Is Making Divorce Easier Bad for Children? Jonathan Gruber, MIT and NBER July, 2000 I am grateful to Amber Batata, Jeffrey Hoffner, Becky Neuschatz, Kokkeong Puah, and Ebonya Washington for excellent research assistance, and to the Smith Richardson Foundation for financial support. One of the most striking trends in postwar social indicators in the U.S. is the rise in divorce rates. Figure 1, updated from Friedberg (1988), illustrates the rate of divorce per 1000 persons in the U.S. over time. After staying at low levels for many years, divorce rates began to rise precipitously in the mid-1960s, with the rate of divorce rising by over 200% in only fifteen years. This “breakdown of the traditional family” has been decried in many circles, particularly due to its perceived negative implications for children. Indeed, there are large literatures in sociology, developmental psychology, and economics which document the negative consequences for children of divorce, both as children, and then later as adults. A common villain in these criticisms is state regulations that increase the ease of divorce. The rise in divorce rates corresponds quite strikingly to the advent of state regulations that allowed for unilateral divorce, or divorce that did not require the explicit consent of both partners. Unilateral divorce laws were passed in a number of states in the wake of the no-fault divorce revolution that moved the basis for divorce from the fault of one spouse to general “irreconcilable differences” (Weitzman, 1985). These unilateral laws substantially increased the ease of divorce by allowing one partner to leave without obtaining the consent of the other. The second line in Figure 1 shows the percent of persons living in states with unilateral divorce laws in place. Exposure to these unilateral divorce regimes rose in tandem with divorce rates over the 1970s. As a result, critics of rising divorce rates have called for a reversal of the 1970s trend towards unilateral divorce regimes, in an effort to maintain the traditional two-parent family and improve child outcomes. Two states (Arizona and Louisiana) have passed “covenant marriage” laws whereby the couple receives pre-marital counseling and signs a covenant that makes divorce 2 more costly through separation periods with intensive counseling (although still possible unilaterally). These laws have been proposed in at least 17 other states as well. And there is a broader movement in religious communities as well to increase the stringency of the marriage contract (Nordinger, 1998). In a recent debate over such regulations in Virginia, the Associated Press Wire on February 14, 2000 quoted the legislation’s author, Rep. McDonnell, as saying that “he was concerned about the rising divorce rate and the impact it was having on families”. The key underlying assumption of this movement is that regulations which increase the ease of divorce have negative implications for children. This line of argument involves three key suppositions, however: that the increase ease of divorce under state regulations contributed to (or even fully caused) the rising rate of divorce in the U.S.; that divorce is actually bad for children, relative to the counterfactual of maintaining potentially damaged marriages; and that changes in divorce regulation do not have any other impacts on families which may offset any direct influences through divorce, such as through the decision to enter into marriage or through influences on the nature of within-family bargaining. The existing evidence on the first of these suppositions is quite mixed; the evidence on the second has yet to convincingly address potential selection biases associated with the decision to divorce; and there is no empirical work on the third supposition. Thus, on net, there is no convincing evidence that unilateral divorce regulations actually have an adverse impact on child well-being. The purpose of this paper is to provide a careful assessment of the implications for children of making divorce easier, in a framework which allows me to integrate all of the issues above. I do so in three steps. I first show that there is a sizeable and significant impact of unilateral divorce on the stock of divorced parents of child-bearing age. I then directly examine 3 the implications of growing up in a unilateral divorce regime for these children when they themselves become adults, allowing for a long run assessment of the ultimate consequences for children of growing up in an environment which divorce is easier. Finally, I extend the basic results to consider the process through which unilateral divorce affects outcomes, by examining how the effects vary with length of exposure to unilateral divorce laws and by age, by exploring the effects of weaker vs. stronger reforms of divorce law, and by considering whether there are were differential impacts on children who were born before and after unilateral divorce was in place. I do so by using data on state divorce regulations, matched to information from the 1960, 1970, 1980, and 1990 censuses. This broad historical sweep allows me to consider both the direct impact of unilateral divorce regimes on the cohorts of adults at the time they are passed, and the later impact on their children when they themselves become adults. And I can do so within models that account for both fixed state preferences towards divorce, as well as changes in those preferences over time. I have several notable results, besides the finding noted above that unilateral divorce increases the stock of divorces. I find that there are important impacts of being exposed to unilateral divorce as a youth on one’s later outcomes. Both women and men who are exposed to unilateral divorce as a youth are: more likely to be married, although this appears to reflect earlier marriages rather than a long-run rise in the marriage rate; less educated, and in particular less likely to be a college graduate and more likely to be a high school graduate instead; and living in families with lower incomes. Women who were exposed to unilateral divorce as a youth are much less attached to the labor force and have lower earnings; men, on the other hand, are 4 marginally more attached to the labor force and earn more. Both the magnitude of the effects, and the patterns with respect to the amount of time exposed to unilateral divorce, suggest that there are important effects of unilateral divorce beyond increased divorce rates. I also find that the impacts of reforms which allowed for unilateral divorce, but only with a requirement of a separation period, have much weaker effects, suggesting that the nature of regulations matter. And I find that the impacts of unilateral divorce on the incomes children born after reforms are in place is much weaker than on children for whom such laws were a “surprise”, suggesting that the long run implications of unilateral divorce may be mitigated relative to what is documented here. The paper proceeds as follows. In Part I, I present background on the potential links between unilateral divorce laws and child outcomes. In Part II, I discuss my data and empirical strategy. Part III presents the results on the how unilateral divorce impacts marital status. Part IV then examines the impact of exposure to unilateral divorce as a youth (and currently) on adult outcomes. Part V considers a number of extensions of the basic results. Part VI concludes. Part I: Background As noted in the introduction, there are three key links between legislation which makes divorce easier, and the outcomes of children. In this background section, I review each link in turn. Before doing so, I provide some background on the unilateral divorce regulations which are the subject of this study. 5 Legislative Background As nicely reviewed by Weitzman (1985), traditional state regulation in the U.S. provided for divorce only for grounds such as infidelity, physical abuse, etc. Moreover, such divorce had to be mutually agreed upon by both partners. This law was widely viewed as inadequate, largely because of the enormous financial and emotional transaction costs that the establishment of fault placed on the divorce process. Indeed, marriages that were viewed by both parties as “broken” for mundane reasons could not be dissolved without more elaborate justification. And fault was viewed as a tool that was often used by one spouse (typically the wife) to “extort” excessive settlements from the other spouse. As Wheeler (1974, quoted in Weitzman, 1985) wrote: It was unanimously agreed that elimination of the present grounds... would conform the law to prevailing reality, eliminate the existing evils of dissimulation, hypocrisy, and outright perjury, and end the use of conduct not formally alleged as a weapon in obtaining extortionate and frequently inequitable and unworkable concessions from the defending spouse. To summarize, Weitzman’s (1985) basic thesis is that these laws were viewed as broken because the process of fault divorce was very messy, and because the system was too biased in favor of the defendant spouse, through the use of the threat mechanism. But, according to Weitzman, the reformers did not appear to recognize that they might be swinging the pendulum too far the other direction, by removing the powerful property rights that mutual consent, fault divorce gave to women. The first steps in these reforms was moving to no-fault divorce, which was in place before 1950 in a number states,1 while maintaining the mutual consent feature. Unilateral 1 Weitzman (1985) claims that the first no fault law was in California in 1970, but this is because her definition of no fault requires that this be the only grounds for divorce. I use the more general definition of no fault which allows for divorce for the general grounds of 6 divorce, which allowed divorce with the consent of just one rather than both spouses, was rare before the late 1960s but was in place in most states by the mid-1970s. My paper focuses on unilateral divorce, following the economics literature growing out of Peters (1986), which is discussed in more detail below. I have documented the availability of unilateral divorce in each state from 1950 to the present, updating the legislative details in Friedberg (1998). The results of doing so are presented in Table 1. States could pass either unrestricted unilateral divorce, or unilateral divorce with the requirement that spouses live separated for some period of time (typically 1-5 years). I focus primarily on unilateral divorce laws that do not include separation requirements. Unilateral divorce laws with separation requirements are quite different types of regulations. Many of the separation requirements are quite lengthy; 95% of states with such regulations have separation lengths of one year or more, and two-thirds have lengths of two years or more. These laws have been in place much longer and many were weakened by the unilateral divorce “revolution” of the early 1970s. Friedberg finds that the impacts of such regulations on divorce flows is positive, but much weaker than for unrestricted unilateral divorce. Does Unilateral Divorce Affect Divorce Decisions? As noted above, unilateral (and no fault) divorce is perceived to be a cause of the dramatic rise in divorce rates in the U.S. Besides the obvious time series parallels, the theory behind such arguments is perceived to be quite straightforward: if there is one partner that wants to terminate the marriage, but the other does not, then a unilateral divorce will cause the marriage “incompability” as one of the options. 7 to end. In a well-known article, Peters (1986) pointed out the flaw in this theory from the perspective of Coasian bargaining. If a marriage is a contract between two partners, and one partner wants to end that contract, he or she can just pay their partner for that privilege. Thus, moving from multilateral to unilateral divorce does not change the fundamental likelihood of divorce; it simply changes the amount of payment that must be made from the partner who wants to leave to the partner who wants to stay. That is, under the typical presumption from the 1970s that men were the ones that wanted to terminate their marriage contracts, unilateral divorce would not lead to rising divorce; it would simply lead to lower alimony and child support payments to the wives left behind. Peters supported her theoretical argument with an empirical analysis of the impact of unilateral divorce regimes on divorce rates. She used a cross-section of data on women to examine whether women were more likely to be divorced in states with unilateral divorce regimes, and she found no significant correspondence between the two. This striking conclusion generated a number of follow-on analyses, with mixed results. Allen (1992) claimed that for alternative specifications of the divorce variable and the model there were impacts of unilateral divorce regimes on divorce, but Peters (1992) disputed his law classification and his omission of important regional controls. Friedberg (1998) carefully revisited this question using panel data on divorce rates by state and year, and found that for the most detailed specification there was an impact of unilateral divorce on divorce rates; similar analysis is found in Reilly and Evenhouse (1997). Wolfers (2000) shows that much of Friedberg’s effect on flows of divorces arose from a large increase in divorces soon after the 8 passage of unilateral divorce laws, and that the long run effect on the divorce flow is quite small. But Gray (1998) does not find an impact of unilateral divorce on divorce rates using the 1970 and 1980 censuses. Thus, the impact of unilateral divorce regimes on divorce rates remains an open question. Is Divorce Bad For Children? The second supposition that drives criticisms of easier divorce regulations is that divorce is actually bad for children. This supposition is supported by an enormous literature which can only be crudely summarized here. After reviewing 92 studies, Amato and Keith (1991) report that children of divorce have more difficulty than children in intact families adjusting both socially and psychologically. Surveys show that children of divorce are more likely to exhibit behavior that is antisocial, impulsive or acting out. They are more likely to become delinquents (Matsueda and Heimer 1987, Zill et al 1993), and perform worse academically; Guidubali, Perry and Cleminshaw (1984) find that first, third and fifth graders from divorced families (compared with children of intact families) scored lowered on IQ, reading, spelling and math scores. They are also more likely to suffer psychological symptoms such as dependency, low self-esteem, anxiety and depression. Children whose parents have divorced, twice as often (20% vs. 10%) in comparison to children from intact households, score above clinical cutoffs on psychological tests of behavior problems. (See for example Achenberg and Edelbrock, 1983, Hetherington and Clingempeel, 1989, Isaacs, 1986). The research on adolescents from divorced families also documents negative consequences. Adolescents of divorce are two to three times more likely to drop out of school, 9 become pregnant, engage in antisocial and delinquent behavior, and score above clinical cutoffs on standardized tests of behavior (Acenback and Edelbroch, 1983). They begin to date and have sex at younger age (Flewelling and Bauman, 1990). Other researchers find that these youngsters are more aggressive, noncompliant, sexually active and likely to use and abuse drugs and alcohol than adolescents from intact households (Baumrin, 1989; Doherty and Needle, 1991, Dornbusch et. al, 1985). Adolescents of divorce are more likely to have lower academic performance and higher dropout rates even after controlling for socio-economic status (Guidubaldi et. al., 1984; Krein and Beller, 1988). By age 23, a British longitudinal study of children shows, children of divorce are more likely to leave home because of friction (Cherlin et. al. 1995). Adult children of divorce are less likely to attend or complete college, more likely to be unemployed and on welfare and to possess fewer resources (Keith and Finley, 1988; McLeod, 1991; McLananahan and Sandefur, 1994; Aquilino, 1994; Cooney, 1994). They are less likely to marry, but if so to marry at a younger age. They are more likely to have marital problems, choose unstable partners, have children out of wedlock and divorce earlier and with greater frequency (Amato and Keith, 1991; Glenn and Kramer, 1985; Hetherington, 1972; Hetherington and Parke, 1986; Kurdek, 1991; McLanahan and Bumpass, 1988; Wallerstein and Blakesley, 1989; Wu and Martinson, 1993). Thus, the negative implications of divorce for children are broadly supported by a large previous literature (and the even larger volume of research not summarized above). But a central limitation of these studies is that divorce is not an exogenous event with respect to other determinants of child outcomes. Another large literature on the determinants of divorce finds that divorce is strongly correlated with socioeconomic characteristics that also determine child 10 outcomes, such as income and family size. For example, divorce rates are higher when men have experienced serious unemployment within the past five years (Ross and Sawhill, 1975), and states with higher male earnings have lower divorce rates (Ferber and Sander, 1989). Moreover, in theory, the implications of divorce for child well being are ambiguous. While depriving the family of one potential earner and caregiver can clearly have negative implications, breaking up emotionally or physically harmful marriages can have benefits for children. A number of the studies above attempt to control for socioeconomic characteristics in assessing the impact of divorce (e.g. McLanahan and Sandefur, 1994). But they do so armed with only a limited set of family background characteristics that might not fully capture underlying differences between families and do and do not divorce. Even conditional on background characteristics such as socioeconomic status, families that choose to divorce are different in unmeasured ways, and these differences can have important implications for children; for example, it may be the families with the weakest tastes for marriage that divorce, and this could lead to lower rates of marriage among their children not because of parental divorce per se, but rather because of inherited weak tastes for marriage. Concerns about selection are heightened by findings such as (a) conditional on any exposure to single parenthood, the total amount of exposure is irrelevant for outcomes, and (b) conditional on having a single parent, having a grandmother present has a large negative impact on outcomes. What is required to appropriately identify the impacts of divorce is an exogenous instrument that causes some families to divorce and others not, based on a factor independent of the determinants of their children’s outcomes. No previous study has been able to uncover such an instrument, making it somewhat hard to interpret causally this large literature. 11 Moreover, the previous literature has focused on the impact of the average divorce, not of the marginal divorce that is induced by a change in the regulatory regime. Even relative to the effects of divorce on average, the divorces induced by a shift to unilateral divorce regulation may have larger or smaller negative implications for children. For example, the average divorce may be taking place for reasons of fault, such as spousal abuse or infidelity, while the marginal divorce that takes place because divorce is made easier is due to spousal incompatibility. If marriages that end due to abuse/infidelity have worse/better implications for children than do marriages that end due to incompatibility, then unilateral law-induced divorces will have better/worse implications than the average divorce. Are There Other (Potentially Offsetting) Impacts of Unilateral Divorce? The third supposition behind thee arguments is that there are not offsetting impacts of unilateral divorce on family structure. In fact, there are at least four reasons why, even if unilateral divorce laws lead to more divorces, the impacts on children may be minimized. First of all, and most obviously, unilateral divorce can lower the incidence of separation, as families substitute official for de facto divorce; that is, parents who are living apart may now formalize a divorce. If it is having two parents in residence that is important for child development, then such a shift may be of little consequence for well-being. Moreover, unilateral divorce may increase the incidence of marriage, by reducing the barriers to exiting that marriage. That is, individuals who are reticent to enter marriage in a regime where unilaterally terminating that marriage is not possible may be more willing when they have a source of exit. That is, unilateral divorce may lead to more “marital churning”, with 12 both more marriages and more divorces, and little impact on the net stock of married couples (or of children living in two-parent households). If what matters for child well-being is being in a two-parent household, then even if divorce is rising in unilateral divorce regimes, child outcomes may be unaffected. This point has not been considered by previous empirical studies, which have focused on divorce, and not marriage, as the outcome of interest; but it is discussed theoretically in Bougheas and Georgellis (1999). Third, even if divorces rise in aggregate, as suggested by Friedberg (1998), the impact may predominantly be among households without children or with a small number of children; as noted above, divorce probabilities fall with the number of children. This would minimize the impact of divorce on children, relative to the overall number of divorces. Moreover, there is a fourth effect that has ambiguous implications, but which nevertheless confounds the simple interpretation of unilateral divorce noted above. Even if families see no change in marital status as a result of unilateral divorce, making divorce easier can change the nature of the bargaining relationships between husband and wife. If these relative positions of power have implications for child well-being, then this is an additional channel through which unilateral divorce can impact children. For example, there is a large literature which suggests that resources in the hands of women are more beneficial to children than are resources in the hands of men (e.g. Strauss & Thomas, 1995). If unilateral divorce weakens the within-household bargaining position of the wife, because the husband can now leave without large compensation, this may have negative implications for children, who benefitted from the relative power of their mothers. Of course, this effect could operate in a number of ways, but it illustrates the difficulty of the simple argument made above. 13 This paper will not be able to separate the direct effects of unilateral divorce through divorce versus the indirect effects through these other channels. But the existence of these other channels highlights that unilateral divorce laws are not useful instruments for divorce. That is, moving to a unilateral divorce regime has impacts that go beyond any simple effect on the divorce rate. Thus, the results from this exercise should be interpreted as the reduced form effects of making divorce easier on outcomes, and not as the structural effects of divorce on outcomes. A new paper by Johnson and Mazingo (2000), written at the same time as this paper, also explores the implications of unilateral divorce for child outcomes as adults. They follow a somewhat different strategy than the one employed here, using cross-sectional data from the 1990 census to compare the effects of unilateral divorce on those exposed to the law for different lengths of time as a child. A number of their key findings echo those reported below, although I do show that a number of the effects that I find do not appear to follow a monotonically increasing pattern with amount of time exposed. Part II: Data The data for this analysis comes from the 1960, 1970, 1980, and 1990 PUMS files from the U.S. census. These data provide very large samples of children and adults so that information can readily be gleaned on the response to state divorce laws. Due to the large samples, and the fact that the relevant legislative variables vary only at the state/year level, the data are collapsed into state/year cells for the analysis. These data are parsed into three data files for the analysis. The first has information by 14 state of residence and year for children 0-18 years old, so that there are 51*19 = 918 cells for each year, or a total of 3,876 cells. All regressions are weighted by cell size to replicate the underlying microdata. This file contains information on the marital status of the parent with whom the child resides, and their per capita family income. These are essentially the same data used by Gruber et al. (1998) for the 1980 census, extended to other years. The second data set is comparable, but it is created for adults of child-rearing age (25-50 years old), to examine the impact of unilateral divorce laws in one’s state of residence on divorce probabilities. The third file is a data set for adults age 25-50 that is organized by age, state of birth, state of residence, and sex. This finer level of detail is necessary for us to examine jointly the impact of unilateral divorce regimes with which the individual grew up and unilateral divorce regimes within which they now reside. The data are also divided by sex because many of the outcomes I examine (e.g. labor force participation) differ significantly across the sexes. These data have information for each of these cells on marital status, family size, family income, individual education, individual work status, and individual earnings. Once again, these are all averages for the year/age/state of birth/state of residence/sex cell, and the regression is run at that level, weighted by cell size. These census data are matched to information on the presence of unilateral divorce regimes across states and over time. The means of the first and third data sets are presented in Table 2; the means in the second and third data sets for adults are identical, since the third is just a more finely parsed version of the second. Among children, 6.6% are living with a divorced mother, and roughly 1% are living with a divorced father. Among all adults of child bearing age, 11% of females and 8.2% of males are divorced. 15 Part III: Does Unilateral Divorce Affect Marital Status? In this section, I address the first of the predicate questions raised above: does unilateral divorce actually affect marital status? I examine both the impact on the likelihood that adults are divorced, and the impact on other marital states that may be affected by this shift in legal regimes. As noted earlier, while a number of papers have focused on the impact of unilateral divorce on the flow of divorces, only Gray (1998) has looked at the stock, and he finds little impact. To assess the impact of unilateral divorce on marital status, I run regressions of the form: (1) DIVORCEajt = + 1UNILATjt-1 + 2 RACEajt + 3 a + 4 j + 5 t+ 6 a * t+ where a indexes ages, j indexes states, and t indexes years; DIVORCE is the cell mean divorce rate (or some other marital status indicator); UNILAT is a dummy for the presence of a unilateral reform law; RACE are dummies for percent black and white, respectively, in the cell; a , j , and t are full sets of dummies for age group, state, and year, respectively; and a * t is a full set of age*year interactions to allow for differential time patterns by age. This model controls for fixed factors that vary by age, location, or year, and is identified by the passage of unilateral divorce laws over time. One concern with this approach, however, is that there may be trends in marital status that are correlated with the passage of unilateral divorce laws. That is, unilateral divorce may pass where divorce is rising, rather than the opposite causal interpretation. The evidence in Wolfers (2000) suggests that there are not large pre-existing trends in divorce rates in the unilateral divorce states, once data back to 1960 is used. Nevertheless, I will attempt to address this 16 concern by including in the model, along with state fixed effect, state-specific trends. This is not a perfect solution given these census data, since there are only four underlying time series observations, so fixed trend models are fairly demanding. Moreover, if the endogeneity is in the short run, then trends that measure decade time spans (as with the census) will miss them. But this is the best feasible approach to the problem, and if the results are robust to the inclusion of trends it offers some comfort that they are not driven by endogeneity of the laws. Another econometric concern is that the variable of interest (unilateral divorce laws) varies only by state and year, while the underlying data here also varies by single year of age. As a result, I correct all standard errors for clustering on state of residence and year. The results of running regressions such as this for various measures of marital status are shown in Table 3. The first set of columns is for women, and the second set for men. Within each set of columns, I show results with and without state-specific trend controls. Within each cell, I present the coefficient, the standard error (in parentheses), and percentage impact (relative to means in Table 2). The coefficients are all multiplied by 100 for ease of interpretability. I find that there is a very sizeable and highly significant impact of unilateral divorce on the likelihood of being divorced. For women, unilateral divorce being place raises the odds of divorce by 0.0124 percentage points, or 11.2%. For men, the increase is 0.0096 percentage points, or 11.7%. The results are even stronger when state-specific trends are included, with the coefficient for females rising to 0.014 percentage points (12.7%) and for men rising to 0.010 percentage points (12.7%). On the other hand, there is no robust evidence on the odds of being either separated or never married. For women, there is some suggestion of a reduction in the odds of being never 17 married, but it is highly insignificant and small when state trends are included; for men, the effect is wrong-signed with trends included. These results show the effects on parents of child-bearing age. But, as noted above, the impacts of divorce reform may be different for parents who actually have children. So the next panel of Table 3 focuses on the marital status of parents of children in the data set. The regression run is the same as above, except that the sample is now children age 0-18, and the dependent variables are whether their mother or father are divorced, separated, or never married. The results for the parents of children echo those for adults of child-bearing age. There is a 0.01 percentage point increase in the odds that a child is living with a divorced mother, which is 14.4% of baseline, and a 0.0014 percentage point increase in the odds of living with a divorced father, which is 14.1% of baseline. Overall, the odds of living with a divorced parent rise by 0.011 percentage points, which is quite similar to the odds for adults of being divorced. These child-based results offer more evidence for increased entry into marriage when divorce is made easier. For mothers, in the model without trends, there is a very large impact on the odds of living with a never married mother; however, this result disappears when trends are included. For fathers, the impact is significant even when trends are included. There are negligible impacts on living with separated parents. Thus, there is clear evidence from the census data that making divorce easier increases the stock of divorced women and men, and that as a result children are more likely to be living with a divorced parent. There is mixed evidence for an offsetting impact on other marital decisions. 18 Part IV: Impact of Unilateral Divorce on Outcomes I now turn to assessing the impact of unilateral divorce as a youth on adult outcomes. To do so, I turn to the data which is created by both state of residence and state of birth, and run expanded regressions of the form: (2) OUTCOMEajbt = + 1 UNILATjt-1 + 2KIDUNIabt + 3RACEajt + 4 a + 5 b + 6 j + 7 t+ 8 a* t + where, in addition to the other indices, b indexes state of birth; OUTCOME is one of the measures of outcomes; KIDUNI is a dummy for having a unilateral divorce law in your state of birth before you were age 18; and b is a full set of state of birth dummies. Thus, this regression framework allows for both effects of contemporaneous and youth laws on outcomes. Once again, I also explore the sensitivity of results to state-specific trends, where now I include trends for both state of birth and state of residence. And the standard errors are now corrected for state of residence * state of birth * year clustering. The results of running these regressions for females age 25-50 are presented in Table 4. The first set of columns shows the coefficient on unilateral divorce law as a youth, with and without state-specific trends included in the model; the second set of columns shows the coefficient on current unilateral divorce laws, with and without state-specific trends. I show the results for several sets of variables, denoted in blocks in the table. Since there are many outcome variables and their effects are likely to be related, I first review below the full set of findings, and then after doing so turn to offering discussion of their implications. For regressions where the dependent variable is a cell mean of a discrete variable (e.g. marital status measures), the coefficients on unilateral variables are multiplied by 100; for continuous variables (e.g. number 19 of children, years of education, earnings), the coefficients are not multiplied by 100. Results The first block of results shows the impact on marital status and number of children; the coefficients for current divorce law parallel those shown in Table 3, and the results are indeed quite similar, suggesting little bias from examining the current laws in a vacuum (ignoring youth laws). But the coefficients for laws as a youth are strikingly different: I find that unilateral divorce as a youth is associated with no rise in the odds of being divorced, but a much higher likelihood of being married. The coefficient indicates that being exposed to unilateral divorce as a youth raises the odds of being married by 0.008 percentage points, or 1.1 percent of baseline. There is a correspondingly much lower likelihood of being never married. There is also a significant rise in the odds of being separated. These results are all very robust to the inclusion of state-specific time trends; indeed, the trends serve to increase the precision with which the coefficients are estimated. Corresponding to the increase in the odds of being married, there is a significant rise in the number of children on average associated with being exposed to unilateral divorce as a youth. This result is somewhat sensitive to the inclusion of trends, with the coefficient doubling to 0.024 more children if exposed to a unilateral divorce regime as a youth. The next block examines the impact of unilateral divorce on educational attainment. There is in fact a significant decline in years of education attained for those exposed to unilateral divorce as a youth, with exposure associated with 0.065 fewer years of education (0.6% of mean). This impact appears to arise mostly from a significant and sizeable increase in the odds 20 of being a high school graduate, and a corresponding large decline in the odds of being a college graduate. The next block examines the impact on living standards. There is a significant deterioration in average income per capita for those women exposed to unilateral divorce as a youth, with exposure associated with a reduction in income per capita of $446, or 3.3%. But there is no effect on the percentage of the cell living below the poverty line, suggesting that the reductions in income are concentrated in middle and higher income families. The next block shows the impact of unilateral divorce on labor supply. Being exposed to unilateral divorce as a youth significantly lowers labor supply. The odds of being employed fall by 0.5 percent, which is 0.8% of the sample mean; weeks worked fall by 0.36, which is 1.2% of the sample mean; and earnings fall by $277, which is 2.6% of the sample mean. The much larger impact on earnings than on labor supplied implies that hourly wages are falling; direct examination of hourly wages yields a negative, but insignificant, coefficient. Table 5 shows corresponding results for males. The basic pattern of results for the first three blocks is quite similar: increased marriage probabilities; reduced educational attainment; and reduced living standards. But the effects on labor supply are quite different: being exposed to unilateral divorce as a youth appears to increase labor supply for men, albeit not significantly. The second set of columns in both tables shows the results of current unilateral divorce regimes. In both cases, except for marital status, there are relatively few significant coefficients. This reflects the much more limited variation in the current unilateral regime than in the state of birth unilateral regime. 21 Interpretation This panoply of results paints an interesting picture of the impact of being exposed to unilateral divorce as a youth. Exposure to unilateral divorce leads for both men and women to more marriage, less education, and lower family incomes. For women, being exposed to unilateral divorce leads to lower labor force attachment and earnings; for men, labor force attachment and earnings actually rise, albeit insignificantly. Distinguishing causal pathways for these effects, however, is somewhat difficult, as educational attainment, marital status, and labor force attachment are all jointly determined. Since, in the raw data, there is a strong positive association between education and marriage, it seems likely that the increase in marriage and reduction in education are both direct effects of unilateral divorce, rather than one being a secondary effect of the other. The reduction in education could arise from at least two channels. The first is liquidity constraints: to the extent that children of divorce have fewer resources, they may be unable to afford higher education. The second is stress in childhood which leads to worse performance in school as a youth, with resulting ramifications for ultimate educational attainment. The fact that the largest impacts appear to be on the college attendance margin would be consistent with liquidity constraints. In addition, as I show below, at least for the cohorts of adults studied here, being a child in a unilateral divorce state leads to lower family incomes as youth, which would be consistent with the liquidity constraints explanation; but this would certainly not rule out impacts through other stress-related channels as well.2 2 A liquidity constraints explanation for reduced educational attainment would suggest that the effect would be largest where the costs of secondary education are the highest. To assess whether this is true, I obtained data on public school tuition levels across the states and estimated 22 Increases in marriage are more striking, given previous evidence that children of divorce are more likely themselves to be in unstable relationships. This may reflect the selection that haunts previous attempts to assess the long run implications of divorce. If one conditions on underlying propensities to be in stable relationships, as this exercise implicitly does, watching one’s parents divorce may increase the desire for a stable relationship of one’s own. Indeed, it is striking that while unilateral divorce leads to no more divorces, it does lead to significantly more separations, suggesting that children of unilateral divorce are more loathe to permanently sever their marital ties. I return to this set of issues below, where I find that the increased propensity towards marriage may reflect timing and not steady state marriage levels. The reduction in labor force attachment and earnings for women could arise either directly through unilateral divorce impacts, or indirectly through either of the marriage or education channels. The impact appears too large to arise solely through reduced education. A simple regression of log earnings on years of education for women in these data shows that each year increase in education leads to a 6.5% increase in earnings, consistent with estimates of the return to education from other studies. I find that unilateral divorce as a youth lowers education of women by 0.065 years, which would imply a $45 reduction in earnings, well below the $277 estimated impact. Moreover, for men I find that education is also falling, but earnings are models where I allowed the education effects to vary by underlying public tuition levels (I am grateful to David Card for providing these data). The data are far from ideal for my purposes; they only go back to 1972, whereas most of my sample graduated high school well before then. I therefore made a ranking of the states in terms of their tuition levels, and used the average ranking over the 1972-1983 period as my regressor for all cohorts, assuming that states are either consistently high or low tuition state. Doing so, I found no consistent evidence of stronger reductions in college attainment from unilateral divorce where tuition ranking was higher, which is inconsistent with a liquidity constraint explanation. But this test is not a strong one due to the limitations of the tuition data for my purposes. 23 increasing, albeit insignificantly. On the other hand, the higher odds of marriage for women can more than explain the reduction in labor force attachment that is seen here. Being married lowers odds of working by 14%; given the 0.8% marriage effect, would lower odds of working by 0.11%. The odds of working fall by 0.054%, which is smaller. Similarly, marriage lowers earnings by 70%, which can easily explain the reduction in earnings from unilateral divorce. The increase in labor force attachment and earnings for men, while not significant, is also consistent with higher incidence of marriage. Moreover, men may be earning more to offset the declining earnings of their wives. I return to this issue below, where the evidence on length of exposure suggests that a third channel outside of either marriage or education is driving labor force attachment effects. Another interpretive issue with these results is the mechanisms through which unilateral divorce regulation leads to later outcomes: is it solely through increased divorce, or through the other mechanisms discussed above? It is impossible to answer this question precisely, given that I only have one instrument and two channels of effects. But back of the envelope calculations from the estimates in Tables 3 and 4 suggest that the effects of divorce must be enormous if unilateral divorce has its impacts through the divorce channel only. For example, I find in Table 3 that the odds for a child of living with a divorced parent is rising by 0.011 percentage points. Putting this together with the findings in Table 4, and assuming that the impacts of unilateral divorce arise through increased divorce only, would imply that coming from a divorced family raises for females the odds of being married by 72 percentage points, lowers education by 5.9 years, lowers the odds of graduating college by 154 percent, lowers average incomes per capita by $40,550, and lowers earnings by $25,180. These substantial effects seem unlikely to be 24 driven solely by parental divorce. Part V: Extensions The initial results are suggestive of important long run effects of unilateral divorce on adult outcomes. In this section, I consider a number of extensions designed to assess the mechanisms through which these effects arise. Exposure to Laws One interesting question is whether the amount of exposure to unilateral divorce laws strengthens the effects that have been shown thus far. To the extent that additional years of exposure to unilateral divorce regimes raises the odds that parents divorce, it could lead to stronger impacts on the outcome of children who grow up in unilateral divorce regimes. On the other hand, if there is a pent-up stock of divorce demand that is satisfied shortly after the unilateral divorce law is passed, and then divorces decline again, there may not be a strong relationship with exposure. Wolfers (2000) suggests that such a “blip” may occur over the first eight years that a unilateral divorce law is in place. In any case, examining impacts by amount of exposure also has the potential to help determine the causal pathways of the effects that were shown above, in two senses. First, if the time pattern of impacts of later outcomes matches that of exposure to divorce, then it lends some credence to the notion that the effects are occurring through the divorce channel and not other (e.g. household bargaining) channels. Second, if the time pattern of some effects (e.g. marriage) matches that of others (e.g. income), it provides some more evidence that can be helpful in 25 interpreting causal pathways of effects of unilateral divorce. To examine exposure effects, I calculate the number of years that each adult in the sample was exposed to unilateral divorce as an adult (after age 18) and as a youth (up through age 18). I then divide these current unilateral and unilateral as youth dummies into three exposure categories: one to four years of exposure; five to eight years of exposure; and nine or more years of exposure. It is important to note that, by the nature of the construction of years exposed, the exposure effects of laws as a youth pick up two effects: amount of time exposed and age first exposed. That is, an individual who is exposed for 8 years is by definition first exposed at age 10. This makes it difficult to interpret the effects below as effects of additional exposure rather than as affects of being exposed at a younger age. Table 6 first explores the impact of additional exposure on marital status, mirroring Table 3. For presentational simplicity, I show only the results with state-specific trends. The results here are somewhat different when examined from the perspective of adults and from the perspective of their children. For adults, in the first two panels, there is a growing odds of divorce with exposure to current unilateral divorce laws. The effect is monotonic, with the odds of divorce roughly doubling at 5-8 years of exposure relative to 1-4 years, and then rising by an equal increment at 9+ years. This finding for divorce stocks appears to contradict the finding of Wolfers (2000) for divorce flows, which would suggest that additional exposure beyond eight years should not affect the number of divorces. Another interesting finding here is that there is a sizeable, although not significant, reduction in the odds of being never married for short exposures, that fades over time. This suggests that for short exposures there may be some of the “marital churning” effect noted earlier, but that this disappears over time, perhaps as persons 26 realize the problems with this approach to marital decision making. For children, on the other hand, there is a much flatter profile with respect to years exposed. As in Wolfers (2000), virtually all of the effect has occurred by eight years after the law is in place, and the vast majority within four years. I also find an interesting pattern of exposure effects for the variables measuring unilateral exposure as a youth, as shown in Table 7; the first set of columns show results for females, and the second set for males, and all results include state-specific trends. For most of the variables, the impacts are roughly constant or only slightly increasing for 1-4 and 5-8 years of exposure, and then growing substantially after eight years. The pattern of effects on years of education is completely flat, but when subcategories of education are examined, there does appear to be a larger effect for long exposures for both high school and college graduation. These results yield a somewhat mixed message. In terms of the causal impact of divorce, as opposed to other channels of unilateral divorce reform, there is much evidence to suggest that other channels are important. Despite an impact of unilateral divorce on divorce levels that is montonically growing (in the top panel of Table 6) or fairly flat after eight years (in the bottom panel of Table 6), the impacts on marriage and education only increase with eight or more years of exposure. There is a monotonic decline in earnings and average incomes, which is consistent with the top panel of Table 6, but not the bottom. In terms of causal channels among the impacts of unilateral divorce, these findings appear to offer further suggestion of a causal effect of marriage patterns on labor supply. Reductions in earnings for women, and increases for men, are occurring mostly after eight years of exposure, which is exactly when the marriage effect grows. 27 Age Effects on Marriage As shown above, exposure to unilateral divorce as a youth leads to a higher likelihood of being married. But, in these cross-sections of a given set of ages, a higher likelihood of being married on average could result from either increased odds of marriage at every age or from a shift forward in the timing of marriage. I explore this issue in Table 8. I show the results, for the marriage variables, of models that interact the current unilateral law and unilateral as youth dummies with age. I focus solely on results with state-specific trends. The first block of the table shows the results for women; the second block for men. In fact, I find that the effects of exposure to unilateral divorce as a youth on marriage appear to mostly arise through marriage timing. The age interaction with the unilateral-as-youth dummy is negative, significant, and sizeable; it indicates that the impact at age 25 is .016 percentage points, but that by age 35 the impact is actually negative. That is, unilateral divorce exposure as a youth raises the odds of being married at younger ages, but actually lowers them at older ages. This does not appear to occur through rising divorce probabilities with age, nor through rising probabilities of separation. Rather, it appears that the impact of unilateral divorce exposure as a youth on the odds of being never married is initially very negative, but rises sharply with age. The pattern of results is very similar for males. Thus, exposure to unilateral divorce as a youth leads to higher marriage rates through reducing the age at which individuals become married. This suggests that these children of easier divorce laws, perhaps craving the stable relationship that they were less likely to get than their counterparts who were not exposed to these laws, move more quickly to form their own unions once they leave the home. It is also interesting to note that, for current exposure to unilateral 28 divorce regimes, there are only very modest increases in the impact of unilateral divorce on divorce propensities with age. Type of Regulation The analysis thus far has focused on the impacts of regulations which made divorce easiest, unilateral divorce laws with no separation requirement. But, as noted earlier, a number of states have explored alternative unilateral divorce laws which included separation requirements. This is an interesting alternative because this is the direction that states are moving as they back away from unfettered unilateral divorce; for example, Louisiana’s covenant marriage law features unilateral divorce with a two year separation requirement. In this subsection, I assess the implications of this alternative relative to the more loose unilateral divorce regulations (without separation requirements) that have been the focus of the paper thus far. Tables 9 and 10 shows the results of models which include indicators for both unrestricted unilateral divorce exposure currently and as a youth (as earlier), and exposure to unilateral divorce with a separation requirement, first for females and then for males. I present results only from models with state specific trends. Focusing first on the results for the impact of current regulations on divorce, I find confirmation for Friedberg’s conclusion that unilateral regulations with separation requirements have much weaker impacts on divorce than do the looser regulations without such requirements. In particular, for women, while unilateral regulations without separation requirements raise the probability of divorce by 0.012 percentage points, regulations with separation requirements raise 29 the probability by only 0.003 percentage points. I also find in general opposite signed impacts on outcomes of exposure to unilateral regulations with separation requirements as a youth. In terms of the major results highlighted earlier, for women I find: negative and insignificant impacts on marriage decisions; positive and often significant impacts on labor supply; positive and significant impacts on family income; and positive impacts on years of education. For men, the positive results on family income and education are somewhat weaker. On the other hand, whereas before there were positive but insignificant impacts on male earnings of loose unilateral regulation, now there are negative and significant impact on male earnings of unilateral with separation requirements. Taken together, the results suggest much weaker negative long run implications for youths of exposure to unilateral regulations with separation requirements. Short vs. Long Run Effects There is an important distinction between the adults studied in this exercise and the youths who are now becoming adults: the adults studied here were virtually all born before unilateral divorce regulations were in place, so that any exposure that they had as youth came as a “surprise” to their parents. However, for subsequent cohorts, the parents knew when the marriage was established that they were living in a looser divorce regime. This can itself have important implications for the decision to bear children. Altruistic parents, understanding when they marry that they face a higher odds of divorce, may be reticent to bear children if they fear that those children will suffer adversely when they divorce. For example, altruistic mothers may only bear children in a unilateral divorce environment if they can be assured that there will be 30 adequate child support payments from their ex-husbands. If this is the case, the implications of unilateral divorce for future generations may be attenuated relative to past generations. While I cannot assess this directly with the current data on adults, I can divide the sample of children into those who were born into regimes where unilateral divorce was in place, versus those born before unilateral divorce was in place and for whom the passage of such laws was a surprise. In both cases, I include those children never exposed to unilateral divorce as controls. The results of doing so are presented in Table 11. The findings here illustrate an important difference between the effects of unilateral divorce when it is a “surprise” and when it is not. In the former case there is a large negative impact on average family incomes from being exposed to unilateral divorce. But in the latter case there is not: children who are born into regimes where unilateral divorce is in place do not have lower family incomes than children who are never exposed to such regimes. This is despite the fact that, in both cases, there are significantly higher odds of living with a divorced parent; indeed, the odds are higher when divorce law is not a surprise. This result, while only suggestive, has important implications for the long run interpretation of the findings: it implies that selective childbearing is mitigating any negative impacts of unilateral divorce on child living standards over time. Part VI: Conclusions Is making divorce easier bad for children? The results in this paper suggest that, for the first cohort of children exposed to unilateral divorce, the answer is a qualified yes. These children are more likely to be living with divorced parents and have lower family incomes as children. As adults, they are less educated and have lower family incomes. These lower family 31 incomes, however, appear to largely arise because of earlier marriage, more children, and reduced labor force attachment by women. Thus the qualification: children of unilateral divorce are living with worse living standards later in life, but largely because women are staying at home with children rather than working. Therefore, the question of whether making divorce easier is bad for children is fundamentally determined by the question of whether having mothers at home rather than in the labor force is social welfare decreasing or improving. In other words, the answer to this question likely gets pushed back yet another generation: is it beneficial or detrimental for the children of this cohort to have lower incomes but more maternal time in the home? In terms of policy, the findings here suggest that the divorce regulation can have a real impact on child living standards, in both the short and long runs. At the same time, I find that tighter unilateral regulations that include separation requirements have more muted effects, suggesting that there may be some value to a move towards the type of “covenant marriage” doctrine embraced by Louisiana. On the other hand, the last section suggests that the negative impacts of unilateral divorce may be mitigated over time. For children born after these laws were in place, there does not appear to be a detrimental impact on living standards on average, despite continued higher rates of divorce. 32 References Achenbach, T. and C. S. Edelbrock (1983). “A Manual for the Child Behavior Checklist and Revised Child Behavior Profile,” New York: Queen City Printers. Allen, Douglas (1992). “Marriage and Divorce: A Comment on Peters,” American Economic Review, 82, 677-685. Amato, P. R. and B. Keith (1991). “Parental Divorce and the Well-Being of Children: A Meta Analysis,” Psychological Bulletin, 110, 26-46. Aquilino, W. S.(1994). “Impact of Childhood Family Disruption on Young Adults’ Relationships with Parents,” Journal of Marriage and the Family, 56, 295-313. Baumrind, D. (1989). “Sex-Differentiated Socialization Effects in Childhood and Adolescence in Divorced and Intact Families,” Paper presented at the biennial meetings of the Society for Research on Child Development, Kansas City, MO. Bougheas, Spiros, and Yannis Georgellis (1999). “The Effect of Divorce Costs on Marriage Formation and Dissolution,” Journal of Population Economics, 12, 489-498. Cherlin, Andrew J., Kathleen E. Kiernan and P. Lindsay Chase-Lansdale (1995). “Parental Divorce in Childhood and Demographic Outcomes in Young Children,” Demography, 32, Issue 3. 299-315. Cooney, T. M.(1994). “Young Adults’ Relations with Parents: The Influence of Recent parental Divorce,” Journal of Marriage and the Family, 56, 45-56. Doherty, W. J. and R. H. Needle (1991). “Psychological Adjustment and Substance Abuse Among Adolescents Before and After a Parental Divorce,” Child Development, 62, 328- 337. Dormbush, S., J. Carlsmith, S. Bushwall, P. Ritter, H. Leiderman, A. Hastor and R. Gross (1985). “Single Parents Extended Households and the Control of Adolescents,” Child Development, 56, 326-341. Ferber, Marianne, and William Sander (1989). “Of Women, Men, and Divorce: Not by Economics Alone,” Review of Social Economy, 47, 15-26. Flewelling, R. L. and K. E. Bauman (1990). “Family Structure as Predictor of Initial Substance Use and Sexual Intercourse in Early Adolesence,” Journal of Marriage and the Family, 52, 171-181. 33 Friedberg, Leora (1998). “Did Unilateral Divorce Raise Divorce Rates? Evidence from Panel Data,” American Economic Review, 88, 608-627. Glenn, N. D. and K. B. Kramer (1985). “The Psychological Well-Being of Adult Children of Divorce,” Journal of Marriage and the Family, 47, 905-913. Gray, Jeffrey (1998). “Divorce Law Changes, Household Bargaining, and married Women’s Labor Supply,” American Economic Review, 88, 628-42. Gruber, Jonathan, Phillip Levine and Douglas Staiger (1998). “Abortion Legalization and Child Living Circumstances: Who is the "Marginal Child?,” Quarterly Journal of Economics, 115, 263-292 Guidubaldi, J., J. D. Perry and H. K. Cleminshaw (1984). “The Legacy of Parental Divorce: A Nationwide Study of Family Status and Selected Mediating Variables on Children’s Academic and Social Competencies,” In B. B. Lahey and A. E. Kazdin (Eds.) Advances in Clinical Child Psychology, 7, New York: Plemum Press. 109-151. Hetherington, E. M.(1972). “Effects of Father Absence on Personality Development in Adolescent Daughters,” Developmental Psychology, 7, 313-326. Hetherington, E. M. “Social Capital and Development of Youth from Nondivorced, Divorced and Remarried Families,” In A. Collins (Ed.) Relationships as Developmental Contexts: The 29th Minnesota Symposium on Child Psychology. Hillsdale, NJ: Lawrence Erlbaum. Hetherington, E. M. and W. G. Clingempeel (1989). “Coping with Marital Transitions: A Family Systems Perspective,” Monographs of the Society for Research in Child Development, 60, 1-14. Hetherington, E. M. and R. D. Parke (1986). “Clinical Psychology: A Contemporary Viewpoint,” New York: McGraw Hill. Isaacs, M. (1986). “The Visitation Schedule and Child Adjustment: A Three-Year Study,” Unpublished Manuscript. Johnson, John and Chris Mazingo (2000). “The Economic Consequences of Unilateral Divorce for Children,” mimeo, University of Illinois. Keith, V. M. and B. Finley (1988). “The Impact of Parental Divorce on Children’s Educational Attainment, Marital Timing and Likelihood of Divorce,” Journal of Marriage and the Family, 50, 797-809. Krein, S. F. and A. H. Beller (1988). “Educational Attainment of Children from Single-Parent 34 Families: Differences by Exposure, Gender and Race,” Demography, 25, 221-234. Kurdek, L. R.(1991). “Divorce History, Marital Status, Gender, and Well-Being,” Journal of Marriage and the Family, 53, 712-718. Matsueda, R. L. and K. Heimer (1987). “Race, Family Structure and Delinquency: A Test of Differential Association and Social Control Theories,” American Sociological Review, 52, 826-40. McLanahan, S. and L. Bumpass (1988). “Intergenerational Consequences of Family Disruption,” American Journal of Sociology, 94,130-152. McLanahan, S. and G. Sandefur (1994). “Growing Up with A Single-Parent,” Cambridge, MA: Harvard University Press. McLeod, J. D. (1991). “Childhood Parental Loss and Adult Depression,” Journal of Health and Social Behavior, 32, 205-220. Nordinger, Pia (1998). “The Anti-Divorce Revolution,” The Weekly Standard, March 2. Peters, H. Elizabeth (1986). “Marriage and Divorce: Informational Constraints and Private Contracting,” American Economic Review, 76, 437-54. Peters, H. Elizabeth (1992). “Marriage and Divorce: Reply,” American Economic Review, 82, 686-693. Reilly, Siobhan and Eirik Evenhouse (1997). “Divorce Laws and Divorce Rates: Evidence from Panel Data,” mimeo, Vanderbilt University. Ross, H. L. and Isabel V. Sawhill (1975). Time of Transition: The Growth of Families Headed by Women. Washington, D.C.: Urban Institute. Strauss, John and Duncan Thomas (1995). “Human Resources: Empirical Modeling of Household and Family Decisions,” In J. Berhrman and T.N. Srinivasan, eds, Handbook of Development Economics, 3A, pp1883-2023. Wallerstein, J. S. and S. Blakeslee (1989). “Second Chances: Men, Women and Children a Decade After Divorce,” New York: Ticknor and Fields. Weitzman, Lenore (1985). The Divorce Revolution. New York: Free Press. Wolfers, Justin (2000). “Did Unilateral Divorce Raise Divorce Rates? Reassessing the Evidence,” mimeo, Harvard University. 35 Wu, L. L. and B. C. Martinson (1993). “Family Structure and the Risk of Premarital Birth,” American Sociological Review, 58, 210-32. Zill, N., D. R. Morrison and M. J. Coiro (1993). “Long Term Effects Of Parental Divorce on Parent-Child Relationships, Adjustment, and Achievement in Young Adulthood,” Journal of Family Psychology, 7, 91-103. 36 Divorce Rate / 1000 Adults % Pop in States w/ Unilateral 8.72688 59.31 % Pop in States w/ Unilateral Divorce Rate / 1000 Adults 3.0415 1.36 1960 1965 1970 1975 1980 1985 1990 1996 year Figure 1: Divorce Rates and Unilateral Divorce Exposure Table 1: Divorce Regulations Across the States State No-Fault Date Unilateral with Separation Unilateral Separation Date Length Date Alabama 1971 1971 Alaska 1963 1963 Arkansas 1950 1950 3 1993 1.5 Arizona 1950 1950 5 1973 California 1970 1970 Colorado 1972 1972 Connecticut 1973 1973 District of 1966 1977 1 Columbia Delaware 1957 1974 Florida 1971 1971 Georgia 1973 1973 Hawaii 1965 1965 2 1973 Idaho 1950 1950 5 1971 Illinois 1984 1984 2 Indiana 1973 1973 Iowa 1970 1970 Kansas 1969 1969 Kentucky 1962 1962 5 1972 Louisiana 1950 1950 2 1979 1 1986 .5 Maine 1973 1973 Maryland 1957 1957 5 1973 3 1983 2 Massachusetts 1975 1975 Michigan 1972 1972 Minnesota 1950 1951 5 1974 Mississippi 1978 Missouri 1974 1974 2 Montana 1973 1973 North Carolina 1950 1950 2 1965 1 North Dakota 1971 1971 Nebraska 1972 1972 Nevada 1950 1950 3 1967 New Hampshire 1971 1971 New Jersey 1971 1971 1.5 New Mexico 1950 1950 New York 1966 1966 2 1972 1 Ohio 1974 1974 2 1982 1 Oklahoma 1953 1953 Oregon 1971 1971 Pennsylvania 1980 1980 3 1988 2 Rhode Island 1950 1950 10 1975 1972 5 South Carolina 1969 1969 3 1979 1 South Dakota 1985 1985 Tennessee 1963 1963 2 Texas 1950 1950 10 1970 1953 7 Utah 1950 1950 3 1987 Virginia 1960 1960 3 1964 2 1975 1 Vermont 1969 1969 3 1971 2 1972 .5 Washington 1950 1950 5 1973 1965 2 West Virginia 1969 1969 2 1977 1 Wisconsin 1950 1959 5 1986 1972 1 Wyoming 1977 1977 Table 2: Means Children Mother Divorced 0.0664 Mother Separated 0.0349 Mother Never Married 0.037 Father Divorced 0.0099 Father Separated 0.00334 Father Never Married 0.0036 Income Per Capita 8432.73 Number of Obs 3876 Female Adults Male Adults Family Structure Married 0.717 0.726 Divorced 0.110 0.082 Separated 0.034 0.023 Never Married 0.120 0.166 Number of Children 2.024 1.169 Educational Attainment Years of Education 11.683 11.933 HS Dropout 0.189 0.195 HS Graduate 0.365 0.308 Some College 0.216 0.216 College Graduate 0.231 0.281 Living Standards Income per Capita 13513.41 15159.06 Below Poverty 0.118 0.086 Labor Supply Work Last Year 0.712 .946 Weeks Worked 29.758 45.040 Earnings 10682.1 26222.66 Number of Observations 159884 159487 Table 3: The Impact of Unilateral Divorce on Marital Status Adults Females Males No Trend Trend No Trend Trend Divorced 1.236 1.394 .957 1.041 (.24) (.367) (.235) (.348) [.112] [.127] [.117] [.127] Separated .196 -.114 -.041 -.058 (.133) (.184) (.082) (.142) [.058] [-.034] [-.018] [-.025] Never Married -.586 -.219 -.302 .066 (.435) (.261) (.525) (.347) [-.049] [-.018] [-.018] [.004] Number of Obs 5304 5304 5304 5304 Children Living with Mother who is Living with Father who is Divorced .607 .956 .198 .14 (.167) (.241) (.041) (.058) [.091] [.144] [.200] [.141] Separated -.008 -.224 -.01 .011 (.184) (.209) (.022) (.022) [-.002] [-.064] [-.030] (.033) Never Married -.883 -.012 -.082 -.094 (.214) (.164) (.031) (.038) [-.239] [-.003] [-.228] [-.261] Number of Obs 3876 3876 3876 3876 Table 4: Unilateral Divorce and Outcomes as Adults - Females Unilateral as Youth Unilateral as Adult No Trend Trend No Trend Trend Family Structure Married .656 .785 -1.193 -.7898 (.241) (.28) (.294) (.391) [.009] [.011] [-.017] [-.011] Divorced .02 -.057 .948 1.107 (.293) (.298) (.126) (.172) [.002] [-.005] [.086] [.101] Separated .344 .323 .123 -.112 (.092) (.08) (.094) (.139) [.101] [.095] [.036] [-.033] Never Married -1.2 -1.257 .03 -.223 (.473) (.504) (.278) (.288) [-.100] [-.104] [.003] [- .019] Number of .013 .024 .052 .017 Children (.01) (.009) (.017) (.02) [.006] [.012] [-.0626] [-.008] Educational Attainment Years of -.06 -.065 .131 .023 Education (.04) (.029) (.072) (.117) [-.005] [-.006] [-.011] [-.002] HS Dropout .575 .555 2.703 .53 (.516) (.397) (.786) (.820) [.030] [.029] [.143] [.028] HS Graduate 1.37 1.402 -4.034 -1.429 (.609) (.422) (1.268) (1.682) [.038] [.038] [-.111] [-.039] Some College -.06 -.263 .72 1.93 (.295) (.211) (.424) (.774) [-.003] [-.012] [.033] [.089] College -1.886 -1.694 .611 -1.031 Graduate (.53) (.38) (1.064) (1.848) [-.082] [-.073] [.027] [-.045] Living Standards Income Per -501.64 -446.47 44.43 -176.61 Capita (151.99) (120.12) (264.74) (455.86) [-.037] [-.033] [.003] [-.013] Below Poverty .001 .098 .598 .378 (.238) (.197) (.731) (.591) [.000] [.008] [.051] [.032] Labor Supply Work Last Year -.80800 -.535 1.181 1.589 (.261) (.254) (.450) (.501) [-.011] [-.008] [.017] [.022] Weeks Worked -.46 -.36 .79 .745 (.140) (.130) (.24) (.259) [-.015] [-.012] [.027] [.025] Earnings -347.57 -276.99 79.1 -107.41 (113.69) (94.37) (193.33) (294.35) [-.033] [-.026] [.007] [-.010] Table 5: Unilateral Divorce and Outcomes as Adults - Males Unilateral as Youth Unilateral as Adult No Trend Trend No Trend Trend Family Structure Married .633 .744 -.963 -.915 (.333) (.392) (.32) (.464) [.009] [.010] [-.013] [-.013] Divorced .047 -.094 .759 .809 (.284) (.28) (.135) (.171) [.006] [-.011] [.093] [ .099] Separated .236 .252 -0.083 -.022 (.057) (.05) (.065) (.097) [.103] [.110] [-.036] [- .010] Never Married -.975 -.954 .28 .169 (.572) (.627) (.32) (.342) [-.059] [-.058] [.017] [.010] Number of 3.924 5.024 -6.762 -4.199 Children (1.22) (1.468) (1.394) (1.793) [.034] [.043] [-.058] [-.036] Educational Attainment Years of -.067 -.074 -.0753 .028 Education (.055) (.04) (.095) (.163) [-.006] [-.006] [-.006] [.002] HS Dropout .884 .852 1.574 .162 (.494) (.377) (.771) (.981) [.045] [.044] [.081) [.008] HS Graduate .889 1.017 -3.446 -1.629 (.514) (.329) (1.136) (1.63) [.029] [.033] [-.112] [-.053] Some College -.303 -.35 .560 2.125 (.208) (.166) (.328) (.578) [-.014] [-.016] [.026] [.098] College -1.47 -1.519 1.312 -.659 Graduate (.646) (.506) (-1.290) (2.388) [-.052] [-.054] [.047] [-.023] Living Standards Income Per -423.66 -292.01 76.84 17.11 Capita (154.55) (138.09) (281.46) (470.04) [-.028] [-.019] [.005] [.001] Below Poverty -.093 .043 .347 .299 (.205) (.153) (.705) (.482) -.011 [.005] [.04] [.035] Labor Supply Work Last Year .084 .173 .239 .077 (.126) (.112) (.143) (.248) [.001] .002 [.003] [.001] Weeks Worked -.015 .112 .189 .171 (.103) (.099) (.108) (.184) [0.00] [.002] [.004] [.004] Earnings 152.88 416.99 -303.63 -163.17 (248.17) (291.26) (433.32) (805.5) [.006] [.016] [-.012] [-.006] Table 6: Amount of Exposure to Unilateral Divorce Regulation and Marital Status Adults Females Males 1-4 Years 5-8 Years 9+ Years 1-4 Years 5-8 Years 9+ Years Divorced .602 1.72 2.152 .93 1.154 1.678 (.248) (.445) (.788) (.207) (.418) (.729) [.055] [.156] [.196] [.113] [.141] [.205] Separated .026 -.183 -.337 .076 -.128 -.292 (.209) (.245) (.313) (.113) (.199) (.267) [.008] [-.054] [.099] [.033] [-.056] [.127] Never -.157 -.151 .487 .336 .067 .738 Married (.279) (.366) (.604) (.252) (.471) (.648) [-.013] [-.013] [.041} [.020] [.004] [.044] Children Living with Mother who is Living with Father who is Divorced .775 1.16 1.089 .044 .192 .271 (.153) (.196) (.305) (.034) (.035) (.058) [.117] [.175] [.164] [.044] [.194] [.274] Separated -.065 -.574 -.683 .032 -.017 -.043 (.147) (.22) (.224) (.014) (.022) (.024) [-.019] [-.164] [-.196] [.096] [-.051] [-.129] Never -.169 .332 .798 -.099 -.145 -.204 Married (.139) (.204) (.259) (.035) (.403) (.09) [-.046] [.09] [.216] [-.275] [-.022] [-.57] Table 7: Amount of Exposure to Unilateral Divorce Regulation and Adult Outcomes Females Males 1-4 Years 5-8 Years 9+ Years 1-4 Years 5-8 Years 9+ Years Family Structure Married .602 .609 1.251 .586 .385 1.311 (.302) (.308) (.507) (.365) (.476) (.79) [.008] [.008] [.017] [.008] [.005] [.018] Divorced -.031 -.031 .03 -.034 0 -.185 (.237) (.322) (.526) (.248) (.282) (.487) [-.003] [-.003] [.003] [-.004] [0] [-.023] Separated .202 .316 .619 .191 .24 .384 (.07) (.093) (.147) (.046) (.063) (.116) [-.059] [.09] [.182] [.083] [.104] [.167] Never -.924 -1.137 -2.173 -.79 -.68 -1.573 Married (.428) (.535) (.998) (.567) (.655) (1.233) [.078] [-.095] [-.181] [-.048] [-.041] [-.095] Number of .011 .025 .052 .038 .045 .08 Children (.008) (.012) (.017) (.013) (.019) (.022) [.006] [.013] [.026] [.033] [.039] [.068] Educational Attainment Years of -.057 -.064 -.067 -.066 -.066 -.069 Education (.028) (.031) (.042) (.041) (.041) (.058) [-.005] [-.006] [-.006] [-.005] [-.006] [-.006] HS .496 .514 .494 .648 .787 1.137 Dropout (.359) (.432) (.642) (.335) (.4) (.691) [.026] [.027] [.026] [.033] [.040] [.058] HS .977 1.517 2.169 .84 .953 1.207 Graduate (.38) (.451) (.604) (.301) (.434) (.487) [.027] [.042] [.059] [.027] [.031] [.039] Some -.047 -.433 -.273 -.091 -.513 -.422 College (.185) (.274) (.373) (.175) (.24) (.264) [-.002] [-.020] [-.013] [-.004] [-.024] [-.02] College -1.426 -1.597 -2.39 -1.397 -1.227 -1.922 Graduate (.329) (.456) (.618) (.43) (.587) (.852) [-.062] [-.069] [-.103] [-.05] [-.044] [.068] Living Standards Income Per -263.23 -520.77 -775.93 -164.89 -262.46 -579.41 Capita (109.87) (147.06) (166.1) (127.92) (170.57) (200.58) [-.019] [-.039] [-.057] [-.011] [-.017] [-.038] Below .005 -.023 .372 -.074 .069 .271 Poverty (.217) (.25) (.323) (.149) (.223) (.248) [0] [-.002] [.032] [-.009] [.008] [.032] Labor Supply Work Last -.347 -.311 -.854 .141 .246 .142 Year (.271) (.305) (.329) (.109) (.141) (.171) [-.005] [-.004] [-.012] [.001] [.003] [.002] Weeks -.07 -.089 -.412 .054 -.052 -.075 Worked (.068) (.092) (.102) (.074) (.061) (.105) [-.002] [-.003] [-.014] [.001] [-.001] [-.002] Earnings -120.55 -276.13 -618.65 359.53 344.70 629.76 (97.84) (119.68) (119.68) (279.97) (303.86) (500.85) [-.011] [-.023] [-.058] [.014] [.013] [.024] Table 8: Age Pattern of Effects of Unilateral Divorce on Marital Status Unilateral as Youth Unilateral as Adult Unilateral Unilateral * Unilateral Unilateral * (Age - 24) (Age - 24) Females Married 1.747 -.156 -.821 -.007 (.44) (.041) (.522) (.024) [.024] [-.002] [-.011] [0] Divorced .276 -.027 .826 .026 (.331) (.022) (.23) (.016) [.025] [.002] [.075] [.002] Separated .416 -.024 -.022 -.01 (.093) (.009) (.162) (.006) [.122] [-.007] [-.006] [-.017] Never Married -2.376 .197 -.336 .024 (.686) (.054) (.451) (.032) [1.32] [.109] [-.198] [.013] Number of .056 -.003 -.041 .003 Children (.017) (.002) (.023) (.001) [.027] [-.001] [-.020] [.001] Males Married 2.035 -.194 -1.127 .008 (.597) (.052) (.644) (.032) [.028] [-.003] [.016] [0] Divorced .272 -.019 .383 .04 (.32) (.027) (.177) (.012) [.033] [.002] [.047] [.005] Separated .274 -.014 .088 -.012 (.076) (.007) (.105) (.004) [.119] [-.006] [.038] [.005] Never Married -2.572 .224 .626 -.029 (.838) (.063) (.6) (.041) [-.155] [.013] [.038] [-.002] Number of .071 -.005 -.028 -.002 Children (.019) (.002) (.022) (.002) [.060] [-.004] [-.024] [-.001] Table 9: Unilateral Divorce vs. Separation Requirement - Females Unilateral as Youth Unilateral as Adult Unilateral Unilateral with Unilateral Unilateral with Separation Separation Family Structure Married .655 -.462 -1.047 -.603 (.276) (.354) (.419) (.348) [.009] [-.006] [-.015] [-.008] Divorced -.133 -.281 1.24 .312 (.278) (.204) (.177) (.149) [-.012] [-.026] [.113] [.028] Separated .316 -.025 -.074 .091 (.078) (.082) (.137) (.117) [.093] [-.007] [-.022] [.027] Never Married -1.035 .8 -.169 .123 (.459) (.497) (.316) (.276) [-.086] [.067] [-.014] [.010] Number of .024 0 -.027 -.025 Children (.009) (.014) (.021) (.016) [.012] 0 [-.014] [-.012] Educational Attainment Years of -.048 .062 -.015 .017 Education (.029) (.025) (.13) (.121) [-.004] [.005] [-.001] [.001] HS Dropout .376 -.655 .763 .548 (.39) (.513) (.888) (.831) [.020] [-.035] [.040] [.029] HS Graduate 1.415 .068 -1.973 -1.273 (.384) (.303) (1.91) (1.689) [.047] [.002] [-.065] [-.042] Some College -.259 .001 2.287 .838 (.204) (.2) (.798) (.678) [-.012] 0 [.106] [.038] College -1.533 .586 -1.078 -.112 Graduate (.36) (.34) (2.04) (1.813) [-.066] [.025] [-.047] [-.005] Living Standards Income Per -381.47 245.04 -447.28 -634.28 Capita (115.05) (103.18) (459.78) (417.26) [-.028] [.018] [-.033] [-.047] Below Poverty -.053 -.539 .202 -.408 (.194) (.173) (.577) (.497) [-.004] [-.046] [.017] [-.035] Labor Supply Work Last Year -.384 .544 1.734 .339 (.242) (.402) (.585) (.534) [-.005] [.008] [.024] [.005] Weeks Worked -.074 .323 .284 .243 (.067) (.089) (.181) (.144) [-.002] [.011] [.010] [.008] Earnings -215.67 225.82 -230.7 -290.86 (91.26) (98.75) (292.07) (258.95) [-.020] [.021] [-.022] [-.027] Table 10: Unilateral Divorce vs. Separation Requirement - Males Unilateral as Youth Unilateral as Adult Unilateral Unilateral with Unilateral Unilateral with Separation Separation Family Structure Married .526 -.722 -1.303 -.893 (.369) (.466) (.481) (.363) [.007] [-.010] [-.018] [-.012] Divorced -.138 -.162 .873 .149 (.266) (.168) (.172) (.14) [-.017] [-.020] [.106] [.018] Separated .228 -.087 .034 .13 (.048) (.058) (.089) (.087) [.099] [-.038] [.015] [.057] Never Married -.669 1.018 .45 .644 (.582) (.59) (.372) (.289) [-.040] [.061] [.027] [.039] Number of .038 -.044 -.051 -.021 Children (.013) (.020) (.020) (.015) [.033] [-.037] [-.044] [-.018] Educational Attainment Years of -.054 .072 .054 .058 Education (.040) (.038) (.185) (.172) [-.005] [.006] [.004] [.005] HS Dropout .576 -.992 .144 -.037 (.382) (.486) (1.076) (.98) [.030] [-.051] [.007] [-.002] HS Graduate 1.159 .521 -1.92 -.678 (.31) (.621) (1.868) (1.696) [.038] [.017] [-.062] [-.022] Some College -.347 .007 2.232 .247 (.16) (.2) (.589) (.378) [-.016] [0] [.103] [.011] College -1.388 .465 -.456 .468 Graduate (.488) (.348) (2.688) (2.453) [-.049] [.017] [.016] [.017] Living Standards Income Per -251.16 155.76 -255.53 -631.16 Capita (132.38) (110.12) (481.92) (444.87) [-.017] [.010] [-.017] [-.042] Below Poverty -.022 -.235 .245 -.124 (.156) (.151) (.477) (.445) [-.003] [-.027] [.028] [-.014] Labor Supply Work Last Year .191 .072 -.054 -.306 (.119) (.196) (.256) (.236) [.002] [.001] [-.001] [-.003] Weeks Worked .015 .056 .029 -.206 (.060) (.044) (.111) (.113) [0] [.001] [.001] [-.005] Earnings 276.48 -488.88 -708.61 -1262.91 (288.54) (310.14) (852.63) (794.03) [.011] [-.019] [-.027] [.048] Table 11 - Short vs. Long Run Effects Children For Whom Law Children Born After Law Was a “Surprise” vs. Was in Place vs. Children Never Exposed Children Never Exposed Living With Divorced Mom .775 1.161 (.183) (.258) [.123] [.179] Living with Divorced Dad .284 .04 (.07) (.067) [.316] [.04] Average Income -731.0 175.77 (292.36) (427.26) [-.021] [.088] Number of Obs 2344 2840