MORTGAGE BACKED SECURITIES AN ACTUARIAL APPROACH TO CASH FLOW ANALYSIS Kyle S. Mrotek, FCAS, MAAA Neal Dihora, ASA, CFA CAS Spring Meeting 1
Disclaimer This presentation contains our views and these views are not necessarily identical to the views of the cosponsors of the program nor the employers or clients of the speakers 2
Agenda Background on the MBS market Current situation Actuarial model presentation 3
Background Gross Issuance Agency vs. Non-Agency Issuance Split by non-agency type (prime, subprime, alt-a) 4
Gross Issuance $3,000,000 $2,500,000 Gross MBS Issuance ($ millions) Agency Total Non Agency Total MBS $2,000,000 $1,500,000 $1,000,000 $500,000 $0 5 Source:Inside MBS & ABS and UBS
Agency vs. Non-Agency MBS Market Share 100% 90% 80% Agency Total Non Agency 70% Percent 60% 50% 40% 30% 20% 10% 0% Source:Inside MBS & ABS and UBS 6
Non-Agency by Type Non Agency Gross MBS Issuance ($ millions) 500,000 450,000 400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000 0 Alt A Jumbo Subprime Other Source:Inside MBS & ABS and UBS 7
Non-Agency by Type Non Agency (% of Total MBS Issuance) Percent 25% 20% 15% 10% Alt A Jumbo Subprime Other 5% 0% Source:Inside MBS & ABS and UBS 8
Current Situation What happened? Liquidity evaporated Market values eroded Why is valuation needed? GAAP Accounting regulations still require a value (FAS 157) Risk quantification Distribution of assumptions and valuations 9
Liquidity Evaporated Broker/Dealers of non-agency MBS unwilling to provide liquidity 1 Forced liquidations of MBS set market prices 1 Pricing vendors find it difficult to obtain real prices Bid - Ask spread is 10-30 points depending on collateral and the depth of distress 2 1 AD&Co's 16th Annual Conference: The Times They Are A-Changin 2 Getting Out of the Mess by Dave Hurt at the Loan Performance Symposium March 11, 2009 10
Liquidity Evaporated 300 280 260 240 220 200 180 160 140 120 100 1/9/04 3/9/04 5/9/04 7/9/04 9/9/04 11/9/04 1/9/05 3/9/05 5/9/05 7/9/05 9/9/05 11/9/05 1/9/06 3/9/06 5/9/06 7/9/06 9/9/06 11/9/06 1/9/07 3/9/07 5/9/07 7/9/07 9/9/07 11/9/07 1/9/08 3/9/08 5/9/08 7/9/08 9/9/08 11/9/08 1/9/09 3/9/09 Mortgage Spread (Conventional Mortgage Loan less 10 year Treasury) Source: Federal Reserve Board Mortgage Spread 11
Erosion of Market Values 225 Real Home Price Index (1890 2008) 200 175 150 125 100 75 50 1890 1910 1930 1950 1970 1990 2010 Source: http://www.econ.yale.edu/~shiller/data.htm 12
Erosion of Market Values 220 200 180 160 140 120 100 Jan 00 Apr 00 Jul 00 Oct 00 Jan 01 Apr 01 Jul 01 Oct 01 Jan 02 Apr 02 Jul 02 Oct 02 Jan 03 Apr 03 Jul 03 Oct 03 Jan 04 Apr 04 Jul 04 Oct 04 Jan 05 Apr 05 Jul 05 Oct 05 Jan 06 Apr 06 Jul 06 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Jan 08 Apr 08 Jul 08 Oct 08 Jan 09 Case Shiller Home Price Index Since January 2000 Jul06: 206.5 Source: Standard and Poor's 'Case Shiller 20 City Compsite' 13 Decline: 29% Jan 09: 146.4
Erosion of Market Values 80 ABX HE AAA 2007 2 Index 70 60 50 Price 40 30 20 10 0 Source: Bloomberg 14
Erosion of Market Values ABX HE AAA 2007-2 Index Components ACE Securities Corp. Home Equity Loan Trust, Series 2007-HE4 Bear Stearns Asset Backed Securities I Trust 2007-HE3 Citigroup Mortgage Loan Trust 2007-AMC2 CWABS Asset-Backed Certificates Trust 2007-1 First Franklin Mortgage Loan Trust, Series 2007-FF1 GSAMP Trust 2007-NC1 Home Equity Asset Trust 2007-2 HSI Asset Securitization Corporation Trust 2007-NC1 J.P. MORGAN MORTGAGE ACQUISITION TRUST 2007-CH3 Merrill Lynch First Franklin Mortgage Loan Trust, Series 2007-2 MERRILL LYNCH MORTGAGE INVESTORS TRUST, SERIES 2007-MLN1 Morgan Stanley ABS Capital I Inc. Trust 2007-NC3 Nomura Home Equity Loan, Inc., Home Equity Loan Trust Series 2007-2 NovaStar Mortgage Funding Trust, Series 2007-2 OPTION ONE MORTGAGE LOAN TRUST 2007-5 RASC Series 2007-KS2 Trust Securitized Asset Backed Receivables LLC Trust 2007-BR4 Structured Asset Securities Corporation Mortgage Loan Trust 2007-BC1 SOUNDVIEW HOME LOAN TRUST 2007-OPT1 WaMu Asset-Backed Certificates WaMu Series 2007-HE2 15 Source: markit.com 3/16/09
GAAP Valuation Still Needed Mark to Market FAS 157 required companies to value holdings Level 1 based on market price Recent observed prices could be due to forced liquidation Level 2 based on related price (ex. spread to treasuries) Spreads can reflect lots of different risks (credit, liquidity, ) Level 3 based on model price Mark to Model pricing developed from loan level data FASB relaxation of mark-to-market rules Perhaps an intrinsic value based on full range of scenarios 16
Risk Quantification The following table has daily percent changes of DJIA under a Normal Distribution assumption and reality Percent Move (1916-2003) Normal Distribution Assumption Reality <>3.4% 58 1001 <>4.5% 6 366 <>7% 1 in 300,000 years 48 17 Source: Benoit Mandelbrot, Economist 1/24/2009
MBS Valuation Flowchart 18
Model Framework Purpose: to model the prepayment and loss rate assumptions to be used in a cash flow engine Prepayment Model Willingness Ability Loss Model Ultimate loss rate development methods Frequency of foreclosure Severity of foreclosure Cash Flow Engine Assigns collateral cash flows to security structure based on triggers Triggers include prepayments, delinquencies and loss rates 19
Model Characteristics Transparent Actuarial Standards of Practice Model documentation Credit Focus Utilize loan level experience Loan Performance or other sources Macro assumptions such as default rates, home price changes 20
Prepayment Model Goal: estimate percentage of loan amounts that will prepay Willingness Interest rate differential (refinancing, cash-out) Loan/Product type Fixed/Adjustable rate Seasonality Ability Home price changes FICO scores LTV original and current Lending standards/policies Federal government initiatives 21
Ultimate Loss Rate Development Methods Goal: estimate percentage of loan amounts that will default and severity of default Paid Loss Development Factor (LDF) Incurred LDF A priori ultimate loss rate (ULR) development Adjusted paid BF method Incurred BF 22
Ultimate Loss Rate Paid LDF Paid losses to date Can calculate from loan level data Providers such as Bloomberg also provide this data Receive data from trustees/servicers of loans Cumulative loss curve by age of loan Examples on next slide What % of the losses should we expect to see at a certain loan age Ultimate loss = paid losses / % expected to be paid 23
Ultimate Loss Rate Paid LDF 100% Illustrative Loss Curves Moody's and Fitch 90% 80% 70% Percent 60% 50% 40% 30% 20% 10% 0% 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99 102 105 108 111 114 117 120 Moody's Alt A FRM/ARM First Lien Fitch Prime/Alt A Fitch Subprime Moody's Subprime FRM First Lien Age (months) 24
Ultimate Loss Rate Incurred LDF 25 Paid losses to date Take current delinquencies to ultimate loss Roll rate projections (project the % of delinquencies that default) Severity (% of loan that is not recoverable) Incurred losses = paid losses + estimate of defaults x severity Utilize incurred loss curves to calculate ultimate loss rate Challenges/pitfalls
A Priori ULR Development Frequency of foreclosure Severity given default Unadjusted a priori ultimate loss rate = frequency x severity Critical considerations for loan level collateral Underwriting characteristics (FICO, LTV, documentation, etc.) Economic conditions the loan is exposed to 26
A Priori Development - Frequency Frequency of Foreclosure Historical data Specific loan characteristics FICO LTV Amortization type (fixed, adjustable rate) Interest only Loan purpose (refinance, purchase) Property type (single family, condo) Occupancy (owner, second home, investor) Loan documentation (full, low, none) Loan size (jumbo, conforming) Future foreclosure estimates Take delinquencies to ultimate loss Economic variables (e.g., home price changes - see chart on slide 32) 27
A Priori Development - Frequency Illustrative Loan Characteristics Amortization Loan Size FICO LTV Prime Alt A Subprime Documentation Interest Only Occupancy Loan Purpose Property Type 28
A Priori Development - Frequency 29 Source: Negative equity and foreclosure: Theory and evidence, Christopher L. Foote, Kristopher Gerardi, Paul S. Willen, May Journal 5, of 2009 Urban Economics 64 (2008), pp. 234 345
A Priori Development - Severity Severity of Default Home price changes Costs of foreclosure (disposal, realtor, legal, upkeep) Accrued interest Current economic situation Home price depreciation results in higher severity Government intervention may impact severity Bankruptcy law changes FHA refinancing Public/private partnerships Interest claw back from 38% to 31% debt to income Others 30
A Priori Development - Severity 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% Illustrative Loan Level Severity Distribution 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% Loan Level Severity 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 31
Ultimate Loss Rate Adjusted Paid BF Paid losses to date A priori persistency adjustment Actual persistency = unpaid balance / original balance A priori persistency = anticipated unpaid balance Adjustment needed to allow for more/less losses based on actual vs. anticipated exposure duration Adjust a priori ultimate loss (frequency x severity) by persistency factor Use loss curve to estimate % yet to be paid 32
Ultimate Loss Rate Incurred BF Utilize incurred loss curve Take a priori ultimate loss rate (from a priori development) Utilize incurred loss curves to estimate % yet to be paid Incurred BF ultimate loss = incurred to date + estimate of yet to be incurred 33
Cash Flow Waterfall Tranche level cash flows based on deal prospectus Model needs to take into account specifics of the deal 34
Cash Flow Waterfall Illustrative NPV of Cash Flow Waterfall Output Net Present Value (NPV) RMBS Tranche Original Rating Scenario 1 Scenario 2 Scenario 3 A AAA 99.71 99.66 99.70 B AAA 77.63 78.52 69.03 C AA 79.09 7.81 1.64 D AA 78.64 9.96 1.66 E A 80.16 2.79 0.70 F BBB 86.83 0.64 0.39 G BBB 85.62 0.49 0.39 H BB 0.94 0.40 0.39 I BB 0.78 0.40 0.39 J Not Rated 5.46 5.34 0.39 K Not Rated 0.40 0.40 0.39 35
MORTGAGE BACKED SECURITIES AN ACTUARIAL APPROACH TO CASH FLOW ANALYSIS Questions? Kyle.Mrotek@Milliman.com Neal.Dihora@Milliman.com 36