1. Adverse Selection in Mortgage Securitization Sumit Agarwal Yan Chang Abdullah Yavas

2. Disclaimer The comments here reflect only the authors ’ views and not those of Freddie Mac or its Board of Directors.

3. Introduction One of the worst financial and economic crises since the Great Depression Triggered by the collapse of the bubble in residential real estate markets Role of securitization Purpose: to empirically examine potential adverse selection problems in mortgage securitization. whether the loans lenders sell into the secondary mortgage market are riskier than the loans they retain in their portfolios. 3

4. Introduction The conventional wisdom – Adverse Selection due to asymmetric information Hard data vs. soft data Mortgage loans involve three kinds of risk: Interest rate risk Prepayment risk Default risk We focus on the prepayment risk and default risk and compare portfolio loans with securitized loans with respect to these two risk types. We ignore interest rate risk, since it is independent of the borrower ’ s characteristics, hence is not subject to potential adverse selection concerns. 4

5. Players Lenders typically sell their mortgage loans in the secondary market to either Fannie Mae or Freddie Mac (GSEs), or to a private sector financial institution, such as subsidiaries of investment banks, large banks and homebuilders. Of the total volume of $7.6 trillion in pooled mortgages at the end of 2008, about $5 trillion is securitized or guaranteed by GSEs while the remaining $2.6 trillion pooled by private mortgage conduits. 5

6. GSEs vs. Private Labels GSEs offer investors guarantees against default risk while private issuers often pass the default risk onto parties willing to bear it. While mortgage securitizations by GSEs typically involve a single form of an investment bond called Pass-Through Certificates, private placement of mortgage backed securities involves multiple forms of investments created by splitting the P and I components of the mortgage pool into various tranches. GSEs have historically purchased only traditional FRM products and only recently begun purchasing alternative mortgages. Private label issuers have been purchasing these alternative mortgages at much larger scales and for a longer period of time. GSE issuance of mortgage backed securities is subject to detailed SEC filings and public reporting requirements while private issuers are not typically subject to any SEC filings or reporting. 6

7. Significance Total volume of outstanding mortgage debt in the U.S., has grown three times as fast as the GDP over the last decade, is about $14.6 trillion With more than $8.8 trillion in mortgage-related securities As the current crisis has illustrated, a jump in default rates in the mortgage market can have dramatic consequences for the real economy. important to identify and investigate the factors that may be contributing to higher default rates in mortgage markets. 7

8. Literature Review A number of recent studies investigate the impact of securitization on the quality of loan screening and servicing in mortgage markets. Using Lender Processing Services (LPS) dataset, Elul (2009) finds that securitized prime loans have higher default rates than portfolio loans. However, securitized subprime loans do not perform worse than portfolio loans. Piskorski, Seru, and Vig (2008) investigate the impact of securitization on loan servicing by examining whether securitization inhibits modifications of loans for distressed borrowers. Studying the loans that are seriously delinquent, they find evidence of moral hazard: significantly lower foreclosure rates for portfolio loans than for securitized loans. 8

9. Literature Review Adelino, Gerardi and Willen (2009) challenge the conclusions of Piskorski, Seru, and Vig (2008) and offer evidence that servicers renegotiate similar fractions of portfolio loans and securitized loans. Keys, Mukherjee, Seru and Vig (2009) compare the performance of subprime loans with credit scores just above and just below the cutoff point of 620. Expected: Lenders will apply weaker screening standards to applicants with credit scores just above the cutoff Indeed, they find that loans with credit scores just below the cutoff point are less likely to default. Bubb and Kaufman (2009) argue and offer evidence that lenders collect more information about borrowers with just below the cutoff credit score because the benefits to lenders of collecting additional information are higher for higher default risk borrowers. 9

10. Literature Review The paper most closely related to ours is Ambrose, LaCour-Little and Sanders (2005). Using data from a single lender, they find that securitized loans of the bank in their data performed better than the loans retained. The authors attribute their result to two factors; reputation concerns and regulatory capital requirements. We: Use data that covers more than 4500 lenders Consider two alternative models of expectations Analyze prepayment as well as default risk Compare GSEs and private issuers Large vs. Small Lenders 10

11. Data LPS Applied Analytics data set As of July 2008, the dataset included loans from 9 of the top 10 servicers and represented around two-thirds of the mortgage market in the United States, or more than 39 million active mortgage loans. Includes agency and non-agency mortgage-backed securities as well as portfolio loans, has extensive information about the loan, property and borrower characteristics at the time of origination as well as how the loans subsequently performed. We merged the LPS dataset with Home Mortgage Disclosure Act (HMDA) data, so that we have access to additional information about the borrower and the lender (income of the borrower, demographic information, differences across lenders) 11

12. Data We focus on conventional, FRMs on single-family residences and condos originated between January 2004 and June 2008. Prime and subprime loans Second mortgages, HELOCs and loans above $650,000 were excluded. Imposed additional restrictions on the prime loans: FICO score above 620 and a loan-to-value ratio of below 95 percent. 78.3% of the loans were agency loans, 7.4% were held in portfolio and 14.3% were non agency loans that were securitized. 12

13. Data Addition of new servicers to the dataset over time means that both loans that have been serviced for years as well as new loans are included in the dataset. This could potentially left-censor the data since earlier loans that have defaulted or been prepaid will not be included while loans that have remained current will. To reduce the extent of left-censoring in the data, we eliminated loans that entered McDash a year after origination. 13

14. Conforming Loans No certain way to do determine if a loan is conforming. We automatically label a loan as conforming if it was held by one of the agencies at some point during the 12 months after origination. If the loan was not held by one of the agencies, we label a loan as conforming if the FICO score was greater than 660 and the origination amount was below the conforming limit for that state and time and that the loan has PMI if the LTV ratio is above 80. Overall, 92.2% of the loans were defined as conforming under this definition. 14

15. Methodology – 4 steps 1. For prime and subprime loans and for each year of origination, we divide our sample population into a random 75% estimation sample and a 25% holdout sample. Based on the 75% estimation sample, we construct a competing risk hazard model using the observed default and prepayment outcomes in the following 24 months. 2. We apply the coefficients obtained from the first step to the holdout sample consisted of the remaining 25% of the population, and calculate their expected default and prepayment probabilities. 3. We account for the degree of over- or under-pricing for the loans in the holdout sample. 4. We regress the securitization choice for the loans in the holdout sample on their expected default probabilities, prepayment probabilities, over-pricing and under-pricing indicators obtained from the previous 2 steps, and other variables controlling for the market environment at the time of origination. 15

16. Methodology This approach: controls for the potential performance difference due to post-securitization factors, such as the moral hazard issues pointed out in Piskorski, Seru and Vig (2009), and approximates lender ’ s ex ante information at the time the securitization decision is made. 16

17. Modeling Lenders ’ Expectations We consider two types of expectation models: a) Perfect Foresight above (where lenders are assumed to have perfect foresight on how loans with similar characteristics will perform) b) Adaptive Expectations where the lenders form their expectations according to previous two years ’ experience. 17

18. Results – Prime Sector Perfect Foresight Coefficient on Expected Cumulative Default / Prepayment Rate Cumulative Default RateCumulative Prepayment Rate

19. Results – Prime Sector Adaptive Expectations Coefficient on Expected Cumulative Default / Prepayment Rate Cumulative Default RateCumulative Prepayment Rate

20. Results – Subprime Sector Coefficient on Expected Cumulative Default / Prepayment Rate Cumulative Default RateCumulative Prepayment Rate

21. Results – Prime Sector Perfect Foresight Small Lenders Coefficient on Expected Cumulative Default / Prepayment Rate Cumulative Default RateCumulative Prepayment Rate

22. Conclusion We find strong evidence that banks sold low default risk loans into the secondary market and retained higher default risk loans on their portfolios. The result holds for subprime as well as prime loans, although the difference is smaller for subprime loans. We find support for adverse selection with respect to prepayment risk; securitized loans had a higher prepayment risk than portfolio loans. In support of the Lenders ’ strategy, they became less and less likely to hold high default loans as we go from 2004 to 2007. Further indicator of adverse selection: Small lenders behave differently. We also find that, compared to loans sold to GSEs, loans sold to private issuers have lower prepayment rates while relative default rates show variations across years. 22

23. Discussion-1 Reputation concerns alone cannot explain these results. Lenders could presumably maintain good reputation by simply randomizing their decisions of which loans to securitize. One could argue that securitization often involves an exchange whereby a large originator assembles a pool of mortgages and trades the pool with the GSEs or private issuers in return for securities backed by the pool. T hus, one can argue that this should significantly reduce the potential for any adverse selection problems. However, owning securities backed by a pool is different than owning the loans directly because while the GSE bears the default risk in the former case, the originator bears the risk in the latter case. Furthermore, not all mortgage loan sales take place through a swap program. 23

24. Discussion-2 One compelling argument: GSEs and private labels have very high underwriting standards. It is possible that a subset of the loans that fail to meet the GSE and private issuer criteria are still acceptable to lenders. Also, if the total amount of loans meeting the secondary market criteria is smaller than the total demand by GSEs, private issuers and originators, originators may choose to approve and retain some of the loans whose risk levels are below the standards of secondary market purchasers. Thus, even if the originators pick a higher-quality portion of the conforming loans for their portfolios, the existence of nonconforming loans in their portfolios may push the average default rate of their portfolios above those of GSEs and private issuers. 24

25. Discussion-3 The above argument is enhanced by the fact that GSEs selectively check their loans that go into default, and if they discover that the lender's representation and warrants were violated, then they can force the lender to purchase the loan back at par. In addition, GSEs also check a random sample of their non-defaulted loans, and can force repurchase of all the loans that had any rep and warrant violations. Furthermore, GSEs keep track of the repurchase record of originators and impose higher fees and capital requirements on originators with high repurchase rates. These high expected costs of having securitized loans go into default may induce lenders to be more conservative with the loans that they sell into the secondary market than the loans that they retain on their portfolios. 25

26. Discussion-4 The most compelling explanation: in return for selling loans with lower or comparable default risks, lenders retain loans with lower prepayment risk on their portfolios. During the high refinancing years, retaining loans with lower expected prepayment rate might have been a much more profitable strategy than retaining the loans with lower expected default rate. Furthermore, trading low default risk for low prepayment risk also helps originators maintain their reputation, minimize the probability and the cost of being required to repurchase loans, and satisfy high underwriting standards of the secondary market, as these concerns all pertain to default risk, but not to prepayment risk. All of this further supported by our results for Small vs. Large lenders 26