IZA DP No. 2157
Divorce, Fertility and the Shot Gun Marriage
zur Zukunft der Arbeit
Institute for the Study
and the Shot Gun Marriage
Harvard University, NBER
International Monetary Fund
and IZA Bonn
Discussion Paper No. 2157
P.O. Box 7240
Any opinions expressed here are those of the author(s) and not those of the institute. Research
disseminated by IZA may include views on policy, but the institute itself takes no institutional policy
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
company supported by Deutsche Post World Net. The center is associated with the University of Bonn
and offers a stimulating research environment through its research networks, research support, and
visitors and doctoral programs. IZA engages in (i) original and internationally competitive research in
all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research
results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available directly from the author.
IZA Discussion Paper No. 2157
Divorce, Fertility and the Shot Gun Marriage*
Using the birth certificates data from the Vital Statistics of the USA between 1968 and 1999,
we construct state level panel data of different measures of fertility and examine the change
in divorce laws. Total fertility declined in states that introduced unilateral divorce, which
makes dissolution of marriage easier. Most of this effect is due to a decline of out-of-wedlock
fertility. We suggest an explanation (and provide supportive evidence for it) based upon the
effect of divorce laws on the probability of entering and exiting marriage. Women planning to
have children marry more easily with an easier “exit option” from marriage. Thus, more
children are born in the first years of marriage, while the total marital fertility does not change,
probably as a result of an increase in divorces and marital instability. The effect of changes in
divorce laws is greater among whites than African Americans.
divorce laws, fertility, marriage
International Monetary Fund
700 19th Street NW
Washington, DC 20431
* We thank Filipe Campante and Azra Pravdic for excellent research assistantship.
The introduction of “no-fault” divorce has been one of the most significant changes in
the structure of the American families of the last thirty years. Unilateral “no fault” divorce
laws allowed one spouse to obtain dissolution of marriage without the consent of the other:
divorce became much easier. The effect of this change in the law has been widely studied
with reference to the frequency of divorce and marriage rates. But not much is known on its
effect on fertility: after all, marriage and divorce must have some effects on the number of
This is what we find. We first look at the effect of unilateral divorce laws on total
fertility, uncovering a significant and large negative effect on fertility rates. If this effect
came from fertility in wed lock that might be explained by marriage instability which lower
the propensity to invest in children (Becker, 1981; Becker, Landes and Michael, 1977). But,
somewhat surprisingly, most of the reduction in fertility, comes from reduction in out of wed
Our explanation is as follows. Imagine an unmarried woman contemplating child
bearing (or in the extreme case already pregnant). Without unilateral divorce, marriage
becomes an irreversible investment; couples are “locked in”. With unilateral divorce the risk
of entering the “wrong” marriage is lower because the exit option is easier. So a woman
contemplating parenthood may choose to enter marriage more easily with unilateral laws; as
a result out of wedlock fertility goes down. Obviously this does not imply that couples stay
married longer on average with unilateral divorce; on the contrary, some of these matches
may be indeed “wrong” and end up in divorce.
We present some supportive evidence for this story. First the number of never
married women goes down with unilateral divorce. Second, the number of marriages per
person is higher in unilateral states, so people marry more frequently. Third, fertility rates for
newly wedded couples (in the first two years of marriage) go up with the adoption of
unilateral laws; in a sense this include a sort of “shot gun marriage effect”: with easier
divorce, the incentive to fight the shot gun is lower.
These findings are therefore consistent with the view that when divorce is easier,
individuals take more chances with marriage, especially at the time of childbearing. This
effect makes welfare analysis of divorce laws even more complex than normally thought, and
we do not venture into any of it.
We are of course not the first to analyze empirically the effect of divorce laws. Many
authors have studied the effects of these laws on divorce rates (Peters, 1986 and 1992; Allen,
1992; Friedberg, 1998; Wolfers, 2006), marriage (Rasul, 2004), children outcome (Gruber,
2004; Johnson and Mazingo, 2000), labor supply (Chiappori, Fortin and Lacroix, 2002) and
general well-being of the couple (Stevenson and Wolfers (2005) and Dee (1999)), with
The available evidence on divorce laws and fertility is instead scant and here lies the
contribution of this paper.3 We use the legislative history of divorce liberalization across
states in the US to identify the effects of this reform on fertility rates. Using births data from
the Natality Files of the Vital Statistics of the US between the years 1968-1999, we fully
exploit cross state and year variation in the timing of adoption of unilateral divorce to
identify the causal effect of a change in divorce laws on fertility rates. The availability of
virtually universal Vital Statistics data on fertility provides enough statistical power that
could not be obtained relying exclusively on other datasets, including the 5 per cent sample
available in the census data. We complement our analysis using census data from 1960 to
2 The impact of unilateral divorce legislations on divorce rates remains an open question. Peters (1986, 1992),
using a cross-section of data on women, finds no effect. Allen (1992) and Friedberg (1998) obtain the opposite
result using an alternative model specification and panel data recording all the divorces by state and year
respectively, while Wolfers (2006) finds only a small long run effect of unilateral divorce regulations. In a
different line of research, Dee (1999) and Stevenson and Wolfers (2005) examine the impact of unilateral
divorce on spousal murders, self-reported domestic violence and suicide, with opposite results. Using a different
empirical strategy, both Gruber (2004) and Johnson and Mazingo (2000) find that exposure to unilateral.
divorce as a youth appears to worsen adult outcomes such as education, labor force participation and family
income. Finally Chiappori, Fortin and Lacroix (2002) analyze the impact of divorce law on labor supply,
finding substantial evidence of a change in bargaining associated with a change in the laws.
3 Some papers have been looking at the impact on childbearing for women exposed to unilateral divorce as a
youth. Gruber (2004) and Johnson and Mazingo (2000) both found a rise in the number of children. We do not
focus on this paper on the exposure to unilateral divorce as a youth, but on current unilateral divorce regime.
Focusing on women resident on states that introduced unilateral divorce, Peters finds no impact of a change in
the law on fertility; her result is probably driven by data limitation. She compares only one pre-unilateral
divorce and one post-unilateral divorce year of treatment.
1990 to confirm that our results can be distinguished from pre-existing trends in fertility, and
we combine birth certificates ad March CPS data to study the impact of the law on marital
versus non-marital fertility and out-of-wedlock ratio, a result never studied in the literature.
Finally, we construct a comprehensive series of marriage rates for the period 1956-1995, and
use the Census 1980 5% State sample to analyze the fertility history of women in their first
two years of marriage.
The paper is organized as follows. After a brief overview of the legislative history of
divorce laws in the USA, section two analyzes the relationship between fertility and divorce
laws, section three sets up the empirical methodology. Section four contains the main results
and specification checks, section five and six investigates more in details the mechanisms
underlying our results on fertility and section seven concludes.
2 Divorce Laws and Fertility
Between 1968 and 1977 the majority of the states in the US passed from divorce with
mutual consent to unilateral divorce. In the previous regime, a spouse’s desire to end a
marriage was not a sufficient reason to be granted a divorce: the spouse petitioning a divorce
had to prove not only that the other spouse was responsible for the marital breakdown, but
also that him/herself was not even partly at fault for the marriage’s failure. Starting in the late
1960’s the states began to enact several legal reforms that simplified legal difficulties in
obtaining a divorce. At first, with “no-fault” divorce laws, divorce could be obtained upon
mutual consent of the parties involved. Immediately after, or contemporaneously, unilateral
divorce statutes made it possible for one spouse to obtain a divorce without the consent of the
other4. Table 1 summarizes the changes in the law in all US states.
There has not been systematic evidence on the impact of divorce laws on fertility. In
theory, one view holds that children constitute “marital capital” (Becker, Landes and
Michael, 1977.) Thus, a couple produces goods which are more valuable inside than outside
4 This paper focuses on unilateral divorce. We do not consider any issue related to the division of property;
unilateral divorce was usually accompanied by an equal division of property, but not the other way around.
the relationship. By reducing the value of marriage, due to a higher probability of divorce,
unilateral divorce law should imply lower fertility. Bargaining models (Brinig and Crafton,
1994, Mc Elroy and Horney, 1981, and Lundberg and Pollak, 1996) also imply a reduction in
fertility: according to these models all family decisions are made in strategic ways that
depend on the enforceability of the contract and the outside opportunities of each partner.
With unilateral divorce outside options become relevant since the contract is now not-
enforceable. The spouse with outside option has a better bargaining position and is able to
obtain a larger share of the couple’s joint production. For that reason the other spouse will
prefer to invest in market activities or in human capital at the expense of marriage specific
investments, including children.
All these models predict then a decline in marital fertility, assuming implicitly that
the incentives for unmarried people remained unchanged. However, a change in divorce law
could imply a change in the composition of individuals in the marginal marriage through a
selection into and out of marriage. There are potentially two effects: If the cost of exiting a
bad marriage goes down one may choose to enter marriage more easily. On the other hand,
if marriage is so easily broken, i.e. the value of commitment is diluted, why marry to begin
The existence of selection makes the prediction of the impact of divorce law on
fertility harder to identify. Marital fertility could go up or down or stay the same depending
upon the relative strength of different forces. First, if there are fewer children in bad
marriages, (which are more likely to end sooner when divorce is easier) we should observe
an increase in fertility due to a selection out of a bad marriage into a good one. Second, since
the cost of exiting marriage is now lower, people will be more likely to marry because they
have an easier exit option. In this case the quality of matches can increase or decrease: it is
possible to imagine that a higher number of marriages will increase the probability of finding
a better match, on the other hand people might be less careful in looking for a partner since
they know that the cost of splitting is lower5. Third, one may argue that since the “value of
5 There is conflicting empirical evidence on this effect. On one hand, Choo and Siow (2003) measure the gains
to marriage over time, using the frequency of matches across different types of market participant and find a
substantial decline from 1970 to 1980. On the contrary, Weiss and Willis (1997) and Mechoulan (2003) using
marriage” declines people choose not to get married, a point made by Rasul (2004). Marital
fertility could then increase/decrease or stay the same, depending on the comparative strength
of these effects.
Lets’ now turn to out of wed lock fertility. On the one hand, reduction in the cost of
exiting marriage will make more people “attempt” a marriage match, especially those who
plan parenting. Thus out of wedlock fertility should go down because some of those who had
children out of wed lock before may now choose to marry if exit from marriage is easier.
This also implies that when unilateral laws are introduced marital fertility rates should go up
immediately after wed lock and that the number of never married women should go down. It
could also imply that the rate of marriage goes up, because more matches are tried at every
point in time. In other words, as divorce become easier, some of those contemplating
childbearing will choose to marry, reducing out of wed lock fertility. On the other hand if the
value of marriage goes down, people could decide to marry less and, therefore, have children
out of wedlock.
We will show below that the first effect vastly dominates.
3 Data and Econometric Specification
We use the births certificates of the National Vital Statistics of the USA to calculate
different measures of fertility. The births certificates data contain individual records on every
birth that took place in the United States between 1968 and 1999 to mothers ages 10 and
older. Prior to 1968 micro data on birth certificates are not publicly available. Birth
certificates contain information on mother’s characteristic including age, race, marital status
and education. We aggregate these data into cells defined by state of residence of the
mother, race and age, to construct state level panel data of total fertility rates, birth rates, and
the ratio of births-out-of-wedlock to total births and marital-non marital fertility from 1968 to
the National Study of the High School Class of 1972 and the CPS respectively find evidence of better matches
associated with the introduction of unilateral divorce.
1999. The total fertility rate (TFR) is the standard way of measuring fertility. It estimates the
number of children a cohort of 1,000 women would bear if they all went through their
childbearing years exposed to the age-specific birth rates in effect for a particular time. The
TFR is calculated using the methodology applied by the National Center for Health Statistics
(described in the appendix). We construct state-year cells containing the average number of
children for women in all their childbearing period. The birth rate is defined as the total
number of childbirths observed per 1,000 women of the appropriate demographic group; it is
a crude measure of fertility but it would allow us to study the impact of the law for marital
status. The fraction of births out-of-wedlock is defined as the ratio of out of wedlock births
over total births6.
Population estimates and age and race composition are obtained by the Bureau of the
Census for the period 1968-1999.7 We also combine birth certificates data with the March
round of the Current Population Survey (CPS) to construct the number of married-unmarried
women by age and race using the CPS weights. CPS data are also used to construct the labor
market and education variables at the state level. Since the micro data on birth certificates are
available only from 1968, we complement our analysis using four decades of Census from
1960 to 1990, to confirm that our results can be distinguished from pre-existing trends in
fertility. Descriptive statistics for adopting and non-adopting states are reported in the
appendix (Table A3).
We construct a very comprehensive series of administrative data for marriages in the
US from 1956 to 1995. Our data comes from the marriage certificates of the United States for
the period 1968-1995 (the marriage certificate data cover roughly 44 states depending on the
specific year, see Appendix for more details), moreover we complement the dataset with
hand-entered data from the annual editions of the Vital Statistics for 1956-1967 and for those
states that are not covered in the marriage certificates dataset for the period 1968-1995. The
6 Some states did not report the information on legitimacy status prior to 1979, (See Appendix 1 for details)
7 Population estimates for the intercensal years are obtained by the U.S. Census Bureau at
count of administrative data is used to construct crude marriage rates- the number of
marriages per 1000 of the population8. Our checks confirm that our marriage rates almost
perfectly match the official numbers reported in the vital statistics. Finally, we use the
Census 1980 5% State sample9 to study the fertility history of women in their first two years
3.2. Econometric Specification
We consider the following panel data regression of the log of the total fertility rate in
state s at time t,
log( f ) , for the period 1968-1999:
log( f ) = ?U + ? + ? + ?X
+? ? ? t + ?
where U is a dummy equal to one if state s has a unilateral divorce regime starting from
year t, ? and ? refer to state and year fixed effects, X is a set of controls and ? ? t
represents state specific trends, where t is a year trend.
Prior to 1967, divorce was mutual in almost all the states in the US. Between 1967
and 1987 almost two thirds of the states introduced unilateral divorce. Hence the causal
effect of unilateral divorce in our specification is identified from variation across states, time
and between adopting and non-adopting states. The impact of a change in divorce law is
captured by the coefficient ? , which represents the change in fertility rate attributable to the
Table I gives the year in which these laws were passed by state. We follow Gruber
(2004) who codes divorce as unilateral when it requires the consent of only one spouse and is
granted on grounds of irreconcilable differences. Since there is some debate in the literature
about this coding and how classify a state’s divorce laws, as well as the timing of the laws,
we have tested our results with different available coding. Our results are robust. We
consider two specifications. In the first, we include state and year fixed effects, but ignore
state-specific trends (? = 0 ), in the second state-specific trends are included.
8 Data on the state population for the period 1956-1998 is obtained by Wolfers (2006)
9 The 1980 5% State sample covers approximately 11,337,000 person records. Data can be downloaded from