The Effects of Union Dissolution on the Economic Resources of Men and Women: A Comparative Analysis of Germany and the United States,1985

Abstract

We analyze the consequences of union dissolution on the economic resources of men and women in the United States and Germany over three decades, using the Panel Study of Income Dynamics (PSID) and another international survey, the German Socio-Economic Panel, which was modeled after the PSID. Measured either by family size–adjusted “pregovernment” or “postgovernment” incomes (incomes to men and women either before or after taxes and cash transfers from social welfare programs), women fare worse than men in both countries in each of the two time periods we study (1985–1993 and 2005–2013). This is primarily due to reductions in access to their partner’s earnings, even though those reductions are somewhat mitigated by increases in their own earnings and by reductions in taxes and family size. German women experienced larger proportionate reductions in postgovernment resources than American women in the first period examined five years after a family split. American women did worse in the later period, experiencing statistically significant declines in their postgovernment incomes relative to the initial period.

Keywords

union dissolution separation divorce PSID SOEP

Increased rates of separation and divorce in the United States have led researchers to consider the differential consequences of union dissolutions on the economic resources of male and female partners and on the children of their unions (Burkhauser and Duncan 1989; Cherlin, Kiernan, and Chase-Lansdale 1995; Corcoran, Duncan, and Hill 1984; Duncan and Hoffman 1985; Hoffman and Duncan 1988; Smock 1993; Tach and Eads 2015). In contrast to other countries, the United States provides less income protection against many adverse life events, and this additionally prompted cross-national comparisons of the economic consequences of union dissolution. The earliest international analysis of the consequences of family dissolution (among both married and cohabiting couples) in the United States and another country compared Americans’ experiences to those of Germans (Burkhauser et al. 1990, 1991). Here, we revisit this topic of family dissolution in the United States and Germany by examining changing impacts of splits that occurred in two intervals marking the beginning and end of a period spanning three decades: 1985 to 1993 and 2005 to 2013.¹

The original analyses of subsequent economic outcomes of split families in the United States and Germany following dissolution (Burkhauser et al. 1990, 1991) in the 1980s drew data from the Panel Study of Income Dynamics (PSID) for the United States and the Socio-Economic Panel (SOEP) for Germany. The SOEP was the first panel dataset outside the United States modeled in a structure similar to the PSID, and it had then recently become available. The analyses focused on changes from one year before dissolution to one year after to assess the extent of change in aggregate monetary resources and their constituent components.

As is now the case, the social welfare system in Germany provided a more extensive system of support than did the social welfare system in the United States. The analyses of Burkhauser et al. (1990, 1991) first examined women’s “pregovernment” incomes (pretax income, adjusted by family size, and measured prior to any government transfers that they may have received) and found that resources fell sharply in the year following divorce in both countries—by 37 percent in the United States and 44 percent in Germany. Important factors in driving these changes for women were the loss of the former spouse’s earnings in both countries and the smaller drop in resources in the United States due to more women entering work.

“Postgovernment” income (posttax income adjusted by family size and measured after the receipt of any government transfers) for German women also fell by 44 percent but only by 24 percent in the United States. Although they lived in a country with a more generous social welfare system, German women fared no better than American women after a marital or cohabiting relationship dissolved. In contrast, men’s access to money did not vary greatly from the year before to the year after in either country based on pregovernment or postgovernment income measures.

The publication of these two papers (Burkhauser et al. 1990, 1991) showed the value of the use of multiple international panel datasets in comparative social science research and formed the basis of a grant from the National Institutes on Aging (Burkhauser and Smeeding 1990) that supported creation of a harmonized collection of international panel datasets, the Cross-National Equivalent File (CNEF). In this analysis, we make use of the CNEF data and other underlying variables from the PSID and SOEP.

Social, economic, and policy changes have occurred in both countries since then. U.S. women on average now have substantially greater levels of educational attainment, and average household size has fallen. These changes suggest women’s earnings should now be higher and help to ensure continued access to monetary resources following a family split, but women may be less able to offset losses by increasing work effort. Higher earnings prior to dissolution and a declining response in terms of more work afterward across successive generations of women in the United States have been confirmed in Couch et al. (2013) and Tamborini, Couch, and Reznik (2015).

Widening wage inequality since the 1980s that has eroded the relative pay of lower-skill males in particular makes them less attractive partners for marriage and reduces the costs to women of separating from a partner. In like manner, the labor force participation of prime-age men (ages 25–54) has substantially fallen over time while it has increased for prime-age women, particularly those with children. From an economic perspective, these factors contributed to compositional changes in those who marry and divorce.

In terms of U.S. policy, the Family Support Act of 1988 requires states to establish child support rules for absent fathers bringing additional resources into families headed by women. However, the shift from entitlement-based cash welfare for low-income families under Aid to Families with Dependent Children (AFDC) to the more limited support of the Temporary Assistance to Needy Families (TANF) program under the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996 may result in less resources for families with absent fathers.

Germany has also experienced broad social, economic, and policy changes over time. Undeniably, the greatest change was German reunification on October 3, 1990. The former Federal Republic of Germany (FRG) and German Democratic Republic (DDR) were joined, and the DDR’s socialist system was transformed to the social market economy of the FRG² following West German laws and institutions (Hauser et al. 1996). Reunification altered various aspects of income inequality and economic mobility in Germany relative to the United States (Bayaz-Ozturk, Burkhauser, and Couch 2013). After years of adjustment, differences between the West and East German societies diminished considerably.

In Germany, alimony and child support is paid by the divorcee with the higher income; however, due to an alimony reform in 2008, it is now expected that alimony will end three years after the divorce unless a former spouse cares for young children or is incapacitated. Bredtmann and Vonnahme (2017) show that, for the most part, this reform did not increase the likelihood that women would enter the labor force to self-insure against a possible future family split, but it did increase the likelihood of divorce, presumably because of the decreased cost of divorce that fell to men.

In 2005, the duration of unemployment benefits was shortened, guaranteed assistance payments were reduced, and a means test was applied (Bröckel and Andreß 2015). This may have reduced postgovernment assistance to low-income families. A legal right to a place for children in kindergarten and day nurseries was also introduced, which should have increased women’s labor force participation.

The divorce rate in Germany is lower than in the United States. In 1990, about 20 couples per 1,000 inhabitants were divorced.³ By 2014, this rate had increased slightly to 21 due to both declining rates of marriage and an increased rate of divorce (see Table 3 in Martin and Kats 2003). The comparable U.S. rate in 2014 was 32 per 1,000 inhabitants based on CDC figures.⁴ The labor force participation rate of women has increased in Germany and is now only a few percentage points lower than men’s (see Table 1 in Bröckel and Andreß 2015); however, women more often work part time, and their hourly wages are on average still about a fifth lower than men’s.

Beyond reunification and the noted changes in social policy, the social security system of Germany for all mandatorily insured persons (blue- and white-collar workers and farmers) remains more generous than its U.S. counterpart and comprises transfer payments in case of sickness, old age, incapacitation, unemployment, and need of nursing home care. Additionally, there are benefits for all children depending on age and number of children in the family (Bröckel and Andreß 2015; Hummelsheim 2009). A system of tax financed and means-tested benefits guarantees a minimum income that supplements the contribution based social security system. Other programs provide means-tested housing allowances and educational grants.

Changes such as these in the United States and Germany since the 1980s may have altered key factors that led to the differential experiences of women and men in both countries following union dissolutions reported in prior studies (Burkhauser et al. 1990, 1991). To examine changes over time in the consequences of family dissolutions, here, we also draw data from the PSID for the United States and from the SOEP for Germany for the periods 1985 to 1993 and 2005 to 2013, making use of comparable cross-national survey measures available in the CNEF and other variables drawn directly from the two surveys.⁵ These time periods mark the beginning and end of a period spanning three decades using the most recently available data.

We examine individuals observed in married or cohabiting unions in the first year of these two intervals. We contrast those who subsequently experience a union dissolution by separation or divorce with those who do not. Extending much of the prior literature, we use the panel data to follow individuals in each relationship for as many as three years prior to dissolution and as many as five years afterward using unbalanced panels. This allows the observation of possible anticipatory responses to dissolutions as well as longer-term impacts on split families. By examining time periods that span three decades in both countries, we can assess how circumstances of women’s and men’s families following separation or divorce change over time.

We proceed by describing the broader literature from which our study springs, the samples we draw from the CNEF and the underlying PSID and SOEP surveys, and the panel data fixed-effects models we use. We then describe the patterns we find in our samples of married and cohabiting families whose relationships dissolve and follow this with our formal estimates of changes in different measures of aggregate well-being in response to family dissolution. We close with a discussion of our findings and their implications for social policy.

Literature Review

The early literature on changes in divorce rates in the United States was based on cross-sectional vital statistics data (e.g., Michael 1978). The focus of that research was on documenting rates of divorce from marriage and changes across cohorts, but the analyses were limited by the cross-sectional nature of the data. This literature was transformed by the advent of the PSID and other panel datasets modeled after it like the SOEP.

Panel surveys that sample families and the people living in them follow the individuals over time and repeatedly collect information on those who are in their families at the time of each survey along with important changes in their lives (Schonlau, Watson, and Kroh 2011). The PSID, together with the development of life course analysis (which required such panel data), provided the basis for the first systematic evidence of the consequences of a union dissolution in the United States on both the subsequent economic well-being of the partners and the children of their union (Corcoran, Duncan, and Hill 1984; Duncan and Hoffman 1985; Hoffman and Duncan 1988). Some key findings from this early literature were that changes in family structure are common, that losses in income are disproportionately larger for female- than for male-headed families following a relationship dissolution, and that the loss of their male partners’ earnings played a key role in differential outcomes for women’s families.

More importantly, the PSID served as a model in the creation of another panel dataset in the 1980s, the SOEP in West Germany. This milestone in international data development made it possible for important life course events to be analyzed from a cross-national perspective. The SOEP data were made available to U.S. researchers in English translation but also provided the original basis in combination for development of the CNEF, a collection of data from international surveys that includes common variables found across them recoded (or harmonized) to make them as consistent as possible. That collection now includes nationally representative panel data from eight countries, including Australia, Canada, Korea, Germany, Great Britain, Russia, Switzerland, and the United States.

This model of replication of the collection of U.S. panel data by other countries and harmonization of measures either through deliberate coordination of surveys or recoding of variables following collection has also been followed for panel data on aging, beginning with the Health and Retirement Study (HRS) for the United States, which was followed by the Survey of Health, Ageing and Retirement in Europe (SHARE), which covers twenty-eight countries. Similarly, the European Union organized the fielding of panel surveys across fourteen member countries in the European Community Household Panel (ECHP) from 1994 to 2001.⁶ The initiation of the PSID followed by the German SOEP began this international movement toward the collection and harmonization of social science data across countries.

The SOEP was initially fielded in 1984, and its first use in comparative international social science research was in studies of family (married or cohabiting) dissolution (Burkhauser et al. 1990, 1991). As previously described, the interest of that research was in comparing the extent of losses of economic resources among the dissolved families of women relative to men in both countries given the more extensive income protections available for common life events in European countries. These studies, like the prior U.S. literature, also reported that women’s families experienced much larger decreases in pregovernment economic resources than men’s following a union dissolution and that the loss of the partner’s labor earnings was the principal driver of this pattern. Despite the greater income supports in response to adverse life events in European countries, proportional declines in postgovernment resources were also found to be larger for German women’s families than for those of American women.

Our research revisits the topic of family dissolution using the CNEF along with other variables taken directly from the PSID and SOEP but will also make comparisons of experiences across cohorts. Smock (1993) examines the experiences of families of divorcing men and women across two generations in the United States using National Longitudinal Survey of Youth (NLSY) and NLSY79 data. The NLSY data capture divorces in the late 1960s through early 1970s, while the NLSY79 measures them in the late 1970s and 1980s. Making comparisons from the year before to the year afterward, she finds that “the economic costs of marital disruption for young women are as severe today as in the 1960s and 1970s. … Men … undergo only modest declines in family income” (pp. 366–67). Thus, Smock reports little change among U.S. families experiencing divorce from the 1970s to the 1980s.

In a more recent study for the United States, Tach and Eads (2015) make use of multiple panels of SIPP data spanning the time period from 1984 through 2007 to examine the experiences of families where children reside with their mothers contrasting the year before a relationship dissolves with the year afterward. With respect to divorce, they report that “the economic consequences of divorce have declined since the 1980s owing to the growth in married women’s earnings and their receipt of child support and income from personal networks” (p. 401). However, they report a divergent trend for cohabiting women whose relationships dissolve, concluding that “in contrast, the economic consequences of cohabitation dissolution were modest in the 1980s but have worsened over time” (p. 401).

Here, we make use of the PSID survey to examine the experiences of dissolving families in the United States. In comparison to the NLSY data, the PSID provides the advantage of information on people of all ages and multiple cohorts of individuals, whereas the NLSY follows one cohort of youths through their midlife years. Relative to each fielding of the SIPP that typically surveys a short time span covering a three- or four-year period, the use of the PSID, which now extends for five decades, provides a longer time frame for observing possible relationship changes and their consequences over many years.

Several studies have also examined union dissolution from a longitudinal perspective in Germany. For the period from 1984 to 1999, Andreß and Bröckel (2007) found that gender inequality in experiences of married men and women who separate persisted in Germany. In short-run examinations of families from two years before to a year after separation, women, on average, end up with much lower disposable (posttax, posttransfer) family incomes than during marriage. Income from labor and public transfers accounts for the greater part of equivalized net income of both partners while private transfers (alimony and child support) play only a minor role. In a follow-up study, Bröckel and Andreß (2015) show that despite institutional changes and an increase in female labor force participation, the situation has not changed much since the turn of the century. Estimated probabilities of income loss following separation among married couples remained about the same for women and decreased slightly for men between the two periods 1984 to 1999 and 2000 to 2012 (Bröckel and Andreß 2015, Fig. 2).

Other studies have made longitudinal comparisons of union dissolutions in Germany to other nations using panel data (Andreß et al. 2009; de Vaus et al. 2017). Most related to this study is the analysis of de Vaus et al. (2017), which considers six OECD countries, making use of SOEP and PSID data to compare married and cohabiting couples that dissolve with those that do not during a period since the turn of the century. The authors use a regression model to predict postgovernment equivalized incomes for those who separate in the counterfactual state of remaining together. For the United States, they use data from 1999 to 2009 and from 2000 to 2011 in Germany. They find that for men in Germany, their postgovernment income increases temporarily around the time of the separation but does not change appreciably afterward. For the United States, the predicted income gap between separating and continuously married men widens throughout the period afterward. Among women in Germany, incomes drop sharply at the time of separation, and losses remain six years afterward; but for U.S. women, there is a sharp drop in income at the time of the family split that widens to roughly double the magnitude observed in Germany six years later. The findings of the study suggest that both U.S. men and women have fared more poorly than Germans following relationship dissolutions since the turn of the century.

The prior literature has disproportionately focused on narrow time intervals such as the year preceding the dissolution of a relationship and the year afterward. Exceptions mentioned here include the work of Duncan and Hoffman (1985) and de Vaus et al. (2017); however, it is reasonable to expect that the duration of adverse economic conditions experienced by children when adults split will matter for their future paths in life. Here, we extend much of the prior literature by looking as much as three years before and five years beyond the time of a relationship dissolution by making use of the available panel data in the PSID and SOEP.

Also, the prior literature typically has not attempted to control for possible biases in attributing changes in outcomes to the dissolution that might be related to experiences of all families at that time or to particular types of people being more likely to separate. Here, we make use of comparison groups of intact families in panel fixed-effect models to control for both observed and unobserved influences on relationship dissolution. This approach has recently been used to examine labor market responses and long-term health impacts following relationship dissolutions (Couch et al. 2013; Couch, Tamborini, and Reznik 2015; Tamborini, Couch, and Reznik 2015; Tamborini, Reznik, and Couch 2016).

Additionally, we focus on measuring changes in experiences of those in a relationship that dissolves from the 1980s into the new millennium, providing a comparative analysis that spans three decades. Other work that has considered changes over time in experiences of those whose relationships end include Smock (1993), Tach and Eads (2015), and Bröckel and Andreß (2015).

Data

The CNEF and the PSID and SOEP surveys

In the analysis, we make use of data drawn from the CNEF for the PSID for the United States and the SOEP for Germany but also take underlying variables as necessary from the primary surveys to supplement the analysis for both countries. Both the PSID and SOEP are national panel surveys that began with representative samples of households and the individuals in them. The PSID began in 1968 with a sample of 4,802 households and 18,231 people. The SOEP began in 1984, surveying 12,245 people in 5,921 households (Wagner, Burkhauser, and Behringer 1993, 430). The SOEP was originally representative of the western states of Germany, and we make use of data from those geographic areas over time here for consistency but would note that the data were extended to also be representative of eastern states of Germany in June 1990. This entailed the addition of another 2,179 households and 4,453 adult respondents. The use of the data from the western states of Germany means that information on those who move from there would no longer be used as would have been the case prior to the addition of the additional states of unified Germany in 1990.

For both the United States and Germany, we consider dissolutions occurring in two periods of equal length, from 1985 to 1993 and 2005 to 2013.⁷ For each period, we consider individuals observed married or cohabiting in the initial year (1985 or 2005) and whose family splits by the last year of the period (1993 or 2013). We separate the sample into those who remain married or partnered throughout the particular time period and alternatively those who end a marital or cohabiting partnership (dissolution). Relationships that remain intact and those that dissolve are primarily determined by examining relationship status variables for each country drawn from the CNEF and comparing the initial status in each period to reports in subsequent years.

For the first conditioning period (1985–1993), we require individuals in both countries to be married or with a partner present in calendar year 1985 and the first year in which a family dissolution can be observed is 1986. We follow individuals as many years as necessary to obtain five years of postdissolution data on them. So for those whose families dissolve in 1993, we follow them for five subsequent years through 1998. For those who split in 1986, we make use of two years of prior outcome information; and similarly, for those families that dissolve in 1987 and beyond (through 1993), we have three prior and five years of subsequent information relative to timing of the family split.

To use the more recent data in the second conditioning period (2005–2013) for Germany, a similar strategy is employed. To enter the sample, persons have to be married in 2005 and the first year of recorded dissolutions is 2006. For those whose families split in 2006, we make use of two years of prior information. For dissolutions occurring in 2007 to 2013, we have three prior years of information. In the second period, however, we cannot follow everyone for five years. For example, for those whose family dissolves in 2012, one year of follow-up information is available, and similarly, two years of information is available for those who split in 2011. Thus, in the second period, the sample is unbalanced in both the years prior to a family dissolution and afterward.

For the United States, the PSID began biennial surveys in 1997. As family splits can occur either in the year of a survey or intervening years, to use the data in the second period (2005–2013) for the United States on an annual basis, it is necessary to determine the year of the dissolution and its relationship to the timing of the earnings and income variables in the year of the survey. In doing this, for those families that dissolve between biennial surveys, the next survey contains information on outcome variables the year of the split. For those whose families dissolve the year of the survey, information on outcome variables collected that year refer to the year prior to the split. So, in alternating years, the individuals whose information is represented in the data are different. Beyond this, the unbalanced pattern of information before and after a family split also occurs as it does in Germany. The appendix contains a description of how the timing of the data in the second period for the United States is aligned (the appendix can be found with the online version of the article). The appendix also contains tables that show the pattern of sample sizes across periods relative to divorce in each sample.

Throughout the analysis, we contrast the experiences of families that remain intact versus those that split within and across time periods. To align the intact families with those that experience dissolution temporally, we randomly assign them an artificial year of dissolution in proportions equal to the odds of a family split in that year among those who experience one. Those who experience other relationship changes such as becoming a widow are not included in the analysis. We have conducted estimates using alternating years of data across both time periods for Germany and the United States and obtained qualitative results similar to those presented here.

Relevant to this analysis, both the PSID and SOEP have survey rules regarding following individuals and the members of their households over time. From the outset, both followed adults in the original sample and their children by birth or adoption. However, the PSID has a relatively low rate of attrition in comparison to other panel surveys and a relatively high rate of formation of new households as children of original sample members become respondents (Schonlau, Watson, and Kroh 2011). Due to these factors, the sample size of the PSID expands over time, while the SOEP would have contracted had a somewhat broader set of following rules not been adopted in wave 9 (fielded in 1992). Both intact and dissolving families are considered here, so the most important consideration was an extension of the SOEP following rule to include cohabiting partners of sample members. An analysis of the impact of the adoption of this rule (Schonlau, Watson, and Kroh 2011) shows that this modification accounts for about 10 percent of the households and sample size of the SOEP about 25 years after the survey began.

The most direct implication of this is that the SOEP would follow both partners as they leave a dissolving family in the second period (2005–2013) examined in the SOEP if one of them was not previously a sample member. While both surveys provide weights used throughout this analysis to create nationally representative samples, and the overall proportion in the sample is relatively small, it is still possible that this difference in following rules may impact the analysis.

While most of the information used in the analysis is drawn from the CNEF, in both the PSID and SOEP, reports of the amounts of alimony and child support paid are taken from the surveys and used to adjust income concepts as explained in the online appendix. Also, due to the biennial surveys of the PSID in the second period, information on changes in marital status, whether a person moves in or out, and the year moved in or moved out are also used to make determinations regarding relationship status in individual calendar years and to construct a measure of family size.

Outcome measures

The primary measures of aggregate economic resources examined here are pregovernment and postgovernment income drawn from the CNEF but adjusted for alimony and child support payments. Pregovernment (pretax, pretransfer) family income consists of the sum of total family income from labor earnings, asset flows, private transfers, and private pensions. Postgovernment family income also includes public transfers and subtracts taxes and social security contributions. The online appendix provides complete definitions of the components of these two measures. Monetary measures are converted to real 2012 figures.

The aggregate resource measures are adjusted (equivalized) to reflect differential access based on family size. We employ a commonly used square root adjustment: for both pregovernment and postgovernment family income, we divide by the square root of the family size.⁸ This means that no adjustment for family size is made when only one person is present and incomes of larger families are adjusted for economies of scale. This adjustment is common in the literature examining economic impacts of divorce (Bröckel and Andreß 2015; Tach and Eads 2015). Prior research has shown that the choice of how to adjust for family size typically does not affect qualitative research findings (Burkhauser, Smeeding, and Merz 1996) for outcomes like income in comparisons of the United States and Germany.

To better understand how individual components of income influence changes in economic well-being following a family split, we also examine changes in broad categories of them. We consider the individual’s earnings and those of their partner. We also examine private and public transfers as well as investment income and other residual sources.

Finally, in both the PSID and SOEP survey, individuals are reporting their income from the previous year. This means the analysis covers the income years 1984 to 2012. Here, we adjust the alignment of data to capture relevant variables for individuals who reside together. In the second period of our PSID analysis that relies on biennial surveys, we examine the year of divorce of the individual and assign the individual to the relevant year as described in the online appendix and the text above. Thus, divorces are observed in all calendar years, but incomes are observed every other year in the second period (2005–2013) in the U.S. data.

Methods

In our analyses, we present descriptive tabulations of the primary outcome variables and constituent components of income over time. We then adopt a panel fixed-effects model to examine changes in pregovernment and postgovernment family income and test for changes in patterns across time periods in the experiences of men and women following divorce. All estimates provided in the analysis make use of survey sample weights.

For the panel fixed-effect analysis, we use a formulation that allows a measure of economic resources, Y_it (pregovernment or postgovernment size-adjusted income), to evolve over calendar time, indexed as t, for individuals denoted by i. We control for a number of time-varying demographic characteristics, including age, age-squared, the number of children present in the household, and years of educational attainment and label the matrix of these demographic covariates as X_it. We also include a set of categorical variables that control for the survey year from which the data are taken that we label as T_it. For the estimation equation written as

Y_{i t} = α + β X_{i t} + \sum_{k = - 3}^{5} γ_{t} d_{i t}^{k} + δ T_{i t} + φ_{i} + ε_{i t},

(1)

the parameters ϕ_i represent individual fixed-effects that are time-invariant. ε_it is an independent and identically distributed (i.i.d.) error term.

The other key variables in this formulation are the categorical indicators for the timing of each record of data relative to dissolution, $d_{i t}^{k}$ . At each point in calendar time, t, if a family splits, $d_{i t}^{k}$ = 1 if the index k denotes the person’s position relative to the year of the split for that record of data and otherwise takes the value 0. If s is the year in which a dissolution occurs, then k = t – s. For example, three years prior to the year of a split, k = −3. If a record of data represents three years before dissolution, then $d_{i t}^{- 3}$ = 1 and otherwise 0. γ is the vector of parameters that captures the effect of a family split on the outcome examined relative to an intact family. δ is a vector of parameters associated with the matrix of categorical time variables T_it. Couch et al. (2013); Couch, Tamborini, and Reznik (2015); and Tamborini, Couch, and Reznik (2015) use this formulation to assess outcomes related to family dissolution.

The advantage of this estimation approach relative to using comparisons of well-being before and after divorce is that even if families do not dissolve, they may experience changes in their well-being. In cases where intact families have increases in income, before-and-after comparisons for families that split will understate losses relative to being in a relationship. If intact families experience declining earnings, before-and-after comparisons of only splitting families will overstate losses.

We estimate equation (1) for both time periods (1985–1993 and 2003–2013), combining information on the intact and dissolving families in each period. We estimate these equations separately for men and women. To perform tests of different experiences following relationship dissolution across periods, we pool the data and interact a dummy for the different periods with all covariates examining statistical significance of the interacted parameter estimates for divorce timing, $d_{i t}^{k}$ .

Descriptive Statistics

Panels A and B of Table 1 provide average family-size adjusted pregovernment and postgovernment incomes in their top two rows along with major constituent components (not adjusted) for women’s families that experience a dissolution for the two periods of the sample, 1985–1993 and 2005–2013, along with average family size. We focus the discussion here on women’s families as the analysis shows that the largest changes over time have come for U.S. women’s families. Appendix Table A1, Panels A and B, contains the same information for families of men that experience splits. Appendix Table A2, Panels A through D, contains similar tables for our sample of intact families of men and women.

Table 1

Resources of Women’s Families before and after Dissolution

	Panel A: Resources of Women’s Families before and after Dissolution, 1985–1993
	−3	−2	−1	0	1	2	3	4	5
United States
Pregovernment income	41,806	41,001	36,994	21,958	36,165	34,292	37,229	38,641	37,334
Postgovernment income	32,148	31,972	29,462	19,490	30,006	28,328	30,967	31,427	31,299
Old/new partner’s earnings	50,210	47,928	35,878	3,625	12,987	14,486	17,279	18,070	18,617
Own earnings	20,973	22,040	24,624	27,092	30,142	29,890	31,833	34,493	33,549
Private transfers	623	644	649	2,828	4,522	4,251	4,361	4,600	3,374
Public transfers	1,086	1,017	1,192	1,692	1,443	1,509	1,290	1,742	3,226
Income from capital	3,011	2,088	2,732	1,998	997	1,112	924	1,703	1,365
All other income	3,029	3,105	4,670	2,531	3,303	2,478	2,604	2,110	2,249
Total household income	77,871	75,896	69,771	38,016	51,269	51,996	56,105	58,875	57,615
Income and payroll taxes	19,708	19,036	14,708	6,827	10,219	10,698	11,454	13,427	12,629
Household size	3.71	3.63	3.63	3.33	2.55	2.70	2.72	2.76	2.74
West Germany
Pregovernment income	36,727	35,649	32,926	23,545	26,355	27,704	28,268	27,869	25,027
Postgovernment income	26,726	26,363	24,816	18,521	21,159	21,732	22,043	22,693	21,527
Old/new partner’s earnings	47,368	43,677	38,023	19,123	19,406	20,758	19,239	17,767	15,753
Own earnings	11,397	13,197	14,984	14,942	14,710	16,360	17,339	16,920	15,538
Private transfers	—	4	144	1,016	1,335	867	888	1,781	1,495
Public transfers	2,204	2,078	1,867	1,969	2,360	2,496	2,176	2,663	2,916
Income from capital	1,160	1,355	1,555	1,151	492	1,291	1,272	1,324	1,599
All other income	7,965	8,293	7,310	5,022	6,623	6,516	7,472	6,542	6,740
Total household income	67,868	66,515	61,085	41,244	42,956	45,124	45,726	44,887	40,557
Income and payroll taxes	21,187	19,926	15,555	10,865	12,020	13,007	13,694	13,161	11,921
Household size	3.71	3.73	3.74	3.27	2.81	2.87	2.86	2.87	2.79
	Panel B: Resources of Women’s Families before and after Dissolution, 2005–2013
	−3	−2	−1	0	1	2	3	4	5
United States
Adjusted pregovernment income	51,004	57,498	37,963	27,057	30,826	38,896	34,055	39,834	36,772
Adjusted postgovernment income	39,829	44,998	29,590	22,736	26,565	33,526	30,566	34,840	33,177
Old/new partner’s earnings	45,708	53,421	26,340	1,306	8,531	5,796	13,323	4,798	17,707
Own earnings	41,230	34,201	31,473	43,635	33,159	38,204	32,012	44,647	34,318
Private transfers	1,366	1,084	800	1,614	2,159	2,818	2,446	3,877	1,196
Public transfers	486	735	461	621	1,119	1,756	1,501	1,882	993
Income from capital	212	9,885	2,442	144	32	2,045	1,012	1,665	989
All other income	1,972	2,141	3,317	1,512	3,380	834	1,320	1,137	1,199
Total household income	89,092	100,477	63,673	47,585	47,546	50,467	50,768	56,904	59,411
Income and payroll taxes	20,293	23,097	13,437	7,816	7,729	7,781	7,849	9,025	10,596
Household size	3.49	3.69	3.51	3.60	3.09	2.27	2.75	2.56	2.84
West Germany
Adjusted pregovernment income	36,497	36,012	35,716	26,960	29,741	30,849	30,863	30,092	31,774
Adjusted postgovernment income	27,809	27,534	28,545	21,351	23,444	24,442	24,536	24,214	25,792
Old/new partner’s earnings	44,048	43,171	41,639	25,546	23,023	23,898	23,271	21,224	22,847
Own earnings	16,103	16,247	16,266	16,535	17,592	17,580	18,074	18,505	19,434
Private transfers	48	106	127	801	1,148	1,644	1,911	1,770	1,309
Public transfers	5,324	4,904	4,601	4,089	3,603	3,837	3,898	4,066	3,987
Income from capital	1,787	2,138	2,230	1,955	1,930	2,030	2,104	2,597	2,069
All other income	4,506	4,668	4,546	3,078	3,736	4,356	3,318	2,679	4,372
Total household income	66,447	66,248	65,583	47,721	47,151	49,227	48,134	46,077	49,530
Income and payroll taxes	21,509	21,415	12,803	14,744	14,406	15,268	14,971	13,655	15,056
Household size	3.50	3.51	3.24	2.62	2.66	2.66	2.51	2.46	2.50

The columns of Table 1, Panel A, contain information on women’s families that experience a split relative to the timing of the dissolution (–3, –2, … 4, 5). The top row of the table contains values of size-adjusted pregovernment income. Unadjusted family income and family size are shown in rows near the bottom of the table. The components of income are also not adjusted for family size.

The table shows that in the United States from 1985 to 1993 relative to the three years prior to the split, a large drop—from $41,806 to $21,958 (47 percent less)—occurs for pregovernment income at the time of the dissolution. However, by five years later, pregovernment income returns to $37,334, a decline of 10 percent. A similar pattern is observed in postgovernment size-adjusted income that drops from $32,148 three years prior to the split to $19,490 (39 percent less) the year of dissolution followed by an almost full recovery to $31,299 (3 percent less) five years later. The components of income that change most in response to the split are large declines in partner’s earnings and increases in the women’s own earnings.

There is a more sustained drop in resources for this period for German women. Three years prior to the dissolution, pregovernment and postgovernment size-adjusted incomes were €36,727 and €26,726, respectively. Sizable drops (to €23,545 and €18,521) of 36 and 31 percent occur at the time of the split; and while some recovery occurs, five years later the pregovernment and postgovernment income levels stand at €25,027 and €21,527, still 32 and 20 percent lower, respectively. Similar to the United States, the loss of a partner’s earnings is the most important contributor to this decline. We observe less of an increase in earnings for German women. Consistent with prior comparative studies using similar methods, we find a relatively smaller labor force response and a larger sustained proportionate loss of resources for German than for American women.

In both countries, it is important to note in viewing the adjusted pregovernment and postgovernment income figures in the tables that there is a nonlinear relationship between the decline in unadjusted incomes and family size. For example, in Germany, the unadjusted household income declines from €61,085 the year prior to the dissolution to €42,956 the year afterward, a reduction of 29.7 percent. The change in average family size across those three years is from 3.74 to 2.81 people, or a reduction of 24.9 percent. Rough calculations based on the average family sizes and unadjusted incomes in the table suggest that pregovernment income the year prior to a dissolution would be (€61,085/[3.74]^1/2) €31,069 and one year afterward of (€42,956/[2.81]^1/2) €25,625, a decline of 17.5 percent. The actual figures for adjusted pregovernment income shown in the tables that are based on averages across the individual families in the sample similarly show that in the year prior to a split, adjusted income was €32,926, and a year afterward it was €26,355, a percentage decline of 20 percent. Thus, due to a reduction in the necessary resources to maintain a smaller family, proportionate changes in size-adjusted incomes are somewhat smaller than proportionate changes in unadjusted incomes.

Also, in both surveys, the timing of collection of information on the family roster is for the year of the survey, whereas the income information is for the prior year. If a former spouse was present part of the year but is not present at the time of the survey, they may be reported as present. Alternatively, the amount of income received from them the year of the split may not be accurately reported the following year (Burkhauser, Holden, and Myers 1986). We have aligned the available information appropriately but have not further corrected the data, for example, by trying to impute income amounts for the presence of former partners for portions of years when their incomes are not reported. Thus, the exact family sizes and incomes reported particularly in the year of dissolution should be viewed with some caution, although the changes from before to after divorce generally align with expectations of a reduction in family size of about one person.

Table 1, Panel B, includes similar information for women’s families from 2005 to 2013. Drops in pregovernment and postgovernment income of 47 and 43 percent, respectively (to $27,057 and $22,736), occur in the United States at the time of the split from levels three years prior ($51,004 and $39,829). Incomes observed five years later ($36,772 and $33,177) represent decreases of 28 and 17 percent, respectively. During the sample period, large drops occur in income from a partner while own earnings remain relatively stable. Income taxes also decline markedly following the dissolution and decline in income.

The sample for the United States in this second period is based on biennial surveys. One implication, as described earlier in the data section, is that the averages presented in alternate columns are based on different samples because a family may split the year of a survey or between them. This is because the survey income measures may refer to the year of dissolution or a year prior. Due to this, there is an observed jaggedness in the decline of family income following dissolution as well as a slower decline in average family size. However, by looking at patterns across each alternate column, one is observing patterns in the same sample and the patterns are smoother. For example, two years prior to dissolution, average adjusted pregovernment income (and family size) is $57,498 (3.69); the year of dissolution, it is $27,057 (3.60); two years later, it is $38,896 (2.27); and four years later, it is $39,834 (2.56). The one somewhat anomalous figure in the table is the family size one year after dissolution (3.09). Due to this, some caution is appropriate with respect to the figures presented for the year of dissolution and for the year afterward. Nonetheless, the pattern of a long-term decline in the economic well-being of women’s families extending for many years following family dissolution is clear in the table.

In Germany, there are survey data available for every year in this second period. Pregovernment and postgovernment size-adjusted incomes are €36,497 and €27,809, respectively, three years prior to the family split. The year of the dissolution, they drop to €26,960 and €21,351 (or by 26 and 23 percent), respectively. Five years later, the levels stand at €31,774 and €25,792, respectively (declines of 13 and 7 percent). These proportionate drops are roughly half the magnitude of those observed among U.S. women. In Germany, the reductions of income from a partner are somewhat offset by increased earnings of the woman and smaller taxes.

Comparing the two periods represented in Panels A and B of Table 1, sustained proportionate losses in household resources following a family split were larger among German women’s families in comparison to those in the United States in the first period. However, these descriptive statistics suggest that pattern has been reversed in the past three decades as the proportionate reduction in access to resources following the dissolution of a relationship among American women has increased.

In examining comparable information for U.S. and German men for the period from 1985 to 1993 contained in appendix Table A1, Panel A, drops in resources are observed the year of a family dissolution in both countries; but by the end of the sample period, men’s incomes increase in both countries. Patterns for men in both countries also descriptively appear similar when examining information for the second period (2005–2013) in appendix Table A1, Panel B. Similar tables for the families of the continuously intact families used as comparisons can be found in appendix Table A2, Panels A through D.

Panel Fixed-Effect Estimates

American families

Because of possible observed and unobserved differences in the composition of those who divorce as well as other idiosyncratic changes that might occur across time or successive generations, we also provide panel fixed-effect estimates in Table 2 considering pregovernment and postgovernment family size-adjusted income as outcomes. All of the estimates use sample weights and cluster standard errors by individual.

Table 2

Fixed-Effect Estimates of Impact of Family Dissolution on Pregovernment and Postgovernment Equivalized Income

	Panel A: United States
	1985–1993				2005–2013
	Men	Men	Women	Women	Men	Men	Women	Women
Yr. from Split	Pregovt.	Postgovt.	Pregovt.	Postgovt.	Pregovt.	Postgovt.	Pregovt.	Postgovt.
−3	2,458.8	2,697.2*	1,355.5	1,783.8*	6,548.6	4,559.7	5,055.0	1,278.1
	(2,419.1)	(1,552.9)	(1,654.2)	(1,053.6)	(5,490.8)	(3,876.1)	(6,727.8)	(4,759.6)
−2	566.7	1,448.2	335.3	1,051.5	5,444.2	4,893.9	8,330.4	4,079.3
	(2,892.6)	(1,735.2)	(1,834.6)	(1,146.2)	(4,960.1)	(3,568.6)	(6,494.4)	(4,114.9)
−1	−2,669.1	−797.1	−3,202.2	−1,825.2	5,807.4	3,164.6	−6,122.1	−8,501.2*
	(2,652.4)	(1,889.3)	(2,067.3)	(1,509.1)	(5,126.3)	(3,385.0)	(6,748.7)	(4,964.8)
0	−8,499.9***	−6,264.4***	−19,753.6***	−13,050.0***	−4,830.7	−5,236.7	−26,494.6***	−20,589.2***
	(2,869.9)	(1,816.5)	(1,967.2)	(1,270.6)**	(6,160.1)	(4,965.7)	(7,880.8)	(4,656.8)**
1	−2,178.3	−1,805.7	−7,725.0***	−4,758.7**	−7,699.0	−8,359.8*	−12,403.2*	−12,205.5**
	(2,691.6)	(1,733.9)	(2,908.0)	(1,892.4)*	(6,905.5)	(4,866.4)	(7,244.6)	(4,993.6)*
2	−5,189.0**	−4,088.7**	−8,878.2***	−5,910.7***	2,687.6	24.08	−11,427.0	−11,965.8**
	(2,559.6)	(1,613.7)	(2,026.5)	(1,328.1)*	(9,693.5)	(6,350.4)	(8,873.4)	(5,387.9)*
3	−2,946.2	−2,300.9	−5,152.5***	−2,961.8**	−7,047.4	−9,543.0*	−11,467.7	−11,379.5**
	(2,851.2)	(1,833.3)	(1,727.2)	(1,194.8)*	(7,390.4)	(5,270.4)	(7,786.3)	(5,374.0)*
4	−3,004.4	−2,244.2	−3,760.4**	−2,582.9**	3,739.4	512.1	−14,547.2***	−14,407.0***
	(2,574.2)	(1,775.9)	(1,685.5)*	(1,144.9)*	(5,967.0)	(3,993.2)	(5,468.8)*	(4,213.7)*
5	−1,135.2	−838.2	−4,347.9***	−2,416.6**	1,220.0	−3,100.4	−9,430.2	−11,273.2*
	(2,191.4)	(1,464.9)	(1,602.4)	(1,193.3)	(9,341.8)	(7,007.8)	(8,822.8)	(5,842.8)
N	22,046	22,041	23,273	23,268	7,807	7,806	8,182	8,182
	Panel B: Germany
	1985–1993				2005–2013
	Men	Men	Women	Women	Men	Men	Women	Women
Yr. from Split	Pregovt.	Postgovt.	Pregovt.	Postgovt.	Pregovt.	Postgovt.	Pregovt.	Postgovt.
−3	−713.0	−486.1	4,488.2	2,828.3*	598.2	550.0	1,875.2	1,073.2
	(2,237.2)	(1,285.6)	(3,509.2)	(1,698.9)	(1,348.6)	(791.7)	(1,908.4)	(1,109.8)
−2	3,411.0	1,705.3	5,113.6***	3,098.1***	2,728.9	1,236.8	3,021.6*	1,344.4
	(2,345.9)	(1,352.7)	(1,772.3)	(979.3)	(1,875.0)	(1,046.7)	(1,550.0)	(901.8)
−1	2,101.2	474.2	2,053.8	808.1	3,134.9*	63.47	1,389.1	902.7
	(2,844.9)	(1,743.6)	(2,057.5)	(1,332.8)	(1,797.0)	(2,026.4)	(1,999.3)	(1,488.9)
0	−3,469.6	−4,288.6**	−6,755.4***	−5,209.4***	882.2	−2,641.3	−8,037.9***	−7,062.4***
	(2,729.4)	(1,662.8)	(1,989.9)	(1,232.2)	(3,060.9)	(1,833.2)	(2,525.7)	(1,569.1)
1	1,731.7	−281.0	−4,227.2**	−3,160.4***	2,402.0	−1,207.5	−5,995.2***	−5,602.4***
	(3,075.8)	(2,023.7)	(1,715.9)	(1,057.6)	(2,298.4)	(1,401.1)	(2,291.0)	(1,401.2)
2	469.7	−1,513.8	−2,322.9	−2,195.4*	2,442.9	−1,864.0	−3,658.3	−4,062.2***
	(2,210.6)	(1,359.7)	(2,005.8)	(1,203.8)	(2,440.7)	(1,505.4)	(2,260.3)	(1,305.8)
3	3,341.4	−445.8	−1,895.2	−2,264.6**	2,004.2	−2,547.3*	−2,791.7	−3,873.1***
	(2,478.5)	(1,411.6)	(1,805.9)*	(1,063.5)	(2,412.9)	(1,425.6)	(2,500.5)	(1,494.6)
4	4,550.2*	1,506.0	−1,789.2	−1,781.0	1,851.0	−2,394.1	−2,939.4	−4,210.9***
	(2,566.6)	(1,476.0)	(2,061.1)	(1,215.4)	(2,639.6)	(1,557.7)	(2,695.0)	(1,626.6)
5	2,957.8	393.0	−3,853.8**	−2,569.9***	8,352.0*	1,777.2	1,141.1	−1,674.9
	(2,174.6)***	(1,062.9)	(1,555.5)***	(841.9)	(4,443.6)***	(2,446.3)	(2,731.3)***	(1,540.6)
N	17,242	17,241	17,335	17,334	16,569	16,568	16,832	16,831

NOTE: Standard errors in parentheses

p < .10. **p < .05. ***p < .01.

In considering the pattern of estimates for American men in the period from 1985 to 1993, the estimated decline in pregovernment equivalized income the year of a family split is $8,500, representing a loss of 18 percent of average income ($46,130) three years prior. A statistically significant drop of $5,189 (or 11 percent) is also observed two years after the dissolution but afterward, size-adjusted pregovernment incomes of men’s families that split do not significantly differ from those of the comparison group.

The pattern of declines in postgovernment income for U.S. men whose families split relative to intact ones from 1985 to 1993 are similar to those for pregovernment income. The year of the split, equivalized postgovernment income falls by $6,264, a decline of 18 percent relative to the average ($35,356) three years prior. The last statistically significant parameter is observed two years after the family split (–$4,089), a drop of 12 percent. Beyond that point, the estimated changes in U.S. men’s postgovernment incomes are statistically insignificant.

For U.S. women in the period from 1985 to 1993, pregovernment incomes declined by a statistically significant $19,754 the year of the split, a drop of 47 percent relative to average income ($41,806) three years prior. Pregovernment equivalized incomes remained lower (statistically significant) for all five years after the family split. The decline five-years after the dissolution (–$4,348) is 10 percent of average income three years prior to the split.

The estimated decline in postgovernment income of $13,050 the year of the split is a reduction of 31 percent relative to the average three years prior ($32,148). Again, statistically significant reductions are observed for all five years after the family split. The value five years later (–$2,417) is a drop of 8 percent relative to the average three years earlier. For men and women in the United States in this period (and for almost all the estimates in Table 2), there is no evidence of declines in family income prior to dissolution. This supports the interpretation that the parameters from the fixed-effect models measure the effect of family dissolution on income.

In the second period (2005–2013), for U.S. men, there is little evidence of persistent declines in pregovernment or postgovernment equivalized incomes. From the year of the dissolution to five years later, twelve parameters are estimated and only two are statistically significant.

In the second period (2005–2013), for U.S. women, pregovernment equivalized income falls by a statistically significant $26,495 the year of the dissolution, a drop of 52 percent relative to the average value three years earlier ($51,004). The last statistically significant parameter is observed in the fourth year following divorce and is a sizable $14,547, a decline of 29 percent.

From 2005 to 2013, postgovernment equivalized incomes of U.S. women’s families decline (statistically significant) by $20,589 the year of dissolution, a drop of 52 percent relative to the average ($39,829) three years earlier. Statistically significant drops are observed each year afterward. The estimated decline five years later is $11,273 (28 percent). The estimated losses in postgovernment incomes for U.S. women’s families appear much larger in the second time period.

German families

For German men in the first period (1985–1993), virtually none of the estimated parameters of impacts of family dissolution on pregovernment or postgovernment income are statistically significant. The only significant parameter is at the time of the split for postgovernment income (–€4,289). The pattern of experiences in the period from 1985 to 1993 for German women is clearly more adverse. Statistically significant declines in pregovernment and postgovernment income are observed the year the family splits (–€6,755 and –€5,209), representing declines of 18 and 19 percent, respectively, relative to average values three years prior (€36,727 and €26,726). Five years later, declines of €3,854 and €2,570 or 10 and 19 percent, respectively, are observed.

In the second period (2005–2013), virtually none of the parameters estimated for changes in the pregovernment or postgovernment equivalized incomes of German men are statistically significant. Thus, little change is observed relative to the period from 1985 to 1993. For German women, larger drops are observed the year the family dissolves (–€8,038 and –€7,062) for pregovernment and postgovernment measures, respectively. These represent declines of 23 and 25 percent, respectively, relative to averages three years prior to the family dissolution (€35,330 and €27,809), although significant declines are only observed for four consecutive years for postgovernment equivalized incomes. The decline four years later of €4,211 is a reduction of 15 percent relative to the average three years before the dissolution.

Finally, we estimate models pooling available data and interacting covariates with a categorical variable for the two time periods. The parameters for the interacted variables measure whether changes occur in patterns of income losses across periods. Statistical significance is denoted with markers placed beside the standard errors in Table 2. In general, differences in the pattern of losses of postgovernment equivalized income for U.S. women are sizable and statistically significant at the 10 percent level across the two periods for five of the six estimated parameters, which indicates a worsening of experiences across the two periods. Although not shown in the table, we also find that across the two periods in the United States, the presence of children leads to lower incomes in the second period for women and higher ones for men.

Discussion and Conclusion

The PSID, together with the development of life course analysis, provided the foundation for the first systematic evidence of the consequences of union dissolution in the United States on the subsequent economic well-being of the partners and the children of their unions. More importantly, the PSID served as a model for creation of other panel datasets such as the SOEP. This milestone in international data development made it possible for important life course events to be analyzed from a cross-national perspective.

The initiation of the PSID and the collection of the SOEP also led to the creation of a data archive containing comparable measures across similar international surveys, the CNEF, which now contains information from eight panel datasets. It also served as an important model for harmonization of data from panel surveys focused on more specific populations such as the aged and for international organizations such as the European Union.

Here, we revisited the topic of the first cross-national social research conducted using panel data on family dissolution among cohabiting and martial relationships in the United States and Germany. Making use of PSID and SOEP data, we examine two time periods (1985–1993 and 2005–2013) that span three decades. We follow families of men and women that dissolve beginning as early as three years before the split and following them up to five years afterward.

In the first period, making comparisons before and after families dissolve, we find that long-term declines in the family size–adjusted pregovernment and postgovernment incomes of women in both countries are more pronounced and sustained than those of men, and that long-term losses are proportionately larger among German women—a conclusion similar to the one reached in the original research on this topic (Burkhauser et al. 1990, 1991). Also consistent with earlier research, reductions in earnings from a partner are the largest contributor to this pattern. Losses are somewhat offset by increased earnings among the women, decreased taxes, and reduced family size. Men’s resources are largely unaffected by a family split.

This pattern changes somewhat in the second period that we examine. In that period, long-term declines in pregovernment and postgovernment incomes become larger for American women relative to Germans. This pattern appears to be principally due to declining contributions of a partner’s earnings and less scope for women to increase work afterward due to their already high levels of labor force participation (Tamborini, Couch, and Reznik 2015).

We also make use of panel fixed-effect estimators to compare the experiences of the dissolving families to those that remain intact over the sample periods. Those analyses reveal similar patterns found in the descriptive analysis with a few exceptions. In comparison to intact couples, the families of men that experience a split in the United States also have mild but statistically significant reductions in their pregovernment and postgovernment equivalized earnings for several years in the first period of the analysis (1985–1993). The panel fixed-effect estimates for the second period (2005–2013) confirm that the declines in the levels of postgovernment income are proportionately larger among the families of American women than they are for German women. Statistical tests confirm that the larger reductions in postgovernment equivalized incomes of U.S. women’s families in the second period are significantly larger than in the first. We find no evidence of changes across the two periods in the outcomes of men in the United States or German families headed by men or women that dissolve.

The increased magnitude of the drop in family size–adjusted incomes of American women following a divorce or separation is a potentially important new finding in the literature. While increased incomes of women’s households without a partner and resulting reductions in cohabitation and remarriage likely explain this finding, confirming these patterns with additional research would be appropriate.

Despite the usefulness of the panel data employed in this analysis in studying the evolving dynamics of families, both surveys report members of the family at the time responses are made, while income information is for the prior year. A year later, if a family split has occurred, there is a concern this may lead to substantial misreporting of income received the year of the split while the former partner was present in the household (Burkhauser, Holden, and Myers 1986). Potentially, this problem may be exacerbated by the recent move to biennial surveys by the PSID. More research and guidance or possibly the provision of relevant income measures the year of a family dissolution would be useful and allow researchers to form a sharper picture of the experiences of families right at the time they split.

Cross-national research allows us to gain a better understanding of the importance of historical context and institutions on relationship formation, changes in them, and resulting economic consequences. The literature using the PSID in the early 1980s focused on the differential consequences of union dissolution on the economic resources of male and female partners and the children of their union. It found that American women systematically experienced worse outcomes than their male partners. Our findings here, that both German and American women continue to do worse than their male partners for many years following a union dissolution, should be of concern given the increasing evidence of a relationship between childhood experiences and later adult outcomes (Currie and Rossin-Slater 2015). The large reductions in the economic well-being of women’s families after a family dissolution point toward the importance of potential approaches to bolster the resources of those families, such as child support payments, dependent deductions from taxes, and other cash and in-kind government benefits.

Footnotes

Notes

Gulgun Bayaz-Ozturk is an assistant professor of economics at CUNY City Tech. The focus of her current research is rather broad, but exclusively relates to older individuals and underprivileged minority groups. Her research papers are published in peer-reviewed journals such as Economic Inquiry, Applied Economics, and Journal of Poverty.

Richard V. Burkhauser is the Emeritus Sarah Gibson Blanding Professor of Policy Analysis in the Department of Policy Analysis and Management at Cornell University. He was appointed a member of President Trump’s Council of Economic Advisors in September 2017. He has published widely on economic issues related to vulnerable populations.

Kenneth A. Couch is a professor of economics at the University of Connecticut and former editor-in-chief of the Journal of Policy Analysis and Management. His research focuses on disadvantaged groups in society, programs to assist them and their effects both in the United States and in comparative analyses.

Richard Hauser is full professor emeritus in the Economics Department of Goethe-University Frankfurt am Main and a fellow of Hanse Wissenschaftskolleg, Delmenhorst, and Wissenschaftskolleg zu Berlin. His published books include Anatomie der Einkommensverteilung (2003); Verteilungseffekte der Hartz-IV-Reform (2006); Soziale Gerechtigkeit – ein magisches Viereck (2009); and Die Verteilung der Vermögen in Deutschland (2010).

References

Andreß

Hans-Jürgen

Borgloh

Barbara

Bröckel

Miriam

Gießelmann

Marco

Hummelsheim

Dina

. 2009. The economic consequences of partnership dissolution: A comparative analysis of panel studies from Belgium, Germany, Great Britain, Italy and Sweden. In When marriage ends: Economic and social consequences of partnership dissolution, eds. Andreß

Hans-Jürgen

Hummelsheim

Dina

, 286–329. Camberley, UK: Edward Elgar.

Andreß

Hans-Jürgen

Bröckel

Miriam

. 2007. Income and life satisfaction after marital disruption in Germany. Journal of Marriage and the Family 69 (2): 500–12.

Bayaz-Ozturk

Gulgun

Burkhauser

Richard V.

Couch

Kenneth A.

2013. Different directions: Economic mobility in the western states of Germany and the U.S. from 1984–2006. Economic Inquiry 52 (1): 431–43.

Bredtmann

Julia

Vonnahme

Christina

. 2017. Less alimony after divorce: Spouses’ behavioral response to the 2008 alimony reform in Germany. IZA Discussion Paper No. 10864, Bonn.

Bröckel

Miriam

Andreß

Hans-Jürgen

. 2015. The economic consequences of divorce in Germany: What has changed since the turn of the millenium? Comparative Population Studies 40 (3): 277–312.

Burkhauser

Richard V.

Duncan

Greg J.

1989. Economic risks of gender roles: Income loss and life events over the life course. Social Science Quarterly 70 (1): 3–23.

Burkhauser

Richard V.

Duncan

Greg J.

Hauser

Richard

Berntsen

Roland

. 1990. Economic burdens of marital disruptions: A comparison of the United States and the Federal Republic of Germany. Review of Income and Wealth 36 (4): 319–33.

Burkhauser

Richard V.

Duncan

Greg J.

Hauser

Richard

Berntsen

Roland

. 1991. Wife or frau, women do worse: A comparison of men and women in the United States and Germany following marital dissolution. Demography 28 (3): 353–60.

Burkhauser

Richard V.

Holden

Karen C.

Myers

Daniel A.

1986. Marital disruption and poverty: The role of survey procedures in artificially creating poverty. Demography 23 (4): 621–31.

10.

Burkhauser

Richard V.

Smeeding

Timothy

. 1990. The well being of the elderly in a comparative context. National Institute on Aging proposal P015R03AG21009-02. Bethesda, MD: National Institute on Aging.

11.

Burkhauser

Richard V.

Smeeding

Timothy M.

Merz

Joachim

. 1996. Relative inequality and poverty in Germany and the United States using alternative equivalency scales. Review of Income and Wealth 42 (4): 381–400.

12.

Cherlin

Andrew J.

Kiernan

Kathleen

Chase-Lansdale

P. Lindsay

. 1995. Parental divorce in childhood and demographic outcomes in young adulthood. Demography 32 (3): 299–318.

13.

Corcoran

Duncan

Greg J.

Hill

M. S.

1984. The economic fortunes of women and children: Lessons from the Panel Study of Income Dynamics. Signs: Journal of Women in Culture and Society 10 (2): 232–48.

14.

Couch

Kenneth A.

Tamborini

Christopher R.

Reznik

Gayle L.

2015. The long-term health implications of marital disruption: Divorce and Social Security disability benefits among men. Demography 52 (5): 1487–1512.

15.

Couch

Kenneth A.

Tamborini

Christopher R.

Reznik

Gayle A.

Phillips

John W. R.

2013. Divorce, women’s earnings, and retirement over the life course. In Lifecycle events and economic security: The roles of job loss, disability and changing family structure, eds. Couch

Kenneth A.

Daly

Mary C.

Zissimopoulos

Julie

, 133–57. Stanford, CA: Stanford University Press.

16.

Currie

Janet

Rossin-Slater

Maya

. 2015. Early-life origins of life-cycle well-being: Research and policy implications. Journal of Policy Analysis and Management 34 (1): 208–42.

17.

de Vaus

David

Gray

Matthew

Lixia

Stanton

David

. 2017. The economic consequences of divorce in six OECD countries. Australian Journal of Social Issues 52:180–99.

18.

Duncan

Greg J.

Hoffman

Saul D.

1985 A reconsideration of the economic consequences of marital dissolution. Demography 22 (4): 485–97.

19.

Hauser

Richard

Glatzer

Wolfgang

Hradil

Stefan

Kleinhenz

Gerhard

Olk

Thomas

Pankoke

Eckart

. 1996. Ungleichheit und Sozialpolitick. Wiesbaden: VS Verlag für Sozialwissenschaften.

20.

Hoffman

Saul D.

Duncan

Greg J.

1988. What are the economic consequences of divorce? Demography 25 (4): 641–45.

21.

Hummelsheim

Dina

. 2009. Die Erwerbsbeteiligung von Müttern: Institutionelle Steuerung oder kulturelle Prägung? Wiesbaden: VS Verlag für Sozialwissenschaften.

22.

Martin

Gary

Kats

Vladimir

. 2003. Families and work in transition in 12 countries. Monthly Labor Review 126 (September): 3–33.

23.

Michael

Robert T.

1978. The rise in divorce rates, 1960–1974: Age-specific components. Demography 15 (2): 177–82.

24.

Schonlau

Matthias

Watson

Nicole

Kroh

Martin

. 2011. Household survey panels: How much do following rules affect sample size? Survey Research Methods 5 (2): 53–61.

25.

Smock

Pamela J.

1993. The economics costs of marital disruption for young women over the past two decades. Demography 30 (3): 353–71.

26.

Tach

Laura

Eads

Alicia

. 2015. The economic consequences of marital and cohabitation dissolution. Demography 52 (2): 401–32.

27.

Tamborini

Christopher R.

Couch

Kenneth A.

Reznik

Gayle L.

2015. The long-term impact of discrete life-course events: A multi-cohort analysis of divorce and women’s earnings. Advances in Life Course Research 26:44–59.

28.

Tamborini

Christopher R.

Reznik

Gayle L.

Couch

Kenneth A.

2016. Divorce and the long-term incidence of work disability among women: A retrospective cohort study. Journal of Health and Social Behavior 22 (2): 489–522.

29.

Wagner

Gert

Burkhauser

Richard V.

Behringer

Friederike

. 1993. The English language public use file of the German Socio-Economic Panel. Journal of Human Resources 28 (2): 429–33.

The Effects of Union Dissolution on the Economic Resources of Men and Women: A Comparative Analysis of Germany and the United States,1985–2013

Abstract

Keywords

Literature Review

Data

The CNEF and the PSID and SOEP surveys

Outcome measures

Methods

Descriptive Statistics

Panel Fixed-Effect Estimates

American families

German families

Discussion and Conclusion

Footnotes

Notes

References