CONSUMPTION, DEBT AND PORTFOLIO CHOICE (cid:3) Testing the Effect of Bankruptcy Law Andreas Lehnert Dean M. Maki Board ofGovernorsofthe Putnam Investments FederalReserve System OnePostOfficeSquare MailStop 93 BostonMA, 02109 WashingtonDC,20551 (617)760-8616 (202)452-3325 Dean [email protected] [email protected] This version: February 2002 February 20,2002 The views are expressedin this paper are oursalone and do not necessarily reflectthose of (cid:3) the Board of Governorsor its staff or Putnam Investments. We thank DarrelCohen, Karen Dy- nan,RonelElul,GaryEngelhardt,DavidLaibson,NickSouleles,RobertTownsendandseminar participants at Syracuse University, the NBER 2000 Summer Institute and the Federal Reserve Board.WereceivedexcellentresearchassistancefromSigurdLund,RichardSauomaandMarcin Stawarcz.Anyremainingerrorsareourownresponsibility. CONSUMPTION, DEBT AND PORTFOLIO CHOICE Testing the Effect of Bankruptcy Law Abstract Consumerbankruptcylaws,whichvaryacrossstatesandovertime,permitdebtors tokeepassetsbelowastatutoryexemptionwhiledebtsareforgiven. Highexemp- tions distort household portfolio decisions and tempt households to default on debts; but they also provide a crude form of consumption insurance. We com- bine information on state-level bankruptcy laws with the Consumer Expenditure Survey from 1984–1999. We find that higher exemptions are associated with (1) Higher bankruptcy rates, (2) Households that are more likely to simultaneously hold low-return liquid assets and owe high-cost unsecured debt, and (3) Slight- ly betterconsumptioninsurancefor renters and worse consumptioninsurancefor homeowners. Journalof EconomicLiteratureclassificationnumbers: H73,H31,K00, D1. Keywords: Bankruptcylaw,householddebt, portfoliopuzzle, consumption. 1 Introduction IntheUnitedStates,consumer(Chapter7andChapter13)bankruptcyisdesigned to provide debtors a “fresh start.” After a household successfully files a Chapter 7 petition, its unsecured debts are erased, but it must forfeit any assets above an exemption level determined by law. Although the United States constitution specifically grants the Federal government the power to set national bankruptcy law,inpracticetheselawsaremostlysetbytheindividualstates.1 Stateconsumer bankruptcy laws vary most dramatically in generosity–i.e. the exemption levels abovewhichhouseholdsforfeitassets. SomestateChapter7bankruptcycodesare extremely generous, allowing households to keep an unlimited amount of assets after bankruptcy, while others are relatively stingy, allowing households to keep, forexample,only$75in assetsafterbankruptcy. Bankruptcy laws of this type provide households with a crude form of insur- ance. If householdsface the possibilitythat their non-capitalincomecould dip to zero for an extended period, they would be unwilling to use unsecured debt un- less somehow insured. Carroll (1992) uses the PSID to confirm that households do face such a risk, suggesting that in the absence of other forms of insurance, households would be less willing to use unsecured debt in states with less gener- ousbankruptcylaws,althoughlenderswouldbemorewillingtoextendunsecured debtinsuchstates. Further,ahouseholdwithrelativelylargedebtswouldcutcon- 1Article I, Section 8 of the U.S. constitution states “The Congress shall have the power ... [t]o establish a uniform rule of naturalization, and uniform laws on the subject of bankruptcies throughouttheUnitedStates.” ForinformationonconsumerbankruptcylawsoutsidetheUnited States,seeAlexopoulosandDomowitz(1998)andZiegel(1997). 1 sumptionmorein responseto an incomeshockinless generousstates. Atthesametime,though,Chapter7bankruptcylawsdistorthouseholds’port- foliochoices. Householdsthatfileforbankruptcyarebetteroffiftheyhaveassets right up to the exemption level set in law. They carry these assets with them in- to their post-bankruptcy life, while their debts (largely) vanish. Thus it is in their interesttosimultaneouslyholdlow-returnliquidassetsevenwhiletheyhaveasig- nificant amountofhigh-interestdebt. Morrison(1999)and Bertaut and Haliassos (2001) have documented the existence of this anomaly using the Survey of Con- sumer Finances. Bertaut and Haliassos present evidence that the portfolio puzzle may be driven by self control, in which the household may be thought of as di- videdbetweentwodecisionmakers,aworkerandashopper. Theworkerchooses nottopay offthehousehold’scredit cards as away to restraintheshopper.2 Finally,allelseequal,householdswillbemorelikelytodeclarebankruptcyin stateswithmoregenerousexemptions. Ofcourse,lendersmayreact toprevailing bankruptcy laws by restricting credit to borrowers living in states with generous laws, so the net effect of bankruptcy law on bankruptcy rates may go in either direction. To test these effects, we collected data on state personal bankruptcy exemp- tions, and other state-level information, from 1984–1999. We matched these da- ta with household-level responses from the Consumer Expenditure Survey (CE) over the same period. The CE contains detailed information about households’ consumption, along with some information about their geographic location, de- 2SeealsoLaibson,Repetto,andTobacman(1998)andHarrisandLaibson(2001). 2 mographic characteristics, finances, income, occupation, employment and health status. The CE asks in detail about holdings of different asset classes (for exam- ple, securities, checking accounts, saving accounts, U.S. savings bonds), and less detailed information about unsecured debt. We use this portfolio information to test whether households in generous bankruptcy law states are more likely to si- multaneously hold low-return liquid assets and high-interest unsecured debt. We refer tothispracticeas as “borrowingtosave.” The CE interviews the same household once per quarter for five quarters; the first interview is excluded from the public use microdata, but all other interviews are available. Thus it is possible to construct a short panel for each household, testing how consumption responds to shocks of various types. We use this infor- mationto testtheinsuranceroleofbankruptcylaw. We find: (1) bankruptcy rates are higher in states with higher bankruptcy ex- emptions;(2)householdsaremorelikelytoborrowtosaveinstateswithgenerous laws;and(3)theconsumptionofrentersisslightlylesssensitivetoincomeshocks in states with generous bankruptcy laws while, by contrast, the consumption of homeowners is slightly more sensitive to income shocks in states with generous bankruptcylaws. The plan of this paper is as follows: in section 2 we briefly review the previ- ous literature and the debate surrounding bankruptcy law, while in section 3 we provide a theoretical framework for our analysis. In section 4 we describe our data sources and the construction of the bankruptcy law database and generosity measures and in section 5 we present our results. Section 6 briefly concludes. 3 An appendix provides the results from alternative specifications of our statistical models and further discussionof our dataset. All tables and figures are located at theend ofthepaper, followingthereferences. 2 The Bankruptcy Debate Much of the current debate over bankruptcy reform is prompted by the extreme- ly high bankruptcy filing rate in the late 1990s.3 Little is known about why the bankruptcyratehasincreased;forexample,usingadatasetofcreditcardrecords, GrossandSouleles(1998)showthattheincreaseindelinquencyratesthroughthe 1990s cannot be explained by observable factors; they posit a decrease in stigma duringthedecade as thecause. 3Thebankruptcyratealsospikedinthefirsthalfof2001.Thismostrecentincreasewaslikely a response to Congressional passage of a comprehensivebankruptcy reform bill. Although the Presidenthassaid hewouldsign thebill, asof thiswriting, a conferencecommitteehadnotyet mettoresolvethesignificantdifferencesbetweentheHouseandSenateversions. 4 Personal Bankruptcy Filings per 100,000 Persons 600 500 400 300 200 100 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 Total U.S. personal bankruptcy filings (Chapter 7 and Chapter 13) per 100,000population, annualrate. Theliteratureonconsumerbankruptcy(and,byextension,consumerdebt)can be roughly divided into two camps. One the one hand, certain authors argue that householdsarenotstrategicintheiruseofbankruptcylaw,whileothersarguethat households value bankruptcy law the same way they would any other financial option, and act strategically given the actions of lenders and the prevailing law. One might characterize the former view as emphasizing the sociological aspects ofconsumption,debtandbankruptcyandthelatterasemphasizingtheireconomic aspects.4 Note that under either view, prevailing bankruptcy law can be critiqued as eithertoo laxortoopunitive. 4Thischaracterizationisabitofanexaggeration,asrecenteconomicresearchhasmovedaway fromtheprevailingmodelofaunitaryrationaldecisionmaker.SeeforexampleHarrisandLaibson (2001). 5 Sullivan, Warren and Westbrook (1989, 1997) advance a theory of consump- tion and bankruptcy in which a beleaguered middle class, overwhelmed by debt, uses bankruptcy as an insurance mechanism. In this view, most households are not strategic in their use of debt and bankruptcy; instead, they are seen as unfa- miliar with the prevailing bankruptcy law in their state. Moreover, in this view, legal culture plays an all-important role in the outcome of a bankruptcy petition. Sullivan, Warren and Westbrook (1997), for example, compare outcomes across district courts in Pennsylvania and find wide disparities, despite having identical statutes. Further, Nelson (1999) finds that legal culture is one of the most impor- tantfactorsaffectingwhetherhouseholdsfileunderChapter7orChapter13ofthe bankruptcycode. Thesestudiescan betaken asevidencethatchangingbankrupt- cylawswouldhavelittleeffectonhouseholdbehaviorsolongasother,intangible culturalfactors remained constant. Gross (1997) also explicitly rejects the rational actor model, instead arguing that most households sincerely wish to meet their obligations and avoid harming theirlocalcommunities. Fromthispremise,shesuggests,tochooseoneexample, thatbankruptcylawsoughttoaimforequalityofoutcomesratherthanequalityof treatment.5 In the context of bankruptcy law, this could mean that all households areabletosustainroughlysimilarconsumptionlevelsafterbankruptcy,nomatter howdifferenttheirdebtsbeforebankruptcy. 5Notalllegalscholarstakethisapproach;BairdandMorrison(2001)usearealoptionsframe- workto discussreformof corporaterestructuringbankruptcylaw; see also Denning, Ferris, and Lawless(2001)foranempiricalanalysisofcorporatebankruptcy. 6