Subsequently published as The Market for Deadbeats, 35 Journal of Legal Studies 201-32 (1996). Margaret F. Brinig George Mason University School of Law 3401 N. Fairfax Drive Arlington, VA 22201 voice (703) 993-8022 FAX (703) 993-8088 web page: http://mason.gmu.edu/~mbrinig THE MARKET FOR DEADBEATS Margaret F. Brinig* & F.H. Buckley** The flow of people from one state to another may provide useful information about both states. While people move for a variety of reasons, net immigration states are in general more attractive than net emigration ones. Thus, many of our intuitions about the attractiveness of different states are based on migration flows. Should we notice a mass exodus from Florida to Castro's Cuba, for example, we might want to reconsider our feelings about socialism. Migration flows are therefore of interest to the economist. Other scholars, too, might profitably study the movement of people. The political theorist who, like Locke, would base political obligation on a contractua foundation must assume that migration is relatively costless. Otherwise, how can a person be said to consent to the authority of his state? Choices are tainted by duress unless there is an alternative to acceptance. The extent to which foundational rights should be imposed in a federal state might also be thought to depend on migration costs. Where emigration is impossible, a thick set of national mandatory rights, like the U.S. Bill of Rights, might commend itself; but where interstate emigration is relatively costless, federally- imposed rights might reasonably be relaxed, with states permitted to craft their own church-state barriers or gun control laws. A subject might then elect his own regime of constitutional rights through the exit option of migration. Such a regime might be thought more protective of liberty than one which removes such choices from the subject Migration flows must also be of interest to the public choice theorist. Migration changes life for emigration and immigration state natives, as wel as the migrant. Politicians in both states might therefore adopt policies to attract a favored class of migrant or to repel a disfavored class. In addition, migrants are voters, and a state's migration policies might plausibly be drafted with an eye to the migrant's anticipated voting sympathies. Through migration incentives, an unpopular politician might seek to prorogue the electorate and elect one more to his liking. Migration flows merit the attention of lawyers, as well as economists and political theorists. Legal institutions and governmental structures may attract or repel migrants. Thus the legal scholar might look to migration flows as a source of evidence about the efficiency of legal rules. Prior law and economics scholarship has ignored such evidence. In this Article, we seek to repair this omission by showing that lawyers may profitably study migration theories. At the same time, we do not minimize the ambiguity of the evidence from migration flows, since states might adopt inefficient as well a efficient laws to attract migrants. Of all people, Americans should be most ready to subscribe to migration theories. Americans are relatively mobile, and migration trends are well- reported in American newspapers. In addition, more than one hundred years ago America's foremost historian explained American history by reference t migration flows. Under Frederick Jackson Turner's Frontier Theory, states adopted liberal laws and democratic values to attract migrants. In this way, the competition for migrants was seen to shape the most basic of legal institutions. Under the Frontier Theory, the competition for migrants is wholly benign. States offer efficient laws to attract migrants, who confer gains on natives. The result is a race for the top, won by the state with the best set of laws and institutions. However, the competition for migrants might be value-decreasing, and states might be led to adopt inefficient laws in order to attract migrants. Thus, a state might compete for migrants on the basis of their expected political sympathies, rather than for their ability to benefit natives. For example, a pro-welfare political party might seek to attract pro welfare migrants through wealth transfer from wealthy natives. This kind o migration is unlikely to result in efficiency gains, and consumes deadweight moving costs. There is a further reason why the competition for migrants might be value-decreasing: States might compete for deadbeats. We define a deadbeat as one who crosses state lines to avoid repayment of a debt. The debt might be a consumer loan owed to a financing company, a personal loan from a friend, or an obligation of spousal or child support. Given collection costs, moving away increases the probability that the creditor will write off the debt as a bad debt. The debtor might also reduce the probability of repayment by moving to a state with pro-debtor insolvency laws. The prospect of deadbeat migratio might indeed lead a state to adopt pro-debtor laws. Deadbeats are ofte entrepreneurs or professionals with high future earnings, and the cost of their default is born primarily by out-of-state creditors. Even if sociall wasteful, then, deadbeat migration might be privately optimal for immigration state natives. Because of this, states might seek to attract deadbeats through inefficient insolvency laws In Section I we describe the market for migrants. The competition fo migrants is benign when states seek to attract value-increasing migrants by enacting efficient laws. However, the competition may be value-reducing when it is for migrant votes or for deadbeats. In Section II, we describe an econometric study of recent interstate migration trends. Our predictors of migration include distance, temperature, economic conditions, welfare and tax policies, as well as pro-deadbeat laws. The results of our econometric analysis are presented in Section III. Our principal finding is that recent migration flows are consistent with deadbeat, tax and welfare- motivated migration theories. I. The Market for Migrant A market for migrants exists when states tailor their policies t attract a favored or repel a disfavored class of migrants. For example, a state might subsidize immigration through homestead rights or increase welfare benefits. Alternatively, a state might reduce welfare payouts below what they would be in a world of closed borders in order to deter welfare- loving migrants. Of greater interest to lawyers are the private law rules or public law institutions a state might adopt to compete for a favored class of migrants. States have competed for migrants throughout American history. Durin the colonial period, and for much of the nineteenth century, America experienced labor shortages, and looked to migration, particularly from Europe, for fresh workers. Through migration, America resolved border problems with Canada and Mexico, and reduced the threat of Indian wars. Fresh migrants also promoted scale economies, with better schools, increased markets for products, and improved methods of transportation. In particular, nineteenth century migration patterns crucially influenced the railroad- building on which a town's fate might depend. In this century, after states adopted social welfare policies, migration tended to effect a wealth transfer from emigration to immigration state natives. Migrants were relatively young, industrious, and healthy. After their schooling in the emigration state, they moved to their new state, where they worked for a relatively long time before they retired. The emigration state subsidized their education, but lost them as taxpayers; while the immigration state gained people who contributed more in taxes and consumed less in welfar benefits than natives. This explains why America sought migrants in the past, most noticeably through federal and state homestead legislation. The federal Homestead Act of 1862 offered public land free of charge to homesteaders who settled it. Many states, particularly in the West and South, made the homestead exempt from seizure for pre-homestead debts. Even before then, states competed for migrants through their insolvency policies. Some states offered a fresh star in insolvency, under which a debtor could shelter future earnings from creditors by consigning his present assets to them. When an emigration state refused to concede such rights, a debtor could still effect a homemade fresh start by leaving his assets behind and moving to a fresh start state The claim that migration benefits natives is more controversial today than in the past. Economies of scale are more speculative, congestion costs are greater, and several states now seek restrictions on international immigration. But this is not to say that the market for migrants ha disappeared. Even states that seek to exclude international migrants might still want internal migrants, the subject of this study. As well, a state will have taken a position in the market for migrants when it seeks to deter their entry through changes in legal policies. Finally, no modern state, Hoxha' Albania excepted, has attempted to deter all migrants. Instead, states have sought to attract a favored while repelling a disfavored class of migrants Moving is costly, and for most of us the migration decision is made deliberately and soberly, not unadvisedly and lightly. From this, we migh wrongly conclude that few people move, and never because of differences in legal regimes. But Americans do move often. In 1990, for example, nearly 40% of Americans were living in a different state from the one they were bor in. This number would be smaller if one excluded the seniors who move on retirement; however, the number would be larger if one included the children born to migrants in the new state. For immigration state natives, then, migration profoundly changes the state they live in; for immigration state politicians, migration profoundly alters the electorate. It is reasonable to suppose, therefore, that a state might seek to adopt implicit migration policies. A. The Race for the Top On race for the top theories, migration is seen to benefit immigration state natives. Because of this, states might compete for migrants by offering them desirable legal and governmental institutions. Even emigration states will copy such institutions, the better to retain valued subjects. Thus th competition for migrants results in a race by states for the best set of laws. Race for the top migration theories were first stated by Frederick Jackson Turner just over one hundred years ago. The 1890 Census Report had noted a curious change in American society. Up to that point, the demographe could always point to a settlement line, beyond which the pioneer had not ventured. But by 1890, this line had disappeared, with only small and isolated frontier areas remaining. Thereafter, the frontier passed into history, and three years later, at the Chicago Exposition, Turner delivered the seminal paper in American historiography on how the frontier had shaped American institutions. States competed for migrants, Turner said. To attract them, states liberalized their laws and adopted free and democratic institutions. Western states led the way, and eastern states followed once they began to lose migrants to the more democratic Western states. "It was the greatest of frontiersmen who declared: 'I believe this Government can not endure permanently half slave and half free. It wil become all one thing or all the other.' ... Mobility of population is death to localism, and the western frontier worked irresistibly in unsettling population. The effect reached back from the frontier and affected profoundly the Atlantic coast and even the Old World. Well before modern public choice scholarship, Turner described how migrants vote with their feet, and how states compete for citizens. Hi insight that the competition for migrants shapes domestic institutions remains relevant today. To be sure, all states have adopted democratic institutions. There is, however, a broad variance in the method and level of taxation, welfare payouts, legal rules and governance structures. Race for the top explanations of the market for migrants resemble race for the top theories in corporate law. Most large American corporations choose to incorporate in Delaware. At one time, it was thought that this was because Delaware law was slanted towards managers and against shareholders. But this view has not withstood empirical scrutiny, and most commentators today believe that the greater popularity of Delaware's corporations law stems from its greater efficiency. In an effort to attract corporations, with their franchise fees and spillover legal business, Delaware jettisoned cumbersome legal formalities and adopted a streamlined corporate statute. It won the competition for corporations by adopting the best set of laws. In the same way, an immigration state wins the competition for migrants by adopting efficient laws, on race for the top theories. B. The Race for the Bottom: Electoral Theories On race for the top theories, state enact efficient laws in order to attract migrants; on race for the bottom theories, states enact inefficient laws to do so. The first race to the bottom theory is an electoral one, under which migrants are inefficiently sought for their future votes. The decision to attract or repel a class of migrants is a political one, made by th gatekeepers elected or appointed to perform this task. Gatekeeper incentives are imperfectly aligned with those of natives, and value-decreasing migrants might therefore be pursued as a consequence of the agency costs of gatekeeper misbehavior. Where gatekeepers must stand for election, then, they might see value-decreasing migrants for their votes, replacing an unhappy electorate with one more to their liking. On electoral theories of migration, a state seeks to attract or repel a constituency of migrants on the basis of their anticipated voting patterns Electoral theories plausibly account for shifts in U.S. immigration policie in the 1920s and 1960s, even though aliens had to wait until they becam citizens before they could vote. Electoral explanations of internal migration trends are more persuasive still, since interstate migrants may vote as soon as they establish their new residence. Thus, a pro-welfare political party might court welfare-loving migrants through the promise of high welfar payouts, while antiwelfare parties might seek welfare-haters through reduced welfare benefits. Similarly, antitax parties will court antitax migrants, while protax parties will court migrants who support government spending, especially if their taxes are low in a progressive regime. The result is a cooperative game in which different parties in different states trade off voters in the manner of the Jack Spratt family at table. In neither state will welfare payouts or tax rates be as low or as high as they would be in a world of closed borders. Electoral explanations of welfare payouts overlap with bureaucratic explanations of welfare policies, in which the crucial support for antipoverty measures comes not from welfare recipients but from welfare administrators On this analysis, welfare administrators seek to expand their caseloads by attracting welfare-motivated migrants from other jurisdictions C. The Race for the Bottom: Deadbeat Apart from vote-seeking, other pathologies may infect the market for migrants. Migration policies which are privately optimal for an immigration state may not be socially optimal, when losses borne in the emigration state are taken into account. Thus, an immigration state might compete for migrants by permitting them, once they have moved, to cut off claims owed to emigratio state creditors. This might be done through fresh start rights in bankruptcy and family law which permit the debtor to shelter future income from hi creditors and first wives and children. While socially wasteful, this may be privately optimal for immigration state natives. The offer to cut off creditor claims will be more valuable for high quality debtors, with high future earnings. Because of the promise to cut off creditor claims, more of them will move to the immigration state; and they will be richer, having left their creditors behind them. Deadbeats are not an unmixed blessing for the immigration state. Having dished his creditors once, the deadbeat may be tempted to do so again, this time at the expense of immigration state creditors. Because of this, some states may wish to discourage their entry by restricting fresh start rights. In this way, states may sort themselves into Kingdoms of Cooperation and Republics of Defection. Any test of the hypothesis that some states seek to attract deadbeat migrants by strengthening fresh start rights is simultaneously a test of the hypothesis that other states seek to deter their entry by weakening such rights. 1. Bankruptcy Deadbeats. Countries may compete for deadbeats through their bankruptcy laws. For example, fresh start policies in the American Bankruptcy Code are more pro-debtor than those of rival immigration states, such as Canada. Barriers to filings by opportunistic debtors who seek to cu off claims which they can afford to repay are very different in the two countries. The Canadian discharge may be suspended or granted conditionally in order to transfer a portion of the debtor's future earnings to creditors. By contrast, American debtors are given the right to elect between Chapter 7, where present assets are surrendered and future earnings retained, and Chapter 13, where present assets are retained and future earnings are committed t creditors. An opportunistic creditor will elect strategically, choosing the option which maximizes his wealth. Thus a debtor with few present assets but great expectations will file under Chapter 7 even if over time he could repa his debts. Debtor opportunism is policed by the Bankruptcy Code's good faith norms. These norms were strengthened in 1984, when 707(b) permitted a court to reject petitions which constituted "a substantial abuse." This has been interpreted to mean that a Chapter 7 filing may be set aside if the debtor proposes to maintain an exorbitant lifestyle, or is able to pay off a substantial portion of his debt under Chapter 13. In spite of this, the 198 amendments coincided with an enormous increase in filing rates. Between 1984 and 1991, a period of substantial prosperity, consumer filings rates under Chapter 7 tripled. This increase might plausibly be attributed to a weakening of good faith norms at trial, even as such norms were ostensibly strengthened by appellate courts. While some bankruptcy districts adhered to the reinforced good faith norms, others paid little attention to them, granting a discharge to debtors who might have repaid all their debts without undue hardship. For example, one court refused to dismiss a Chapter 7 petition under 707(b) even though the debtors had enough income to repay all claims withi three years. Another court refused to impeach a filing after finding that the debtor's estimated expenses of $700 a month on entertainment were unreasonable and that the debtor had disposed of a $10,000 tax refund two months before the petition. Cross-country differences in legal norms appear to explain why Canadian consumer bankruptcy filing rates are lower than those in the United States, for the higher American rates cannot be attributed to differences in economic variables. On deadbeat theories, then, the two countries compete fo different kinds of immigrants through their bankruptcy laws. Similarly, differences in state filing rates within the United States might be explained on internal migration theories. Informal studies have reported differences in the way in which bankruptcy trustees interpret good faith norms, and there is substantial variation in state filing rates. By weakening good faith norms, a judge will attract deadbeat migrants to his jurisdiction. With this will come increased work for the bankruptcy bar from which he was appointed, and to which he may feel continued loyalty. In addition, should the judge seek reappointment, the support of the bankruptcy bar might count for mor than the good opinion of appellate judges. At the same time, judges in other districts might wish to strengthen good faith norms in order to lighten thei case load. During the period of the Articles of Confederation, the American provinces appear to have competed for migrants in this way through their insolvency legislation. The Contracts Clause of the Constitution was adopted to prevent interstate exploitation of this kind, but state laws fettered creditor collection efforts throughout the nineteenth century. Today, what remains of such laws are state law exemptions from bankruptcy. Bankruptcy Code 522(d) offers debtors a presumptive list of assets which they may shelte against unsecured creditors. However, 522(b) permits states to opt out of this list, and about two-thirds have done so, some to narrow and some to widen the list of exempted assets. In some states, for example, debtors may shelter only $7500 in their house from unsecured creditors; but in other states there is no ceiling on the value of the homestead exemption. Not surprisingly, wealthy debtors have taken advantage of these differences to move to high exemptions states 2. Family Deadbeats. Family deadbeats seek to avoid legal obligations, abandoning their wives and children to public welfare or private charity. They are the stuff of Dickensian novels and of Grimm's fairy tales They were the immigrants who never sent back for their wives, and the pioneers who cast off their families to move West. They were Theseus on the coastal road and Leatherstocking on the prairie. They lived lives without second acts, and gave us the laws of support, alimony and divorce. For them, the West offered freedom from family responsibilities as well as the political freedom of the Frontier Thesis. Deserted wives could assert a variety of remedies against their spouses. Desertion was a ground of divorce, and states mandated child support obligations. Children were also protected through child abuse and compulsory education legislation. There is, however, a wide variance amongst states in family support obligations, and deadbeat spouses have an incentive to move to low-payout states. The deserted spouse could still sue for support in her hom jurisdiction, and seek to enforce the order in the state to which her spouse had moved. However, this was never an entirely effective remedy. In the past, many states were unwilling to enforce support decrees from a foreig jurisdiction, particularly since this required the application of crimina extradition laws. Under the prodding of the federal government, this changed in the 1950s, when states enacted Uniform Reciprocal Enforcement of Suppor legislation. Where the deadbeat can be located, these statutes provide an expeditious remedy for the deserted spouse, without court filings and attorney costs. The social service agency in the home state simply sends a copy of the decree mandating the obligation to the state where the defendant lives. He is permitted to present a defense -- he never owed the money, he has already paid, or he has no money. The defendant may also argue that the foreign award should be reduced because the plaintiff needs less money. Some states make a general practice of reducing foreign awards, particularly when the defendant has remarried and has a new family in the destination state. Even with reciprocal enforcement of support laws, then, a state might compete for divorced spouses by offering to reduce support obligations. II. The Empirical Test These hypotheses offer different predictions about migration patterns. Race for the top theories predict that states with expanding economies will attract migrants, and that stagnant states with high tax will repel them. States might also compete for migrants through their legal institutions, and on race for the top theories the winner of the competition will have the most desirable set of laws. Electoral theories predict that some states will see to attract migrants through high welfare payouts, while other states will attempt to export them through low welfare payouts. In the same way, states might trade off voters through differential tax burdens. Finally, deadbeat theories predict that more migrants will move to deadbeat havens, which offer debtors increased protection against creditors A. The Migration Flow Our migration model examines the 1985-90 migration flow in the continental United States. The dependent variable, based on data provided b the Census Bureau, is the total number of people during that period who moved from StateI to StateJ divided by StateJ population. Our predictors were StateJ variables. Since we looked at the migrants from the perspective of StateJ, our model was an immigration rather than an emigration one. We sought to identify what motivated migrants to settle in one state rather than another, not why they left. We looked for pulls rather than pushes. We excluded those who moved from their state after 1985 but returned to it by 1990. Because of the higher moving costs, we also excluded migration into or out of Alaska, Hawaii and offshore territories. For each of the remaining forty-eight states, then, migrants might have moved to any one of forty-seven different states. In this way, we constructed a table with 225 observations. Because we wished to test the effect of differences in state laws, we focused on interstate migration rather than migration between Standard Metropolitan Statistical Areas (SMSAs). The 1985-90 migration flow is the most current available data on interstate migration in the United States. It is also a useful sample upon which to test migration theories. The period coincided with an enormous increase in American wealth, though the growth was stronger in some states than others. It was also a time of tax revolt, when state tax burdens differed markedly from each other. In addition, there was substantial variance in state welfare payouts. Finally, the period coincided with an enormous increase in bankruptcy filings and divorce petitions, with a substantial variance in state filing rates. B. The Determinants of Migration We considered several different classes of predictors of migration patterns. Our environmental predictors included distance, temperature, population density, and the per cent of the population over 65. Our economi predictors were based on figures for construction starts and unemployment. To test electoral theories, we included welfare, state tax and state debt variables. Finally, to test deadbeat theories, our predictors included Chapter 7 consumer bankruptcy petitions and in-state child support collection rates. Unless indicated, the source for the independent variables was the Statistical Abstract of the United States 1986-93, with 1985 figures selected for our independent variables. While the dependent variable in our model is total 1985-90 interstate migration, using 1985 predictors reduced endogeneity concerns. Under endogeneity, causation works in both directions, and a change in the value of the dependent variable results in a change in the value of a predictor. For example, migrants are attracted to states with booming economies, as measured by relatively high construction starts. But fresh migrants also mean increased construction starts, since they will need new homes to live in and offices to work in. Had we used post-1985 figures, then, a finding that the construction start variable was significant would have been ambiguous. We also transformed the values of the non-Dummy variables into their natural logs, after determining through a Box-Cox test that this wa appropriate. 1. Environmental Predictors. Since migration costs increase with the distance travelled, we included a Distance predictor, measured by the driving distance between state capitals. We also included a Temperature variable to test whether the trend to the Sunbelt continues, independent of other predictors. As a measure of Temperature, we took the average January high temperature in the largest city of each state from 1961 to 1990. To measure migration flows to and from urban states, we included a Metropolitan variable, representing the extent to which the destination state was urbanized. Finally, we adopted an Elderly variable for the per cent of the destination state's over 65 population. Some of the migration flow represents older people moving to retirement havens. One would expect that states which were attractive to retirees in 1985 would remain attractive to them over the next five years. We therefore assumed that Elderly might serve as a suitabl proxy for the factors which attract retiree immigration 2. Economic Predictors. Race to the top theories predict that migration patterns are sensitive to economic conditions. The model's economic proxies were Construction Starts, the dollar value of completed commercial an residential construction contracts for new structures and additions, divided by the adult population; and Unemployment, the average monthly unemployment rate. There was substantial variance in both figures: The mean Construction Start value was 1.310, with a standard deviation of 0.464; and the mean Unemployment rate was 7.060, with a standard deviation of 1.938. 3. Electoral predictors. Vote-seeking politicians might compete for welfare-loving migrants by offering them relatively high welfare payouts. A the same time, politicians who seek the electoral support of welfare-haters might be tempted to reduce welfare payouts below what they would be in a world of closed borders. As a proxy for welfare, AFDC represents the average welfare payout under the Aid for Families with Dependent Children programme. Once again, there is a substantial variance in state AFDC payments, with a mean payout of $293.60 a month, and a standard deviation of $102.50. Using dat provided by the U.S. Census Department, we adjusted the AFDC variable fo differences in the cost of living amongst states. States also compete for migrants by lowering or raising their tax burden. The Tax variable represents total state and municipal tax receipts from all sources excluding federal transfer payments, divided by the adult population. Under Ricardian Equivalence theories, a state's debt load represents anticipated future taxes. Since we sought to test whether the total tax burden attracts or repels migrants, we added a Debt variable, representing outstanding state indebtedness as at 1985. The mean Tax figure is 11.613, wit a standard deviation of 0.4722; the mean Debt figure is 1.372, with a standard deviation of 0.8922. 4. Bankruptcy Deadbeats. As seen in the previous Section, American provinces and states appeared to compete for migrants through lax insolvency laws in the eighteenth and nineteenth centuries. Such laws were found to be unconstitutional by the end of the nineteenth century, and a lasting national bankruptcy law adopted in 1898. However, the competition for bankruptcy deadbeats might still continue, through differences in the way federal judge interpret the federal law. We suggest that bankruptcy judges might seek to attract (repel) deadbeats by weakening (strengthening) good faith barriers to debtor opportunism. With weaker (stronger) good faith barriers will come higher (lower) consumer bankruptcy filing rates. On deadbeat theories, then, a filing rate variable might predict migration flows. While consumers may petition themselves into bankruptcy under several different chapters, we restricted our study to Chapter 7 filings. We believe that Chapter 13 plans, which consign a portion of future earnings to pay off present debts, are less likely to be tainted by debtor opportunism than Chapter 7 plans. For this reason, we excluded Chapter 13 petitions, along with the less frequently employed consumer petitions under Chapters 11 (reorganizations) or 12 (farmers). As a measure of state filing rates, one might select either 1985 filings or 1985-90 total filings. While the first choice minimizes endogeneit concerns, it misses the spectacular run-up of consumer filings over the period of the migration study. For this reason, our Chapter 7 variable reflects total filings from 1985-90. Since children do not go bankrupt, we arrived at our per capita filing rate by dividing the adult population into the total filings figure. We then addressed the endogeneity problem through a Three-Stage Least Squares system of equations procedure, estimating the Chapter 7 filing rate through independent instrumental predictors. A Hausman test confirmed our suspicions that the Chapter 7 variable, as well as the Family Deadbeat predictor, were endogenous. The move to 1985-90 filings assisted in another way. Bankruptcy filing rates might ordinarily be thought to signal a depressed economy more than a lax legal regime. If so, we would expect bankruptcy predictors to be negatively correlated with migration flows. During the 1985-90 period, however, consumer bankruptcy filing rates seemed divorced from economic reality. The threefold run-up in filing rates largely overlapped with th seven fat years of the Reagan economic recovery. Our claim, then, is that the Chapter 7 variable is a legal more than an economic predictor, and that during the period of our study it measured lax good faith norms more than a depressed economy. To evidence this, we conducte an econometric study of Chapter 7 filing rates, with the results reported in Table II. Our principal finding, that economic predictors explain little of the variance in Chapter 7 filing rates from 1985 to 1990, is consistent wit our assumption that the Chapter 7 variable is primarily a deadbeat one. We employed eight instrumental variables to predict Chapter 7 Filing Rates in our Three-Stage Least Squares procedure: Metropolitan, Construction Starts, Unemployment, AFDC, Employment Growth, Gross State Product, Exemptions, and Divorce. Employment Growth was the per cent change in non-farm employment in a state from 1984 to 1985. The Gross State Product figure was arrived at by divided the total figure by the adult population Exemptions represents our estimate of the value of state law exemptions from bankruptcy for unsecured creditors. The Divorce predictor represents the average number of divorces per 1,000 marriages. Relying on economic predictors is straightforward, since one would expect a greater number of bankruptcies in an economic downturn. For similar reasons, we would expect higher bankruptcy rates where the social safety net, as measured by AFDC payments, is more generous. With higher exemption levels, bankruptcy might also be more attractive to some debtors. Finally, we would also expect divorce and bankruptcy rates to be correlated, since some bankruptcies result from divorce. 5. Family Deadbeats. We suggest that deadbeat migrants might also be attracted to jurisdictions which permit them to scale back family obligations. Because states differ widely in the extent to which they enforce support obligations and subsidize collection efforts, we chose the percentag of unpaid child support orders as our Family Deadbeat variable. Since data for prior years was unavailable, we used 1990 figures. Such unpaid debts were as high as 85% in Florida and as low as 0% in Iowa. The recent example of Virginia shows how sensitive collection rates are to state collection efforts. In the early 1980s, Virginia ranked thirteenth in the child support collection. But after state funding was cut in 1986, the state fell to forty- ninth place. After this became a political issue in 1989, Virginia collection rates improved dramatically. Since our use of 1990 data introduced endogeneity concerns, we separately estimated Family Deadbeats in our Three-Stage Least Squares equation, using ten instrumental variables: Metropolitan, Unemployment, AFDC, Divorce, Unwed Births, Under 18, Child Abuse, Infant Mortality, Working Wives, and Joint Custody. We might expect parents to be less likely to comply with enforcement orders in high Unemployment states. In addition, the temptation to default on personal support obligations would presumably be stronger when AFDC payments are relatively high, and there is a greater incentive to share support duties with the state. We would also expect to see a larger number of family deadbeats in high-divorce states, on the assumption that those more likely to breach marriage vows are more likely to cheat their families in other ways. Similarly, we would anticipate a positive correlation between family deadbeat variables and the Unwed Birth rate, measuring the percentag of births to unwed mothers. A man who is unwilling to marry the mother of his children is less likely to support them financially. We might also expect to see a positive correlation between family deadbeats and the Child Abuse rate, since a disregard for a child's physical wellbeing might translate into fiscal disregard. On the positive side, we would expect to see fewer family deadbeats in states which prescribe joint custody orders for children on divorce. Noncustodial parents are less likely to pay child support since it is harder for them to monitor the way their payments are spent. As well, the more time one spends with a child, the stronger the attachment and the greater the willingnes to pay child support. Our Joint Custody Dummy variable took the value of if the relevant family law statute promoted joint custody orders, and 0 if not. We also included Under 18, the under-18 population in the destination state. The greater the ratio of children, the more they are prized. More children might also mean better collection mechanisms, assuming economies of scale. With more children, then, we might expect fewer family deadbeats. Other variables, such as Metropolitan, might cut either way. We might perhaps expect to see a positive correlation between the Family Deadbeat variable and the Infant Mortality rate. Many studies have reported a relationship between a parent's attachment to his children and the likelihoo that they will survive to adulthood. On the other hand, we may be measuring different parents here. Infant mortality is more closely related to maternal behavior, while family deadbeat variables generally measure behavior by fathers. The Working Wives variable, measuring the percentage of married women in the labor force, is also ambiguous. We might expect more family deadbeats with a higher Working Wives figure, since women will have a stronger incentive to work when their husbands fail to support them. However, working women ar less likely to need spousal support, have more power within individual marriages, and are more apt to share child rearing responsibilities with their spouse III. Results A. Migration Flows Our empirical results are found in Table I, which reports on a system of regression equations in which the values of the family and bankruptc variables, as well as the migration flow to particular states, are jointly estimated. Average immigration into a state is the dependent variable in the first equation, which predicts about 30% of the variance in migration flows. Equation 2, in the second column, estimates Chapter 7 Bankruptcy Filings, and explains more than half of the variance in state bankruptcy rates. The third column estimates the Family Deadbeat variable. Our principal result is that 1985-90 migration flows are consisten with deadbeat theories. Both predictors are significant and positive, suggesting that deadbeat laws attract migrants. This finding is particularly interesting for Chapter 7 Bankruptcies. Even in the United States, bankruptc filings will to some extent reflect economic conditions, with higher rates observed in a depressed economy. We might then expect migrants to be deterred by high bankruptcy rates, like high unemployment rates. However, the sign of the Chapter 7 Bankruptcies coefficient is positive, while that for Unemployment is negative. For migrants, the positive deadbeat effect appears so strong that it swamps the negative economic signal. To say that deadbeat variables explain migration patterns does not mean that states seek deadbeat migrants. In particular, good faith norms i bankruptcy are set by federal judges and not by state politicians. If filings are high in a particular state, then, this is not because the state has sought to increase immigration. At most, the bankruptcy judge may have sought to increase the level of bankruptcy litigation. Alternatively, courts might simply differ as to interpretation to be given to good faith standards. The results as to other migration predictors are largely those one would have expected. Of the environmental predictors, Distance is significant and negative. Given differential moving costs, migrants not surprisingly prefer short to long moves. Temperature is positive but not significant suggesting that any migration trend to the sunbelt is no longer attributable simply to the climate. At first, we thought that the contribution of the Temperature variable might have been masked by the Elderly variable. However, the correlation between the two coefficients was only -0.093. The Elderly coefficient was not significant, suggesting that retiree migration decision resemble those for non-retirees. The significant, negative coefficient for Metropolitan suggests a trend away from urbanized areas in the late 1980s. The significant, positive coefficient for Construction Starts was expected, with migrants moving to states with booming economies. Surprisingly the Unemployment predictor was not significant, though it had an expected negative sign. We might perhaps have seen a significant, negative coefficient for Unemployment had we used 1985-90, though this would have raised endogeneity concerns.Table I. Interstate Migration Equation 1 (Dependent Variable- Interstate Immigration/Pop) Equation 2 (Dependent Variable -Chapter 7 Bankruptcies) Equation 3 (Dependent Variable- Family Deadbeat) Variable Nam Coef. T- Statisti c Coeffic ien T Statist ic Coeffic ien T- Statist i Chapter 7 Bankruptcies 0.44403 5.31 Family Deadbeats 0.21199 2.456* Distance -0.9521 3 -28.74* Temperature 0.27433 1.592 Metropolitan -0.5000 -6.857* 0.09822 7 4.156* -0.2042 5 -3.487* Elderly 0.18959 0.9031 Constructio Starts 1.1194 8.375* 0.3346 7.82* Unemployment -0.2170 1 -1.173 0.952 19.53 -0.8862 8 -9.344* AFD 0.51233 5.239* 0.18418 5.972* 0.33605 3.704* Ta -0.5160 9 -4.408* Debt 0.0208 8 0.4092 Employment Growth -706.71 -11.59* Gross State Product 0.1242 1.71 * Exemptions -0.0001 -2.044 Divorce 0.80809 17.86* 1.1569 11.15* Unwed Births 1.0818 7.7* Under 1 -0.0887 -0.2915 Child Abuse -0.0727 9 -1.301 Infant Mortality 0.04590 7 0.1834 Working Wive -0.818 -4.414* Joint Custody Dummy -0.427 6 -9.598* Constant 2.5872 2.774* 325 11.58* 3.102 1.943 R2 0.2979 0.528 0.126 System 2 2830 NOTES: Number = 2162. Equation 1's dependent variable is total migration from State I to State J divided by State J population. The independent variables are provided for State J. Chapter 7 Bankruptcies are personal bankruptcy filings under Chapter 7 of the U.S. Bankruptcy Code divided by the adult population. Family Deadbeats represents the per cent of unpaid family support obligations. Distance is the driving distance between state capitals, and Temperature is the average mean January high temperature from 1960-90. Metropolitan is the per cent of people in a state living in an urban area, and Elderly is the per cent of the population over 65. Construction Starts is the value of commercial and residential construction contracts for projects completed within the year, divided by adult population. Unemployment is th average monthly unemployment rate. AFDC is the average payment under the Ai for Families with Dependent Children program. Tax is state taxes from al sources, and Debt is total state indebtedness, both divided by adult population. Employment Growth refers to per cent non-agricultural job growth from 1984 to 1985. The Gross State Product figure is divided by adult population. Exemptions represents an estimate of the value of state law exemptions from seizure by unsecured creditors on bankruptcy. Under 18 is the per cent of the population under the age of 18, and Child Abuse is the total number of reported child abuse cases divided by the under 18 population. Working Wives is the per cent of wives who are employed. The Joint Custody Dummy takes the value of 1 if state law promotes joint custody, and otherwise. All variables have been transformed into their natural logs except for the Joint Custody Dummy. T-statistics which are significantly different from zero at the 05% level (two-tailed test) are denoted with an * The significant, positive coefficient for AFDC suggests that som migration decisions are motivated by the prospect of increased welfar payouts. The Tax coefficient was significant and negative, as expected, though the Debt coefficient was positive and insignificant. There are at least two possible explanations for this latter result. First, high debt states might offer higher levels of public spending, and this might attract welfare- loving migrants. States might sort themselves out into high and low tax jurisdictions, each competing for a different set of migrants, with the two effects cancelling out. On this analysis, it is interesting that the Tax and AFDC predictor remained significant. The second possible explanation for the insignificant Debt predictor is that anticipated future taxes will b reflected in land values. Thus migrants to low tax states would have a higher after-tax income, but would pay more for their houses. To the extent that this happens, the incentive to move to take advantage of lower taxes is weakened. Once again, therefore, the significant, negative Tax coefficient is striking, and consistent with electoral explanations of migration flows. B. Chapter 7 Bankruptcies In Equation 2, our instrumental predictors explain more than half of the variance in Chapter 7 filing rates. As expected, the coefficients for Unemployment, Divorce and AFDC are significant and positive. Those who are unemployed or recently divorced are more likely to be in financial distress, while a high AFDC payout might make default more attractive. The significant, negative coefficient for Employment Growth was also expected. We might have expected the Exemptions coefficient to be positive, since broader exemptions might take the sting out of bankruptcy. However, the Exemptions coefficient was significant and negative, and higher exemptions were associated with lower bankruptcy filing rates. The most plausible explanation for the absence of a positive correlation is that tax policies bias debtors against relying on high homestead exemptions in states where this is permitted. The exemption does not avail against secured lenders, and most debtors will have secured their homestead to take advantage of home mortgage interest deductibility. We might have expected the coefficients for Construction Starts and Gross State Product to be significant and negative, on the assumption that a booming economy will be associated with a lower bankruptcy filing rate. However, the coefficient for both variables was significant and positive. There are two possible explanations for this result. First, if Chapter 7 Bankruptcies is a deadbeat and not an economic variable, we should not be surprised to see higher filing rates during prosperous times. Second, the anomalous findings might result from our use of 1985 data, While this reduced endogeneity concerns, it also weakened our model, since we were asking a 198 variable to predict bankruptcy filing rates for the following five years. Since our findings hinge on our assumption that Chapter 7 Bankruptcies is a deadbeat variable, we sought to distinguish between these two explanations. We therefore estimated annual bankruptcy filing rates through the Ordinary Least Squares regression presented in Table II. Unlike Equation 2 of Table I, the dependent variable in Table II is consumer bankruptcy filings for each year from 1985 to 1990 divided by adult population, rather tha average filings for the entire period. Instead of using 1985 figures for ou independent variables, then, we provided separate figures for each year. Only the Exemptions variable is unchanged, since exemption policies remained the same during the period. The table also includes a time trend Year, with different value for each year of the period Our principal result in Table II is that the variance in the Chapter 7 Bankruptcy filing rates is not attributable to economic factors. This is consistent with our hypothesis that the variance results from the difference in legal regimes. Of the economic predictors, only Unemployment is significant in regressions (1) to (3), and by itself in regression (4) explains only 8.6% of the variance in filing rates. By contrast, the Divorce predictor and the time trend Year better explain filing rates. The welfare predictor AFDC is significant in regressions (1) and (2), suggesting that debtors are more willing to run the risk of default in regimes with a safety net for them or their family. The Exemptions variable was not significant, likely for the reasons discussed above. These results are consistent with other studies of American consumer bankruptcy filing rates Our study says nothing about the determinants of bankruptcy prior to 1985. Other studies, using macroeconomic and demographic predictors, hav explained considerably more of the variance in pre-1985 filing rates than w were able to do in our study. We believe that the reason for this is that the decisive legal shift came after 1985, through a relaxation of good faith norms, and not in 1980 with the introduction of the new Bankruptcy Code. We did not employ predictors such as personal leverage ratios for which state data was unavailable. We are in any event highly suspicious of personal leverage ratios as a predictor of consumer bankruptcy. On a shift towards a laxer legal regime, rational debtors will respond by increasing personal leverage ratios. We should therefore expect filing rates and leverage ratios to be strongly endogenous.Table II. Consumer Bankruptcy Filing Dependent Variable: Annual Consumer Bankruptcy Filing Rates Independent Variables (1) (T- statistics) (2) (T- statistics) (3) (T- statistics ) (4) (T- statistics) (5) (T- statistics) Year 0.206 (10.77)* 0.208 (12.15) 0.226 (11.88)* 0.159 (7.65)* Constructio n Starts -0.041 (-0.36) Unemploymen t 0.798 (6.09)* 0.811 (7.58)* 1.0 (10.17)* 0.62 (5.29) Employment Growth 0.089 (0.26) Gross State Product 0.013 (0.08) AFDC 0.330 (3.50) Divorc 1.0 (7.66)* 0.983 (8.74)* Exemptions -0.023 (-0.81) Constant -5.5 (-3.07) -5.21 (-7.87)* -2.38 (-10.46)* -0.62 (-2.94)* -0.238 (-2.38) R2 (adjusted) 50.8% 51.4 38.7 8.6% 0.16 NOTES: N=288. The dependent variable represents total consumer Chapter 7 filings for each year from 1985 to 1990 divided by the adult population for that year. Year represents each year during that period. The other independent variables are defined in the same manner as in Table I, except that separate values are given for each year from 1985-90 in this Table. T-statistics which are significantly different from zero at the 0.05 level (two-tailed test) are denoted by an *. C. Family Deadbeats There were few surprises amongst the Family Deadbeat predictors in Equation 3. The Unwed Births and Divorce coefficients were both significant and positive. A spouse who is faithless in one respect would appear to be faithless in others as well, and states would seem to sort themselves out along a continuum of promise-keeping and promise-breaking. As expected, the Joint Custody Dummy was negative and significant. Parents would appear more likely to support their children when they maintain close contact with them through joint custody orders. The Working Wives coefficient was also significant and negative. Husbands whose wives are working spend more time with their children and are likely to have a closer attachment to them after divorce. As well, a working wife is less in need of support. This might also explain why the AFDC predictor is significant and positive. Non-working wives are more likely to be abandoned when deadbeat dads can offload suppor obligations onto the state. The insignificant Metropolitan coefficient suggests that family deadbeats are not a particularly urban phenomenon. Other results were less expected. We did not observe a link between goo faith and the Under 18 ratio, or between bad faith and violence against children. Neither the Infant Mortality nor the Child Abuse coefficients were significant. We did not expect the Unemployment predictor to be significan and negative, since it is harder for the unemployed to keep up suppor payments. On the other hand, unemployed deadbeats are easier to trace, and their welfare benefits are easier to garnish. Conclusion This Article outlines three explanations for why states seek migrants, and tests them by reference to 1985-90 interstate migration. On race for the top theories, states compete for value-increasing migrants by offering them healthy economies and efficient laws. On electoral theories, states compete for clienteles of voters, with some states seeking to attract and some to deter welfare- or tax-loving migrants. On deadbeat theories, states compete for high human capital debtors by offering them a fresh start from foreign creditors. Our findings support deadbeat and electoral migration theories. We ourselves are migrants, as are most of the people we know. Fellow migrants seem to us more interesting, and perhaps a little brighter than the few native Virginians to be found in Arlington and Alexandria. As we grow older, however, we find we increasingly prefer natives to migrants. THE MARKET FOR DEADBEATS Margaret F. Brinig & F.H. Buckley George Mason University School of Law 3401 North Fairfax Drive Arlington, Va. 22201 Internet: mbrinig@gmuvax.gmu.edu, fbuckley@gmuvax.gmu.edu Abstract: This Article outlines three explanations for why states seek migrants, and tests them by reference to 1985-90 interstate migration flows. On race for the top theories, states compete for value-increasing migrants by offering them healthy economies and efficient laws. On electoral theories, states compete for clienteles of voters, with some states seeking to attract and some to deter welfare- or tax-loving migrants. On deadbeat theories, states compete for high human capital debtors by offering them a fresh start from foreign creditors. Our findings support electoral and deadbeat theories. October 15, 1994 Please do not quote without permission. Comments welcome. This Article was written with the generous help of the George Mason University School of Law. We thank Doug Allen, Lloyd Cohen, Esther Goldberg, David Levy, Larry Ribstein and Larry Weiss for their helpful comments. Previous drafts of this Article were presented at INSEAD, Fountainebleau, France, and at the Canadian Law and Economics Association. Table III. Bankruptcy, Family Support and Exemptions State Nam Chapter Seven Filings 1986-9 / Adul Percentage of Child Support Noncompliance Estimated Value of Exemptions from Bankrupt Estate Alabama 14.1153 44 1835 Arizona 18.6008 50.3 13500 Arkansas 8.4991 47.4 51600 California 17.1746 59.6 106525 Colorad 19.5885 42.1 18350 Connecticut 5.21 56.6 18350 Delaware 5.4014 41.3 18350 Florida 9.1643 84.7 18350 Georgi 14.2022 50.7 18350 Idaho 18.631 55. 15550 Illinois 14.6459 51 10000 Indiana 21.1031 * 11450 Iowa 9.9433 0.1 18350 Kansas 15.399 53.4 58700 Kentucky 15.6882 63.4 200000 Louisiana 14.6504 35.7 1835 Main 3.954 47.3 8850 Maryland 7.342 93.6 90200 Massachusetts 3.1685 60.5 200000 Michigan 8.3779 30 17000 Minnesota 11.5748 22.3 200000 Mississippi 14.937 64.1 18350 Missour 11.077 37.5 8500 Montana 11.57 69.3 28000 Nebrask 10.4709 45.5 14950 Nevada 25.4323 51 22750 New Hampshire 4.4736 13.8 18350 New Jerse 4.886 46.8 7800 New Mexico 12.4229 8. 200000 New York 6.8369 43.3 5300 North Carolina 3.0382 38.7 10000 North Dakota 8.9958 51.1 8000 Ohio 14.5486 50. 85000 Oklahom 22.6885 79.6 18350 Oregon 17.9202 51.5 35000 Pennsylvania 5.6742 31.2 18350 Rhode Island 6.5813 53.5 20000 South Carolina 3.8423 41.1 52700 South Dakota 7.8 55.1 29000 Tennessee 15.1011 69.8 1700 Texas 8.5388 15 113300 Utah 19.8464 51.7 43500 Vermont 3.0405 61 137400 Virginia 10.731 43.4 1835 Washington 18.8447 60.2 105500 West Virginia 9.1561 79.1 43700 Wisconsin 10.4212 * 27100 Wyoming 17.2425 80. 531 ENDNOTES * Professor of Law, George Mason University. ** Professor of Law, George Mason University.