Deregulation, Competition and the Race to the Bottom

Transcription

1 Deregulation, Competition and the Race to the Bottom Marco Di Maggio Amir Kermani Sanket Korgaonkar February 28, 2015 The latest version can be found here. Abstract We take advantage of the pre-emption of national banks from state anti-predatory lending laws as a quasi-experiment to study the effect of deregulation and its interaction with competition on the supply of complex mortgages (interest only, negative amortization, and teaser mortgages). We first show that following the pre-emption ruling, national banks significantly increased their origination of loans with prepayment penalties and negative amortization features, relative to non-occ regulated lenders, and lenders in states without anti-predatory lending laws. This increase in the supply of complex mortgages is significantly more pronounced for banks that poorly performed in the previous quarters. Further, we highlight a competition channel: first, a higher degree of competition induce OCC lenders to originate more riskier loans; second, in counties where OCC regulated lenders had larger market share prior to the pre-emption, even non-occ lenders responded by increasing the presence of predatory terms to the extent permitted by the state anti-predatory lending laws. Overall, our evidence is suggestive that the deregulation of credit markets ignited a race to the bottom among distressed financial institutions, working through the competition between lenders. Columbia University; md3226@columbia.edu University of California - Berkeley; amir@haas.berkeley.edu University of California - Berkeley; sanketk@berkeley.edu 1

2 1 Introduction The financial deregulation of the last two decades is the subject of a heated political and scholarly debate as it might have played an important role in creating a permissive lending environment. In fact, critics sustain that regulators incentivized looser underwriting standards in order to encourage the making of more and more marginal loans. Effective regulation of lending practices could have also prevented the aggressive lenders from abusing vulnerable borrowers by offering riskier and more complex mortgages. 1 Moreover, on the one hand market forces and lenders reputation concerns may discipline banks; behavior, on the other hand fiercer competition could induce lenders to race to the bottom by originating even riskier loans. A crucial challenge in empirically identifying the effects of deregulation in fueling the increase in mortgage origination is the fact that policy interventions usually affect all lenders at the same time. This precludes the possibility to distinguish between the direct effects of the policy and other confounding factors affecting mortgage originations, such as changes in the demand for mortgages. In this paper, we are able to overcome these difficulties by exploiting the 2004 pre-emption of state anti-predatory lending laws for lenders regulated by the Office of Comptroller and Currency as an exogenous shock to the competitive landscape. Specifically, this was a shock which expanded the set of loans OCC-regulated lenders were allowed to originate, while leaving unchanged the set that non-occ regulated lenders were allowed to originate. The pre-emption ruling creates an ideal environment to test for the effects of deregulation by providing us with a clean set of affected banks, i.e. the ones regulated by the OCC, and a set of unaffected banks, i.e. those regulated by the state regulators as well as by the department of housing and urban development (HUD). Thus, 1 President Barack Obama justified the need for a Consumer Financial Protection Agency by claiming that predatory lending by unregulated mortgage brokers was a cause of the financial crisis: Part of what led to this crisis were not just decisions made on Wall Street, but also unsustainable mortgage loans made across the country. While many folks took on more than they knew they could afford, too often folks signed contracts they didn t fully understand offered by lenders who didn t always tell the truth (White House news release, September 19, 2009, available at press office/weekly-address- President-Obama-Promotes-Tougher-Rules-on-Wall-Street-to-Protect-Consumers).. 2

3 we can exploit it to understand how lenders respond to de-regulation, as well as how deregulation might have spillover effects on other lenders, due to intensified competition among mortgage originators. Our first result uses a difference-in-difference analysis on a sample of loans issued in states with anti-predatory lending laws (henceforth APL laws ) to show that the pre-emption of these laws for OCC regulated lenders led them to increase the issuance of loans with predatory terms, such as prepayment penalties, negative amortization, balloon payments and lengthy prepayment penalty terms. Our most conservative estimation shows that OCC regulated lenders were 10% more likely, relative to non-occ regulated lenders, to issue loans with prepayment penalties following the pre-emption ruling. Compared to an unconditional probability of prepayment penalties of about 30% in our sample, this represents an economically significant increase. These prepayment penalties are particularly important as they enable other features of the mortgage, such as negative amortization, teaser rates and balloon payments to be used profitably. Our results remain robust to using a triple diff-indiff analysis, which also uses as a control group loans made in states where anti-predatory lending laws were not in effect. A large literature in household finance studies the demand-side determinants of the different loan contracts observed in the data. This literature takes important steps towards understanding the types of borrowers who take on different forms of debt, such as adjustable rate mortgages (ARM), fixed rate mortgages (FRM) and interest-only mortgages (IO). 2 However, much less is known about the lenders supply of these loan contracts. The deregulation in 2004, by differentially affecting different types of originators, gives us the unique opportunity to show that the supply of these mortgages significantly changed in the years preceding the crisis. Next, we further explore the important sources of heterogeneity in the responses of OCC regulated lenders to the pre-emption ruling. We investigate whether OCC lenders responded 2 See Campbell (2006) for a survey of this literature. 3

4 differently across geographies based on local mortgage market competition. We find that a 1 standard deviation decrease in the Herfindahl Index of a county (a more competitive market) predicts an increase in the probability of origination of a loan with prepayment penalties of 4% (an effect that is 17% of the unconditional probability of having a prepayment penalty), and a 2.6% increase in the origination of negative amortization loans (25% of the unconditional probability of having a negative amortization loan). Additionally, we show that, consistent with the risk-shifting-hypothesis, OCC lenders, following the pre-emption ruling, became more responsive to their recent stock price returns. In other words, following poor returns OCC lenders were more likely to issue loans with predatory features following the pre-emption than prior to it. One important questions is whether this deregulation by affecting the competitive landscape also affected the non-occ lenders depending on the degree of competition with OCC lenders. The hypothesis is that in counties where OCC lenders have a smaller market share, the pre-emption rule should not have much of an effect, as most of a non-occ lender s competitors in the county are unaffected. Thus, a priori, one might expect the competition channel to operate in changing the lending behaviour of non-occ regulated lenders when they face more intense competition from OCC lenders. We find that a 1 standard deviation increase in a county s market share of OCC lenders predicts a 6.4% increase in the probability of originating a loan with negative amortization, a 6.6% increase in the probability of originating a loan with interest only payments and ARM respectively. Our results point out that rather than attenuating the effects of deregulation, competition might induce also the banks not directly affected by the preemption to compete by issuing riskier and more complex mortgages. Taken together, our results indicate two main channels through which deregulating the mortgage market might have an effect. First, it directly increased the origination of loans with predatory -like features by OCC-regulated lenders. Second, our tests also show that preemption rule induced a response even from those lenders who remained subject to the 4

5 regulation in the same markets. Our results are suggestive of a race to the bottom which began with the OCC regulated lenders, worked it s way through the local mortgage market, and forced the hand of the non-occ regulated lenders to alter their mortgage terms as a competitive response. 1.1 Related Literature Our key contribution is to directly estimate the causal effect of deregulation on the supply of riskier and complex mortgages through both a direct channel, the behavior of the deregulated national banks, and through an indirect one, the response of their non-national competing institutions. We borrow the same identification strategy proposed by Di Maggio and Kermani (2014), based on the introduction of the preemption rule in 2004 by the OCC and the variation across states with and without anti-predatory laws. However, our paper differs both in focus and results. The main results of Di Maggio and Kermani (2014) are about the real effects of an outward shift in the credit supply, specifically, the possibility to induce a boom and bust cycle in economic activity at the county level. Our paper exploits, instead, individual-level data to first show the effect of the preemption rule on the features of mortgages originated after the preemption rule by national banks. We then investigate the response of the non- OCC regulated banks, such as state banks and credit unions, to show how competition might shape the response to deregulation. After the crisis, a novel literature relating the changes in the mortgage market conditions and the real economy emerged. For instance, in their seminal paper, Mian and Sufi (2009) show that zip codes with a higher fraction of subprime borrowers experienced unprecedented relative growth in mortgage credit and a corresponding increase in delinquencies. Our own paper advances this literature by exploiting an exogenous shock supply of credit and the competitive environment, to estimate how the specific contracting features offered by the financial institutions significantly changed. Our paper also related to the several studies 5

6 investigating the changes in lending behavior during the years preceding the crisis. Few studies, such as Jiang et al. (2014), Agarwal et al. (2014), Haughwout et al. (2011), Chinco and Mayer (2014) and Barlevy and Fisher (2010), have pointed out that weakened lending standards is one of the main causes behind the subprime crisis; while others, such as among others Rajan et al. (2010), Purnanandam (2011), Nadauld and Sherlund (2013) and Keys et al. (2010), have highlighted the failure of ratings models and the rapid expansion of non-agency securitization markets as one of the main driving factors. We complement these studies by providing evidence that deregulation might have ignited a race to the bottom among lenders in the years preceding the crisis. Another recent paper analyzing how policy intervention affects the mortgage market is the study by Agarwal and Ben-David (2012). They analyze the effect of the Community Reinvestment Act (CRA) on banks lending activity. They find that adherence to the act led to an increase in lending by banks, in fact, during the six quarters surrounding the CRA exams lending is 5 percent higher, but these loans default more often. We share the focus on the effect of deregulation on the pre-crisis loan origination, however, we exploit loan-level data to study how lenders modified key features of the mortgages they originated to remain competitive. Moreover, we also complement these findings by showing that the poor-performing banks were significantly more likely to take advantage of the deregulation. Other related papers include Jayaratne and Strahan (1996), Favara and Imbs (2010), Greenstone and Mas (2012), and Adelino et al. (2012). Jayaratne and Strahan (1996) show that per capita growth rates in income and output increased significantly following the relaxation of bank branch restrictions in the United States. We share with Favara and Imbs (2010) the use of a deregulation as quasi-experiment, in fact, Favara and Imbs (2010) exploit the passage of the Interstate Banking and Branching Efficiency 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 main difference with the current paper is that we document an increase in credit supply due to the preemption 6

7 rule of 2004, which in contrast to the IBBEA targeted subprime lending and riskier borrowers. Greenstone and Mas (2012) investigate the importance of the credit channel for employment by assessing the role of bank lending to small businesses in the employment decline during the Recession. Adelino et al. (2012) exploits changes in the conforming loan limit as an instrument to gauge the effect of the availability of cheaper financing on house prices. We complement these studies by showing how the mortgage originators directly affected by the deregulation significantly changed the contracting features of the mortgages offered, which also made other market participants compete by adopting complex and predatory lending practices as well. We share with Amromin and Kearns (2014) its focus on the effect of policy changes on the competitive landscape. Amromin and Kearns (2014) explore whether market competitiveness affects mortgage interest rates exploiting the introduction of the Home Affordable Refinancing Program (HARP). Specifically, lenders that currently service loans eligible for refinancing enjoyed substantial advantages over their competitors under HARP. They show a significant increase in mortgage interest rates, about 15 to 20 basis points, precisely at the HARP eligibility threshold. Amromin et al. (2013), instead, investigates what type of borrowers are more prone to take on complex mortgages during the years preceding the crisis. They show that these riskier loans were chosen by prime borrowers with high income levels seeking to purchase expensive houses relative to their incomes. However, these borrowers tend to default more often than borrowers with traditional mortgages with similar characteristics. We complement these findings by showing that when competition is more intense, as measured by the Herfindahl index or the share of deregulated lenders, the lenders not affected by the preemption rule tend to adjust not only the interest rate but also a number of other different mortgage features. 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 outlines our main research design, and provides the first results on the effect of the deregulation. 7

8 Section5 investigates geographic variation in the response of OCC regulated lenders. Section 6 investigates a competition mechanism by which non-occ lenders also changed their origination behavior, Section 7 performs some robustness checks, while Section 8 concludes. 2 Regulatory Framework 2.1 Mortgage Regulators In the United States, residential mortgage lenders are regulated by national and local agencies. Specifically, national banks, Federal thrift institutions and their subsidiaries are supervised by the OCC or the Office 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 companies are regulated by the Department of Housing and Urban Development (HUD) and the Federal Trade Commission. One potential source of concern is the possibility for mortgage companies to shop for the most lenient regulator. However, Agarwal et al. (2012) show that federal regulators are significantly less lenient, downgrading supervisory ratings about twice as frequently as state supervisors, while banks under federal regulators report higher nonperforming loan ratios, more delinquent loans, higher regulatory capital ratios, and lower ROA. Banks accordingly have an incentive to switch from Federal to state supervision, if they are allowed to do so. Moreover, Rosen (2005)explores switching in regulatory agencies between 1970 and 2003, and finds that in the early part of the period most of the switches were due to new banking policies, such as the easing of the ban on interstate banking, whereas after the initial period the main reason for switching was merger with a bank chartered at a different level. Further, the banks that switched tended to be small banks with assets of less than $1 billion. These findings corroborate our own identification strategy; moreover, the granularity of 8

9 our dataset allows us to track the banks that changed regulatory agencies, so that we can address any further concerns related to this issue. 2.2 Anti-predatory laws This dual banking system generated conflicting regulations when several states passed antipredatory-lending laws and the OCC issued a preemption rule for national banks. In 1994, Congress had passed the Home Ownership and Equity Protection Act (HOEPA) which imposed substantive restrictions on terms and practices for high-priced mortgages, based either on APR or on total points and fees. This regulation aimed to redress abusive high charges for refinancing and home equity loans. However, the thresholds for classifying mortgages as predatory or high cost were very high, which significantly reduced the applicability of the restrictions; these high cost mortgages, in fact, accounted for just 1 percent of subprime residential mortgages; they represented the most abusive sector of the subprime mortgage market (Bostic et al. (2008)). Many states later adopted stronger anti-predatory regulations than federal law requires. Anti-predatory laws seek to prevent various unfair and deceptive practices, such as steering borrowers into loans with a higher interest rate than they could qualify for, making a loan without considering repayment ability, charging exorbitant fees, or adding abusive subprime early repayment penalties, all of which can increase the risk of foreclosure significantly. 3 The first comprehensive state APL law was that of North Carolina in 1999, which was targeted at the subprime mortgage market. As of January 2007, 20 states and the District of Columbia had APL laws in effect. Potentially, APLs may have different kinds of effects on mortgage market outcomes. On the one hand, the laws might ration credit and raise the price of subprime loans. On the other, they might serve to allay consumer fears about dishonest lenders and ensure that 3 Agarwal and Evanoff (2013) provide evidence of unscrupulous behavior by lenders such as predatory lending during the housing boom of the 2000s. They show that lenders steered higher-quality borrowers to affiliates that provided subprime-like loans, with APR between 40 and 60 basis points higher. 9

10 creditors internalize the cost of any negative externalities from predatory loans, which could increase the demand for credit. There is strong recent evidence that anti-predatory laws had an important role in the subprime market. Ding et al. (2012), for instance, find that they are associated with a 43% reduction in early repayment penalties and a 40% decrease in adjustable-rate mortgages; they are also correlated with a significant reduction in the riskier borrowers probability of default. In subprime regions (those with a higher fraction of borrowers with FICO scores below 680) these effects are even stronger. Using 2004 HMDA data, Ho and Pennington-Cross (2006) find that subprime loans originated in states with laws against predatory lending had lower APRs than in unregulated states. Ho and Pennington-Cross (2008) provide additional evidence, focusing on border counties of adjacent states with and without APL to control for labor and housing market characteristics. Using a legal index, they examine the effect of APLs on the probability of subprime applications, originations, and rejections. They find that stronger regulatory restrictions reduced the likelihood of origination and application. Similarly, Elliehausen et al. (2006), using a proprietary database of subprime loans originated by eight large lenders from 1999 to 2004, find that the presence of a law was associated with fewer subprime originations. More recently, Agarwal et al. (2014) estimate the effect on mortgage default rates of a pilot anti-predatory policy in Chicago that required low-credit-quality applicants and applicants for risky mortgages to submit their loan offers from state-licensed lenders for third-party review by HUD-certified financial counselors. This policy significantly affected both the origination rates and the characteristics of risky mortgages. 4 Finally, the anti-predatory laws are likely to have had significant impact on the banks incentives for securitization. In fact, credit rating agencies stated explicitly that after the APLs were enacted they began to require credit enhancement from lenders who might have been in violation of state APLs: To the extent that potential violations of APLs reduce the 4 For a theoretical model of predatory lending see Bond et al. (2009). 10

11 funds available to repay RMBS investors, the likelihood of such violations and the probable severity of the penalties must be included in Moody s overall assessment. 5 Evidence of this is also provided by Keys et al. (2010) who study the effect of securitization on lenders screening decisions and exploit the passage and subsequent repeal of anti-predatory laws in New Jersey (2002) and Georgia (2003) that varied the ease of securitization. They find strong evidence that the incentives to screen the borrowers significantly increased during a period of strict enforcement of anti-predatory lending laws. We follow this literature employing the measure constructed by Ding et al. (2012), which considers only the states that passed anti-predatory laws that were not just small-scale home ownership and equity protection acts implemented to prevent local regulation. 2.3 Preemption Rule On January 7, 2004 the OCC adopted sweeping regulations preempting, with regard to national banks, a broad range of state laws that sought to regulate the terms of credit. The measure preempted laws that regulate loan terms, lending and deposit relationships or require a state license to lend. The final rule also provided for preemption when the law would obstruct, impair, or condition a national bank s exercise of its lending, deposit-taking, or other powers granted to it under federal law, either directly or through subsidiaries. The new regulations effectively barred the application of all state laws to national banks, except where (i) Congress has expressly incorporated state-law standards in federal statutes or (ii) particular state laws have only an incidental effect on national banks. The OCC has said that state laws will be deemed to have a permissible incidental effect only if they are part of the legal infrastructure that makes it practicable for national banks to conduct their federally-authorized activities and do not regulate the manner or content of the business of banking authorized for national banks, such as contracts, torts, criminal law, the right to 5 Available at US.html. 11

12 collect debts, property acquisition and transfer, taxation, and zoning. 6 Specifically, the OCC preempted all regulations pertaining the terms of credit, including repayment schedules, interest rates, amortization, payments due, minimum payments, loanto-value ratios, the aggregate amount that may be lent with real property as security or term to maturity, including the circumstances under which a loan may be called due and payable after a certain time or upon a specified external event. This means that starting in 2004 the subprime mortgage market in states with antipredatory laws was no longer a level playing field: national banks were significantly less constrained by APLs in providing credit to riskier borrowers. 3 Data We collected data from a number of different sources. The primary source of our data is the ABSNet Loan Database. This database covers almost 90% of the private-label Residential Mortgage Backed Securitization issuances and provides data on the underlying loans, as well as, data on key borrowers characteristics. The main advantage of this dataset over the other standard datasets used in the literature, such as LPS and Blackbox, is the possibility to identify the mortgage originator, which is key to our identification. In fact, this allows us to use a classification of the lenders into those who were regulated by federal agencies (henceforth OCC Lenders ) and all other lenders (henceforth Non-OCC Lenders ). 7 We 6 For instance, New Century mentioned in its K filing the following: Several states and cities are considering or have passed laws, regulations or ordinances aimed at curbing predatory lending practices. In general, these proposals involve lowering the existing federal HEPA thresholds for defining a high-cost loan, and establishing enhanced protections and remedies for borrowers who receive such loans. [...] Because of enhanced risk and for reputational reasons, many whole loan buyers elect not to purchase any loan labeled as a high cost loan under any local, state or federal law or regulation. This would effectively preclude us from continuing to originate loans that fit within the newly defined thresholds. [...] Moreover, some of our competitors who are, or are owned by, national banks or federally chartered thrifts may not be subject to these laws and may, therefore, be able to capture market share from us and other lenders. For example, the Office of the Comptroller of the Currency issued regulations effective January 7, 2004 that preempt state and local laws that seek to regulate mortgage lending practices by national banks. (available at pag. 45). 7 This classification has been graciously provided to us by Nancy Wallace and the Fisher Center for Real Estate and Urban Economics at the Haas School of Business. 12

13 consider all first-lien mortgages originated in the pre-period, January 2002 to January 2004, and in the post-period, February 2004 to December 2005, with a final sample including close to 7 million individual loans. Another main advantage of this fine-grained data, is the possibility to observe all the specific features of these loans at the origination date. For instance, the first part of our analysis will exploit this by analyzing how the national banks changed the presence of prepayment penalties, length of the prepayment penalty term, balloon payment, negative amortization, and interest rates in response to the preemption rule. We shall show that the ability to impose prepayment penalties enabled lenders to issue more complex mortgages such as those with negative amortization or balloon payments, and those that were interest only or had adjustable rates. One shortcoming of the data, however, is that we do not observe the loan fees and points so as to classify loans into those that were high cost. Additionally, we do not observe the amount or size of the prepayment penalty. Table 1a displays summary statistics for our sample of loans. As our sample comes from private label securitizations, which were the way in which a large quantity of subprime, nonconforming loans were securitized, the average LTV is at 84%. Most loans are for Owner Occupied houses, with 40% being used for a purchase of a home, and the remaining largely for cash out or rate refinances. Unconditionally, 29% of the loans in our sample have a prepayment penalty, a variable that will constitute a key focus of the analysis. 16% of the loans in our sample also have silent second liens, leading the combined LTV in our sample to be 86%. 64% of the loans have ARMs while 17% are interest only loans. Only 1% of the sample are loans with balloon payment features. Finally, the average credit score in our sample is , once again reflecting the fact that private label securitization largely funded subprime origination. There are about 3.6 million loans in our sample that were originated in states that had APL laws in place. Table 1a shows us the summary statistics for loans that were in states with APL laws and those states without APL laws. 13

14 4 Identifying the effect of the pre-emption ruling of 2004 In this section, we look to estimate and quantify the impact that the pre-emption ruling had on the presence a variety of mortgage lending terms. We do so using both a difference-indifference approach, as well as a more robust triple difference-in-difference approach. This analysis identifies an exogenous change in the incidence of mortgage terms brought about by the pre-emption ruling. Our results show that the pre-emption led to a higher incidence of loans with prepayment penalties and negative amortization, as well as longer prepayment penalty terms. First, in order to test the response of OCC lenders to the pre-emption of APL laws, we run a difference-in-difference regression. Here we restrict the sample to loans originated in those states where the APL laws were in effect prior to the pre-emption. We further restrict attention to those loans issued after the anti-predatory lending laws were in effect in the states in question. The specification considered is: Y icgt = β 0 + β 1 P ost t OCC i + +β 2 P ost t + β 3 OCC i + β 4 X it + η c + η month + ɛ icgt where OCC i is an indicator for whether the lender originating loan i was regulated by the OCC; P ost t is an indicator equal to 1 for t February 2004, X it are loan level covariates 8, and η c and η month are county and month fixed effects respectively. In this specification, and in those that follow, Y icgt is either an indicator of whether the loan had a prepayment penalty, whether the loan was a negative amortization loan, whether the loan had a balloon payment, whether the prepayment penalty term of the loan would have been in violation of 8 Loan level covariates included are: LTV ratio at origination, Log of appraised value at origination, FICO score, indicator for the presence of second liens, low or no documentation indicator, indicator for loan purpose (i.e. cash out refinance, rate refinance or other), and indicator for the presence of PM. 14

15 existing APL laws that applied to high cost 9, and whether the loan classified as low or no documentation. β 1 is the coefficient of interest. It estimates: [Ȳ AP L OCC,P ost Ȳ ] [Ȳ OCC,pre AP L AP L Non OCC,P ost Ȳ ] Non OCC,P AP L re Table 2 presents the regression results. The results in Column 1 suggest that an OCC lender in an APL state was 10% more likely to make a loan with a prepayment penalty relative to a non-occ lender following the pre-emption. This compares to an unconditional mean of the presence of prepayment penalties of 25.6%. This result suggests that the pre-emption led to an economically important increase in the presence of this loan term. Similarly, relative to non-occ lenders, OCC lenders made 6.2% more negative amortization loans (unconditional mean of 7.5%), 1.2% more loans with balloon payments (unconditional mean of 1.4%), and 4% more loans with prepayment penalty terms that would have violated APL laws (unconditional mean of 14.7%). Additionally, they made prepayment penalty terms 2 months longer relative to non-occ lenders. The results for prepayment penalties, negative amortization and prepayment penalty terms are particularly striking, as β 3 suggests that OCC lenders issued loans with lower incidence of predatory terms in the pre-period. To gain further insight into how the effects propagated over time and to check our data for any pre-trends, we perform an event study version of the diff-in-diff regression. The specification we use is: Y icgt = β 0 + τ β 2τ OCC i 1 t=τ + τ β 3τ X it 1 t=τ + η county halfyear + ɛ icgt where η county halfyear indicates county by half year fixed effects. Figures 1a and 1b plot the coefficients β 1τ from the regressions with Y icgt equal to first, an indicator for whether the loan had a prepayment penalty, and second, for whether the prepayment penalty term of 9 For this purpose we use the Bostic et al. (2008) classification of prepayment penalty term related APL laws. See Table 2 of Bostic et al. (2008). 15

16 the loan would be in violation of the state anti-predatory lending laws. Figure 2a and 2b do the same for the length of the prepayment penalty term and the indicator for negative amortization loans. We normalize to zero the coefficient for the second half of 2003, the half year period before the pre-emption ruling occurred, and plot the remaining coefficients relative to it. The event study highlights two main points. Firstly, in the pre-period, there does not appear to be a large persistent difference among the loans originated by the two types of lenders. The coefficients for the half year prior to the pre-emption (2nd half 2003) are significantly different from zero. However, the effect is small in magnitude compared to the response following the pre-emption ruling. Secondly, looking at the dynamics of the coefficients plotted, we see that all the coefficients become significantly positive following the pre-emption ruling. The effect of the pre-emption on each of these outcomes appears to be economically large for a year and a half to two years following the event. Having considered the dynamics in detail, we move on to a more robust test of this effect. The specification above has identified an exogenous change in the loan contracts issued in states with APL laws, induced by the pre-emption ruling via the channel of the expanded choice set of OCC lenders relative to non-occ lenders. We can go one step further to generalize our results. The specification above does not control for the fact that states with APL laws were perhaps different in certain ways than states without these laws. Arguably, local mortgage market conditions may have led to states implementing these laws, and the effects of the pre-emption identified reflect the effects of these pre-existing local conditions as opposed to the effect of a larger OCC lender choice set. In a more robust approach, we would like to use as a control set not only the loans made by lenders with a different regulator (non-occ) but also those loans made in states where the pre-emption should not have had any effect as no APL laws were in place. In order to do so, we use a triple difference in difference approach. We use the following 16

17 specification: Y icgt = β 0 + β 1 P ost t OCC i AP L gt + β 2 P ost t OCC i + β 3 OCC i AP L gt + β 4 P ost t AP L gt + β 5 P ost t + β 6 OCC i + β 7 AP L gt + β 5 X it + η c + η month + ɛ igt where AP L gt indicates whether state g had a anti-predatory lending law in effect at time t, the month of origination of the loan. We define AP L gt to be equivalent to the ineffect variable of Ding et al. (2012). The coefficient of interest is β 1. It estimates: ([Ȳ AP L OCC,P ost Ȳ ] [Ȳ OCC,pre AP L AP L Non OCC,P ost Ȳ ]) Non OCC,P AP L re ([Ȳ Non AP L OCC,P ost Ȳ ] [Ȳ Non AP L OCC,pre Non AP L Non OCC,P ost Ȳ ]) Non AP L Non OCC,P re Results appear in Table 3. Our results remain largely robust to this specification. The magnitude of the effect on the origination of loans with prepayment penalties is mitigated, but remains statistically significant and economically significant. The effect in Column 1 is 26% of the unconditional mean of the dependent variable. Once again, we see that this result is particularly striking as OCC regulated lenders in APL states were in fact less likely to issue loans with prepayment penalties (as indicated by β 3 of Column 1). The results for the presence of negative amortization loans, presence of prepayment penalty term violations, and length of prepayment penalty terms respond similarly to this specification. In unreported results we see that the origination of low or no documentation loans by OCC regulators reduced relative to that of non-occ regulated lenders in non-apl states. This suggests that while in the pre-period, OCC lenders made riskier loans by issuing to low documentation borrowers, in the post period they may have resorted to more predatory lending, while at the same time being more stringent on documentation requirements. Next, we consider the effect of this pre-emption ruling on the interest rates of loans 17

18 issued by OCC and non-occ lenders. Interest rates vary across the range of mortgage products offered by the lenders, and so we alter the specifications used so far. Ideally, we want to perform the test by comparing loans of a similar type issued at a similar point in time. Thus, we construct 15 groups of loans (for example 30, 20 and 15 year fixed rate mortgages form their own groups, as do mortgages that are interest only and have various lengths of interest only terms and so on) which cover 91% of the mortgages in our original sample. Thus instead of county and month fixed effects, we use Loan Type x Quarter FE, and separately include county FE as well. Table 4 shows the results for interest rates using our difference-in-difference and triple difference-in-difference specification. As we can see, OCC lenders issued loans with interest rates that were 2 basis points higher than those issued by non-occ lenders in the pre-period. Together with the results in the earlier table, it suggests that while they were increasing the origination of these riskier loans, they were raising interest rates as well. The results align with others in the literature which show that the effect of APL laws was to lower the interest rates on loans in those states. In this section, have identified an exogenous component to the change in loan contract features in our data from the pre-period to the post-period. In the next set of analysis, we focus on how deregulation interacted with local mortgage market competition to drive the response of lenders. 5 Heterogenous treatment effects on OCC Lenders The previous section demonstrates that following the pre-emption, OCC lenders increased the origination of loans with predatory terms compared to non-occ lenders and controlling for the difference between states with and without APL laws. In this section, we more closely investigate how OCC regulated lenders responded to the pre-emption ruling. First we document considerable heterogeneity in the response of OCC regulated lenders in our sample. Next, we construct a measure of local mortgage market concentration and test 18

19 whether the response of nationally chartered lenders varied on this measure. Finally, we examine whether the pre-emption encouraged poorly performing lenders to more aggressively expand the origination of mortgages with prepayment penalties and negative amortization terms. 5.1 Documenting variation in lender response To facilitate a first test of originator response heterogeneity we restrict the sample to loans originated by the top 20 OCC lenders in our sample, in those states with APL laws in place, and then estimate the following specification: Y iclt = η c + η t + η l + β 1 X it + β 2 X it + l β 3l P ost t 1 Originator=l + ɛ iclt where the new terms, η l are originator fixed effects, and the coefficients in the summation capture separately for each originator, the effects of the various borrower level characteristics we have included as controls. Figures 3a and 3b below then plot the originator fixed effects for the regressions using Y iclt equal to either a 1-0 indicator for whether the loan had a prepayment penalty, and for Y iclt equal to the length of the loans prepayment penalty term. The omitted lender is Citibank N.A. Figure 3 then plots β 3l for each lender. The plots below highlight that there does appear to be substantial cross-sectional response heterogeneity in the loan terms extended by originators. We also run the test including a term l β l X it 1 Originator=l. Adding this term does add explanatory power, yet there remains variation in lender response. 5.2 Mortgage market competition and response of de-regulated lenders In this subsection, we explore one potential source of the variation in response of OCC lenders - the competitiveness of the local mortgage market. As we have described, the effect 19

20 of the pre-emption ruling was to expand the set of possible mortgage products nationally chartered lenders could originate, while holding fixed the set that state-chartered lenders could originate. We now wish to consider whether a OCC lender located in a highly competitive market would be more likely to take advantage of this bigger set. A lender in a very concentrated market may either have captured the market or have much ground to gain. Such a lender may be less likely to use this expanded choice set as compared to a lender in a more competitive market. In such an environment an OCC lender may be more inclined to use the prepayment penalties he was prevented from doing so. This would allow him to more profitably issue mortgages with deferred amortization features and expand market share in the county. We thus hypothesize that, as a response to the deregulation, the use of these features expanded more in competitive markets. One measure that captures competition at the county level is the Herfindahl-Hirschman Index (HHI). We use the sample of ABSNet loans originated in 2000 to 2001 to compute the Herfindahl index for each county. We use the Herfindahl as per the time period before our analysis begins to mitigate concerns about endogeneity. In performing our analysis we drop counties which had fewer than 50 loans originated in our sample. As a robustness check, we also compute the Herfindahl using HMDA data which covers the universe of loans originated and purchased in 2000 to 2001, and our results remain qualitatively unchanged. The HHI over 2000 to 2001 for each county c is calculated as: HHI c = i L c s 2 ic where s ic is the market share of lender i in county c, and where L c is the set of lenders in county c. Note that in computing the Herfindahl, we include all lenders, OCC regulated and non-occ regulated lenders. The HHI index is thus a measure of market concentration. A lower HHI corresponds to a more fragmented, more competitive market. The mean HHI in the final sample of APL state counties used in the regression is while the standard 20

21 deviation is We then estimate the following difference in difference specification Y OCC icgt = β 0 + β 1 P ost t HHI c + β 2 P ost t +β 3 X it + η c + η month + ɛ ict on the sample of loans originated by all OCC lenders. The coefficient of interest is β 1 which describes how the post period response of OCC lenders varied depending on the concentration of mortgage lending in county c. The results appear in table 5. The negative coefficient on β 1 demonstrates that the response of OCC lenders identified in Section 4 effect was larger in counties that had a lower Herfindahl Index (more competitive local mortgage market). The coefficient suggests that a 1 standard deviation decrease in the HHI is associated with a 4% increase in the presence of loans with prepayment penalties, and a 0.98 month increase in the length of the prepayment penalty term. Additionally, this is associated with a 2.6% increase in the origination of negative amortization loans, which is about 25% of the unconditional mean of the negative amortization indicator, and is thus economically significant. The results of Section 4 suggest that de-regulation on it s own did have an effect on the presence of certain mortgage features. The results of this section shed some light on the mechanism that was at play. The origination of mortgages with these terms appears to have been higher in the more competitive markets. Thus, deregulation and local competition interacted to drive the response of the OCC regulated lenders. They took advantage of the larger set of potential mortgages to compete more effectively and attempt to gain market share. 5.3 Risk-shifting by de-regulated lenders In this section, we look to further investigate which lenders used this larger set of mortgage products made possible by the pre-emption ruling, and what determined whether they did so. 21

22 To motivate our next analysis we look to the risk-shifting hypothesis. Landier et al. (2011) outline this hypothesis, wherein they describe how a lender who has received a negative shock (measured in their case by exposure to the real estate market) would want to expose themselves to loans that were more sensitive to house prices. Loans with features such as negative amortization, balloon payments, and interest only mortgages are those that were more sensitive to house prices; as most of the payments were back-loaded via these features and depended on the ability of the homeowner to refinance their mortgages. Pre-emption of the state APL laws would mean that OCC lenders would now have more of an opportunity to take advantage of such risk-shifting opportunities. These opportunities would be enabled by using prepayment penalty terms which locked borrowers out from early prepayments. We wish to test whether OCC regulated lenders exhibited more of this risk-shifting behaviour following the de-regulation. In other words, were OCC regulated lenders more likely to respond to lower returns by issuing riskier loans following the pre-emption ruling. We design our test along the lines of Titman and Tsyplakov (2010). We follow their methodology in constructing for each of our top 20 federally chartered originators, a time series measuring for each month, the cumulative returns over the past 6 months. Each loan level observation now has an additional variable measuring the returns of the originator of the loan over the six month period prior to the month of origination. In order to facilitate our analysis, we focus on the top 20 OCC lenders in our sample. Due to the concentration of lending among the OCC lenders, this accounts for about 98% of the OCC originated loans in the sample. For each of the lenders, we obtain the monthly adjusted price from Datastream, and form a time series of lagged 6-month-returns for each month that the lender originated a mortgage in the sample. We construct Return l,t 7,t 1, i.e, the 6 month return of lender l over the period, t 1, the month prior to the origination of the loan, going back to t We first consider a difference in difference specification to 10 More specifically, this is calculated as; Return t 6,t = P ricet P ricet 6 P rice t 6. 22

23 test the risk-shifting hypothesis. More specifically, we test Y OCC icgt = β 0 + β 1 P ost t Return l,t 7,t 1 + β 2 P ost t +β 5 Return l,t 7,t 1 + β 4 X it + η c + η month + η l + ɛ ict where, compared to the previous specifications, we also add lender fixed effects η l. The coefficient of interest is β 1 which indicates the sensitivity of an OCC lender s post period response to past 6 month returns, in APL states vs. non-apl states. Columns 1, 2, 4 and 5 of Table 6 show that in the post period relative to the pre-period, the lower the prior 6 month return of an OCC lenders, the more likely they were to originate loans with prepayment penalties, and longer prepayment penalty terms. In addition, they were more likely to make loans with negative amortization features. To further ensure the robustness of these results, we look to a triple difference in difference specification that will also control for the lending behavior by OCC regulated lenders in states that did not have anti-predatory lending laws. These results appear in Table 7. The results in Columns 1, 2 and 5 on the origination of loans with prepayment penalties, negative amortization loans and the length of the prepayment term remain robust to this specification. The results of this subsection indicate that lenders did appear to engage in risk-shifting behavior following the pre-emption ruling by using prepayment penalties, and longer prepayment penalty terms to make their loan portfolios more sensitive to house prices, and to issue more loans with features such as negative amortizations. Additionally, as the earlier results in this section suggest, the increase in origination of mortgages with these features was concentrated in more competitive local mortgage markets. 23

24 6 Competition and the response of non-occ regulated lenders In the previous section, we demonstrate that OCC regulated lenders responded to the preemption ruling by increasing the origination of loans with prepayment penalties and deferred amortization features. We now wish to consider the response of Non-OCC regulated lenders. The pre-emption created an un-level playing field whereby the non-occ regulated lenders still had to adhere to the state APL laws. There are two possible ways they may have responded to the change in competitive landscape. On one hand they could have increased the presence of loans with prepayment penalties, changed the prepayment penalty terms, and originated more loans with deferred amortization, while staying within the limits of the state APL laws. On the other hand, they may have increased the origination of loans with features such as interest only payments, or ARMS that were not directly governed by the laws, but were still riskier in nature. 6.1 Non-OCC Response by County Market Concentration We follow the analysis of 5.2 and construct for each county a Herfindahl Hirschman Index to measure the concentration of mortgage lending. Once again, we argue along similar lines that one should expect to see more of a response from non-occ lenders particularly in those counties where they face a competitive (less concentrated) market. To understand their behaviour in the local mortgage market, we estimate the same difference-in-difference specification as 5.2, now using the sample of loans issued by non-occ regulated lenders in states with APL laws. The coefficient of interest is once again the one on interaction between P ost t and HHI c. The results appear in table 8. The coefficients are all negative and statistically significant. They indicate that in the post period, in more competitive markets, non-occ lenders responded by increasing the origination of mortgages with the features we have investigated. In particular, a one standard 24

25 deviation decrease in Herfindahl is associated with a 2% increase in origination of mortgages with prepayment penalties, and an average increase of 0.92 months in the length of the prepayment penalty term length. Additionally a 1 S.D. decrease in Herfindahl index is related to a 3% increase in the origination of mortgages with negative amortization. Additionally, we check whether non-occ lenders increased the issuance of mortgages with Interest Only and ARM features, which were not directly covered by the state APLs. We see that in the post period, a 1 standard deviation decrease in the HHI predicts a 4% increase the origination of loans with ARM features (relative to 60% unconditional probability of ARM loan in the regression sample) and a 3.5% increase in the origination of loans with Interest Only payments (relative to 23% unconditional probability in the sample). 6.2 Non-OCC Response Heterogeneity by County Market Share To further understand the importance of local mortgage market competition, we look at another variable which captures the exposure of a non-occ lender to the change in the competitive environment induced by the pre-emption ruling. We compute within our sample a measure of the market share in each county captured by OCC regulated lenders in 2000 and We then incorporate this term into our difference-in-difference specification in place of HHI c. In essence, we are comparing the response of a non-occ regulated lender in a county where he faces a steeply un-level playing field following the preemption, to that county in which OCC lenders do not originate a lot of loans. We should expect them to respond particularly in those counties where OCC market share is high, as this is where the competitive landscape has changed the most. The specification we use is: Y Non OCC icgt = β 0 + β 1 P ost t OCC Share c + β 2 P ost t +β 3 X it + η c + η month + ɛ ict 25

26 where OCC Share c is the OCC-lender market share as described above. Table 9 shows the results. β 1 is the coefficient of interest. The positive coefficients on β 1 suggest that the higher the county market share of OCC lenders the more likely the presence of predatory terms in loans, in particular prepayment penalties, longer prepayment penalty term length, and negative amortization features. The mean OCC Share c is with a standard deviation of This suggests that a 1 S.D. increase in OCC Share c is associated in the post period with a 2% increase in the origination of loans with prepayment penalties, a 6.4% increase in the origination of loans with negative amortization, a 6.6% increase in the origination of loans with Interest Only Payments and a 6.5% increase in the origination of loans with ARM features. The results point to the possibility that the pre-emption ruling may have had perverse implications for how non-occ regulated lenders responded in particular markets, via the channel of competition. Note that the pre-emption ruling did not directly affect non-occ regulated lenders. The set of loans they could possibly originate was the same before and after the pre-emption ruling. What did change with the ruling was the set of loans that their OCC regulated competitors could issue. In the post-period, in a county with a smaller OCC market share, the effect of the pre-emption ruling is smaller. Counties with a higher OCC market share saw non-occ regulated lenders use prepayment penalties more frequently, and additionally, originate more mortgages with features such as Interest Only payments and ARMs. These were dimensions on which they could compete with the OCC regulated lenders as state APL laws did not restrict the use of IO and ARM loans. It is possible that non-occ lenders in high OCC share counties responded by using these terms up to the limits imposed by the APL laws that still applied to them. A closer analysis can be performed by looking specifically at each states APL law and assessing how close non-occ lenders came to violating various dimensions of it before and after the preemption ruling. To investigate this, we construct a measure that captures the distance from 26

27 the loan contract s prepayment term length to the limit of the state APL law. 11 For example, if the law states that the prepayment penalty term for covered loans cannot be more than 36 months, our measure for a loan with 24, 36, 48 month prepayment penalty term will be -12, 0, and 12 respectively. We then divide our sample into 5 quantiles based on the OCC market share, and consider how the distribution of our measure for loans originated by non-occ regulated lenders changes from the pre-period to the post period. We see some suggestive evidence that non-occ lenders issued a larger percentage of loans closer to the state APL limits following the pre-emption ruling. A more robust analysis is required to determine the precise magnitude and nature of this shift. The results of this and the previous section point to a competition channel at play which induced a response even from those lenders who were not subject to the deregulation. It is further evidence of this interaction between deregulation and local mortgage market competition. Loans with features that the APL laws looked to curb performed poorly in the aftermath of declining house prices. Our results are thus suggestive of a race-to-the-bottom in which deregulated OCC lenders took advantage of a larger set of possible mortgage products and induced more aggressive lending from their non-occ regulated competitors. 7 Robustness In this section, we further test the validity of our identification strategy. In particular, we restrict our sample to high-cost loans and run our baseline difference-in-differences specification. This test is complematary to that of Di Maggio and Kermani (2014), who show that the amount of high-cost loans and mortgages with high debt-to-income ratio increased significantly in states with APL laws and in regions with a higher presence of national banks. The state APL laws defined the set of loans the laws were applicable to using either an interest rate threshold, or a points and fees threshold. We do not observe the 11 using the classification as in (Bostic et al. 2008) of the strength of predatory payment term related state APL law. 27

28 points and fees paid by borrowers, but we do observe the contract rate or APR on the loan. The rate thresholds were defined in terms of the yield on Treasury Securities of similar maturity. However, given the difficulty in determining the maturity of a mortgage contract with various features, we use an alternative definition of high-cost loans as used by HMDA following In HMDA,...a loan is considered high cost if the difference between loan APR and a survey-based estimate of APRs currently offered on prime mortgage loans of a comparable type is equal to or greater than 1.5 percentage points for a first-lien loan.... To apply this definition, we obtain average prime rates on Fixed Rate and Adjustable Rate Mortgages from FHFA s Monthly Interest Rate Survey. Table 10 shows the results of our baseline specification on this sub-sample. As can be seen, our results remain robust to this specification. The coefficient of interest on the Prepayment Penalty Indicator remains economically significant, as do those for the Negative Amortization and Balloon regressions. One may be concerned that the effect of the pre-emption ruling on loan features was limited to those borrowers who were able to obtain prime rates, putting into question the predatory nature of the national bank response. The test here shows that this was not the case. The economic significance of the effect of the pre-emption ruling on this subsample of high cost loans is comparable to the broader sample. 8 Conclusion In this paper, we the pre-emption of state anti-predatory lending laws for banks regulated by the OCC - as a quasi-experiment to test for the effect of deregulation on the supply of complex mortgages. This was a shock which expanded the set of loans OCC-regulated lenders were allowed to make while leaving unchanged the set that non-occ regulated lenders were allowed to make. To further make our results robust, we are able to also use as a control group those loans made in states that did not have anti-predatory lending laws in place. Our research design and identification strategy allows us to explore the increase in supply of 28

29 these loans in further detail. We first show that the supply of loans with negative amortization features, prepayment penalties and balloon features increased as a response to the deregulation. Additionally, the length of prepayment penalties increased. Prepayment penalties enable the profitable use by the lenders of features such as interest only, or negative amortization. Additionally, while most state APLs did not fully restrict prepayment penalties, they did curb the length of the prepayment penalty term. Our initial results confirm our initial hypothesis that the supply of complex, and perhaps predatory, mortgages increased in response to the deregulation. Next, we look to further investigate the mechanism via which these effects are propagated. In particular, we study separately the lending response of OCC regulated lenders and non- OCC regulated lenders. Considering the response of OCC-regulated lenders, we find that, following the pre-emption ruling, they increased the origination of loans with the features discussed particularly in those markets where they faced the most competitive environment. Additionally, for each lender, the supply of these complex mortgages became more responsive to recent returns - the higher the recent returns, the less likely a lender was to issue these risky loans. Our results indicate that lenders were more likely to respond to risk-shifting incentives as a result of the deregulation. This suggests that poorly performing lenders appeared to have taken advantage of the de-regulation in an attempt to compete for rents in local mortgage markets. Finally, we explore how local mortgage market competition between lenders regulated by different agencies may have had perverse effects. We show that in counties where OCC lenders had a high market share, non-occ lenders became more aggressive in the origination of loans with prepayment penalties to the extent permitted by state APL laws. In addition, they increased the origination of loans with interest only payments and ARMs, features not directly controlled by the state APL laws. We find a similar response of non-occ regulated lenders in mortgage markets that were more competitive (less concentrated). This is striking because these non-occ regulated lenders were not directly affected by the pre- 29

30 emption ruling. Our evidence is suggestive of a competition channel that ignited a race to the bottom and induced a potentially adverse response even from those lenders who continued to fall under the regulation. 30

31 References Adelino, M., A. Schoar, and F. Severino (2012). Credit supply and house prices: Evidence from mortgage market segmentation. Agarwal, S., G. Amromin, I. Ben-David, S. Chomsisengphet, and D. D. Evanoff (2014). Predatory lending and the subprime crisis. Journal of Financial Economics 113 (1), Agarwal, S. and I. Ben-David (2012). Did the community reinvestment act (cra) lead to risky lending? Available at SSRN Agarwal, S. and D. D. Evanoff (2013). Loan Product Steering in Mortgage Market. Available at SSRN Agarwal, S., D. Lucca, A. Seru, and F. Trebbi (2012). Inconsistent regulators: Evidence from banking. Amromin, G., J. Huang, C. Sialm, and E. Zhong (2013). Complex mortgages. Technical report, National Bureau of Economic Research. Amromin, G. and C. Kearns (2014). Access to refinancing and mortgage interest rates: Harping on the importance of competition. Barlevy, G. and J. D. Fisher (2010). Mortgage choices and housing speculation. Technical report, Working Paper, Federal Reserve Bank of Chicago. Bond, P., D. Musto, and B. Yilmaz (2009). Predatory mortgage lending. Journal of Financial Economics 94 (3), Bostic, R. W., K. C. Engel, P. A. McCoy, A. Pennington-Cross, and S. M. Wachter (2008). State and local anti-predatory lending laws: The effect of legal enforcement mechanisms. Journal of Economics and Business 60 (1), Campbell, J. Y. (2006). Household finance. The Journal of Finance 61 (4),

32 Chinco, A. and C. Mayer (2014). Misinformed Speculators and Mispricing in the Housing Market. Technical report, National Bureau of Economic Research. Di Maggio, M. and A. Kermani (2014). Credit-Induced Boom and Bust. Columb. Ding, L., R. G. Quercia, C. K. Reid, and A. M. White (2012). The impact of federal preemption of state antipredatory lending laws on the foreclosure crisis. Journal of Policy Analysis and Management 31 (2), Elliehausen, G., M. E. Staten, and J. Steinbuks (2006). The effects of state predatory lending laws on the availability of subprime mortgage credit. Favara, G. and J. Imbs (2010). Credit Supply and the Price of Housing. Available at SSRN: Greenstone, M. and A. Mas (2012). Do Credit Market Shocks affect the Real Economy? Quasi-Experimental Evidence from the Great Recession and Normal Economic Times. MIT Department of Economics Working Paper No Haughwout, A., D. Lee, J. Tracy, V. der Klaauw, and H. Wilbert (2011). Real estate investors, the leverage cycle, and the housing market crisis. Staff Reports, Federal Reserve Bank of New York. Ho, G. and A. Pennington-Cross (2006). The impact of local predatory lending laws on the flow of subprime credit. Journal of Urban Economics 60 (2), Ho, G. and A. Pennington-Cross (2008). Predatory lending laws and the cost of credit. Real Estate Economics 36 (2), Jayaratne, J. and P. E. Strahan (1996). The finance-growth nexus: Evidence from bank branch deregulation. The Quarterly Journal of Economics, Jiang, W., A. A. Nelson, and E. Vytlacil (2014). Liar s loan? Effects of origination channel and information falsification on mortgage delinquency. Review of Economics and Statistics 96 (1),

33 Keys, B. J., T. Mukherjee, A. Seru, and V. Vig (2010). Did securitization lead to lax screening? Evidence from subprime loans. The Quarterly Journal of Economics 125 (1), Landier, A., D. Sraer, and D. Thesmar (2011). The risk-shifting hypothesis: Evidence from subprime originations. TSE Working Paper 11. Mian, A. and A. Sufi (2009). The consequences of mortgage credit expansion: Evidence from the US mortgage default crisis. The Quarterly Journal of Economics 124 (4), Nadauld, T. D. and S. M. Sherlund (2013). The impact of securitization on the expansion of subprime credit. Journal of Financial Economics 107 (2), Purnanandam, A. (2011). Originate-to-distribute model and the subprime mortgage crisis. Review of Financial Studies 24 (6), Rajan, U., A. Seru, and V. Vig (2010). The failure of models that predict failure: distance, incentives and defaults. Chicago GSB Research Paper (08-19). Rosen, R. (2005). Switching primary federal regulators: is it beneficial for US banks? Economic Perspectives (Q III), Titman, S. and S. Tsyplakov (2010). Originator performance, cmbs structures, and the risk of commercial mortgages. Review of Financial Studies, hhq

34 A Additional Results on OCC Lender Response Heterogeneity A.1 OCC Lender Response by Geographic Characteristics Here we wish to further explore the heterogeneity in the cross sectional response First, for each county, we collect data from HMDA and the Census to compute the share of borrowers in each county in 2001 that had a FICO score below 680. We want to answer the question of whether the de-regulation and the response of the lenders subject to this de-regulation was averse to a specific group of borrowers. Bond et al. s (2009) motivate their model of predatory lending by pointing out that lenders may possess superior information about the borrower s expected future income, and potentially use this to make predatory loans. This suggests that financially unsophisticated borrowers should be relatively more susceptible to predatory lending, and thus, the effects of the pre-emption. We use the both the difference in difference and the triple diff-in-diff specification to test this hypothesis. The triple diff-in-diff specification is as follows: Y icgt = β 0 + β 1 P ost t AP L gt F raction < 680 c +β 2 P ost t F raction < 680 c + β 3 F raction < 680 c AP L gt + β 4 P ost t AP L gt + β 5 P ost t + β 6 F raction < 680 c + β 7 AP L gt + β 5 X it + η c + η t + ɛ icgt Table 5 shows the results for the difference-in-difference and the triple difference in difference (including the non-apl states as a control group) for dependent variables related to prepayment penalties. In the triple diff-in-diff specification the interaction of interest is P ost t AP L gt F raction < 680 c. Column 4 shows that the OCC lenders were more likely to extend loans with prepayment penalties in those counties that had a higher fraction of borrowers with a credit score below 680. We follow the literature in using this credit score as the cut-off for subprime borrowers. Additionally, the higher the fraction of subprime 34

35 borrowers in a county, the larger the increase in prepayment penalty term lengths following the pre-emption ruling (as seen in Column 6). The coefficients on AP L gt F raction < 680 c result from the fact that states that implemented APL laws perhaps did so to protect these unsophisticated borrowers. Table 7 shows the results of these specifications on negative amortization and balloon loans. We see that counties with larger populations of subprime borrowers were in fact less likely to see an increase in the origination of mortgages with negative amortization features, but more likely to see origination of balloon loans However, only the effect on balloon loans remains robust to using a triple diff-in-diff specification. Next, we collect data from the census on the log of the median county income in We run the equivalent specification as above, with LnMedIncome c as the variable in place of F raction < 680 c. Table 8 shows the results for prepayment related dependent variables and table 9 shows the results for other important loan features as the dependent variables. We find similar results; in the APL states vs. the non-apl states, OCC lenders increased the issuance of loans with predatory terms such as prepayment penalties and balloon payments, and increased the lengths of the prepayment penalty terms offered. The diff-in-diff results using only the sample of loans issued in states with APL laws supports this evidence, however the coefficient for Prepayment penalties is smaller and not statistically significant. This is consistent with the fact that APL laws did not always prohibit prepayment penalties but severely limited their amount (on which we don t have data) and their term lengths. Thus we see the effect on the length of the term in this simpler specification, but not on the presence of this particular contractual feature. The results of the prior section show that OCC lenders were associated with origination of more loans with predatory and complex features following the pre-emption ruling as compared to lenders who still had to adhere to state anti-predatory lending laws. This section explores further the geographies in which they did this increased complex lending. The results suggest that the lending did in fact seem to be predatory in nature, with the likelihood of the origination of the loans increasing in the fraction of a county s borrowers 35

36 who were subprime, and decreasing in a counties median income. 36

37 Figure 1: Event Study Regression Coefficient Plots (a) Pre-emption of national banks from state anti-predatory lending laws and origination of loans with prepayment penalties (b) Pre-emption of national banks from state anti-predatory lending laws and origination of loans with prepayment penalty term violations

38 (c) Pre-emption of national banks from state anti-predatory lending laws and length of prepayment penalties (d) Pre-emption of national banks from state anti-predatory lending laws and origination of loans with negative amortizations

39 Figure 2: Pre-Emption Response by Originator (a) Prepayment Penalty Indicator

40 (b) Length of Prepayment Penalty Term