lottery-based counterfactuals in panel A are constructed by subtracting from the observed GFR the quantity
0.0799*0.289*0.697*1000 in repeal states and 0.0093*0.25*0.706*1000 in control states. The polynomial
counterfactuals in panel B are constructed by fitting a sextic interacted with state groups for the 1955-1965 and 19721980 periods and predicting fertility rates in the Vietnam era. See text for details.
Demographic Legacy of the Vietnam War
36
Table 1. The Effect of Draft Risk on Childbearing 9 to 11 Months after the 1969 Draft Lottery
(1)
(2)
(3)
A. Sample: Women with 1969 Draft-Eligible Husband
Mean of Dep. Var. for Draft # 0-122
0.040
Draft # 123-244
0.016
0.018
0.018
[0.008]
[0.009]
[0.009]
Draft # 245-366
0.006
0.005
0.005
[0.008]
[0.008]
[0.008]
Observations
R-squared
4,574
4,574
4,567
0.001
0.002
0.006
B. Sample: Childless Women with 1969 Draft-Eligible Husband
Mean of Dep. Var. for Draft # 0-122
0.033
Draft # 123-244
0.029
0.030
0.029
[0.012]
[0.012]
[0.012]
Draft # 245-366
0.007
0.008
0.007
[0.010]
[0.010]
[0.010]
Observations
R-squared
2,518
2,518
2,512
0.004
0.006
0.016
C. Sample: Women with Children and 1969 Draft-Eligible Husband
Mean of Dep. Var. for Draft # 0-122
0.036
Draft Number 123-244
-0.004
-0.002
-0.003
[0.012]
[0.012]
[0.013]
Draft Number 245-366
0.005
-0.0002
-0.002
[0.013]
[0.013]
[0.013]
(4)
0.017
[0.009]
0.005
[0.008]
4,567
0.029
0.029
[0.013]
0.007
[0.010]
2,512
0.052
-0.005
[0.011]
-0.003
[0.012]
Observations
2,056
2,056
2,055
2,055
R-squared
0.000
0.007
0.013
0.076
Covariates
Husband Birth Month FE
X
X
X
Survey Year (1976 Dummy)
X
X
X
Age of Birth (Categorical Dummy)
X
X
Race FE
X
X
Education FE
X
X
State FE (Residence at Ages 6 - 16)
X
Notes: Robust standard errors in brackets. The dependent variable is an indicator for a birth in September, October or
November 1970. In panel B, the sample is restricted to NSFG respondents (women) who have not had a first birth
before September 1970. In panel C, the sample is restricted to NSFG respondents who already had a child before
September 1970. Source: 1973 and 1976 NSFG (Smock et al. 2012).
Demographic Legacy of the Vietnam War
37
Table 2. The Effect of Draft Risk on Childberaing after the 1970, 1971, and 1972 Draft Lotteries
(1)
Mean of Dep. Var. for Draft # 0-122
Draft # 123-244
Draft # 245-366
Observations
R-squared
Covariates
Husband Birth Month FE
Survey Year (1976 Dummy)
Age of Birth (Categorical Dummy)
Race FE
Education FE
State FE (Residence at Ages 6 - 16)
(2)
(3)
(4)
Sample: Childless Women with Husband Eligible for 19701972 Draft Lotteries
0.034
0.006
0.008
0.010
0.010
[0.019]
[0.018]
[0.019]
[0.019]
0.003
0.004
0.004
0.009
[0.016]
[0.016]
[0.016]
[0.017]
872
0.000
872
0.019
862
0.029
862
0.099
X
X
X
X
X
X
X
X
X
X
X
X
X
Notes: This table uses the sample of men subject to the 1970, 1971, and 1972 draft lotteries. See Table 1 and text for
more details. Source: 1973 and 1976 NSFG (Smock et al. 2012).
Demographic Legacy of the Vietnam War
38
Table 3. Heterogeneity in the Impact of Draft-Induced Childbearing in 1970
(1)
Dependent Variable
Sample
Mean of Dependent Variable
Draft # 123-244
Draft # 245-366
(3)
(4)
(5)
(6)
1= Birth 9 to 11 Months After 1969 Draft Lottery
No Prior
Births
0.033
0.029
[0.013]
0.007
[0.010]
Reform
Reform × Draft # 123-244
Reform × Draft # 245-366
Repeal
Repeal × Draft # 123-244
Repeal × Draft # 245-366
Observations
R-squared
Covariates
Husband Birth Month FE
Survey Year (1976 Dummy)
Age of Birth (Categorical Dummy)
Race FE
Education FE
State FE (Residence at Ages 6 - 16)
2,512
0.052
X
X
X
X
X
X
All Women
with Draft
Eligible
Partners
0.034
0.002
[0.010]
0.0006
[0.010]
No Prior
Births
0.033
0.010
[0.015]
-0.001
[0.013]
-0.015
[0.016]
0.023
[0.029]
0.004
[0.024]
-0.028
[0.014]
0.080
[0.035]
0.043
[0.029]
2,512
0.021
No Prior
Births
0.033
0.009
[0.015]
-0.003
[0.013]
At Least
One Prior
Birth
0.036
-0.015
[0.015]
0.0000
[0.017]
0.024
[0.030]
0.006
[0.023]
0.0189
[0.0309]
-0.012
[0.028]
0.019
[0.021]
-0.003
[0.018]
0.080
[0.036]
0.043
[0.029]
2,512
0.056
0.029
[0.028]
-0.0003
[0.023]
2,055
0.077
0.062
[0.025]
0.024
[0.019]
4,567
0.031
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Notes: Robust standard errors in brackets. See Table 1 notes. Source: 1973 and 1976 NSFG (Smock et al. 2012).
Demographic Legacy of the Vietnam War
39
Table 4. Impact of Abortion Legalization on Actual and Counterfactual General Fertility Rates
(1)
Repeal × 71-73
Repeal × 74-75
Repeal × 76-79
Years
Observations
R-squared
(2)
(3)
(4)
Dependent Variable: Log(GFR)
(5)
(6)
Lottery
Lottery
Counterfactual
Polynomial
Counterfactual
for 1970 (omit Counterfactual
for 1970
1971-1972)
Reprinted
GLS (1999)
Table 1
Replication
1970-1980
Sample
-0.056
[0.006]
-0.015
[0.009]
0.01
[0.014]
-0.062
[0.015]
-0.038
[0.019]
-0.012
[0.028]
-0.061
[0.011]
-0.034
[0.018]
-0.011
[0.025]
0.087
[0.014]
0.114
[0.020]
0.137
[0.029]
0.089
[0.018]
0.114
[0.021]
0.138
[0.030]
-0.018
[0.003]
-0.020
[0.009]
0.010
[0.015]
1965-1979
750
N/A
1965-1979
750
0.945
1970-1980
500
0.954
1970-1980
500
0.952
1970-1980
400
0.955
1970-1980
500
0.970
Notes: Estimates are the effect of abortion legalization on fertility rates obtained using Gruber at al. (1999)’s preferred specification. Column (1)
reprints the Gruber et al. (1999) results. Column (2) replicates the Levine et al result in our data. Column (3) uses only data from 1970 to 1980.
Column (4) uses the lottery-based counterfactual fertility rate. Column (5) estimates the same regression excluding 1971 and 1972. Column (6) uses
instead the polynomial counterfactual. All models include the unemployment rate, percent of population that is nonwhite, per capita income, state
fixed effects and year fixed effects. Source: 1950 to 1967 Vital Statistics Volumes (Bailey 2010) and 1968 to 1980 Natality Files (US DHHS 2002).
Table 5. Impact of Abortion Legalization on Actual and Counterfactual First Birth Rates
(1)
GLS (1999)
Specification
Repeal × 71-73
Repeal × 74-75
Repeal × 76-79
Years
Observations
R-squared
(2)
(3)
(4)
(5)
Dependent Variable: Log(First Birth Rate)
Lottery
Lottery
Counterfactual
Polynomial
1970-1980
Counterfactual
for 1970 (omit Counterfactual
Sample
for 1970
1971-1972)
-0.135
[0.078]
-0.089
[0.023]
-0.028
[0.031]
-0.092
[0.057]
-0.055
[0.023]
-0.010
[0.026]
-0.013
[0.060]
0.024
[0.021]
0.069
[0.025]
0.022
[0.13]
0.026
[0.021]
0.073
[0.026]
-0.024
[0.003]
0.0007
[0.004]
0.073
[0.005]
1965-1979
750
0.945
1970-1980
500
0.954
1970-1980
500
0.952
1970-1980
400
0.955
1970-1980
500
0.970
Notes: Estimates are the effect of abortion legalization on fertility rates obtained using Levine at al. (1996)’s preferred specification. Column
(1) uses the same specification as in Gruber et al. (1999). Column (2) uses only data from 1970 to 1980. Column (3) uses the lottery-based
counterfactual fertility rate. Column (4) estimates the same regression excluding 1971 and 1972. Column (5) uses instead the polynomial
counterfactual. Column (6) replaces the 1970 birth rate with the 1971 birth rate. All models include the insured unemployment rate, percent of
population that is nonwhite, per capita income, state fixed effects and year fixed effects. Source: 1950 to 1967 Vital Statistics Volumes (Bailey
2010) and 1968 to 1980 Natality Files (US DHHS 2002).
Demographic Legacy of the Vietnam War
41