The real value of welfare: Why poor families do not migrate

Sanford Schram, Joe Soss

    Research output: Contribution to journalReview articlepeer-review

    6 Scopus citations
    Original languageEnglish (US)
    Pages (from-to)39-66
    Number of pages28
    JournalPolitics and Society
    Volume27
    Issue number1
    DOIs
    StatePublished - Mar 1999

    Bibliographical note

    Funding Information:
    SCHRAM SANFORD SOSS JOE 03 1999 27 1 39 66 sagemeta-type Journal Article search-text The Real Value of Welfare: Why Poor Families Do Not Migrate SANFORD SCHRAM and JOE SOSS Like the impoverished families they purport to describe, tales of welfare migration have traveled from one state to another over the past two decades, gravi- tating especially toward states offering higher benefits.' Responding to these tales, state-level policy makers have become increasingly concerned that welfare recipients might move to their jurisdictions in search of more generous benefits. Such fears of becoming a "welfare magnet" have proven to be a potent political force. Over the past decade, they have repeatedly prompted demands for greater discretion to deter recipients from other states.2 Indeed, the perceived need for greater flexibility to combat the migrant poor played a major role in the campaign that produced the landmark Personal Responsibility and Work Opportunity Rec- onciliation Act of 1996. The most controversial provision of the 1996 act ended the sixty-one-year-old federal program that structured how states provided cash assistance to single mothers with children (Aid to Families with Dependent Children, or AFDC). In place of AFDC, there is a new program (Temporary Assistance for Needy Fami- lies, or TANF) that mandates time limits and work requirements for the receipt of welfare but also gives states freedom to administer federal TANF block grants.3 As a central motive for this transfer of authority, the threat of welfare migration was addressed explicitly in several provisions of the act itself. Title 42 of the U.S. The authors' names are listed in alphabetical order; both made substantial contributions to both the writing and data analysis contained herein. The authors thank Jim Baumohl, Nathan Dietz, Monica Friar, Gary Krueger, Frances Fox Piven, Tom Vartanian, and members of the editorial board of Politics & Society for their helpful comments and suggestions. POLITICS & SOCIETY, Vol. 27 No. 1, March 1999 39-66 C 1999 Sage Publications, Inc. 39 POLITICS & SOCIETY Code, for example, was amended by the 1996 law to allow states to provide wel- fare to recent migrants according to the rules and benefit levels existing in their states of origin.4 Other changes gave states the freedom to treat families who move across state borders "differently than other families" for their first twelve months of residence (Section 404(c)).5 The considerable influence that the welfare migration thesis has come to exert in policy-making circles has not abated under the new TANF system. On the con- trary, some observers now predict that powerful migration anxieties will combine with enhanced discretion to produce a "race to the bottom" in which states cut their benefits to avoid becoming attractive targets for welfare recipients.6 In light of its pivotal role in recent and current welfare politics, we suggest that a critical examination of the welfare migration thesis is long overdue. Whereas claims of welfare migration have provoked much debate, their central premises have tended to be accepted almost without dissent. Stripped to its essen- tials, a generous reading of the thesis would have to include at least three critical assumptions: (1) The incentive assumption: welfare benefits vary sufficiently across states to provide an incentive for welfare recipients' residential decisions. (2) The salience assumption: welfare recipients attend to this benefit incentive to a degree that outweighs other factors. These two assumptions serve to define the decision-making environment that, in turn, helps to specify the set of real-life alternatives from which welfare recipients must choose. The third and final assumption, then, is a behavioral one describing how recipients choose among and act upon their induced preferences. (3) The instrumental assumption: welfare recipients make their residential decisions on the basis of a means-ends rational calculus. When combined, these premises suggest the politically potent expecta- tion that welfare recipients will move across state borders in search of more gener- ous welfare benefits.7 Yet, in recent years, researchers have produced a large number of empirical studies suggesting that the welfare poor rarely engage in welfare migration. In what follows, we offer an explanation for these findings-for why, in fact, poor families do not appear to migrate in search of higher benefits. We suggest that these findings force a reconsideration of the generally accepted assumptions about the role of welfare benefits in the migration decisions of welfare recipients. Our analysis draws on existing research regarding welfare recipients' lives to offer a critique of the first two of these three core assumptions. We argue that regardless of whether one embraces the third assumption and chooses to view welfare recipi- ents through the lens of rational actor theory (a subject of debate that we cannot resolve here), it is not possible to sustain the incentive and salience assumptions that make welfare migration appear to be a rational act. By interrogating these two assumptions, we hope to illuminate, first, why the migration thesis has proved to be so powerful and, second, why it has proved to be inaccurate. 40 SCHRAM and SOSS In what follows, we systematically pull back layers of considerations until we reach the "heart of the matter" and demonstrate how welfare benefits variation is not the important factor it is commonly made out to be. We begin with an exami- nation of the historical roots of the contemporary politics of welfare migration. This discussion locates recent debates within a longer history of local responses to impoverished outsiders and clarifies how the modern migration thesis has blended old themes with new ideas. Following this history, we provide a detailed examination of the thesis itself. We outline its key assumptions and show how these assumptions have been operationalized in ways that conveniently, but mis- leadingly, suggest strong incentives for welfare migration. Finally, we review the evidence indicating that, nonetheless, these incentives have not, in fact, given rise to welfare migration. In the second half of our paper, we question the incentive and salience assump- tions of the welfare migration thesis. First, we challenge the assumption that cash benefits define the structure of incentives for recipients. Beginning with a wider perspective on both poor people's lives and residential migration, we explore a variety of countervailing factors that are ignored by the migration thesis but that research suggests should be central to recipients' residential decisions. Examin- ing each in turn, we show how social-, economic-, and welfare-related factors pro- vide recipients with alternative considerations that undermine the importance of benefit-based incentives. Second, we argue that even if one takes the dubious step of ignoring these other factors, there are good reasons to question the most basic premise of the migration thesis-that welfare benefits vary enough across the states to provide incentives for recipients. We suggest that welfare benefit varia- tion has not stimulated welfare migration because, from recipients' perspectives, these benefits do not vary all that much in real terms. We conclude by discussing some of the ways in which migration incentives are likely to change in ironic ways under the new TANF system. We suggest that although welfare migration has previously existed more as a fear than as a fact, new uses of state-level discretion may change this state of affairs. Ironically, as policy makers use their newfound discretion to impose time limits and additional restrictions on aid, they may wind up creating the very welfare migration they have hoped to avoid. By basing public policy on questionable assumptions about poor people's behavior, policy makers may force poor families across state bor- ders in a desperate search for the means of survival. LOCAL RELIEF AND OUTSIDERS: A BRIEF HISTORY OF POOR PEOPLE AND MOVEMENT The problem that migration poses for welfare provision is an old one with deep historical roots in the local control of public assistance.8 The tradition of treating public assistance as a local matter in the United States has always reflected a vari- ety of political forces. It has served business interests by regulating local labor 41 POLITICS & SOCIETY markets through the expansion and contraction of relief.9 It has allowed white elites to maintain greater control over local race relations.10 It has served to mark the distinction between the deserving and undeserving poor. 1 And it has kept con- trol over incendiary moral issues at the local level, where they are felt most keenly, and away from the federal level, where they might prove to be politically volatile for elected officials.'2 As Joel Handler notes, "this has been the historic pattern. The more ambiguous or more deviant the perceived moral character of the poor, the more local the control of relief."'13 Along with these political forces, local control of public assistance has also flowed from a longstanding cultural commitment to the idea that helping the poor is a local obligation. This belief asserts a moral obligation to assist one's neigh- bors while suggesting that poor people from other communities are not "our responsibility."'4 Indeed, the fear that someone else's poor might become the bur- den of one's own jurisdiction has traditionally been a signature characteristic of local poor relief. This fear has been rooted not only in financial interests but also in the traditional tendency for community members to view unfamiliar poor people with a stronger degree of moral suspicion.'5 The questionable status of poor "outsiders" has been codified in public welfare laws since at least as far back as the Elizabethan Poor Law of 1601 and the English Law of Settlement in 1662.16 These laws established a tradition of organizing poor relief at the local level and affirmed that poor people were the independent respon- sibility of each home parish. This same belief in community-based obligation also prevailed in the U.S. colonies during the seventeenth and eighteenth centuries, stimulating a variety of local attempts to exclude poor "strangers." 17 In the nineteenth century, poor outsiders continued to be viewed with suspi- cion, and local communities continued to take legal action to bar them from receiving assistance. In The Mayor of the City of New York v. Miln (1837), for example, the Supreme Court upheld the constitutionality of a state law prohibiting poor people from entering New York City, declaring it a reasonable measure to guard against "the moral pestilence of paupers."'8 Following the Civil War, pen- sions for northern, but not southern, veterans once again grounded the moral standing for public assistance in the litmus test of residence.'9 During the Great Depression of the twentieth century, widespread poverty prompted a renewed interest in stemming the migration of poor people. In 1931, the California legislature instituted a three-year residency requirement for "pau- per aid." In February 1936, the city of Los Angeles "became sufficiently desperate to send 150 police officers to state border crossings to turn back indigents." This so-called bum blockade was politically popular but clearly illegal, as the Ameri- can Civil Liberties Union effectively demonstrated in short order.2" When the Social Security Act created a new set of federally administered social insurance programs in 1935, it left local control over poor relief relatively undisturbed. The nationalization of social insurance provision made local 42 SCHRAM and SOSS residence virtually irrelevant for a privileged class of beneficiaries (primarily white males with proven work histories) and thereby protected them from local efforts to limit relief and deter outsiders. By contrast, as Linda Gordon notes, pov- erty programs have continued to be administered at the state and local levels and hence have remained "far more vulnerable to political attacks, declining tax bases and interstate competition. Imagine states trying to rid themselves of elderly resi- dents by lowering Social Security old-age-pension benefits."2' Under the system created in 1935, poor people confronted a patchwork quilt of public assistance programs characterized by "length-of-residence requirements, pervasive invasion of privacy, and unregulated state discretion over eligibility conditions and the amounts of grants."22 In the 1960s, however, the wide latitude afforded to the states under this system (and the resulting efforts to deter and disci- pline welfare recipients) became a target for welfare rights activists. Legal advo- cates won a series of landmark decisions that strengthened national standards for poor relief, including King v. Smith, 392 U.S. 309 (1968); Shapiro v. Thompson, 394 U.S. 618 (1969); and Goldberg v. Kelly, 397 U.S. 254 (1970). The Shapiro decision is particularly noteworthy because it directly addressed the question of local obligations to poor outsiders. In its decision, the Supreme Court struck down forty state residency requirement laws, ruling that states that imposed a waiting period before allowing new residents to claim benefits penalized the constitu- tional right to interstate travel and thereby denied equal protection of the law.23 In the wake of these legal decisions, it might have been possible in the mid- 1970s to describe an ever-expanding trend toward national control in U.S. welfare history: from local charity early in the country's history, to the federal provision of Civil War pensions to northern soldiers, to a national system of social insurance benefits, to the stronger enforcement of equal protection and national standards for public assistance. Today, however, in the aftermath of the "new federalism" of the 1980s and the "devolution revolution" of the 1990s, such an account would seem naive and teleological. The past two decades have been marked by a rather dramatic reassertion of the doctrine of local obligation along with a resurgence of attention to poor outsiders and efforts to deter them. The two developments have gone hand in hand. The renewed emphasis on local control has raised the specter of the migrant poor, and anxieties over the migrant poor have fueled the perceived need for greater local control. Locating contemporary discussions of welfare magnet effects within this broader history is critical for an adequate understanding of why the welfare migration thesis has exerted so much influence in recent American politics. The power of this thesis derives from the fact that it blends very old commitments to local obligation and suspicions of poor outsiders with rational actor theories of human nature. Local efforts to deter poor outsiders have historically been based on images of indigent migrants as aimless drifters-paupers who presumably had been shunned by their own communities and who would stop wherever they could 43 POLITICS & SOCIETY receive some assistance. The poor people that local communities sought to deter were viewed as rootless and undesirable but not as rational actors who targeted specific communities in an effort to maximize benefits. Indeed, one of the most striking features of the modem thesis of welfare migra- tion is the extent to which it departs from traditional stereotypes of the poor. The contemporary migration thesis assumes that welfare recipients pursue rational long-term strategies, and although it attributes suspect goals to recipients, it pre- sumes that they make the substantial efforts needed to achieve them. Welfare recipients are purported to absorb the immediate costs of moving in order to "invest" in a future return of higher welfare benefits. In this regard, the actors por- trayed by the migration thesis contradict old stereotypes of the welfare poor as passive, unmotivated, and lazy.24 They also contradict traditional "culture-of- poverty" arguments suggesting that welfare recipients remain poor precisely because they are unable to defer gratification and engage in long-term rational planning.25 Building on neoclassical economic theories of human behavior, the welfare migration thesis presumes that recipients are purposive actors who behave in a way that maximizes individual or family utility. Regarding residential decisions, this assumption is broadly taken to mean that poor family heads survey their avail- able options, rank order these options according to the benefits each would pro- vide, and make residential choices in a manner that maximizes family benefits. In this manner, the welfare migration thesis transforms the aimless, short-sighted indigent into a more strategic actor-a homo economicus who will take advantage of interstate differences in economic benefits to achieve the highest possible income. The difference is critical because it is precisely this attribution of ration- ality that leads to the expectation that relatively generous benefits will attract the welfare poor. WELFARE MIGRATION: ASSUMPTIONS, MEASUREMENT, AND EVIDENCE The assumption that welfare recipients behave as rational actors, however, is only a starting point for the migration thesis. It does not yield any behavioral pre- dictions until one ties it to a particular calculus of choice based on a set of concrete considerations. Consequently, the welfare migration thesis depends on two addi- tional assumptions regarding the structure of incentives that recipients confront. For the thesis to make sense, one must assume that (1) welfare benefits define the structure of incentives confronting recipients, and (2) these benefits vary across states in ways that present recipients with meaningful gains and losses. In most research on welfare migration, the first of these two assumptions has been accepted without discussion. Regarding the second premise, the vast major- ity of welfare migration studies have either concluded or assumed that benefit variation is substantial enough to induce migration. Paul Peterson and Mark Rom, 44 SCHRAM and SOSS for instance, argue that benefit variation under AFDC was quite wide and that it remained stable for forty years, right into the late 1980s.26 Others, such as Thomas Dye, have agreed with this assessment.27 These and most other observers suggest that variation in AFDC benefits persisted through the 1970s, 1980s, and 1990s, even as the real value of welfare benefits declined.28 To their credit, most contemporary studies of welfare migration have taken some steps to guard against the possibility of overestimating benefit variation. For example, in an important attempt to assess the size of benefit variation in real terms, Robert Albritton used per capita income to adjust for differences in the cost of living in each state.29 Although this adjustment accounted for some of the inter- state benefit differences, Albritton still found considerable variation. Since Albritton's analysis, better measures of cost of living have become available.30 Because these measures indicate that the cost of living varies less across the states than one would assume based on per capita income, they suggest that welfare benefits probably varied even more than Albritton concluded. At present, the study of welfare migration continues to be marked by widespread agreement that substantial benefit variation remains after one accounts for state-level costs of living.31 A second way in which researchers have guarded against overstating benefit variation is by including additional benefits available to recipients in each state, primarily Food Stamps. Since 1972, Food Stamps has been a nationwide program fully funded by the federal government and adjusted for yearly increases in the cost of living. Because these benefits are based on a family's total available income (including cash benefits, such as AFDC), recipients receive lower Food Stamp payments in states that provide higher AFDC benefits.32 As a result, com- bined AFDC-Food Stamp benefits vary much less than AFDC benefits alone. Even after this adjustment, however, researchers have tended to conclude that combined AFDC-Food Stamp benefits vary more than the cost of living and hence vary across the states enough to induce welfare migration.33 A third and final way in which researchers have adjusted their calculations is by excluding Hawaii and Alaska and limiting their analyses to the continental forty-eight states and Washington, D.C. For years, Hawaii and Alaska have had distinctively high costs of living that have been matched with relatively high bene- fits. Moreover, given their geographical remoteness, neither of these states have proven to be likely targets for welfare migration. Removing these states from analysis, however, does not greatly reduce the amount of variation found for either AFDC alone or for combined benefits (though it does weaken the correlation between per capita income and benefit levels).34 Thus, welfare migration studies have tended to display a near-total consensus that welfare benefits vary substantially across the states, and this variation remains after applying appropriate statistical controls. To demonstrate how this conclusion is typically supported, Table 1 presents measures of benefit variation across the states for 1993 and 1996. This table shows the mean, standard 45 POLITICS & SOCIETY Table 1 Interstate Variation in Welfare Benefits and Cost of Living, 1993 and 1996 (in dollars) 1993 1996 Coefficient of Coefficient of Mean SD Variation Mean SD Variation AFDC benefits 374.04 126.25 .473 382.45 128.15 .335 Combined AFDC and Food Stamp benefits 617.90 95.64 .155 682.10 111.44 .163 The Friar-Leonard COLA (cost-of-living index) .9774 .0698 .071 .9822 .0653 .067 Combined AFDC and Food Stamp benefits adjusted for the Friar-Leonard COLA 630.53 75.28 .119 692.32 90.27 .130 Source: AFDC and Food Stamps-U.S. House of Representatives, Committee on Ways and Means, Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means, "The 1996 Green Book" (Washington, DC: Government Printing Office, 1993, 1996). Cost- of-Living Index-Herman Leonard and Monica Friar, State Cost f Living Indexes (Cambridge, MA: Kennedy School of Government, Harvard University, 1996). Note: Table entries are based on the maximum monthly welfare benefit for a family of three in each state. AFDC = Aid to Families with Dependent Children. deviation, and coefficient of variation for the value of AFDC benefits in the United States (based on state monthly maximums for a family of three and exclud- ing Alaska and Hawaii).3" The first line presents these descriptive statistics for nominal AFDC benefits; the second line presents them for AFDC benefits com- bined with Food Stamps. The third line of Table 1 presents descriptive statistics for the Friar-Leonard cost-of-living index, and the last line presents them for com- bined AFDC and Food Stamp benefits adjusted for state cost-of-living differences. The evidence in the first row of the table suggests that AFDC benefit variation is substantial. The standard deviation for the nominal value of benefits is $126.25 for 1993 and $128.15 for 1996. A comparison of this row with the one below it, however, shows that combined benefits vary far less than AFDC benefits alone. In 1993, the coefficient of variation for AFDC alone was .473, but it was only .155 for combined benefits. Likewise, this coefficient was .335 for AFDC in 1996 but was only .163 for combined benefits. Still, even the data for combined benefits could be viewed as a confirmation of the assumption that there is a fair amount of benefit variation across states. In 1996, for instance, a family of three moving to a state one standard deviation higher in combined benefits would have brought in roughly $1,300 in additional annual benefits. The potential gains of migration fall a bit more if we adjust for interstate differ- ences in the cost of living, using an index created by Herman Leonard and Monica Friar.36 However, even after this adjustment, the coefficient of variation for 1993 is still .119, and for 1996 it is .130. Moreover, if we compare the coefficient of 46 SCHRAM and SOSS variation for combined benefits to the coefficient of variation for the cost-of- living index, we find that the former is consistently larger (.155 vs. .071 in 1993; .163 vs. .067 in 1996). This last comparison, more than any other, appears to clinch the case for benefit variation. Peterson and Rom, as well as others, have emphasized that this comparison suggests that benefits vary more than twice as much as costs of living.37 They and others therefore conclude that even after com- bining Food Stamps and AFDC, welfare benefits vary more across the states than do the costs welfare recipients would incur by migrating. Based on these sorts of comparisons, scholars and policy makers have gener- ally accepted the assertion that rational welfare recipients have clear incentives to move to states with more generous benefits. The resulting predictions of welfare migration, however, have not fared so well when put to the test in empirical research. Indeed, the scholarship on welfare migration offers a classic case of an elegant theory running into increasingly uncooperative evidence. Until recently, empirical research has appeared to offer mixed answers regard- ing the question of whether welfare recipients migrate to states with higher bene- fits. Successive waves of studies served to shift the consensus among analysts. The earliest studies were based on aggregate data and tended to conclude that wel- fare migration did not exist to any great degree.38 A second wave of studies, how- ever, offered reasons to suspect that welfare migration might exist after all.39 Writ- ing in 1992, at the tail end of this wave, Robert Moffitt concluded a review of the welfare incentive literature by stating that the research on migration was "sugges- tive but inconclusive."40 Over the past five years, however, a third wave of studies has illuminated a number of weaknesses in earlier measurement techniques and has consistently pointed to the conclusion that welfare migration is largely nonexistent.41 In a typi- cal summary of this last wave of research, the authors of one recent study conclude that "the annual level of welfare migration for the 1985-90 period, and probably for other periods, is quite small . . . [and] the migration routes of poor single moth- ers with children are not associated with higher welfare benefits."42 This study and others in the most recent wave of scholarship do not support the claim that even moderate numbers of recipients move out of state in order to obtain higher bene- fits. Indeed, poor single mothers, the primary group of adults eligible for welfare, do not tend to move from one community to another with great frequency; when they do, more than 90 percent of their moves are within state.43 Thus far, few researchers have tried to explain why recipients' behaviors do not follow the predictions of the theory. Instead, the lack of evidence for migration has tended to be accepted without any corresponding attempt to figure out where the theory went wrong. It is not difficult to see why. The claim that families living in poverty might act instrumentally to enhance their income certainly has some claim to validity on its face, and the evidence has strongly suggested that real material gains could be achieved through interstate moves. Indeed, despite the 47 POLITICS & SOCIETY lack of evidence that welfare migration actually occurs, the conventional wisdom at present continues to hold that the benefit variation needed to produce this behavior is an empirical reality. To many, this conclusion has taken on the com- fortable status of an objective fact. In the remainder of this article, however, we argue that the extent and salience of welfare benefit variation are actually highly contestable and depend heavily on the particular social standpoint adopted by welfare researchers. WELFARE VARIATION: VIEWED FROM ABOVE AND BELOW The descriptions of welfare migration incentives found in most existing research, and reenacted in the preceding section, reflect what might be termed a top-down perspective. 44That is, researchers have sought to describe and measure migration incentives without reflecting on the actual conditions of recipients' lives and how these conditions actually appear to recipients. The top-down per- spective employed in most research in this area implicitly adopts the standpoint of the welfare policy maker. The assessment of incentives is generally based on information that can be readily obtained without leaving the offices of legislators, administrators, and advising social scientists. Little or no attempt has been made to understand how poor people actually make residential decisions-what factors matter to them and how they reason and feel their way through these factors.45 The top-down perspective is readily apparent in the ways welfare analysts have calculated benefit variation. For example, the nominal value of cash assistance can be used as a measure of benefit variation without any knowledge of how recipients actually spend their money. The use of combined AFDC and Food Stamp benefits to measure recipient income does not require any attempt to find out the actual sources of income that allow recipients to make ends meet. Simi- larly, the use of generic cost-of-living differences to recalculate benefits frees researchers from the need to ask what sorts of expenditures actually determine the cost of living in a given state for a family receiving welfare. These and other maneuvers that have been undertaken to get an accurate portrait of benefit varia- tion have not required researchers to ask what value welfare benefits have in real terms for recipients. The question of interstate variation appears more complex if one considers the life conditions and concerns of recipients-an approach to research that might be termed a bottom-up perspective.46 Researchers adopting this perspective assume that to understand migration decisions, one must begin with the considerations that are most salient to the actors involved. Put differently, bottom-up analyses are based on the particular needs and concerns that provide a frame of reference for poor people.47 In ethnographic and interview studies, researchers adopting this perspective directly engage recipients in order to illuminate social processes,48 explain the 48 SCHRAM and SOSS tactics and decisions of welfare recipients,49 and provide direct evidence needed to assess the effects of welfare policies and the assumptions made in welfare research.S These sorts of studies are readily accessible to quantitative researchers and provide them with materials that can be used to guide research design, meas- urement, model specification, and the interpretation of results. In this manner, quantitative researchers attempting to explain recipients' decisions and actions from a bottom-up perspective can ground their studies in the considerations that community-based research suggests are most likely to matter to recipients. The extent to which recipients focus on state benefit levels when making their residential decisions can and should be addressed as an empirical question. Thus far, however, most researchers have foregone observation in favor of the assump- tion that poor family heads either limit their economic considerations to welfare benefits or at least give priority to these benefits in their decision-making calcu- lus. This assumption makes it possible to dramatically simplify a complex struc- ture of incentives. Not only are social and psychological influences discarded, but so too are economic incentives that lie beyond the dollar amounts of welfare bene- fits. In the analysis that follows, we suggest that the failed predictions of the wel- fare migration thesis can be explained by a combination of two weaknesses in the top-down perspective: the failure to consider countervailing pressures in recipi- ents' lives and the failure to assess welfare benefit incentives in terms of their real value for recipients. Our discussion is structured to make its way toward the relatively narrow scope of considerations included in most analyses of welfare migration. We begin with a relatively wide perspective on migration decisions, one that allows for a diversity of social and psychological factors. After establishing the importance of these factors, we ask what incentives confront recipients if we artificially narrow the scope of considerations to economic concerns. Next, we restrict our purview even further by considering only welfare-related factors. At each of these stages, we find countervailing incentives that help to explain why poor families do not migrate in search of higher benefits. Finally, we demonstrate that even if one sticks to a generous treatment of the migration thesis, by focusing solely on varia- tion in welfare benefits and costs of living, a bottom-up perspective still offers rea- sons to doubt those who claim that there are significant welfare migration incentives. To construct our argument, we rely on community-based studies of the welfare poor, existing scholarship on migration, and an intriguing survey-based study of welfare recipients who migrated across state borders. In a major exception to the prevailing approach, Paul Voss, Thomas Corbett, and Richard Randell argue that "if you want to know why the welfare poor migrate, then ask them."51 Voss, Cor- bett, and Randell designed their survey to maximize the likelihood of identifying welfare-motivated migrants. They drew their sample solely from a subpopulation of poor family heads who applied for welfare benefits in Wisconsin shortly after 49 POLITICS & SOCIETY moving into the state (a group that, as we have already noted, represents a very small minority of welfare recipients). In what follows, we combine evidence from this survey with other community-based studies, making it possible to clarify the relative place of welfare benefits among the incentives confronted by poor families. Social and Psychological Factors The solitary focus on benefit levels in research on welfare migration implicitly assumes that the allure of higher benefits is not balanced against a variety of noneconomic considerations. This assumption contrasts sharply with the findings of research on how families generally make their residential decisions.52 At a minimum, people contemplating a move might be expected to consider relation- ships with family and friends, their own safety and security of well-being, social support in meeting daily needs, and loyal attachments to "home" places. Scholar- ship on migration suggests that these sorts of factors are critical for an adequate understanding of residential movement. As Patrick Jobes, William Stinner, and John Wardell conclude in their edited volume on "noneconomic migration in America," Migration undoubtedly always has been a complex process, motivated by such diverse fac- tors as spiritual treks, going back home, adventure and employment.... [S]ocial founda- tions of migration continue to be instrumental in motivating people either to move or to stay. Despite the prevalent metaphorical acceptance of an economic model as the explana- tion governing how people behave, the analyses of migration presented here indicate that noneconomic factors continue to help determine why, when, where, and who moves.53 It is, of course, reasonable for researchers to ask whether the economic depri- vation experienced by welfare recipients might make them more likely than other people to focus narrowly on income considerations. But in response to this ques- tion, existing scholarship on migration strongly suggests that to ignore noneco- nomic factors would be unwise. To begin with, most heads of recipient house- holds are women, and research suggests that women are even more likely than men to make their residential decisions based on noneconomic quality-of-life considerations.54 More to the point, however, research on migration clearly indi- cates that social and psychological factors play a critical role in the residential decisions of poor people. In two major works of ethnographic research, Carol Stack has provided detailed evidence of how the major life decisions of poor African Americans are governed by bonds of kinship.55 In the more recent of these two studies, Call to Home, Stack directly addresses the question of migration, providing eloquent descriptions of how powerful feelings of home bind families to socially and emo- tionally significant places. Such feelings are powerful enough to lead poor people to move their families to rural counties where economic conditions are among the so SCHRAM and SOSS worst in the nation. More generally, Carol Stack and John Cromartie report that "longitudinal analysis has uncovered the importance of kinship networks and cyclical migration in many migration systems, especially among the poor."56 Other studies suggest that people with lower socioeconomic status tend to engage in more restricted and localized residential searches, sticking close to home when contemplating a move.57 Is it possible that, relative to other poor people, welfare recipients are less likely to be influenced by these sorts of social and psychological factors? Here, it is instructive to turn to the survey results obtained by Voss, Corbett, and Randell in their study of recently arrived welfare recipients in Wisconsin. Using open-ended survey questions, the authors find a strong preponderance of noneconomic over economic motives in the migration decision-making process.... By far the most important single reason given [for moving to Wisconsin] ... was the importance of relatives.... [T]he single most important factor for the majority of these migrants was their attachment to their friends and family.58 Closed-ended items in the survey corroborate this conclusion and underscore the extent to which kinship considerations outweigh economic and welfare-based incentives. Voss, Corbett, and Randell find that being too far from family and friends was the only factor cited by more than half of the respondents as being a consideration in their decision to move.... Somewhat surprisingly, low wages, the accessibility and generosity of welfare, and a general concem about the overall local cost of living did not play an important role in the decision to relocate.59 Beyond the agreement between closed- and open-ended items, our faith in these findings is also bolstered by their agreement with results from one of the few other studies to address the motives of migrating welfare recipients. Based on interviews with 152 Pennsylvania welfare recipients, Gordon De Jong and Zafar Ahmad conclude that "the single most important reason welfare clients gave for migration was the presence of relatives or friends in the community."6o Thus, one reason why we suspect that poor families do not behave in the man- ner predicted by the welfare migration thesis is that bonds of kinship and connec- tions to home serve as countervailing forces. The influence of these considera- tions derives as much from practical considerations as from emotional attachments. Family and friends are generally important factors in the residential decisions of people at all socioeconomic levels. For welfare recipients, they also provide sorely needed assistance in raising children and serve as critical sources of many types of support that shape the overall quality of life. From a bottom-up perspective, then, it is little wonder that "the single most important factor that influences [recipients' migration decisions] is the presence of family and friends"-not welfare benefit levels.6' 51 POLITICS & SOCIETY Economic Factors Even if one takes the questionable step of assuming away social attachments and focusing narrowly on economic considerations, it still would make little sense to limit one's attention to welfare benefit levels. By doing so, the welfare migra- tion thesis ignores the variety of costs associated with neighborhoods that provide inadequate services and jobs, the short-term transaction costs of moving across state lines, the economic benefits of the superior job opportunities offered by some locales, and the significant economic contributions that recipients tend to receive from extended kin networks. It is precisely these omissions that make it possible to hypothesize that when welfare benefits are higher in a nearby state, recipients will have an economic incentive to move there. The need to consider economic factors beyond the welfare check and Food Stamps is amply demonstrated by research that has asked how well public assis- tance actually covers the expenses of poor families. Because of declines in the real-dollar value of cash benefits, combined benefits now cover less than two- thirds of the average recipient family's expenditures.62 Research by Kathryn Edin and Laura Lein has persuasively demonstrated that recipients are only able to make ends meet each month by drawing significant amounts of additional income from kin and (frequently unreported) jobs.63 The growing gap between what recipients spend each month and what they get from government means that there is a large amount of "slippage" in studies that measure the economic allure of a state by its welfare benefits. To the extent that welfare migration forces recipients to give up the informal networks of support they enjoy in their home communities,64 higher benefits in a neighboring state may actually represent a lower overall income. Insofar as recipients cannot be sure what types of informal support might lie across the state border, welfare benefits are not likely to be a clear signal to a better standard of living. Thus, from a bottom-up perspective, it is clear that welfare benefits are only part of the income mix and that they must be evaluated within a broader context of economic incentives. In addition to overlooking the economic centrality of social networks, the wel- fare migration thesis also embraces a pessimistic view of welfare recipients' eco- nomic outlooks that is, in fact, quite misleading. In considering the economic incentives that confront unemployed welfare recipients, why assume that it is welfare rather than employment that offers the incentive? After all, research strongly sug- gests that most welfare recipients hope to leave the welfare rolls forjobs that will allow them to support their families.65 On this point, we can once again return to the survey of migrant welfare recipients conducted by Voss, Corbett, and Randell.66 If one would expect to find any group of poor people that places greater impor- tance on welfare benefits than job opportunities, it would almost certainly be this 52 SCHRAM and SOSS study's sample of recipients who moved to Wisconsin and immediately applied for welfare. In fact, Voss, Corbett, and Randell report that 27.9 percent of their respondents listed inadequate welfare benefit levels among the reasons why they left their states of origin. By comparison, "a perceived lack of good employment opportunities was a factor in the decision to move for almost half of the respon- dents." Turning to the question of why respondents chose to make Wisconsin their destination, the authors find that while 28.2 percent listed the fact that welfare was easier to get, 44.4 percent listed the fact that there were more jobs to choose from. Similarly, while 22.6 percent listed higher welfare benefits in Wisconsin, 46.0 percent listed higher wages in Wisconsin. (To put both of these economic factors in perspective, relative to the social factors discussed in the preceding section, consider that 65.6 percent of the respondents listed the fact that Wisconsin was a better place to raise children and 64.5 percent listed the presence of family and friends in Wisconsin.) Although the evidence does not indicate that welfare benefits were irrelevant for these recipients, it clearly suggests the greater importance of jobs among eco- nomic motives for moving. Given this study's focus on migrants who applied for welfare, one would hardly expect to find a larger role for welfare benefit levels in a more representative sample of welfare recipients. Thus, even if one focuses solely on economic considerations, evidence regarding welfare recipients' lives and pri- orities suggests good reasons to be skeptical of those who focus too narrowly on the incentives created by welfare benefits. Because public assistance rarely covers the cost of necessities, the reality is that welfare recipients can expect to go broke before the end of the month in every state. Under these conditions, it makes little sense to assume that there are no countervailing incentive effects associated with the close kin and off-the-books employment that allow most recipients to cover the costs of necessities.67 It seems equally unwise to assume that recipients are more interested in staying in this situation than in finding a job that offers a living wage. Welfare-Related Factors What if one narrows the scope of considerations even further, limiting them to the incentives created by state welfare policies? Even in this case, we would argue that the welfare migration thesis errs in assuming that the incentives created by welfare policies flow solely from benefit levels. Interviews with welfare recipi- ents suggest that administrative procedures and requirements are highly salient among the criteria that recipients use to evaluate the attractiveness of a welfare program.68 Thus, from a bottom-up perspective, we must ask whether the alleg- edly clear signal created by benefit variation is made incomprehensible by varia- tion in other salient aspects of state welfare policies. Despite a tide of discussion about how the 1996 welfare law has fostered procedural experimentation in the 53 POLITICS & SOCIETY states, surprisingly little consideration has been given to how interstate differ- ences in administration might influence the likelihood of migration. Significant procedural differences across the states can be traced back to long before thepassage of the 1996 law. From 1962 to 1996, Section 1115 of the Social Security Act authorized the secretary of Health and Human Services to waive fed- eral requirements in order to enable a state to initiate an experimental, pilot, or demonstration project that could help better realize the objectives of AFDC.69 By mid-February 1996, all but ten states (Alaska, Idaho, Kansas, Kentucky, Maine, Nevada, New Hampshire, New Mexico, Rhode Island, Tennessee) and the Dis- trict of Columbia had approval to test departures from specified provisions of AFDC.70 These waivers have combined with new forms of state discretion under the TANF system to produce a complex and rapidly changing landscape of welfare provision. For recipients, this environment has the potential to foster a paralyzing uncertainty, especially regarding decisions to move to a new state with an unfa- miliar welfare program. Eligibility requirements have become so complex and now vary so much that recipients cannot be sure that they will receive any benefits in a neighboring state. Indeed, many states have exercised their new freedom to set up different eligibility standards for people who have lived in the state for less than twelve months. Likewise, the proliferation of new participation requirements, oversight mechanisms, and rewards and punishments all make it more difficult for a recipient to predict the treatment he or she might receive in a new state. States also differ significantly in how they calculate assets and income, grant exemptions from burdensome requirements, and determine access to benefits like housing vouchers. Beyond these barriers, recipients may also find that changes in broader rules make their families eligible in one state but not another. For example, twenty-one states have instituted "family caps," which deny additional aid for children born while the mother is receiving public assistance. In addition, states vary in how long they will provide assistance to two-parent families who are needy because their principal wage earner is unemployed. After September 30, 1998, states are free to end such aid altogether-a provision of the new law that returns states to the flexibility they had before the Family Support Act of 1988 required them to offer assistance to two-parent families for at least six months in a year. As of Feb- ruary 1996, twelve states already had time limits on this support: Arizona, Arkan- sas, Colorado, Florida, Georgia, Nevada, North Dakota, South Dakota, Texas, Utah, Virginia, and Wyoming. These states are likely candidates to think about terminating assistance to two-parent families in 1998. Recipients may also find that they are ineligible in a nearby state because of individual characteristics. For example, the 1996 law requires states to deny aid to parents if they are under eighteen unless they live in an "adult supervised setting" 54 SCHRAM and SOSS but also gives states latitude in determining exemptions. Consequently, young parents are likely to be eligible in some states but not others. Under the new sys- tem, states can also now choose to deny aid to persons convicted of drug felonies after August 22, 1996. In 1997, thirty-three states said they would do so. Eight states, including New York, also exercised their option to require drug testing. Diversity and restrictiveness in state welfare programs have also been enhanced by the proliferation of new conditions for the continued receipt of aid. In addition to the well-known workfare requirements, states used waivers under the old AFDC system to add conditions such as learnfare (which makes aid contin- gent on children's school attendance) as well as other "fares" that imposed addi- tional constraints. Under the new system, states can specify all of these conditions and more. For example, states can now require recipients to sign "Individual Responsibility Plans" explaining how they will make the transition from welfare to work and penalize recipients for not fulfilling provisions of the plans. As time goes on, recipients will also increasingly confront restrictions arising from the 1996 law's emphasis on time limits for the receipt of welfare; these too will vary across states. Although states cannot spend federal TANF funds on any recipient who has participated for more than two years without engaging in "work-related activities," they are free to set limits that are stricter than the two- year mark. Twenty-seven have done so, with thirteen requiring immediate work as a condition of aid. Similarly, while states cannot use federal TANF funds to assist any family with a recipient who has exceeded the five-year lifetime limit, states are once again allowed to specify stricter time limits. Twenty-one have done so, with twelve specifying a two-year lifetime limit. In fifteen states, recipients would find both shorter lifetime limits and stricter work requirements than are mandated by the federal law. In the area of work requirements, states vary tremendously in what they count, in the support services they provide, and in the extent to which they grant exemp- tions. Whereas some states have continued to help recipients pursue an education by allowing their activities to fulfill the work requirement, most have put more stress on immediate paid employment with no support for education and training after the first year. States also vary in the extent to which they provide child care services. Because large numbers of recipients have traditionally combined wel- fare and work for at least part of their time on public assistance, the availability of quality child care has always been an important consideration for many welfare recipients. It is likely to become an even more essential support service now that recipients are forced by new rules to devote greater portions of their time to seek- ing out and preparing for paid employment. Finally, states also vary considerably in how they grant exemptions from work requirements. Under the 1996 law, states can exempt up to 20 percent of their caseloads from work requirements and can exercise some choice over how to allo- cate these exemptions. For instance, some states may exempt parents with 55 POLITICS & SOCIETY children below a certain age, while others may implement the domestic violence option to exempt women at risk of battering and abuse. By the end of 1997, exemption criteria included domestic violence in twenty-seven states, medical disability in thirty-three states, mental health problems in thirty states, and AIDS in twenty-six states. Such procedural differences across states are, of course, criti- cal determinants of the value of welfare for the large numbers of recipients who confront these problems. In sum, the administration of welfare programs has always differed across states, and this procedural variation has been greatly expanded under the new TANF system. In this climate, it strikes us as untenable to focus narrowly on bene- fit variation when attempting to determine how state welfare programs might influence the residential decisions of poor families. A broad array of other institu- tional factors is likely to matter to recipients; many of these are critical determi- nants of whether a family will actually be able to receive benefits in a given state. Our point, of course, is not that recipients are aware of all these details regard- ing state programs. On the contrary, our argument is that the staggering rate and number of these restrictive policy changes force recipients to reason through their residential decisions under conditions of massive uncertainty. Under these condi- tions, "a bird in the hand" may be worth more to an impoverished recipient than the potential for "two" in a neighboring state.71 Indeed, as James Scott has empha- sized, the economically and politically marginal person is likely to choose secu- rity over advantage.72 Welfare migration simply does not seem very likely if one considers the plight of a poor single parent who currently survives by relying on informal support and a familiar welfare program, who confronts uncertain condi- tions across the state border, and who has everything to lose by making the wrong decision. The Heart of the Matter: Benefit Levels Reconsidered Finally, what if we adopt the most generous approach to assessing the welfare migration thesis, ignoring all of the factors discussed in previous sections and fol- lowing its narrow comparison of welfare benefits and costs of living? With these blinders firmly in place, the key question becomes whether benefits vary across states (1) more than the costs recipients would incur by moving and (2) enough to provide a clear economic incentive. If benefits do not vary enough to meet these two criteria, then models of welfare migration have little possibility of being correct. To approach these questions from a bottom-up perspective, one must begin by asking what sorts of benefits and costs are likely to matter most to recipients. If we agree to leave aside the other sources of income that matter to recipients, conven- tional studies of welfare migration seem right to assume that poor families experi- ence the value of a state's welfare package as a combination of cash and Food Stamp benefits they receive each month. The issue of costs, however, is not so 56 SCHRAM and SOSS straightforward. Recipients are likely to recognize that their ability to stretch AFDC and Food Stamp benefits over an entire month will hinge on the cost of liv- ing in some particular locale. We question, however, whether these costs are ade- quately reflected in a generic cost-of-living measure that is based on many items that welfare recipients cannot afford and do not buy. A bottom-up perspective offers a useful standpoint for questioning the assumption that an abstract set of costs is salient to recipients. We argue that researchers should not simply adjust combined benefits in light of an overall cost-of-living index: they should calculate the real value of these benefits accord- ing to the cost of the market goods that most directly affect the quality of welfare recipients' lives. For recipients, the value of a state's welfare grant is determined less by the general cost of living than by the cost of items they actually use their benefits to purchase. Research on poor people's spending patterns suggests that welfare recipients use a disproportionate amount of their combined benefits to cover food and hous- ing expenses.73 Because they do not receive enough Food Stamps to pay for the full amount and variety of food they require, most families are forced to use part of their cash grant to cover some of their food expenses.74 After these expenses are covered, most of the remaining cash assistance goes for housing. The cost of hous- ing is especially important for analyses of welfare migration because it is an expense that one would expect to be highly salient for recipients: unlike food or clothing costs, the cost of rent is encountered and paid as a lump sum rather than trickling out in day-to-day expenditures. The survey conducted by Voss, Corbett, and Randell confirms this expectation. The authors find that "the overall local cost of living did not play an important role in the decision to relocate. Rather . . . an interest in quality housing at a reasonable price appear[s] most important."75 This finding is hardly surprising if one considers the extent to which housing costs affect the economic lives of welfare recipients.76 A majority of families receiving welfare do not receive any housing subsidy.77 After paying for food costs with their Food Stamps and some of their welfare grant, housing is the larg- est single item in these families' budgets.78 In 1995, more than half of all poor renter households, regardless of welfare receipt, spent over 50 percent of their income for housing.79 Even the minority of recipient families that do receive hous- ing subsidies typically devote nearly half of their welfare grant to the cost of shel- ter.80 Thus, if we hope to assess the real value of welfare benefits for recipients (especially for the bulk of recipients who do not receive housing subsidies), we should compare variation in welfare grants to differences in housing costs rather than to general differences in cost of living. The best available indicator of housing costs for the welfare population comes from the Fair Market Rent estimates created by the U.S. Department of Housing and Urban Development (HUD) and used to determine rent calculations in HUD- financed projects that often include welfare recipients. Relative to a generic cost- 57 POLITICS & SOCIETY Table 2 Interstate Variation in Welfare Benefits, Generic Costs of Living, and Fair Market Rents (housing costs), 1993 and 1996 (in dollars) 1993 1996 Coefficient of Coefficient of Mean SD Variation Mean SD Variation Combined AFDC and Food Stamp benefits 617.90 95.64 .155 682.10 111.44 .163 The Friar-Leonard COLA (cost-of-living index) .9774 .0698 .071 .9822 .0653 .067 Average major metro Fair Market Rent (housing costs) 565.69 111.20 .197 577.29 111.54 .193 Source: See Table 1 sources for combined benefits and cost-of-living indexes. Fair Market Rents-1993: 57 Federal Register 45481 (1 October 1992) and 1996: 61 Federal Register 6705 (21 February 1996). Note: Table entries are based on the maximum monthly welfare benefit for a family of three in each state. AFDC = Aid to Families with Dependent Children. of-living index that does not focus on items that recipients actually buy, variation in the average Fair Market Rent provides a better measure of how far recipients are able to stretch their grants in each state. We use the average Fair Market Rent for 1993 and 1996 for a two-bedroom apartment in the major metropolitan area of each state that had the largest number of welfare recipients.8" Table 2 presents statistics describing interstate variation in combined welfare benefits, the generic cost-of-living index, and the average Fair Market Rent. As Table 2 shows, in both 1993 and 1996, housing costs varied more widely across states than the generic cost of living. In 1993, housing costs had a coefficient of variation of .197 compared to .071 for the generic cost of living. A similar dispar- ity existed in 1996: the coefficient of variation for housing costs, .193, is almost three times the size of the coefficient for the generic cost-of-living index, .067. The difference in variation here turns out to be critical. Recall that the existence of welfare migration incentives has typically been supported by showing that combined welfare benefits vary more across the states than does the cost of liv- ing.82 By contrast, Table 2 shows that, unlike the generic cost-of-living index, housing costs varied more than combined benefits in both 1993 and 1996. The coefficient of variation for housing costs in 1993, .197, was larger than the one obtained for combined benefits that year, .155. In 1996, variation in the cost of housing, .193, again exceeded variation in the value of combined welfare benefits, .163. Thus, the well-known claim that welfare benefits vary more across states than the cost of living does not hold when we compare welfare benefits to housing costs-the expense that is felt most keenly by the welfare poor. To assess the real incentives for migration confronted by recipients, however, we must look beyond these measures of overall variation and estimate the actual 58 SCHRAM and SOSS losses a recipient would incur due to housing costs if he or she sought additional benefits across a state line. In most studies of welfare migration, researchers take the curious step of dividing the combined benefit by their measure of costs in each state. This maneuver allows them to control for cost-of-living differences and hence produce a more accurate measure of overall variation. From a bottom-up perspective, however, it is difficult to see how this procedure offers insight into the incentives for poor families. Recipients contemplating an interstate move do not actually encounter overall patterns of variation. Rather, they confront some specific trade-off between wel- fare benefits and housing costs. From this perspective, the real value of the bene- fits obtained through migration can only be known by taking account of the "asso- ciated losses" a recipient experiences in moving across a state line. Thus, to accurately capture the trade-offs confronted by would-be migrants, we should not create a new measure of benefits that controls for variation in costs; we should examine the relationship between benefits and costs to ascertain the magnitude of losses associated with each dollar gained through migration. To do this, we need to calculate a regression coefficient rather than the tradi- tional coefficient of variation. From the recipient's point of view, the regression coefficient answers the pivotal question: how much will extra costs in housing be associated with each additional increment of benefits in a new state? Using an unstandardized bivariate regression coefficient to estimate the aver- age increase in housing costs associated with each dollar of benefits, we find that in 1993, migrating welfare recipients would have incurred an average increase of 69 cents in housing costs for each additional dollar of welfare benefits received in their new states (b = .692, SE = .136). In 1996, a recipient's housing costs would have increased an average of 62 cents for each additional dollar of welfare benefits (b = .62 1, SE = . 114). As the relatively small standard errors indicate, these posi- tive relationships meet conventional standards of statistical significance for both years. These results allow us to demonstrate just how small benefit variation really is. The first line of Table 2 shows that in 1993, a family of three moving to a state one standard deviation higher in combined benefits would have received an average of $95.64 per month in additional income. Based on our regression results, we esti- mate that extra housing costs would have eaten up an average of $66.18 of this income, leaving a net gain of only $29.46 per month. The average net gain pro- duced by a similar move in 1996 would only have been a little higher, at $42.24 per month-hardly a lucrative payoff for uprooting one's family and spending the resources needed to move to a new state. In fact, factoring in moving expenses would undoubtedly reduce this already limited payoff further. To sum up, then, we find little evidence suggesting that welfare migration could be driven by real disparities in welfare benefits across the states. Our analy- sis of state-level data suggests that even if we exclude all other relevant social-, 59 POLITICS & SOCIETY economic-, and welfare-related factors, the alleged cause of welfare migration offers a pitifully weak incentive for recipients. From this perspective, it should hardly be surprising that the behavior alleged in the welfare migration thesis is an exceedingly rare phenomenon. Put simply, the real value of welfare for recipients does not vary all that much across the states. As a final note, we hasten to add that although our analysis of benefits is based on AFDC data, there is little reason to believe that these benefit-based incentives have changed very much under the new TANF system. States always had discre- tion to set AFDC benefits; thus far, few states appear to have responded to the new TANF block-grant system by dramatically revising their benefit schedules. In 1997, as in past years, most states left their benefit levels unchanged: three states raised benefits, while six lowered them.83 Thus, the structure of incentives we have described for 1993 and 1996 has remained substantially the same up to the present. CONCLUSION In this article, we have suggested a variety of plausible reasons why, contrary to the conventional wisdom, benefit levels may not motivate welfare migration. The welfare migration thesis is built on a long history of local poor relief that has been characterized by a commitment to local obligation and a tendency to view poor people from other jurisdictions as undeserving interlopers. To this powerful mix, the modern thesis of welfare migration has added the assumption that welfare recipients are driven by an instrumental rationality to maximize their welfare benefits. The resulting image of undeserving outsiders acting on their self-interest by moving to communities that offer more generous benefits has proven to be remarkably influential in policy-making circles. It played a major role in the cam- paign that abolished the old AFDC system in 1996, and it continues to be a major factor influencing the ways states respond to the TANF system. The specter of welfare migration currently casts a long shadow over postre- form deliberations, stifling the imagination of legislators and undercutting the discretion that welfare reform has given the states. Haunted by the image of the rational welfare migrant, state-level policy makers now hesitate to offer benefits and services that might prove too helpful or generous to the poor. To the extent that this image leads policy makers to try to repel would-be migrants by mimicking the restrictive policies of their neighbors, they retreat from the challenge of serving as innovative "laboratories" seeking out new ways to combat poverty. In this man- ner, one of the oldest fears in American welfare politics seems to be undermining the ambitious hopes of contemporary reformers. That the migration thesis continues to have corrosive effects on American wel- fare politics, even as researchers have produced evidence that it does not exist, suggests the need for stronger substantive explanations for why welfare migration is not "rational" for poor families. In an attempt to fill this void, we have drawn on 60 SCHRAM and SOSS research regarding clients' lives to develop a bottom-up critique of the assump- tions that make the welfare migration thesis seem plausible. This approach has yielded two important insights. First, it is simply not tenable to believe that the scope of relevant residential incentives for recipients is limited to welfare benefits. State differences in welfare administration matter greatly to recipients and vary in ways that produce tangled and contradictory incentives. Economic considerations are shaped by additional factors such as job availability, wages, the transaction costs of moving, the quality of local services, and the presence or absence of kin-based exchange networks. Finally, one must also consider the social and psychological considerations of home, family, and friendship that recipients cite as the most important factors in their residential decisions. Regardless of whether one accepts the assumption of instrumental rationality that underlies the welfare migration thesis, these cross- cutting incentives must be recognized as factors capable of undercutting the influ- ence of welfare benefits. Second, even if one focuses narrowly on the welfare benefits and costs incurred by recipients, it is not all clear that interstate differences in welfare bene- fits offer meaningful incentives for migration. When benefits are examined in terms of their real worth to recipients, their variation appears smaller and less cer- tain than is commonly assumed. When we begin our analysis by asking what costs are felt most keenly by recipients, we find that, on average, housing costs eat up roughly two-thirds of the financial gains recipients might achieve through migra- tion. In 1993, for example, this would have left recipients with a net gain of only $29.46 per month. And, of course, to place even this small net gain in perspective, one must remember that welfare checks rarely cover the cost of necessities and that recipients must look to close kin, off-the-books employment, and elsewhere to cover additional expenses.84 Looking to the immediate future, it appears that the new intergovernmental system of welfare provision will foster some conditions that we suspect undercut welfare migration. In the short term, we expect that there will continue to be an absence of incentives strong enough to motivate welfare migration. Cash benefits continue to vary too little to induce migration, and welfare administration now varies so much that it can hardly be expected to present clear incentives to recipi- ents. This balance of incentives could all be changed, however, in the long term. If absolute time limits come to affect larger numbers of recipients and if states pur- sue increasingly stringent exclusionary practices, recipients will no longer have to consider subtle differences between state programs. As recipients find their own access to welfare restricted and simultaneously see the welfare-based resources in their kin networks dry up, the weak "pull" of greater benefits may be replaced by the strong "push" of extreme deprivation. Like the families who fled the American dust bowl in the 1930s, recipients may move toward states that offer assistance 61 POLITICS & SOCIETY simply because they cannot survive on "nothing" and hope that "something" is at least possible elsewhere. Although evidence for welfare migration remains scarce, states continue to fear it. And they now have a welfare system that, like water, can seek its lowest level. Although one can never make predictions with certainty, the incentives of the new system and fears of welfare migration combine to make it highly likely that the states will drift toward more benefit restrictions, shorter time limits, and stricter work requirements. The net result may be analogous to a cat chasing its tail: as states reshuffle themselves to the bottom, they may actually create the very welfare migration they fear.85 States that move expeditiously to tighten time limits and work requirements may force poor people to flee to other states out of sheer necessity. 6 Welfare migration has probably been no more than a ghost haunting welfare reform in the past. But if that ghost succeeds in generating a race to the bottom, it may create a class of impoverished welfare migrants that is very real. NOTES 1. See Russell Hanson and David T. Hartman, "Do Welfare Magnets Attract?" (Dis- cussion Paper #1028-14, Institute for Research on Poverty, University of Wiscon- sin-Madison, 1994). 2. Sanford F Schram and Gary Krueger, " 'Welfare Magnets' and Benefit Decline: Symbolic Problems and Substantive Consequences," Publius: The Journal of Federalism 24, no. 4 (1994): 61-81. 3. Irene Lurie, "Temporary Assistance for Needy Families: A Green Light for the States," Publius: The Journal of Federalism 27, no. 2 (1997): 73-88. 4. See Helen Hershkoff and Stephen Loffredo, The Rights of the Poor (Carbondale: Southem Illinois Press, 1997), 45-46. Title 42 was amended to include a new Section 402(a), stating that "a State operating a program funded under this part may apply to a fam- ily the rules (including benefit amounts) of the program funded under this part of another State if the family has moved to the State from the other State and has resided in the State for less than 12 months." This option is now under review by the U.S. Supreme Court in the case Anderson v. Roe (98-97). 5. Ibid., 45. 6. See John D. Donahue, Disunited States: What's at Stake as Washington Fades and the States Take the Lead (New York: Basic Books, 1997), 133-36. The contemporary debate about a "race to the bottom" in welfare was initiated by Paul E. Peterson and Mark C. Rom, Welfare Magnets: A New Casefora National Standard (Washington, DC: Brook- ings Institution, 1990). 7. Peterson and Rom, Welfare Magnets, 6-13. 8. As Peterson and Rom have argued, policy makers could quickly undercut the poten- tial for welfare migration by creating a national program offering uniform benefits in each state. See Peterson and Rom, Welfare Magnets, 119-50. 9. Francis Fox Piven and Richard A. Cloward, The Breaking of the American Social Compact (New York: The New Press, 1997), 72. 10. Jill Quadagno, The Color of Welfare: How Racism Undermined the War on Poverty (New York: Oxford University Press, 1994); Theda Skocpol, "African Americans in U.S. Social Policy," in Paul E. Peterson, ed., Classifying by Race (Princeton, NJ: Princeton Uni- versity Press, 1995). 62 SCHRAM and SOSS 11. Joel F. Handler and Yeheskel Hasenfeld, The Moral Construction of Poverty: Wel- fare Reform in America (Newbury Park, CA: Sage, 1991). 12. Joel F. Handler, The Poverty of Welfare Reform (New Haven, CT: Yale University Press, 1995), 91. 13. Ibid.; Handler and Hasenfeld, The Moral Construction of Poverty. 14. Regarding the foundations of distributive justice, Michael Walzer writes, "The pri- mary good we distribute to one another is membership in some human community. And what we do with regard to membership structures all our other distributive choices." See Michael Walzer, Spheres of Justice (New York: Basic Books, 1983), 31. 15. Handler and Hasenfeld, The Moral Construction of Poverty. 16. Karl de Schweinitz, England's Road to Social Security, 1349-1947 (Philadelphia: University of Pennsylvania Press, 1947). 17. Walter Trattner, From Poor Law to Welfare State: A History of Social Welfare in America (New York: Free Press, 1974), chap. 2. 18. Rand E. Rosenblatt, "Legal Entitlement and Welfare Benefits," in David Kairys, ed., Politics of Law: A Progressive Critique (New York: Pantheon, 1982). 19. Theda Skocpol, Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States (Cambridge, MA: Harvard University Press, 1992),139,149. 20. Jim Baumohl, "Now We Won't Call It Lobbying: The Federal Bureau of Narcotics and the Depression-Era Maintenance Controversy in California and Washington" (paper presented at the Conference on Historical Perspectives on Alcohol and Drug Use in Ameri- can Society, 1800-1997, Philadelphia, PA, May 1997). 21. Linda Gordon, Pitied but Not Entitled: Single Mothers and the History of Welfare (New York: Free Press, 1994), 294. 22. Rosenblatt, "Legal Entitlement and Welfare Benefits," 266. 23. Martha F. Davis, Brutal Need: Lawyers and the Welfare Rights Movement, 1960-1973 (New Haven, CT: Yale University Press, 1993), 72. 24. For instance, see Charles Murray, Losing Ground: American Social Policy 1950-1980 (New York: Basic Books, 1984); Lawrence Mead, The New Politics of Poverty: The Nonworking Poor in America (New York: Basic Books, 1992); and Mickey Kaus, The End of Equality (New York: Basic Books, 1992). 25. See Edward C. Banfield, The Unheavenly City Revisited (Boston: Little, Brown, 1974). For a critique, see Michael Katz, The Undeserving Poor: From the War on Poverty to the War on Welfare (New York: Pantheon, 1989), 9-35. 26. Peterson and Rom, Welfare Magnets, 12. 27. Thomas R. Dye, "The Policy Consequences of Intergovernmental Competition," Cato Journal 10, no. 1 (1990): 59-73 at 69. 28. From 1970 to 1996, the real value of the maximum welfare benefit for a family of three in the average state fell 46 percent. See U.S. House of Representatives, Committee on Ways and Means, Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means, "The 1996 Green Book," WMCP: 104-14 (Washing- ton, DC: Government Printing Office, 1996; available at http://www.access.gpo.gov/con- gress/wmOO1.html) (hereafter The 1996 Green Book). 29. Robert B. Albritton, "Impacts of Intergovernmental Financial Incentives on State Welfare Policymaking and Interstate Equity," Publius: The Journal of Federalism 19, no. 2 (1989): 127-41. 30. See Peterson and Rom, Welfare Magnets, 12; Walter W. McMahon, "Geographical Cost of Living Differences: An Update," Journal of the American Real Estate and Urban Economics Association 19, no. 3 (1991): 426-50; and Herman Leonard and Monica Friar, 63 POLITICS & SOCIETY State Cost of Living Indexes (Cambridge, MA: Kennedy School of Government, Harvard University, 1996). 31. See Dye, "The Policy Consequences of Intergovernmental Competition," 59-73; Robert B. Moffitt, "Incentive Effects of the U.S. Welfare System: A Review," Journal of Economic Literature 30, no. 2 (1992): 32-36. 32. Food Stamps are reduced approximately $.30 for each additional dollar of income. 33. See Peterson and Rom, Welfare Magnets, 12-13; Todd Zubler, "The Right to Migrate and Welfare Reform: Time for Shapiro v. Thompson to Take a Hike," Valparaiso University Law Review 31, no. 3 (1997): 893-950 at 935; and Donahue, Disunited States, 133. 34. See Sanford Schram, Lawrence Nitz, and Gary Krueger, "Without Cause or Effect: Reconsidering Welfare Migration as a Policy Problem," American Journal of Political Sci- ence 42, no. 1 (1998): 217-19. 35. The coefficient of variation is the standard deviation divided by the mean; it enables us to compare data "entirely different in magnitude or units of measure," according to Patri- cia E. Gaynor and Rickey C. Kirkpatrick, Introduction to Time-Series and Forecasting in Business and Economics (New York: McGraw-Hill, 1994), 38. This coefficient is appro- priate for comparing welfare benefits to the cost of living because these items are measured in entirely different units. By contrast, the standard deviation is more appropriate for com- paring variation in welfare benefits, including benefits measured in constant dollars over time. Inappropriate use of the coefficient variation can lead researchers to underestimate compression in the real value of welfare benefits. By dividing a declining standard devia- tion by a declining mean, researchers are likely to incorrectly produce evidence that there has been no decline in variation. This is one reason why benefit compression has been neglected by other researchers. See Peterson and Rom, Welfare Magnets, 8. 36. We adjusted combined Aid to Families with Dependent Children (AFDC) and Food Stamp benefits by dividing them by a cost-of-living index score for each state. See Leonard and Friar, State Cost of Living Indexes. 37. Peterson and Rom, Welfare Magnets, 12. 38. Richard J. Cebula, "A Survey of the Literature on the Migration-Impact of State and Local Government Policies," Public Finance/Finages Publiques 37, no. 1 (1979): 69-84. 39. Peterson and Rom, Welfare Magnets, 50-83; Dye, "The Policy Consequences of Intergovernmental Competition," 59-73. 40. Moffitt, "Incentive Effects of the U.S. Welfare System," 36. 41. Hanson and Hartman, "Do Welfare Magnets Attract?"; James R. Walker, "Migra- tion among Low-Income Households: Helping the Witch Doctors Reach Consensus" (Dis- cussion Paper #1031-94, Institute for Research on Poverty, University of Wiscon- sin-Madison, 1994); William H. Frey, Kao-Lee Liaw, Yu Xie, and Marcia Carlson, "Interstate Migration of the U.S. Poverty Population: Immigration 'Pushes' and Welfare Magnet 'Pulls'" (Research Report #95-331, Population Studies Center, University of Michigan, 1995); Phillip B. Levine and David J. Zimmerman, "An Empirical Analysis of the Welfare Magnet Debate Using the NLSY" (Working Paper #5264, National Bureau of Economic Research, Cambridge, 1995); Carole Roan Gresenz, "An Empirical Investiga- tion of the Role of AFDC Benefits in Location Choice" (Working Paper Series 97-05, DRU-161 1-RC, RAND Corporation, Santa Monica, CA, 1997); and Schram, Nitz, and Krueger, "Without Cause or Effect," 210-30. 42. Schram, Nitz, and Krueger, "Without Cause or Effect," 227. 43. According to the Public Use Microdata Set of the 1990 U.S. census, all poor single mothers in 1990 were no longer living in the house in which they had resided in 1985, but only 9 percent of them had moved out of state. See ibid., 220. 64 SCHRAM and SOSS 44. See Sanford F. Schram, Words of Welfare: The Poverty of Social Science and the Social Science of Poverty (Minneapolis: University of Minnesota Press, 1995), 59-76. 45. See Carol B. Stack, Call to Home: African-Americans Reclaim the Rural South (New York: Basic Books, 1996), xi-xix. 46. Schram, Words of Welfare, 59-76. 47. See Kathryn Edin and Laura Lein, Making Ends Meet: How Single Mothers Survive Welfare and Low-Wage Work (New York: Russell Sage, 1997). 48. Elliot Liebow, Tally's Corner: A Study of Negro Streetcorner Men (Boston: Little, Brown, 1967); Stack, All Our Kin. 49. See Edin and Lein, Making Ends Meet, 20-59; Joe Soss, "Unwanted Claims: Poli- tics, Participation and the U.S. Welfare System" (Ph.D. dissertation, University of Wiscon- sin-Madison, 1996); and Stack, Call to Home, xi-xix. 50. Carol B. Stack, "A Critique of Method in the Assessment of Policy Impact," Research in Social Problems and Public Policy 4 (1987): 137-47. 51. Paul Voss, Thomas Corbett, and Richard Randell, "Interstate Migration and Public Welfare: The Migration Decision Making of a Low Income Population," in P. C. Jobes, W. F. Stinner, and J. M. Wardell, eds., Community, Society, and Migration (New York: Univer- sity Press of America, 1992). 52. See Thad A. Brown, Migration and Politics (Chapel Hill: University of North Caro- lina Press, 1988); Lawrence A. Brown, Susan A. Gustavus, Edward J. Malecki, and Karen Walby, Choice Mechanisms in Migration Decisions (Columbus: Center for Human Resource Research, Ohio State University, 1985). 53. Patrick C. Jobes, William F. Stinner, and John M. Wardell, eds., Community, Soci- ety, and Migration (New York: University Press of America, 1992), 22-23. 54. Patrick C. Jobes, "Economic and Quality of Life Decisions in Migration to a High Natural Amenity Area," in Jobes, Stinner, and Wardell, eds., Community, Society, and Migration, 354. 55. See Carol B. Stack, All OurKin: StrategiesforSurvival in a Black Community (New York: Harper & Row, 1974); Stack, Call to Home. 56. Carol B. Stack and John B. Cromartie, "The Joumeys of Black Children: An Inter- generational Perspective," in Jobes, Stinner, and Wardell, eds., Community, Society, and Migration, 364. 57. L. A. Brown and J. Holmes, "Search Behavior in an Intra-Urban Migration Context: A Spatial Perspective," Environment and Planning 3, no. 3 (1971): 307-26. 58. Voss, Corbett, and Randell, "Interstate Migration and Public Welfare," 125. 59. Ibid., 129. 60. Gordon F. De Jong and Zafar M. N. Ahmad, "Motivation for Migration of Welfare Clients," in Anthony Richmond and Daniel Kubat, eds., Internal Migration: The New World and the Third World (Beverly Hills, CA: Sage, 1976). 61. Voss, Corbett, and Randell, "Interstate Migration and Public Welfare," 141. 62. Edin and Lein, Making Ends Meet, 43. 63. Ibid. 64. Stack, Call to Home, xiii-xix. 65. Among many others, see Edin and Lein, Making Ends Meet; Joel F. Handler, The Poverty of Welfare Reform; Mark R. Rank, Living on the Edge: The Realities of Welfare in America (New York: Columbia University Press, 1994). 66. All of the figures cited in the following paragraph can be found in Voss, Corbett, and Randell, "Interstate Migration and Public Welfare," 128-33. 67. See Edin and Lein, Making Ends Meet. 65 POLITICS & SOCIETY 68. Joe Soss, "Welfare Application Encounters: Subordination, Satisfaction, and the Puzzle of Client Evaluations," Administration & Society, forthcoming (1999). 69. The 1996 Green Book. 70. The state breakdowns in this section are from American Public Welfare Association (APWA), State-by-State Welfare Reform Policy Decisions (Washington, DC: APWA, 1997), 22-24. 71. As cognitive theories of decision making suggest, the value of what one already pos- sesses is likely to outweigh the value of what one might (or might not) be able to obtain under changed circumstances-especially if changing those circumstances entails the potential for substantial losses. See Deborah Stone, Policy Paradox: The Art of Political Decision-Making (New York: Norton, 1997). 72. Frances Fox Piven suggested James Scott's argument that the economically mar- ginal will choose "security over advantage." See James C. Scott, The Moral Economy of the Peasant: Subsistence and Rebellion in Southeast Asia (New Haven, CT: Yale University Press, 1976). 73. Edin and Lein, Making Ends Meet, 20-45. 74. Rebecca M. Blank, It Takes a Nation: A New Agenda for Fighting Poverty (New York: Russell Sage, 1997), 163. 75. Voss, Corbett, and Randell, "Interstate Migration and Public Welfare," 129. 76. See Barbara Sard, Ed Lazare, Robert Greenstein, and Jennifer Daskal, Housing Bills Could Weaken Welfare Reform and Create Problems for the Working Poor (Washing- ton, DC: Center on Budget and Policy Priorities, 1997); Tracy L. Kaufman, Out of Reach: Rental Housing at What Cost? (Washington, DC: National Low Income Housing Coali- tion, 1997). 77. Approximately 70 percent of the welfare population does not receive rent subsidies. See The 1996 Green Book and Kaufman, Out of Reach. 78. Edin and Lein, Making Ends Meet, 40. 79. Sard et al., Housing Bills Could Weaken Welfare Reform. 80. Ibid. 81. See 57 Federal Register 45468 (1 October 1992) and 61 Federal Register 6690 (21 February 1997). The U.S. Department of Housing and Urban Development (HUD) uses the fortieth percentile rent for the 1996 Fair Market Rent and so do we. 82. Peterson and Rom, Welfare Magnets, 12-13. 83. APWA, State-by-State Welfare Reform Policy Decisions, 22-24. 84. See Edin and Lein, Making Ends Meet. 85. Mark Alan Hughes, "The Welfare Dustbowl," The Honolulu Advertiser, 20 October 1996, B1. 86. The concem that reform may push recipients from one state to another has already intensified in adjacent states where people are cut from the rolls. See Stacy Milbouer, "Mass. Welfare Deadline Causes Concem: Some Could Seek Benefits in N.H.," Boston Globe, 22 November 1998, p. 8. In its Amicus Curiae brief submitted to the Supreme Court inAnderson v. Roe (98-97), the U.S. Govemment argues that Congress anticipated how the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 could well induce migration. Unfortunately, the Solicitor General went on to write that this justified allowing states to establish two-tier benefit schedules to discourage migration. Whether the Court agrees and upholds this option at the expense of the right to travel is yet to be determined. A better argument would be to agree that welfare reform may well cut people off from services and induce migration, but that is exactly why states should not be able to create barriers and deny access to such needed services. 66 1. See Russell Hanson and David T. Hartman, “Do Welfare Magnets Attract?” (Discussion Paper #1028-14, Institute for Research on Poverty, University of Wisconsin-Madison, 1994). 2. Sanford F. Schram and Gary Krueger, “ `Welfare Magnets' and Benefit Decline: Symbolic Problems and Substantive Consequences,” Publius: The Journal of Federalism 24, no. 4 (1994): 61-81. 3. Irene Lurie, “Temporary Assistance for Needy Families: A Green Light for the States,” Publius: The Journal of Federalism 27, no. 2 (1997): 73-88. 4. See Helen Hershkoff and Stephen Loffredo, The Rights of the Poor (Carbondale: Southern Illinois Press, 1997), 45-46. Title 42 was amended to include a new Section 402(a), stating that “a State operating a program funded under this part may apply to a family the rules (including benefit amounts) of the program funded under this part of another State if the family has moved to the State from the other State and has resided in the State for less than 12 months.” This option is now under review by the U.S. Supreme Court in the case Anderson v. Roe (98-97). 5. Ibid., 45. 6. See John D. Donahue, Disunited States: What's at Stake as Washington Fades and the States Take the Lead (New York: Basic Books, 1997), 133-36. The contemporary debate about a “race to the bottom” in welfare was initiated by Paul E. Peterson and Mark C. Rom, Welfare Magnets: A New Case for a National Standard (Washington, DC: Brookings Institution, 1990). 7. Peterson and Rom, Welfare Magnets , 6-13. 8. As Peterson and Rom have argued, policy makers could quickly undercut the potential for welfare migration by creating a national program offering uniform benefits in each state. See Peterson and Rom, Welfare Magnets , 119-50. 9. Francis Fox Piven and Richard A. Cloward, The Breaking of the American Social Compact (New York: The New Press, 1997), 72. 10. Jill Quadagno, The Color of Welfare: How Racism Undermined the War on Poverty (New York: Oxford University Press, 1994); Theda Skocpol, “African Americans in U.S. Social Policy,” in Paul E. Peterson, ed., Classifying by Race (Princeton, NJ: Princeton University Press, 1995). 11. Joel F. Handler and Yeheskel Hasenfeld, The Moral Construction of Poverty: Welfare Reform in America (Newbury Park, CA: Sage, 1991). 12. Joel F. Handler, The Poverty of Welfare Reform (New Haven, CT: Yale University Press, 1995), 91. 13. Ibid.; Handler and Hasenfeld, The Moral Construction of Poverty. 14. Regarding the foundations of distributive justice, Michael Walzer writes, “The primary good we distribute to one another is membership in some human community. And what we do with regard to membership structures all our other distributive choices.” See Michael Walzer, Spheres of Justice (New York: Basic Books, 1983), 31. 15. Handler and Hasenfeld, The Moral Construction of Poverty. 16. Karl de Schweinitz, England's Road to Social Security, 1349-1947 (Philadelphia: University of Pennsylvania Press, 1947). 17. Walter Trattner, From Poor Law to Welfare State: A History of Social Welfare in America (New York: Free Press, 1974), chap. 2. 18. Rand E. Rosenblatt, “Legal Entitlement and Welfare Benefits,” in David Kairys, ed., Politics of Law: A Progressive Critique (New York: Pantheon, 1982). 19. Theda Skocpol, Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States (Cambridge, MA: Harvard University Press, 1992), 139, 149. 20. Jim Baumohl, “Now We Won't Call It Lobbying: The Federal Bureau of Narcotics and the Depression-Era Maintenance Controversy in California and Washington” (paper presented at the Conference on Historical Perspectives on Alcohol and Drug Use in American Society, 1800-1997, Philadelphia, PA, May 1997). 21. Linda Gordon, Pitied but Not Entitled: Single Mothers and the History of Welfare (New York: Free Press, 1994), 294. 22. Rosenblatt, “Legal Entitlement and Welfare Benefits,” 266. 23. Martha F. Davis, Brutal Need: Lawyers and the Welfare Rights Movement, 1960-1973 (New Haven, CT: Yale University Press, 1993), 72. 24. For instance, see Charles Murray, Losing Ground: American Social Policy 1950-1980 (New York: Basic Books, 1984); Lawrence Mead, The New Politics of Poverty: The Nonworking Poor in America (New York: Basic Books, 1992); and Mickey Kaus, The End of Equality (New York: Basic Books, 1992). 25. See Edward C. Banfield, The Unheavenly City Revisited (Boston: Little, Brown, 1974). For a critique, see Michael Katz, The Undeserving Poor: From the War on Poverty to the War on Welfare (New York: Pantheon, 1989), 9-35. 26. Peterson and Rom, Welfare Magnets , 12. 27. Thomas R. Dye, “The Policy Consequences of Intergovernmental Competition,” Cato Journal 10, no. 1 (1990): 59-73 at 69. 28. From 1970 to 1996, the real value of the maximum welfare benefit for a family of three in the average state fell 46 percent. See U.S. House of Representatives, Committee on Ways and Means, Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means, “The 1996 Green Book,” WMCP: 104-14 (Washington, DC: Government Printing Office, 1996; available at http://www.access.gpo.gov/congress/wm001.html) (hereafter The 1996 Green Book ). 29. Robert B. Albritton, “Impacts of Intergovernmental Financial Incentives on State Welfare Policymaking and Interstate Equity,” Publius: The Journal of Federalism 19, no. 2 (1989): 127-41. 30. See Peterson and Rom, Welfare Magnets , 12; Walter W. McMahon, “Geographical Cost of Living Differences: An Update,” Journal of the American Real Estate and Urban Economics Association 19, no. 3 (1991): 426-50; and Herman Leonard and Monica Friar, State Cost of Living Indexes (Cambridge, MA: Kennedy School of Government, Harvard University, 1996). 31. See Dye, “The Policy Consequences of Intergovernmental Competition,” 59-73; Robert B. Moffitt, “Incentive Effects of the U.S. Welfare System: A Review,” Journal of Economic Literature 30, no. 2 (1992): 32-36. 32. Food Stamps are reduced approximately $.30 for each additional dollar of income. 33. See Peterson and Rom, Welfare Magnets , 12-13; Todd Zubler, “The Right to Migrate and Welfare Reform: Time for Shapiro v. Thompson to Take a Hike,” Valparaiso University Law Review 31, no. 3 (1997): 893-950 at 935; and Donahue, Disunited States , 133. 34. See Sanford Schram, Lawrence Nitz, and Gary Krueger, “Without Cause or Effect: Reconsidering Welfare Migration as a Policy Problem,” American Journal of Political Science 42, no. 1 (1998): 217-19. 35. The coefficient of variation is the standard deviation divided by the mean; it enables us to compare data “entirely different in magnitude or units of measure,” according to Patricia E. Gaynor and Rickey C. Kirkpatrick, Introduction to Time-Series and Forecasting in Business and Economics (New York: McGraw-Hill, 1994), 38. This coefficient is appropriate for comparing welfare benefits to the cost of living because these items are measured in entirely different units. By contrast, the standard deviation is more appropriate for comparing variation in welfare benefits, including benefits measured in constant dollars over time. Inappropriate use of the coefficient variation can lead researchers to underestimate compression in the real value of welfare benefits. By dividing a declining standard deviation by a declining mean, researchers are likely to incorrectly produce evidence that there has been no decline in variation. This is one reason why benefit compression has been neglected by other researchers. See Peterson and Rom, Welfare Magnets , 8. 36. We adjusted combined Aid to Families with Dependent Children (AFDC) and Food Stamp benefits by dividing them by a cost-of-living index score for each state. See Leonard and Friar, State Cost of Living Indexes. 37. Peterson and Rom, Welfare Magnets , 12. 38. Richard J. Cebula, “A Survey of the Literature on the Migration-Impact of State and Local Government Policies,” Public Finance/Finages Publiques 37, no. 1 (1979): 69-84. 39. Peterson and Rom, Welfare Magnets , 50-83; Dye, “The Policy Consequences of Intergovernmental Competition,” 59-73. 40. Moffitt, “Incentive Effects of the U.S. Welfare System,” 36. 41. Hanson and Hartman, “Do Welfare Magnets Attract?”; James R. Walker, “Migration among Low-Income Households: Helping the Witch Doctors Reach Consensus” (Discussion Paper #1031-94, Institute for Research on Poverty, University of Wisconsin-Madison, 1994); William H. Frey, Kao-Lee Liaw, Yu Xie, and Marcia Carlson, “Interstate Migration of the U.S. Poverty Population: Immigration `Pushes' and Welfare Magnet `Pulls' ” (Research Report #95-331, Population Studies Center, University of Michigan, 1995); Phillip B. Levine and David J. Zimmerman, “An Empirical Analysis of the Welfare Magnet Debate Using the NLSY” (Working Paper #5264, National Bureau of Economic Research, Cambridge, 1995); Carole Roan Gresenz, “An Empirical Investigation of the Role of AFDC Benefits in Location Choice” (Working Paper Series 97-05, DRU-1611-RC, RAND Corporation, Santa Monica, CA, 1997); and Schram, Nitz, and Krueger, “Without Cause or Effect,” 210-30. 42. Schram, Nitz, and Krueger, “Without Cause or Effect,” 227. 43. According to the Public Use Microdata Set of the 1990 U.S. census, all poor single mothers in 1990 were no longer living in the house in which they had resided in 1985, but only 9 percent of them had moved out of state. See ibid., 220. 44. See Sanford F. Schram, Words of Welfare: The Poverty of Social Science and the Social Science of Poverty (Minneapolis: University of Minnesota Press, 1995), 59-76. 45. See Carol B. Stack, Call to Home: African-Americans Reclaim the Rural South (New York: Basic Books, 1996), xi-xix. 46. Schram, Words of Welfare , 59-76. 47. See Kathryn Edin and Laura Lein, Making Ends Meet: How Single Mothers Survive Welfare and Low-Wage Work (New York: Russell Sage, 1997). 48. Elliot Liebow, Tally's Corner: A Study of Negro Streetcorner Men (Boston: Little, Brown, 1967); Stack, All Our Kin. 49. See Edin and Lein, Making Ends Meet , 20-59; Joe Soss, “Unwanted Claims: Politics, Participation and the U.S. Welfare System” (Ph.D. dissertation, University of Wisconsin-Madison, 1996); and Stack, Call to Home , xi-xix. 50. Carol B. Stack, “A Critique of Method in the Assessment of Policy Impact,” Research in Social Problems and Public Policy 4 (1987): 137-47. 51. Paul Voss, Thomas Corbett, and Richard Randell, “Interstate Migration and Public Welfare: The Migration Decision Making of a Low Income Population,” in P. C. Jobes, W. F. Stinner, and J. M. Wardell, eds., Community, Society, and Migration (New York: University Press of America, 1992). 52. See Thad A. Brown, Migration and Politics (Chapel Hill: University of North Carolina Press, 1988); Lawrence A. Brown, Susan A. Gustavus, Edward J. Malecki, and Karen Walby, Choice Mechanisms in Migration Decisions (Columbus: Center for Human Resource Research, Ohio State University, 1985). 53. Patrick C. Jobes, William F. Stinner, and John M. Wardell, eds., Community, Society, and Migration (New York: University Press of America, 1992), 22-23. 54. Patrick C. Jobes, “Economic and Quality of Life Decisions in Migration to a High Natural Amenity Area,” in Jobes, Stinner, and Wardell, eds., Community, Society, and Migration , 354. 55. See Carol B. Stack, All Our Kin: Strategies for Survival in a Black Community (New York: Harper & Row, 1974); Stack, Call to Home. 56. Carol B. Stack and John B. Cromartie, “The Journeys of Black Children: An Intergenerational Perspective,” in Jobes, Stinner, and Wardell, eds., Community, Society, and Migration , 364. 57. L. A. Brown and J. Holmes, “Search Behavior in an Intra-Urban Migration Context: A Spatial Perspective,” Environment and Planning 3, no. 3 (1971): 307-26. 58. Voss, Corbett, and Randell, “Interstate Migration and Public Welfare,” 125. 59. Ibid., 129. 60. Gordon F. De Jong and Zafar M. N. Ahmad, “Motivation for Migration of Welfare Clients,” in Anthony Richmond and Daniel Kubat, eds., Internal Migration: The New World and the Third World (Beverly Hills, CA: Sage, 1976). 61. Voss, Corbett, and Randell, “Interstate Migration and Public Welfare,” 141. 62. Edin and Lein, Making Ends Meet , 43. 63. Ibid. 64. Stack, Call to Home , xiii-xix. 65. Among many others, see Edin and Lein, Making Ends Meet ; Joel F. Handler, The Poverty of Welfare Reform ; Mark R. Rank, Living on the Edge: The Realities of Welfare in America (New York: Columbia University Press, 1994). 66. All of the figures cited in the following paragraph can be found in Voss, Corbett, and Randell, “Interstate Migration and Public Welfare,” 128-33. 67. See Edin and Lein, Making Ends Meet. 68. Joe Soss, “Welfare Application Encounters: Subordination, Satisfaction, and the Puzzle of Client Evaluations,” Administration & Society , forthcoming (1999). 69. The 1996 Green Book. 70. The state breakdowns in this section are from American Public Welfare Association (APWA), State-by-State Welfare Reform Policy Decisions (Washington, DC: APWA, 1997), 22-24. 71. As cognitive theories of decision making suggest, the value of what one already possesses is likely to outweigh the value of what one might (or might not) be able to obtain under changed circumstances—especially if changing those circumstances entails the potential for substantial losses. See Deborah Stone, Policy Paradox: The Art of Political Decision-Making (New York: Norton, 1997). 72. Frances Fox Piven suggested James Scott's argument that the economically marginal will choose “security over advantage.” See James C. Scott, The Moral Economy of the Peasant: Subsistence and Rebellion in Southeast Asia (New Haven, CT: Yale University Press, 1976). 73. Edin and Lein, Making Ends Meet , 20-45. 74. Rebecca M. Blank, It Takes a Nation: A New Agenda for Fighting Poverty (New York: Russell Sage, 1997), 163. 75. Voss, Corbett, and Randell, “Interstate Migration and Public Welfare,” 129. 76. See Barbara Sard, Ed Lazare, Robert Greenstein, and Jennifer Daskal, Housing Bills Could Weaken Welfare Reform and Create Problems for the Working Poor (Washington, DC: Center on Budget and Policy Priorities, 1997); Tracy L. Kaufman, Out of Reach: Rental Housing at What Cost? (Washington, DC: National Low Income Housing Coalition, 1997). 77. Approximately 70 percent of the welfare population does not receive rent subsidies. See The 1996 Green Book and Kaufman, Out of Reach. 78. Edin and Lein, Making Ends Meet , 40. 79. Sard et al., Housing Bills Could Weaken Welfare Reform. 80. Ibid. 81. See 57 Federal Register 45468 (1 October 1992) and 61 Federal Register 6690 (21 February 1997). The U.S. Department of Housing and Urban Developmnet (HUD) uses the fortieth percentile rent for the 1996 Fair Market Rent and so do we. 82. Peterson and Rom, Welfare Magnets , 12-13. 83. APWA, State-by-State Welfare Reform Policy Decisions , 22-24. 84. See Edin and Lein, Making Ends Meet. 85. Mark Alan Hughes, “The Welfare Dustbowl,” The Honolulu Advertiser , 20 October 1996, B1. 86. The concern that reform may push recipients from one state to another has already intensified in adjacent states where people are cut from the rolls. See Stacy Milbouer, “Mass. Welfare Deadline Causes Concern: Some Could Seek Benefits in N.H.,” Boston Globe , 22 November 1998, p. 8. In its Amicus Curiae brief submitted to the Supreme Court in Anderson v. Roe (98-97), the U.S. Government argues that Congress anticipated how the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 could well induce migration. Unfortunately, the Solicitor General went on to write that this justified allowing states to establish two-tier benefit schedules to discourage migration. Whether the Court agrees and upholds this option at the expense of the right to travel is yet to be determined. A better argument would be to agree that welfare reform may well cut people off from services and induce migration, but that is exactly why states should not be able to create barriers and deny access to such needed services.

    Cite this