Sector-Specific Criteria Report Residential Mortgage / Asia-Pacific Inside This Report Scope 1 Key Rating Drivers 1 Data Adequacy 2 Representations, Warranties and File Review 3 Data Sources and Models Used to Derive Criteria Assumptions 3 Rating Approach 4 1. Asset Analysis 5 2. Cash Flow Analysis 23 3. Operational Risk 29 4. Counterparty Risk 31 5. Legal Analysis 32 Rating Sensitivity Analysis 32 Surveillance 34 Limitations 36 Appendix 1: Related Criteria a 37 Appendix 2: Revolving Periods 38 Appendix 3: Loan Level Gross Loss Calculation an Example 42 Country-Specific Criteria Addenda 44 Australian RMBS Model Assumptions 45 New Zealand RMBS Model Assumptions 53 This report updates and replaces Fitch s APAC Residential Mortgage Criteria and the Australia and New Zealand Addenda, all dated 23 June 2014. Related Criteria Appendix 1 Analysts David Carroll +612 8256 0333 david.carroll@fitchratings.com Natasha Vojvodic +612 8256 0350 natasha.vojvodic@fitchratings.com Helen Wong +852 2263 9934 helen.wong@fitchratings.com Atsushi Kuroda +813 3288 2692 atsushi.kuroda@fitchratings.com Scope This report outlines Fitch Ratings framework for analysing credit risks inherent in residential mortgage-backed securities (RMBS) transactions, and collateral within covered bonds supported by residential mortgages in Asia-Pacific (APAC). The criteria provide the basis of Fitch s rating approach, both when Fitch is assigning a rating and when monitoring existing ratings. The approach includes an analysis of risk relating to the underlying assets, the structure of the issuer from both a financial and a legal standpoint and operational risk, in particular servicing and trust management arrangements. Related criteria are listed in Appendix1. Key Rating Drivers Asset Performance: RMBS portfolios are exposed to performance variations within the residential mortgage portfolios. Fitch believes the key performance parameters are: (i) the frequency of foreclosure rates, which are driven by the loan/value ratio (LVR or LTV), debt-toincome (DTI) ratio, and adverse borrower, loan and product characteristics; and (ii) the loss severity rates, which are driven by house price declines, the degree of liquidity in the residential property market, residential foreclosure costs and the timing of recoveries. Cash Flow Analysis: The main purpose of the cash flow analysis is to test the ability of the issuer to meet its obligations under the transaction documents within the context of the transaction s financial structure under various stress scenarios. The key drivers of Fitch s cash flow analysis are: the asset foreclosure frequency and recovery rate, the timing of defaults and receipt of recoveries, prepayment rates, the interest rate and currency hedging structures and interest rates, transaction and servicing fees; and the transaction s structural features. Operational Risk: The quality of the underwriting and servicing of the residential mortgagerelated assets that are securitised is key when evaluating and rating a securitisation of a portfolio of residential mortgage loans. In Fitch s experience, there is a direct relationship between the quality of the origination and servicing process and the performance of the collateral portfolio. As part of the rating analysis of each transaction, Fitch analysts complete an originator and servicer review to assess the quality of the origination and servicing procedures. Counterparty Risk: RMBS transactions are expected to be sufficiently isolated from the insolvency risk of the transactions counterparties. Fitch expects RMBS structures to include structural mitigating factors that minimise counterparty risk, in line with its Counterparty Criteria for Structured Finance and Covered Bonds, dated 14 May 2014. Legal Analysis: The transaction structure is intended to de-link the performance of the issued notes or other securities from the credit quality of the originator and from the insolvency risks of the originator and the other parties to the transaction. Fitch may adjust its rating analysis as appropriate to reflect the added risk for any part of Fitch s analysis related to legal aspects of the transaction not confirmed to the agency s satisfaction in a supporting legal opinion. This criteria report provides an overarching framework for the analysis of APAC RMBS transactions; detailed country-specific criteria assumptions are highlighted in the Country Addenda section later in this report. Performance Analytics Alison Ho +852 2263 9937 alison.ho@fitchratings.com www.fitchratings.com 23
Data Adequacy Fitch uses available market data and the historical performance data provided by the originator and servicer to perform its credit analysis and to form a view on the future performance that could be expected on a particular APAC RMBS transaction. Fitch expects to receive loan-by-loan collateral information and historical mortgage performance information from lenders to assess the credit quality of a mortgage-loan portfolio Fitch expects to receive loan-by-loan collateral information to enable it to assess the credit quality of a mortgage loan portfolio when rating a new transaction. Although fundamentally uniform across APAC, the collateral information may vary depending on the mortgage loans country of origin. These variations are detailed in the country-specific addenda later in this report. In accordance with its ratings methodology, Fitch expects portfolio data to be supported through the provision of an agreed-upon procedures (AUP) report. AUP reports are prepared by auditing firms typically selected by the arranger or originator, and assess the error rate in loan-level collateral data when checked against information in the individual loan files of the originator. Alternatively, in the absence of the AUP report, the agency will perform a file review in order to assess the general reliability of the data provided. In addition, Fitch expects to receive the following historical mortgage performance information from lenders as part of its rating process: loan-level or aggregate static arrears data and/or default data by origination vintage, eg by arrears bucket and/or default definition (90+ days, 180+ days etc.) depending on the servicing practices in the relevant jurisdiction and by distinctive sub-segment specific to the jurisdiction, eg subsidised loans; prepayment data, eg dynamic voluntary prepayments; and loss or recovery data, eg loan or aggregate level information on foreclosed properties. For purposes of rating new transactions, the agency will use historical performance information from the lender s whole portfolio and, if relevant, recent transactions, to benchmark the lender s performance and underwriting quality with the results from Fitch s default model. Please also refer to the section titled Portfolio and Data Quality in the report titled Criteria for Rating Caps and Limitations in Global Structured Finance Transactions, dated 28 May 2014. The agency expects data to be at a sufficiently granular level to enable it to reach specific conclusions regarding the portfolio that is being analysed. Fitch assesses the robustness and accuracy of the available data before including it in its analysis. Fitch considers, among other factors, whether the historical data are deemed sufficient for a proper comparison with the default model results. Where data do not meet expected standards, Fitch may consider proxy or supplementary information in order to derive its assumptions in a conservative manner. In issuing and maintaining its ratings, Fitch utilises the accuracy of the factual information it receives from the transaction parties and from other sources Fitch considers to be credible, such as the originator s external accountants. Fitch conducts a reasonable investigation of the factual information relied upon by it in accordance with its ratings methodology, and seeks reasonable verification of that information from independent sources (such as an AUP) to the extent such sources are available for a given security or in a given jurisdiction. Fitch will assess its model-derived foreclosure frequency assumptions and its loss severityrelated assumptions using the available data. The agency may apply, in conjunction with lender and servicer adjustments, conservative adjustments to account for any insufficiencies in the quality or quantity of the data which have been provided, eg lack of loan-by-loan collateral information on specific product types, or performance which deviates from the assumptions on 2
which the agency s default model is calibrated. In addition, Fitch may apply a rating cap or decide not to assign a rating if it has not received adequate data. Fitch s cash flow modelling assumptions are based on available historical and market data. As the main purpose of Fitch s cash flow analysis is to test the ability of the structure to meet its obligations under various stress scenarios, some of the assumptions represent extreme conditions which go beyond the historically observed data. Following the closing of a transaction, Fitch expects to receive regular investor/servicer reports which should include both asset portfolio information and liability information including payment distribution reports, note balances and account balances. In addition loan-by-loan level data should be provided not less than once each quarter for all transactions, regardless of the timing of payments made on the notes in the transaction. The data to be provided should include loanby-loan loss and recovery information for loans repossessed post-closing. If the transaction deviates from Fitch s initial expectations the agency will request that the issuer begin sending loan-by-loan level data more regularly. Fitch may use any data received for future criteria development. Representations, Warranties and File Review APAC RMBS transactions typically feature a country-specific standard set of representations and warranties (R&Ws) with regards to the mortgage collateral provided by the seller. Fitch s analysis is focused on assessing the risk of R&W breaches. Fitch expects, an extensive performance history, a solid track R&W record, or structural protection. Effectively aligned incentives between originator and investor interests through retention arrangements can be viewed as an additional mitigant to the potential for R&W breaches. Fitch conducts checks on the origination quality for its standard originator reviews and reasonable investigation analysis, including file reviews and portfolio-level checks. For the file reviews, Fitch at a minimum selects files for 10 mortgage loans, but may choose to request and review files for more cases where deemed necessary. Alternatively or on a supplementary basis, Fitch may expect a more detailed third-party agreed-upon procedures report. Data Sources and Models Used to Derive Criteria Assumptions In its asset analysis, Fitch combines data with its knowledge and experience in developing analytical assumptions as inputs into its loan-level default model. The agency typically derives its criteria assumptions to evaluate portfolios of residential mortgages within the APAC region by taking into consideration information from several sources. Fitch uses performance and recovery data from individual lenders and macroeconomic information on the countries to derive its criteria assumptions Fitch uses the following sets of information to derive foreclosure frequency assumptions and adjustments to the base foreclosure frequency on a standard loan: aggregate and/or loan-level issuer/lender information on delinquencies, defaults/foreclosures/arrears and losses for Fitch-rated RMBS transactions in the relevant jurisdiction; aggregate static portfolio performance information by origination vintage from lenders and from Fitch-rated RMBS transactions in the relevant jurisdiction; historical national loan default/arrears and foreclosure statistics; other economic information reported by government statistics offices and central banks and similar institutions, eg GDP growth, unemployment trends and interest rates; and other research studies. Fitch s loss severity assumptions, including market value declines (MVDs), foreclosure timing and foreclosure cost assumptions are derived using the following information sources: 3
loan-by-loan foreclosure data from the lender and from Fitch-rated RMBS transactions in the relevant jurisdiction reporting recovery and loss data; historical national house price indices; other economic information reported by government statistics offices and central banks and similar institutions, eg household growth trends, interest rates and construction activity that may affect demand and supply for housing; other research studies; and discussions with arrangers, lawyers and other third parties. Fitch assesses the general reliability of the available data and conducts a reasonable investigation into the factual information relied upon by it before including the information in its analysis, as described in the Data Adequacy section above. The foreclosure frequency and loss severity criteria assumptions are derived using a combination of different modelling techniques which may include: (i) logistic regressions; (ii) econometric modelling; (iii) time series analysis; and (iv) standard descriptive statistics. The choice of methodologies is dictated by, among other factors, the appropriateness of different methods and the extent to which robust modelling can be undertaken given the nature of the available data. Specific techniques undertaken for each country are highlighted in jurisdictionspecific criteria addendum reports. Fitch uses control procedures in each part of the model development and maintenance process in order to reduce model risk. Rating Approach Fitch s rating process begins with a review of the residential mortgage portfolio and of the historical mortgage performance history of the lender. As part of its initial analysis, Fitch will apply a filtering process to determine the key risks associated with the portfolio and the structure; this is particularly relevant for transactions with atypical risks such as very high LVR loans. The key areas reviewed as part of the filtering process include: the transaction s background, eg the purpose of the transaction; the originator of the assets, the type of assets and the portfolio performance data provided by the arranger and originator; the indicative financial structure; the counterparty risks in the proposed transaction; and the specific criteria to be used in the rating process. The originator and servicer review is an integral part of Fitch s rating process. Fitch focuses on understanding the originator s underwriting criteria and procedures, the servicer s credit risk management policies and its ability to manage its portfolio. Fitch then undertakes a loan-by-loan analysis of the portfolio and an analysis of the lender s historical performance. As discussed in more detail in the Cash Flow Analysis section below, Fitch performs a cash flow analysis to test the ability of the assets and the structure to withstand various rating stress scenarios. Although the cash flow model output is an important factor in determining the final ratings, ratings are ultimately assigned by a Fitch rating committee, based on both quantitative and qualitative factors. 4
Transaction documentation is reviewed to understand whether the documents reflect the transaction structure as presented to Fitch. Legal opinions are reviewed to confirm that the assumptions which Fitch factors into its analysis are supported by the transaction legal opinions (as described below in the section Legal Analysis). A diagrammatic representation of Fitch s rating approach is shown in Figure 1. Figure 1 Asset Analysis Portfolio Original Loan-to-Value Debt-to-Income Borrower Specifics Loan Specifics Indexing Property Value Quick Sale Adjustment Market Value Decline Illiquid Property Adjustment Foreclosure Frequency Recovery Rate Recovery Time Default Model Origination & Servicer Review Gross Credit Enhancement = Default Probability X (1-Recovery Rate) Cash Flow Analysis Foreclosure Frequency Recovery Rate Recovery Time Portfolio Amortisation Cash Flow Model New Credit Enhancement Structural Features and Stresses Waterfall (Use of Excess Spread) Legal Analysis Fitch may publish transaction-specific presale and new issue reports when assigning expected and final ratings. Gross Loss The gross loss is a measure of the expected loss on each loan for a particular rating level, each of which corresponds to a particular stress scenario 1. Asset Analysis Fitch uses a default model to calculate the estimated gross loss for each loan in the portfolio as the product of its foreclosure frequency and loss severity. The gross loss is a measure of the expected loss on each loan for the associated rating level, each corresponding to a particular stress scenario. By way of illustration, if a loan under a particular stress scenario had a foreclosure frequency of 20% and, upon foreclosure, a loss severity of 30% of the cut-off 5
balance, or current principal balance, the gross loss rate expected for that stress scenario would be 20% x 30% = 6% of the current principal balance. After calculating loan-by-loan figures, Fitch aggregates them to derive the portfolio figures. A step-by-step example of the estimated gross loss calculation is shown starting in Example 1 on page 9 and in aggregate in Appendix 3. If the portfolio comprises flexible mortgage loans, where the borrower can re-draw the loan amount already paid down or draw up to a predefined limit, Fitch uses the maximum amount that the borrower can redraw over time to determine loss severity. The default model calculates loan-level gross loss, a product of the foreclosure frequency and loss severity Fitch also calculates a recovery rate, being the recovery of principal and interest as a percentage of the current principal balance upon foreclosure of the loan. The recovery rate is calculated on a loan-by-loan basis and aggregated using the foreclosure frequency as the weight at the borrower level to arrive at the weighted-average recovery rate (WARR). While loss severity treats accrued interest as a loss in addition to the current loan balance, the recovery rate calculation assumes the lender will also recover this accrued interest. The weighted-average foreclosure frequency (WAFF) and WARR are the key outputs from the asset analysis that are used when Fitch models the cash flows of a transaction. The formulae for calculation of loan-by-loan foreclosure frequency, loss severity and recovery rate and an illustrative example are set out in the Loss Severity section below. Rating Scenario Multipliers and Forward-Looking Ratings Fitch uses a forward-looking approach in its rating criteria. In structured finance transactions, this means that lower rating levels can be expected to be more responsive to changes in the economic cycle due to their relatively greater proximity to potential default. In contrast, higher rated notes are expected to experience a higher degree of rating stability and lower rate of default, except in very severe stresses; their ratings are expected to remain stable through the cycle to a much greater degree. Assumptions at lower rating scenarios take into account immediate macroeconomic expectations Assumptions at high rating scenarios are derived to represent more remote and stressed conditions Base frequency of foreclosure is defined by: willingness to pay; and ability to pay Establishing Expected Foreclosure Frequency and House Price Decline Assumptions At lower rating levels, Fitch derives an expected frequency of foreclosure that considers expected economic and housing market dynamics. During benign periods, the expected foreclosure frequency may be floored at the through-the-cycle long-term average. With respect to loss severity calculations, Fitch s house price expectation is used to reflect the immediate expectations for a fall in value of a repossessed property and feeds into the market value decline (MVD) assumptions. House price decline (HPD) expectations are similarly adjusted based on Fitch s housing market expectations determined from an analysis of price trends and market liquidity. As such, MVDs may be dynamically adjusted based on these expectations as the house price index moves up and down. Establishing AAAsf Foreclosure Frequency and House Price Decline Assumptions At the higher rating scenarios, the frequency of foreclosure, loss severity and gross loss, are anticipated to remain relatively stable over time. AAAsf foreclosure frequency and HPD assumptions are constructed from an analysis of historical peak-to-trough performance data in extreme stress conditions. Where the AAAsf scenario continues to remain remote, any revision of the expected scenario is expected to lead to revisions in the levels of the lower rating scenarios. In other words, AAAsf assumptions are formed such that the gross loss rates are expected to remain fairly stable for the AAAsf rating level while, for lower rating levels, the assumptions may vary depending on the position in the economic cycle. To avoid undue compression between the remote and expected scenarios, Fitch ensures a minimum is maintained between AAAsf and lower rating categories. The level of compression that is acceptable between AAAsf and lower rating categories may vary on a jurisdiction- 6
specific basis and will depend on, among other factors, the strength of the sovereign and nature of the legal environment. More detail regarding general scaling of assumptions between rating categories can be found in Frequency of Foreclosure, and Loss Severity below. Specific assumptions for each jurisdiction can be found in the individual Country-Specific Criteria Addendum section later in this report. Frequency of Foreclosure The base frequency of foreclosure reflects the risk of a standard mortgage loan defaulting in a particular jurisdiction. A standard mortgage loan, while defined differently across APAC jurisdictions, is intended to reflect typical borrower characteristics, eg LVR and borrower financial strength, usually measured through DTI, and product characteristics, eg interest-rate type, payment frequency, that are prevalent in the market. Based on evidence from its data analysis across various jurisdictions, Fitch believes the primary indicator of a borrower s propensity to default is a combination of their ability to pay and willingness to pay. Fitch s APAC RMBS methodology assigns each standard borrower a base foreclosure frequency that is derived from a matrix defined by these key characteristics: the borrower s financial strength, usually represented by the DTI, which reflects a borrower s ability to pay, and the LVR, which is expected to be correlated with a borrower s willingness to pay. Ability to Pay A borrower s ability to meet its periodic mortgage payment is dictated by the relationship between its income and monthly debt service. The perceived affordability of a loan for a borrower can be measured by the DTI, which represents the ratio of the debt burden to borrower income. Empirical data have shown that borrowers with lower DTIs at origination have a greater ability to withstand financial shocks caused by divorce, unemployment and interestrate rises, which tend to be the primary reasons for borrowers defaulting on their mortgage payments in a normal economic cycle. In APAC jurisdictions, where information on income and debt burden (interest and principal) are available, the DTI is calculated by Fitch. The DTI is generally calculated as the ratio of mortgage debt service payments, either including or excluding other debt the borrower may have, depending on the jurisdiction, to gross (or net) income, depending on the underwriting practices in the jurisdiction. Where the DTI is calculated by Fitch, it uses the contractual interest rate on fixed-rate loans to calculate debt service payments. For floating-rate loans, in most countries Fitch uses an equilibrium interest rate of the corresponding currency, plus a margin, to ensure a stable DTI is calculated. This can: either correspond to a currency s standard reference interbank offer rates; or be a country-specific long-term equilibrium floating-rate mortgage rate calculated by Fitch (please refer to Criteria for Interest Rate Stresses in Structured Finance Transactions and Covered Bonds, dated 19 December 2014, for more details on how this equilibrium interest rate is derived). In some jurisdictions, the DTI is obtained directly from lenders who employ various estimations, such as the inclusion of debt service with additional liabilities and/or income that may be adjusted to reflect additional earnings and regular expenses. If loan-by-loan DTI is not available which is the case for some APAC jurisdictions then the assessment of the borrower s ability to pay is made in other ways such as by evaluating regional per capita income differences and also reviewing sample or aggregate portfolio level DTI distributions and the DTI origination policies. 7
As lack of DTI data makes the loan-by-loan analysis less accurate, assumptions for these jurisdictions tend to be more conservative to account for the shortcomings involved in measuring borrowers financial strength with a variable which may be less reliable than DTI. In all cases, Fitch reviews the individual affordability assessment guidelines of lenders when deriving the lender adjustment and degree of exceptions to affordability limits (see below). For second-lien loans, Fitch uses information on the first lien when computing the DTI for the borrower. Where this information is unavailable, the agency may decide not to rate the transaction. For investment loans also known in some jurisdictions as buy-to-let loans which are underwritten on the rental-income stream, Fitch may consider alternative measures in addition to the DTI such as the interest coverage ratio (ICR) on the rental stream. LVR is used as a measure of willingness to pay Willingness to Pay Based on its analysis, Fitch believes willingness to pay is best indicated by the amount of equity invested in the property, as determined by LVR, defined as the current outstanding loan amount as a percentage of the latest valuation at or before origination. The borrower s perception of the magnitude of their own equity, or wealth, invested in the property significantly affects the likelihood of default when the borrower is in financial distress. Where portfolios have a concentration of borrowers with negative equity Fitch will treat these portfolios on an individual basis. Its assessment of the default behaviour of borrowers in negative equity depends, among other elements, on the relevant legal regime, in particular whether the lender has full recourse to the borrower for shortfalls following foreclosure, which can be an important driver of default behaviour. Negative equity can increase the likelihood of default for borrowers already in financial distress. However, borrowers in full recourse regimes are unlikely to default solely because the value of their house is less than the outstanding balance of their mortgage. An economic environment with the combination of increasing macroeconomic stress combined with negative equity can have a strong effect on foreclosure frequency. Fitch accounts for this impact on the expected foreclosure frequency on a jurisdiction-specific basis. Most countries have legal regimes where a secured lender has recourse to an individual borrower if the property securing a loan is repossessed and sold for less than the full outstanding amount owed by the borrower to the lender. Data received by Fitch suggest that recoveries received from borrowers following the sale of the property tend to be negligible and it is only on rare occasions where the amount received clears the shortfall in full. However, Fitch believes that recourse to the borrower for outstanding amounts can provide a disincentive for a borrower to default. Consequently, the agency believes that borrower default behaviour is positively affected by the presence of full recourse to the borrower. Each standard borrower is assigned a base foreclosure frequency that is taken from an LVR-borrower financial strength matrix Base Frequency of Foreclosure Matrix and Rating Scenario Multipliers Each standard borrower is assigned a base foreclosure frequency that is derived from an LVRborrower financial strength matrix. The method for measuring a borrower s financial strength differs depending on the jurisdiction where the borrower is located. In those countries where the measurement method consists of assessing the relationship between LVR and DTI, the matrix splits LVR and DTI into a range of classes to which different foreclosure frequencies are assigned. Separate base foreclosure frequency matrices are derived for non-conforming loans in jurisdictions where relevant market data are available. If data regarding the non-conforming segment have limited market coverage, Fitch will develop a separate matrix for each nonconforming deal using data provided by the originator. 8
The relationship between foreclosure, LVR and DTI is typically determined by the results of a regression analysis performed on a loan-by-loan data set covering several transactions in each jurisdiction. Where loan-by-loan data are not available, Fitch will typically use time series analysis to derive base foreclosure frequencies. Fitch may apply a rating cap where data are limited or may not rate a transaction at all from specific market segments if an originator s historical data on its mortgage loan portfolio prove too limited. Fitch applies rating scenario multipliers to the foreclosure frequency matrix to derive the base foreclosure frequency for each rating level Figure 2 Example 1: Loan Level Gross Loss Calculation for AAAsf Rating Scenario This illustrative example shows the calculation of a gross loss for a loan in a AAAsf rating scenario. It will be developed further throughout the text. The example does not pertain to a specific country. The example is summarised in Appendix 3 (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (B) Loss severity (C) Gross loss The LVR buckets and DTI classes can differ by jurisdiction as they depend on the range of distribution of these variables. Rating Scenario Multipliers for Frequency of Foreclosure Fitch applies rating scenario multipliers to the foreclosure frequency matrix to derive the base foreclosure frequency for each rating level. The agency scales the base foreclosure frequencies from the Bsf rating level to other rating levels using a set of multipliers. The AAAsf base foreclosure frequencies are generally equal to approximately seven times the observed long term default rate observed in each jurisdiction. Recent experience since the global financial crisis indicates that this benchmark remains generally appropriate against Fitch s expectations. However, if expected default rates are greater than the long-term average due to macroeconomic stress, the differential between the multipliers for the various rating scenarios is contracted, reducing the distance between expected default rates and a remote scenario. If expected default rates drop, the differential between the rating scenario multipliers expands. Countries where the mortgage and housing markets are expected to experience more stress, accordingly have a smaller multiplier between rating scenarios, as opposed to jurisdictions where the forward-looking expectations are relatively benign. Figure 3 Example 1 (Cont.): Loan Level Gross Loss Calculation for AAAsf Rating Scenario (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (2) Rating scenario multiplier for AAAsf scenario Factor 5.0 10.50 Fitch adjusts the base foreclosure frequency on a loan-by-loan basis in each rating category to account for individual borrower, loan and property characteristics (B) Loss severity (C) Gross loss Rating assumptions for stress scenarios assume a degree of stress to the entire economy and thereby also to the finances of the sovereign. Fitch may decide to assign ratings higher than the Local-Currency Issuer Default Rating (LC IDR) of the corresponding sovereign as long as 9
an assessment of the scenario under consideration of a sovereign default can be made. The agency will consider several factors which can influence the scenarios it assesses that include the sovereign default, such as the profile of government finances, the relationship of the government to international networks, the strength of the legal system and the vulnerability of the financial system. In any event, Fitch will apply a cap to its structured finance ratings which is a maximum of six notches above the sovereign s LC IDR. Fitch will continue to evaluate the appropriateness of its existing assumptions at rating levels higher than the sovereign IDR whenever rating actions on the sovereign are taken. In case of severe downgrades towards lower investment-grade levels or below, Fitch may decide not to rate to the highest international rating categories for either new or existing transactions and in any event will apply a cap which is a maximum of six notches above the sovereign s LC IDR. Adjustments to Base Frequency of Foreclosure Fitch adjusts the base foreclosure frequency on a loan-by-loan basis in each rating category to account for individual borrower, loan and property characteristics. These adjustments are derived from detailed analyses of loan-level and aggregate-level performance data using logistic regressions and univariate analysis. Where available data for carrying out statistical analyses are limited, Fitch also uses its professional judgement and benchmarking to other countries (where sufficient data are available) to supplement its derivation of these adjustments. Where the portfolio includes a significant share of loans displaying unusual characteristics, eg LVRs outside of the standard range in the country, non-standard products, Fitch can define transaction-specific adjustments. In order to do so, the agency would expect the lender to provide static default performance data on loans affected by the unusual feature. Various adjustments are applied as multiplicative uplifts to the base foreclosure frequency. Please refer to the country-specific criteria addenda, later in this report, for more details on the specific adjustments applied in particular jurisdictions. Fitch may also apply a lender adjustment to the portfolio to account for the quality of underwriting, based on its originator review and historical default performance of the lender Adjustment for Underwriting Quality (Lender Adjustment) A key factor when evaluating and rating a securitisation of a portfolio of mortgage loans is the quality of the underwriting, as Fitch believes there is a direct relationship between this process and the performance of a collateral portfolio. This evaluation affects the lender adjustment as described in Lender Adjustment below. Originator reviews, including evaluations of policies and procedures, on-site visits and loan file reviews, are performed as part of the agency s analysis of a particular proposed transaction. For repeat issuers, Fitch may not conduct an on-site visit or file reviews, if there have been no material changes in the origination/servicing policies and in the products included in the portfolio. In those instances, Fitch conducts update calls and reviews documents describing origination and servicing policies. Nonetheless, Fitch carries out originator reviews of repeat issuers at least once a year. Originator Review The focus of the originator review is the underwriting process. The way in which loans are sourced, eg through branch networks or intermediaries such as brokers, and the assessment of the borrower s creditworthiness are key to judging the quality of an originator. The technology that the originator uses to process the application is another important aspect of underwriting a mortgage loan. The agency also assesses data on staffing, such as levels of experience, ongoing training and productivity incentives. In addition, valuation procedures are evaluated and quality control issues are discussed to find out how well an originator adheres to its own guidelines and procedures. 10
An assessment of the financial condition of an originator, its underwriting guidelines, appraisals and collection procedures are of particular importance when assessing its quality and lending practices. Fitch analysts will also complete a targeted file review to better understand the operational implementation and consistency of the originator s practices and policies. The agency reviews data received for consistency compared to, among other things: its expectations; knowledge of the originator/data provider; its knowledge of the market; and any other source that it considers a meaningful comparison. If data inconsistencies are observed and not reasonably explained, Fitch may decide not to rate the transaction or to apply a rating cap. Lender Adjustment Fitch s base foreclosure frequency assumes the underwriting criteria and origination practices of a standard, average lender originating mortgages in a particular jurisdiction. Adjustments are applied on a portfolio-wide basis to reflect variances from the originating practices of an average standard lender at the time of the criteria review. A number of factors will be considered in this assessment, including: the general robustness or otherwise of underwriting standards; origination practices, such as internal quality controls; valuation techniques; underwriter experience and performance track record; historical default performance of the lender s mortgage book; and any other relevant historical data. Fitch will also review recent material changes to origination and underwriting practices that are applicable for vintages specific to the transaction analysed and the lender adjustment may be adjusted accordingly. The agency uses the data received from lenders as described in the Data Adequacy section above, along with information gathered at the on-site review visit, to benchmark the lender against its industry peers and to determine the appropriate lender adjustment. Figure 4 Example 1 (Cont.): Loan Level Gross Loss Calculation for AAAsf Rating Scenario (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (2) Rating scenario multiplier for AAAsf scenario Factor 5.0 10.50 (3) Lender adjustment Factor 0.95 9.98 (B) Loss severity (C) Gross loss Examples of borrower adjustments: self-employed; adverse credit characteristics; and foreign residence. Borrower-Specific Adjustments Borrower profile adjustments to the base foreclosure frequency are made and vary by country, depending on differences in borrower characteristics and whether there is supportive evidence in the data. Examples of borrower-specific adjustments include employment type, eg selfemployed or employed, and adverse credit characteristics. For example, in several APAC jurisdictions, data analysis indicates that self-employed borrowers have a greater probability of default than those who are paid a salary. In Fitch s view, a borrower who earns a fixed monthly salary is more likely to be able to make periodic mortgage payments than a self-employed borrower who generates income from his or her 11
own business and is more susceptible to economic cycles and business interruption, eg as a result of personal illness. For these reasons, Fitch makes an upward adjustment to the base foreclosure frequency on a loan made to a self-employed borrower. In the non-conforming sector, Fitch also adjusts, eg for adverse credit characteristics of the borrower, such as prior defaults, bankruptcies or court judgements. Borrowers who have had payment problems in the past as indicated by some of the above measures are typically more likely to default. If the portfolio contains loans to individuals residing outside the jurisdiction, Fitch may increase the foreclosure frequency and decrease recoveries for the related loans. This is because the different jurisdictions involved mean that this type of borrower is generally not subject to the same motivations and incentives to pay and is more likely to abandon the property. In addition, cross-border credit checks are more difficult to make and potentially weaker. Figure 5 Example 1 (Cont.): Loan Level Gross Loss Calculation for AAAsf Rating Scenario (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (2) Rating scenario multiplier for AAAsf scenario Factor 5.0 10.50 (3) Lender adjustment Factor 0.95 9.98 (4) Borrower adjustments to base frequency of foreclosure Employment type: Self-employed borrower Factor 1.25 12.47 (B) Loss severity (C) Gross loss Examples of loan-specific adjustment: interest-only loans; self-certified loans; and semi-annual payment frequency Loan-Specific Adjustments Based on its data analysis, Fitch makes adjustments to the base foreclosure frequency on each loan for various loan characteristics, which can include loan purpose (eg debt consolidation loans), repayment type (eg interest-only, amortising loan), interest-rate type (eg variable-rate loans), payment frequency (eg quarterly, semi-annual), low documentation (or self-certified) loans and specific products in the non-conforming sector. Similar to borrower-specific adjustments, the types and size of product-specific adjustments can vary significantly across countries and are largely dependent on the nature of products that are available in that specific mortgage market. Detailed examples of typical loan-specific adjustments across some jurisdictions are set out below: Figure 6 Example 1 (Cont.): Loan Level Gross Loss Calculation for AAAsf Rating Scenario (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (2) Rating scenario multiplier for AAAsf scenario Factor 5.0 10.50 (3) Lender adjustment Factor 0.95 9.98 (4) Borrower adjustments to base frequency of foreclosure Employment type: Self-employed borrower Factor 1.25 12.47 (5) Loan adjustments to base frequency of foreclosure Interest frequency: Quarterly payments Factor 1.10 13.72 Repayment type: Annuity loan Factor 1.00 13.72 Interest-rate type: Fixed interest Factor 1.00 13.72 Loan purpose: Debt consolidation Factor 1.30 17.83 (B) Loss severity (C) Gross loss 12
Fitch may apply an adjustment to interest-only loans to account for their higher risk of foreclosure due to the payment shock borrowers may experience when a balloon repayment becomes due or a shortened amortisation period commences. Also, the borrower builds up less equity in the property compared to a standard fully-amortising mortgage, meaning there is less incentive to keep payments current. This is for two reasons: First, borrowers with potential affordability problems are more likely to select an interest-only loan, to reduce the size of the initial instalment payment. Second, interest-only loans can be construed to be riskier than amortising loans because of the greater risk that the borrower may be unable to repay the debt in full at maturity. In countries where there are a mix of long-term fixed-rate and variable-rate loans, an adjustment may be applied to loans where the interest rate is variable. This is because borrowers with variable-rate mortgages can experience payment shocks due to underlying interest rate volatility. Mortgage lenders usually factor in this risk when assessing the affordability of the loans by stressing the current interest rate. Fitch may make an adjustment on top of the base foreclosure frequency for variable-rate loans to account for their potentially greater sensitivity to changes in interest rates if this is not adequately taken into account during the underwriting process. Conversely, in countries where variable-rate mortgage loans are the norm, no such adjustment is made as such vulnerability to interest-rate volatility is already captured in the base foreclosure frequency of a standard loan. Instead, Fitch may consider applying an adjustment for long-term fixed rate loans, where supported by data evidence. Based on findings from its analysis, Fitch views loans where interest payments are made on a less frequent basis such as quarterly, semi-annual or annual as higher risk than loans where the payment frequency is monthly. This is because of a mismatch between frequency of income and mortgage payments and also because borrowers who chose such payment arrangements are more prone to mismanagement of their finances. Fitch increases the base foreclosure frequency for such loans. Low documentation or selfcertified borrowers are generally not required to provide full documentation to verify their income declaration and lenders perform varying degrees of due diligence on the declared income level. Due to the risks associated with borrowers misusing self-certification of income and consequently overstating affordability, Fitch increases the base foreclosure frequency on such loans. Property-Specific Adjustments Figure 7 Example 1 (Cont.): Loan Level Gross Loss Calculation for AAAsf Rating Scenario (A) Foreclosure frequency Calculation step Foreclosure frequency (FF) (%) (1) Base frequency of foreclosure from Bsf DTI/LVR matrix 60% LVR/30% DTI 2.10 (2) Rating scenario multiplier for AAAsf scenario Factor 5.0 10.50 (3) Lender adjustment Factor 0.95 9.98 (4) Borrower adjustments to base frequency of foreclosure Employment type: Self-employed borrower Factor 1.25 12.47 (5) Loan adjustments to base frequency of foreclosure Interest frequency: Quarterly payments Factor 1.10 13.72 Repayment type: Annuity loan Factor 1.00 13.72 Interest-rate type: Fixed interest Factor 1.00 13.72 Loan purpose: Debt consolidation Factor 1.30 17.83 (6) Property adjustments to base frequency of foreclosure Property use: Investment property Factor 1.25 22.29 (B) Loss severity (C) Gross loss 13
Examples of property specific adjustments: second homes; and investment property Fitch believes that mortgage loans for second homes or investment properties are more susceptible to default, since a financially distressed borrower is more likely to default on the mortgage of a second home or investment property than on a mortgage on their primary residence. For these reasons, foreclosure frequencies of such loans are adjusted upwards. Other Adjustments Employer Concentration Transaction portfolios sometimes include a substantial proportion of loans granted to borrowers employed by a unique employer, often the originator itself. Fitch s standard analysis assumes a good diversification; this assumption is not appropriate when portfolios contain a high proportion of loans granted to borrowers employed by a single employer. Instances of financial difficulties, redundancy plans or default of the employer may affect the creditworthiness of all related borrowers at the same time. For these reasons, Fitch may assign a higher foreclosure frequency and/or loss severity assumptions to any employee concentration, in order to simulate the effect an adverse event related to the employer may have on their creditworthiness. A significant employer concentration may lead Fitch to link, or even cap, the rating of notes to that of the employer. In addition, Fitch s default rate assumption for portfolios or sub-portfolios with large employer concentration will consider: a) The probability of redundancy: In rating scenarios above the employer s IDR, Fitch would typically assume a significantly higher default likelihood for concentrated employee portfolios. Below the employer s IDR, higher default likelihood may be assigned as well to reflect concentrated exposure to financial distress of the same entity. The rate of redundancy assumed by Fitch at the higher rating scenarios will also depend on the systemic importance of the employer, and consequently the likelihood that it will be liquidated or significantly sized down after a default on its obligations. b) The probability of default following redundancy: Borrowers made redundant may be able to maintain their mortgage payments for a period of time making use of savings, unemployment benefits and other sources of support. Fitch will also take into account reemployability, which tends to vary depending on the employer s industry, region and type of organisation, as well as levels of regional unemployment and employees qualifications. Similarly, and unless the rating of the notes is closely linked or capped to the rating of the employer-originator, Fitch will consider the set-off risk arising from the originator s default. The agency may assume losses resulting from the set-off against mortgage balances of due and unpaid salaries, severance payments or pension entitlements, depending on the legal protections available against this risk. As part of its cash-flow analysis, Fitch may also test the sensitivity of the transaction structure to some particular time concentration of the borrowers redundancies and defaults. Geographic Concentration Fitch assumes that the mortgage portfolio is generally broadly diversified geographically within each country or reflects the average geographic distribution of the country s property stock. The agency may apply adjustments to its base assumptions to account for portfolios with significantly higher than average regional concentrations to reflect the greater vulnerability to local economic downturns compared to a diversified portfolio. The magnitude of the adjustment will depend on an individual analysis, since some regions may be more/less sensitive to economic downturns and/or other developments in the property and mortgage market. Fitch may also consider a rating cap in cases of excessive geographic concentration. 14