Does Bankruptcy Protection Affect Asset Prices? Evidence from changes in Homestead Exemptions
DDoes Bankruptcy Protection Affect Asset Prices?Evidence from changes in Homestead Exemptions ∗ Yildiray Yildirim and Albert Alex Zevelev † BaruchOctober 15, 2019
Abstract
Does the ability to protect an asset from unsecured creditors affect its price? Thispaper identifies the impact of bankruptcy protection on house prices using 139 changesin homestead exemptions. Large increases in the homestead exemption raised houseprices 3% before 2005. Smaller exemption increases, to adjust for inflation, did notaffect house prices. The effect disappeared after BAPCPA, a 2005 federal law designedto prevent bankruptcy abuse. The effect was bigger in inelastic locations. ∗ All errors are our own. † Email: [email protected], [email protected] a r X i v : . [ ec on . GN ] F e b Introduction
Bankruptcy is one of the largest social insurance programs in the US. Americans dischargemore debt (formally and informally) than all unemployment benefits combined. The sharprise in personal bankruptcy from .3% of households annually in the 1980s to 1.5% in theearly 2000s raised concern about strategic behavior and motivated the 2005 BankruptcyAbuse Prevention and Consumer Protection Act (BAPCPA). The bill raised the costs andreduced the benefits of filing for bankruptcy.A large literature studies strategic bankruptcy. Pattison and Hynes (2019) found thatrises in homestead exemptions are followed by a rise in chapter 7 filings by debtors with homeequity. Helland, Jena, Ly, and Seabury (2016) found that in states with unlimited homesteadexemptions physicians invest 13% more in their homes compared to other professionals withsimilar income and demographics. Additionally, the response of physicians to unlimitedhomestead exemptions is larger in areas with higher liability risk. While the literature hasfound that households prefer to own assets which give them greater protection, we studywhether the market price of these assets reflects this protection.There are several advantages of using house prices to detect strategic behavior. First,studies that use formal bankruptcy filings do not capture a lot of strategic behavior sincethe majority of defaulting consumers do not file for bankruptcy, and most debt collectiontakes place outside of the courtroom (Dawsey, Hynes, and Ausubel (2013)). However, thesettlement negotiated between creditors and debtors outside the courtroom is influenced bylocal exemption laws under the “threat-point” of bankruptcy. Second, changes in houseprices in locations with a rise in homestead exemptions can help quantify forward lookingstrategic behavior via demand for assets that provide this implicit insurance.This study uses annual house price data in 55,316 census tracts from the FHFA com-bined with 139 changes in homestead exemptions between 1990-2017, collected from otherauthors, legal guide books (Elias, Renauer, and Leonard (1989)), and legal statutes. Theidentifying assumption is that changes in homestead exemptions are uncorrelated with un- Lefgren, McIntyre, and Miller (2010); Auclert, Dobbie, and Goldsmith-Pinkham (2019); Pattison andHynes (2019). Gross, Kluender, Liu, Notowidigdo, and Wang (2019); Albanesi and Nosal (2018). Mahoney (2015); Skeel (2003); Pattison (2017). We thank Mariela Dal Borgo, Richard Hynes, Paul Goldsmith-Pinkham, and Jeffrey Traczynski forsharing their data. We thank Albert Levi for helping us search statutes. . . , .
82% and is not statistically significantfor small changes. Finally, big changes Pre-BAPCPA raise house prices 3 . onhouse prices indicating an asymmetric effect. In addition, census-tracts in MetropolitanStatistical Areas with relatively inelastic housing supply had experienced bigger effects,consistent with standard theory. Moreover, tracts in counties with higher pre-treatmentunemployment rates had smaller effects. This doesn’t mean that households in countieswith higher unemployment rates don’t value bankruptcy protection, but rather that thetypes of households who strategically protect their assets from creditors tend to be moreprosperous.Finally, falsification tests find that changes in the homestead exemption don’t affectlevels and changes in unemployment rates, income per capita, and single family buildingpermits. There is a small drop in population levels and homeownership rates. The dropin homeownership rates suggests that the rise in house prices caused by wealthier strategichouseholds reduced housing affordability. In 1993, Minnesota reduced its homestead exemption from unlimited to $ Institutional Setting
This section describes the institutional setting and relevant details of the bankruptcy sys-tem in the United States. If a debtor defaults on secured debt the creditor can seize thecollateral. If there is a deficiency and recourse is possible then the debtor is personallyliable for the deficiency. If a debtor defaults on unsecured debt (e.g. credit card debt,legal judgment, student loan debt) the creditor can seize the debtor’s non-exempt assetsand income. However, asset and income exemptions (i.e. wage garnishment limits) dependon local and federal laws which vary over time and space.For example, suppose Ann is a homeowner with a house worth $ $ Heq i,t = $120 ,
000 in home equity. Ann lives in a state with ahomestead exemption of H i,t = $50 , Heq i,t − H i,t =$70 ,
000 in unprotected home equity. Suppose Ann defaults on $ $ $ H i,t = $50 ,
000 of her exempt homeequity, the unsecured lender will receive the remaining $ $ $ $ Heq i,t = $120 ,
000 all her home equity would be protected, and her creditors would knowthis as well.The homestead exemption has been studied extensively in the history, legal, and eco-nomics literature. The first homestead exemption was incorporated into statutory and con-stitutional law in the Republic of Texas in 1839 (London (1954)). Following the example ofTexas, Mississippi passed the second homestead exemption law in 1841. Over the next few A deficiency is when the collateral is worth less than the debt balance. Recourse is possible when it is legal and not restricted in the original debt contract. The deficiency on the secured debt becomes unsecured debt. hasargued that changes in homestead exemptions arise from a legislative process that dependon idiosyncratic historic and geographic factors, a process that does not depend on states’economic conditions.In addition to the history and legal literature, an empirical literature has also stud-ied determinants of changes in bankruptcy protection. Severino and Brown (2017) findthat lagged changes in house prices, medical expenditure, unemployment rates, state GDP,bankruptcy filings, share of democrats, and income do not predict changes in homesteadexemptions. We find further evidence that economic variables don’t predict changes inhomestead exemptions, over a longer sample. Other bankruptcy protection laws, includingstatutes of limitations on debt, Tenancy by Entirety laws, and wage garnishment restric-tions were stable over our sample period. This paper estimates the impact of changes in homestead exemption laws on changes inhouse prices. The main outcome variable, real house price growth, is measured using therecently available Federal Housing Finance Agency (FHFA) census tract data. This datasetcontains 55,316 tracts in the US. Like the S&P/CoreLogic/Case-Shiller home price indices,the FHFA series corrects for the changing quality of houses being sold at any point in timeby estimating price changes with repeat-sales. The dataset only includes tracts and yearswith enough repeat-sales to construct the index. The main treatment variable is the homestead exemption in location i in year t denoted H i,t . We define the homestead exemption as the maximum home equity that is protected: H i,t ≡ max (cid:110) H Local, Married i,t , Fed i,t
Fed
Married t (cid:111) (1) Goodman (1993); Skeel (2003). By 1860, there were seven states with homestead exemptions written into the constitution (in additionto statutory exemption). Six out of seven of these states were in the west. See the Turner hypothesis(London (1954)). Pattison and Hynes (2019); Severino and Brown (2017). Traczynski (2019). For details about this new dataset see Bogin, Doerner, and Larson (2019). This data was collected from other authors (Mariela Dal Borgo, Richard Hynes, Paul Goldsmith-Pinkham, and Jeffrey Traczynski), legal guide books (Elias et al. (1989)), and legal statutes. H Local, Married i,t is the local homestead exemption for married households, 1
Fed i,t is anindicator variable equal to one if local laws allow households to use the federal exemption,and Fed
Married t is the federal homestead exemption level for married households in that year.Hence, H i,t is the maximum amount of home equity a household can legally protect in agiven location at a given time. The non-homestead exemption denoted N H i,t is the sum ofthe vehicle and wildcard exemptions, and is computed the same way as H i,t . The wildcardexemption lets the debtor choose which property to protect such as a vehicle, bank deposits,and art.Data used for controls, heterogeneity analysis, and other outcome variables come fromseveral different sources. Supply elasticity data are available at the MSA level from Saiz(2010). Employment data at the county level are from the BLS. Income data at the countylevel are from the BEA. Population, single family building permits, and homeownershipdata are from the census. The population and permit data are at the county level, whereasthe homeownership rate data are at the MSA level. Median house price data at the zip codelevel and rent data at the MSA level are from Zillow. One must be careful in merging thedatasets since the same zip code can be in more than one county. Each zip code is assignedto the county with the maximum allocation factor (e.g. if 75% of zip code z is in county A and 25% in county B , then zip code z is assigned to county A ). US oil price data are fromthe EIA. US interest rates are constructed as in Himmelberg, Mayer, and Sinai (2005) bycorrecting the 10 year Treasury bond rate for inflation with the Livingston Survey. Nominalvariables are deflated using the CPI for all urban consumers from the BLS as in Glaeser,Gottlieb, and Gyourko (2012). Table 1 presents descriptive statistics and Figure 1 plots median homestead and non-homestead exemption levels in the US 1989-2017. A few stylized facts are immediatelyclear: 1 both the homestead and non-homestead exemptions grew considerably over thissample, 2 the homestead exemption grew a lot more and is currently much higher than thenon-homestead exemption, 3 changes in the homestead exemption are less frequent, occur- Note our data on levels begins 1989 and our data on changes begins 1990. $ $ $ This section presents the main results, the impact of changes in homestead exemptions onreal house price growth. We estimate: y i,t = β H { ∆ H i,t > } + g ( X i,t ) + u i,t (Static) y i,t = (cid:88) k = − k (cid:54) = − η k { ∆ H i,t > } + g ( X i,t ) + u i,t (Dynamic)where the main outcome variable y i,t is real house price growth in tract i in year t , themain treatment variable 1 { ∆ H i,t > } is an indicator for years when tract i experienceda change in the homestead exemption, and X i,t are controls including tract and year fixedeffects, the unemployment rate, population, and income per capita. The dynamic regressionis used for validation, to check whether pre-trends in the static model are parallel, and toinvestigate persistence in the treatment. In addition to using an indicator variable for themain treatment, we re-estimate the main equations using a continuous variable equal tochange in homestead exemptions in the appendix.Table 2 presents the main results. Column 1 finds that an (unconditional) averagerise in homestead exemptions raises real house prices 0 . t ≤ . t > , .
82% and is not statistically significant for smallchanges mostly due to inflation adjustment. Finally, big changes Pre-BAPCPA raise house7rices 3 . H i,t divided by the average big (over $ $ This section investigates heterogeneity in the treatment effect – that is, whether the treat-ments had different impacts in different locations. To study the sensitivity of the effect tovarious observable measures of heterogeneity G i , this paper estimates: y i,t = β H, { ∆ H i,t > } + β H { ∆ H i,t > } × G i + g ( X i,t ) + ε i,t (DDD)In this specification the average treatment effect (ATE) is an affine function of G i ATE ( G i ) = β G, + β G G i The coefficient β G, is the estimated average treatment effect if G i = 0 and β G = ∂ ATE( G i ) ∂G i is the sensitivity of the average treatment effect to a rise in G i .For example, theory predicts that a rise in demand should have a smaller impact onhouse prices in elastically supplied locations where it is easier to build real estate (Figure 6).This corresponds to the hypothesis β Elasticity <
0. The coefficient β Elasticity, is the estimatedimpact of the law change on prices in a hypothetical location where the asset (housing) isin perfectly inelastic supply. 8able 4 investigates treatment effect heterogeneity in positive versus negative changesin the homestead exemption, supply elasticity, pre-treatment unemployment rates, popu-lation, real income per capita, median real house value, home ownership rates, and singlefamily building permits. Pre-treatment variables are set equal to their value in the yearbefore the treatment to make sure they are unaffected by the treatment. The negativechange in homestead exemption in Minnesota in 1993, had a modest effect of -0.31%, how-ever it is not statistically significant. Consistent with theory (Figure 6) supply elasticityattenuates the treatment effect. A 1% higher supply elasticity corresponds to 0.62% smallereffect. The implied treatment effect for a hypothetical city with perfectly inelastic housingsupply is 1.68%. The only other significant source of heterogeneity is the pre-treatmentunemployment rate. Locations with higher pre-treatment unemployment rates had muchsmaller effects. This doesn’t necessarily mean that households in these locations don’t valuebankruptcy protection, it only suggests that these types of households are less likely to bestrategic in protecting their assets (possibly because they don’t have the same access tofinancial advisors and tax attorneys as households in wealthier areas). This section explores the mechanism through which changes in homestead exemptions af-fect house prices. On the one hand, the effect could be driven by a rise in demand forbankruptcy protection. On the other hand, the rise in bankruptcy protection can increaseentrepreneurship affecting local demand. Table 5 examines the impact of changes in thehomestead exemption on alternative outcome variables including levels and first differencesof: unemployment rates, population, real income per capita, single family building permits,and home ownership rates. A rise in homestead exemptions has a small negative effect onpopulation levels (but not changes), and on home ownership rates (but not changes). Levelsand changes in unemployment rates and real income per capita are not affected. Togetherthese results indicate that changes in homestead exemptions likely have a very small, if any,impact on local demand.
Table 6 examines determinants of changes in homestead exemptions using various predic-tors including lagged: real house price growth, unemployment rates, population, real income9er capita, homeownership rates, homestead exemption levels, and non-homestead exemp-tion levels. The only consistent, statistically significant predictor of changes in homesteadexemptions is the lagged homestead exemption level. Locations with high (or unlimited)homestead exemptions are less likely to raise them, compared to locations with low (or zero)exemptions that want to catch up.
A large body of literature studies the impact of bankruptcy protection on bankruptcy filings.These studies do not capture a lot of strategic behavior since the majority of defaultingconsumers do not file for bankruptcy, and most debt collection takes place outside of thecourtroom (Dawsey et al. (2013)). In contrast, we use house prices to quantify demand forbankruptcy protection.We find that the average rise in homestead exemptions raises real house prices 0 . t ≤ . t > , .
82% and is not statistically significantfor small changes, mostly due to inflation adjustment. Big changes Pre-BAPCPA raise houseprices 3 . eferences Albanesi, Stefania, and Jaromir Nosal, 2018, Insolvency after the 2005 bankruptcy reform,Technical report, National Bureau of Economic Research.Auclert, Adrien, Will S Dobbie, and Paul Goldsmith-Pinkham, 2019, Macroeconomic effectsof debt relief: Consumer bankruptcy protections in the great recession, Technical report,National Bureau of Economic Research.Bogin, Alexander, William Doerner, and William Larson, 2019, Local house price dynamics:New indices and stylized facts,
Real Estate Economics
47, 365–398.Dawsey, Amanda E, Richard M Hynes, and Lawrence M Ausubel, 2013, Non-judicialdebt collection and the consumer’s choice among repayment, bankruptcy and informalbankruptcy,
Am. Bankr. LJ
87, 1.Elias, Stephen, Albin Renauer, and Robin Leonard, 1989,
How to File for Bankruptcy (NoloPress Berkeley, CA).Glaeser, Edward L, Joshua D Gottlieb, and Joseph Gyourko, 2012, Can cheap credit explainthe housing boom?, in
Housing and Financial Crisis , 301–359 (University of ChicagoPress).Goodman, Paul, 1993, The emergence of homestead exemption in the united states: Accom-modation and resistance to the market revolution, 1840-1880,
The Journal of AmericanHistory
80, 470–498.Gross, Tal, Raymond Kluender, Feng Liu, Matthew J Notowidigdo, and Jialan Wang, 2019,The economic consequences of bankruptcy reform, Technical report, National Bureau ofEconomic Research.Helland, Eric, Anupam B Jena, Dan P Ly, and Seth A Seabury, 2016, Self-insuring againstliability risk: Evidence from physician home values in states with unlimited homesteadexemptions, Technical report, National Bureau of Economic Research.11immelberg, Charles, Christopher Mayer, and Todd Sinai, 2005, Assessing high houseprices: Bubbles, fundamentals and misperceptions,
Journal of Economic Perspectives
19, 67–92.Lefgren, Lars, Frank L McIntyre, and Michelle Miller, 2010, or 13: Are client or lawyerinterests paramount?,
The BE Journal of Economic Analysis & Policy
The Southwestern HistoricalQuarterly
57, 432–453.Mahoney, Neale, 2015, Bankruptcy as implicit health insurance,
American Economic Review
Virginia Law and Economics Research Paper .Saiz, Albert, 2010, The geographic determinants of housing supply,
Quarterly Journal ofEconomics
Available at SSRN 2447687 .Skeel, David A, 2003,
Debt’s dominion: A history of bankruptcy law in America (PrincetonUniversity Press).Traczynski, Jeffrey, 2019, Personal bankruptcy, asset risk, and entrepreneurship: Evidencefrom tenancy-by-the-entirety laws,
The Journal of Law and Economics
62, 151–179.12
Appendix: Figures
A.1 Bankruptcy Protection Laws
Figure 1:
Bankruptcy Protection Laws
Bankruptcy Protection in the US, 1989-2017
Note . This figure plots median homestead and non-homestead exemptionsin the US between 1989-2017. The data was collected from other authors,bankruptcy guidebooks, and statutes as described in the paper. .2 Homestead Exemption Law Changes Figure 2:
Homestead Exemption Law Changes
WyomingWisconsinWest VirginiaWashingtonVirginiaVermontUtahTexasTennesseeSouth DakotaSouth CarolinaRhode IslandPennsylvaniaOregonOklahomaOhioNorth DakotaNorth CarolinaNew YorkNew MexicoNew JerseyNew HampshireNevadaNebraskaMontanaMissouriMississippiMinnesotaMichiganMassachusettsMarylandMaineLouisianaKentuckyKansasIowaIndianaIllinoisIdahoHawaiiGeorgiaFloridaDelawareDCConnecticutColoradoCaliforniaArkansasArizonaAlaskaAlabama 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Change Hex No Change
Note . This figure presents all state changes in homestead exemption lawsfrom 1990-2017. This dataset includes all 50 states and Washington DC. In2011, New York became the first state to offer county level exemptions forthree groups of counties. Beginning April 2012, the New York homesteadexemptions will be updated in April every 3 years to keep pace with infla-tion as measured by the New York-Newark-Jersey City CPI-U. The datawas collected from other authors, bankruptcy guidebooks, and statutes asdescribed in the paper. .3 Homestead Exemption Levels Figure 3:
Homestead Exemption Levels
WyomingWisconsinWest VirginiaWashingtonVirginiaVermontUtahTexasTennesseeSouth DakotaSouth CarolinaRhode IslandPennsylvaniaOregonOklahomaOhioNorth DakotaNorth CarolinaNew YorkNew MexicoNew JerseyNew HampshireNevadaNebraskaMontanaMissouriMississippiMinnesotaMichiganMassachusettsMarylandMaineLouisianaKentuckyKansasIowaIndianaIllinoisIdahoHawaiiGeorgiaFloridaDelawareDCConnecticutColoradoCaliforniaArkansasArizonaAlaskaAlabama 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Hex=$0 Hex<50k Hex<$200kHex<=$550k Hex unlimited
Note . This figure presents all homestead exemption levels from 1989-2017.The data was collected from other authors, bankruptcy guidebooks, andstatutes as described in the paper. .4 Bankruptcies and Foreclosures Figure 4:
Bankruptcies and Foreclosures
Foreclosures BankruptciesThousands
Number of Consumers with New Foreclosures and Bankruptcies
Thousands
Source: New York Fed Consumer Credit Panel/Equifax
Note . This figure plots the number of consumers with new bankruptciesand foreclosures per quarter in the US. The data is from the New York FedConsumer Credit Panel/Equifax. .5 Pre-Trends Figure 5:
Impact of Changes in Homestead Exemptions on Real HousePrice Growth -6-4-20246 -3 -2 -1 0 1 2 3
All Positive Changes -6-4-20246 -3 -2 -1 0 1 2 3
Pre-BAPCPA -6-4-20246 -3 -2 -1 0 1 2 3
Big Changes -6-4-20246 -3 -2 -1 0 1 2 3
Big Changes Pre-BAPCPA
Note . This figure plots point estimates ˆ η k and 95% confidence intervals fromthe dynamic regression in Table 3. There is a vertical red line in the yearof the law change. House price data is at the census-tract year level fromthe FHFA, deflated by the CPI-U. The bankruptcy law data was collectedfrom other authors, bankruptcy guidebooks, and statutes as described inthe paper. .6 Supply Elasticity Theory Figure 6:
The impact of a rise in demand on house prices in cities withdifferent supply elasticities Q Supplyinelastic Q Supplyelastic Q Demandold Q Demandnew
Note . This figure compares the impact of a rise in housing demand onhouse prices in two cities with different supply elasticities. Price (P) is onthe vertical axis and quantity (Q) in on the horizontal axis. Initially, theprice of housing is the same in both cities. A rise in demand causes pricesto rise more in the relatively inelastic city. Appendix: Tables
B.1 Descriptive Statistics
Table 1:
Descriptive Statistics
Variable N Min Median Mean Max Freq Change (%) H i,
51 0.00 16,000.00 109,373.00 550,000.00 H i,
51 0.00 50,000.00 151,559.00 550,000.00 H i,
51 10,000.00 100,000.00 197,513.00 550,000.00∆ H i,t if > H i,t if < H i,t if > NH i,
51 0.00 9,100.00 9,327.00 40,000.00 NH i,
51 0.00 18,200.00 18,476.00 60,000.00 NH i,
51 5,000.00 28,400.00 24,695.00 60,000.00 12.68∆ NH i,t if > NH i,t if < NH i,t if > Note . This table reports descriptive statistics summarizing the homestead and non-homestead exemptionin the US between 1989-2017. The bankruptcy law data was collected from other authors, bankruptcyguidebooks, and statutes as described in the paper. H i,t is the maximum homestead exemption in censustract i in year t . Similarly, N H i,t is the maximum non-homestead exemption (vehicle and wildcard) incensus tract i in year t . The percent change is undefined for two changes in H i,t (Delaware 2006, Maryland2011) which followed zero homestead exemption levels. The percent change is undefined for one such changein N H i,t (Delaware 2006). The final column gives the frequency for all (positive and negative) changes ofeach type. Minnesota had a negative change in H i,t in 1993 and Louisiana had a negative change in N H i,t in 2003. .2 Main Estimates Table 2:
Impact of Changes in Homestead Exemptions on Real House PriceGrowth (1) (2) (3) (4)VARIABLES1 { ∆ H > } { ∆ H > } × Pre-BAPCPA 1.066**(0.418)1 { ∆ H > } × Post-BAPCPA 0.488(0.332)1 { ∆ H ≥ k } { ∆ H < k } { ∆ ≥ k } × Pre-BAPCPA 3.043***(0.923)1 { ∆ H < k } × Pre-BAPCPA 0.395(0.471)1 { ∆ ≥ k } × Post-BAPCPA 0.742(0.955)1 { ∆ H < k } × Post-BAPCPA 0.400(0.313)N 1266056 1266056 1266056 1266056R2 .339 .34 .34 .341std-err state state state stateRobust standard errors in parentheses*** p < < < Note . This table reports estimates of the impact of a change in homestead exemption on real house pricegrowth. Each column reports a separate regression estimated at the census tract year level where thedependent variable is the annual percent change of the real house price index. All specifications includecensus tract and year fixed effects. Standard errors, clustered at the state level, are reported in parentheses.1 { ∆ H > } is an indicator equal to one if census tract i had a rise in the homestead exemption that year.Pre-BAPCPA is an indicator equal to one for years up to and including 2005. Post-BAPCPA is an indicatorequal to one for years after 2005. 1 { ∆ H > = 50 k } is an indicator equal to one if census tract i had a risein the homestead exemption of at least $ .3 Validation: Dynamic Estimates Table 3:
Impact of Changes in Homestead Exemptions on Real House PriceGrowth (1) (2) (3) (4)VARIABLES X t − -0.346 -0.765 -1.399 -1.674(0.411) (0.560) (1.063) (1.607) X t − -0.058 0.022 -0.751 0.258(0.357) (0.395) (1.055) (1.021) X t X t +1 X t +2 X t +3 { ∆ H > } { ∆ H > }× Pre-BAPCPA 1 { ∆ H ≥ k } { ∆ H ≥ k }× Pre-BAPCPAN 1012386 1012386 1012386 1012386R2 .378 .378 .384 .381std-err state state state stateRobust standard errors in parentheses*** p < < < Note . This table reports estimates of the impact of a change in homestead exemption on real house pricegrowth. Each column reports a separate regression estimated at the census tract year level where thedependent variable is the annual percent change of the real house price index. All specifications includecensus tract and year fixed effects. Standard errors, clustered at the state level, are reported in parentheses.1 { ∆ H > } is an indicator equal to one if census tract i had a rise in the homestead exemption that year.Pre-BAPCPA is an indicator equal to one for years up to and including 2005. 1 { ∆ H ≥ k } is an indicatorequal to one if census tract i had a rise in the homestead exemption of at least $ .4 Heterogeneity Analysis Table 4:
Heterogeneity in the Impact of Changes in Homestead Exemptionson Real House Price Growth (1) (2) (3) (4) (5) (6) (7) (8)VARIABLES1 { ∆ H > } { ∆ H < } -0.311(0.375)1 { ∆ H > } × Elasticity -0.615***(0.217)1 { ∆ H > } × ur t − -0.195**(0.094)1 { ∆ H > } × pop t − { ∆ H > } × rincpc t − { ∆ H > } × rzhvi t − { ∆ H > } × hown t − -0.031(0.044)1 { ∆ H > } × permits1 t − -0.000*(0.000)Observations 1,266,056 1,012,642 1,258,840 1,266,530 1,266,530 932,422 1,282,709 1,256,147R-squared 0.339 0.374 0.324 0.330 0.317 0.382 0.329 0.341Robust standard errors in parentheses*** p < < < Note . This table reports estimates of the impact of a change in homestead exemption on real house pricegrowth. Each column reports a separate regression estimated at the census tract year level where thedependent variable is the annual percent change of the real house price index. All specifications includecensus tract and year fixed effects. Standard errors, clustered at the state level, are reported in parentheses.Each column reports a separate regression in which the treatment effect is allowed to vary based on sevenmeasures of heterogeneity: supply elasticity, pre-law unemployment rate, population, real income per capita,real Zillow House Value Index, home ownership rate, and single family building permits. House price datais at the census-tract year level from the FHFA, deflated by the CPI-U. The bankruptcy law data wascollected from other authors, bankruptcy guidebooks, and statutes as described in the paper. Elasticitydata are from (Saiz (2010)), unemployment rates are from BLS, population, homeownership, and buildingpermit data are from the Census, income per capita data are from the BEA, and House Value data are fromZillow. .5 Mechanism: Alternative Outcome Variables Table 5:
Impact of Changes in Homestead Exemptions on AlternativeOutcomes (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)VARIABLES R HPG ur ∆ur pop ∆pop incpc ∆incpc permits1 ∆permits1 hown ∆hown1 { ∆ H > } < < < Note . This table reports estimates of the impact of a change in homestead exemption on five outcomevariables and their differences. All specifications include census tract and year fixed effects. Standarderrors, clustered at the state level, are reported in parentheses. R HPG denotes real house price growth.Each column reports a separate regression in which the outcome variable is the level and first difference of:unemployment rates, population, income per capita, single family building permits, and home ownershiprates. House price data is at the census-tract year level from the FHFA, deflated by the CPI-U. Thebankruptcy law data was collected from other authors, bankruptcy guidebooks, and statutes as describedin the paper. Unemployment rates are from BLS, population, homeownership, and building permit dataare from the Census, income per capita data are from the BEA. .6 Predictors of Changes in Homestead Exemptions Table 6:
Predictors of Changes in Homestead Exemptions (1) (2) (3) (4) (5) (6) (7) (8) (9)VARIABLES 1 { ∆ H > } { ∆ H > } { ∆ H > } { ∆ H ≥ k } { ∆ H < k } { ∆ ≥ k } { ∆ H < k } { ∆ ≥ k } { ∆ H < k }× Pre-BAPCPA × Post-BAPCPA × Pre-BAPCPA × Pre-BAPCPA × Post-BAPCPA × Post-BAPCPARHPG t − t − -0.002 -0.005* 0.003 -0.003 0.001 -0.003 -0.003 -0.000 0.003(0.005) (0.003) (0.004) (0.003) (0.004) (0.002) (0.003) (0.002) (0.003)pop t − t − -0.003 0.001 -0.004** -0.001 -0.001 -0.001* 0.002* -0.000 -0.004**(0.002) (0.001) (0.002) (0.001) (0.002) (0.001) (0.001) (0.001) (0.002)hown t − -0.007 -0.000 -0.006 -0.002 -0.004 0.000 -0.001 -0.003 -0.003(0.005) (0.004) (0.004) (0.003) (0.004) (0.002) (0.003) (0.002) (0.004)H i,t − -0.007*** -0.003*** -0.004*** -0.005*** -0.003** -0.003*** -0.000 -0.001*** -0.003***(0.001) (0.001) (0.001) (0.001) (0.001) (0.000) (0.001) (0.000) (0.001)NH i,t − < < < Note . This table reports estimates of the impact of several lagged predictors on indicators for a changein homestead exemption. All specifications include census tract and year fixed effects. Standard errors,clustered at the state level, are reported in parentheses. Each column reports a separate regression inwhich the outcome variable is an indicator equal to one if a census tract experienced a rise in its homesteadexemption. The predictors are lagged: real house price growth, unemployment rates, population, real incomeper capita, homeownership rates, homestead exemption levels, and non-homestead exemption levels. Thebankruptcy law data was collected from other authors, bankruptcy guidebooks, and statutes as describedin the paper. House price data is at the census-tract year level from the FHFA, deflated by the CPI-U.Unemployment rates are from BLS, population, homeownership, and building permit data are from theCensus, income per capita data are from the BEA. or Online Publication: Robustness Appendix B.7 Main Estimates: Robustness
Table 7:
Impact of Changes in Homestead Exemptions on Real House PriceGrowth (1) (2) (3) (4)VARIABLES∆ H H × Pre-BAPCPA 2.164***(0.736)∆ H × Post-BAPCPA 0.094(0.843)∆ H × { ∆ H ≥ k } H × { ∆ H < k } H × { ∆ H ≥ k } × Pre-BAPCPA 2.182***(0.743)∆ H × { ∆ H < k } × Pre-BAPCPA 1.331(4.108)∆ H × { ∆ H ≥ k } × Post-BAPCPA 0.008(0.894)∆ H × { ∆ H < k } × Post-BAPCPA 1.611(2.144)Observations 1,266,056 1,266,056 1,266,056 1,266,056R-squared 0.340 0.340 0.340 0.340Robust standard errors in parentheses*** p < < < Note . This table reports estimates of the impact of a change in homestead exemption on real house pricegrowth. Each column reports a separate regression estimated at the census tract year level where thedependent variable is the annual percent change of the real house price index. All specifications includecensus tract and year fixed effects. Standard errors, clustered at the state level, are reported in parentheses.∆ H is the change in the homestead exemption in a given year. 1 { ∆ H > } is an indicator equal to oneif census tract i had a rise in the homestead exemption that year. Pre-BAPCPA is an indicator equal toone for years up to and including 2005. Post-BAPCPA is an indicator equal to one for years after 2005.1 { ∆ H > = 50 k } is an indicator equal to one if census tract i had a rise in the homestead exemption of atleast $ .8 Main Estimates: Robustness Table 8:
Impact of Changes in Homestead Exemptions on Real House PriceGrowth (1) (2) (3) (4)VARIABLES 1 { ∆ H > } { ∆ H > } × Pre-BAPCPA 1 { ∆ H > = 50 k } { ∆ H > = 50 k } × Pre-BAPCPA X t − -0.322 -0.038 -0.400 -0.007(1.199) (1.540) (1.259) (1.561) X t − -0.620 0.211 -0.747 0.201(1.096) (0.581) (1.146) (0.583) X t X t +1 X t +2 X t +3 < < < Note . This table reports estimates of the impact of a change in homestead exemption on real house pricegrowth. Each column reports a separate regression estimated at the census tract year level where thedependent variable is the annual percent change of the real house price index. All specifications includecensus tract and year fixed effects. Standard errors, clustered at the state level, are reported in parentheses.∆ H is the change in the homestead exemption in a given year. 1 { ∆ H > } is an indicator equal to oneif census tract i had a rise in the homestead exemption that year. Pre-BAPCPA is an indicator equal toone for years up to and including 2005. Post-BAPCPA is an indicator equal to one for years after 2005.1 { ∆ H > = 50 k } is an indicator equal to one if census tract i had a rise in the homestead exemption of atleast $ or Online Publication: Appendix Referencesor Online Publication: Appendix References