Credit-Induced Boom and Bust (cid:3) Marco Di Maggio Amir Kermani Columbia Business School University of California - Berkeley [email protected] [email protected] April 15, 2015 Abstract Can a credit expansion induce a boom and bust in house prices and real economic activity? This paper exploits the federal preemption of national banks in 2004 from local laws against predatory lending to gauge the e⁄ect of the supply of credit on the real economy. Speci(cid:133)cally, we exploit the heterogeneity in the market share of national banks across counties and in state anti-predatory laws to instrument for an outward shift in the supply of credit. First, a comparison between counties in the top and bottom deciles of presence of national banks in states with anti-predatory laws suggests that the preemption regulation produced a 15% increase in annual lending in the 2004-2006 period. Our estimates show that this lending increase is associated with a 5% rise in annual house price growth rate and a 3% expansion of employment in the non-tradable sectors. These e⁄ects are followed by a decline in loan origination, house prices and employment of similar magnitude in subsequent years. Furthermore, we show that the increase in the supply of credit reduced mortgage delinquency rates during the boom years but increased them in bust years. Finally, these e⁄ects are stronger for subprime and inelastic regions. We thank Andreas Fuster, Dwight Ja⁄ee, Benjamin Keys, Ross Levine, Chris Mayer, Adair Morse, (cid:3) Tomasz Piskorski, Giorgio Primiceri, Amiyatosh Purnanandam, Christina Romer, David Romer, Philip Schnabl, Anna Scherbina, Johannes Stroebel, James Vickery, Nancy Wallace, and seminar participants at the 2014 NBER Summer Institute Monetary Economics and Real Estate meetings, the 16th Annual Texas Finance Festival, the 10th CSEF-IGIER Symposium on Economics and Institutions, the 2014 Summer Real Estate Symposium, the SF Fed-UCLA conference on Housing and Monetary Policy, the Columbia-NYU Finance meeting, the Joint Central Bank Conference on Monetary Policy and Financial Stability at Bank of Canada, New York Fed, Cornell University, Columbia University and UC Berkeley. We also thank Katrina Evtimova and John Mondragon for excellent research assistance. All remaining errors are our own. We are alsogratefultothePaulMilsteinCenterforRealEstateatColumbiaBusinessSchoolforsharingtheirdata. 1 1 Introduction The (cid:147)Great Recession(cid:148)was preceded by a very rapid expansion of credit and followed by a collapseinhousepricesandconsumption,whichdidnotregainitspre-recessionlevelforthree years. The resulting job decline was sharper than in any recession of recent decades, with unemployment peaking at 10% in October 2009. What role did the (cid:133)nancial markets play in these severe (cid:135)uctuations? Speci(cid:133)cally, does an outward shift in the credit supply during the up-phase of the business cycle explain the observed disruptions in the real economy? Thispaperinquiriesintothewayinwhichanincreaseincreditsupplytoriskierborrowers was responsible for the boom and bust cycle in house prices and economic outcomes during the recession. This is important to understand how (cid:133)nancial markets a⁄ect real economic activity and how (cid:135)uctuations may be ampli(cid:133)ed by changes in the supply of credit. But it is hard to isolate credit as a causal variable because of omitted variables and reverse causality. The latter is especially important: counties with faster growth will have higher consumption and higher house prices, but they will also have greater demand for credit. As a result, house price and employment movements will be strongly correlated with credit supply, even if the latter has no direct e⁄ect on real estate prices or consumption. In this paper we estimate the e⁄ect of an increase in credit supply to riskier borrowers on economic outcomes using signi(cid:133)cant changes to US banking regulation in the early 2000s. Starting in 1999, a number of states adopted anti-predatory-lending laws (APL) restricting the terms of mortgage loans to riskier borrowers by such means as requiring veri(cid:133)cation of ability to repay as well as limits on fees, rates and early repayment penalties. How- ever, in 2004, in an e⁄ort to increase home ownership, the O¢ ce of the Comptroller of the Currency (OCC) enacted a preemption rule, barring the application of state anti-predatory- lending laws to national banks. In other words, national banks and their mortgage lending subsidiaries were exempted from state APLs and enforcement, while mortgage brokers and independent non-depository lenders, as well as state-chartered depository institutions and 2 their subsidiaries, were still required to comply. Thissettingo⁄ersanexcellentopportunitytoexploitregulatoryvariationsbetweenstates and types of lenders to investigate the role of credit supply shocks. The key to our identi(cid:133)ca- tion strategy is the possibility of comparing economic outcomes in states with and without APLs before and after the OCC preemption rule, but also taking advantage of the sub- stantially uneven presence of national banks in di⁄erent counties. In particular, APL-state counties in which a large proportion of loans were originated by national banks before 2004 had a positive credit supply shock in the wake of the OCC regulation, as national banks could now grant credit to riskier borrowers with fewer restrictions than other (cid:133)nancial insti- tutions. But states with APLmight di⁄er fromthose without, while counties with a stronger presence of national banks might be subject to di⁄erent shocks than those dominated by lo- cal banks. To control for these di⁄erences, we compare counties within APL states, thus excluding di⁄erences between counties with more and fewer OCC lenders in non-APL states. That is, we use a triple di⁄erence-in-di⁄erence estimator to gauge the e⁄ect of the credit increase on the real economy, pinpointing the e⁄ect of the preemption on the availability of lending to riskier borrowers, and then using this as an instrument for the supply of credit during the period preceding the Recession. Mostoftheexistingliteratureinvestigateshowanunderlyingincreaseinthecreditsupply, due for instance to the rise of securitization, lax screening standards, or foreign capital in(cid:135)ows, propagates through the economy by using static regional variations orthogonal to the initial credit shock, such as the elasticity of housing supply developed by Saiz (2010) or the fraction of subprime borrowers in the region. In contrast, our key contribution is to provide an instrument aimed to directly capture an outward shift in the credit supply to riskier borrowers, which allows us to investigate, controlling for regional di⁄erences, how lending to riskier borrowers a⁄ects several sectors of the economy. Moreover, the source of variation we use also highlights the potentially adverse consequences of deregulation on the real economy. 3 There are four main (cid:133)ndings. First, comparing counties in the top and bottom deciles of presence of national banks instates withanti-predatory-lending laws, we showthat the OCC preemption resulted, through a local general equilibrium e⁄ect, in an increase of 11-15% in annual loan issuance.1 To control for county characteristics in all speci(cid:133)cations, we include county and year (cid:133)xed-e⁄ects. We also include other controls such as the county(cid:146)s median income and population, as well as the elasticity measure proposed by Saiz (2010) and the fraction of subprime borrowers to control for the increase in credit demand and collateral values. This is needed in order to show that our instrument is not capturing inter-county di⁄erences in the propensity of house price to increase but the variation due to the expansion of credit. To shed light on how this e⁄ect varied over time, we examine the boom period 2003-2005 and the bust period 2007-2009 separately, con(cid:133)rming that the counties with a greater presence of OCC lenders in states with APLs had a more pronounced boom-bust cycle in loan origination. These estimates constitute our (cid:133)rst stage regression; now we can instrument the supply of credit with the interaction between the presence of national banks in APL states and the post indicator for the period after 2004. Second, using this as an instrument for the supply of credit to riskier borrowers, we estimate its e⁄ect on house prices and (cid:133)nd it to be substantial. A 10% increase in loan origination, through a local general equilibrium e⁄ect, leads to a 3.3% increase in house prices growth rate, which resulted in a total increase of 10% in house prices during the 2004- 2006 period; what is more, our interaction signi(cid:133)cantly predicts the bust in housing prices. Our estimate is robust to extensive controls for demographics and income di⁄erences. And all the speci(cid:133)cations explicitly control for the elasticity of house prices, which means that our estimates are not a⁄ected by confounding e⁄ects unrelated to the shift in credit supply. Third, we explore the e⁄ect of the increase in credit on employment in non-tradable 1We emphasize that these estimates are the result of a local general equilibrium e⁄ect, because even if the initialshock increasedthe credit available tosubprime borrowers, prime borrowersmighthaveincreased their demand for credit as well. For instance, subprime borrowers(cid:146)higher demand for houses, by increasing collateral values, would indirectly increase the credit available to prime borrowers. 4 sectors (as de(cid:133)ned by Mian and Su(cid:133) (2012)), in order to focus on the sectors that are a⁄ected mainly by local demand. We (cid:133)nd that employment expands signi(cid:133)cantly more in counties with a large presence of national banks in APL states, even controlling for county characteristics. Speci(cid:133)cally, our IV estimates suggest that a 10% increase in loan origination leads to a 2% increase in employment in the non-tradable sectors. And focusing solely on the boom and bust period, the predicted increase in lending is associated with a stronger boom and a sharper bust. Finally, we examine the e⁄ect of the expansion of credit on delinquencies. Interestingly, we(cid:133)ndthatincountieswithmoreloansoriginatedbyOCClendersinAPLstatesdelinquency rates were signi(cid:133)cantly lower during the boom but surged in the bust period. Comparing counties in the top and bottom deciles of presence of national banks in APL states, the OCC preemption diminished delinquencies by 30% during the boom and increased them by a similar amount during the Recession. Presumably, the increase in lending enabled households to avoid defaults during the upswing by relaxing their borrowing constraints, but aggravated their (cid:133)nancial situation during the downturn, making them more fragile. Wealsoprovideevidenceofinterestingheterogeneouse⁄ectsacrosscounties. Speci(cid:133)cally, if the e⁄ects we uncover in the data are due to the relaxation of the borrowers(cid:146)credit constraint, we should then expect them to be stronger for regions where borrowers face tighter (cid:133)nancial constraints. Our proxies to capture the extent of these constraints are the fraction of subprime borrowers in a county; a measure of house a⁄ordability, i.e. the ratio of median house price to median income; and the elasticity of housing supply. In all three instances, the results show that the preemption signi(cid:133)cantly increased the availability of credit to riskier and more constrained borrowers, which con(cid:133)rms our posited mechanism of credit-induced (cid:135)uctuation. To check the robustness of our results and weigh potential alternative mechanisms we show several additional results. One concern about our results is the possibility that the presence of national banks could be correlated with the rise in securitization that occurred 5 during the same period. To rule this out, for each county we compute the fraction of loans that were securitized and use this to proxy for the banks(cid:146)incentive to increase lending due to securitization. Although securitization is an important predictor for the boom and bust cycle, all the results are completely una⁄ected, which suggests that our instrument is not correlated with securitization.2 Second, to show that the results are driven by changes in the states with anti-predatory laws, due to the increase in lending by national banks, rather than by loan origination by other (cid:133)nancial institutions in states without APLs, we focus on the states that passed anti-predatory laws by 2004 and estimate a di⁄erence-in-di⁄erences regression. This test exploits variation between counties, but controls for potential di⁄erent regional trends due to the rise in securitization or the heterogeneous presence of subprime borrowers. It shows that after 2004 the counties with a higher fraction of national banks are the ones that experienced a more signi(cid:133)cant increase in loan issuance, house prices and employment and a larger decline in delinquencies. Third, to control further for potential unobserved heterogeneity across counties, we can restrict attention to those at state borders. Since counties on the West coast are much larger than those on the East coast, and the sample of counties close to the borders is small, we construct our main variables at census tract level. Even restricting the investigation to census tracts within (cid:133)fteen miles of state borders, the results stand con(cid:133)rmed. Fourth, we show that the main e⁄ects derive from the increase in loans to households, rather than by lending to small businesses.3 Furthermore, since lenders within the same bank-holding company might exploit a form of regulatory arbitrage by switching, for instance, between a lender regulated by the Depart- 2Forinstance,Kermani(2012)showsthatregionsthatexperiencedlargerincreaseinthefractionofloans that were securitized also exprienced a larger boom and bust in house prices and consumption. 3The robustness of our results is further demonstrated by the fact that the predicted lending increases are not associated with an increase in employment in the tradable sectors. Furthermore, we can eliminate the states with the highest delinquency rates and largest housing bubbles, Arizona and Nevada, and show that our results are not driven by those states. In unreported results, we also show that our estimates are robust even when only states that eventually passed an anti-predatory-lending law are considered. In other words, if the concern is that APL and non- APL states are fundamentally di⁄erent, our results hold when only the timing of the adoption of APL, and not the di⁄erence between two (cid:147)types(cid:148)of state, is considered. 6 ment of Housing and Urban Development and an OCC lender, we re-estimate our e⁄ects assigning to OCC all the subsidiaries of the bank holding companies, if there is at least a subsidiaryregulatedby OCCwithinthe bankholding company. Finally, analyzing loan-level data we show that the introduction of the OCC preemption rule resulted in a signi(cid:133)cant in- crease in the issuance of (cid:147)high-cost loans(cid:148)and mortgages with debt-to-income ratios in the top decile by OCC lenders.4 This provides further evidence corroborating the mechanism behind our results.5 1.1 Related Literature Our key contribution is to use the deregulation of lending restrictions to directly estimate the causal e⁄ect of an increase in credit supply to riskier borrowers on house prices and real economic activity, and its role in generating a distinct boom and bust pattern. An emerging literature related to this paper studies the e⁄ects of house price booms on real economic activity. The most closely related paper is Mian and Su(cid:133) (2009). In their seminal paper, Mian and Su(cid:133)(2009) show that zip codes with a higher fraction of subprime borrowersexperiencedunprecedentedrelativegrowthinmortgagecreditandacorresponding increase in delinquencies. Our own paper makes three signi(cid:133)cant advances: (1) exploiting an exogenous variation in the supply of credit, we estimate the e⁄ect of credit supply on house prices, controlling for local economic shocks; (2) our data enable us to track employment and delinquency rates as well; and (3) we (cid:133)nd that the outward shift in the credit supply that followed the preemption regulation signi(cid:133)cantly predicts both the boom and the bust in real economic activity. There has been abundant evidence of changes in lending during the years preceding the 4(cid:147)High-costloans(cid:148)arede(cid:133)nedasloanswithanannualpercentagerate3percentagepointsormoreabove theTreasuryratefor(cid:133)rst-lienmortgageswithcomparablematurities. Mortgageswithdebt-to-income(DTI) ratios in the top decile usually exhibit DTI above 4. 5Inacomplementarypaper,DiMaggioetal. (2015)showthat,afterthepreemptionrule,nationalbanks signi(cid:133)cantly increased the origination of mortgages featuring prepayment penalties, negative amortization and balloon payments by about 10%. 7 crisis due to di⁄erent reasons. There are studies on the weakened lending standards (Jiang et al. (2014), Agarwal et al. (2014), Haughwout et al. (2011), Chinco and Mayer (2014) and Barlevy and Fisher (2010)), on the increase in misrepresentations and fraud (Ben-David (2011), Garmaise (2014), Piskorski et al. (2013) and Gri¢ n and Maturana (2014)), on the failure of ratings models and the rapid expansion of non-agency securitization markets (Rajan et al. (2010), Purnanandam (2011), Nadauld and Sherlund (2013) and Keys et al. (2010)). We complement these studies by showing how a signi(cid:133)cant fraction of the increase in lending can be attributed to changes in the regulatory framework. Otherpapersontheinterplaybetweencredit, housepricesandconsumptionincludeMian et al. (2012), Mian et al. (2013), Greenstone and Mas (2012) and Chodorow-Reich (2014).6 Mian et al. (2012) instrument foreclosure with the di⁄erence between judicial and non- judicial foreclosure states to show that foreclosures cause a signi(cid:133)cant decline in house prices and residential investment. Mian et al. (2013) show that zip codes where households are more highly leveraged experienced a more severe decline in consumption and employment in the non-tradable sector. The importance of the credit channel for employment is highlighted by Greenstone and Mas (2012), which assesses the role of bank lending to small businesses in the employment decline during the Recession and by Chodorow-Reich (2014) who relates the availability of credit with the employment decline at small and medium (cid:133)rms in the year following the Lehman bankruptcy. Our own paper, by contrast, instruments variations in lending with regulatory changes to show the e⁄ect of the increase in lending on the boom andbustinseveralsectorsoftheeconomythroughmortgageorigination. Furthermore, Mian and Su(cid:133)(2012) show that job losses in the non-tradable sector between 2007 and 2009 are signi(cid:133)cantly more severe in high-leverage counties that experienced sharp demand declines while Adelino et al. (2012) exploits changes in the conforming loan limit as an instrument to gauge the e⁄ect of the availability of cheaper (cid:133)nancing on house prices. We employ the same 6Another related paper is Kleiner and Todd (2007), which (cid:133)nds that the requirement, in place in many states, that mortgage brokers maintain a minimum net worth is associated with smaller numbers of brokers, fewersubprimemortgages, higherforeclosurerates, andahigherpercentageofhigh-interest-ratemortgages. 8 di⁄erentiation as Mian and Su(cid:133)(2012) between tradable and non-tradable sectors to show that the increase in lending boosted local demand, which in turn increased employment in the non-tradable sectors. In contrast toAdelino et al. (2012), which employs a local source of variation,oursisatthecountylevelandthereforewewouldexpectalocalgeneralequilibrium e⁄ect. For instance, our (cid:133)ndings on employment are the result of a local general equilibrium e⁄ect, which includes the (cid:133)rms(cid:146)increased borrowing capacity through a collateral channel and its e⁄ects on the (cid:133)rms(cid:146)investment policy (Chaney et al. (2012)). Finally, Jayaratne and Strahan (1996) show that per capita growth rates in income and output increased signi(cid:133)cantly following the relaxation of bank branch restrictions in the United States. Favara and Imbs (2015), instead, use the passage of the Interstate Banking and Branching E¢ ciency Act (IBBEA) in 1994 to show that this deregulation triggered an increase in the demand for housing, that is, that house prices rose because the supply of credit in deregulating states expanded. The present paper, instead, observes an increase in credit supply due to the preemption rule of 2004, which in contrast to the IBBEA targeted subprime lending and riskier borrowers, and shows how it helped to trigger the boom and bust cycle in both real estate and employment for a di⁄erent sample period. Thispaperalsocontributestothegrowingliteratureonthee⁄ectsofthedeclineinlending duringtheRecession. IvashinaandScharfstein(2010), forinstance, documentthatnewloans to large borrowers fell by 79% between the second quarter of 2007 and the fourth quarter of 2008. They argue that this drop was largely (cid:147)supply-driven(cid:148), because of the decline in banks(cid:146)access to short-term funding following the Lehman Brothers failure. Similarly, Cornett et al. (2011) point out that the banks(cid:146)e⁄orts to manage the liquidity crisis led to a decline in credit supply. Using Community Reinvestment Act data, Huang and Stephens (2011) and Berrospide and Edge (2010) show that multi-market banks(cid:146)exposure to markets in which there were housing busts a⁄ected the supply of small business loans within all MSAs. Goetz and Valdez (2010) (cid:133)nd evidence that di⁄erences in the liability structure of small U.S. commercial banks, particularly the use of (cid:147)non-core(cid:148)funding, a⁄ected lending 9 patterns during the 2008 crisis. Dagher and Fu (2011) show a positive correlation between the presence of independent mortgage companies and the increase in foreclosure (cid:133)ling rates at the onset of the housing downturn. We complement these studies by providing evidence that part of the decline in lending is the reversal of the initial boom. Finally, we contribute to the literature on credit booms and (cid:133)nancial crisis (see, among others, Jorda(cid:146)et al. (2011), Schularick and Taylor (2012) and Rajan and Ramcharan (2012)) by showing that the credit supplied by national banks during the economic upswing explains the surge and subsequent collapse in house prices and employment. The remainder of the paper is organized as follows. Section 2 gives background on the US credit market and regulation. Section 3 provides details on the data sources. Section 4 explains the research design and how it is made operational. Section 5 describes and interprets the main results. Section 6 presents the heterogeneity of treatment e⁄ects across regions to provide further evidence of our credit-supply mechanism. Section 7 discusses a number of robustness checks, while Section 8 provides an estimate of the aggregate impact of our results. Section 9 concludes. 2 Regulatory Framework 2.1 Mortgage Regulators In the United States, residential mortgage lenders are regulated by national and local agen- cies. Speci(cid:133)cally, national banks, Federal thrift institutions and their subsidiaries are super- vised by the OCCor the O¢ ce of Thrift Supervision (OTS). State banks and state-chartered thrift institutions are supervised by either the Federal Reserve System, the Federal Deposit Insurance Corporation (FDIC) or by their chartering state. Credit unions are supervised by the National Credit Union Administration (NCUA), while non-depository mortgage compa- nies are regulated by the Department of Housing and Urban Development (HUD) and the 10