Literature Review

Divorce and Labour Force Participation

      A simultaneous rise in divorce rates and female LFP rates occurred during the post-World War II
era. As noted by Becker, et al. (1977), one possible explanation for these trends is that wage increases and
greater earnings ability among females increased the opportunity costs of being married. In addition, wives
working outside the home may invest less in marriage-specific capital than their non-working counterparts,
reducing the gains from marriage for both men and women. This interpretation of Figure 1 suggests that
the increase in LFP among women was a causal factor that led to the increased divorce rate. However, it is
also possible that the causality moves in the other direction. If a woman becomes divorced, she may need
to enter the workforce to support herself and her family.

      The number of divorced women working would naturally increase as divorce rates increase, thus
raising the female LFPR. As more women observe the incidence and impacts of divorce among their
mothers and peers, they may adjust their own expectations regarding divorce risk upward, and subsequently
increase their own LFP while married to insure financial independence. Becker, et al. (1977, pg. 1181 )
state, “...the secular growth in wages, which contributed significantly to the growth in the labor force
participation of women, especially married women, probably also contributed significantly to the growth
in divorce rates. Again causation probably flows both ways: divorced women (and women who anticipate
divorce) have higher wages because they spend more time in the labor force.” To examine the relationship
empirically, 

      Michael (1985) included lagged divorce rates and lagged LFP rates as independent variables
in time-series equations explaining subsequent divorce and LFP rates. He found that lagged divorce rates
were positively correlated with subsequent LFP rates, but there was no significant relationship between
lagged LFP rates and subsequent divorce rates. Johnson and Skinner (1986) analyzed data from the
Michigan Panel Study of Income Dynamics (PSID), which allowed them to estimate predicted divorce
probabilities based on actual divorce. They then used this predicted probability of divorce and actual
divorce in estimates of female LFP, and though both variables had positive coefficients, they were
statistically insignificant.

      Green and Quester (1982) utilized U.S. Census’ Survey of Economic Opportunity data to generate a
predicted divorce probability based on the demographic characteristics of married women. They found
married women’s labor supply increases with divorce risk. In addition, although it is often hypothesized
that working could increase marital instability for married women, this relationship was found to be
statistically insignificant in both papers (Becker, et al. 1977). Haurin (1989) used a dynamic model to
estimate the labor market reactions of women who experience a deviation in husband’s actual work hours
from the expected amount of work hours, as would occur when a husband loses his job, falls ill, passes
away, or when a couple separates. He concluded that there is a significant increase in a woman’s LFP
following a divorce or separation. He also confirmed Johnson and Skinner’s (1986) finding that increases
in the likelihood of divorce lead to increases in married women’s labor force activity.

      More recent studies on divorce suggest that divorce risk may not be as influential on female labor
supply as the earlier studies suggested. Sen (2000) compared responses to divorce risk among 1944-1954
and 1957- 1964 birth-cohorts using sub-samples from the National Longitudinal Surveys (NLS). The panel
structure of this data allowed Sen to proxy current divorce risk using actual divorce in the future, which
was then included in an LFP regression. Age at the time of marriage was also used in the LFP estimation
as an instrument for divorce risk because research suggests that divorce risk decreases with age at the time
of marriage. The results indicated that the impact of divorce risk on labor supply is significantly smaller for
the younger cohort than the older group. Though divorce had a positive effect on both cohorts by each
measure, divorce risk had a substantial impact on LFP for the older cohort while having a small (yet still
significant) effect on the young cohort. Because having a female child is associated with higher divorce
risk, Bedard and Deschenes (2003) used sex of first born child as an instrumental variable for divorce risk
when testing for labor market outcomes of divorce.

      Though the estimates of person-adjusted household income without this variable indicated that the
economic well-being of divorced women is lower than married women’s economic well-being, the
instrumental variable results refute this. They found that ever-divorced mothers have higher levels of
income than neverdivorced mothers and they concluded that divorce does not affect the decision to
participate in the labor force, but does increase the hours and weeks worked by mothers. As noted in the
introduction, the potential endogeneity of using actual divorce or divorce rates to measure divorce risk
makes research based on these variables somewhat suspect. In response, researchers have used exogenous
changes in divorce law to examine the relationships between divorce and female LFP.

      Johnson and Skinner (1986) included residence in a state with no-fault divorce legislation in their
predicted labor supply equations. They concluded that living in a state with a no-fault divorce law has a
negative impact on women’s labor supply. However, Johnson and Skinner’s (1986) analysis used PSID
data from 1972, when only a few states had changed their divorce laws to no-fault; this created a small
comparison group and a limited amount of time for the laws' effects to occur. Alternatively, Peters (1986),
using 1979 CPS data, found a higher probability of women participating in the labor force in no-fault states.
Parkman (1992) also used the 1979 CPS and a similar methodology to Peters (1986) to estimate the effect
of no-fault divorce legislation on married women’s labor supply. Consistent with Peters (1986), he also
found about 2 percent higher rates of married women’s LFP in no-fault states.

      However, he attributed this difference to lower compensation for women’s marriage-specific
investments in states with no-fault divorce law. In particular, he used differences in property division laws
to illustrate that married women’s LFP is greater in states with no-fault divorce because at the time of
divorce, women’s human capital losses from not engaging in market work are compensated at lower rates
than in mutual consent states. In later research using. the Time Use Longitudinal Panel Study, 1975-1981,
Parkman (1998) again concluded that no-fault divorce laws were associated with a greater number of hours
worked among married women. He suggested that women work to protect themselves from the potential
costs of divorce. Gray (1998) utilized Census and CPS data, and he tested for the effect of no-fault divorce
law in 1980 across states. His results indicated that the probability that a married woman participates in the
labor force is 1.6 percent higher in states with no-fault divorce law. He then used data from 1960 to 1980
to measure the change in married women’s LFP in states with and without no-fault divorce laws. The results
from this analysis found a small and insignificant impact of no-fault divorce law and Gray suggested that
without considering the marital property laws of a state, “divorce legislation has no significant impact on
married women’s increasing labor-force participation rates during the 1970’s” (Gray 1998, pg. 634).

      Gray (1998) categorized states into three types of property law: common property, community
property and equitable distribution. With the enactment of no-fault divorce laws, common property favors
the wife in divorce settlements, community property tends to lead to redistribution of assets to the husband,
and equitable distribution gives the court discretion on property division and thus does not favor either
party. The property laws were accounted for in Gray’s (1998) analysis by interacting them with the no-fault
variable. Consequently, the results from analyses using the Census, CPS and PSID all indicated that the
adoption of a no-fault divorce law in a common property states is associated with decreases in married
women’s LFP. In addition, married women in states with community property laws had significant
increases in LFP after no-fault legislation was enacted. Chiappori, et al. (2002) used PSID data from 1988
to test the effect of divorce and property legislation on married women’s labor supply.

      They created a “divorce index” comprised of four attributes associated with the favorability of each
state’s divorce law towards women: no-fault divorce laws, property division laws, support order
enforcement, and the settlement value of educational degrees. This index was then used in a regression of
the hours worked by married women. In their analysis, the “divorce index” was negatively correlated with
hours worked. In other words, in states where the divorce laws are more favorable toward women, married
women are likely to work fewer hours than in states where divorce law is less favorable towards married
women. The research presented in this subsection indicates the myriad of findings regarding the relationship
between divorce rates and female LFP. Empirical results consistently show that increases in divorce rates
are associated with increases in LFP rates for females. Though a causal relationship between female LFP
and divorce is theoretically possible, empirical research has not found strong evidence for this; increases in
female LFP do not appear to have led to increases in divorce rates. There has also been evidence suggesting
that women’s LFP response to divorce risk is not as strong for recent cohorts of women when compared to
earlier cohorts.

      Research addressing the effect of divorce law changes in the United States on the LFP of married
women was also presented in this section. Though Johnson and Skinner’s (1986) analysis suggested that
nofault divorce had a negative impact on married women’s LFP, later research has consistently found that
married women living in states with no-fault divorce laws are more likely to work, but the effects are small.
Recent research in this area has also highlighted the complexity of the LFP decision for married women
because marital property laws and other legal factors are found to be related to the effects of no-fault divorce
laws on women’s LFP. Previous research has ignored the possibility that the various laws have differential
impacts among married women with and without children. The impact of no-fault divorce law on female
LFP could be larger than suggested by previous research because the results reported were the average
effect for mothers and non-mothers. By separating married women into mothers and non-mothers, this
paper allows for varying impacts of no-fault divorce law on the two groups.