Subprime Loan Performance

Disclosure Regulation on Mortgage Securitization and Subprime Loan Performance Lantian Liang Harold H. Zhang Feng Zhao Xiaofei Zhao October 2, 2014 Abstract Regulation AB (Reg AB) enacted in 2006 mandates disclosure by originators of 20% or more of the pool assets on information regarding the size and composition of the originator s origination portfolio as well as information material to an analysis of the performance of the pool assets such as the originators credit-granting or underwriting criteria. Using data on the non-agency mortgage backed security (MBS) market, we find that after Reg AB a higher fraction of mortgage deals consists of loans from low-stake originators just below the disclosure threshold. Deals with these low-stake originators have significantly larger losses than those without and this is only so for deals issued after Reg AB. In particular, deals with originators who change from highstake to low-stake have larger losses. Analysis on loan level data provides further evidence that securitized loans show worse performance when their originators increase their participation of deals as a low-stake originator after Reg AB. Overall, our evidence suggests that mortgage securitizers circumvent the disclosure requirement in Reg AB and this has adverse impact on mortgage performance. Keywords: Regulation AB; Disclosure Threshold; MBS Performance Naveen Jindal School of Management, University of Texas at Dallas, 800 West Campbell Road, Richardson, Texas, 75080, email: lantian.liang@utdallas.edu, harold.zhang@utdallas.edu, feng.zhao@utdallas.edu, xiaofei.zhao@utdallas.edu

1. Introduction Regulation reform on disclosure in the private-label MBS market remains a virtually unexplored area in the aftermath of the 2007-2008 subprime mortgage crisis. To meet the insatiable demand from global investors reaching for higher yields, the entire supply chain of mortgage securitization increasingly expanded lending to riskier borrowers. Compelled by the rapid growth of loan securitization market and the lack of explicit regulations directed towards the distinguishing features of this market, Securities and Exchange Commission introduced Regulation AB (Reg AB). While Reg AB became effective in January 2006, no study has devoted to examining its effects on loan securitization market. Our investigation represents the first attempt to evaluate how Reg AB affected the structure of securitization and securitized loan performance. In this study, we focus on the effect of disclosure regulation under Reg AB on the residential mortgage securitization market. Existing studies attribute the financial crisis to the sharp increase in mortgage loan defaults. 1 Securitized residential mortgages accounted for a large fraction of new issuance of securitized loans in the period leading to the financial crisis. 2 We collect detailed deal and loan level data on securitized mortgages which facilitate an in-depth investigation and allows quantitative assessment on the regulation effects. Specifically, we examine the effect of disclosure threshold on loan originators under Reg AB, the structural change of the MBS issuance and the associated impact on the performance of securitized mortgages. To identify the effect on securitized mortgages, we resort to cross sectional variation in deal strucuture and loan originators with different reactions to the regulation. Specifically, we focus on its effect on loan originators because Reg AB sets different disclosure requirements 1 See, Mian and Sufi (2009), Nadauld and Sherlund (2009), Keys, Mukherjee, Seru, and Vig (2010), Keys, Seru, and Vig (2012), Purnanandam (2011), among others. 2 According to former International Monetary Fund chief economist Simon Johnson, the total volume of private mortgage-backed securities (excluding those issued by Ginnie Mae, Fannie Mae and Freddie Mac) grew from $11 billion in 1984 to over $200 billion in 1994 to close to $3 trillion in 2007. 1

on loan originators based on the percentage of loans included in a mortgage deal by each originator. Reg AB Item 1110 requires identification of any originator or group of affiliated originators that originated, or is expected to originate, 10% or more of the pool assets; and requires disclosure of information regarding the size and composition of the originator s origination portfolio as well as information material to an analysis of the performance of the pool assets, such as the originator s credit-granting or underwriting criteria for the asset types being securitized, if the originator originated or is expected to originate 20% or more of the pool assets. We collect information on loan originators from mortgage deal prospectuses of privately securitized residential mortgages between 2003 and 2007 from Bloomberg. For each mortgage deal, its prospectus provides information on the composition of loans originated by different originators. To link individual loans to a particular originator in a deal, we utilize First American Corelogic LoanPerformance database. Corelogic data provides the name of the original lender for each loan. We collect identity and affiliation information for the original lender of each loan to determine if the original lender is one of the mortgage deal originators or is affiliated with one of these deal originators. Using this information, we assign individual loans to the originators listed in the prospectus supplements, which we use to perform loan level analysis on the impact on loan performance of deal structure change due to Reg AB. Our investigation demonstrates several significant effects of the regulatory disclosure mandate on certain loan originators and the performance of securitized mortgages. First, we find that the number of financial institutions originated less than 20% of loans in each mortgage deal (referred to as low-stake originators hereafter) increased significantly post Reg AB. In particular, the percentage of deals having low-stake originators more than doubled after Reg AB relative to before Reg AB. This unintended consequence of Reg AB has a significant effect on performance of securitized mortgages. Deals with low-stake originators have significantly larger cumulative net losses than those without and this is only so for deals issued after Reg AB, but not before Reg AB. In particular, deals with originators who 2

change from high-stake to low-stake have larger cumulative net losses. Analysis on loan level data provides further evidence that securitized loans show worse performance when their originators increase their participation of deals as low-stake originators after Reg AB. This result also suggests that the strategic use of the regulatory threshold is unlikely due to the motive of avoiding SEC compliance costs, rather, it is more likely driven by the intention of avoiding scrutiny and potentially withholding some adverse information on riskier loans. Our paper contributes to two strands of research. First, it offers the first empirical investigation on the effect of changes in regulations on mortgage securitization on the practices of financial institutions participated in this market and their consequences on loan performance. Second, it sheds light on the effect of mandatory disclosure on financial institutions and the performance of securitized loans issued by these financial institutions. While early studies argue that firms disclose bad news to avoid lawsuits in the future (Skinner (1994)), more recent evidence suggests that firms disclosing more also have more frequent litigation (Skinner (1997)). Kothari, Shu, and Wysocki (2009) argue that firms withhold bad news up to certain threshold and provide evidence using different magnitude of stock market reaction to negative and positive news. The rest of the paper is organized as follows. Section 2 describes the most relevant item on originator information disclosure under Reg AB. Section 3 describes data and provides summary statistics. Section 4 presents and discusses our empirical findings at deal level. Section 5 provides loan level evidence on the impact of low-stake originators on loan performance. Finally, section 6 concludes. 2. Regulation AB Securities and Exchange Commission defined asset-backed securities (ABS) as securities that are backed by a discrete pool of self-liquidating financial assets. ABS market experienced a rapid growth in the last two decades. One source estimates that annual issuance of U.S. 3

public non-agency ABS grew from $46.8 billion in 1990 to $416 billion in 2003. 3 Another source estimates that new issuance for 2003 was at $800 billion. 4 Prior to the introduction of Reg AB, there have been few SEC initiatives directly related to ABS. In this section, we describe the most relevant item of Reg AB on originator information disclosure for mortgage securitization and discuss the implications for mortgage deal structure and securitized loan performance. Asset-backed securitization is a financing technique in which financial assets are pooled and converted into instruments that may be offered and sold more freely in the capital markets. In a basic securitization structure, a financial institution known as sponsor constructs a pool of financial assets, such as mortgage loans, self-originated or acquired directly or through an affiliate. Securities backed by the pool of financial assets are then sold to investors by investment banks known as underwriters. Payment on the asset-backed securities depends primarily on the cash flows generated by the assets in the underlying pool and other rights designed to assure timely payment such as guarantees known as credit enhancements. Asset-backed securities and ABS issuers differ from corporate securities and operating companies in that there is generally no business or management to describe in offering these securities. Instead, information about the transaction structure and the quality of the asset pool and servicing is often what is most important to investors. 5 According to SEC, prior to Reg AB, many of the SEC existing disclosure and reporting requirements, which were designed primarily for corporate issuers, did not elicit the information that is relevant for most asset-backed securities transactions. Reg AB which became effective in January 2006 thus represents a comprehensive treatment of asset-backed securities under the Securities Act and the Exchange Act. It consolidates and codifies SECs positions and industry practice which it has done through no-action letters and the filing review process over time leading to Reg AB. 3 See Bank One Capital Markets, Inc., 2004 Structured Debt Yearbook. 4 See Asset Securitization Report (pub. by Thomson Media Inc). 5 See Securities and Exchange Commission Asset-Backed Securities Proposed rule Release NOS. 33-8419; 34-49644. 4

Reg AB covers four primary regulatory areas: Securities Act registration; disclosure; communications during the offering process; and ongoing reporting under the Exchange Act. 6 The new rules on disclosure represent the most dramatic changes on the ABS markets. Prior to Reg AB, there was no disclosure items specifically tailored to asset-backed securities. While eliminating unnecessary boilerplate and de-emphasizing unnecessary legal recitations of terms, Reg AB requires that issuers disclose information material to an asset-backed securities transaction, such as the background, experience, performance, and roles of various transaction parties, including the sponsor, the servicing entity and the trustee. It also requires, for the first time, that certain statistical information on a static pool basis be provided if material to the transaction to aid in an investors analysis of current and prior performance. Specifically on loan originators, Reg AB Item 1110 requires identification of any originator or group of affiliated originators that originated, or is expected to originate, 10% or more of the pool assets; and requires disclosure of information regarding the size and composition of the originator s origination portfolio as well as information material to an analysis of the performance of the pool assets, such as the originator s credit-granting or underwriting criteria for the asset types being securitized, if the originator originated or is expected to originate 20% or more of the pool assets. Thus, loan originators of 20% or more of the collateral pool represents an important disclosure threshold which did not exist prior to Reg AB. Our empirical investigation on the effect of Reg AB on securitized loan performance will focus on the change in loan originators in the disclosure threshold and associated cross sectional variations in loan performance of these originators. Prior to Reg AB, the SEC positions on the ABS issuance was done through no-action letters. These positions and industry practice are consolidated and codified under Reg AB. Therefore, in the post-reg AB period, riskier loans in a deal from certain originators may be used strategically to keep their fraction in the deal below the threshold in order to 6 See Securities and Exchange Commission Regulation AB Final Rule 33-8518. http://www.sec.gov/rules/final/33-8518.pdf. 5

avoid mandatory information disclosure. As a consequence, we expect the number of deals involving below threshold low-stake originators (20 percent under Reg AB) to increase in mortgage deals after Reg AB. This gives us the following testable hypothesis. Hypothesis 1: All else equal, the likelihood of a deal involving low-stake originators is higher after Reg AB than before Reg AB. There are two main motives on the use of low-stake originators. The first is to achieve lower SEC compliance costs. Apparently, Reg AB imposes higher compliance costs on issuers of mortgage deals consisting of loans from originators with stakes over 20% in a deal. If possible, ABS issuers may use loans from low-stake originators to minimize SEC compliance costs. However, the presence of low-stake originators should have no impact on deal or loan performance under this motive. The second possible motive is to avoid disclosure of information on riskier loans which constitute a low-stake in a deal. In pre-reg AB period, a deal does not need to limit an originator s loans to under 20% of the deal to avoid information disclosure because of no disclosure threshold requirement. After Reg AB, if ABS issuers take in riskier loans and also avoid scrutiny, it is much more likely to limit these loans to under the 20% threshold in the deal to avoid scrutiny by investors and regulators. In other words, if avoiding information disclosure is the main motive, we would expect the performance to be worse for deals with the increased presence of more low-stake originators, after Reg AB. Hypothesis 2: All else equal, the increased presence of low-stake originators should be associated with worse performance of securitized mortgages after Reg AB than before Reg AB. 3. Data description and summary statistics Our data comes primarily from two sources: Bloomberg and First American Corelogic Loan- Performance. We collect information on deal characteristics from Bloomberg which provides information on mortgage originator(s) and underwriter(s) extracted from deal prospectuses. 6

Our sample consists of privately securitized mortgage deals that were issued between 2003 and 2007, a period immediately preceding the financial crisis. Each deal in our database has detailed information on deal characteristics at issuance. In the meantime, our loan level data consists of information on privately securitized mortgages constructed by Corelogic LoanPerformance. It provides information on loan origination date, the mortgage loan pool, the identity of the securitizer, the MBS where the loan is placed, and detailed information on borrower and loan characteristics. We also construct variables from various sources on regional housing and economic conditions at the time of loan origination. Bloomberg reports the identities of originators and the percentage of loans that each of them originates for the deal. Not every deal provides origination information, thus we focus on a sample of 2,248 deals for which origination information is available for our investigation. From the detailed origination information, we identify deals that have originators that contributed 10-20% or below 20% of the collateral pool. Considering the disclosure requirements of Reg AB, we use 10-20 percent as the main measure of a low-stake originator and use below 20 percent as an alternative measure. We also calculate the aggregate percentage of loans originated by these low-stake originators for each deal. In our sample period, 18% (22% if we use below 20% as the criterion) of the deals have low-stake originators. The unconditional mean of the aggregate percentage of low-stake loans for a deal is 4.8% to 5.6%. The highest aggregate low-stake origination percentage is 100%. In other words, in the extreme case, a deal could consist of loans entirely from low-stake originators. For deals with originators contributing 10-20% loans in a deal, the percent of loans from these originators are on average 25.8%. For deals with originators contributing less than 20% collateral pool, the percent of loans from these originators are on average 24.5%. Our deal level performance measure is the cumulative net loss rate measured as the sum of all losses of principal suffered until December 2010 divided by the total original balance of all mortgages. We utilize an extensive list of deal characteristics as control variables. These include deal original collateral balance, an indicator for high reputation following 7

Griffin, Lowery, and Saretto (2014), the number of tranches, average share of loans that are low or no documentation in the collateral, average FICO score, weighted average loanto-value (LTV) ratio, percentage of adjustable rate mortgages in the deal, an indicator for negative amortization, percentage of purchase loans, percentage of loans for single family house, percentage of loans for owner-occupied house, and percentage of second lien loans. Bloomberg provides information on mortgage originator(s) collected from the deal prospectus supplements, but not on individual loans. To assign individual loans to a particular originator in a deal with multiple originators, we use Corelogic LoanPerformance database that provides the name of the original lender for each loan, where it can be a direct lender or a mortgage broker/correspondent. We collect identity and affiliation information for the original lender of each loan to determine if the original lender is one of the mortgage deal originators or is affiliated with one of these deal originators. When such a link can be made, we can assign individual loans to the originators listed in the prospectus supplements. When the original lenders cannot be linked to any of the originators as is often the case with the loans acquired by the originators, we set originator information for these loans as missing and exclude them in our loan level analysis. Definitions for all variables at both deal and loan levels are described in the Appendix. Since changes in house prices have an impact on mortgage performance, we include additional control variables in our analysis. For deal level analysis, we calculate the house prices change for the representative geographic area using the housing price index for the corresponding state reported by the Federal Housing Finance Agency (FHFA). Specifically, we compute weighted-average house price change associated with a deal from the quarter that the deal was issued to the last quarter of 2010. We also compute pre-deal housing price appreciation over the four quarters preceding the issuing quarter. For loan level data, we compute housing price appreciation over the 24 months after origination using the housing price index for the borrowers metropolitan statistical area (MSA) reported by the Office of Federal Housing Enterprise Oversight (OFHEO). We also compute the change in the 8

state-level unemployment rate over the 24 months after origination using data reported by the Bureau of Economic Analysis and collect the median household income in 1999 for the borrowers zip code as reported by the U.S. Census Bureau in 2000. Additionally, we include credit spread and 10-year Treasury yield as macro control variables. We begin our investigation with deal level analysis. Table 1 reports summary statistics for deal level variables. Table 1 about here Table 2 reports the correlation coefficients on the main variables of interest at deal level. The cumulative net loss is significantly positively correlated with the presence of low-stake originators and the aggregate percentage of loans originated by low-stake originators in these deals. The results are very similar for both measures of low-stake originators: percentage of loans in a deal within 10-20% or percentage of loans in a deal below 20%. Consistent with our intuition on higher risk associated with worse performance, deal cumulative net loss is positively correlated with original collateral asset value, the average loan-to-value ratio, percentage of adjustable rate mortgages, the presence of negative amortization loans, percentage of purchase loans, and percentage of loans with second lien. It is also negatively correlated with the average FICO score and the percentage of single family home loans. Table 2 about here 4. The change in origination structure and its impact on deal performance We start our empirical analysis by examining the impact of Reg AB on the use of low-stake originators. We then focus on investigating the impact of this origination structure change on the performance of securitized mortgages at the deal level. 9

4.1. The change in origination structure under Reg AB We define low-stake originators as those who contributed an amount to the total collateral mortgage pool in an MBS deal that is less than the threshold necessitating mandatory disclosure by SEC under Reg AB. Specifically, to test our hypothesis on the impact of Reg AB, we define low-stake originators as those who contributed between 10 and 20 percent to a mortgage pool. As a robustness check, we also use an alternative low-stake originator definition as those with less than 20 percent loans in the underlying collateral pool. There is an empirical issue related to the below 10 percent disclosure, especially in the post 2006 period because according to Reg AB, there is no requirement on the disclosure of identity for these originators. Therefore, the disclosure of below 10 percent is voluntary and we do not observe all below 10 percent originators. 7 This may lead to a measurement error on whether a deal has and who are the below 10 percent originators and the total percentage of loans originated by this type of originators. Consequently, we focus primarily on the low-stake originators contributed 10-20% of the collateral pool. To visualize the change in origination structure, we plot the number and percentage of deals with low-stake originators in our sample period in Figure 1. The top panels of Figure 1 present the plots for deals with originators contributing 10-20 percent of a collateral pool before and after Reg AB. Both the number and the percentage of deals with low-stake originators show similar pattern sourrounding Reg AB. Specifically, the number of deals with low-stake originators experienced a sharp increase from 121 in the pre-reg AB regime (before 2006) to 303 in the post-reg AB regime (after 2006). In percentage terms, the increase is more than doubled from around 11% before Reg AB to 27% after Reg AB. Moreover, from the bottom panels of Figure 1, it is clear that the percentage of low-stake deals is relatively stable in the pre-reg AB period and the sharp jump occurred right after Reg AB became effective and remained high. 7 Though not directly related, Lee and Mason (2012) show that affiliation matters for the loan-level selective disclosure of originators. 10

Figure 1 about here The increase in the use of low-stake originators pre- and post-reg AB is statistically significant. In Table 3, we use logistic regressions to evaluate this change by controlling for other factors that may affect the deal structure. In column (1), the dependent variable is a dummy variable which takes a value of one if a deal has at least one 10-20 percent originator and takes a value of zero otherwise. The result shows a very significant increase in probability (more than tripled) that a deal would involve at least one 10-20 percent originator in the post Reg AB period (e 1.32 = 3.74). We find similar result in column (2) when we use the alternative dependent variable to capture the presence of at least one low-stake originator contributing less than 20 percent to the collateral pool. Table 3 about here To demonstrate the change in origination structure around the 20% threshold, we examine the difference in the percentage of mortgage deals with originators contributing loan fractions just below 20%, say [10,20)%, [15,20)%, and [18,20)%, and the percentage of mortgage deals with originators contributing loan fractions just above 20%, say [20,30]%, [20,25]%, and [20,22]%, respectively, before and after Reg AB. Figure 2 shows the difference between the percentage of mortgage deals in respective brackets. We observe a jump up after Reg AB in 2006 and remained high for 2007. This pattern is robust for all comparison brackets: before Reg AB, the difference between the brackets around the 20% threshold is negative or marginally positive; after Reg AB, this difference becomes positive in all comparison brackets, indicating increases in deals in the brackets just below 20% than just above 20%. Considering that the loan pools before Reg AB may be different from that after Reg AB, we apply a difference-in-difference test to the differentials in the percentages of deals with just below 20% originators and just above 20% originators. We observe a differential of 6.6% for [10,20)% versus [20,30]%, 5.8% for [15,20)% versus [20,25]%, and 2.8% for [18,20)% versus [20,22]%, respectively. Our test results show that the increase in the percentage of deals with 11

just below 20% originators relative to that with just above 20% originators is statistically significant at 1% test level for all three comparison brackets. Figure 2 about here Next, we evaluate the increase in the use of low-stake originators quantitatively, controlling for the other factors that may affect the structure as well as the lead underwriter fixed effect. Table 4 reports the results for OLS (panel A) and ordered logit (panel B) regression analysis. Clearly, there is a significant increase in the fraction of mortgage deals that involves originators changing from contributing just above the threshold 20% of the total collateral pool to just below the threshold after Reg AB. Our OLS estimation shows that controlling for deal characteristics, issuer reputation, and macroeconomic variables, the fraction of deals with originator contributions from just above the threshold of 20% to just below the threshold increased by 15% from [20,30]% to [10,20)%, 8% from [20,25]% to [15,20)%, and 3% from [20,22]% to [18,20)%, respectively, from before Reg AB to after Reg AB. Given that the average fractions of deals with low-stake originators in [10,20)%, [15,20)%, and [18,20)% brackets are 11%, 5.7%, and 2.1%, respectively, before Reg AB, our estimates show that the percentage of deals with low-stake originators just below the threshold increased by 136% for [10,20)%, 140% for [15,20)%, and 142% for [18,20)%, respectively, under Reg AB. Qualitatively similar results are also observed using ordered logit regression analysis. For instance, the log-odds ratio of mortgage deals with originators reducing their contributions from just above the threshold to just below the threshold is higher by 92% for [20,30]% to [10,20)%, 65% for [20,25]% to [15,20)%, and 54% for [22,22]% to [18,20)%, respectively, after Reg AB. Table 4 about here 12

4.2. The impact of origination structure change on deal performance Now that we have documented a significant increase in the low-stake originators just below the 20% threshold in MBS deals after Reg AB, we seek to understand whether this origination structure change has any impact on mortgage performance. Due to lack of regulation specific to MBS disclosure in pre-2006 period, if MBS issuers engage in riskier loans and attempt to avoid disclosing adverse information, they did not have to limit riskier loans from a particular originator to the 20% threshold. After Reg AB, due to the required information disclosure regulation on originators with 20% or more loans in a deal, if MBS issuers take in riskier loans and attempt to avoid disclosure on the originators of these riskier loans, they are much more likely to strategically use the statutory 20% threshold to achieve the objective. Therefore, for deals issued after Reg AB, we would expect deals with increases in low-stake originators to have worse performance if the low-stake position is used strategically to avoid adverse information disclosure about the origination process and loan quality. On the other hand, we would expect the presence of low-stake originators to have no impact on deal performance if it were primarily driven by the motive of reducing SEC compliance costs. To test our hypothesis, we regress deal cumulative net loss on variables that capture the presence of low-stake originators and their interactions with a post Reg AB dummy variable. The inclusion of the interaction term allows us to assess if the change in low-stake originators in mortgage deals has an incremental effect post-reg AB than pre-reg AB. We use two measures for the presence of low-stake originators in our regression analysis: (1) a dummy variable representing the presence of at least one low-stake originator in a mortgage deal; and (2) a continuous variable that captures the aggregate percentage of loans originated by the low-stake originators in a deal. We do so for low-stake originators contributing 10-20 percent of a collateral pool and for low-stake originators contributing less than 20 percent collateral pool, respectively. 13

The results are reported in Table 5. Columns (1) to (4) present the findings for the low-stake originators contributing 10-20 percent of a collateral pool. It is quite clear that prior to Reg AB, the disclosure threshold does not have any significant impact on deal performance. However, after Reg AB, deals with low-stake originators have significantly worse performance. Specifically, the estimate in column (2) indicates that the presence of at least one 10-20 percent low-stake originator corresponds to 1.83 percentage points higher deal cumulative net loss. This represents 27% of the average cumulative net loss in our full sample period (1.83/6.74). When using the aggregate percentage of loans originated by 10-20 percent originators as the measure of low-stake originator involvement, our estimate shows that a one standard deviation increase in this aggregate percentage of low-stake loans leads to a 0.8% increase in the cumulative net loss which represents 12% average cumulative net loss for our full sample (0.8/6.74). Our results are robust if we define the low-stake originators using less than 20 percent collateral pool contribution in a deal (see columns (5) to (8)). Table 5 about here To provide further evidence that the effect on mortgage deal cumulative net loss associated with the presence of low-stake originators pre- and post-reg AB is due to the regulatory disclosure threshold, we conduct regression analysis including both the presence of originators contributing 10-20% collateral pool and originators contributing 20-30% collateral pool, a bracket just above the threshold. Table 6 reports the results of our analysis. For both measures of the low-stake originators (the dummy variable representing the presence and the continuous variable representing the percentage of collateral loan pool), we find that originators contributing 20-30% collateral pool had no significant effect on deal cumulative net loss. On the other hand, low-stake originators contributing 10-20% collateral pool are associated with significantly larger deal cumulative net loss. More important, the larger deal cumulative net loss is concentrated in deals with the presence of low-stake originators contributing 10-20% post-reg AB. This finding highlights the effect on deal performance of 14

loan originators who contributed 10-20% collateral pool post-reg AB, an amount just below the disclosure threshold. Table 6 about here 4.3. Deal performance and strategic use of the threshold Next, we investigate the impact of the strategic use of the 20 percent threshold on mortgage deal performance. For this analysis, we introduce a dummy variable Strategic Originator 10-20% orig. increase to represent the increase in the number of 10-20% originators in each deal pre- and post-reg AB. Specifically, for each originator, we compute the change in the percentage of deals for which the originator contributed a low stake (10-20%) in a collateral pool before and after Reg AB. For each deal, we define the dummy variable which takes a value of one if there are originators experienced an increase in the number of deals for which these originators contributed a low-stake in a collateral pool pre- and post-reg AB, and takes a value of zero otherwise. Similarly, we define a dummy variable Strategic Originator below 20% orig. increase. As an alternative, we also use Strategic Originator (High 10-20% orig. increase) to represent a dummy variable which takes a value of one if there are originators experienced larger than the average increase in low-stake originators and takes a value of zero otherwise. Similarly, we define a dummy variable Strategic Originator (High below 20% orig. increase). For our sample of 149 originators, the average increase in low-stake originators is 2.0% for the 10-20% loan contribution group and 6.4% for below 20% loan contribution group post Reg AB. Table 7 reports the results of our analysis. Our estimate shows that deals with increased strategic originators in 10-20% origination contribution group are associated with 0.7% higher cumulative net loss than deals with no strategic originators (column (1)). Similar result is found when we use the dummy variable Strategic Originator below 20% orig. increase (column (2)). When using the alternative measure Strategic Originator (High 10-20% orig. increase), we find that deals with more than average increase in strategic originators in 15

10-20% origination fraction group are associated with 1.1% higher cumulative net loss than other deals (column (3)). Similar effect is found for Strategic Originator (High below 20% orig. increase) (column (4)). Consistent with our intuition, we observe a larger impact on the cumulative net loss on deals with originators shown larger than average increase in low-stake originators. Table 7 about here 4.4. The impact of origination structure change on deal yields and credit enhancement One question is whether the higher cumulative net loss of mortgage deals that experienced increases in low-stake originators is reflected in deal initial yield spreads and credit enhancement. This is relevant for assessing the cost to investors of the disclosure threshold under Reg AB. For credit enhancement, we focus on subordination which is measured as the percentage of the face value of trust securities not rated AAA by Moodys or Standard & Poors at deal close. For deal yields, we use the initial average yield spread of all securities issued by the trust of mortgage deals. This is the difference between the average yield of all securities issued by the trust weighted by the face value of the securities and the yield on the 10-year Treasury bond. The former is calculated using the standards of the Bond Market Association and reported by Bloomberg. Table 8 reports the results of our regression analysis. Panel A shows that the presence of low-stake originators has no effect on deal yield spread. The deal structure change associated with the 20% threshold under Reg AB does not change the result. This is true for both low-stake originators contributing 10-20% collateral pool (column (1)) and contributing less than 20%collateral pool (column (2)). However, the presence of low-stake originators has a significant impact on credit enhancement measured by mortgage deal subordination. Under both measures of low-stake originators, their presence is associated with a higher average 16

subordination before Reg AB and a lower suboridination post Reg AB (columns (3) and (4)). This shows a sharp contrast on the impact of the presence of low-stake originators on subordination with respect to Reg AB. Combining the findings on deal yield spreads and subordination, we have evidence suggesting that investors may have not impounded the increase in low-stake originator risk due to the 20% disclosure threshold under Reg AB in yields and credit enhancement. Table 8 about here Panel B reports results of analysis on whether the strategic use of the 20% threshold under Reg AB is reflected in deal yields and credit enhancement. For both measures of the strategic use of the threshold under Reg AB, we find no evidence that the increase in the strategic use of the threshold in deals with larger than average increase in the number of deals with low-stake originators contributing 10-20% or below 20% collateral pool is reflected in deal yield spreads or subordination. Overall, our results on Reg AB have two very important implications. First, the 20% disclosure threshold was used strategically after Reg AB. Second, this strategic use is likely driven by avoiding scrutiny and potentially withholding some adverse information on riskier loans. The policy implication is that although alleviating certain compliance costs may be beneficial for certain ABS issuers, a cost (of larger magnitude) may be transferred to the investors. Therefore, it may be beneficial to further tighten the regulation of ABS market by not leaving so much room for certain non-disclosures. 5. The change in origination structure and its impact on loan performance We now turn to investigating the impact of origination structure change on mortgage performance at loan level. We first directly compare the loans originated by the strategic originators 17

with those originated by others and then examine how the performance difference is related to strategic originators increasing their number of mortgage deals with 10-20% stakes. We recognize that loans from originators with different stakes in a deal may have different quality. To control for this, we introduce the variable loan origination percentage ( Origpct ) as follows. For each loan i in mortgage deal j, we compute the percentage of loans in deal j originated by the same originator that originated loan i and assign this percentage to all loans originated by the same originator in deal j. Following the definition of a strategic originator in Table 7, we compute the change in the percentage of deals for which an originator contributed 10-20% collateral pool before and post Reg AB ( Origchg ). Using loan origination percentage we can further divide loans by their originators stake size when these loans were placed in mortgage deals. Merging the deal level originator information with loan level data and excluding missing observations, we have more than three and a half million loans in 1,603 deals. The average loan origination percentage is 86%, suggesting that majority loans were originated by originators with stake size larger than the threshold. In the meantime, three percent of all loans belong to the 10-20% stake group. The average of the variable Origchg is 4.1% with a standard deviation 11.3% for all loans, slightly different from those at deal level. Following the standard practice, we use securitized loan delinquency, defined as 60 days or more past due within 24 months of loan origination, as loan performance measure in our loan level analysis. In Table 9, we report the summary statistics for the loan level variables for the whole sample and subsamples of loans from 10-20% and 20-30% stake groups, respectively. We use loans from 20-30% stake group for comparison due to their vicinity in stake size. We observe that the sample averages for these variables are close between the whole sample and subsamples, and even closer between the two subsamples. Table 9 about here We take two steps in our loan level analysis. In the first step, we examine whether strategic originators have worse performing loans than others. Since the strategic originators 18

are more likely to increase the 10-20% stake deals after Reg AB, we expect that loans from the originators with larger increases in 10-20% stake deals are riskier than loans by other originators. This suggests that loan delinquency will increase in the variable Origchg. In the second step, we investigate whether loans in the 10-20% stake deals by strategic originators are worse than loans by other originators. We use loans in the 20-30% stake deals as a control group because of their close proximity to the 10-20% stake deals. If MBS issuers strategically place riskier loans in the 10-20% deals, we expect the variable Origchg has a larger positive effect on delinquency for loans in the 10-20% deals than those in the 20-30% deals. We conduct this analysis separately for pre and post Reg AB subsamples of loans because we expect the strategic use of the 10-20% deals occurs post Reg AB but not before Reg AB. Table 10 reports the marginal effects from probit regression for the whole sample with Origpct (column (1)), the whole sample with both Origpct and Origchg (column (2)), the subsample of loans with the stake size 10-20% (column (3)), and the subsample of loans with the stake size 20-30% (column (4)). Our estimation shows that for the full sample the loans from larger stake sizes have lower delinquency. This finding makes it necessary to control for the stake size in our subsequent analysis. As expected, for the strategic originators the variable Origchg is positively assoicated with delinquency, controlling for the stake size and various loan level controls. The economic magnitude is significant in that loans from an originator who increased the fraction of low-stake deals by 10% are 0.5% more likely to be delinquent. For the subsample of loans from 10-20% stake deals we observe the same effect for Origchg that the low-stake loans from strategic originators are worse than the low-stake loans from other originators. On the other hand, we find the difference for loans from 20-30% stake deals between strategic originators and others has the opposite sign, a striking contrast around the 20% threshhold. Table 10 about here Next we explicitly test whether the strategic originators utilize 10-20% stake deals dif- 19

ferently from their closest 20-30% stake deals post-reg AB versus pre-reg AB. Table 11 presents the results of a probit regression on loans from 10-30% stake size for the pre-reg AB subsample (2003-2005) and the post-reg AB subsample (2006-2007). Our estimation for the pre-reg AB subsample loans shows that the variable Origchg is positively associated with loan delinquency, yet statistically insignificant for the pre-reg AB period. However, the effect on the loans from the 10-20% stake deals is actually weaker than on those from 20-30% stake deals. This suggests no strategic use of the 10-20% stakes pre-reg AB, a finding support our intuition. In contrast, our estimation for the post-reg AB subsample shows that the effect of Origchg is much stronger for loans from 10-20% stake deals than those from 20-30% stake deals. This indicates that the strategic use of low stakes concentrates in the 10-20% deals. We note that Origchg is negatively associated with delinquency for loans from 20-30% stakes. This is consistent with the explanation that the strategic originators shift riskier loans into the 10-20% stakes from 20-30% stakes. The sharp contrast in loan delinquency rate pre- and post-reg AB illustrates the strategic originator s use of 10-20% stakes for riskier loans. This lends support to our finding on the impact of strategic originators on the deal cumulative net loss documented above. Table 11 about here 6. Conclusion How to design and implement effective regulation has received widespread attention following the 2007-2008 subprime mortgage crisis. Very little is known about the impact on the nonagency MBS market of the regulation on ABS, Reg AB, implemented during the height of the housing boom right before the crisis. Even less is known about the effects of these regulations on the market participants and resulting economic impact. In this paper we fill in the void on these issues. 20

One of the most important aspects of Reg AB is the disclosure requirement on the part of the mortgage originators. Specifically those originators who contribute more than 20% of the loans in the collateral pool are required to provide detailed financial information material to the investor analysis of the collateral assets. The purpose of this requirement is to encourage transparency and therefore accountability. Using the mortgage deals before and after Reg AB, we find that certain originators circumvent this requirement by staying below the 20% threshold. Furthermore, it is exactly these originators that contributed to the worse performance of the MBS deals. Our loan level analysis provides support evidence on the findings of deal analysis. This suggests that the beneficial effect of the 20% disclosure threshold requirement has been somewhat mitigated and its effectiveness curtailed. Overall, our study on how these regulations change the market participants behavior and the ensuing economic impact can shed light on future research and policy-making regarding the asset-backed securities markets. There are other aspects of these regulations that could potentially change these markets and we leave those topics for future research. 21

References Griffin, John, Richard Lowery, and Alessio Saretto, 2014, Complex securities and underwriter reputation: Do reputable underwriters produce better securities?, Review of Financial Studies. Keys, Benjamin J., Tanmoy Mukherjee, Amit Seru, and Vikrant Vig, 2010, Did securitization lead to lax screening? evidence from subprime loans, The Quarterly Journal of Economics 125, 307 362. Keys, Benjamin J., Amit Seru, and Vikrant Vig, 2012, Lender screening and the role of securitization: Evidence from prime and subprime mortgage markets, Review of Financial Studies 25, 2071 2108. Kothari, S. P., Susan Shu, and Peter D. Wysocki, 2009, Do managers withhold bad news?, Journal of Accounting Research 47, 241 276. Lee, Hong, and Joseph R Mason, 2012, Selective disclosure of originators in mortgage securitizations, Working Paper. Mian, Atif, and Amir Sufi, 2009, The consequences of mortgage credit expansion: Evidence from the U.S. mortgage default crisis, The Quarterly Journal of Economics 124, 1449 1496. Nadauld, Taylor D, and Shane M Sherlund, 2009, The role of the securitization process in the expansion of subprime credit (Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board). Purnanandam, Amiyatosh, 2011, Originate-to-distribute model and the subprime mortgage crisis, Review of Financial Studies 24, 1881 1915. 22

Skinner, Douglas J, 1994, Why firms voluntarily disclose bad news, Journal of accounting research pp. 38 60. Skinner, Douglas J., 1997, Earnings disclosures and stockholder lawsuits, Journal of Accounting and Economics 23, 249 282. 23

Appendix: Variable definitions Deal and macro variables: Cumulative net loss: Historical percentages of cumulative loss on the underlying loans comprising the entire collateral that backs the deal Has originator 10-20% (d): Equals 1 if a deal has (an) originator(s) originate(s) percentage of loans between 10 Has originators < 20% (d): Equals 1 if a deal has (an) originator(s) originate(s) loans below 20%; 0 otherwise Total percentage of origination 10-20%: Total percentage of originations that are between 10% and 20% Total percentage of origination below 20%: Total percentage of originations that are below 20% Original collateral balance (in Billions): The original balance of the underlying loans comprising the entire collateral High reputation: Equals 1 if the deal has an underwriter IPO reputation score greater than or equal to 8 (from Professor Jay Ritter s website); 0 otherwise No. of tranches: Number of securities in a deal Low documentation: Dummy variable indicating underlying loans with limited, as distinguished from full, documentation FICO: Weighted average original credit score of the underlying loans LTV: Original loan to value percentage of the loan Adjustable rate mortgage: The percentage of the adjustable rate mortgage loans Negative amortization: Equals 1 if the deal consists of mortgages with negative amortization features; 0 otherwise Purchase loans: The percentage of the Loan Purpose (the reason for the loan) for Purchase Single family: The percentage of Single Family Mortgaged Properties, the type of properties against which the loans were written Owner occupied: The percentage of the Occupancy (the purpose of the property) for Owner Occupied Second lien: The percentage of the loans comprising the collateral that are second lien House prices change: We compute the average house price changes from issue quarter to the last quarter of 2010 using the state level Federal Housing Finance Agency s (FHFA) seasonally adjusted quarterly house price index. The weighted average for each deal is taken over the top five states by their mortgage balances assuming the remaining 45 states have equal representation House prices run-up: We use the same data and method as in House prices change to calculate the weighted average price change associated with a deal during the four quarters preceding the quarter the deal was closed Credit spread: The spread between BBA and AAA corporate bond yields in issue month 24