Understanding the Subprime Mortgage Crisis

Title Introduction Empirical Analysis Boom conclusions Understanding the Subprime Mortgage Crisis Yuliya Demyanyk1 1 Federal 2 New Otto Van Hemert2 Reserve Bank of St. Louis York University, Stern School of Business December 6, 2007 contact: Yuliya.Demyanyk@stls.frb.org The views expressed are those of the authors and do not necessarily reflect the official positions of the Federal Reserve Bank of St. Louis or the Federal Reserve System. Available at http : //papers.ssrn.com/sol3/papers.cfm?abstract id = 1020396 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 1/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data What is a Subprime Mortgage? 1 Subprime borrowers1 previous delinquency, foreclosure, bankruptcy credit score < 620; debt-to-income ≥ 50% 2 Subprime Lenders2 fewer originations, more refinances, sell less to GSEs, mostly high-cost loans 3 Subprime Securities riskier than Prime and Alt-A 4 “Riskier” borrowers “good” credit history but “zero-down” or “no doc”... Board of Governors FRS, OCC, FDIC, OTS HUD, HMDA. Issues: subprime lenders originate prime loans and prime lenders issue subprime loans; not all high-cost are subprime and not all subprime are high-cost Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 2/ 17 1 2 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data The Crisis in Subprime Mortgage Market Actual Delinquency Rate (%) 18 16 14 12 10 8 6 4 2 0 2 4 6 8 10 12 14 16 Age (Months) 18 20 22 24 2007 2006 2005 2004 2003 2002 2001 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 3/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Delinquency Rates of Outstanding Subprime Mortgages Actual Delinquency Rate (%) 20 FRM Hybrid 15 10 5 0 2000 2001 2002 2003 2004 Year 2005 2006 2007 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 4/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Delinquency Rates of FRM Subprime Mortgages Fixed−Rate Mortgage Loans 12 Actual Delinquency Rate (%) 2007 2006 2005 2004 2003 2002 2001 10 8 6 4 2 0 2 4 6 8 10 12 14 16 18 20 22 24 Age (Months) Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 5/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Delinquency Rates of ARM Subprime Mortgages Adjustable−Rate Mortgage Loans 20 18 16 14 12 10 8 6 4 2 0 2 4 2007 2006 2005 2004 2003 2002 2001 Actual Delinquency Rate (%) 6 8 10 12 14 16 18 20 22 24 Age (Months) Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 6/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Delinquency Rates by Purpose and Documentation Purchase−Money Mortgage Loans 20 18 16 14 12 10 8 6 4 2 0 2 4 2007 2006 2005 2004 2003 2002 2001 Cash−Out Refinancing Morgage Loans 20 18 16 14 12 10 8 6 4 2 0 2 4 2007 2006 2005 2004 2003 2002 2001 Actual Delinquency Rate (%) 6 8 10 12 14 16 18 20 22 24 Age (Months) Actual Delinquency Rate (%) 6 8 10 12 14 16 18 20 22 24 Age (Months) Full−Documentation Mortgage Loans 20 18 16 14 12 10 8 6 4 2 0 2 4 2007 2006 2005 2004 2003 2002 2001 Low−or−No−Doc Mortgage Loans 20 18 16 14 12 10 8 6 4 2 0 2 4 2007 2006 2005 2004 2003 2002 2001 Actual Delinquency Rate (%) 6 8 10 12 14 16 18 20 22 24 Age (Months) Actual Delinquency Rate (%) 6 8 10 12 14 16 18 20 22 24 Age (Months) Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 7/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Evolution of Subprime Mortgage Market Year Subprime Share (%) Securitized Share of Subprime Loans(%) 2001 7.8 54.1 2002 7.4 62.9 2003 8.4 61.1 2004 13.5 75.7 2005 21.3 76.3 2006 20.1 74.8 Source: Mortgage Market Statistical Annual, 2007 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 8/ 17 Title Introduction Empirical Analysis Boom conclusions Subprime The Crisis Market Data Data: Loanperformance Inc. & OFHEO Loanperformance: “a self-standing division of First American CoreLogic, Inc., the nations largest and most comprehensive source of real estate business information.” Contains data on about 85% of all subprime securitized mortgage loans (≈ 50% of all subprime loans) Information included: underwriting criteria, origination date, loan structure, loan status (monthly), property location. OFHEO data for the House Price Appreciation (MSA-level, quarterly) Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 9/ 17 Title Introduction Empirical Analysis Boom conclusions Factor Impact Descriptives Contribution Factor-Adjustment What “Explains” Delinquency and Foreclosure One Year After Origination Marginal Impact of Individual Factors on Probability of Delinquency and Foreclosure, % Credit Score (FICO) House Price Appreciation Full Documentation Provided Prepayment Penalty Charged If a Loan is an ARM (relative to FRM) If a Loan is a Balloon (relative to FRM) Margin for ARM and Hybrids If a Loan is a Hybrid (relative to FRM) Origination Amount Loan−to−Value Ratio (First−Lien) Missing Debt−to−Income Ratio Value of Debt−to−Income Ratio (if not missing) Initial Rate (APR) −2 Demyanyk, Van Hemert −1 0 1 2 10/ 17 Understanding the Subprime Mortgage Crisis Title Introduction Empirical Analysis Boom conclusions Factor Impact Descriptives Contribution Factor-Adjustment First-Lien Mortgages/Vintage Number of Loans (*1000) Average Loan Size (*$1000) FRM (%) ARM (%) Hybrid (%) Balloon (%) Purchase (%) Refi. (cash out) (%) Refi. (no cash out) (%) FICO Score Combined LTV (%) DTI (%) Missing DTI (%) With Full Doc. (%) With Prep. Penalty (%) Mortgage Rate (APR) (%) Margin (%) 2001 624 151 41.4 0.9 52.2 5.5 35.1 52.1 12.3 620.1 80.0 37.8 41.6 68.5 66.3 9.4 6.2 2002 974 168 39.9 1.9 55.9 2.2 33.9 51.2 14.6 630.5 79.9 38.1 44.1 63.4 63.8 8.3 6.3 2003 1676 180 43.3 1.3 54.7 0.8 32.9 51.6 15.1 641.4 80.6 38.2 38.3 59.8 61.4 7.3 5.9 2004 2743 201 28.2 4.3 67.3 0.2 42.0 47.9 10.0 645.9 82.8 38.5 35.1 57.2 60.1 6.7 5.3 2005 3440 234 25.1 10.3 62.0 2.6 45.7 45.7 8.6 653.7 83.5 39.1 39.2 51.8 60.6 6.6 5.0 2006 2646 259 26.1 12.8 46.2 14.9 45.4 44.8 9.8 654.7 84.4 39.8 31.7 44.7 61.6 7.2 4.9 11/ 17 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis Title Introduction Empirical Analysis Boom conclusions Factor Impact Descriptives Contribution Factor-Adjustment “Contribution” of Individual Factors in 2001 and 2006 to Explaining Delinquency and Foreclosure If individual factors equal their values in 2001 and 2006, what would be the change in probability of delinquency and foreclosure? Contribution 2001 (%) CLTV Loan Size Full Doc DTI Margin Cash−Out Investor ARM Hybrid Balloon Prep. Penalty Missing DTI House price FICO Mortgage Rate −.5 Graphs by vintage Contribution 2006 (%) FICO Missing DTI Hybrid Margin Prep. Penalty ARM Investor Cash−Out DTI Mortgage Rate Balloon Full Doc CLTV Loan Size House price 0 .5 1 1.5 2 −.5 0 .5 1 1.5 2 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 12/ 17 Title Introduction Empirical Analysis Boom conclusions Factor Impact Descriptives Contribution Factor-Adjustment “Adjusted” Delinquency Rates Adjusting for the actual values of individual factors for each loan reveals GRADUAL DETERIORATION. 14 12 10 8 6 4 2 0 2 4 6 8 10 12 14 16 Age (Months) 18 20 22 24 2006 2005 2004 2003 2002 2001 Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 13/ 17 Title Introduction Empirical Analysis Boom conclusions Symptoms Subprime International Evidence of Lending Booms Leading to Crises Argentina 1980 Chile 1982 Sweden, Norway, Finland 1992 Mexico 1994 Thailand, Indonesia, Korea 1997 “Lending Booms and Lending Standards” by Dell’Ariccia and Marquez in JF2006 Sizable booms lead to: lowering underwriting standards, decreasing collateral requirement decreasing (lender’s) profit margin decreasing “unobserved” borrower quality Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 14/ 17 Title Introduction Empirical Analysis Boom conclusions Symptoms Subprime Was there a “boom” in Subprime Mortgage Market? Market increased dramatically over the past 6 years “Accepted” CLTV increased “Unobserved” quality of borrowers deteriorated Price per unit of risk decreased CLASSIC LENDING BOOM SCENARIO. Historically, booms like that have often led to crises or distress. Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 15/ 17 Title Introduction Empirical Analysis Boom conclusions Symptoms Subprime Could we Have Known in Advance? Quality deterioration was a monotonic process over 2001-2006 Masked by high house price appreciation in 2004-2005 Could have been detected in 2005 with a statistical analysis Riskiness of high-LTV borrowers gradually increased Interest rates charged by lenders gradually became more sensitive to the LTV ratio... Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 16/ 17 Title Introduction Empirical Analysis Boom conclusions Concluding remarks Understanding the Subprime Mortgage Crisis: Investor’s supply-driven increase in credit availability dramatic and monotonic increase in subprime mortgage market dramatic and monotonic increase in “unobserved” riskiness of borrowers decrease in the price of risk disaster was waiting to happen... why did it happen now? No more house-appreciation “mask” Demyanyk, Van Hemert Understanding the Subprime Mortgage Crisis 17/ 17