Methodology. Rating Canadian Residential Mortgages, Home Equity Lines of Credit and Reverse Mortgages

Transcription

1 Methodology Rating Canadian Residential Mortgages, Home Equity Lines of Credit and Reverse Mortgages november 2014

2 CONTACT INFORMATION Kevin Chiang Senior Vice President CDN ABS, RMBS & CBs, Global Structured Finance Tel Jamie Feehely Managing Director CDN Structured Finance, Global Structured Finance Tel DBRS is a full-service credit rating agency established in Privately owned and operated without affiliation to any financial institution, DBRS is respected for its independent, third-party evaluations of corporate and government issues, spanning North America, Europe and Asia. DBRS s extensive coverage of securitizations and structured finance transactions solidifies our standing as a leading provider of comprehensive, in-depth credit analysis. All DBRS ratings and research are available in hard-copy format and electronically on Bloomberg and at DBRS.com, our lead delivery tool for organized, Web-based, up-to-the-minute information. We remain committed to continuously refining our expertise in the analysis of credit quality and are dedicated to maintaining objective and credible opinions within the global financial marketplace.

3 Rating Canadian Residential Mortgages, Home Equity Lines of Credit and Reverse Mortgages TABLE OF CONTENTS Scope and Limitations 5 Introduction 5 Application of Methodology 5 Loan-Level Analysis 7 Default Frequency (Probability of Default) 7 Key Factor: The Combination of Loan-to-Value (LTV) and Credit Score 7 Base-Case Mortgage and Base-Case Default Curve 7 Examples of Default Frequency Assumptions 7 Other Factors 10 Loss Severity (loss given default) 18 (1) Unpaid Principal Balance 18 (2) Accrued Interest 18 (3) Recovery on the Property 19 (4) Foreclosure Costs 22 (5) Other Recoveries 22 (6) Second-Lien Mortgages 22 Portfolio-Level Analysis 22 (1) Geographic Concentration 22 (2) Pool Size 24 Cash Flow and Structural Analysis 24 (1) Excess Spread and Cash Flow Analysis 24 (2) Other Structural Features 26 Analysis for Insured Mortgages/HELOCs and NHA-MBS 27 Legal Review 27 Notice and Perfection 27 Registration 28 Foreclosure and Recourse 29 Operational Review 29 HELOCs 30 Characteristics of a HELOC 30 Terms and Payments 30 Credit Limit 31 HELOCs in Canada and the United States 31 Transaction Structure 32 Trust Structure (Master Trust) 32 Co-Ownership Interest (Non-Discrete Purchase) 32 Seller s Interest 32 Commingling 33 3

4 Addition and Removal of Accounts 33 Revolving Period 33 Amortization Period 33 Controlled Accumulation and Controlled Amortization Periods 33 Early Amortization Period 34 Priority of Payments (Waterfalls) 34 Loan-Level and Cash Flow Analysis 34 (1) Loan-Level Analysis 34 (2) Excess Spread and Cash Flow Analysis 35 ABCP liquidity 36 Reverse Mortgages 36 Expected Occupancy Term (EOT) 37 Property Value and Home Price Appreciation (HPA) 37 Mortgage Rates 37 Stress Testing and Cash Flow Analysis 38 EOT Assumption 38 Stressed Property Value and HPA 39 Excess Spread and Cash Flow Analysis 39 Accrual of Note Interest 39 Surveillance 40 Appendix 1: Glossary 40 4

5 Scope and Limitations DBRS evaluates both qualitative and quantitative factors when assigning ratings to a Canadian structured finance transaction. This methodology represents the current DBRS approach for rating mortgage-related securitizations issued in Canada with collateral originated in Canada. It describes the DBRS approach to analysis, which includes (1) a focus on the quality of the originator/servicer, (2) evaluation of the collateral pool and (3) utilization of historically employed credit evaluation techniques. This report also outlines the asset class and discusses the methods DBRS typically employs when assessing a transaction and assigning a rating. It is important to note that the methods described herein may not be applicable in all cases. Further, this methodology is meant to provide guidance regarding the DBRS methods used in the sector and should not be interpreted with formulaic inflexibility, but understood in the context of the dynamic environment in which it is intended to be applied. Introduction DBRS s approach to rating Canadian residential mortgage-related transactions considers the key characteristics of residential mortgages and their corresponding risk metrics, together with the various structural features. In addition, as market value declines from the Canadian RMBS Model (the RMBS Model) are used for the analysis of reverse mortgages, DBRS incorporates in this methodology the approach to rating Canadian reverse mortgages, which was previously a separate appendix in the Rating Canadian Structured Finance Transactions methodology. The main body of this methodology continues to focus on the discussion of regular residential mortgage securitization, with additional discussions of home equity lines of credit (HELOCs) and reverse mortgages. The fundamental approach used for Canadian residential mortgage transactions has remained consistent since the first publication of the methodology in First, the RMBS Model assesses credit risk for an individual mortgage relative to a base-case mortgage, which is a 30-year fixed-rate mortgage for purchase of an owner-occupied single-family property underwritten to a full documentation standard. The expected loss at each rating level is then increased or decreased by distinctive risk factors (risk layering) of the individual loan. Second, after the RMBS Model sums up the expected loss for each individual mortgage in a pool through a risk-weighting formula, the RMBS Model adjusts the expected loss, if necessary, to reflect the characteristics of the pool as a whole (namely, the size and the geographic concentrations of the asset pool). Finally, DBRS uses the RMBS Model output to run the cash flow model. In addition, the quality and experience of the mortgage originator and servicer and the legal and operational aspects of the transaction are examined to determine if any further adjustments are necessary. APPLICATION OF METHODOLOGY The following diagram describes the overall process for analyzing residential mortgage- and HELOCbacked transactions: (1) DBRS conducts loan-level and portfolio-level analysis using the RMBS Model. The resulting output of the model is the expected gross credit loss of the loan. The RMBS Model is a substantial component of the DBRS rating process. (2) DBRS performs a cash flow analysis based on the output from the RMBS Model by incorporating assumptions regarding prepayment for residential mortgages or base-case principal payment rate and stress multiples for HELOCs, timing of defaults and interest rates in order to estimate the excess spread available over the life of the transaction and the appropriate credit support for each rating level. 5

6 (3) The legal and operational aspects of the transaction are also reviewed with the understanding that the adequacy of the credit enhancement available is also subject to the legal structure of the transaction and the results of an operational review. Loan-Level Analysis Canadian RMBS Model + Portfolio-Level Analysis Prepayment Assumptions Gross Credit Loss Interest Rate Assumptions Cash Flow Analysis Timing of Default Assumptions Base Case Principal Payment Rate and Stress Multiple Assumptions (HELOC) Operational Review Credit Enhancement/Ratings Legal Review Surveillance 6

7 Loan-Level Analysis DEFAULT FREQUENCY (PROBABILITY OF DEFAULT) Key Factor: The Combination of Loan-to-Value (LTV) and Credit Score The combination of LTV and credit score (the most common credit score provided in Canada is the BEACON score from Equifax Inc.) is the most important factor in determining the default risk of a residential mortgage. LTV is positively correlated with mortgage defaults, which means the higher the LTV, the greater the probability of default. Credit score, however, is negatively correlated with mortgage defaults, which means the lower the credit score, the greater the probability of default. These two variables are used in tandem within the RMBS Model. Base-Case Mortgage and Base-Case Default Curve DBRS defines the base-case mortgage as a 30-year fixed-rate mortgage for purchase purposes, underwritten to a full documentation standard and assumed by an A -grade borrower who occupies the single-family detached house as the primary residence. This base-case mortgage of 30 years is based on the U.S. market where a large amount of mortgage data is available. When the characteristics of a mortgage deviate from the base-case mortgage, the default risk associated with this mortgage changes accordingly. For example, a 25-year mortgage with all other characteristics being the same has a lower probability of default than a 30-year mortgage because if a borrower chooses a shorter amortization period, it implies that the borrower is able to afford a larger monthly mortgage payment and is confident in paying off the debt sooner. In other words, such a borrower is considered to have a stronger financial ability and willingness to pay and, therefore, is less risky in terms of credit risk. The default probability of a base-case mortgage can be estimated by the borrower s credit score in combination with the LTV of the mortgage at origination. The base-case default curves are subsequently derived from DBRS interpretation of the modified Fair Isaac Corporation (FICO) bad-rate table and the statistical study of the historical mortgage performance data in the United States. DBRS internal analysis reveals that, for the same credit scores between 510 and 680, Canadian borrowers tend to perform better, resulting in adjustments for Canadian loans with a slight credit lift embedded in the base-case default curve. As credit score decreases, the ability or willingness to pay is considered to decrease exponentially, causing the bad rate of mortgages 90-plus days in arrears or defaulted mortgages to increase at an exponential rate. The RMBS Model allows for credit scores in the range of 440 and 880. Any score above or below this range is considered statistically immaterial from the cut-off points and is replaced by the cap (880) or floor (440) score. If a credit score is not available for any given mortgage, DBRS typically assigns a score of 580 unless the historical performance or credit quality of the pool proves otherwise. Examples of Default Frequency Assumptions (1) The graph below illustrates a base-case mortgage with a BEACON score of 820 at different LTV levels for rating categories from AAA (sf) to B (sf). 7

8 820 BEACON Score by LTV 3.0% Base Default Frequency 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 60% 70% 80% 90% 100% LTV AAA (sf) AA (sf) A (sf) BBB (sf) BB (sf) B (sf) (2) The graph below shows a base-case mortgage with a BEACON score of 620 at different LTV levels for rating categories from AAA (sf) to B (sf). 620 BEACON Score by LTV Base Default Frequency 18.0% 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 60% 70% 80% 90% 100% LTV AAA (sf) AA (sf) A (sf) BBB (sf) BB (sf) B (sf) 8

9 (3) The graph below shows a base-case mortgage of 90% LTV with different BEACON scores for rating categories from AAA (sf) to B (sf). 90% LTV by BEACON Score 25.0% Base Default Frequency 20.0% 15.0% 10.0% 5.0% 0.0% BEACON Score AAA (sf) AA (sf) A (sf) BBB (sf) BB (sf) B (sf) (4) The graph below shows a base-case mortgage of 60% LTV with different BEACON scores for rating categories from AAA (sf) to B (sf). 60% LTV by BEACON Score 8.0% Base Default Frequency 7.0% 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% BEACON Score AAA (sf) AA (sf) A (sf) BBB (sf) BB (sf) B (sf) 9

10 Other Factors (1) Type of Mortgage Product A mortgage product contains various characteristics such as mortgage rate, amortization period, term to maturity and interest-only (IO) period (if applicable). The RMBS Model compares these characteristics to the base-case mortgage, determines the deviation from the base-case mortgage and any related penalty (or benefit) factor of each characteristic and multiplies these factors to obtain the overall assessment of default risk of a particular mortgage product. (a) Mortgage Rate: Fixed-rate mortgages (FRMs) are the most popular type of mortgage in Canada, while the popularity of adjustable-rate mortgages (ARMs) depends on interest rate levels. ARMs are considered riskier in the RMBS Model because the mortgage rate could potentially be reset to a higher level when the Bank of Canada rate (and consequently the index rate, usually the lender s prime rate) moves up and a higher monthly payment will be required (the payment shock). DBRS uses FRMs as the base and applies an ARM penalty factor, based on the reset frequency of the interest rate. A mortgage with interest rate reset more frequently than two years (monthly, quarterly, semi-annually, annually or biennially) is considered most risky as the borrower is subject to more frequent potential payment shocks. A mortgage with a three-year reset period is considered riskier than a five-year reset as historical experience in Canada shows that the likelihood of default is highest within three years of loan origination. With a reset to a potentially higher rate, the likelihood of default increases. In comparison, a common mortgage in Canada with a fixed five-year term or longer (i.e., mortgage rate reset at the end of five years or later) is considered virtually the same as an FRM throughout the entire amortization period. See the Timing of Defaults section in Cash Flow and Structural Analysis for more details. Frequency Adjustment for ARMs 1.60 Frequency Adjustment Factor ARM Reset in Years (b) Amortization Period: In Canada, a common amortization term is 25 years. Using 30-year amortization as the base, DBRS applies a frequency (upward) adjustment for mortgages with longer amortization terms and a frequency benefit factor for mortgages with shorter amortization terms. As discussed above, if a borrower chooses a shorter amortization (all else being equal), it implies that the borrower is likely in better financial condition or more committed to pay off the debt sooner and, therefore, the borrower is less likely to default on the loan. On the other hand, if a borrower opts for a longer amortization, which results in lower monthly payments and increases affordability, it could mean the borrower likely would not otherwise qualify under the standard amortization term and is less creditworthy. The likelihood of default is therefore assumed to be higher. Another feature of amortization is negative amortization, which allows unpaid accrued interest to be added to the principal balance of the mortgage, causing the outstanding loan balance to grow instead of being paid down over time. Mortgages with negative amortization features are considered risky since the effective interest rate is below the mortgage rate and the monthly 10

11 payments are kept artificially low to stretch the affordability. These borrowers will face payment shocks when the outstanding mortgage balance reaches a predetermined LTV limit and starts to amortize. To properly assess this risk, the frequency penalty factor for negative amortization is 1.2 times (x) within the RMBS Model. Frequency Adjustment for Full Amortization Mortgages 1.40 Amortization Frequency Factor Full Amortization in Years (c) Term to Maturity: In Canada, a mortgage usually carries a term to maturity much shorter than its amortization term, which means a mortgage is subject to a review (and renewal) at the end of the mortgage term. Such partially amortizing mortgages (or balloon mortgages) carry additional default risk as the mortgagor may not be able to arrange a new mortgage when the term of the mortgage ends and the final payment (the balloon payment) becomes due. In general, the longer the term to maturity and the closer its length to that of the amortization term, the less risky the mortgage is because a mortgage with longer term to maturity will carry a smaller balloon when the mortgage matures and the LTV at maturity is likely to be lower; therefore, it is considered easier for the borrower to obtain another mortgage at the end of the mortgage term. The RMBS Model assigns a penalty factor for partially amortizing mortgages if the term to maturity is less than five years as a five-year mortgage term is the norm and borrowers with shorter loan terms are potentially perceived as less creditworthy by the lenders to assume their credit risk. In addition, there is less time to pay down the mortgage balance for shorter term loans. Changing market conditions, lender liquidity and refinancing criteria have caused some non-conventional lenders to be unable or unwilling to refinance a mortgage at maturity, even if a borrower has been making the mortgage payments on time during the entire mortgage term. This caused some borrowers to default on the balloon payments as a result of their inability to renew the maturing mortgage, in spite of the borrowers clean payment history. On the other hand, with respect to prime mortgages, DBRS was not aware of any conventional mortgage lender failing to renew performing loans during the recent financial crisis and considers it highly unlikely that a prime borrower of conventional mortgages with a clean payment history would be unable to obtain refinancing at loan maturity. As the risk of defaulting on maturity (balloon risk) is different from the risk of defaulting on regular mortgage payments (credit risk), it is additive to the default frequency generated from the RMBS Model. The goal of balloon risk assessment is to determine a realistic probability of non-refinancing at loan maturity. Lender-specific liquidity lies at the heart of balloon risk; therefore, the higher the lender is rated, the less likely a liquidity shortage would occur and the more likely maturing mortgages could be renewed. DBRS attempts to assess the potential default concentration at the maturity of the loans by estimating the lender s liquidity (or lack thereof) at the end of the loan tenure and the percentage of loan defaults as a result of non-renewal. 11

12 For surveillance of existing transactions where the eligibility by mortgage insurers or lenders underwriting criteria may have changed since loan origination, DBRS may estimate future LTV and credit scores at loan maturity. The estimated LTV takes into account the future scheduled mortgage payments, potential prepayments and home price appreciation (HPA) since origination, based on publicly available information such as Teranet-National Bank House Price Index (Teranet Index) or Canadian Real Estate Association (CREA) data with DBRS adjustments. When a non-conventional mortgage at maturity is considered to fall outside refinancing guidelines of mortgage insurers or most lenders and a balloon risk exists, the loan is assigned a minimum default probability of 50% in the RMBS Model, based on the credit score of the borrower. Partial Amortization Frequency Adjustment Factor Partial Amortization Adjustment Mortgage Term to Maturity (Years) (d) IO Period: The IO mortgages offered in Canada are different than those offered in the United States. As such, IO mortgages in Canada begin with an IO period (typically three years or five years) and end with an obligation of full principal repayment at the end of the IO period; therefore, these mortgages are IO balloon loans, while in the United States, IO mortgages begin with an IO period and at the end of the IO period, principal amortization will occur for the remaining mortgage term. The IO feature brings an additional layer of risk to the mortgage because these mortgages do not amortize during the IO period and the borrower solely depends on the appreciation of the property to increase his/her equity by the end of the IO period to refinance or fully repay the principal. There is a risk that the property value may not increase during the IO period if there is a serious market downturn. In addition, borrowers face a payment shock when the IO period ends as the mortgage payments start to include principal repayment in addition to interest expenses. Within the RMBS Model, the penalty for the IO feature depends on the term to maturity. For example, IO mortgages with a 15-year term to maturity are considered to have a higher default probability than those with a 25-year term to maturity. This is because a shorter term to maturity means a shorter period of time for full principal amortization after the IO period ends and results in a larger principal repayment monthly, increasing the magnitude of payment shock when principal (re)payment is required. For loans with an IO period of between five years and ten years, the penalty is relatively low because the borrower has a longer period for its property to appreciate and, therefore, more opportunities during the ensuing years to sell or refinance the property. For an IO period of less than five years, the risk of payment shock increases significantly as interest rates may move upward when the IO period ends, increasing the payment obligations (both principal and interest) and the increased payment also becomes effective during the period of greater credit risk (within five years of origination). For an IO period greater than ten years, information on historical performance is limited and the increasing default frequency penalty factors reflect a conservative bias. As such, the DBRS IO adjustment curve is U-shaped. 12

13 IO Frequency Adjustment Factor 1.8 IO Frequency Adjustment Factor IO (5-year Term to Maturity) IO Period in Years IO (15-year Term to Maturity) IO (25-year Term to Maturity) (2) Loan Purpose The base-case loan purpose is to purchase. Purchase loans are the least risky of all loan purposes as no borrower is willing to overpay for a property, especially when the borrower needs to provide a down payment. Thus, the value of the property should be very close to the true market value, minimizing the risk of an inflated property price and an underestimated LTV, which would result in lower credit risk. Refinance mortgages are considered riskier than purchase mortgages, whether the refinance is a rate refinance or a cash-out refinance. Refinance/cash-out mortgages are expected to have more aggressive property appraisals than purchase mortgages as there is only appraised value without an actual sale and the LTV may be underestimated because of an inflated assessment so that a larger equity takeout can be obtained. Also, as no down payment is required for a rate refinance and only some costs are incurred for a cash-out refinance, the refinance becomes more of a financial transaction, reducing the borrower s commitment to the property and turning the home into a financing vehicle using the equity in the house. Nevertheless, there is generally no property reappraisal for a rate refinance transaction, reducing the risk of inflated property valuation. In addition, rate refinancing is not common in Canada as there is usually a substantial penalty on full prepayment, which discourages borrowers from seeking out the lowest mortgage rates even when the interest rate drops. The RMBS Model uses continuous equations based on borrowers credit scores to calculate default penalty factors for both rate and cash-out refinances. The adjustment factors for rate refinance range from 1.2x for borrowers with high credit scores to 1.5x for borrowers with weaker credit scores. This is intuitively correct because the penalty is modest for strong credits and more substantial for weaker credits as rate refinance is geared toward sophisticated borrowers. The penalty factors for cash-out refinance range from more than 2.0x for borrowers with high credit scores to 1.6x for borrowers with weak credit scores. That is, the stronger the borrower credit-wise, the larger the penalty for a cash-out refinance as prime credit borrowers are assumed to have many means of cost-effective borrowing rather than having to resort to cash-out refinance. If a prime credit borrower chooses a cash-out refinance, it is likely that the borrower is running out of other financing options to the point of being over-leveraged and, therefore, a larger penalty (up to 2.44x) is warranted than for a non-prime borrower for whom cash-out refinance is probably the only financing option available. The adjustment is relatively smaller at 1.6x for non-prime borrowers. 13

14 Frequency Adjustment Factor - Loan Purpose Frequency Factor BEACON Score Rate Refi Purpose Factor Other Purpose (Cash Out, Debt Consolidation etc.) (3) Documentation The degree of documentation verification is important in the evaluation of mortgage default risks. DBRS considers a mortgage to be one with full documentation if the mortgage meets the documentation requirements of uninsured mortgages at a traditional Canadian financial institution (e.g., a bank, credit union or trust company). Full documentation usually includes the verification by the lender of the borrower s income, assets, employment and rental/mortgage payment history. DBRS assigns a penalty factor to mortgages with reduced or low documentation since those features introduce additional credit risks into the mortgage. The less documentation verified as part of the mortgage application process, the greater the uncertainty about the borrower s financial ability and propensity to pay. Similar to loan purposes, the frequency penalty for documentation is continuous based on the borrowers credit scores. The penalties for reduced documentation increase as the borrower s credit score improves, similar to the adjustment for cash-out refinance loans. This may not be intuitively obvious: DBRS considers that prime borrowers should have no problem providing full documentation and the use of low documentation in mortgage applications implies non-standard practice, potential misrepresentation and adverse self-selection. There is a disconnection between a prime credit and low documentation. As for non-prime borrowers, the use of reduced or low documentation mortgages is considered part of their credit nature, meaning incomplete or unverifiable credit documents contribute to weak credit. The penalty, then, is lower than for prime borrowers with low documentation. If a low doc or no doc (Doc Code = 1 or 2) mortgage happens to carry other layers of risk, such as having a high LTV (over 80%), being an investment property or having junior liens, there is an additional frequency penalty of up to 3.0x on top of the documentation frequency adjustment factor. Mortgage Documentation Codes Documentation Type A full doc loan with all verifications completed in a manner that satisfies usual Schedule I bank standards. Four pieces of verification are completed: income, assets, employment and mortgage/rental history Doc Code 4 Usually three verifications were performed or one verification less than a full doc 3 A stated income, stated assets loan with verification of mortgage/rental history and employment 2 Minimum document verification 1 14

15 Frequency Adjustment for Various Documentation Standards 2.50 Frequency Factor Full Doc 3 Complete Verifications BEACON Score 2 Complete Verifications 1 Complete Verifications or None (4) Borrower Credit Grade DBRS base-case borrower credit grade is A, which means that the borrower has had no delinquencies on his or her credit card or on other personal debts in the past seven years, according to the credit bureaus in Canada. Mortgagors whose credit histories are not as good are subsequently classified as A-, B, C or even D borrowers based on their credit scores or their delinquency/default history. DBRS assigns a frequency penalty factor for mortgages without an A credit grade. DBRS may examine the underwriting standards, available performance data and the borrowers credit scores to assign a credit grade to each mortgage. As the underwriting standards vary among different mortgage originators, DBRS typically uses the more objective credit scores presented in the following table as general guidelines to determine the credit grade for mortgages. Frequency Factors Credit Grade Credit Score Frequency Factor A Greater than x A- 640 to x B 580 to x C/D Less than x (5) Occupancy As mentioned above, a base-case mortgage is secured by a property occupied by the owner as the primary residence. Non-owner-occupied properties, such as second homes and investment properties, are considered riskier since the borrower is more likely to default on a non-owner-occupied property when financial resources are constrained. For example, a second home is not a necessity to satisfy an immediate, basic need for housing and represents additional commitment (a luxury) on the part of a mortgagor. Given the choice between making the mortgage payments on a primary residence and on a second home, most borrowers would be expected to maintain their primary residences if their ability to pay is diminished. The penalty factor for second homes is 1.2x. For an investment property, the investor is dependent on rental income to cover the mortgage payments and other expenses and rental income is subject to changes in the rental market. In addition, renters are generally not as financially or emotionally committed to the property and if the investor-owner is more distant from the property s maintenance needs, an investorowned property is more likely to be poorly maintained than a primary residence or a second home. As it is by nature an investment, subject to investment return goals and reflecting a certain appetite for risk, vacancy and/or rental market risk may alter the likely investment outcome, property value and borrower s 15

16 interest. The frequency factor for investor-owned properties is 1.7x, harsher than for second homes, which typically have a strong commitment from the owners. Frequency Adjustment for Different Occupancy Types Frequency Factor Primary 2nd Home Investor (6) Property Type Detached single-family homes, the most common and preferred property type in Canada, are the basecase property type. Semi-detached houses, row houses and freehold townhouses are also considered the same as detached single-family properties. The default penalty factor increases when the property houses more than one family (e.g., a duplex) or ownership is somehow shared and an individual mortgagor is closely tied to neighbouring properties (e.g., condominiums and co-operatives). Historically, non-singlefamily properties tend to have less market liquidity than single-family properties, take longer to sell and are, therefore, considered riskier in a market downturn, although the popularity of multi-unit residential dwellings such as condominiums has increased over the past decade. Specifically, condominiums and co-operatives, even though they are intended as single-family housing, are different in the sense that a mortgagor owning a condominium or co-operative unit is tied to the neighbouring units for the financial and physical maintenance of the entire building in which the unit exists. This codependency creates additional risk, which is especially apparent in circumstances where the building association is financially troubled, even though the individual unit owners are not. To account for this increased risk, the DBRS penalty factor for both condominiums (including condominium townhouses) and co-operatives is 1.5x. Frequency Adjustment Factor for Various Property Types Property Frequency Adjustment Factor Single-family Townhouse Duplex Condo 16

17 (7) Location (a) Provinces/Territories: Each province/territory in Canada has its own distinct real estate laws and some provinces/territories are more creditor friendly than others. DBRS assigns an adjustment factor to account for the impact of different legislation. For example, in Alberta, property foreclosures occur under judicial sale so that when a lender obtains a court order for foreclosure on a property, the mortgage debt is entirely extinguished and the lender can no longer pursue the borrower for any deficiency if the foreclosure proceeds do not fully satisfy the debt. In other words, the lender has no recourse back to the borrower (except for insured mortgages). Therefore, DBRS applies a penalty factor for mortgages with high LTVs in Alberta as a distressed borrower will likely be better off surrendering the property to the lender and walking away than trying to sell the property, paying off the loan and realizing the equity (if any) in the property. The likelihood of default in this scenario is considered high. Some provinces/territories have less creditor-friendly legislation, which complicates the delinquency management process and increases the default risk. (b) Market Liquidity: According to Statistics Canada, more than half of Canadians live in a municipality with a population of 100,000 or more and more than two thirds of Canadians live in a municipality with a population of 20,000 or more. Properties located in the less populated areas are considered riskier because the liquidity of the local real estate market is limited and there is a higher dependence on one industry. To assess the risk related to the reduced market liquidity, DBRS uses population cut-offs to classify mortgages into the following categories and assigns a penalty factor to mortgages located in nonurban locations. Location Adjustment Factors Category Population Range Frequency Adjustment Factor Urban 100,000 and over 1.0x Suburban 45,000 to 99, x Tertiary 20,000 to 44, x Rural Less than 20, x (8) Loan Size Large mortgages (usually associated with expensive properties) exhibit higher default propensities because expensive properties have limited market liquidity and take longer to sell during a market downturn. DBRS classifies mortgages according to their sizes and locations and assigns adjustment factors as follows: a penalty factor of 1.15x to loans with sizes above the large loan limit and 1.1x to loans within the size limit. DBRS also reviews the concentration of large loans from a portfolio perspective and may adjust the expected loss levels if necessary. Mortgage Size Restrictions by Location Large Mortgage Limit Locations $600,000 to $1,000,000 Toronto and Vancouver (city proper) $600,000 to $800,000 Oakville, Thornhill and Unionville (Ontario) Victoria, West Vancouver, North Vancouver, Richmond and Burnaby (British Columbia) Calgary (inner city only) Montréal $400,000 to $600,000 Rest of Canada 17

18 (9) Seasoning Seasoned loans are considered less likely to default than newly originated loans with similar features because, for a seasoned loan, the borrower has already demonstrated his/her ability to pay. DBRS review of historical mortgage performance data indicates that, in general, mortgage defaults tend to be front-end loaded; therefore, there is no credit given for loans with less than two years of seasoning as these loans are not considered to be over the default peak. DBRS gives credit to loans with more than two years of seasoning (up to 60% frequency reduction). DBRS expects loans to have an updated property value at renewal, as indicated in the Office of the Superintendent of Financial Institutions Canada (OSFI) Guideline B-20, and the origination date to be reset to the renewal date, essentially considering it a new loan, in the data provided to DBRS. For a very seasoned portfolio without updated property values or renewal dates such as the portfolios of mortgage insurers, DBRS may mark to market the property value for HPA since the last appraisal date or the origination date based on Teranet Index or CREA data with DBRS HPA adjustments and prepayment assumptions to assess the appropriate LTV. If the property value or the origination date is not updated at renewal, the loan is subject to seasoning benefit since the last appraisal date, which is generally less beneficial than HPA benefit depending on price appreciation over the assessment period. (10) Delinquency Status The default frequency for loans that are over 30 days delinquent is equal to the greater of (1) the default estimate based on the credit attributes of the loan (without considering delinquency status) and (2) the default estimate indicated in the table below (regardless of rating category): Default Frequency Estimate for Delinquent Loans Delinquency Status Credit Score 680 Credit Score 740 Credit Score > 740 2nd Lien days 60% 50% 40% 60% days 75% 70% 50% 90% 90 days and over 90% 80% 60% 100% Bankruptcy 90% 90% 90% 100% Foreclosure 95% 95% 90% 100% Real Estate Owned 100% 100% 100% 100% LOSS SEVERITY (LOSS GIVEN DEFAULT) Loss severity is defined as the mortgage loss divided by the unpaid principal balance. The mortgage loss is the sum of unpaid principal balance and accrued interest, net of recovery on the property after adjustments for foreclosure costs or any other recoveries. (1) Unpaid Principal Balance Unpaid principal balance is the loan amount at the time of securitization. To be conservative, DBRS assumes an immediate default on the original securitized amount without making any assumption for the timing of default or amortization. (2) Accrued Interest The total amount of interest accrued during the mortgage arrears and subsequent property foreclosure periods depends on the length of the liquidation period and the level of the interest rate on the mortgage. The total liquidation period includes a delinquency period, a property marketing period and a foreclosure period, with the length of each period varying among the provinces/territories because of different provincial/territorial regulations. For mortgage rates used in the calculation of the carrying costs during the liquidation period, DBRS uses the maximum of the current rate, the initial rate plus 4% and the life cap minus 2% for ARMs and the current mortgage rate for FRMs. 18

19 Foreclosure Process in Canada In Canada, judicial sale and power of sale are two main ways a lender recovers mortgage debt when a borrower defaults. Judicial sale is a sale conducted under the supervision and authority of the court, where a lender must apply to the court to get the court s permission to sell the property. On the other hand, power of sale allows a lender to sell the property without the involvement of the court. The lender has the right to sell the property according to the mortgage documents and/or the provincial/territorial legislation that allows power of sale in that province/territory. Power of sale is mainly used in Ontario, Newfoundland and Labrador, New Brunswick and Prince Edward Island. Judicial sale has been adopted in British Columbia, Alberta, Saskatchewan, Manitoba, Nova Scotia and Québec. In addition to legislative guidelines, the actual length of foreclosure (either for judicial sale or power of sale) varies in each province/territory. During normal or good economic periods, it can range from days to several months, but during economic downturns, it can be stretched to up to a full year, with some smaller communities experiencing even longer time frames. The following table shows the DBRS assumptions on the liquidation period for each province/territory. Liquidation Period by Province/Territory Province/Territory Foreclosure Period (Months) Total Liquidation Period (Months) Alberta 4 14 British Columbia 8 18 Manitoba 1 11 New Brunswick 2 12 Newfoundland and Labrador 6 16 Nova Scotia 3 13 Northwest Territories Nunavut Prince Edward Island 2 12 Ontario 4 14 Québec 9 19 Saskatchewan 4 14 Yukon 4 14 (3) Recovery on the Property Market value decline (MVD) is important in determining loss severity when the mortgage defaults. Loss severity, similar to default frequency, is correlated to the borrower s credit quality. Losses are likely smaller for borrowers with better credit quality and vice versa as borrowers of good credit quality are assumed to pay more attention to and take better care of the property than those with poor credit, resulting in lower losses upon recovery. Accordingly, the borrower s credit score is used as a proxy for the condition of real estate collateral during the repossession of property and recovery process. In the RMBS Model, MVD is a continuous function of borrower credit scores by the rating categories as follows. 19

20 Market Value Decline by Rating Categories MVD % 55% 50% 45% 40% 35% 30% 25% 20% 15% 10% BEACON Score AAA (sf) AA (sf) A (sf) BBB (sf) BB (sf) B (sf) As shown above, DBRS base MVD (B rating) represents a quick-sale or distressed-sale discount on the property, with a minimal level of stress beyond the individual property s distressed state. The base MVD for borrowers with credit scores greater than 700 is approximately 20% and rises steadily to 30% as credit scores decrease. At the AAA (sf) level, the RMBS Model assumes that the collateral will lose its value at origination from approximately 38% to 51%, depending on the borrower s credit score. Such stress levels are consistent with the worst historical home price declines observed in the United States (financial crisis of , California in the early 1990s, oil bust in Texas) and certain market segments in Canada. Delinquency Status If the delinquency status of the loan is either 90 days and over, bankruptcy, foreclosure or real estate owned (REO), the RMBS Model changes the credit score of the loan to 580 or keeps the original credit score if it is lower so that the MVD calculation reflects the possible rundown condition of the property. Further Adjustments on MVD Property Type DBRS base-case property is a single-family detached home (including freehold townhouses). Multifamily buildings, such as a duplex or condominium (including condominium townhouses), are considered riskier as they appeal to a more limited buyer population and tend to exhibit more price volatility historically. As a result, they tend to take longer to sell, especially during a market downturn. Condominiums and co-operatives carry additional risks because the ownership of a common property means (1) escalating association or maintenance fees may push levered borrowers beyond their means and (2) the borrower partially depends on the neighbours in maintaining the property value. The following graph shows the penalty factors for different property types. 20

21 MVD Adjustment Factor by Property Type 1.6 MVD Adjustment Factor Single-family Townhouse Duplex Condo Relative Price Level (Dearness) Affordability and desirability are relative measures in the RMBS Model and may be perceived differently when different market conditions exist. DBRS has incorporated three levels (provincial/territorial, municipal and forward sortation area (FSA, the first three characters of a Canadian postal code) of housing price indices based on 2006/2007 resale transaction data as a benchmark to gauge the relative dearness (or cheapness) of a collateral property. Extremely expensive or cheap properties are modelled at a higher loss at foreclosure as expensive properties have a limited resale market and very cheap properties will incur a higher percentage of foreclosure costs because of some fixed foreclosing costs, regardless of the property value. Properties with a price closer to the benchmark index (i.e., cookie cutters) are assumed to experience less severe value declines thanks to their broader appeal in resale markets. The benchmark indices used in the RMBS Model were based on the market values in 2006/2007. Given the over 30% positive HPA in the national housing index since 2006/2007 to date, collateral properties for mortgages originated or renewed in recent years are more likely to be considered dear or expensive compared with the benchmark index and are therefore more likely to have less benefit or even be subject to penalty. To be conservative, DBRS has not updated the benchmark index to reflect positive HPA. Should there be a housing price correction up to 30% in the future, the benchmark index used in the RMBS Model would still be considered appropriate to measure a property collateral s dearness. MVD Adjustment Factor by Relative Dearness MVD Adjustment Factor Relative Dearness Ratio (Property Value/Property Index) 21

22 The RMBS Model incorporates benchmark indices based on average resale prices at the (i) FSA level (with 86, 21, 196 and 185 benchmark indices for the Greater Toronto Area, Greater Montréal Area, Calgary and the entire province of British Columbia, respectively); (ii) municipality level (142 benchmark indices); and (iii) provincial/territorial level (13 benchmark indices). The property value is compared with the most granular corresponding benchmark index available to calculate the dearness ratio of the property. An adjustment factor based on the dearness ratio is a continuous function implemented in the RMBS Model as illustrated in the chart above. Properties too cheap or too expensive relative to the index are penalized up to 1.65x, while property prices close to the index receive a benefit of up to 15%. (4) Foreclosure Costs There are additional costs incurred in the foreclosure and liquidation processes. These include legal fees, real estate broker fees, property taxes, hazard insurance premiums (if applicable), eviction and routine maintenance. The RMBS Model assumes foreclosure and liquidation costs are equal to 10% of the stressed property value with a minimum of $5,000. (5) Other Recoveries Aside from property sale, recoveries may occur from pledged accounts, mortgage insurance or other arrangements that provide potential cash flows to offset losses on the property value. The analysis of insured mortgages is discussed in the Analysis for Insured Mortgages section. (6) Second-Lien Mortgages As a result of the increased sensitivity of second mortgages to property value declines, the impact of additional costs, the involvement of additional stakeholders and the subordinated access to recovery proceeds, a 100% loss is assumed for second-lien mortgages plus disposal costs. This may result in a loss severity of more than 100%. In addition, the RMBS Model usually assigns a punitive default frequency close to 100% on second-lien loans because of the high combined LTV of the first- and second-lien loans and the adverse selection nature of second-lien loans (i.e., over-leverage). Depending on the combined LTV limit (for example, up to 65% or 80% with a second-lien HELOC), DBRS may adjust the default frequency downward to address the benefit of at least 20% equity in the property. Portfolio-Level Analysis (1) Geographic Concentration Geographic concentration increases the risk of a mortgage pool as the concentration increases the dependence on local economies and reduces the benefits of diversification compared with a geographically diversified pool. A geographically concentrated pool is also more likely to be subject to other shocks such as environmental issues and natural disasters, which can depress housing prices in the affected areas. DBRS assumes that the base-case pool is geographically diversified according to the actual population distribution in Canada. Expected loss for a pool with substantial geographic concentration is adjusted based on the following two factors: province/territory and city. (a) Province/Territory If the concentration of one province/territory (in terms of mortgage values) is more than 2.0x that of the province/territory s actual population distribution, the excess portion is subject to a 20% penalty. The threshold of 2.0x is considered sufficient to account for property value differences in each province/territory across Canada. The following table shows the population distribution of each province/territory and the maximum concentration without penalty. For example, a pool can contain almost 80% of loans in Ontario without being penalized for any provincial/territorial-level concentration. 22

23 Concentration Levels by Province/Territory Province/Territory Population Percentage 1 Maximum Concentration without Penalty Ontario 38.9% 77.7% Québec 23.4% 46.9% British Columbia 13.2% 26.4% Alberta 10.4% 20.9% Manitoba 3.6% 7.2% Saskatchewan 3.0% 6.0% Nova Scotia 2.9% 5.7% New Brunswick 2.3% 4.6% Newfoundland and Labrador 1.6% 3.1% Prince Edward Island 0.4% 0.8% Northwest Territories 0.1% 0.3% Yukon 0.1% 0.2% Nunavut 0.1% 0.2% 1. According to a Statistics Canada 2006 survey. The three largest provinces/territories in the pool (in terms of mortgage values) are evaluated according to the maximum concentration allowed listed above and the penalty factor is the sum of the three provincial/territorial concentration factors, if any. A pool with 60% of its loan balances in Québec results in a 2.64% relative increase in the expected loss figure, assuming the other two largest provinces/territories do not incur concentration penalties. If the original expected loss is 15% prior to the provincial/territorial concentration assessment, the final expected loss after such adjustment is 15.4% (15% * (1+2.64%)). Sample Provincial/Territorial Population Concentration Calculation A B C D E Provincial/territorial % in the Pool Provincial/territorial population % in Canada 60% 23.4% 2.0 * 23.4% = 46.8% Threshold (2.0 x B) Excess (A C) Provincial/Territorial Concentration Factor 60% % = 13.2% 13.2% * 20% = 2.64% (b) City If the concentration of a city (in terms of mortgage values) is greater than 2.5x the city s population distribution, the excess portion is subject to a 20% penalty. Similarly, the threshold of 2.5x is considered sufficient to account for the property value differences in municipalities across Canada. The ten largest cities in the pool (in terms of mortgage values) are evaluated and the total city concentration adjustment factor is the sum of the ten city concentration factors, if any. Consider a pool with 40% of its loans in Toronto. Such concentration in Toronto results in a relative 4.05% increase in the expected loss, assuming the other nine largest cities do not incur concentration penalties. If the level of expected loss is 15% prior to the city concentration assessment, the final level after such adjustment is 15.6% (15% * ( %)). Sample City Population Concentration Calculation A B C D E City % in the Pool City Population % in Canada Threshold (2.5 x B) Excess (A C) City Concentration Factor 40% 7.9% 2.5 * 7.9% = 19.75% 40% % = 20.25% 20.25% * 20% = 4.05% 23