A New Look at the U.S. Foreclosure Crisis: Panel Data Evidence of Prime and Subprime Lending. Preliminary Draft: Feb 23, 2015

Transcription

1 A New Look at the U.S. Foreclosure Crisis: Panel Data Evidence of Prime and Subprime Lending Preliminary Draft: Feb 23, 2015 Fernando Ferreira and Joseph Gyourko The Wharton School University of Pennsylvania and NBER Abstract Much of our understanding about the U.S. foreclosure crisis is based on research focused exclusively on the subprime sector at the beginning of the housing bust. But subprime mortgages are a relatively small part of the mortgage market (15%) and the largest share of foreclosures occurred over a longer period of time. In this paper we investigate how all types of home purchase financing (prime lending, subprime lending, special government-insured lending (FHA/VA), small lenders, and all-cash transactions) contributed to the foreclosure crisis by creating a nationwide micro panel data set of over 33 million home ownership sequences that includes 96 MSAs, and extends from Subprime foreclosures did surge one year ahead of the other types of financing, and subprime mortgages also have unconditional probabilities of foreclosure around two times higher than prime and government-insured loans. However, prime mortgages have almost twice the number of foreclosures of the subprime sector and more than five times when compared to FHA/VA loans. Furthermore, the empirical investigation finds that a large part of the higher subprime probability of foreclosure can be explained by negative equity and borrower liquidity--two factors more associated with the economic cycle and emphasized by the traditional mortgage default literature. Other observed factors that have been proposed by the literature as potential determinants of foreclosures, such as owner race and initial income, whether the buyer is a speculator, and housing features, have little or no economic impact. Overall, the recent foreclosure crisis was much less of a uniquely subprime mortgage market event than is commonly perceived. We appreciate the excellent research assistance of Matt Davis, Lindsay Relihan, Yitong Wang, and Chen Zheng. We thank Raphael Bostic, Nancy Wallace, Paul Willen and of participants at the NBER Real Estate Summer Institute and Homer Hoyt Institute meetings for comments and suggestions. We are grateful to the Research Sponsors Program of the Zell/Lurie Real Estate Center at Wharton for financial support. 1

2 I. Introduction Much economic analysis of the foreclosure crisis during the recent housing bust focuses on the role of subprime mortgages. 1 Our review of the literature presented in section II shows that roughly three-quarters of the published articles in this area only use subprime lending data. Mian & Sufi (2009) conclude that a salient feature of the mortgage default crisis is that it is concentrated in subprime ZIP codes throughout the country. Viewing the housing bust largely through a subprime lens also has facilitated research on issues ranging from incentive conflicts in mortgage securitization (Keys, et. al. (2009, 2010)) to risky contract design (Demyanyk & Van Hemert (2011); Piskorski, Seru & Vig (2010)). That subprime loans only constitute 15% of the overall mortgage market in our data from , and never more than 20% in any one year raises questions about the representativeness of results using information only from this sector. Prime loans make up the majority of loans in any year (61% overall), with other types of financing such as Federal Housing Administration (FHA)- or Veterans Administration (VA)-insured loans and cash only transactions have non-trivial market shares of 10%-11% in many years. Moreover, the literature mostly uses data up to 2008, but most foreclosures happened after that year. This suggests an analysis of the entire mortgage market over a longer period of time may provide a better understanding of the foreclosure crisis. The traditional literature on mortgage default also calls for a broader approach. It emphasizes the role of two factors negative equity (arising from house prices dropping below mortgage balances) and borrower illiquidity (arising from negative income shocks) independent of the type of mortgage contract or whether the loan was securitized. Negative equity affects the desirability of repaying the debt from the borrower s perspective, and illiquidity affects the borrower s ability to repay. Foote, et. al. (2010) term this combination of 1 There is no legal definition of what constitutes a subprime loan or borrower. Colloquially, the subprime sector is meant to reflect loans made to borrowers who could not quality for a conforming loan - because of an insufficient down payment, unstable income, or low credit scores. Researchers employ one of two strategies to distinguish subprime from prime borrowers: i) Credit score cutoffs somewhere between 600 and 650; or ii) lists of subprime lenders compiled by the Department of Housing and Urban Development (HUD) and industry publications. In this paper we use the latter method and provide all details in the data section. Both methods have shown that subprime mortgages have very high foreclosure rates (e.g., see Chomsisengphet and Pennington-Cross (2006), Foote, Gerardi and Willen (2008), Mayer, Pence and Sherlund (2009), Mian and Sufi (2009), in addition to the current paper.) ). 2

3 factors the double trigger hypothesis. 2 From this perspective, the foreclosure crisis is more of a consequence of the economic cycle that would likely impact the entire housing market, not just one segment such as subprime. This paper provides an empirical analysis of the foreclosure crisis that integrates the different strands in the literature. Our goals are to provide key stylized facts about the complete housing market including subprime and prime lending - and to identify which factors, if any, can explain the different propensities to foreclose across types of housing financing. In addition to negative equity and borrower illiquidity, this includes a host of other potential factors mentioned in the recent literature, such as housing structure traits, household traits such as race and buyer initial income, and whether the buyer of the house is a speculator. We do so by amassing a new panel data set that captures the entire owner-occupied housing market across 96 MSAs from This micro data set contains information on over 33 million unique ownership sequences on just over 19 million distinct owner-occupied housing units in these markets, resulting in almost 800 million quarterly observations. Owners are categorized by financing type, with Figure 1 plotting the share of the four most prominent types. The rise of the subprime sector clearly is visible in this plot, as it doubled between to just over a 20% share of all owners in our sample. The rise of Subprime owners largely is at the expense of Government-insured owners, as Prime owners slowly kept increasing their market share. Cash buyer owners are a consistent 10%-12% of the sample until the beginning of the financial crisis, after which their share increases. Figure 2 then plots the share of each type of owner in distress, which we define as having lost the home through foreclosure or a short sale. 3 The subprime share in home losses was higher than that for prime mortgage borrowers only for a brief time in 2006 and 2007 as the housing cycle crested and began to turn down. By 2008, however, distress among prime mortgage borrowers had spiked, and their share continued to rise for years. The only drop in homes lost to foreclosure is among owners who took out FHA/VA-insured loans, and that is allegedly due to their sharp decline in overall market share (see Figure 1). The distress among 2 The terminology arises as follows. Negative equity is one trigger, but is a necessary, not a sufficient, condition for default. If the owner subsequently suffers a negative income shock that renders him unable to make monthly debt service payments in a timely manner and generally illiquid, that is the second trigger which guarantees default because not even a sale of the property can pay off the outstanding mortgage balance. 3 Foreclosures are defined by the timing of when title to the property is transferred to another entity. A short-sale means that the value of the house transaction is less than the balance of the mortgages held by the seller. We discuss both measures in the data section. 3

4 all cash buyers is mostly due to local taxes not being paid. Over the complete time period, prime mortgages have twice the number of foreclosures observed in the subprime sector. Given the much lower number of subprime mortgages, the subprime sector has an unconditional probability of foreclosure well over two times higher than the prime sector over our full sample period. Figure 3 shows how these unconditional probabilities vary over time. Until the latter half of 2006, the highest rate of foreclosure and short sales occurred among FHA/VA-insured borrowers. The year with the lowest risk of distress is 2005, but conditions began to deteriorate in 2006, and distress rates spiked sharply in That year, 2.94% (or 1- in-34) Subprime owners lost their homes while 0.80% (or 1-in-125) of Prime owners did so. Peak distress rates among Subprime owners occurred in 2008, when 6.60% or 1-in-15 lost their homes via foreclosure or short sale. Over 2% of Prime owners lost their homes that year, but their rate did not peak for another two years, at 3.03% (1-in-33) in What factors explains those differential propensities to foreclose? We model distress as a function of the home purchase financing types: subprime, prime, government-insured loan, small lender, and all-cash buyer (the omitted category). The micro panel data allows the estimation of propensities to foreclose within ownership sequences (more on this below) and within neighborhoods. The within-neighborhood variation could be potentially important since comparisons across geographies may suffer from omitted factors or differential trends. It turns out that the subprime prime foreclosure gap is only slightly reduced within neighborhood (as opposed to across) in part because we observe that both forms of financing are widespread across neighborhoods in our data. We then directly investigate the impact of negative equity within the same framework. Negative equity materially increases the probability an owner will lose the home, and its influence increases with the degree of negative equity. For example, owners with extremely high loan-to-value ratios (LTVs) above 2 are about ten times more likely to lose their home via foreclosure or short sale than are owners with only slight negative equity (LTVs barely above 1). That these results hold within neighborhoods and within ownership sequence suggests that the estimated effects are not due to omitted factors. Moreover, negative equity can explain nearly three-quarters of the higher subprime propensity to foreclose relative to all cash owners and the totality of the prime and governmental propensities (also relative to all cash owners). It also reduces the subprime-prime gap in the propensity to foreclose by about one-third. 4

5 We also analyze other measures that influence current LTV which have received attention in the recent literature: Initial LTV, whether the owner refinanced or took out a second loan, and cohort of last purchase or mortgage transaction. Initial LTV was shown to explain about 60% of the foreclosure crisis according to simulations based on a macro model of housing markets developed in Corbae and Quintin (2015); Mian and Sufi (2011) found that home-equity based borrowing may explain one-third of mortgage defaults between 2006 and 2008; and Bayer, Ferreira, and Ross (forthcoming) and Palmer (2013) show that cohort effects may explain some of the movements in defaults and prices respectively. Our estimates corroborate the importance of the more detailed negative equity variable: initial LTV and cohort dummies play a role in explaining the foreclosure gaps, but it is much smaller than for current LTV. Refinancing activity, on the other hand, explains a negligible part of the subprime-prime gap. This is not surprising since both prime and subprime owners had extremely high rates of refinancing activity during the housing boom. A weakness of all empirical research in this area is the absence of information on individual homeowner unemployment status, a key illiquidity factor emphasized in the traditional default literature. 4 Our huge data base allows us to address this issue indirectly by estimating a household fixed effects specification. These fixed effects allow us to control for permanent unobserved factors such as wealth, general employability and attachment to the labor force at the household level. Naturally, household fixed effects also capture other features, such as initial LTV and the permanent neighborhood effects. Just using the within-ownership sequence variation reduces the subprime prime gap in foreclosure rates by 71% (when compared to the within neighborhood model). Adding negative equity increases the reduction in the gap to 81%. Interestingly, other observed fixed factors such as race, gender, self-reported income at the time of purchase, whether the buyer is a speculator, and house size, have negligible effects. Overall, the subprime and prime mortgage labels matter much less once we account for fixed individual features likely associated with the illiquidity channel. 4 The exception is Foote et al (2013) who use the Panel Survey on Income Dynamics (PSID), which does report individual unemployment status. The problem with that data source is the very small sample of households it yields. We also experimented with aggregate employment measures, but as expected, they had little or no impact on further reducing the gap between subprime and prime foreclosure rates. Large attenuation bias is to be expected when an aggregate measure is used to proxy for individual unemployment status. For example, if the local unemployment rate doubles from 5% to 10% in a quarter, 90% of the labor force still is employed, so regressing whether one lost a home to foreclosure or engaged in a short sale on that aggregate variable is not likely to be statistically significant (Gyourko and Tracy (2014)). 5

6 It is noteworthy that the household fixed effects specification is identified by variation in households that switched their type of financing at some point during their ownership spell. These switchers comprise one-quarter of all ownership sequences in our data, and are particularly prevalent in the big refinancing wave during the build-up to the peak of the housing boom. If they are a random subsample of the population, then the results would apply for the whole population. However, one can envision two-sided selection leading to higher distress rates in the sample of Subprime switchers. That is, households with declining credit quality could be switching from prime to subprime mortgages, while borrowers with improving credit prospects could be switching from subprime to prime mortgages. But our empirical estimates do not find that subprime borrowers are more likely to lose a home after controlling for household fixed effects. While the switchers are not a random sample, they do have similar homes in physical terms (i.e., same units sizes, number of bedrooms, etc.) to the rest of the population, and they only slightly differ along other lines: switchers have 3.8 percentage point higher shares of Blacks and Hispanics, and they have 7 percent smaller median self-reported income. 5 In sum, the recent foreclosure crisis was much less of a uniquely subprime event than what might be expected given the focus of much recent research on this sector. The large unconditional propensities to lose one s home to foreclosure or short sale observed among subprime borrowers can be explained largely by the negative equity and borrower illiquidity channels. Our estimates also have a number of potentially important implications for policy. For example, they suggest that macroprudential regulation focused on the details of the mortgage contracts in the subprime sector (i.e., banning some types of adjustable rate mortgages for example) does not appear to address the majority of taxpayer or cyclical risk satisfactorily. Also, there is evidence consistent with the conclusion that being unlucky in terms of buying a home, especially a highly leveraged one, or refinancing near a price peak can be vitally important, regardless of mortgage type. 6 Many issues remain for future research. For example, we presume the choice of financing type is exogenous and estimate how much of the variation in home loss by different types of financing can be explained by observed factors suggested by the literature. Our 5 Those switchers could also be different from pure subprime borrowers with respect to foreclosure behavior. But we show that not only switchers correspond to a large fraction of subprime borrowers, but also that pure subprime borrowers are not very different than switchers based on our observables. 6 Bayer, Ferreira, and Ross (forthcoming) also find that the racial gap in defaults and foreclosures is correlated with financing a home close to the peak of local housing booms. 6

7 empirical strategy benefits from the fact that prime and subprime borrowers are more comparable within neighborhoods and within ownership sequences, but that still is not direct randomization. This is an important step towards a fuller understanding of the foreclosure crisis, but future work could also try to model the own versus rent decision as well as the choice among different types of financing. Addressing the data limitation associated with measuring borrower illiquidity status at the household level is another important item. It will no doubt be difficult, but an essential line of research will involve linking labor market micro data with real estate micro data in order to pin down the relationship between unemployment shocks and home loss. 7 The paper proceeds as follows: Section II discusses the related literature, section III describes out data, section IV provides the empirical results, and section V concludes. I. Related Literature: Implications of the Focus on Subprime Because default and foreclosure rates first spiked among subprime mortgage borrowers, researchers have paid particular attention to that sector. Appendix Table 1 reports a list of published papers on the subprime crisis since While not exhaustive, it is representative of primarily empirically-oriented work on the fallout from the housing bust. 8 There are a number of noteworthy patterns in this research. First, three-quarters of the papers focus exclusively on the subprime sector (see column 2) in their data and analysis. 9 Second, the vast majority of the studies analyzing borrower payment behavior use loan level data (see column 3), with the predominant data source being First American Core Logic Loan Performance (LP, hereafter; see column 4 for the specific data provider). This source is based on subprime mortgages which were used to collateralize private label mortgage-backed securities 7 Such data would also be useful for estimating house lock-in effects, which is the opposite of the foreclosure effect. See Ferreira, Gyourko, and Tracy (2010, 2012). 8 The articles listed here are published pieces in three urban/real estate journals (Journal of Urban Economics, Real Estate Economics, Journal of Housing Economics), select finance journals (Journal of Finance, Review of Financial Studies, Journal of Financial Economics), select banking-related journals (Journal of Banking and Finance, Journal of Monetary Economics) and general interest economics journals (American Economic Review, various American Economic Journals, Quarterly Journal of Economics, and the Review of Economics and Statistics). This list does not contain unpublished working papers. It also excludes a host of published work on related topics including local spillovers of foreclosures (e.g., Biswas (2012), Campbell, Giglio and Pathak (2011), Chan, et. al. (2013), Cheung, Cunningham and Meltzer (2014), Harding, Rosenblatt and Yao (2009), Schuetz, Been and Ellen (2008), and Whitaker and Fitzpatrick (2013)) and macroeconomic effects of the housing boom and bust (e.g., Mian and Sufi (2011, 2014)). 9 The closest anybody comes to our goal of analyzing distress in the broader mortgage or housing market is Ascheberg, et. al. (2013). It notes the rise borrower defaults as the housing bust unfolded, but takes a very different research approach from ours. Those authors construct a dynamic simulation model to impute the spillover effects of subprime defaults on prime defaults. 7

8 (MBS). The strength of the LP data is that they are rich in detail on loan traits. 10 Creative researchers also merged these data with credit bureau files (from Equifax typically) which provides detailed credit risk information on the borrower. This combination provides the foundation for much of the empirical work listed in the table on the role of loan and borrower traits in explaining subprime default patterns in particular. A countervailing weakness of this type of data is that it cannot be used to generate panels of ownership sequences, unless the owner never changes the debt it uses. Hence, these studies generally cannot control for the addition of new debt or link refinancing of original debt across a unique owner. Cumulative LTVs cannot be known with accuracy, as a result, unless these data are merged with a credit bureau panel. Most of these studies also use data from a relatively short window of time (see column 5). The private label subprime mortgage-backed securities market did not really boom until late in the last housing cycle, so these data do not become nationally representative until the middle of the last decade. The information in the table also shows that the vast majority of studies do not use originations from later than 2008, and it is extremely rare for borrower outcomes to be tracked past that year, too. As noted above, this is two years prior to the peaking of prime borrower distress rates. 11 Geographic coverage tends to be fairly wide, as many studies use national (or multi-state) samples, and some have made considerable effort to control for differences in economic (and housing price) conditions across metropolitan areas. However, there has not yet been an extensive effort to control for detailed location effects within a metropolitan area. Finally, there have been a number of studies on special topics or features of the subprime market, such as the impact of securitization on the all-on costs of origination, on lender screening incentives, and on incentives to renegotiate or work out distressed loans. There also have been separate studies on predatory lending and the prevalence and impact of untruthful data (e.g., liar loans ). II. Data Description 10 See Mayer, Pence and Sherlund (2009) for an excellent overview of this data source. Other similar data providers listed in the table include LPS Applied Analytics (which includes data on prime and subprime mortgages), Black Box Logic LLC, the HMDA files (which cover prime and subprime borrowing), and OCC-OTS Mortgage Metrics (which also covers prime and subprime mortgages). 11 Palmer (2013), unpublished, is an exception in its tracking subprime payment histories through April

9 The home purchase and financing transactions files compiled by DataQuick are the foundation of the rich micro data used in this paper. They permit us to observe sales transactions, the financing associated with those purchases, and non-purchase financings (i.e., refinancings and the taking on of subordinate mortgages) of existing owners. Our sample includes these data from the 96 metropolitan areas listed in the Appendix Table 2, along with their start and ending dates. As the appendix table notes, different metropolitan areas enter the sample at different times, some as early at 1993(1). Our analysis focuses on the (3) period for which we have complete data on most MSAs. Detailed information is provided on the following variables (among others): (a) transaction date; (b) name of buyer if the observation is for a purchase; name of owner if the observation is for a refinancing or other debt; (c) name of seller if the observation is for a home purchase; (d) name of up to three lenders for any type of transaction involving new debt; (e) sales price for all home purchases; (f) mortgage amounts for up to three loans on all observations using any type of financing; (g) street address and census tract of the underlying home; (f) various home characteristics including age of the home, size as reflected in the number of bedrooms, bathrooms, and square footage, etc.; and (g) codings provided by DataQuick indicating whether a transaction involves a home being foreclosed by a creditor, as well as whether the home is being sold out of foreclosure to a new owner; in both cases, names of the principals are reported, along with a purchase price for the latter type of transaction. Because individual owners and all their financings can be tracked over time, we use these data to create a panel of individual ownership sequences. An ownership sequence is the complete span of time a unique owner owns a given residence. Our final panel contains 33,545,252 ownership sequences on 19,648,475 homes. There are just under 800,000,000 quarterly observations on these ownership sequences. Summary statistics on a wide variety of variables for our final sample are reported in the first two columns of Table 1. A. The Number and Types of Transactions Sales transactions are explicitly identified in DataQuick so that we can distinguish between sales of new and existing homes. The predominant type of transaction is an arms-length purchase of an existing home (coded as Arms Length Resale by DataQuick). These constitute 80.2% of all our home sales transactions. Arms-length sales of new homes from the builder (or 9

10 other entity) to a household make up another 11.2% of all purchases (typically coded as Arms Length Other by DataQuick). 12 The remaining sales observations are comprised of purchases out of foreclosure (8.6%). DataQuick does not code these as arms length trades between two disinterested parties, but they are readily identifiable from another variable categorizing distress transactions. 13 Some do not consider these normal sales, but they certainly are home purchases, and we count them as such. Their transaction prices also are included in the price series described below, although we can do all our analysis excluding them. Our panel contains almost 36 million purchases of one of these three types. We also observe 48 million financings not associated with a home purchase. These include refinancings and the taking on of junior debt. First, second and third loans at purchase are clearly identified, so it is straightforward to sum them to compute a loan-to-value ratio when the house is bought. But DataQuick does not identify whether a later financing represents a refinancing of existing debt or the taking on of net new debt. We adopt the following rule to distinguish between a refinancing and adding a junior lien to prior debt. If a new mortgage taken out subsequent to purchase has an initial loan balance that is more than 50% of the total mortgage balance taken out at purchase or is more than 50% of the imputed current price of the home, we assume the new loan is a refinancing. That is, we presume it replaces the prior debt; otherwise, we assume it represents junior debt, which is added to the outstanding loan balance for the purposes of computing loan-to-value ratios. Using this rule, we observe 34 million refinancing and over 14 million second loans. B. Classifying Owners Each ownership sequence is classified as one of five types based on the type of financing used by the owner. The most straightforward is those who own in a given quarter without the use of any debt. These are referred to as Cash owners in Table 1 and in Figures 1-3 above. They constitute a relatively stable 10%-11% share of our sample until 2010, after which their share increases to over 16% in We can confirm new home sales by analyzing another variable indentifying the year the home was built, as well as the name of the seller. The former allows us to exclude land sales (which occur prior to time the structure was built). For new homes, the seller usually is a home builder. 13 The seller in these cases typically is some type of financial entity, while the buyer usually is a household. See the discussion below for more on these transactions. 10

11 All other owners used some type of debt. There is no legal definition of what constitutes a subprime or prime loan. The research cited in Section II using loan level data typically uses a credit score cutoff to distinguish subprime from prime borrowers. Other research relies on lender lists compiled annually by HUD since 1997 or industry publications such as Inside Mortgage Finance, which reports the top 20 subprime lenders each year from 1990-onward, to make this distinction. 14 We employ the latter strategy because we do not use loan level data with credit scores. More specifically, we define a borrower as subprime if it obtained its loan(s) from a lender on either the HUD or Inside Mortgage Finance lists, but the loan was not insured by FHA or VA. This group is called Subprime in Table 1 and in all figures. 15 Borrowers whose loans were guaranteed by FHA or VA (regardless of lender identity) are classified separately as Government owners. These loans often are considered of subprime quality because of the very high loan-to-value ratios usually involved, but we treat them separately from the private subprime group. As shown above, the time series on their shares in our panel almost are the mirror images of one another. A fourth category of owners were financed by individuals and households, or firms that issued less than 100 loans throughout our sample period. We label them as Small owners because they obtained their debt from small entities of one type or another. They also always constitute a small share of our sample, never amounting to more than 2%-3% of all observations in any one year. They are treated separately from the final category of owners financed with prime mortgages to help ensure that we do not conflate subprime and prime borrowers. 16 Our reasoning is that those owners who obtain financing from individuals or other entities that clearly are not traditional banks and financial institutions could be riskier, and thus more subprime-like. Unconditionally, they do have a modestly higher probability of losing a home to foreclosure or via short sale, as shown below in Table Inside Mortgage Finance, which previously was called B&C Mortgage Finance, claims to capture up to 85% of all subprime originations in most years. See Chomsisengphet & Pennington-Cross (2006) for more detail on this particular list. 15 The entities on the subprime lender list generally distinguish among the several units of a lender. For the HUD list in particular, identification was based on HMDA identification number of the entity, and different subsidiaries of a large bank typically had different ID numbers. Thus, having a subsidiary of (say) Bank of America that HUD believes specializes in subprime lending on the list does not mean that all of Bank of America s mortgage issuance gets classified as subprime. Banks and subsidiaries also enter and leave the HUD list over time. The HUD list also ends in The Inside Mortgage Finance also lists specific units, but we also consider those units as subprime if they ever show up on that publication s list. 16 Some of these small lenders could also have individual names because of measurement error in the way Dataquick assigns the names of lenders in the data. 11

12 All remaining owners with debt are Prime borrowers by definition. Their share always exceeds 50%, and it rises as the boom built, from a low of 56.8% in 2000 to a high of 66.2% in Thus, the rise in Subprime share over the same period is at the expense of the Government sector, not the Prime sector (Figure 1). C. Distress: Losing One s Home Via Foreclosure or Short Sale We define distress as the home being lost to foreclosure or short sale. As noted above, foreclosed homes are explicitly identified in the DataQuick files by a distress code that indicates the exact date when the home was lost by the previous owner. We are able to confirm this by looking at the name of the new owner, which typically is some type of financial institution, not a household. In the empirical work below, we also include short sales as exemplifying distress because the owner also left the home. Dataquick has a variable that indicates when a sale is considered a short-sale, but our conversations with the company revealed that such information is based on a proprietary model. Given that, we imputed that a short sale occurred if the sales price is no more than 90% of the outstanding balance on all existing debt. Our proxy for shortsales matches the Dataquick indicator x% of the time. We use our version of short-sales because the Dataquick variable is only populated since D. Constant Quality House Prices Constant quality nominal house price series are used throughout our analysis. We use hedonic price indexes, rather than repeat sales indexes popularized by Case and Shiller (1987), because of their much less onerous data requirements. 17 This becomes relevant because we create semi-annual price indexes for groups of census tracts. 18 These data are intended to proxy for neighborhoods within a metropolitan area. There is significant variation in price growth over time across tract groups, and we use that heterogeneity in creating loan-to-value ratios in our subsample of homes for which we have panel data. We do not include observations for which the reported sales price is less than $10,000 or greater than $5 million. Nominal price in logarithmic form is modeled as a function of the 17 It is worth noting that, at the metropolitan area level, the correlation between our hedonic price indexes and repeat sales indexes typically is extremely higher than Because there are few home sales within a given tract in any period, we aggregate tracts into groups of 4-6 (average is 4.5 tracts per group). The grouping is done to make the tracts as contiguous as possible. The variation in numbers is driven by this, as well as the need to have a sufficient number of transactions in each six month period. 12

13 square footage of the home entered in quadratic form, the number of bedrooms, the number of bathrooms, and the age of the home. We also include a dummy for condominiums or houses located in subdivisions and interact these dummies with the linear and quadratic terms for square footage. The hedonic index values are derived from the coefficients on the semi-annual dummies included in the model the actual equation and additional details are available upon request. The estimated indexes are then normalized to 100 in 2000(S1) for all neighborhoods. Figure 4 reports the graphs of the neighborhood-level semi-annual hedonic price series for four metropolitan areas. Not surprisingly, tract groups in a given metropolitan area tend to move together over time. However, there is considerable variation both on the upside and downside of the cycle. For example, the mean nominal price appreciation from 2000(1) to the cyclical peak across the 154 tract groups for which we estimated neighborhood-level hedonic price indexes in the Boston metropolitan area was 87.1%, with a standard deviation of 24.3 percentage points. The lowest neighborhood-level price growth to the peak was 48.6%, compared to 167.2% for the highest. There is less spatial heterogeneity in the bust, where the mean price decline from the peak to trough was 69.1%, with the spread from lowest to highest decline only 20 percentage points (from -61.1% to -82.6%). Naturally, this means that some submarkets in the Boston metro performed materially better than others since 2000(1). The mean price growth since 2000(1) was 45.8%, with the full range across neighborhoods running from 1.5% to 92.4%. This type of spatial heterogeneity is typical across tract groups within a given metropolitan area (larger ones in particular), and we exploit this variation in local price changes to compute loan-to-value ratios as discussed next. E. Leverage at Purchase and Over Time Loan and purchase price data are combined to compute leverage. Doing so at purchase is straightforward: divide the sum of all mortgages taken out at purchase by the arms-length purchase price recorded by DataQuick. Figure 5 shows how initial LTV varies over time by different types of home purchase financing. Governmental loans have much higher initial LTVs (close to 1) than both prime and subprime lending during the whole time period. Even though subprime initial LTVs increased from around 80% to 85% in the mid-2000s, this measure of leverage is somewhat stable over the housing cycle. 13

14 For current LTV, both numerator and denominator must be imputed. Two features of our data allow for a more accurate estimation of current LTVs: the complete history of home financings, including refinancings and second loans, and our own neighborhood house price indexes. For the numerator, all debt is observed at initial purchase and assumed to be fully amortizing, 30-year, fixed rate product. This method almost certainly understates true LTV, particularly on the subprime product which often involved adjustable rate mortgages (ARMs) and terms that did not require immediate amortization of principal. Then we additionally adjust the mortgage balance with information about refinances and second loans. The denominator, current house value, is estimated in a similar manner. Start with house price at purchase, and then update it on a half-year basis using the local price index described in the previous subsection. Prices may be overestimated for distressed properties because such houses may have been receiving less investment in maintenance, and this is another reason for our current LTV being under estimated. However, it turns out that the variation provided by refinances, second loans and the local price index are likely to overshadow any small measurement error due to this factor. Figure 6 shows how current LTV varies over time by type of borrower. Average current LTVs steadily declined from 1987 until Initial LTVs were mostly flat in this period, so the decline is associated with the house price increases observed in Figure 4. But that rapidly changed after 2006, when house prices fell dramatically, and current LTVs increased to unprecedented levels. By 2009 subprime and government borrowers had average current LTVs above 1.2, and prime borrowers had average LTV above 1. F. Identifying Speculators Nonacademic commentators in particular have argued that speculators played an important role in the building of the last housing boom, helping make its ultimate demise worse. We identify speculators in one of two ways. First, we follow Chinco and Mayer (2012) who reasoned that since speculators would not be living in the purchased unit, they would have their tax bills sent to another address. We compare the precise street address of the housing unit with the address to which the tax bill is sent the Tax Address in the DataQuick files. Whenever 14

15 the two are appreciably different, we call that purchaser a speculator. 19 The second way we identify whether a purchaser is a speculator is by whether the buyer has a name that is a business. This includes corporate or commercial names that include LLC or INC in them, homebuilders, or trusts (especially mortgage-backed securities trusts which are typically identified by a four-digit number in their names). 20 Figure 7 shows that the share of speculators by type of borrower increases for all categories until 2002, but then remains stable for Prime, Subprime, and Governmental loans, while it keeps escalating for Cash and Small borrowers. G. Demographics and Income of Borrowers We have described the many strengths of the DataQuick files, but one weakness is that they do not contain any information on the owners beyond their names. To gain more insight into borrower demographic characteristics (race and gender of the head of the household) and the self-reported income levels, we match individual sales transactions to loan application data in the Home Mortgage Disclosure Act (HMDA) files. Observations are merged in steps using a straight-forward matching process. In the first step, each transaction was matched to a loan using the year in which the transaction occurred, the full 11 digit Census tract number, the lender name, and the exact loan amount. In cases where there were multiple matches one of them was randomly assigned as being a true match while the rest were considered unmatched. The remaining unmatched observations were then merged based only on year, Census tract and exact loan amount with multiple matches being randomly assigned as in the first step. This two-step process was repeated several times allowing for the loan amounts to differ from each other in increments of $1,000 up to a total allowable difference of $10,000. Any observations remaining after this process then went through an identical matching procedure using 9 digit Census tract numbers. Observations surviving that procedure are considered to be unmatched. In total 92.7% of the sales transactions in DataQuick were matched at some point in the procedure. Of those, approximately 60% were matched in the first step. Because we are unsure about the quality of the matches in the other steps, in the empirical work below we always distinguish the demographics in two groups perfect and imperfect matches and include both in the 19 By appreciably different, we generally mean that more than one number in the street address before the zip code differs. 20 Other academic research has identified speculators by whether the flip properties quickly (e.g., Bayer, Geissler, and Roberts (2011)). We also investigated those cases, but more than 99% of them were already encompassed by our measures of tax address and names of business. 15

16 estimation. Finally, the demographic data for Cash buyers will be missing because they never took out a loan, and hence, cannot be matched with any HMDA observation. H. Summary Statistics Table 1 reports summary statistics for a number of variables of interest in our empirical work for the full sample and also for each of the five type of financing categories, with their shares in our overall sample in parentheses: Prime (61%), Subprime (15%), Government (10%), Cash (11%), and Small (2%). The top panel notes distress rates. Foreclosures always are at least twice as prevalent as short sales in leading to the loss of a home. And, the unconditional probability of a Subprime owner losing its home is well more than double the rates a Prime owner does. The sample-wide mean is 0.73% for Subprime owners (1-in-137) versus only 0.33% (1-in-303) for Prime owners, but Figure 2 above shows that masks substantial heterogeneity over time. The group of owners that borrowed from small lenders has a distress rate of 0.42%, above that for Prime owners, but well below that for Subprime owners. The same is true of borrowers using FHA/VA-insured mortgages, who have a distress rate of 0.38%. Not surprisingly, all cash owners have the lowest rate of distressed home loss at 0.14% or 1-in-714. Panel 2 reports data on housing traits. There is no evidence here that Subprime owners purchased systematically smaller or appreciably older units. Subprime sector homes are only slightly older than Prime owner units on average. The size of Government borrowers homes are the smallest by about 1,000 square feet compared to Subprime borrowers. Demographics are reported in Panel 2. There are only modest differences in the fraction of male borrowers (defined as the head of household) across the four borrower types. Racial differences are bigger. The White share of Subprime owners is 10 percentage points below that of Prime owners, at 62.8% versus 72.8%. The Government category has a similarly low White share. Reported income on the loan application recorded in the HMDA files shows Subprime owners to be less rich than Prime owners, but the difference is only 6%. It is borrowers using FHA/VA-insured loans who have appreciably lower incomes on average: $68,792 versus $117,537 for Subprime, $124,930 for Prime, and $141,700 for Small sector owners. Panel 2 also reports a substantial number of speculators, with 22.6% of all ownership sequences being speculators. The vast majority are identified by the housing unit and tax addresses being different. There is not an appreciable difference between the share of Subprime 16

17 and Prime owners who we classify as speculators. Speculators are most rare among the Government category, at 13.8%, and more prevalent among the all-cash group at 41%. Figure 5 depicts the share over time of each of the five owner categories that we consider to be speculators. The high average share of the Cash owners is due to a sharp increase as the housing bust deepened. Note also that there is no increase in the shares of Subprime or Prime owners we classify as speculators after The fourth panel reports summary statistics on house prices and LTV at the time of purchase. Transaction prices are higher and similar for both prime and subprime borrowers, and smaller for owners who took governmental loans. However, the data indicate that subprime owners are substantially more likely to use high leverage than prime owners at purchase. Less than 40% of all our ownership sequences start off with initial LTVs<80%, but subprime owners have the lowest probability of being lightly leveraged, at 25.2%. Another quarter of our ownership sequences have between 10%-20% equity at purchase. Just over 30% have leverage between 90% and 100% of their purchase price. Among Subprime owners, 30.7% start out their ownership sequence with at least a 95% LTV. This is ten percentage points above the analogous figure of 20.6% for Prime owners. By no means are Subprime owners the most leveraged owners in our sample. Borrowers of FHA/VA loans tend to use much more debt at purchase. Nearly two-thirds (63.8%) have 95%-100% LTVs. The fifth panel of Table 1 breaks down the financings by whether they are for a purchase or any subsequent financing. For the overall sample, 52% of all ownership sequences never altered the debt they took on at purchase (see column 1). Another one-third (33.2%) refinanced and nearly 15% took out a second mortgage. Across owner types, the Subprime group was the least likely to not refinance or take on a second mortgage. This breakdown highlights the benefit of having a panel structure, as nearly half of all ownership sequences involve some type of debt adjustment beyond simply amortizing the loan. Finally, the last panel shows averages for current LTV. Twenty-two percent of our observations had negative equity, i.e., current LTVs greater than 1. And these percentages are much higher for subprime and governmental borrowers, at 30% and 34%, respectively. III. Empirical Results A. Empirical Strategy 17

18 We estimate linear probability models of whether an owner lost a home via foreclosure or short sale. Our panel data of the complete housing and mortgage market is indexed by housing ownership sequence h and year-quarter t. The dependent variable is a binary distress outcome y ht which equals one if the owner lost the home via foreclosure or short sale that quarter and equals zero otherwise. This is first modeled as a function of the five owner/financing types (O k ): Subprime, Prime, Government, Small and Cash. In all reported regression results, Cash is the omitted category. A given owner is classified as one (and only one) of these types each quarter, with k indexing the five ownership types. 21 We next include neighborhood fixed effects (S tn ) in order to compare outcomes across Subprime and Prime owners within tracts, and interact with year/quarter in order to capture local economic conditions at time t. This is followed by adding a D ht that contains indicators for ranges of current LTV on the property. Due to the very large number of observations, we estimate simple linear probability models as in equation (1): (1) y ht = α k O htk + S tn + ρd ht + μ ht It still could be that differences between households with different types of loans within tracts remain. We deal with this issue by including available household features (e.g., race, initial selfreported income, whether the owner is a speculator, and house characteristics) in some specifications. We use the same setup to take a deeper look at how different aspects of borrower leverage beyond the current LTV impact the probabilities of foreclosure. These variables include the initial LTV, whether the homeowner refinance the mortgage or took out additional loans, and the timing of house purchase or refinance. As discussed above, we are like the rest of the literature in not being able to directly measure some other likely determinants of foreclosures or short sales such as owner wealth and employability. However, these variables have components that are invariant over time, and the panel data allows us to flexibly control for them by including a set of ownership sequence fixed effects h : 21 Classification is straightforward for any quarter in which an owner has no more than one mortgage. If there are multiple loans and at least two have different classifications, we use the following order to determine ownewr type: Subprime, Government, Prime, Small. Thus, if the owner has at least one Subprime loan and any number of other types of loans, it is classified as Subprime. If it does not have any subprime loans, but has at least one Government loan and any other loans, it is classified as Government, and so forth. 18

19 (2) y ht = α k O htk + I h + t + ρd ht + μ ht. We also include a set of year-quarter effects t. In this specification, identification of the parameters α is based on ownership sequences in which the owner switched from one type of financing (including all cash) to another during the sequence. Below we show that a large fraction of homeowners switched financing type during the housing boom, and document how representative they are. Even though we do not observe details of the mortgage contract, such as whether the interest rate is adjusted or fixed or the set of documents required by the lender, the household fixed effects model allow us to compare the behavior of households that switched the complete package of mortgage terms from prime to subprime and vice-versa. B. Panel Structure Estimates: Negative Equity and Illiquidity Table 2 contains our first set of results from the models described above. The first column presents unconditional estimates of distress for Subprime, Prime, Government, and Small owners in the absence of any other controls. Cash owners always are the omitted category, so that the estimated coefficients must be interpreted as relative to the 0.14% home loss rate for that group of owners (see the means in the top panel of Table 1). Because we are particularly interested in the extent to which observables can account for differences in the degree of distress among Subprime owners versus Prime (and other) owners, the bottom panel of the table reports the estimated Subprime Gaps in home loss. Unconditionally, the Subprime gap with Prime distress is (or 0.40%), which is economically large when compared to the unconditional mean probability of distress among Prime owners of (0.33%). The Subprime gap with Government owners is very similar, and that with respect to Small owners also is large economically. The second column includes census tract-by-quarter fixed effects. Their inclusion means that the comparison between Subprime and Prime owner distress rates now is being made for ownership sequences in the same neighborhood that faced similar average local economic shocks. The Subprime Prime Gap is only modestly reduced by about 8%, from to This suggests that both Subprime and Prime owners are dispersed across census tracts within a MSA rather than spatially concentrated within a few. The first column of Appendix 19

20 Table 3 supports this assertion with its listing of the average share of tracts within a MSA that never had a Prime owner (top panel) or a Subprime owner (bottom panel) within its boundaries at any point in our sample period. On average, virtually no tracts never had a Prime owner in them, while only 3.5% of tracts have never had a Subprime owner. The remaining three columns of that table report the share of tracts containing 25%, 50% and 75% of each owner type, respectively. Neither Prime nor Subprime owners are randomly dispersed geographically, but both are pretty widespread. For example, 45.3% of tracts contain 75% of Prime owners on average; the analogous figure for Subprime owners is 46.6%. 22 In sum, we can rule out a dense spatial concentration of subprime loans in select neighborhoods that experienced some type of strong negative economic shock as explaining why subprime distress rates were so high. Column 3 shows that current LTV is much more influential. It reduces the Subprime- Prime distress gap by an additional one-third, drives down the Subprime Government gap by another quarter, and sharply shrinks the Subprime Small gap by two-thirds. The economic importance of current LTV in explaining the rate of home loss is also apparent in the Subprime coefficient itself, which implies that the Subprime-Cash gap is reduced by three-quarters. Note also that controlling for current leverage results in Prime and Government owners having slightly lower distress propensities than owners who purchased with all cash. Figure 8 s plot of current LTV coefficients this particular specification as well as for other panel models discussed below indicates that a large fraction of the distress in residential real estate markets, regardless of the type of mortgage finance, was concentrated among borrowers living in homes that were underwater. Starting with the model that includes neighborhood fixed effects, we find that current LTV coefficients tend to be very close to zero if the current LTV is below 1. The slope steepens once that threshold is crossed, and reaches 2%- 3% for very high LTVs. The intuition behind this is straightforward: underwater homeowners generally cannot sell a house without a net monetary loss. Since most homeowners do not have savings available to cover big losses, they ended up locked-in their own homes (Ferreira, Gyourko, and Tracy 2010, 2012) or have to surrender the house in a foreclosure procedure when making mortgage payments is not feasible or desirable. Note that there is relatively little difference between estimates across different models: within neighborhood, within ownership 22 This general pattern of geographic dispersion of loan types does not vary much over time. Those data are available upon request. 20

21 sequence, controlling for household features, or for other components of leverage. This strongly suggests that our negative equity variable is well measured and its effects on foreclosures are unlikely due to omitted factors. Columns 4 and 5 of Table 2 then present two versions of the panel data model that include household fixed effects. Household fixed effects are added to our baseline model along with year-by-quarter fixed effects in column 4. Tract-by-quarter fixed effects are not included here due to computational constraints. The Subprime-Prime gap is reduced by nearly three quarters in this specification relative to our baseline ( / ~ 0.28), and the Subprime-Cash gap is only 16% of the gap that only controls for neighborhood. Column 5 additionally includes current LTV, reducing even more the Subprime-Prime gap in distress rates, which becomes 19% of the within neighborhood gap. Moreover, the positive differences between Subprime and Prime with respect to Cash disappear. In sum, controlling for the permanent unobserved features of homeowners produces the largest reduction in the Subprime Prime foreclosure rates gap. Table 3 provides information on Switchers, which are households that refinanced their original mortgage at purchase or took out another mortgage later in their ownership sequences. Nearly one-third of all ownership sequences involve some type of adjustment to the original debt used to purchase the home. There is wide variation in the share of Switchers across owner types. The share of Subprime owners who switch (62.0) is well more than double the analogous share of Prime owners (27.0), and it exceeds the propensity of the other three categories of owners. The remainder of this panel provides details on the frequencies of the myriad ways that switching can occur. Figure 9 shows how the total number of switchers varies over time. The largest wave of switching happened between 2002 and 2006, a period of declining interest rates and increasing house prices (Bhutta and Keys, 2014). Both provide the necessary economic incentive for refinancing activity to blossom, and as a consequence, we observe this enormous switching behavior. However, one could also envision two-sided selection leading to higher distress rates in the sample of Subprime switchers. That is, households with declining credit quality could be switching from prime to subprime mortgages, while borrowers with improving credit prospects could be switching from subprime to prime mortgages. But our empirical estimates do not find 21

22 that subprime borrowers are more likely to lose a home after controlling for household fixed effects. The results do not necessarily mean that Switchers constitute a random sample. Appendix Table 4 provides some descriptives of switchers versus non-switchers. While the Switchers are not a random sample, they have similar homes in physical terms (i.e., same units sizes, number of bedrooms, etc.) to the rest of the population, and they only slightly differ along other lines: Switchers have a 3.8 percentage point higher shares of Blacks and Hispanics, and they have 7 percent smaller median self-reported income. Moreover, the switchers are more similar to subprime owners in many dimensions, and therefore our estimates are unlikely to be driven by selection into switching (as opposed to always staying a subprime owner). One weakness of the household effects specification is that it does not distinguish among different fixed factors. But we can test if some observed fixed features of homeowners matter such as household and housing traits - and we proceed with this exercise in the next section. C. Panel Structure Estimates: Other Potential Factors Table 4 reports estimates for specifications where other factors, such as housing traits and household traits, are included as potential common factors that can explain the subprime-prime propensity to foreclose gap. First, we look at housing unit traits, such as house size, number of bathrooms and bedrooms. Much was written during the housing boom about homeowners buying bigger and bigger homes, and keeping up with the Joneses potentially without the proper means to make the mortgage payments. Column 1 reports results for the specification where only the house traits are added to the specification with neighborhood-by-quarter effects. Note that there is virtually no change in the Subprime Prime Gap (or any other gap). Thus, there is no evidence that differences in house size can explain any of the difference in distress across the different types of owners. Next we look at household traits. They could impact foreclosures in a number of ways. Race, for example, could be important since minorities have a larger share of subprime mortgages relative to prime, and usually have less wealth than non-minorities (Bayer, Ferreira, Ross, forthcoming). Speculators could react faster to the first sign of negative equity and stop making monthly payments early in order to avoid future bigger losses. Low self-reported income could indicate a lower likelihood to sustain mortgage payments in the future. While 22

23 these are all plausible mechanisms, we find that household traits only minimally shrink the Subprime-Prime gap and do not change the Subprime coefficient itself (column 2). Column 3 s results then show that including housing and household features does not materially change the impact of current LTV on the probabilities of foreclosure. The next three columns investigate whether other measures of leverage that influence current LTV can explain the foreclosure probabilities. The first component is initial LTV, which mechanically corresponds to the first current LTV observation in any household sequence. Initial LTV was shown to explain about 60% of the foreclosure crisis according to simulations based on a macro model of housing markets developed in Corbae and Quintin (2015). The results reported in column 4 find an impact for initial LTV, but it only explains 14% of the Subprime-Prime gap. While this is much larger than housing and household traits, it still is much smaller than the effect of current LTV (which explains 34% of that gap). Column 5 shows if taking a refi or second loan (not at purchase) influences the Subprime Prime gap. Refi and second loans directly contribute to the variation in current LTV by providing discrete changes in the mortgage balance at certain periods of an ownership sequence. This mechanism could be potentially important not only because of the large number of refi s and second loans observed in our data, but also because Mian and Sufi (2011) found that homeequity based borrowing may explain one-third of mortgage defaults between 2006 and 2008 in the United States. We find that these two variables play a negligible role in explaining the Subprime - Prime gap, presumably because both types of borrowers refinanced at very high rates during the boom. We also test whether cohort dummies based on the quarter of the last purchase or mortgage transaction within an ownership sequence impact the probability of foreclosure. These cohorts affect current LTV because houses bought or refinanced near the peak of the housing boom had the largest declines in prices after the beginning of the recession, and therefore suffered the largest increases in current LTV. As shown in Figure 4, these cohort effects may be quite relevant in MSAs where all neighborhood prices moved in sync with the rest of the market during the housing boom. Also, Bayer, Ferreira, and Ross (forthcoming) and Palmer (2013) show that cohort effects may explain some of the movements in defaults and prices respectively. Estimates including the cohort dummies are shown in Column 6: they explain 19% of the Subprime - Prime gap, which is a significant but still smaller part of the effect from current LTV. 23

24 Finally, we include the three components of current LTV in column 7. Together they explain 31% of the subprime-prime gap (as opposed to 35% for current LTV only). Column 8 shows estimates of all individual components and current LTV in the same model. These results reveal that the three subcomponents add some extra explanatory power, as the Subprime - Prime gap is reduced by 39%. D. Panel Estimates: Heterogeneity (TBD) E. Panel Estimates: Robustness Tests (TBD) IV. Conclusion The housing bust and its consequences are among the defining events of the past quarter century. Much remains unknown about their very nature because the crisis had such a long duration, and most popular and academic commentary and analysis viewed it through a subprime lens. Incomplete interpretations and outright mistakes are more likely if one only looks at the start of the crisis and/or focuses on one type of financing. We brought a new and very large micro data set spanning the cycle and all sectors of the mortgage market to the investigation. The panel structure facilitated the estimation of models that permit us to tell which of a host of different factors can explain the large unconditional differences in the rate of home loss across subprime versus prime (and other) owners. Our key findings are readily summarized as follows. The crisis was much less of uniquely subprime event than is recognized by scholars and policy makers. Subprime distress did spike first as previous research has showed. And, it occurred at relatively and absolutely high rates. However, distress and home loss among prime borrowers occurred with a lag and became quite extensive itself. Negative equity and owner illiquidity explain much of this variation, as predicted by the traditional mortgage default literature. Other factors, such as race or whether the owner is a speculator, do not have much independent explanatory power. The potential implications for public policy are especially difficult to identify in the absence of complete data over the cycle. We find that our measures of negative equity and borrower illiquidity indicate a bigger role for the economic cycle. But links to the economic cycle leave policy makers with a very challenging problem. For example, Ferreira and Gyourko 24

25 (2010) find that local booms started as fundamentally based due to real income shocks. Lenders reacted to those shocks by increasing the availability of credit. It is difficult to effectively control or regulate that scenario. It probably is not wise to restrict lending per se, as that may have perverse consequences (e.g., it could make affordability would be even harder for lower and middle income households to handle). Moreover, even 80% initial LTV loans had a lot of defaults, in part because prices can fall so quickly and steeply in a recession that you suffer from negative equity. If you suddenly don t have a job to pay the bills, there is nothing to be done if you have not amassed savings ahead of time. Regulation of details of mortgage contracts may not help much either. Prior literature has shown very high defaults among borrowers with more exotic subprime loans, for example. We do not have loan contract data, but it turns out that we do not need it to account for differences in the propensity to lose one s homes across subprime and prime borrowers. The question of whether homeowners should have been bailed out is a fraught one because it involves political and philosophical positions, not just economic data. However, bailout of homeowners was not done in the last cycle at least in part because many people appear to have thought that the crisis was a subprime one. Home owner distress became much more widespread over time, geographically and by owner type. And, our results suggest that the high rates of home loss among subprime borrowers can be accounted for by two standard factors suggested by the traditional mortgage default literature negative equity and borrower illiquidity. These affect all borrowers, not just subprime ones. We cannot know if this knowledge would have changed the public s view on the appropriateness of home owner bailouts in the crisis, but we do know that distress in the housing market had large negative macroeconomic effects (e.g., Mian & Sufi, 2014). A number of important issues remain for future research. Foremost among them is the need to directly measure employment status and wealth at the household level. Better linking individual conditions in the labor and housing markets should be the first order of business if we are to improve our understanding of what explains owner distress in the great housing bust. 25

26 Selected References Adelino, Manuel, Kristopher Gerardi, and Paul Willen. Renegotiating Home Mortgages: Evidence From the Subprime Crisis, Federal Reserve Bank of Boston Working Paper, March 31, Why Don t Lenders Renegotiate More Home Mortgages? Redefaults, Self-Cures, and Securitization, Journal of Monetary Economics, Vol. 60, no. 7 (2013): Agarwal, Sumit, Gene Amromin, Itzhak Ben-David, Souphala Chomsisengphet, and Douglas Evanoff. The Role of Securitization in Mortgage Renegotiation, Journal of Financial Economics, Vol. 102, no. 3 (2011): Predatory Lending and the Subprime Crisis, Journal of Financial Economics, Vol. 113, no. 1 (2014): Aschberg, Marius, Robert Jarrow, Holger Kraft, and Yildiray Yildirim, Government Policies, Residential Mortgage Defaults and the Boom and Bust Cycle of Housing Prices, Real Estate Economics, forthcoming (2013 web version available). Bayer, Patrick, Fernando Ferreira, and Stephen Ross, The Vulnerability of Minority Homeowners in the Housing Boom and Bust, American Economic Journal: Economic Policy, forthcoming. Ben-David, Itzhak. Financial Constraints and Inflated Home Prices During the Real Estate Boom, American Economic Journal: Applied Economics, Vol. 3, no. 3 (2011): Bhardwaj, Geetesh and Rajdeep Sengupta. Subprime Cohorts and Loan Performance, Journal of Banking & Finance, Vol. 41 (2014): Biswas, Arnab. Housing Submarkets and the Impacts of Foreclosures on Property Prices, Journal of Housing Economics, Vol. 21, no. 3 (2012): Bubb, Ryan and Alex Kaufman. Securitization and Moral Hazard: Evidence from Credit Score Cutoff Rules, Journal of Monetary Economics, Vol. 63 (2014): Campbell, John, Stefano Giglio, and Parag Pathak. Forced Sales and House Prices, American Economic Review, Vol. 101, no. 5 (2011): Capozza, Dennis and Robert Van Order. The Great Surge in Mortgage Defaults : The Comparative Roles of Economic Conditions, Underwriting, and Moral Hazard, Journal of Housing Economics, Vol. 20, no. 2 (2011):

27 Chan, Sewin, Michael Gedal, Vicki Been, and Andrew Haughwout. The Role of Neighborhood Characteristics in Mortgage Default Risk: Evidence from New York City, Journal of Housing Economics, Vol. 22, no. 2 (2013): Cheung, Ron, Chris Cunningham, and Rachel Meltzer. Do Homeowners Associations Mitigate or Aggravate Negative Spillovers from Neighboring Homeowner Associations?, Journal of Housing Economics, Vol. 24, no. 1 (2014): Chomsisengphet, Souphala and Anthony Pennington-Cross. Evolution of the Subprime Mortgage Market, Federal Reserve Bank of St. Louis Review, Vol. 88, no. 1 (2006): Coleman IV, Major, Michael LaCour-Little, and Kerry Vandell. Subprime Lending and the Housing Bubble: Tail Wags Dog?, Journal of Housing Economics, Vol. 17, no. 4 (2008): Corbae, Dean, and Erwan Quintin, Leverage and the Foreclosure Crisis, Journal of Political Economy, Vol.. 123, n.1 (2015): Demyanyk, Yuliya and Otto Van Hemert. Understanding the Subprime Mortgage Crisis, Review of Financial Studies, Vol. 24, no. 6 (2011): Deng, Yongheng, John Quigley, and Robert Van Order. Mortgage Terminations, Heterogeneity, and the Exercise of Mortgage Options, Econometrica, Vol. 68, no. 2 (2000): Elul, Ronel, Nicholas Souleles, Souphala Chomsisengphet, Dennis Glennon, and Robert Hunt. What Triggers Mortgage Default?, American Economic Review, Vol. 100, no. 2 (2010): Foote, Christopher, Kristopher Gerardi, Lorenz Goette, and Paul Willen. Just the Facts: An Initial Analysis of Subprime s Role in the Housing Crisis, Journal of Housing Economics, Vol. 17, no. 4 (2008): Reducing Foreclosures: No Easy Answers, NBER Macroeconomics Annual University of Chicago Press, April Foote, Christopher, Kristopher Gerardi, and Paul Willen. Negative Equity and Foreclosure: Theory and Evidence, Journal of Urban Economics, Vol. 64, no. 2 (2008): Ferreira, Fernando, Joseph Gyourko, and Joseph Tracy, Housing Busts and Household Mobility, Journal of Urban Economics, Vol. 68, n.1 (2010): , Housing Busts and Household Mobility: an Update, Economic Policy Review, Vol. 18, n.3 (2012). 27

28 Foster, Chester and Robert Van Order. An Option-based Model of Mortgage Default, Housing Finance Review, Vol. 3, no. 4 (1984): Goodman, Allen and Brent Smith. Residential Mortgage Default: Theory Works and So Does Policy, Journal of Housing Economics, Vol. 19, no. 4 (2010): Guiso, Luigi, Paola Sapienza, and Luigi Zingales. The Determinants of Attitudes toward Strategic Default on Mortgages, The Journal of Finance, Vol. 68, no. 4 (2013): Harding, John, Eric Rosenblatt, and Vincent Yao. The Contagion Effect of Foreclosed Properties, Journal of Urban Economics, Vol. 66, no. 3 (2009): Haughwout, Andrew, Richard Peach, and Joseph Tracy. Juvenile Delinquent Mortgages: Bad Credit or Bad Economy?, Journal of Urban Economics, Vol. 64, no. 2 (2008): Jiang, Wei, Ashlyn Aiko Nelson, and Edward Vytlacil. Liar s Loan? Effects of Origination Channel and Information Falsification on Mortgage Delinquency, The Review of Economics and Statistics, Vol. 96, no. 1 (2014): Kau, James, Donald Keenan, and Taewon Kim. Default Probabilities for Mortgages, Journal of Urban Economics, Vol. 35, no. 3 (1994): Kau, James, Donald Keenan, Constantine Lyubimov, and V. Carlos Slawson. Subprime Mortgage Default, Journal of Urban Economics, Vol. 70, no. 2 (2011): Keys, Benjamin, Tanmoy Mukherjee, Amit Seru and Vikrant Vig. Financial Regulation and Securitization: Evidence from Subprime Loans, Journal of Monetary Economics, Vol. 56, no. 5 (2009): , Did Securitization Lead to Tax Screening? Evidence From Subprime Loans, The Quarterly Journal of Economics, Vol. 125, no. 1 (2010): Keys, Benjamin, Amir Sufi, and Vikram Vig. Lender Screening and the Role of Securitization: Evidence from Prime and Subprime Mortgage Markets, Review of Financial Studies, Vol. 25, no. 7 (2012): LaCour-Little, Michael, Charles Calhoun, and Wei Yu. What Role Did Piggyback Lending Play in the Housing Bubble and Mortgage Collaspe, Journal of Housing Economics, Vol. 20, no. 2 (2011): Mayer, Christopher and Karen Pence. Subprime Mortgages: What, Where, and to Whom?, NBER Working Paper (2008). Mayer, Christopher, Karen Pence, and Shane Sherlund. The Rise in Mortgage Defaults, 28

29 Journal of Economic Perspectives, Vol. 23, no. 1 (2009): Mian, Atif and Amir Sufi. The Consequences of Mortgage Credit Expansion: Evidence from the U.S. Mortgage Default Crisis, Quarterly Journal of Economics, Vol. 124, no. 4 (2009): House Prices, Home Equity-Based Borrowing, and the US Household Leverage Crisis, American Economic Review, Vol. 101, no. 5 (2011): House of Debt. The University of Chicago Press: Chicago and London, Moulton, Shawn. Did Affordable Housing Mandates Cause the Subprime Mortgage Crisis?, Journal of Housing Economics, Vol. 24, no. 1 (2014): Nadauld, Taylor and Shane Sherlund. The Impact of Securitization on the Expansion of Subprime Credit, Journal of Financial Economics, Vol. 107, no. 2 (2013): Palmer, Christopher. Why Did So Many Subprime Borrowers Default During the Crisis: Loose Credit or Plummeting Prices?, Job Market Paper (2013). Piskorski, Tomasz, Amit Seru, and Vikrant Vig. Securitization and Distressed Loan Renegotiation: Evidence from the Subprime Mortgage Crisis, Journal of Financial Economics, Vol. 97, no. 3 (2010): Rose, Morgan. Geographic Variation in Subprime Loan Features, Foreclosures, and Prepayments, The Review of Economics and Statistics, Vol. 95, no. 2 (2013): Schuetz, Jenny, Vicki Been, and Ingrid Gould Ellen. Neighborhood Effects of Concentrated Mortgage Foreclosures, Journal of Housing Economics, Vol. 17, no. 4 (2008): Smith, Brent. Stability in Consumer Credit Scores: Level and Direction of FICO Score Drift as a Precursor to Mortgage Default and Prepayment Journal of Housing Economics, Vol. 20, no. 4 (2011): Vandell, Kerry. How Ruthless is Mortgage Default? A Review and Synthesis of the Evidence, Journal of Housing Research, Vol. 6, no. 2 (1995): Whitaker, Stephan and Thomas Fitzpatrick IV. Deconstructing Distressed-Property Spillovers: The Effects of Vacant, Tax-Delinquent, and Foreclosed Properties in Housing Submarkets, Journal of Housing Economics, Vol. 22, no. 2 (2013):

30 Figure 1: Borrower/Mortgage Type Share Over Time Figure 2: Borrower/Mortgage Type Foreclosure Share Over Time 30

31 Figure 3: Unconditional Distress Probabilities (Foreclosures + Short Sales) Over Time Detroit-Warren-Livonia, MI Las Vegas-Paradise, NV price index price index h1 1997h1 2000h1 2003h1 2006h1 2009h1 time 1994h1 1997h1 2000h1 2003h1 2006h1 2009h1 time Orlando, FL San Francisco-Oakland-Fremont, CA price index price index h1 1997h1 2000h1 2003h1 2006h1 2009h1 time 1994h1 1997h1 2000h1 2003h1 2006h1 2009h1 time Figure 4. Hedonic Price Indexes by Neighborhood; Selected MSAs 31

32 Figure 5. Average Initial LTV by Borrower/Mortgage Type Over Time Figure 6. Average Current LTV by Borrower/Mortgage Type Over Time 32

33 Figure 7: Share Over Time of Owner Types Classified As Speculators Figure 8. Current LTV Estimated Coefficients 33

34 Figure 9: Share of New Switchers in Total Ownership Sequences Each Quarter 34

35 Table 1. Descriptives Notes: TBW 35

36 Table 2. Panel Data Estimates Notes: TBW 36