Evaluation Review http://erx.sagepub.

com/

The Effect of Economic Stability on Family Stability among Welfare Recipients

Alisa C. Lewin Eval Rev 2005 29: 223 DOI: 10.1177/0193841X04272558 The online version of this article can be found at: http://erx.sagepub.com/content/29/3/223

Published by:
http://www.sagepublications.com

Additional services and information for Evaluation Review can be found at: Email Alerts: http://erx.sagepub.com/cgi/alerts Subscriptions: http://erx.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://erx.sagepub.com/content/29/3/223.refs.html

>> Version of Record - Apr 28, 2005 What is This?

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

EVALUATION 10.1177/0193841X04272558 Lewin / THE EFFECT REVIEW OF/ ECONOMIC JUNE 2005 ST ABILITY

THE EFFECT OF ECONOMIC STABILITY ON FAMILY STABILITY AMONG WELFARE RECIPIENTS

University of Haifa and University of Chicago The main rationale for defining two-parent families eligible for welfare was to keep families intact by eliminating an incentive for union dissolution. But there are other reasons for family instability, most notably womens reduced economic gain from marriage associated with having a chronically unemployed husband. This article explores the hypothesis that husbands unemployment increases union dissolution among welfare recipients. The analysis uses data from Californias Link-Up demonstration project. A discrete-time event-history methodology was employed to examine family instability. The findings show that husbands unemployment and the familys long-term welfare dependency lead to breakup, net of race, age, and number of children. Keywords: poverty; divorce; welfare; AFDC; Link-Up

ALISA C. LEWIN

Claims have been made that welfare policy aimed at single mothers is a disincentive to marriage (Murray 1984). Policy makers response to such claims was to define two-parent families as eligible for welfare. Providing welfare to two-parent families supports these families in periods of extreme economic hardship and eliminates an incentive for union dissolution. But making aid available to two-parent families does not eliminate other reasons for family instability, most notably, womens reduced economic gain from marriage associated with having a chronically unemployed husband. In this study, I examine the effect of income and mens unemployment on family instability among welfare recipients.
AUTHORS NOTE: This project was supported by the Alfred P. Sloan Center for Working Families at the University of Chicago. This study benefited greatly from comments and suggestions by Richard Berk, Dahlia Elazar, Guy Enosh, Zeke Hasenfeld, Julia Henly, Yona Rubenstein, Judah Matras, Valerie Oppenheimer, Haya Stier, Linda Waite, and anonymous referees of Evaluation Review. I thank Michael N. Mitchell, Marina Shapira, and Tal Oron for excellent research assistance and data management throughout the project. An early version of this article was presented at the annual meeting of the Population Association of America, October 2004, in Boston.
EVALUATION REVIEW, Vol. 29 No. 3, June 2005 223-240 DOI: 10.1177/0193841X04272558 2005 Sage Publications

223

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

224

EVALUATION REVIEW / JUNE 2005

This study contributes to the debate on the relationship between welfare policy and family stability in two ways; the first contribution is methodological and the second is theoretical. In terms of method and analysis, in this study I use household-level data on welfare recipients to examine the relationship between household income and family instability. This analytical strategy provides a better test than possible in studies that use aggregate data and ecological regressions (Bitler et al. 2004), or studies that use householdlevel data and aggregate measures of welfare policy (Blackburn 2003; Hoynes 1997). For example, Hoynes (1997) used Panel Study of Income Dynamics (PSID) data and a general measure of Aid to Families With Dependent Children (AFDC) grant, and Blackburn (2003) used National Longitudinal Study of Youth (NLSY) data and a measure of mean level of welfare grant in the state of residence. Both studies examined the effect of welfare policy on union dissolution in the entire population. I argue that the test in the current study improves upon previous studies for two reasons. First, the research sample in the current study represents welfare recipients rather than the entire population. Welfare recipients are more likely to be directly affected by welfare policy than the general population, a large portion of which is not at risk of receiving welfare, even upon breakup. Therefore, a sample of welfare recipients is more appropriate for studying the effects of welfare on family structure than a nationally representative sample of the entire population. Second, I use a household-level measure of welfare policy that indicates the amount of welfare the household received. The amount of welfare a family received is affected by family size and earned income from work. This measure allows for a more direct estimation of welfares effect on the gain from marriage and its incentive to break up a family than more general contextual variables indicating the average level of welfare in the state. This studys theoretical contribution is its focus on mens economic activity rather than womens economic activity. In this study, I ask whether husbands chronic unemployment reduces the gain from marriage and increases the likelihood of breakup. I focus on families with unemployed or underemployed husbands and investigate the effect of income from husbands earnings, wives earnings, and welfare grants on family instability. Many studies on family instability have focused on changes in womens earnings, perhaps because the rise in married womens employment coincided historically with a rise in marital dissolution. But predictions on the effect of womens work on marital stability have been ambiguous (Weiss 1997). Although it is clear that from the wifes perspective, her husbands income increases her gain from marriage, the effect of her own income on her gain from marriage is unclear,

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

225

because her own income raises her utility both in the married and unmarried states (Bitler et al. 2004). Thus, depending on tastes for marriage and how income is distributed in the household, the effect of wifes earnings on divorce may have an independence effect, making being single more attractive (Becker 1973, 1981; Becker, Landes, and Michael 1977) or a stabilizing effect, making marriage more attractive and unions more stable (Oppenheimer 1997). Recent policy interest has set out to promote two-parent families and has defined two distinct goals: first, to encourage marriage among single mothers; and second, to increase the stability of low-income two-parent families. The current study has implications for the second goal. Though the Temporary Aid for Needy Families (TANF) program differs from its predecessor, AFDC, in many respects related to work incentives, time limits, and block grants, I argue that the question of the relationship of welfare, unemployment, and family instability remains as timely as ever, theoretically and politically.1

ECONOMIC GAIN FROM MARRIAGE AND UNION DISSOLUTION The debate on the economic causes of divorce is grounded in Beckers (1981) argument that the increased economic independence of women, resulting from their labor force participation, led to a reduction in their economic gain from marriage. This theory sparked much research, but the findings have been ambiguous; womens employment has been found to increase dissolution, yet wives income reduces it (Ruggles 1997; South 2001; South and Spitze 1986; Spitze and South 1985; Greenstein 1990; Hoffman and Duncan 1995; Tzeng and Mare 1995). By Beckers (1981) same theoretical logic, the advantages of the traditional sexual division of labor in the family are reduced when the husband is unemployed. Here there are fewer studies, but the findings are more consistent. For example, Tzeng and Mare (1995) found that couples where the wife works much more than her husband have higher rates of marital disruption than couples where the husband works much more than his wife, suggesting that husbands low work effort and earnings may have a negative effect on marital stability. Cherlin (1979) found that the stability of husbandsemployment, rather than the absolute level of earnings, is an important determinant of marital dissolution. In these terms, having an unemployed husband greatly reduces the wifes gain from marriage and is likely to make divorce (or

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

226

EVALUATION REVIEW / JUNE 2005

separation) more attractive, especially if the wife has an alternative source of income. Next, Becker (1981, 252) argued that welfare substitutes for husbands earnings and that welfare may reduce the gain from marriage. Though studies have shown that economic hardship is associated with marital instability (Conger et al. 1990) and that marital dissolution occurs more frequently in families on welfare than in the general population (e.g., Rank 1994), there is little evidence that welfare destabilizes marriages (Blackburn 2003; Cherlin 1979; Danziger et al. 1982; Draper 1981; Garfinkel and McLanahan 1986; Hoffman and Duncan 1995; Moffitt 1992; Rank 1987). Instead, studies have demonstrated that welfare reduces divorce and that the rise in female headship could not be explained by welfare support (Bitler et al. 2004; Darity and Myers 1984; Fitzgerald and Ribar 2004; Hoynes 1997; Lichter, McLaughlin, and Ribar 1997; Sweezy and Tienfenthaler 1996; Zimmerman 1994). In fact, recent studies suggested that more relaxed welfare policies may have a positive effect on family stability. For example, findings from the Minnesota Family Investment Program (MFIP) show that increasing work incentives among welfare recipients increases their employment and earnings and has a positive effect on family stability among two-parent families on aid (Gennetian 2003). In this study, I explore the effect of income and husbands unemployment on family instability among welfare recipients. I examined a large sample of women in California, living in two-parent families and on welfare due to husbands unemployment (or underemployment), and I focused on the choices available to them. I traced three alternative paths open to two-parent families on welfare: They could go off aid entirely, they could remain together as a family and on aid, or they could dissolve their unions and qualify for aid as single parents.

DATA, METHOD, AND VARIABLES

DATA

Aid to Families With Dependent ChildrenUnemployed Parent (AFDCU) was established in 1961 and allocated federal funds to states that extended aid to children with able-bodied unemployed parents. Giving states the option of extending eligibility to unemployed two-parent families was a measure to eliminate any previous incentive there might have been to family breakup in order to qualify for welfare. In 1990, all states were mandated to

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

227

have an AFDC-U program. Still, AFDC and AFDC-U participants differed substantially in their work history. Eligibility for the AFDC-U program required that at least one parent demonstrated significant attachment to the labor market, and this parent was restricted to work no more than 100 hours per month while receiving aid. By imposing this 100-hour work limit, the AFDC-U program targeted assistance to children with unemployed parents. Californias Link-Up demonstration project (1992-1994) examined the effect of eliminating the 100-hour work limit by randomly assigning AFDCU recipients to experiment and control groups, where the experiment group had the 100-hour rule waived. One of the rationales for the 100-hour rule waiver was to make the AFDC-U program more similar to the AFDC singleparent program and to eliminate an incentive for two-parent families to break up to qualify for aid. In this study, I used data compiled for evaluating the Link-Up demonstration project (Hasenfeld, Lewin, and Mitchell 1996; Lewin 2001) to examine the effect of husbands unemployment and longterm welfare dependency on family instability among two-parent families on AFDC-U. The data for this study were generated before the 1996 welfare reform initiative. Yet this particular sample of two-parent AFDC families in California is similar enough to two-parent TANF families in California today to render the current study both timely and relevant. The Link-Up experiment group, which had the 100-hour rule waived, was exposed to more similar rules to those currently implemented in California than other AFDC families nationwide. Therefore, in this investigation I selected cases that were randomly allocated to the Link-Up experiment group. The sample for the current study included 2,679 two-parent families receiving welfare in seven counties in central California in 1992 and followed them for 2 years, measured in eight quarters. Welfare benefits (AFDC-U and TANF) are available to two-parent families, regardless of whether the parents are married, as long as they are the biological (or adoptive) parents of the children in the household (see Winkler [1995] for details of AFDC-U eligibility and Moffitt, Reville, and Winkler [1998] on estimates of cohabitation among AFDC mothers). Therefore, in this study I examined stability of cohabiting and marital unions on aid. In terms of policy, this sample adequately represents the population of twoparent families on aid and is therefore the appropriate sample for study. As with many other studies examining the relationship between policy and family outcomes, the data became available for the initial goal of evaluating welfare reform and not for studying union dissolution (Gennetian and Knox 2003). Nonetheless, the data set evinced some notable advantages for the current investigation because it included more detailed information than

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

228

EVALUATION REVIEW / JUNE 2005

is usually available to researchers using survey data. For example, rather than being based on retrospective data, welfare history was obtained from welfare records, and earnings were reported by employers rather than by welfare recipients. The drawback of using administrative records rather than survey data was that other important variables, such as education and year of marriage, were not available for analysis because they were not collected by welfare agencies. The study drew from three data sources: County Administrative Data (1992-1994); California Employment Development Department (EDD) data, including Unemployment Insurance and Disability Insurance (UI/DI) data (1985-1994); and Medi-Cal Eligibility Data System (MEDS) data (1987-1994). County Administrative Data provided information on monthly welfare grant amount, race, and age. EDD data provided quarterly reports of wages by employers of all employees covered by UI (Unemployment Insurance) or DI (Disability Insurance). Welfare recipients in the sample with no employer-reported earnings are recorded as having no earnings. Though certain types of employees are exempt from UI/DI (e.g., casual labor earning less than $50 in a quarter), these employer-reported earnings are expected to be much more reliable than self reported earnings that tend to be underreported because they are deducted from the welfare grant. The state of California administers MEDS, which is a longitudinal database of Medi-Cal recipients. MEDS maintains a record for each individual who has been reported as Medi-Cal eligible, and because AFDC recipients were automatically eligible for Medi-Cal, MEDS records include information on all AFDC recipients. These files include annual welfare history dating back to 1987 and monthly aid codes, which were used to identify cases that moved from AFDC-U (two-parent family) to AFDC-FG (Aid to Families With Dependent ChildrenFamily Group) (single parent), the main dependent variable in the study.
METHOD

The current analysis of family instability employed a discrete-time eventhistory methodology, with the household-quarter as the unit of analysis (Allison 1982; Yamaguchi 1991). In household-quarter files there is a separate record for each household for each quarter that the mother remains in a union and on aid, plus the quarter in which a transition occurs. Once a union dissolves or a family goes off aid, it is no longer exposed to the risk of dissolution or going off aid, and the following quarters observed for the case are removed from the data set.

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

229

The dependent variable in the analysis is whether a transition, either going off aid or breaking up a union, occurred between the two quarters. There are two types of explanatory variables in the current analysis, constant and time varying. Explanatory variables that do not change by quarter are race and welfare history. Other explanatory variables are time-varying covariates (e.g., mothers employment and earnings, fathers employment and earnings, welfare grant amount) and the values used to predict transition are those in the preceding quarter, each quarter at risk. A set of multinomial logistic regressions were calculated to estimate the odds of going off aid or breaking up a union during a quarter, given that the respondent was in a union and on aid the preceding quarter.
VARIABLES

The dependent variable has three categories representing the options available to welfare recipients in two-parent families: (1) remaining in a union and on aid, (2) being on aid as a single parent, meaning that a breakup has occurred, and (3) being off aid. Breakup is measured as a movement from one type of aid (AFDC-U, i.e., two-parent family) to another (AFDC-FG, i.e., single parent). This dependent variable allows for the time it may take to reapply for aid and for some administrative churning and was defined as follows: The category off aid indicates the case was off aid for the entire 3month period in a quarter. The category on AFDC-U indicates the case was on AFDC-U for at least 1 month during the quarter, and was never on AFDCFG during the quarter. The category on AFDC-FG indicates the case was on AFDC-FG for at least 1 month during the quarter. Movement from one aid code to another indicates that a family applied for aid and the eligibility worker moved the case from AFDC as a two-parent family to AFDC as a single-parent family. This proxy for union dissolution is therefore problematic in two ways. First, this measure does not capture changes in marital status that occur after the family goes off aid; it captures changes only among families that reapply for aid after the separation. Therefore, there is no attempt here to generalize to families not on aid. Instead, this study examines the relationship between welfare and breakup and focuses on families who reapply for aid after the breakup. The second problem is that because separation occurs more frequently than dissolution and divorce, and may be more temporary, the dependent variable may overstate the incidence of union dissolution. Nonetheless, the transition to a single-parent family in the welfare records indicates that both the parent and the case worker identified the separation as long-term enough to justify a change in the welfare

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

230

EVALUATION REVIEW / JUNE 2005

files. Therefore, I argue that this measure is a good proxy for family instability. Independent variables. Aid history is measured as a binary variable distinguishing cases new to aid at Q1 (the first quarter in the study) from cases with a recent aid history (measured as cases receiving welfare support any month during the 2-year period prior to the study). Aid history was important in the current investigation because it is an indicator of the chronicity of unemployment and poverty in the household. Other independent variables are womans race, age, and number of children in the household. I also controlled for quarter (duration) in the experiment. I controlled for race in the analysis, because studies have shown different rates of marital dissolution by race (e.g., Hoffman and Duncan 1995). I included womans age as a control variable because it is a proxy of duration of marriage, and for young women it also indicates a young age at marriage. Both age at marriage and duration of marriage are related to the propensity to divorce (South 2001; Becker, Landes, and Michael 1977; Lewin 2004). Unfortunately, it is not possible in this study to control for duration of marriage or age at marriage because this information is not collected by welfare agencies and therefore is not available in the current data set. Having children in the household is a condition of eligibility to welfare; therefore, all couples in the sample at Q1 had at least one dependent child in the household. Number of children is expected to have a negative effect on dissolution, because children have a stabilizing effect on families and because couples who feel they are at risk of separating may have fewer children (Lillard and Waite 1993). I included time-varying covariates measuring fathers and mothers employment, earnings, and grant amount (per person). All the time-varying covariates were measured at the quarter preceding the transition. Mothers and fathers employment was a binary variable indicating whether the parent had any earnings during the previous quarter. Mothers and fathers earnings were the dollar quarterly earnings, as reported by the employer. I used welfare grant amount per person in the household to reduce the correlation between total welfare grant and number of children, because welfare grants increase with family size. Reported earnings were deducted from the welfare grant at a rate of about 67% to 100%, depending on the number of months of employment (the thirty and a third rule); therefore, there is a negative relationship between grant and earned income, and high earnings correspond to low grant and to high odds of being off aid the following quarter. To contain the offsetting relationship between earnings and welfare grant and to test for the overall effect of income

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

231

on family stability, I included a measure of total household income, which combined all earned income and welfare grant. The relationship between welfare grant and the odds of breakup is even more complex because of the time it may take to reapply for aid. If the case goes off aid during a quarter and reapplies as a single-parent family the same quarter, there is no problem in predicting breakup with grant in previous quarter. However, if the case goes off aid during a quarter and reapplies as a single-parent family the following quarter, the case may appear to have a low grant the quarter prior to breakup. In this case, the apparent relationship between grant and breakup reflects measurement problems rather than a stabilizing effect of welfare grants. In this study, I have devised the dependent variable to allow for some administrative churning, but data are not available to solve this measurement problem fully. Therefore, I advise caution in interpreting the effects of the coefficients of welfare grant. Fortunately, income and employment data do not suffer this measurement deficiency because they were derived from an independent source (the employer). Information on earnings and employment was available regardless of whether the case was on aid; therefore, results regarding earnings and employment are more reliable.

FINDINGS Table 1 shows the percentage distribution, means, and descriptions of the variables included in the analysis at Q1. The great majority of cases in the sample had been on aid in the 2-year period prior to the study, and only 16% were new to aid at the beginning of the study (Q1). Almost half of the mothers in the sample were Latina (47%), and 37% were White. Asians represented 14% of the sample of two-parent families on welfare in this period in Californias Central Valley, mainly due to Hmong (or Laos) refugees, and African Americans represented a small minority (2%) of this population. About one third of the mothers were younger than 25 years at the time of the study, and about 22% were older than 35. The mean number of children in a two-parent family on aid was almost three. About one third of the fathers (32%) and 13% of the mothers had some income during the first quarter under investigation; the mean fathers income for the first quarter was $627, and the mean mothers income was $202 for that quarter. Table 2 shows the marginal distribution of the dependent variable by quarter. Table 2 shows that between 4% and 6% of cases moved from being a twoparent family on aid (AFDC-U) to being a single-parent family on aid

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

232

TABLE 1: Variable Definition and Percentage Distribution, Means (Standard Errors) of Variables in the Analysis, First Quarter (N = 2,679)

Variable Name
Off aid Breakup (on AFDC-FG, single parent) On AFDC-U (reference category)

Variable Definition

Dependent Variable

Aid status

Independent variables Aid history (new-to-aid) Mothers ethnicity

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Mothers age

Number of children Father employed Mother employed Fathers income Mothers income Welfare income Total household income Quarter

Binary variable indicating that neither parent had been on aid in the 2-year period prior to the investigation 16% Categorical variable representing four ethnic groups White (reference category, omitted from regressions) 37% Latina 47% African American 2% Asian 14% Younger than age 25 35% Between ages 25 and 35 43% Older than age 35 (reference category, omitted from regressions) 22% Continuous variable indicating number of children in the household 2.88 (1.68) Binary variable indicating father had earned income in quarter prior to outcome 32% Binary variable indicating mother had earned income in quarter prior to outcome 13% Continuous variable indicating fathers income ($) in quarter prior to outcome 627. (1,269) Continuous variable indicating mothers income ($) in quarter prior to outcome 202. (694) Continuous variable indicating welfare grant amount($) per person in quarter prior to outcome 259. (178) Continuous variable indicating total income (earned and welfare grant) ($) in quarter prior to outcome 1,901. (1,413) Variable that ranges from one to eight, indicating period

NOTE: AFDC-FG = Aid to Families With Dependent ChildrenFamily Group; AFDC-U = Aid to Families With Dependent ChildrenUnemployed Parent.

Lewin / THE EFFECT OF ECONOMIC STABILITY

233

TABLE 2: Marginal Distribution of Dependent Variable by Quarter

Quarter
1 2 3 4 5 6 7 8

Transition to Transition AFDC-FG (on Off Aid Aid as Single Parent)

136 (5.1%) 225 (8.9%) 198 (9.0%) 139 (7.2%) 144 (8.4%) 110 (7.3%) 67 (5.0%) 66 (5.3%) 115 (4.3%) 134 (5.3%) 125 (5.7%) 118 (6.1%) 92 (5.4%) 70 (4.7%) 58 (4.3%) 58 (4.7%)

No Transition (on Aid as Two-Parent Family)

2,428 (90.6%) 2,155 (85.7%) 1,884 (85.4%) 1,684 (86.8%) 1,470 (86.2%) 1,318 (88.0%) 1,214 (90.7%) 1,110 (90.0%)

n 2,679 2,514 2,207 1,941 1,706 1,498 1,339 1,234

NOTE: AFDC-FG = Aid to Families With Dependent ChildrenFamily Group.

(AFDC-FG) each quarter, and between 5% and 9% of cases went off aid each quarter. Though some of these transitions may have been followed by additional transitions, in the current event history analysis I focus on first transitions. In the following analysis, I tested whether income, fathers unemployment, and long-term aid dependence among two-parent families on aid affected the odds of family breakup. Tables 3 to 5 show multinomial logistic regression coefficients predicting the odds of going off aid and breaking up, in contrast to staying in a union and on aid. I compared three models: Model 1 (Table 3) included a measure of total household income, whereas Model 2 (Table 4) decomposed household income and examined the net effects of each of its components: mothers earnings, fathers earnings, and welfare grant (per person). Model 3 (Table 5) measured the net effect of parents employment (vs. unemployment) on family breakup. The first two columns in each model represent the odds of going off aid versus staying on aid as a two-parent family, and the second two columns in each model represent the odds of being on aid as a single-parent family versus being on aid as a twoparent family. The results concerning the odds of going off aid versus staying on aid as a two-parent family were as expected; having no recent aid history increased the odds of going off aid in contrast to staying on aid as a two-parent family. Table 3 showed that the higher the total household income, the higher the odds of going off aid. Table 4 showed that earned income had a positive effect on going off aid, whereas welfare grant had a negative effect. This is not surprising, as cases with earned income have lower welfare grants, because as explained earlier, earnings (67%-100%, depending on number of months on

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

234

EVALUATION REVIEW / JUNE 2005

TABLE 3: Multinomial Regression Coefficients (Standard Errors) Predicting Leaving Aid and Breakup Versus Staying in a Union and on Aid (N = 14,914)

Model 1 Off Aid Versus AFDC-U

Quarter Number of children New to aid Asian African American Latina Younger than age 25 Between ages 25 and 35 Household income Intercept Chi square df p 0.002 0.246** 0.348** 0.736** 0.161 0.069 0.030 (0.015) (0.026) (0.087) (0.131) (0.266) (0.071) (0.095)

Odds Ratio
0.998 0.782 1.416 0.479 0.851 0.934 0.970 1.014 1.017

Breakup Versus AFDC-U

0.017 (0.017) 0.026 (0.028) 0.774** (0.148) 1.062** (0.147) 0.354 (0.204) 0.634** (0.083) 0.353** (0.111) 0.054 (0.106) 0.022** (0.003) 2.176

Odds Ratio
1.017 1.026 0.461 0.346 1.425 0.528 1.424 1.056 0.978

0.014 (0.089) 0.017** (0.002) 2.181 471.048 18 .000

NOTE: Household income, mothers income, fathers income, and welfare grant amount were divided by 100. AFDC-U = Aid to Families With Dependent ChildrenUnemployed Parent. **p < .01.

welfare) are deducted from the grant, and months off aid receive no grant at all. Finally, Table 5 showed that having employed parents in the household increased the odds of going off aid. The odds of staying on aid as a single mother (breakup) versus staying on aid as a two-parent family are presented in the second two columns of each table and were the focus of the current study. The results here show that being new to aid reduced the odds of breakup by more than half. Families new to aid had about 44% the odds of breaking up than families with an aid history. These findings suggest that families that have suffered long-term economic hardship are less stable and have higher odds of breakup than families new to aid, net of the other variables in the equation. This is consistent with the hypothesis that chronic unemployment and poverty have a destabilizing effect on families. There were also differences in odds of breakup by race: Latino and Asian families had lower odds of breaking up than white families (the reference category). Differences in union dissolution by race and ethnicity may be due to a

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

235

TABLE 4: Multinomial Regression Coefficients (Standard Errors) Predicting Leaving Aid and Breakup Versus Staying in a Union and on Aid (N = 14,914)

Model 2 Off Aid Versus AFDC-U

Quarter 0.054** Number of children 0.272** New to aid 0.190* Asian 0.070 African American 0.434 Latina 0.437** Younger than age 25 0.033 Between ages 25 and 35 0.017 Mothers earnings 0.025** Fathers earnings 0.022** AFDC grant per person 0.551** Intercept Chi square df p 0.711 1,600.955 22 .000 (0.016) (0.027) (0.092) (0.140) (0.280) (0.077) (0.101) (0.095) (0.004) (0.002) (0.023)

Odds Ratio
1.056 0.762 1.209 1.073 0.648 0.646 0.967 1.017 1.025 1.022 0.576

Breakup Versus AFDC-U

0.038* (0.017) 0.069* (0.027) 0.839** (0.149) 0.876** (0.150) 0.300 (0.205) 0.777** (0.084) 0.330** (0.112) 0.044 (0.106) 0.009 (0.007) 0.007* (0.004) 0.275** (0.022) 1.513

Odds Ratio
1.039 0.934 0.432 0.417 1.350 0.460 1.391 1.045 0.991 0.993 0.760

NOTE: Household income, mothers income, fathers income, and welfare grant amount were divided by 100. AFDC-U = Aid to Families With Dependent ChildrenUnemployed Parent. *p < .05. **p < .01.

variety of reasons, such as religion and cohabitation. For example, Latino families may tend to be Catholic, and Catholics may be less tolerant of divorce than members of other religions. The Asians in this sample are mostly Hmong refugees from Laos, who also may have lower tolerance of divorce than native White Americans in California. Hmong refugees may also have lower rates of cohabitation than Whites, and cohabiting unions may be less stable than marriages. Unfortunately, the effects of religion and cohabitation cannot be examined with the current data set because information on religion and marital status is not collected by welfare agencies. Families with younger mothers had higher odds of experiencing family breakup than families with older mothers, net of the other variables in the equation. Mothers young age at time of experiment may be indicative of young age at marriage and also of recent marriage. Such marriages tend to be more fragile than marriages contracted at an older age and marriages of longer duration (Becker, Landes, and Michael 1977; Lewin 2004). Again, it is

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

236

EVALUATION REVIEW / JUNE 2005

TABLE 5: Multinomial Regression Coefficients (Standard Errors) Predicting Leaving Aid and Breakup Versus Staying in a Union and on Aid (N = 14,914)

Model 3 Off Aid Versus AFDC-U

Quarter Number of children New to aid Asian African American Latina Younger than age 25 Between ages 25 and 35 Mother employed Father employed AFDC grant per person Intercept Chi square df p 0.061** 0.259** 0.173 0.003 0.476 0.414** 0.054 0.026 0.311** 0.143* 0.607** (0.137) (0.027) (0.091) (0.140) (0.276) (0.076) (0.100) (0.094) (0.087) (0.071) (0.023)

Odds Ratio
1.063 0.772 1.188 1.003 0.621 0.661 0.948 1.026 1.364 1.153 0.545

Breakup Versus AFDC-U

0.039* (0.017) 0.071** (0.028) 0.830* (0.149) 0.884** (0.150) 0.299 (0.205) 0.763** (0.084) 0.346** (0.112) 0.050 (0.106) 0.132 (0.119) 0.307** (0.086) 0.280** (0.022) 1.452

Odds Ratio
1.040 0.931 0.436 0.413 1.348 0.466 1.413 1.051 0.876 0.735 0.756

0.488 1,477.615 22 .000

NOTE: Household income, mothers income, fathers income, and welfare grant amount were divided by 100. AFDC-U = Aid to Families With Dependent ChildrenUnemployed Parent. *p < .05. **p < .01.

not possible to test for the effect of age at marriage or marriage duration as this information is not available in the current data set. The results presented in Table 3 showed that total household income reduced the odds of breakup. The effect of household income on breakup is statistically significant, an increase in one unit of household income ($100) reduces the odds of breakup by a factor of .98. Though this effect is weak, it has theoretical significance because it suggests that overall economic stability improves family stability. Moreover, Table 4 showed that all the components of household income reduced the odds of breakup, though the effects of parents incomes are very weak. For example, fathers earnings reduce the odds of breakup, an increase in one unit of fathersincome ($100) reduces the odds of breakup by a factor of .99, and this effect is statistically significant at the .05 level. Mothers earnings also have a weak effect, reducing the odds of breakup, but this effect is not statistically significant. Though the effect of

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

237

mothers earnings is weak and statistically insignificant, this finding has theoretical importance because it does not support the hypothesis that womens earnings have a destabilizing effect on families. Grant amount (per person) reduces the odds of breakup, and the effect is statistically significant, but it is difficult to interpret this coefficient because grant amount is affected by earnings from work. The effect of parents employment on family breakup are presented in Table 5. The findings showed that fathers employment (vs. fathers unemployment) reduced the odds of breakup net of the other variables in the equation. Employed fathers had 74% the odds of breaking up than unemployed fathers. This finding is both statistically significant and substantially meaningful. The finding that long-term welfare dependence and fathers unemployment increase the odds of family instability are consistent with the theory that fathers chronic unemployment reduces the gain from marriage and increases the odds of breakup. These findings run counter to the argument that by substituting for fathers earnings, mothers earnings induce breakup. In sum, these results showed that fathers employment had a negative and statistically significant effect on breakup, suggesting that income has a stabilizing effect on families. Cases new to aid had lower odds of breakup than cases with welfare histories at Q1, suggesting that, among welfare recipients, the chronicity of poverty increases breakup. These findings support the hypothesis that long-term poverty and fathers unemployment increase family instability. The findings do not support the hypothesis that mothers earnings reduce the gain from marriage and increase breakup.

DISCUSSION Recent welfare policy directed at promoting two-parent families has focused on two different types of low-income families, with two different policy goals; First, there is the policy goal of promoting marriage and cohabitation among single mothers; and second, there is the goal of increasing the stability of low-income two-parent families. The current study contributes to the latter goal. In this study, I asked whether economic stability improves family stability among welfare recipients and found, overall, that it does. Household income reduces the odds of union dissolution, whereas fathers unemployment and chronic poverty increase these odds. This study has important implications for the politically loaded debate on the relationship between the welfare state and the retreat from marriage. The findings suggest that income from all sources increase family stability of welfare recipients

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

238

EVALUATION REVIEW / JUNE 2005

and that fathers nonemployment (or underemployment)leads to breakup, net of race, age, and number of children. Thus, this research has implications for the current dispute regarding the reauthorization of TANF. Specifically, if a policy goal is to encourage the stability of two-parent families, these findings suggest that policy efforts should focus on promoting stable employment among parents in two-parent families. These findings suggest that making welfare available to two-parent families strengthens these families during periods of economic hardship due to unemployment or underemployment. Studies have shown that being on welfare may have the indirect effect of creating dependency, as recipients become less employable because they lose work skills (see Harris [1993, 1996] for a discussion of the complex relationship of recent job experience, welfare, and employment of single mothers). The current study has shown that having a welfare history is detrimental to the stability of families. The deterioration of employability and labor market skills may be the mechanism through which this occurs. The policy implication is, again, to focus attention on increasing the job skills and market opportunities of unemployed and low-income parents and thus indirectly to strengthen two-parent families. This study also has important theoretical implications. The finding that mens unemployment leads to family instability shifts the emphasis from womens economic activity to mens. The finding that womens earnings and employment do not weaken families is inconsistent with the theory that womens economic independence reduces the gain from marriage and leads to union dissolution. Instead, these findings suggest that among welfare recipients, womens gain from marriage is reduced when the man in the household is economically inactive. The general theoretical implication is that studies of union stability should focus not only on changes in womens economic independence and the reduced gain from marriage due to womens earnings but also on fluctuations in mens earnings and the effect of mens employment (or lack thereof) on family well-being.

NOTE
1. Many studies have focused on the policy initiative of promoting marriage among single mothers as a means to reduce poverty (Thomas and Sawhill 2002; Harknett and Gennetian 2003; Wood, Rangarajan, and Deke 2003; Gennetian and Knox 2003).

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

Lewin / THE EFFECT OF ECONOMIC STABILITY

239

REFERENCES
Allison, P. D. 1982. Event history analysis. Beverly Hills, CA: Sage. Becker, G. 1973. A theory of marriage: Part I. Journal of Political Economy 81:813-46. . 1981. A treatise on the family. Cambridge MA: Harvard University Press. Becker, G., E. M. Landes, and R. T. Michael. 1977. An economic analysis of marital instability. Journal of Political Economy 85:1141-87. Bitler, M. P., J. B. Gelbach, H. W. Hoynes, and M. Zavodny. 2004. The impact of welfare reform on marriage and divorce. Demography 41:213-36. Blackburn, M. L. 2003. The effects of the welfare system on marital dissolution. Journal of Population Economics 16:477-500. Cherlin, A. 1979. Work life and marital dissolution. In Divorce and separation: Context, causes and consequences, ed. G. Levinger and O. Moss. New York: Basic Books. Conger, R. D., G. H. Elder Jr., F. O. Loren, K. J. Conger, R. L. Simons, L. B. Whitbeck, S. Huck, and J. Melby. 1990. Linking economic hardship to marital quality and instability. Journal of Marriage and the Family 52:643-56. Danziger, S., G. Jakubson, S. Schwartz, and E. Smolensky. 1982. Work and welfare as determinants of female poverty and household headship. Quarterly Journal of Economics 97:519-34. Darity, W. A., and S. L. Myers. 1984. Does welfare dependency cause female headship? The case of the Black family. Journal of Marriage and the Family 46:765-79. Draper, T. W. 1981. On the relationship between welfare and marital stability: A research note. Journal of Marriage and the Family 43:293-99. Fitzgerald, J. M., and D. D. Ribar. 2004. Welfare reform and female headship. Demography 41:189-212. Garfinkel, I., and S. S. McLanahan. 1986. Single mothers and their children: A new American dilemma. Washington, DC: Urban Institute Press. Gennetian, L. A. 2003. The long-term effects of the Minnesota Family Investment Program on marriage and divorce among two-parent families. New York: MDRC. Gennetian, L. A., and V. Knox. 2003. Staying single: The effects of welfare reform policies on marriage and cohabitation. New York: MDRC. Greenstein, T. N. 1990. Marital disruption and the employment of married women. Journal of Marriage and the Family 52:657-76. Harknett, K., and L. A. Gennetian. 2003. How an earnings supplement can affect union formation among low-income single mothers. Demography 40:451-78. Harris, K. M. 1993. Work and welfare among single mothers in poverty. American Journal of Sociology 99:317-52. . 1996. Life after welfare: Women work and repeat dependency. American Sociological Review 61:407-26. Hasenfeld, Y., A. Lewin, and M. Mitchell. 1996. LINK-UP evaluation of the 100-hour rule waiver for AFDC-U. Impact analysis revisited. Los Angeles: School of Public Policy and Social Research, University of California, Los Angeles. Hoffman, S. D., and G. J. Duncan. 1995. The effect of incomes, wages, and AFDC benefits on marital disruption. Journal of Human Resources 30:19-41. Hoynes, H. W. 1997. Does welfare play any role in female headship decisions? Journal of Public Economics 65:89-117. Lewin, A. C. 2001. Changing work ethic and welfare dependence through welfare reform: The 100-hour rule waiver experiment for AFDC-U. Evaluation Review 25:371-89.

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014

240

EVALUATION REVIEW / JUNE 2005

. 2004. Matches and mismatches: Homogamy and divorce in Israel. Manuscript. Lichter, D. T., D. K. McLaughlin, and D. C. Ribar. 1997. Welfare and the rise in female-headed families. American Journal of Sociology 103:112-43. Lillard, L. A., and L. J. Waite. 1993. A joint model of marital childbearing and marital disruption. Demography 30:653-81. Moffitt, R. 1992. Incentive effects of the U.S. welfare system: A review. Journal of Economic Literature 30:1-61. . 2000. Policy effects on family structure and fertility: Welfare benefits and female headship in the U.S. time series. AEA Papers and Proceedings 90:373-77. Moffitt, R., R. Reville, and A. E. Winkler. 1998. Beyond single mothers: Cohabitation and marriage in the AFDC program. Demography 35:259-78. Murray, C. 1984. Losing ground: American social policy, 1950-1980. New York: Basic Books. Oppenheimer, V. K. 1997. Womens employment and the gain to marriage: The specialization and trading model. Annual Review of Sociology 23:431-53. Rank, M. R. 1987. The formation and dissolution of marriages in the welfare population. Journal of Marriage and the Family 49:15-20. . 1994. Living on the edge: The realities of welfare in America. New York: Columbia University Press. Ruggles, S. 1997. The rise of divorce and separation in the United States, 1880-1990. Demography 34:455-66. South, S. J. 2001. Time-dependent effects of wives employment on marital dissolution. American Sociological Review 66:226-45. South, S. J., and G. Spitze. 1986. Determinants of divorce over the marital life course. American Sociological Review 51:583-90. Spitze, G., and S. J. South. 1985. Womens employment, time expenditure, and divorce. Journal of Family Issues 6:307-29. Sweezy, K., and J. Tienfenthaler. 1996. Do state variables affect divorce rates? Review of Social Economy 54:47-65. Thomas, A., and I. Sawhill. 2002. For richer or for poorer: Marriage as an antipoverty strategy. Journal of Policy Analysis and Management 21:587-99. Tzeng, J. M., and R. D. Mare. 1995. Labor market and socioeconomic effects on marital stability. Social Science Research 24:329-51. Weiss, Y. 1997. The formation and dissolution of families: Why marry? Who marries whom? And what happens upon divorce? In Handbook of population and family economics, ed. M. R. Rosenzweig and O. Stark, 81-123. New York: Elsevier. Winkler, A. E. 1995. Does AFDC-UP encourage two-parent families? Journal of Policy Analysis and Management 14:4-24. Wood, R. G., A. Rangarajan, and J. Deke. 2003. Marriage patterns of TANF recipients: Evidence from New Jersey. Mathematica Issue Brief 5:1-4. Yamaguchi, K. 1991. Event history analysis. Beverly Hills, CA: Sage. Zimmerman, S. L. 1994. The role of the state in family life: States AFDC payments and divorce rates in the United States. International Journal of Sociology and Social Policy 14:4-28.

Alisa C. Lewin is a lecturer in the Department of Sociology and Anthropology, University of Haifa, and is visiting at the Sloan Center for Working Families at the University of Chicago. Her research interests are marriage, divorce, poverty and welfare. She also studies program evaluation and the effect of policy on various demographic outcomes.

Downloaded from erx.sagepub.com at Alexandru Ioan Cuza on March 4, 2014