The Preventable Foreclosure Crisis

Transcription

1 The Preventable Foreclosure Crisis Roberto G. Quercia Department of City and Regional Planning Center for Community Capital University of North Carolina, Chapel Hill, USA 1

2 Serving LMI/minority borrowers Traditionally: Federal Housing Administration -FHA insurance Starting in the late 1980s: Community reinvestment type products are mostly FRM with flexible but carefully underwritten standards. Limited. Starting in the late 1990s: Subprime loans are products characterized by risky underwriting reliance on rapid appreciation, low doc or no-doc, low-down or no-down, high debt ratios, originated by brokers. Starting in the late 2000s: FHA loans has gained market share after the inset of the current crisis, it is expected to gain even more market share (~30% of originations in 2008?) 2

3 How the Financial System Worked in the 1930s Depositors, lenders, mortgage borrowers 3

4 Community Reinvestment Lending 1977 Community Reinvestment Act 1992 Affordable housing goals of the government sponsored enterprises (GSEs) 30 year fixed rate mortgages With escrows Reduced dowpayments Only one underwriting gguideline at the time was liberalized Kept in portfolio lack of liquidity led to limited availability 4

5 Secondary Market In the past, loan remained in bank s balance sheet until it was repaid Today, party originating the loan will likely sell loan to a third party (Ginnie Mae, Fannie Mae, Freddie Mac) Third party packages mortgages and sells the payment rights to investors Some investors use their payment rights to back other securities they issue Securitization converts mortgages to mortgage backed securities 5

6 Mortgage Backed Securities A MBS is a bond whose payments are based on the payments of a collection of individual mortgages Initial sales of the bonds are put together either by the two GSEs or by private financial institutions such as Countrywide, Lehman Brothers or Wells Fargo As payments are made on the mortgages, they are passed through the trust to bondholders Rating agencies rate risk quality of bonds (e.g., Moodys, Standard and Poors) and insurance companies (e.g., AIG) 6

7 In summary 7

8 Enter Subprime Lending In the mid 1990s, subprime market emerged due to Availability of cheap credit Technological innovations which allowed for risk based pricing Need for new markets Push to deregulation Wall Street driven $35b in subprime originations in 1996 $625b in subprime originations in 2006 Refinance products for the most part 8

9 Enter the Subprime Market Subprime mortgages are characterized by Broker originated Teaser rates/adjustable rates No downpayment (up to 125% of property value) No/low documentation ti Prepayment penalties No escrows Balloon payments 9

10 Subprime Lending and Securitization 2006 Ginnie Mae facilitates the securitization of mortgages backed federally It guaranteed the mortgages underlying only 4% of all MBS issued Fannie and Freddie purchased conforming loans (<=$417,000) to back MBS Accounted for 40% of MBS issued Remaining 56% packaged by private sector financial institutions MBS backed by prime (good credit borrowers), subprime (lower credit borrower) or Alt-A loans (good credit but loans riskier, e.g. no doc) 71% of total private sector MBS issuances are subprime and Alt-AA 10

11 Structure of A MBS Security: Example MBS issuer collects 1000 mortgages, each worth $100,000 with a 30 year maturity and a fixed rate of 6.5% This $100 m pool can be used to back 10,000 bonds, each worth $10, with a 30 year term and a fixed coupon rate of 6% Pi Principal i repayments on the mortgages are used dto pay down the principal on the bonds 11

12 Structure of a MBS Security Bond Class Percent of Pool Rating A 94.15% AAA B1 2.00% AA B2 1.50% A B3 1.00% BBB B4 0.65% BB B5 0.40% B B6 0.30% Not rated 12

13 Re-Securitization Often pools of MBSs were collected and securitized Bonds that are themselves backed by pools of bonds are referred to as collateralized debt obligations i (CDOs) 13

14 Structure Investment Vehicles (SIV) SIV are similar to CDOs They are back by pools of assets such as MBSs and CDOs SIVs issue short and medium term debt (1 to 90 days) rather than the longer term debt of most CDOs SIVs are referred to as asset backed paper: pp backed by credit from the sponsoring bank SIV are sold to money market funds 14

15 In summary 15

16 Structure of a Toxic CDO Security MBS Class MBS Rating CDO Rating A AAA AAA B1 AA AA B2 A A B3 BBB BBB B4 BB BB B5 B B B6 Not rated Not rated 16

17 Performance of Different Loan Products Portfolio 90+day Delinquency Rate 20% 15% sub_arm 10% sub_frm FHA 5% prime_arm CAP prime_frm 0% Ma r-04 Jun n-04 Sep p-04 Dec c-04 Ma r-05 Jun n-05 Sep p-05 Dec c-05 Ma r-06 Jun n-06 Sep p-06 Dec c-06 Ma r-07 Jun n-07 Sep p-07 Dec c-07 Source: Mortgage Bankers Association and Self-Help. 90+day delinquencies include loans in different foreclosure stages. N=44,973 for the CAP portfolio. 17

18 High default rate of subprime mortgages: Risky borrowers? Risky products? Borrowers holding subprime loans are generally weaker across key underwriting criteria: collateral (down payment), credit history (credit score), and repayment capacity (debt ratios). Yet, many of these borrowers could have obtained a prime mortgage (Freddie Mac 2005) Subprime loans have features found to significantly add risk prepayment penalties and balloon payments (Quercia, Stegman, and Davis, 2007; Danis and Pennington-Cross, 2005) adjustable rate (Calhoun and Deng, 2002) hybrid ARMs (Ambrose, LaCour-Little, and Huszar, 2005; Pennington-Cross and Ho, 2006) It is difficult to isolate the impact of loan features unless we are comparing borrowers with similar risk characteristics. This study compares the relative risk of loan products (CRA and subprime) for borrowers with the same risk profile. 18

19 Data (I) McDash Analytics Loan level info about 15 residential mortgage servicers; Tracking over 20 million active loans on a monthly basis Good coverage on conventional prime loans and FHA market Modest coverage in the subprime market Definition of subprime loans B&C loans High-cost ARMs (arm margin greater than 300 basis points) 19

20 Data (II) Community Advantage Program (CAP) Partnership between Self-Help, Fannie Mae, Ford Foundation, and a group of lenders. Started in 1998, over 50,000 loans Secondary market outlet t for CRA-type loans Prime mortgage products serving borrowers with Low- to moderate- income Weak credit history and/or low down-payments A sample of 9,221 CAP loans originated during

21 Methodology: Propensity Score Match In the first phase, PSM method summarizes all of the background (covariate) information about taking out a subprime into a scalar which is called propensity score. In the second phase, subprime loans are matched to CAP loans based on the estimated propensity scores and the resultant sample usually is a subset of the original sample. In the last step, regression models can be applied to the matched group to compare performance of subprime and CAP loans. Advantages of PSM: it summarizes multidimensional covariates into a one-dimentional score; it provides an adjustment for selection bias to help support strong inferences about program effects using observational data. 21

22 Propensity Score Match: focusing on borrowers with similar risk characteristics but holding different products: CAP or subprime Prime Loans Subprime Loans Prime Borrowers Subprime Borrowers CAP Subprime 22

23 This following characteristics are same/similar: CAP Subprime Lien status first-lien first-lien Loan purpose home purchase home purchase Occupancy status owner occupied owner occupied Loan amount conforming conforming Property type single-family single-family Loan term 30-year 30-year Documentation full/alternative doc full/alternative doc Geography CAP in zip codes with CAP N ( ) 9,221 42,065 23

24 However, underwriting criteria and some loan characteristics are different: CAP Subprime Credit score DTI Underwriting 40.0% 43.1% LTV 96.5% 82.9% ARM % Broker originated % Prepayment penalty % Note-Rate 6.66% 7.87% N ( ) 9,221 42,065 24

25 Credit Score Distribution CAP subprime 0.04 Den nsity FICO Score 25

26 Explanatory variables to predict propensity of taking out subprime Underwriting variables directly affect credit risk, and therefore affect mortgage choice/assignment since higher credit risk is hypothesized to be associated with a greater probability of taking out a subprime mortgage: FICO score; DTI; LTV. We used the average FICO score for mortgages originated in the preceding gyear in the zip code as a proxy of local credit risk. Educational distribution is included as a proxy of residents financial knowledge/literacy. Herfindahl- Hirschman Index ( HHI ) is constructed to measure the extent of market competition in the market (the sum of squared market shares of lenders in a zip code). Share of minority is included to consider the local demographics. Quarterly calendar dummy variables. 26

27 Results of the logistic regressions Coef. P-value Coef. P-value Coef. P-value Coef. P-value dti< dti dti cscore< cscore cscore cscore loan_amt qtr qtr qtr HHI (in 10,000) 000) area credit score pctmin pct_high pct_somecoll pct_coll _cons N=7,350 N=20,96 1 N=13,95 4 N=9,021 27

28 Results of the propensity score match Sh Scheme Description of matching method N of original sample N of the new sample CAP CAP Subprime Match1 Nearest 1-to-1 using caliper=0.1 9,221 5,558 5,558 Match3 Nearest 1-to-many using caliper=0.1 9,221 3,943 35,971 28

29 Performance of the matched sample (multinomial logit model for the first 90-day Delinquency and prepayment) Sample Sample Mode l 1 Model 2 Model 1 Model 2 Variable Coef. P>z Coef. P>z Coef. P>z Coef. P>z Default subprime sub_arm sub_arm&ppp sub_bro sub_bro&pppbro&ppp sub_bro&arm sub_bro&arm& ppp cap Prepay subprime sub_arm

30 Results and Discussion (I) Cumulative Serious Delinquency Rate 24 Months after Origination (CAP vs Subprime) 50% CAP Subprime 40% 30% 20% 10% 0% 2004 Orig 2006 Orig Note: Estimation is based on a borrower with a FICO score between with the mean value of other regressors. 30

31 Results and Discussion (II) Cumulative Serious Delinquency Rate 24 Months after Origination (CAP vs Retail Originated Subprime Loans) CAP 50% sub_arm sub_arm&ppp 40% 30% 20% 10% 0% 2004 Orig 2006 Orig Note: Estimation is based on a borrower with a FICO score between with the mean value of other regressors. sub_arm represents retail originated subprime ARMs without prepayment penalties; sub_arm&ppp represents retail originated subprime ARMs with prepayment penalties. 31

32 Results and Discussion (II) Cumulative Serious Delinquency Rate 24 Months after Origination (CAP vs. Broker Originated Subprime Loans) 60% 50% 40% CAP sub_brobro sub_bro&ppp sub_bro&arm sub_bro&arm&ppp 30% 20% 10% 0% 2004 Orig 2006 Orig Note: Estimation is based on a borrower with a FICO score between with the mean value of other regressors. sub_bro&ppp and sub_bro represent broker-originated subprime FRMs with and without prepayment penalties respectively; sub_bro&arm&ppp and sub_bro&arm represent broker-originated subprimearms with and without prepayment penalties respectively. 32

33 Results and Discussion (IV) The broker origination channel, the terms of adjustable rate, and prepayment penalty all increase the default risk of mortgages. Why broker originated subprime loans are more likely to default? - Subprime loans are much more complex than conventional fixed-rate prime mortgages. It is difficult for borrowers understand complicated loan features and terms and compare costs of different products. - Are broker originated subprime loans have significantly higher prices? - Are broker originated subprime loans more likely to have high-risk features? - Are broker originated subprime loans more likely to have a looser underwriting standard? 33

34 Conclusions The higher default risk of subprime loans may not be attributed to borrower credit risk only, instead it is significantly associated with characteristics of loans products and the origination channel in the subprime market. For borrowers with similar risk profiles, the estimated default risk would be much lower for community reinvestment loans instead of subprime mortgages. 34