U. S. Bureau of the Census
Washington, D.C. 20233
Working Paper Series No. 46
This paper reports the results of research and analysis undertaken by Census Bureau Staff. It has undergone a more limited review than official Census Bureau publications. This report is released to inform interested parties of research and to encourage discussion.
TABLE OF CONTENTS
The Effect of Work and Welfare on Living Conditions in Single Parents Households.
Working and non-working households
Welfare, work, proverty and living conditions
Multiple regression analysis
The Effect of hardship on subsequent labor force behavior.
Discussion and conclusion
Population Working Papers
Recent changes to programs of income support for the poor have focused attention on how work requirements and incentives affect earnings and employment of welfare recipients. The predominant way of thinking of these issues, at least in broader political discourse, assumes that obtaining work or improving wages are desirable goals for welfare recipients and their families. However, recent research has begun to indicate that single parents and their families are not always better off in the labor force. This paper uses the 1991 and 1992 panels of the Survey of Income and Program Participation to examine welfare, work and well-being in a broader context. The paper finds an apparent advantage of work over welfare for most households, but not for single parent households. In addition, material hardship is found to have strong effects on subsequent labor market participation and welfare use.
Recent changes to programs of income support for the poor have focused attention on how work requirements and incentives affect earnings and employment of welfare recipients. The predominant way of thinking of these issues, at least in broader political discourse, assumes that obtaining work or improving wages are desirable goals for welfare recipients and their families. However, recent research has begun to cast doubt about this assumption.
When the choice between welfare and work (or a combination) is voluntary, as was commonly taken to be the case in past research, it was not always essential to have a full accounting of the costs of moving from welfare to work. Unmeasured costs could be inferred from decisions. Now that most welfare programs are limiting eligibility in a way that makes work mandatory, it is more important to be able to account for these costs. These include direct costs such as foregone benefits and daycare, which have become a subject of closer scrutiny. However, other costs remain difficult to measure. These include loss of time, variance of income, unreliability of daycare and increased problems in complying with program rules.
Recent research has begun to indicate that single parents and their families are not always better off in the labor force. In a prominent example of this research, Kathryn Edin and Laura Lein showed that material well-being of single mothers does not increase when they are in the workforce, even when their income is higher (1997a, b; Edin 1991). Research by Beverly (2000, Forthcoming) provides additional evidence supporting the idea that hours of work directly decrease the well-being of single-mother families. This paper uses the 1991 and 1992 panels of the Survey of Income and Program Participation to examine welfare, work and well-being in a broader context. The paper examines how presence of children, marital partners, work and welfare use affect well-being, and how well-being, in turn, affects subsequent participation in labor market activities and welfare programs.
Most recent changes in welfare policy over the last decade have been premised on the idea that work is generally beneficial for those who use it as a means of support and welfare is generally not. These changes have subsequently led to a sharpened research focus on the validity of this premise.
The benefits and drawbacks of work.
Authors who have examined the contrast between work and welfare have often found ways that the former improves the lives of those involved. Employment does usually provide higher income than most types of welfare. Employment also potentially extends the range of a person's interpersonal network and often includes people with higher status and resources. It usually expands the availability of credit and the type of housing available. The "intangibles" of work as a normatively acceptable life may be beneficial as well (Mead 1992; Wilson 1996; Wilson 1987). Finally, those who are employed may reasonably expect to have greater resources in the future, as better employment opportunities arise Corcoran and Loeb 1999).
The contrast between welfare and work does not favor work unequivocally, however. This is evident from recent research measuring costs and benefits of employment using new and different indicators of well-being. For example, child development indicators have shown that poverty, family structure and other factors interact to produce different outcomes in different situations (Clark 1983; Guo Brooks-Gunn and Harris 1996; McLanahan and Sandefur 1994). Work by Edin and Lein used indicators of material well-being to show how poor mothers use time, work, and resources from friends, relatives and others to balance a family budget (1997a, b; Edin 1991). Taken together, these studies have helped overturn the assumption that increased income and family well-being move together unproblematically when comparing families on welfare with families that work, and opened the door to understanding other mechanisms that are important to the healthy functioning of households and families.
That single mothers in the workforce may have lower material well-being despite higher income is a potentially interesting finding. However, the evidence on this point is mixed. Research by Beverly (2000, Forthcoming) shows a negative relationship. Other research has found household well being to increase with labor force participation (Bauman 1998). Identifying how work is related to well-being requires taking a closer look at effects of employment, and examining how other indicators of well-being vary with work and income.
The hidden costs of employment.
If work decreases household hardship, it is because there are costs to working. In fact, much of welfare policy prior to the 1980s was aimed at understanding these costs and looking for ways to reduce barriers for those moving from welfare into work. The primary focus of these policy (and research) concerns was loss of welfare income and loss of benefits from other support programs such as medicare and food stamps. However, work has other costs, and these have been getting increasing attention. To begin with, there are direct financial costs of working paying for child care, transportation and related expenses that have not been adequately accounted for in current measures of poverty (Citro and Michael 1995; Short et al. 1998; Fisher 1999). While it is possible to measure some of these costs, this subject remains controversial. Some authors argue that although they can sometimes be large, these expenses don't usually represent a serious burden or disincentive to work for households (Mead 1992).
Another cost experienced by working households might be the loss of time available to take care of household needs. Work takes up time otherwise available to produce other income or to reduce expenditures. For single parents, the time constraints involved in entering the labor force are especially severe. The presence of younger children, for example, significantly decreases work and increases welfare use among single women (Danziger et al. 1999; Harris 1991, 1996). Households with more than one adult present may be able to reap the benefits of employment while suffering fewer costs.
Finally, there is the possibility that working households experience less stability and predictability of living circumstances. Income is subject to fluctuation when working hours change, when daycare is unreliable, or when sickness or other problems arise. In some cases, program rules such as those forbidding collection of benefits while also earning income may create unpredicted fluctuations.
It may be possible to clarify the roles of these different mechanisms by examining well-being among working and non-working households of different types. For example, if day-care costs are a major component of the unmeasured cost of working, then the presence of children in low-income households should help explain lower levels of well-being associated with work. Most other direct costs (commuting, uniforms, cost of prepared foods) would be shared approximately equally across different types of households. If time constraints were a key factor affecting well-being, then single-parent households might be expected to find their well-being more adversely affected by work than other types of households.
Material hardship and household well-being.
A number of indicators could be used to measure household well-being. For this paper, income (and poverty status, which is closely related to income) cannot serve as a measure, because income needs to serve as a control variable when comparing well being of working and non-working households. A common alternative has been to operationalize well-being in terms of long-term outcomes such as respondent's career growth, children's behavioral problems or children's academic success. While these measures are ideal in certain respects, the linkages between these and current conditions of welfare use or employment have not been clearly established. There is some profit, therefore, in focusing on more immediate aspects of material well-being, which is the approach taken here.
Recently there have been attempts to systematize knowledge about immediate material hardships, such as inability to pay bills or lack of material possessions (Mayer and Jencks 1989; Mayer 1995; Bauman 1998; Meyers, Weissman and Garfinkel 2000; Beverly Forthcoming). Measures of material hardship often include indicators of problems that are serious in their own right, such as insufficient food or failure to go to the doctor or dentist when needed. Other indicators included along with these, such as failure to pay utility bills, seem on their face to have greater association with behavioral choices than with simple material need. However, Bauman (1998) has shown that indicators typically used to measure material hardship are related to household economic and social conditions in nearly identical ways. This implies that they tap into a the same underlying set of processes, despite apparent differences.
Material hardship is only weakly correlated with income and poverty (Beverly 2000). This is both a cause for optimism and for concern, in terms of this analysis. If income and material hardship were the same thing, then it would be impossible to find lower material well-being for working than for non-working families with similar incomes. On the other hand, a significant portion of households with substantial income report material hardship, which raises basic questions about what, in fact, is being measured by hardship. Although the answer is still far from complete, there may be two basic processes taking place. First, hardship may reflect attitudes or competence on the part of household members, as when relatively well-to-do households neglect to pay some of their bills. Second, hardship may reflect the impact of unforseen events, as with major automobile repair expenses.1
By contrast to material hardship, there has been a long history of using other indicators such as housing and neighborhood conditions as a measure of well-being. Neighborhoods have been associated with a number of individual and household problems such as dropout, unwed childbearing and weak labor force attachment (Crane 1993; Wilson 1987; Massey and Denton 1992; Rosenbaum and Popkin 1991). Neighborhood conditions would tend to be associated with longer-term economic status rather than immediate work or welfare reliance. Unlike neighborhood conditions, hardship has been shown to be a relatively short-lived condition, more linked to temporary income fluctuations than to long-term income (Cook et al. 1986; Mayer and Jencks 1989).
Although neighborhood conditions and material hardship are not the same as poverty, households that are affected by them are plainly worse-off than those that are not.2 Beyond this, neighborhood conditions and hardship might discourage employment. If material hardship were higher for working-poor households than for those on welfare, or if working-poor households were forced to live in less desirable neighborhoods, there would be a clear incentive to leave work or remain on welfare. The extent to which hardship or neighborhood conditions have an impact on employment and welfare use is unclear. Although previous work supports the idea that neighborhood conditions do affect work (Rosenbaum and Popkin 1991), there has been little or no research on the hardship-employment linkage.3
Welfare use, work and well-being
Comparing work with welfare requires consideration of how welfare is used. Static comparisons of those on welfare and those not on welfare contain many who would be in a different state at another point in time (Duncan 1984; Ellwood 1989). Those who "emerge" from welfare often re-enter the welfare system a short time later (Harris 1996). Past research using longitudinal surveys has classified welfare users into three or four groups (Duncan 1984; Bane and Ellwood 1986; Spalter-Roth et al. 1992). One group uses welfare temporarily, returning to employment or other means of support over the long term. A second group, of particular concern to those involved in fashioning policy, consists of those who remain on welfare for extended periods. Finally, there are two residual groups, those who move back and forth between welfare and work, and those who combine welfare and work simultaneously.4 Recent research on welfare and well-being has not distinguished these groups clearly. It is important to attempt these distinctions in relating welfare, work and well-being because of the differences to be expected in how they are related when welfare use is short-term, long-term or sporadic.
People who use welfare differ in many ways from those who work. If there are differences in material well-being between welfare-using and working households, it is not immediately clear whether differences in the exigencies of work or differences between individuals are to blame. While working-poor and welfare-dependent households commonly trade places, it is clear from the previous discussion that only certain segments of the welfare population commonly move between welfare and work. In the end, therefore, it is not known whether differences in material well-being represent a barrier to work for most single parent families.
As with material well-being, most factors that affect the path from welfare to employment aren't easily categorized as characteristics of the path or of the individual. These factors include lack of job skills, transportation problems, perceptions of discrimination and depression (Danziger et al. 1999). It is unlikely that any of these operate independently of pre-welfare characteristics or of experience on welfare. Disentangling the cause-and-effect nature of these barriers opens up difficult empirical problems that may never be completely solved.
It is important to establish the time ordering of events even if this does not show causality in an ultimate sense. While it would not be surprising to find that losing a job decreases material well-being, it is important to know whether declines in well-being lead to job instability even if the linkage has more to do with respondent characteristics than to situational factors.
The research presented in this paper attempts to set the experience of material well-being in a broader context than that shown in previous research. In so doing it addresses the following questions:
The balance of this paper proceeds as follows: the next section discusses the data to be used to answer these questions. The following section examines patterns of welfare receipt and work. In the next section of the paper, these patterns become the basis for exploring the relationship between welfare, work and well-being among single parent households and other households. Subsequently, there is a test of whether the negative relationship between work and well-being becomes positive when control for income is added, and whether differences between single parent households and other households remain. The last empirical section focuses on the effects of material well-being on work and welfare transitions. This is done in two ways. First there is a global examination of the impact of material well-being on whether a transition was observed in the 12 to 18 months after well-being status was determined. Second, there is examination of a discrete-time hazard model for those at risk of welfare recidivism or terminating spells of work. Finally, there is a discussion of the implications of the findings made here.
The data for this study are from the 1991 and 1992 panels of the U.S. Census Bureau's Survey of Income and Program Participation (SIPP). The regression analyses make use of the topical module on "Extended Measures of Well-being" administered in wave 6 of the 1991 panel and wave 3 of the 1992 panel. The topical module was administered to the reference person in each household (or his/her proxy), providing data on many conditions affecting the household including housing quality, neighborhood conditions and material well-being. An identical questionnaire was administered in each panel. These panels overlapped so that both sets of questionnaires were in the field at the same time winter of 1992-1993. There were 13,162 households in the 1991 panel and 18,634 in the 1992 panel for a total of 31,796 cases for analysis. Households with no members under age 55 were excluded, so that analyses on work and welfare would not be colored by the choice of retirement. This brought the number of households being analyzed to 24,147. In many analyses, missing data on hardship measures reduced the sample further to 22,397.
Respondents were asked about several types of conditions affecting their households. One set of questions dealt with neighborhood conditions. These included street noise, streets in need of repair, litter in the streets, rundown or abandoned buildings, industrial or other nonresidential uses, and smoke or fumes. Respondent were also asked if they would like to move because of these conditions. If two or more neighborhood problems were cited, the household was counted as having "neighborhood problems." This was the case for around 24 percent of the sample.
Questions on material hardship asked in the 1991 and 1992 SIPP were very similar to those used by Mayer and Jencks in their analysis of poverty and material hardship in Chicago (Mayer and Jencks 1989). Household heads were asked, "During the past 12 months, has there been a time when your household did not meet its essential expenses? By essential expenses, I mean things like the mortgage or rent payment, utility bills, or important medical care." They were then asked about instances when the household did not pay the full amount of rent or mortgage, did not pay the full amount of utility bills, had telephone service cut off due to nonpayment. Next they were asked if someone had needed to go to the doctor or hospital but did not go, and if someone had needed to see a dentist but did not go. Finally, household heads were asked to categorize the food eaten in their household as "enough of the kinds of food we want," "enough but not always the kinds we want to eat," "sometimes not enough to eat," or "often not enough to eat." The latter two answers were taken to indicate hardship. A summary measure of hardship indicated households where at least two types of hardship were reported (around 14 percent of households).
Other questions in the SIPP dealt with household possessions, quality of housing, and crime. Although most of these were analyzed along with the measures just discussed, the results did not shed additional light on work, welfare and household well-being. For the sake of simplicity, they are not presented here.
For this research, single parenthood was treated as a characteristic of a household. Any household that contained the unmarried parent of a child under age 18 during the interview period was considered a single parent household. All other households containing parents were considered married parent households. Households where no parents were identified were non-parent households.
Receipt of several different program types was considered 'welfare' for the analysis of welfare, work and well-being. These included all means tested income support programs included on the survey: AFDC, SSI, general assistance and several less common programs. Analyses predicting welfare use focused narrowly on AFDC receipt, as use of this program is more responsive to external influences. Households that received money from one or more of these programs during a month were classified as on welfare.
Work status was measured as a dichotomy at the household level. In order for a household to be considered not working in any month it had to have a non-working household reference person or contain at least one person between the ages of 25 and 64, inclusive, who did not work and was not married to the household reference person. Work could be either part-time or full-time.
Because there was only one respondent per household, individual characteristics of all household members were not used in the construction of control variables. Instead, characteristics of the household itself (household income, home ownership) and characteristics of the household reference person (age, race, sex, education) were used. Health insurance status was recoded into three categories: complete coverage for all household members in all months, coverage for at least some household members in some months, or no coverage for any household member. Household income (in logarithmic form) was measured both with and without food stamps, with little difference in results. The analyses using the former version were included here. After some experimentation with others, two measures of asset holdings were included: tenure and number of asset types. The latter was measured by adding together the total number of types of asset income received (there were 12 types, including such things as money market accounts, US government securities, dividends, rental income).
The SIPP is based on a multistage sample, rendering standard statistical tests inappropriate. In the tabulations presented below, standard errors are calculated using an adjustment for sampling based on the Taylor-series approximation, using the SAS software package. Statistical tests for regression models were adjusted to reflect a design effect of 1.4.5
Working and non-working households.
The most striking finding that emerged from the research cited above was that working mothers needed to bring in substantially more income to avoid experiencing hardship than did non-working mothers. To confirm these findings using the SIPP, a classification of households with respect to living arrangements, work and welfare is needed. There a re two major complicating factors. First, the SIPP accounts for the living conditions of all households, not just poor single-mother households. Second, the SIPP follows households over time, capturing movements between welfare and work, and changing household living arrangements.
To classify household living arrangements, the primary consideration was the degree to which households could be expected to be homogenous in terms of major expenditure patterns and time available for work and non-work activities. One major division was between households with and without children, since these entail different expenditure patterns and time expenditures (Betson 1990; Robinson and Godbey 1997). When there were children in a household, but no parent identified for the child, it was decided to count the household as one without children, thus the term "non-parent household" is used in most discussion below. A second major division was between single parent households and married parent households, since having a spouse present also makes a great difference in making time available (Robinson and Godbey 1997). Because households change their living arrangements over a year, and the SIPP keeps track of those changes, there was the necessity of deciding how to handle households who fell into two or three categories. After testing several variations, a household was classified as having a parent if an identified parent of a child in the household was present any month, and as a single parent household if a parent was not married in any month.
A similar set of decisions needed to be made on classifications of welfare and work. Twelve months of income, labor force and household data presented a daunting number of combinations. Looking at patterns of welfare receipt over a year, the distinction is clearest between those with complete reliance on work and those with complete reliance on welfare. Those who combined work and welfare or who used welfare sporadically might be expected to be in-between those who worked and those who relied on welfare along most dimensions of interest here. However, their instability in work and welfare patterns may result from events that affect household well-being such as marriage, divorce, job change, job loss, changes in health, or movement of members into and out of the household. To allow for these types of effects it was necessary to break households into subgroups. Based on some exploration of the data, the following welfare patterns were chosen: (1) those who received welfare throughout the last year, (2) those moving from work (or non-reliance on welfare) to welfare reliance during the last year, (3) those moving from welfare to work or some other means of support in the last year, and (4) those who had a spell of welfare during the last year. For some analyses categories (1), (3) and (4) were combined.
Table 1 shows the distribution of households across categories of work/welfare and parental status. The weighted total number of households in this data set is 72 million which is less than that in other estimates, due to restrictions explained earlier. There were around 35 million non-parent households, a smaller number of married-parent households and about ten million single parent households. Overall, around 85 percent of households relied exclusively on work, while over 10 percent received welfare. The remainder were not in the labor force for all or part of the year. Single parent households had much higher use of welfare than either of the other parental groups. The share that received welfare can be broken down further into the categories just described. Approximately 7 million households reported some receipt of welfare in the last year prior to interview. The largest portion of these received welfare over the course of the entire year. The balance were evenly divided between those who moved onto welfare, moved off welfare or experienced a spell.
Unfortunately, this breakdown does not correspond exactly to the categorization from previous research mentioned above. Those who moved to or from welfare in the previous year contain a mixture of types some are using welfare for temporary relief, some are cycling on and off welfare on a regular basis, and a few are at one end or another of a long welfare spell. Those who were on welfare the entire year were most likely long-term recipients, but contained some portion who were only receiving welfare for relatively short spell. To better capture information on those who "cycle" on and off welfare or combine welfare and work over a longer period, additional information on welfare use was taken from the recipiency history questions administered in the first two waves of the SIPP. All those who had a spell of welfare before the current year (or current spell if the spell extended beyond one year in the past) were classified as "previous recipients." This category crosscut all the others, so that households dependent on work, dependent on welfare moving off or on welfare or combining welfare and work each could have had a prior spell.
The last panel of table 1 shows those who had a welfare spell in years prior to last year, including those who received welfare in immediate past year and those who did not. Most of the households that received welfare before last year did not receive it in that year. An exception is single parent households, who were evenly divided. This may indicate that this group is one that is at continuing risk of being pushed onto welfare.
Welfare, work, poverty and living conditions.
Table 2 shows the percentage of households experiencing three types of material conditions and circumstances, broken down by parenthood status, labor force status and welfare use. Part A of the table takes up the first measure of material circumstances, poverty. Poverty was strongly related to welfare use, labor force status and parental status. Poverty was high among households that relied on welfare all of last year, did not contain an adult who worked or did contain a single parent around 60 percent, 57 percent and 30 percent were poor, respectively. At the other extreme, there was a poverty rate of between 3 and 6 percent among non-working households, non-parent households and households that did not receive welfare. There were few notable differences in poverty across different patterns of welfare use, although those who had spells of welfare were somewhat less likely to be poor than other welfare users. In addition there are no immediately apparent interactions between welfare use and work or parenthood status. A formal test of interactions leads to the same conclusion.6
Neighborhood problems (table 2B) followed a pattern very close to that observed with poverty status, although the differences across categories were not as pronounced. While married parent households had slightly higher levels of poverty than non-parent households, the non-parent households reported greater neighborhood problems. Non-parent households may have shown relatively high poverty rates due to the peculiarities of the current treatment of single-person households in poverty equivalence scales (Betson 1996). Alternatively, households with children may live in better neighborhoods because they invest a greater share of their income toward obtaining housing in these neighborhoods. Among those who received welfare, there was some tendency for those who received welfare the entire year or who received welfare in previous years to report a greater number of neighborhood problems. A likely explanation is that both these situations are likely to reflect long-term economic conditions, which have a greater effect on the ability to locate in a favorable neighborhood.
Poverty and neighborhood conditions were characterized by a gradient where those on welfare the entire year had the greatest problems, followed those with part-year patterns of welfare and, finally, those who did not receive welfare. This same gradient was not observed for the last measure of material circumstances, material hardship (table 2C). Households relying on welfare generally had high prevalence of material hardship. However, households that moved to welfare, and those with prior welfare receipt did not fall solidly between welfare reliant and work-reliant households. Poverty and neighborhood conditions were also characterized by a gradient where those who did not work in the last year had the greatest problems, followed those with part-year work, followed by those households with working adults the entire year. This gradient was also not observed in table 2C. Households in which adults worked part year experienced greater hardship than households in which no adults worked. As discussed earlier, greater material hardship in households with part year work and with recent moves onto welfare may be related to crises, unpredictable circumstances or instability affecting the household.
In summary, two basic patterns of well-being were observed here. Poverty and neighborhood problems were least problematic for those who worked and didn't rely on welfare, somewhat problematic for those who worked part year or had part-year welfare use, and most problematic for those who didn't work and/or relied on welfare alone. Material hardship was least problematic for those who worked, but quite high among those with evidence of instability in their economic situation. Overall, parental status did not seem to make as much difference as economic behavior in household well-being, and most interactions were not significant.
The next section presents regressions showing the relationships of welfare, work and single parenthood with material circumstances net of controls for many influences that are already known to influence well-being. While this does not isolate the effect of welfare and parenthood status as influences on material circumstances in any causative sense, it does help limit the range of explanations that might be offered for the patterns that appeared in table 2.
Multiple regression analysis.
Table 3 shows the results of logistic regression models of well-being measures on parental status, welfare status, work status and other variables. The columns labeled "A" show regressions with no additional controls. The columns labeled "B" add controls for income, health insurance coverage, age, race, education, and several other factors that might be expected to influence household hardship. The welfare-use categories used in tables 1 and 2 were collapsed somewhat in these regression models, based on trial and error. In addition, an indicator of current work or welfare status (at the time of the interview) was added to the model. This indicator is not designed to be exclusive, so a household could have positive codes for up to three welfare use indicators, with "moved onto welfare" and "welfare last year" defined to be mutually exclusive.
Shown is a set of regressions with the main effects, along with the interaction between single parenthood and working. These interactions provide a direct test of the proposition that single parents who work face greater material difficulties than those who do not, once account is taken for their higher incomes.
The regressions without controls (the models with the label "A") in table 3 look a little different for each of the three dependent variables. The material hardship regression stands out because of the large coefficient associated with moving onto welfare in the past year. Poverty and neighborhood conditions, by contrast, were not significantly related to moving onto welfare. They were, however, significantly associated with other patterns of welfare receipt last year, while material hardship was not. This contrast reinforces the idea that changes in living circumstances have immediate connection with material hardship, but delayed relationships with other indicators.
The neighborhood conditions regression stood out because of the small association with work in the previous year and with married parent status. Households in which an adult worked all 12 months didn't rate their neighborhoods significantly better than those with lower work levels. Likewise, married households did not rate their neighborhoods any better than did non-parent households.
A main point of interest in table 3 is the interaction between parental status and work. Single parent households that worked experienced much higher levels of material hardship than did working non-parent households, and both of these experienced greater hardship than married households that worked. These associations were not significant in the other two regressions (poverty and neighborhood conditions). The strong interaction of work and parental status shows that many of the "hidden costs" of work are associated with differences in household structure. Single parents who work must deal with hardship levels that are much higher than would be expected from the example of other working households. On the other hand, it seems work is not detrimental to single parent households relative to non-work in an absolute sense. The net effect of work is to reduce hardship for all three types of households in the regressions without control for income and background factors.
The second point of interest is the strong impact of having received welfare prior to last year on all three dependent variables. These households were poorer, lived in worse neighborhoods and experienced greater hardship than other households, even after taking into account their current work and welfare status. It seems there was something about this group of households that made them vulnerable to short term and long term economic setbacks (neighborhood problems and material hardship), as well as medium term economic setbacks (poverty). This point will be discussed in greater detail below.
The results of regressions with controls (labeled "B" in table 3) test whether the coefficients of work and welfare are affected by control for income, assets and other household characteristics. Once controls are introduced for material ircumstances and other household characteristics, none of the indicators of work, welfare or parental status have any effect on neighborhood conditions. By contrast, many of these indicators retain a strong effect on material hardship. With controls, the coefficient of moving onto welfare falls just below the significance cutoff (the t statistic is 1.95), but is still quantitatively large. The association with single parent status falls somewhat. The interactions between parental status and work remain unchanged. The only variable that loses its explanatory power is work. The coefficient of full year work fell from being one of the largest to near zero, while part-year work became associated with increased material hardship.
The control variables in these equations generally conformed to results found in previous research (Bauman 1998, 1999). Households with higher income had a much lower probability of experiencing difficulties either because income decreases hardship or because low hardship facilitates income generation. Having members of a household covered by health insurance was also associated with lower hardship. Other control variables included age, education, gender, and race/ethnicity of the householder, and home ownership. Those factors generally associated with higher income were also associated with fewer problems with material circumstances.
Overall, it can be said that, aside from its effect on income, work is associated with lower hardship and improved neighborhoods only among married-parent households. Working households enjoyed no better neighborhood conditions than non-working households. Single parent households that worked had marginally (and insignificantly) higher levels of hardship, non-parent households that worked had marginally (and insignificantly) lower levels of hardship than non-working households with the same parental status.
A second major conclusion is that certain patterns of welfare receipt and work had a strong associated with hardship. Households that moved onto welfare seemed to have higher hardship (although the effect did not quite reach conventional levels of significance once income was controlled). Hardship was higher in households that received welfare prior to last year. Also, households that worked only part of the year had higher levels of hardship once income is controlled.
The effect of hardship on subsequent labor force behavior.
The evidence just presented clearly supports the proposition that working single parent households have greater material hardship than would be expected from the patterns observed among other types of households. However, it is not clear what consequences this has for the households involved.
As stated earlier, material hardship might be a condition that reflects unmeasured costs of work or it might simply reflect characteristics of working households that live near the poverty line. However, the greatest concern is that single parent households are discouraged from work due to the types of hardships observed here. A first step is to examine whether households that experience hardship are affected in their subsequent participation in the workforce and the welfare system. If hardship is associated with later welfare use or detachment from the labor force, it is not necessarily because hardship plays a causal role. However it does tell us that a hardship is a marker of some process or influence that discourages work.
Because of the panel nature of the SIPP, respondents who answered questions about their material conditions in the fall and winter of 1992-1993 continued to be interviewed up to 1 « years later. Table 4 shows a set of regressions of labor market behavior and welfare use observed subsequent to the determination of hardship status. The regression models include controls for nearly all the variables included in the long regressions of table 3. The idea was to try to isolate the effects that material hardship might have above and beyond the economic conditions that existed in the household at the time that hardship was recorded. In particular, measures of prior year welfare and work behavior were controlled, so the coefficients on other variables can be interpreted roughly as their influence on change in these behaviors.
Looking across the first line of table 4, it can be seen that material hardship has an influence on most of the welfare and labor market transitions. Households that experienced hardship had a greater likelihood of leaving work and of moving onto welfare. These are the effects one would expect if hardship were involved in processes that discouraged work. The effect of hardship on beginning work is also significant, and it is positive in sign, indicating that households that experience hardship move into and out of jobs.
By contrast to hardship, the effects of poverty are significant only in the regression of moving on and off welfare. As would be expected, poor households are more likely to move onto welfare and less likely to leave. Neighborhood problems have no significant impact on any of these transitions.
The balance of the coefficients in the regression conform to expectations, generally speaking. Work transitions were affected by work in the last year, welfare transitions by welfare in the last year. Certain factors seemed to encourage work and discourage welfare use, including assets, home ownership and age. Other factors followed the pattern observed with hardship, encouraging transitions of all types except leaving welfare. These included single parent status, lack of health insurance and lack of education. It is interesting to note that these variables have elements that reflect instability of life situations (especially single parenthood, lack of health insurance), as well as elements that are more reflective of long-term household characteristics (education and single parenthood).
As an extension to the analysis in table 4, an analysis was conducted to determine whether hardship has an especially strong impact on households who recently went off of welfare or recently began work. This involved the use of a discrete-time event history model.7 The coefficients were not measured with enough statistical accuracy to make conclusions about the relative magnitude of effects. However, it provided no support for the proposition that hardship has an especially large impact on people who are moving off welfare or onto work.
The major finding of table 4 is the strong impact of material hardship on employment transitions experienced by adults. Hardship seems to be associated with patterns of living that make it difficult to establish a permanent footing in the world of work. For example, households that experience hardship may have been exposed to instability in their recent past, and continue to experience it as time moves on.
Discussion and conclusion.
This paper examined the relationship between work, welfare and material well-being to more closely examine the unmeasured costs of working in single-parent households. As with previous research, several types of households receiving welfare were identified. Single parent households were much more likely to have received welfare, and were especially likely to have been receiving welfare the entire year. They were also especially likely to have received welfare in the past, and to be currently receiving welfare conditional on past receipt.
Poverty was greatest among those with the strongest attachment to the welfare system and among those with the weakest attachment to the labor force. Neighborhood problems followed much the same pattern. Material hardship, on the other hand, was greatest in households that were in-between. Households that worked part-year had higher levels of hardship than those that did not work at all. Households that moved onto welfare in the previous year had higher levels of hardship than those that received welfare the entire year. Interestingly, households that received welfare in prior years (before the last year) also had higher levels of hardship than others.
The different behavior of material hardship as an indicator of household well-being was confirmed in regression models that added controls for income, parental status, marital status, age and other indicators of economic capacity. The effect on material hardship of having received welfare in previous years also stood out.
Work was associated with lower levels of material hardship for married households, even when all controls were in place. Not so for single parent households; their hardship increased with work, though not significantly. However, the difference between single parent households and others was significant and quite large compared with other variables in the model.
Material hardship also played a unique role in its effects on subsequent work and welfare use. As with poverty, material hardship was associated with transitions onto welfare in the months after hardship status was determined. However, hardship was associated with work transitions both starting and ending work spells while poverty was not. Neighborhood conditions had no significant effect on any of these transitions with the controls used here.
Two things stand out from these findings. First, it was confirmed that single parent households show a negative association of work with well-being, at least compared with other households. Compared to households that don't work, single parent households that work face significantly higher hardship levels than would be expected from the pattern observed among two-parent and non-parent households. This difference in the experience of hardship may be due to heterogeneity among households or it may be due to time constraints and other barriers to work. In either case, work does not mean the same thing for single-parent households as it does for most other Americans.
The mechanism by which work creates hardship is unclear, but some insight is available from this research. If health insurance were the main issue, then control for insurance status would have greatly decreased the differentials in material hardship, which did not happen. If complying with program rules were the main issue, then there would have been a much larger impact of recent welfare receipt relative to receipt in prior years. If day care were the decisive factor, households without children would get the greatest benefit from work, not married parent households, as was o bserved. None of these tests is conclusive, but they do point to the possibility that there are other major forces at work.
Second, material hardship showed itself to measure aspects of well-being not captured by more traditional measures. Aside from single parent households, the other group that stood out as being affected by material hardship was households with a history of receiving welfare in the past. Looking at this group can give better insight into the meaning of hardship.
There are at least three reasons that households that received welfare in the past would be more likely to experience material hardship in the present. First, it may be that they have basic characteristics, including human capital, that makes it difficult for them to earn sufficient money or manage it well. However, the evidence in the SIPP data is that households with past welfare use have slightly better human capital characteristics than households that depended on welfare the immediate past year.8
A second possibility is that welfare dependence creates behaviors and attitudes that make it more difficult for households to survive without welfare support. However, if welfare dependence were the issue, then households with steady welfare use over the immediate past year would have high levels of hardship. Instead, those with welfare use in years prior to the last year were the ones with high levels of hardship.
The third possibility is that previous recipients have continuing difficulty managing their economic situations. Descriptions abound in both the academic literature and in popular accounts of working-poor families facing repeated setbacks as they try to keep job and family from falling apart under the weight of financial obligations and life's catastrophes, small and large. This lends some plausibility to this explanation, though it would take additional work to substantiate.
The idea of instability can be used as an explanation of nearly all the relationships with material hardship found in this research. Single parent households that support themselves through low-wage jobs may be stretching their resources and laying themselves open to various types of disruption. Instability may explain the effect of moving onto welfare on material hardship, as well as the higher material hardship of households that worked part-year. These households have experienced disruptive events: a loss of employment, divorce, or other change in household membership. Also, the idea of instability fits well with an explanation for the influence of hardship on subsequent welfare and work behavior especially the observation that starting employment and ending employment are both positively related to past hardship.
That hardship itself seems to contribute to future difficulty in leaving welfare or holding onto work points to the possibility of a self-reinforcing pattern. Instability creates hardship and hardship creates instability. However, it may be that instability and hardship are in some ways epiphenominal. That is, both may be signs of underlying difficulties that cannot be described simply in terms of hardship or instability. In either case there seems to be a group of low-income households mostly single-parent households that stay at the juncture between welfare and work, expose themselves to financial instability and hardship, and have a difficult time leaving this situation.
The contrast between work-reliant and welfare-reliant households may be less important than the contrast between those who have adapted themselves to living with welfare and those who have, instead, developed other means of support that lift them partially or temporarily away from the welfare system. Although much of the policy focus of recent years has been on long-term welfare recipients, it may be that the group of "cyclers," that enter the workforce only to fall back again, are the key group on which to focus if we are to achieve true success in welfare reform.
1. The impact of both these processes is evident among older adults, who report little hardship despite low incomes. (Bauman 1998) claims that older people have learned to protect themselves from budget problems by better planning and by establishing predictable living conditions (but see Mirowsky and Ross 1999).
2. Recent work has shown that material hardship, like poverty, is predictive of outcomes usually viewed as negative -- such as out-of-wedlock childbearing and high school dropout (Mayer 1997, Bauman 1998).
3. However, factors that influence hardship (income instability, changes in household composition) also affect welfare use and work. Family instability has also been shown to contribute to school problems and other developmental problems in households where it occurs (Haveman et al. 1991, Wu and Martinson 1993, Tucker et al. 1998).
Some findings from studies linking family instability with child outcomes might be partially explained by family hardship. For example, Tucker and colleagues found that instability of living arrangements had effects on children's academic outcomes only in single parent families. What they might be observing is the much higher prevalence of unstable poor households with single parents (28.6 percent of households) compared with the prevalence of unstable poor households with married parents (5.3 percent). The former have much higher levels of hardship, as shown later in the paper.
4. The division between wage reliant and welfare reliant single mothers, used by Edin and Lein in their research, captures a mixture of these groups -- the greater part of "wage reliant" mothers had received welfare in the past, and the greater part of "welfare reliant" mothers had some degree of work experience (Edin and Lein 1997a).
5. Standard errors were estimated using the Surveymeans procedure in SAS and the Wesvar software package. All tabular results were estimated with the Surveymeans procedure and initial regression models were estimated with jackknife standards errors. The average design effect for the regression results was found to be 1.4. Because this estimation method was cumbersome for testing large numbers of models, the balance of the regressions were done using the design effect to adjust standard regression output from SAS.
6. I tested several models of the data in table 3, alone with similar tables constructed for some of the other well-being indicators available with these data. Interactions between work/welfare status, previous welfare status and parenthood status of households were generally not significant. By standard fit criteria, some interaction were significant, but the improvements in fit were not impressive, given the large sample size available. Of the models with interactions, the interaction between work status and single parenthood always provided the best fit.
7. The model used four-month periods as the unit of time to avoid the "seam bias" of using the SIPP for this type of analysis (for a similar approach see Blank and Ruggles 1996). Households were at risk of leaving work from the point that they entered work during the first year of observation.
8. Of households with prior-year welfare use, 32 percent had less than a high school diploma. By contrast, of households that received welfare the entire year during the immediate past year, 48 percent had less than a high school diploma. The former group also had lower percentages of blacks and Hispanics (characteristics associated with lower earnings capacity). Strangely enough, the one exception to this pattern of greater human capital of prior-year welfare users had to do with female-headed households. Almost 80 percent of prior-year welfare using households were headed by women, compared to 65 percent of households who received welfare the immediate past year.
|Bane, Mary Jo, and David Ellwood|
|1986||"Slipping into and out of Poverty: the Dynamics of Spells." Journal of Human Resources 21:1-23.|
|Bauman, Kurt J.|
|1998||Direct measures of poverty as indicators of economic need: Evidence from the Survey of Income and Program Participation. Working Paper Series. Washington, D.C.: Population Divison, U.S. Census Bureau. 43.|
|1999||"Shifting Family Definitions: The Effect of Cohabitation and Other Nonfamily Household Relationships on Measures of Poverty." Demography 36(3):315-325.|
|Betson, David M.|
|1990||"Alternative Estimates of the Cost of Children from the 1980-86 Consumer Expenditure Survey." Report to the US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation.|
|Betson, David M.|
|1996||"Is Everything Relative? The Role of Equivalence Scales in Poverty Measurement." Poverty Measurement Working Papers. Washington, DC: U.S. Census Bureau.|
|Beverly, Sondra G.|
|2000||"Material Hardship in Working-Poor Households with Children." Population Association of America, Annual Meetings. Los Angeles, March.|
|Forthcoming.||"Measures of Material Hardship: Rationale and Recommendations." Journal of Poverty.|
|Blank, Rebecca M., and Patricia Ruggles|
|1994||"Short-term recidivism among public assistance recipients." American Economic Review 84(2, May):49-53.|
|Citro, Constance F., and Robert T. Michael|
|1995||Measuring Poverty: A New Approach. Washington, D.C.: National Academy Press.|
|Cook, Fay Lomax, Christopher Jencks, Susan Mayer, Ernesto Constantino, and Susan Popkin|
|1986||Stability and Change in Economic Hardship: Chicago 1983-1985. Evanston, IL: Center for Urban Affairs and Policy Research, Northwestern University.|
|Corcoran, Mary, and Susanna Loeb|
|1999||"Welfare, Work Experience and Economic Self-sufficiency." Population Association of America, Annual Meetings. New York, March.|
|1983||Family Life and School Achievement: Why Poor Black Children Succeed or Fail. Chicago : University of Chicago Press.|
|1993||"The Epidemic Theory of Ghettos and Neighborhood Effects on Dropping Out and Teenage Childbearing." American Journal of Sociology 96(5):1226-59.|
|Danziger, Sandra, Mary Corcoran, Sheldon Danziger, Colleen Heflin, Ariel Kalil, Judith Levine, Daniel Rosen, Kristin Seefeldt, Kristine Siefert and Richard Tolman.|
|1999||"Barriers to the Employment of Welfare Recipients." IRP Discussion Paper no. 1193-99. Madison, WI: Institute for Research on Poverty.|
|Duncan, Greg J.|
|1984||Years of Poverty, Years of Plenty: The Changing Economic Fortunes of American Workers and Families. Ann Arbor, Mich.: Institute for Social Research, University of Michigan.|
|1991||"Surviving the Welfare System: How AFDC Recipients Make Ends Meet in Chicago." Social Problems, 38(4), November, 462-474.|
|Edin, Kathryn, and Laura Lein|
|1997a||Making Ends Meet: How Single Mothers Survive Welfare and Low-Wage Work. New York: Russell Sage Foudation.|
|1997b||"Work, Welfare and Single Mothers' Economic Survival Strategies." American Sociological Review 62(2, April):253-66.|
|Ellwood, David T.|
|1989||Poor Support: Poverty in the American Family. New York: Basic Books.|
|1999||An Overview of Developments Since 1995 Relating to a Possible New U.S. Poverty Measure. Poverty Measurement Working Papers. Washington, D.C.: U.S. Census Bureau.|
|Guo, Guang, Jeanne Brooks-Gunn, and Kathleen Mullan Harris|
|1996||"Parents' Labor Force Attachment and Grade Retention among Urban Black Children." Sociology of Education 69(3):217-236.|
|Harris, Kathleen Mullan|
|1991||"Teenage Mothers and Welfare Dependency: Working Off Welfare."Aja Abstract of Journal Article. Journal of Family Issues 12(4, Dec):492-518. United-States.|
|1996||"Life after Welfare: Women, Work, and Repeat Dependency." American Sociological Review 61(3, June):407-426.|
|Haveman, Robert, BarbaraWolfe, and James Spaulding|
|1991||"Childhood Events and Circumstances Influencing High School Completion." Demography. 28:133-157.|
|Massey, Douglas S., and Nancy A. Denton|
|1993||American Apartheid: Segregation and the Making of the Underclass. Cambridge, Massachusetts: Harvard University Press.|
|Mayer, Susan E.|
|1995||"A Comparison of Poverty and Living Conditions in the United States, Canada, Sweden and Germany." Poverty, Inequality and the Future of Social Policy: Western States in the New World Order, ed. Katherine McFate, Roger Lawson and William Julius Wilson.109-52. New York: Russell Sage Foundation.|
|1997||What Money Can't Buy: Family Income and Children's Life Chances. Cambridge, Massachusetts: Harvard University Press.|
|Mayer, Susan E., and Christopher Jencks|
|1989||"Poverty and the Distribution of Material Hardship." Journal of Human Resources 24(1):88-114.|
|McLanahan, Sara and Gary D. Sandefur|
|1994||Growing Up with a Single Parent: What Hurts, What Helps. Cambridge, Mass.: Harvard University Press.|
|1992||The New Politics of Poverty: The Nonworking Poor in America. New York: Basic Books.|
|Meyers, Marcia K., Joanna Weissman, and Irwin Garfinkel|
|2000||"A Comparison of New Measures of Poverty and Hardship." Population Association of America, Annual Meetings. Los Angeles, March.|
|Mirowsky, John, and Catherine E. Ross|
|1999||"Economic Hardship across the Life Course." American Sociological Review 64(4):548-569|
|Robinson, John P. and Geoffery Godbey|
|1997||Time for Life: The Surprising Ways Americans Use Their Time. University Park, Pa.: Pennsylvania State University Press|
|Rosenbaum, James E., and Susan J. Popkin|
|1991||"Blacks Who Move to Middle Class Suburbs." In The Urban Underclass, ed. Paul E. Peterson, 348-349. Washinton, DC: Brookings.|
|Short, Kathleen, Martina Shea, David Johnson and Thesia I. Garner.|
|1998||Poverty Measurement Research Using the Consumer Expenditure Survey and the Survey of Income and Program Participation (with supporting estimates from the Current Population Survey). Poverty Measurement Working Papers. Washington, D.C.: U.S. Census Bureau.|
|Spalter-Roth, Roberta M., Heidi I. Hartmann, and Linda Andrews|
|1992||Combining Work and Welfare: An Alternative Anti-poverty Strategy. Washington, DC: Institute for Women's Policy Research.|
|Tucker, C. Jack, Jonathan Marx and Larry Long,|
|1998||"Moving On: Residential Mobility and Children's School Lives." Sociology of Education. 71(2):111-29.|
|Wilson, William Julius|
|1987||The Truly Disadvantaged: The Inner City, the Underclass and Public Policy. University of Chicago Press.|
|1996||When Work Disappears: The World of the New Urban Poor. New York: Vintage Books.|
|Wu, Lawrence L. and Brian C. Martinson.|
|1993||"Family Structure and the Risk of Premarital Birth." American Sociological Review. 58(April):210-32.|
Patterns of Welfare and Work in Households, by Parental Status
|Non-parent||Married parent||Single parent||All households|
|Welfare and work||3.9||4.6||18.8||6.2|
|No work or welfare||5.4||0.9||3.1||3.4|
Households that received any welfare last year
|On welfare entire year||59.4||44.2||71.7||62.1|
|Moved onto welfare||14.4||25.7||10.7||15.1|
|Moved off welfare||16.0||19.1||11.6||14.6|
|Had spell of welfare||10.2||11.1||5.8||8.3|
Households that received any welfare in past years (before last year)
|Received welfare last year||32.2||27.0||49.5||42.4|
|No welfare last year||67.8||73.0||50.5||57.6|
by Parental Status and Welfare Experience.
|Percent in proverty|
|Parental Status||Labor Force Status||TOTAL|
|Married Parent||Single Parent||Worked full year||Worked part year||Did not work|
|No welfare last year||4.0||5.1||11.2||2.8||19.2||31.2||5.1|
|Welfare entire year||39.0||51.3||73.9||14.5||57.4||82.8||60.2|
|Moved onto welfare||15.7||26.1||46.4||10.7||47.7||64.7||30.4|
|Moved off welfare||9.7||15.3||40.5||9.8||39.3||48.2||23.2|
|Had spell of welfare||12.2||25.6||24.5||7.6||39.1||41.1||20.2|
|Received before last year||20.1||20.9||47.4||13.9||59.2||79.2||38.1|
|(Standard errors in parentheses)|
by Parental Status and Welfare Experience.
|Percent experiencing neighorhood problems|
|Parental Status||Labor Force Status||TOTAL|
|Married Parent||Single Parent||Worked full year||Worked part year||Did not work|
|No welfare last year||22.5||20.5||26.5||21.4||28.3||26.0||22.1|
|Welfare entire year||34.6||31.5||43.9||30.0||37.8||44.1||39.3|
|Moved onto welfare||29.2||22.5||40.7||26.3||32.2||44.8||30.8|
|Moved off welfare||29.1||28.7||34.6||29.0||37.2||25.0||31.1|
|Had spell of welfare||35.4||30.1||23.0||28.7||30.9||31.6||29.6|
|Received before last year||21.7||35.9||40.6||34.3||37.7||41.8||36.8|
|(Standard errors in parentheses)|
by Parental Status and Welfare Experience.
|Percent with material hardship|
|Parental Status||Labor Force Status||TOTAL|
|Married Parent||Single Parent||Worked full year||Worked part year||Did not work|
|No welfare last year||9.9||13.4||23.0||10.9||27.1||22.9||12.6|
|Welfare entire year||19.0||34.5||33.6||20.3||42.3||30.1||29.4|
|Moved onto welfare||21.7||35.9||44.1||27.6||40.7||48.3||34.8|
|Moved off welfare||14.2||24.9||34.8||22.1||29.2||30.2||25.2|
|Had spell of welfare||20.8||27.7||36.4||24.5||33.7||33.3||28.1|
|Received before last year||23.2||30.9||40.4||31.3||44.8||39.7||35.9|
|(Standard errors in parentheses)|
The Effect of Work, Welfare and Parental Status on Household
Poverty, Neighborhood Conditions and Material Well-being
|Currently receiving welfare||0.68*||0.13||0.08||0.05||0.05|
|Moved onto welfare in last year||0.33||0.11||0.00||0.58*||0.35|
|Welfare last year||0.78*||0.29*||0.12||0.20||-0.01|
|Welfare prior to last year||0.69*||0.32*||0.14||0.64*||0.33*|
|Worked 12 months||-3.15*||-0.19||0.17||-0.96*||-0.22|
|Worked some months||-1.06*||0.08||0.20||0.09||0.33*|
|Single parent household||1.46*||0.36*||0.12||0.37*||0.21|
|Single parent *worked||-0.03||-0.20||-0.22||0.57*||0.40*|
|Married parent household||0.65*||-0.01||-0.04||1.00*||0.93*|
|Married parent *worked||0.36||-0.09||-0.04||-0.53*||-0.48*|
|Household income (log)||-0.20*||-0.35*|
|Number asset types owned||-0.07*||-0.42*|
|Rented living quarters||0.34*||0.36*|
|Number of adults in hhld.||0.01||0.18*|
|Number of children in hhld.||0.01||0.20*|
|All in hhld. have health insur.||-0.06||-1.13*|
|Some in hhld. health insur.||-0.07||-0.55*|
|Education of household head||-0.04*||-0.05*|
|Male household head||-0.12*||-0.19*|
|Black household head||0.34*||0.20*|
|Hispanic household head||0.14*||-0.21*|
|Head age 15-24||0.16||0.58*|
|Head age 25-34||0.23*||0.60*|
|Head age 35-44||0.19*||0.65*|
|Head age 45-54||0.09||0.42*|
|SIPP panel 1992||-0.02||0.04||0.04||0.06||0.06|
|Null model likelihood (-2ll)||9180.5||16107.5||16107.5||12102.3||12102.3|
|Fitted model likelihood||5709.0||15879.6||15484.2||11339.0||10246.7|
|Degree of freedom||12||12||27||12||27|
|Number of observations||22397||22397||22397||22397||22397|
|(Standard errors in parentheses)|
The Effect of Poverty, Neighborhood Conditions and Material
Well-being on Work and Welfare Transitions
|Move onto welfare last yr||-0.12||0.38*||1.19*||0.80*|
|Other welfare last year||-0.02||0.03||0.94*||-0.09|
|Welfare prior to last year||-0.20||0.06||0.57*||0.16|
|Worked 12 months||-0.09||-1.02*||-0.29||-0.13|
|Worked some months||0.23||0.32||-0.0004||0.34|
|Number of asset types||-0.07||-0.08*||0.23*||0.02|
|Single parent household||0.37||0.45*||1.99*||0.18|
|Married parent household||0.58*||0.14||1.18*||0.30|
|Number of children in hhld.||-0.001||0.01||0.07||-0.03|
|Number of adults in hhld.||0.21*||0.11*||0.31*||0.15|
|All hhld have health insurance||-0.46*||-0.62*||-0.67*||0.17|
|Some in hhld have health ins||-0.29||-0.22*||0.04||0.53|
|Work disability of hhld head||0.14||0.30*||0.13||-0.40|
|Education of household head||-0.08*||-0.04*||-0.10*||-0.03|
|Male household head||0.19||0.18*||-0.32||0.44|
|Hispanic household head||0.01||0.15||0.14||-0.48|
|Black household head||-0.10||0.06||0.75*||-0.18|
|Head age 15-24||-0.99*||0.79*||1.52*||-0.09|
|Head age 25-34||-0.39||0.92*||0.70*||-0.23|
|Head age 35-44||-0.19||0.79*||0.39||-0.18|
|Head age 45-54||-0.13||0.78*||0.52||0.02|
|SIPP panel 1992||0.18||0.29*||0.75*||0.44*|
|Null model likelihood||2370.7||7185.7||2159.9||863.5|
|Fitted model likelihood||2184.6||6545.2||1611.3||758.0|
|Degrees of freedom||27||27||27||27|
|Number of observations||2561||23238||23504||1146|