The Racial Character of Welfare Reform

Race bias has many manifestations. Individuals can be biased consciously or intentionally. They can also be implicated unconsciously or unintentionally in racially biased institutional settings—what has been called “institutionalized racism.”³⁰

Policy too can be race-biased in different ways. It can explicitly invoke racial categories, making what the courts have called “invidious distinctions” based on race. It can also avoid such explicit distinctions but still lead foster racial disadvantage.³¹

In addition, when it comes to welfare as a policy regime in the federal system of the United States, it is important to account for the levels of policy making and implementation. The source of race bias can be national, state, or local. Its source may also be agency implementation more than policy-making deliberation.

It is also important to account for how different dimensions of racial bias may interact. Welfare reform could be explicitly biased in its stated policy as enacted at the federal level, or it could be biased through promulgation of a national policy that is implicitly detrimental to blacks rather than whites. It could be biased in the way states implement the program in terms of what policies they adopt for which populations. It could be biased in the ways caseworkers administer the program for individual clients. Bias could involve all these dimensions. For instance, a caseworker may, without consciously intending to discriminate, implement inherently unfair state or national policy choices such that the needs of nonwhite recipients for support are systematically dismissed to the point of discouraging recipients' self-sufficiency. These racially biased effects may be unintentional, but they end up reproducing racial disadvantage in the form of less economic opportunity and less overall well-being for nonwhites.

I review the available evidence on race bias in welfare reform based on three distinct interpretations: (1) welfare reform is not race-biased; (2) welfare reform is explicitly race-biased; and (3) welfare reform is race-neutral but structured and operating in ways that reproduce racial disadvantage. My argument is that whether the data reviewed here should be seen as supporting any of the three hypotheses depends on how well the data plus an interpretation fit together.

Evidence for the first hypothesis would include the absence of race-related categories in the stated policy, participation rates indicating equal access to the program by different racial groups (relative to need or some other pertinent standard), and other data suggesting that nonwhites are not singled out for disadvantage. Even evidence of racial disparities in treatment and outcome would be consistent with this interpretation if such disparities are interpreted as more a result of differences between clients than differential treatment by federal policy makers or state administrators.

Evidence supporting the second hypothesis might be found in the language and structure of federal welfare policy, in the way states choose options to participate in the program, and in the way state agencies and their contractors administer it. This interpretation would be credible if we found, for example, that the new welfare law specified differential treatment for white and black recipients, that states relied on race for deciding how to structure their programs, or that caseworkers and other administrators treated black recipients differently than whites. Data on differential outcomes for blacks and whites would provide supporting evidence if it could be shown that racially disparate outcomes result from race being invoked to structure welfare reform policy, state program choices, or agency administration.

Evidence for the third hypothesis would come from data indicating that the implementation of welfare reform treats black and white clients similarly but that this “equal” treatment results in blacks being systematically disadvantaged. Supporting evidence would show that welfare reform fails to ensure that its “equal” treatment of clients does not leave black families more likely to remain poor, unemployed, with lower wages, needing to cycle back into the system, and less able to improve their standard of living.

Welfare Reform as a Race-Neutral Policy Regime

This interpretation stems from the fact that welfare reform is a post–civil rights public policy, and neither the national legislation nor states' policies explicitly favor whites over blacks. Probably the best data showing that welfare reform is not race-biased can be found in participation rates. From the 1930s into the 1960s, states could discriminate against blacks seeking or receiving assistance with relative impunity. Southern states in particular did so, but they were not alone.³²

Civil rights legislation and litigation attacked such practices, and in the post–civil rights era, there is less evidence that states erect barriers designed to systematically exclude nonwhites from receiving assistance. Welfare reform has focused on reducing the rolls, and states have used the increased latitude given them in the welfare reform period to establish “diversion” programs intended to discourage families from enrolling in the TANF program.³³ Yet the available data do not provide strong evidence that indicates that welfare reform has led to blacks being systematically denied welfare at rates higher than whites.

One way to assess whether blacks are underrepresented on the welfare rolls in the welfare reform period is to compare poverty and welfare rates by race. Poverty is unevenly distributed across the U.S. population. Nonwhites are overrepresented in the poverty population; therefore, in a racially unbiased welfare system, we would expect nonwhites to comprise a relatively large number of welfare recipients. The poverty rate for the U.S. population in 2003 was 12.5 percent, up from 11.7 percent in 2001.³⁴

In 2003 people below the poverty thresholds numbered 35.9 million. For non-Hispanic whites, the poverty rate rose from 7.8 percent in 2001 to 8.2 percent in 2003. For African Americans the poverty rate was much higher, increasing in 2003 to 24.4 percent up from 22.7 percent in 2001; for Hispanics the rate increased from 21.8 percent in 2001 to 22.5 percent in 2003; and for Asians and Pacific Islanders it increased from 10.0 to 11.8 percent. While African Americans have the highest poverty rate among these groups, their numbers were among the lowest they have been since the federal government began measuring poverty. Yet nonwhites—African Americans and Hispanics in particular—are much more likely to have incomes below the poverty line; thus we should not be surprised that nonwhites comprise a larger proportion of the welfare population than of the population overall. And in fact, that has been the case for several decades.³⁵

Table 1 compares the racial breakdown of families with incomes below the poverty line to the racial composition of families receiving welfare. Forty-three percent of all families with incomes below the poverty line in 2003 were non-Hispanic whites, compared to only 32 percent of the families receiving TANF in 2002.³⁶

Only 27 percent of all poor families were black, compared to 38 percent of TANF families. Twenty-five percent of all poor families were Hispanic and the same percent of the families receiving TANF were also Hispanic. It seems, then, that blacks were by no means underrepresented in their receipt of welfare in the reform period.

Nonetheless, these data are hardly definitive. Welfare is a program largely for low-income female-headed families with children under 18. Therefore, the more relevant statistics concern the racial breakdown for these families. Comparing columns 2–4 to column 5, we see a different picture emerge: the poorer the population, the greater the proportion of nonwhites. The percentage of families with children under 18 and incomes below 100 percent of the poverty line in 2003 which were white was 37 percent, while it was 36 percent for both those below 75 percent of poverty and 50 percent of poverty. The percentage of families who were black increased from 39 to 41 percent for those below 1.00 of the poverty line compared to those below 0.75 of the poverty line, and remained at 41 percent for those families below 0.5 of the poverty line.

In addition, when we compare these data to the last column, which indicates the racial breakdown of families receiving TANF, we find that blacks are underrepresented. Only 38 percent of the families receiving welfare were black, while 39 percent of families with children were under the poverty line and 41 percent under 0.5 of the poverty line were black. Therefore, while there is reason to believe that in our post–civil rights era racial barriers to accessing welfare have been repealed, the data do not prove that black underrepresentation on welfare is a thing of the past. Instead, they suggest that blacks may still not be receiving assistance at a rate comparable to their level of need.

Yet, these data do not enable us to determine definitively whether blacks are underrepresented and, more importantly, whether systematic bias in policy and program implementation is involved. They do, however, show that each measure of poverty implicitly indicates a standard for evaluating the welfare program.³⁷

For instance, if we focus on the overall poverty rate, we imply that welfare is about assisting all poor persons. If, alternatively, we focus on the proportion of all families with children below 0.5 percent of the poverty line, then we imply a goal for the program that focuses on assisting the poorest of the poor. Which statistics are used determines what normative standard is assumed. As the data in table 1 suggest, blacks can be seen as under- or over-represented depending on which poverty standard is referenced. The decision of which statistic and standard to use is a normative issue that depends on who we decide should be the primary target population of the welfare program—poor people in general, or the only poorest families. Until we determine that, we cannot decide the issue of racial representation.

Other evidence, however, can be used to suggest that welfare programs are open to nonwhites, such as statistics indicating that poor nonwhites are more likely to receive a variety of forms of government assistance compared to whites. Table 2 presents the percentage of poor persons in 2004 living in households where at least one member received some form of government assistance. According to these data, 68 percent of the poor lived in a household where at least one member was receiving some kind of means-tested benefit from the government, and 62 percent were in households that received a means-tested benefit other than school lunches. Fifty-five percent were in households that received Medicaid, and 36 percent lived in a household where someone was receiving food stamps. Only 23 percent were in households with means-tested cash assistance, and just 18 percent of the poor in 2004 were in a household that received subsidized public housing.

A similar pattern holds for blacks, but the rates of receiving were higher, suggesting that they are not being systematically prevented from receiving assistance more than whites. The rates for poor Hispanics fell between those for blacks and the total poverty population for most means-tested benefits, but lagged behind for Food Stamps, cash means-tested benefits and subsidized housing. Rates for whites and Asians were consistently below the rates for the population overall.

Table 2 indicates that nonwhites in the post–civil rights era can receive public assistance benefits at rates on par with or even exceeding whites. One possible interpretation is that there is less racial discrimination in welfare than several decades ago. Yet, these data are also not definitive. Table 1 showed that the proportion of the poor who are black rises, the further down the income scale we go. Poor black families have lower average incomes than poor white families and are more likely to be among the poor- est of the poor; therefore, they are more likely to have a greater need for public assistance. The higher public assistance participation rates for poor blacks may be an artifact of their lower incomes compared to poor whites. This also suggests that once we account for their lower incomes, we might find that poor blacks actually participate in public assistance programs at rates below those of poor whites.

Yet the issue of whether welfare policies enable eligible black families to access assistance at rates comparable to whites requires more refined measures than poverty and participation rates. We would need to measure whether eligible blacks participate at rates comparable to eligible whites, and also whether any differences in participation rates by race are due to policy and program administration, including, for instance, discrimination by intake workers. While there have been many studies on racial differences in welfare participation, only one accounts for racial differences in eligibility rates.³⁸

None measures racial differences in application, acceptance, and rejection rates—let alone whether these rates have changed under welfare reform.³⁹ Further, blacks' participation rates that are comparable to or higher than whites may be attributable to racial bias in the broader political economy, producing their greater inability to access other, better social welfare programs or decent paying jobs. Without further study, the question of racial bias in welfare participation remains open.

Additional evidence regarding racial bias in accessing welfare can be gleaned from data on changes in the racial and ethnic composition of the welfare population. Table 3 indicates that since the mid-1990s, the number of nonwhite welfare recipients has grown. In 1992, 39 percent of welfare families were white; by 2002 only 32 percent were. The percentage of black families increased slightly, from 37 to 38 percent, while the percentage of Hispanic families increased, along with its growing share of the overall population, rising from 18 percent in 1992 to 25 percent in 2002. Thus welfare rolls have become considerably less white since TANF was implemented. In particular, the gap between the percentage of recipients who are non-Hispanic whites and blacks has widened. If Hispanics were further subdivided into white and nonwhite, it is likely that the proportion of nonwhite recipient families would exceed 50 percent.⁴⁰

We could read these data as suggesting a growing racial liberalism in social welfare policies of the post–civil rights era—but that would be a mistake. There are good reasons to interpret these data as a sign that new welfare-to-work programs have worked better for whites than nonwhites, as I discuss later.⁴¹

Systematic racial barriers to welfare assistance and related government benefits have declined, and no available data on participation suggest that barriers have been rebuilt in the reform era. Yet the data do not allow us to dismiss the possibility that blacks are underrepresented in welfare programs; nor is it clear whether such underrepresentation, if it exists, results from race bias in policy and program implementation. The data offer an array of implicit normative standards and, in the end, are simply too porous to prevent multiple interpretations. Even evidence of the growing proportion of nonwhite welfare recipients can be used to support competing interpretations. A further issue, which the data cannot even begin to address, has to do with the limitations of the welfare program. Even if all eligible families were receiving welfare benefits, we would need to assess how the program missed ineligible but needy families. Such “target inefficiency” may itself be race-related. Therefore, while there may be good reasons to believe that welfare as reformed does not perpetuate the race-biased practices of earlier periods, it will take more than available statistical data to make this question as moot as it is often treated in public policy deliberations.

Welfare Reform as an Explicitly Race-Biased Regime

The available data on welfare participation do not prove that welfare in the reform period is either race-neutral or race-biased. To find evidence on the second hypothesis—that welfare reform is overtly race-biased—we must look elsewhere.

We can start with the welfare legislation itself. As a post–civil rights era public policy, we would not expect to find evidence of a Jim Crow–era distinction between black and white recipients. Yet the law explicitly invokes nationality and thereby indirectly reinscribes racial and ethnic otherness as a way of disqualifying whole categories of persons who might need assistance. The welfare law and the immigration reform law, both passed in 1996, restrict immigrants' access to welfare. Immigrant single mothers who entered the United States after 1996 are most often ineligible not only for TANF but also for food stamps and Medicaid. Where immigrants are eligible for assistance, states' welfare programs have often in recent years reduced access by not providing crucial support services. For example, not offering translation services to non-English speakers constitutes “equal” treatment that reinforces disparities in access to benefits.⁴²

At least two factors, however, complicate the case of immigrants. First, immigrants are not citizens, and whether or not they should be accorded the same rights as citizens does not, at least in theory, necessarily have anything to do with race. Second, immigrants are not necessarily nonwhite, though increasingly they are Latino and Asian. Whether restricting immigrant access to welfare is tantamount to race bias is, then, still subject to interpretation.

At the state level, race may be more clearly seen as a salient factor in policy adoption. Elizabeth Lower-Basch notes that “to the extent that welfare recipients are perceived as predominantly of a different racial or ethnic group than the majority of the taxpaying/voting citizens of a state, lawmakers may be more ready to impose stringent conditions on welfare receipt.”⁴³

Recent studies find a strong statistical relationship between the racial composition of the welfare population and whether a state aggressively adopts get-tough reforms, including full-family sanctions, relatively short time limits, and family cap policies.⁴⁴ (The family cap denies aid for any child born to a family already receiving welfare—thereby “capping” the family benefit at its current level.) Figure 1 presents the results from one such study,⁴⁵ indicating the effect of an increase in the percentage of the state's welfare population that is black on the probability of stringent reforms. As the percentage of the welfare population that is black rises, the probability of adopting full-family sanctions increases from 54 to 97 percent, the probability of adopting a family cap rises from 5 percent to 96 percent, and the probability of adopting a time-limit shorter than five years increases from 10 to 88 percent.

In addition, nonwhites are more likely to live in states that adopt tough welfare policies.⁴⁶

Therefore, one reason that national studies have found that blacks are more likely to be sanctioned may be because they live in states with tougher policies.⁴⁷ As Figure 1 suggests, this correlation may be the result of its own circular causality; the prevalence of nonwhites on the rolls may increase the chances that a state will choose to impose an extreme form of the get-tough reforms which in turn results in blacks being more likely to be affected by tougher time limits, sanctions and family cap policies.

Yet whether the relationship between the racial composition of the welfare population and the state's choices is evidence of race bias is subject to interpretation. In one sense, that the racial composition of a state's welfare population is related to how tough a state chooses to be can itself be interpreted as a form of race bias. Yet white recipients in states where blacks comprise a relatively large proportion of recipients are subject to the same get-tough policies, making it harder to characterize states' structuring of reform policy choices as explicit bias. Perhaps a more nuanced interpretation is that the get-tough state choices are race-neutral public policies triggered by racial considerations. Even this interpretation needs further exploration, though, because, while race is correlated with state welfare reform policy choices, it may not be clear whether state policy makers have intentionally acted in a racially biased way. A more contextual understanding of the role of race bias in welfare reform may better explain the correlation.

Bureaucratic encounters between clients and caseworkers suggests additional evidence of explicit race bias. Susan Gooden has published a series of studies on racial discrimination in the administration of welfare reform.⁴⁸

Her research suggests that the differential rates at which white and nonwhite clients access education and training under welfare reform result not from coincidence or differences in client backgrounds and attitudes, but from unequal treatment by caseworkers. For example, “in two Virginia counties 41 percent of white clients reported their caseworker continued to promote education, compared to none of the black clients.”⁴⁹ Gooden also examined the degree to which caseworkers assisted clients in securing child care and transportation—key factors affecting a client's ability to get and keep a job. While she did not find differentials in clients being directed to child care solutions, she did find that caseworkers tended to make extra efforts to help white, but not black, clients solve transportation problems. Further study is needed to document whether caseworkers systematically treat black clients less favorably than whites. Gooden's work is suggestive, but also indicates that race bias might exist on some dimensions more than others.

Thus at each level of government there is evidence of explicit race bias. Yet in each case the data allow for multiple interpretations and therefore fail to provide a definitive evidentiary base that welfare reform is an explicitly race-biased policy regime.

Ostensibly Race-Neutral but Producing Racial Disadvantage

Much of the data already used to examine the first two hypotheses applies to the third as well. For instance, the growing preponderance of nonwhites on welfare reported in table 3 can be read to suggest that nonwhites needing public assistance are more vulnerable under an ostensibly neutral welfare reform regime that in fact fails to address the barriers they confront for leaving welfare. As a result, blacks on welfare might be treated equally to white welfare recipients in terms of work requirements, time limits, the imposition of sanctions, referrals for education and training, et cetera, but given their relative disadvantages in average education, readiness for employment, access to other sources of financial support, and other confounding factors, such as residential isolation and employer discrimination, “equal” treatment ends up producing more racial disadvantage.

Other data can be similarly reinterpreted. Blacks are more likely to be sanctioned off welfare than whites. Concern about racial disparity in sanction rates is heightened by other research that indicates that families forced off welfare do relatively poorly compared to families who leave for other reasons.⁵⁰

Recipients, especially blacks, who leave welfare involuntarily have higher recidivism rates.⁵¹ One possible reading of these data is that sanctions are forcing off nonwhites who are generally not as prepared to secure jobs as other recipients are, for want of education, work experience, and skills, or because of greater personal problems, need to care for young children, and likelihood of confronting social barriers, such as employer discrimination.⁵² Blacks are also more likely to leave welfare because they have exhausted their time allowance under welfare reform;⁵³ as a result, they may face serious economic jeopardy, further compounding the inequities they confront in the welfare system.

The data on sanctions and time limits might be read as indicating that blacks do not try as hard as whites to succeed in the world of work.⁵⁴

But, it is important to note that they are more likely to be working when leaving welfare and working more hours at increased earnings compared to white leavers, even though they have, on average, less education.⁵⁵

Another possible reading of the same data is that white single mothers are more likely to marry or take up living arrangements with an employed man than are black single mothers who leave welfare. This interpretation points to disparities in employment rates between white and black single, low-income men. Low-income black males are much more likely than whites to be unable to support their families for a variety of reasons, ranging from incarceration to employer discrimination.⁵⁶

Therefore, both the higher employment rate and lower marriage and cohabitation rates for black women who leave welfare may be associated with barriers to employment confronting poor black males. Welfare reform's failure to account for this connection in a society that still privileges the two-parent family contributes to its failure to better support low-income black women who may require added assistance to leave welfare successfully.

Recidivists, black or white, clearly have difficulty transitioning off welfare; between a fifth and a third of leavers return to welfare within a year. Parents who return have significantly less education and work experience as well as poorer health than parents who do not return.⁵⁷

They are also less likely to have used transitional support services, and were more likely to have been sanctioned off welfare. One possible interpretation for this finding is that the punitive approach reflected in sanctions is frequently applied to families that continue to need assistance.

Pamela Loprest found that nationwide while 22 percent of all leavers returned within a year of leaving welfare, the same was true for 32 percent of black and 24 percent of Hispanic leavers.⁵⁸

Only 13 percent of white leavers returned within a year's time.⁵⁹ More research is needed to account for racial disparities related to welfare recidivism, but these data indicate a policy that is insensitive to addressing problems that make it harder for nonwhite recipients to succeed under the “work first” regime.⁶⁰

There is evidence that employer discrimination partly explains why black welfare recipients have on average a harder time leaving welfare for work without returning. Harry Holzer and Michael Stoll note that “[m]inority welfare recipients are hired somewhat less frequently than their representation in the population of poor female-headed households would lead us to predict. Conditional on hiring welfare recipients, the relative tendencies of establishments to hire minorities (especially African Americans) appear to be less closely related to overall demand conditions or employer skill needs than to location, and possibly to discriminatory employer preferences as well.”⁶¹

Jared Bernstein and Heidi Hartmann add:

Overall, there is good reason to believe that the data on racial disparities best support my third hypothesis. The case is further buttressed once we go beyond the limits of conventional public policy research and place welfare reform in historical and social context. Doing so, we see that welfare reform has turned a blind eye to how its own race-neutrality reinscribes racial subordination.⁶³

Post–Civil Rights Policy as Malign Neglect

As Frances Fox Piven and Richard A. Cloward have noted, welfare has historically been a secondary institution calibrated to serve the primary needs of the state and the market in the United States. It was always structured to serve two potentially conflicting goals: to reinforce the work ethic so that the poor would take whatever low-paying jobs the market provided, and to offer support to low-income families to recreate the conditions for maintaining the state's political legitimacy.⁶⁴

It is important to add that welfare has also historically operated so as not to undermine the racial regime of white-dominated America.

Mothers' pensions programs were the precursor of AFDC and were developed to reduce the numbers of children being placed in orphanages because their mothers could not care for them.⁶⁵

Yet many of the reformers who championed these pensions did so in the name of safeguarding “republican motherhood” and fighting back the “race death” threatened by growing numbers of immigrant families who were then considered nonwhite.⁶⁶ In addition, mothers' pensions operated more to Americanize immigrant mothers than to assist indigenous blacks, who were largely considered undeserving.⁶⁷

AFDC was added to the Social Security Act of 1935, which was officially titled the Economic Security Act but rather quickly came to take on the name of its most popular program of old-age insurance for retirees.⁶⁸

Its name underscores the important, if troubling, role that federalism has played in the American welfare state. The Social Security Act gave rise to a two-tiered welfare state that privileges national social insurance over state public assistance programs.⁶⁹ In addition, this tiered system tends to reinforce differences in society along class, race, and gender lines, with the upper tier social insurance programs disproportionately benefiting families associated with more economically privileged white male workers, and the lower tier programs benefiting poorer female-headed families who were disproportionately nonwhite.

As Robert Lieberman has amply demonstrated, the system developed in accord with the role of federalism in our policy-making system.⁷⁰

Southern congressmen in particular lobbied hard to ensure that public assistance titles be federal programs that gave states substantial discretion in determining eligibility and setting benefits. They especially wanted to ensure that public assistance to poor families could be calibrated to the needs of the still powerful sharecropping system, which relied on impoverished black families to work the fields. With substantial discretion, southern states could ensure that these families could be moved on and off the welfare rolls with beginning and end of the planting and harvesting seasons. Public assistance could in that way take its place in the political economy of the ancien regime that arose in the apartheid of the south. As a result, the main public assistance program, AFDC, was a federal program that standardized and upgraded the limited mothers' pensions programs already in existence in many states. In addition, the top-tier social security program was originally structured to exclude coverage of those in certain occupations, such as agricultural workers and domestics, which disproportionately employed nonwhites. It thereby limited access to this privileged program in a way that favored whites. Though revised many times, both tiers of the Social Security Act still perpetuate class, race, and gender bias.

As blacks began to gain rights to assistance in the 1960s, AFDC came to be seen as a “black program,” making it more politically vulnerable and increasing calls for its retrenchment, which finally came in the 1990s.⁷¹

Nonetheless, Americans now imagine that they are done with the civil rights struggle, that the battle to defeat racial injustice has been won, and that the nation no longer need to consider race in making public policy. This is a period when it is politically incorrect for public policy to explicitly invoke race to legislate either privileges for whites or disadvantages for blacks. In this period of “race blind” public policy, racial hierarchy must be recreated by stealth. The encoded discourse of welfare reform law is larded with euphemisms of “personal responsibility” and “self sufficiency” that implicitly suggest the undeservingness of welfare mothers who just happen to be disproportionately black. Such racially neutral euphemisms not so ironically discourage the idea that black recipients are much more likely to confront societal barriers when making the transition from welfare to work.⁷² Some of the major effects of welfare reform, therefore, arise from the ways in which its euphemistic discourse tends to be obtuse to the distinctive needs of disproportionately nonwhite low-income, single mothers.

Welfare reform's work-first regime overlooks the fact that many low-income single mothers of all races and ethnicities lack what they need to make what welfare reformers call “rapid attachment” to the workforce and end a life of “welfare dependency.” Its assumption that all single mothers can leave welfare quickly, practice “personal responsibility,” and achieve “self sufficiency” by participating in the labor market is destined to fail many single mothers. Given society's overall racial inequalities—the fact that some are born in poverty, attend under-funded schools, grow up in poorer communities with fewer economic opportunities, and confront more barriers in looking for work—it just happens that the racially neutral but socially obtuse policies of welfare reform are likely to adversely affect blacks more than whites.

The concrete practices of welfare-to-work programs starkly illustrate how these programs produce racial disadvantage. Increasingly, welfare policy has focused on targeting resources to address what are euphemistically called “barriers” that prevent recipients from leaving welfare.⁷⁶

Welfare-to-work programs concentrate substantial energy on screening, diagnosing, and treating clients for various personal deficiencies so that they can take jobs and reduce their reliance on public assistance. Yet, as the research reviewed has indicated, nonwhite welfare recipients confront discrimination in hiring,⁷⁷ which is not addressed by most welfare-to-work programs.⁷⁸ At best, there are self-esteem and dress-for-success programs designed to get welfare recipients to ingratiate themselves with prospective employers. Recipients who complete these classes and still fail to secure paid employment risk being penalized and even terminated from assistance depending on their time spent on welfare.

Such programs could be interpreted as a perversion of efforts to address the racial barrier confronting welfare recipients. To the extent that they can be interpreted as attempting to overcome the racial barrier to employment, they constitute yet another instance of welfare reform blaming individuals for the structural problems of society; these programs seek to solve social problems by making victims change their behavior.

The ostensibly race-neutral discourse of welfare reform makes racial bias a reality in the treatment of both individual clients and black recipients as a group.⁷³

At the level of the individual client, even unbiased caseworkers who carry out ostensibly race-neutral policies are implicated in this self-defeating cycle of proving some recipients less personally responsible than others. Without saying so, caseworkers give welfare recipients an ultimatum: go to work or lose what little social support we offer. Given the organizational context in which they find themselves, they practice a latent race-biased form of what Herbert Simon once called “bounded rationality,” where they consider only certain factors and not others.⁷⁴ At the level of black recipients as a group, all these racial disparities operate in public discourse to demonstrate that nonwhites are less deserving. Along with caseworkers, the public is encouraged to think the unthinkable: it was wrong to assume that blacks would do as well as whites. While conservatives claim such race-neutral social policies “assume the best” about blacks—that is, that they can compete with whites—policies like welfare reform, with their false neutrality, justify “assuming the worst” about low-income blacks who rely on welfare.⁷⁵

The vicious cycle that blames black recipients as individuals and as a group also operates at the policy level. It is not just that the latent race bias of ostensibly race-neutral policies produces racial disadvantages, but, as the welfare population increasingly comes to be seen as a largely nonwhite population, the entire program itself is at risk of being marginalized to the point where program administrators cannot help black recipients even if they recognize the problem. With the darkening of the welfare rolls, welfare risks increasingly being seen as a program for those “other” people who are not conforming to the work and family rules of white middle-class society.⁷⁹

As welfare reform leads to nonwhites increasingly making up a growing proportion of the families on welfare, policy is becoming more steadily focused on recipients' behaviors. Moreover, welfare reform has intensified state discretion, such that states are more likely to impose strict behavioral modification regimes for implementing reform. Thus the most dramatic welfare reforms come to be imposed on populations that are disproportionately nonwhite, and blacks are most likely to feel the negative effects of these policies in ways that imply that they are not as able as whites to transition out of welfare. The cycle repeats itself: changes are simultaneously rationalized by and reinforce the idea that welfare is a program for blacks who do not adhere to white middle-class values.

Ultimately this cycle may keep spinning to the point that welfare reform's “success” in cracking down on the problem of “welfare dependency” becomes a rationalization for white neglect of what is seen as a black problem.⁸⁰

Whites then will have an added reason for supporting welfare reform: it highlights blacks' disproportionate reliance on welfare as a sign of black inferiority. Welfare therefore helps promote what Loic Wacquant calls “racemaking,” whereby a particular social condition like poverty or welfare reliance becomes a way of suggesting that the races are different.⁸¹ Welfare today makes issues of economic need out to be problems associated with a particular race. Yet it is the economic marginalization of low-income black families that helps to reinforce the idea of black inferiority rather than the other way around.⁸² Welfare helps invert the causal arrows and makes race seem to be a cause of economic destitution when in fact it is economic destitution that helps create the idea of race.⁸³