OVERVIEW AND SCOPE

1. S&P Global Ratings' methodology and assumptions for assessing pools of residential loans describe its approach to rating residential mortgage-backed securities (RMBS) and covered bonds globally. These criteria follow "Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Korea)," published Feb. 13, 2023, and the criteria are effective for Korea as of June 2, 2023. For a comparison of the changes between the Request For Comment (RFC) article and the final criteria, see "RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Korea)," published June 2, 2023. See the Related Research section for previous RFC and RFC Process Summary articles related to these criteria.

2. These criteria apply to residential mortgage-backed securities (RMBS) and covered bonds globally, except for jurisdictions (listed in Appendix IV) where S&P Global Ratings currently has separately published criteria, and for jurisdictions where an analytical approach has been established using "Principles of Credit Ratings," published Feb. 16, 2011, on RatingsDirect. If such jurisdictions were to become in scope in the future, we would first seek comments via a request for comment (RFC) publication.

3. These criteria supersede the articles listed in Appendix III as they become effective in each applicable market. Some markets require prior notification to and/or registration by the local regulator, where the criteria will become effective when so notified by S&P Global Ratings and/or registered by the regulator. These criteria are accompanied by a guidance article, which sets out considerations for our application of the criteria in selected jurisdictions in Europe and Latin America (see "Guidance: Global Methodology And Assumptions: Assessing Pools Of Residential Loans," published Jan. 25, 2019). Furthermore, appendices of this article discuss the application of these criteria in other relevant jurisdictions. These appendices are each associated with a corresponding Sector And Industry Variables (SIVR) article that provides additional information on the variables we use when applying the criteria in corresponding jurisdictions.

4. The criteria framework for analyzing residential mortgage credit risks in individual jurisdictions includes:

• A defined archetypal pool against which other mortgage pools are compared;
• The assessment of mortgage risk based on loan attributes and borrower characteristics relative to the archetype; and
• The calculation of a repossession market value decline to estimate recoveries.

5. We also use the Mortgage Market Assessment (MMA) to calibrate the foreclosure frequency assumptions associated with the archetypal pool of a given country. The MMA provides a relative risk ranking of mortgage markets across jurisdictions based on macroeconomic and mortgage industry considerations.

METHODOLOGY AND ASSUMPTIONS

Criteria Framework

6. The framework for our analysis of securities backed by residential mortgage loans considers the risks associated with the credit quality of the underlying assets. For the analysis of payment structure and cash flow mechanics, operational and administrative risk, counterparty risk, and legal and regulatory risk, see the "Related Criteria And Research" section.

7. Charts 1A and 1B below summarize the approach used for our credit analysis under these criteria.

Credit quality of the underlying assets

8. To assess the credit quality of the underlying assets, we determine projected losses for each pool of residential mortgage loans under conditions of stress commensurate with certain defined issue ratings. Projected losses are the product of foreclosure frequency and loss severity:

• Foreclosure frequency measures the probability of default. It represents the proportion of loans in a pool that are likely to go into foreclosure.
• Loss severity measures loss-given default, expressed as a percentage of the loan balance. It refers to the loss on foreclosure (i.e., the amount by which a loan balance and foreclosure costs exceed the property sale proceeds and other proceeds [e.g., lender's mortgage insurance]).

9. To determine projected losses, the framework establishes an archetypal pool of residential mortgage loans for a given jurisdiction based on global and country-specific considerations, in which we associate foreclosure frequency assumptions to each rating level. We apply adjustment factors for variations from the archetype to reflect varying degrees of credit risks in individual portfolios. The criteria calculate the rating-specific market value decline likely to be endured when foreclosing on a property, and adjust it to determine loss severity assumptions upon default of a mortgage in a pool (including foreclosure and other associated costs).

10. The outcome of these calculations is a stressed foreclosure frequency, loss severity, and loss coverage at each rating level for a pool. The rated tranche is generally expected to withstand the stressed loss level without defaulting in order to achieve a certain rating.

11. To calibrate the foreclosure frequency assumptions associated with the archetypal pool for a given country in a systematic, transparent manner, we use the Mortgage Market Assessment (MMA). The MMA considers key factors in the calibration of country specific foreclosure frequencies.

12. In addition to the MMA, the application of these criteria, when analyzing pools in individual countries, also considers the specificities of the local mortgage markets. For this reason, the specification for the archetypal pool varies across individual countries, and where warranted, may include considerations not already made under these criteria. Similarly, the criteria provide general principles for assessing variations from the archetype. Depending on the situations, additional considerations may be made in the more detailed application of the criteria, or in order to reflect risk factors that are unique to a specific mortgage market.

Global Approach For The Analysis Of Mortgage Default Risk – Calculating The Foreclosure Frequency Of A Mortgage Pool

MMA methodology: country specific assessment

13. The criteria use the MMA to calibrate the archetypal pool's foreclosure frequencies for each individual mortgage market (typically at a country level). The MMA framework is designed to evaluate the relative risk of residential mortgage markets globally.

14. Our MMA analysis results in an assessment ranging across six categories from "very low risk" to "extremely high risk." The MMA has three components: economic risk, mortgage industry risk, and provisions for lender recourse to the borrower.

MMA methodology: deriving the archetypal foreclosure frequencies

Table 1

Deriving Foreclosure Frequency Assumptions Based On MMA
MMA	'AAA' anchor (%)	'B' minimum (%)
Very low risk	8-9	0.50
Low risk	9-13	1.00
Intermediate risk	10-15	1.25
High risk	15-20	2.00
Very high risk	20-30	3.00
Extremely high risk	30 or more	4.00
MMA--Mortgage Market Assessment.

15. Each MMA maps to a range of potential 'AAA' foreclosure frequency assumptions and a minimum default assumption at the 'B' rating level, as shown in table 1.

16. These assumptions reflect a rank ordering based on macroeconomic considerations to benchmark relative credit risks across jurisdictions. They are the basis for determining the default assumptions assigned to an archetypal pool of residential mortgages in a given country, as described further below.

17. For more detail on our assessment of the MMA components and the determination of the final MMA for a country, see Appendix I.

Setting the 'AAA' archetypal foreclosure frequencies

18. The ranges of applicable foreclosure frequencies at the 'AAA' rating level are calibrated taking into account extreme historical stress scenarios as a reference point. We considered the U.S. Great Depression, as well as more recent stress scenarios such as those experienced in Greece and Argentina.

19. As such, these 'AAA' foreclosure frequency ranges (see table 1 above) are intended to provide stability to our default expectations through normal economic cycles. However, as the ranges indicated in table 1 show, 'AAA' levels could increase (or decrease) if the level of risk in the market increases (or decreases) significantly. For example, we would expect such change in the case of a moderate-to-substantial level of economic stress and/or a material structural shift in the mortgage market.

20. In order to determine the specific 'AAA' foreclosure frequency assumption within the range shown in table 1 assigned to a particular country based on its MMA, we consider a number of factors to assess the default risk relative to peers with comparable MMAs. Some examples of the types of factors we may consider include:

• Idiosyncratic market features such as product links to inflation, mortgage penetration levels, etc.;
• Long-term performance history relative to peers; and/or
• Data limitations, such as in relation to long- and short-term performance history, or availability of granular pool level data.

21. Based on this relative risk-ordering approach, if we observe a given mortgage market's performance over cycles is stronger (or weaker) relative to peers (i.e., countries with a comparable MMA), we typically assign a 'AAA' anchor at the lower (or higher) end of the range for such jurisdiction. Similarly, where there are limitations to the available data, we typically assign a 'AAA' anchor at the higher end of the range for a given MMA.

22. In certain circumstances, the 'AAA' foreclosure frequency assumption assigned for a given jurisdiction's archetypal pool may exceed the range associated with the relevant MMA as indicated in table 1. For example, this could occur if observed defaults on loans with archetypal characteristics in a severe (or less stressful) economic downturn exceed the levels associated with the relevant range of 'AAA' foreclosure frequencies. Alternatively, we do not increase the 'AAA' foreclosure frequency beyond the range indicated by table 1 if the negative performance was driven by factors we are otherwise adjusting for in our analysis (e.g., poor underwriting standards, very high loan-to-value ratios [LTVs], etc.).

Determining our expected-case foreclosure frequency assumptions at the 'B' rating level

23. The 'B' foreclosure frequency minimums in table 1 serve as the starting point for our 'B' expected-case foreclosure frequency for the archetypal pool of a given country.

24. Our expected case reflects observed trends as well as our outlook for the prevailing market over the next three-to-five years. The outlook takes into account considerations such as:

• Market structure and underwriting standards (and any relevant changes thereto);
• Forecast unemployment rates, inflation, and interest rates; and
• Prevailing mortgage loan performance, defaults, delinquencies, and their roll rates (e.g., transition rates).

25. Therefore, default assumptions at the 'B' rating level adjust over time to reflect changing economic conditions, when we believe they may materially affect levels of credit risk in the mortgage market.

26. If the 'B' (or more rarely, the 'AAA') foreclosure frequency assumptions for the archetypal pool are updated, intermediate foreclosure frequency assumptions ('B+' to 'AA+') are also updated through an interpolation based on the relevant 'B' and 'AAA' values.

Pool Analysis Methodology: Establishing An Archetypal Pool Of Residential Mortgage Loans

27. We perform a loan-level analysis to assess a residential mortgage pool's credit quality based on a comparison of its borrower attributes and loan characteristics to those of the archetypal pool. Pool foreclosure frequency levels are derived from archetypal projected defaults (see the Criteria Framework section above) adjusted for relevant variances from the archetype as described further below in the Pool Analysis Methodology: Adjustment Factors For Variations From The Archetype section.

28. The archetypal pool in each jurisdiction or market represents a somewhat idealized version of the average pools historically observed. While reflecting local features, the archetypal pool for a given country generally includes the below features:

• At least 250 loans (at the inception of the transaction that the pool is backing);
• Prudently/conservatively underwritten (i.e., no originator adjustment is applicable)--more specific characteristics may feature in local markets;
• Borrower has no adverse credit history (or has a prime-like credit score);
• Loan's purpose is for property purchase (not a refinancing loan or cash-out);
• Diversified nationally;
• LTV of the loan reflects prudent underwriting in the given market (e.g., 75%);
• Loan affordability was assessed at origination;
• Pool reflects an idealized product mix (for example, share of fixed- versus floating-rate loans, no loans with payment shock features, etc.);
• Borrower is a private individual;
• Loan is current;
• Borrower is not self-employed;
• Loan is denominated in domestic currency of lender and borrower;
• Unseasoned loan;
• Fully amortizing loan;
• First-lien mortgage (or equivalent in certain jurisdictions);
• Residential property, owner-occupied; and
• Fully appraised property valuation.

29. In addition, where relevant in a given market, we specify the archetypal pool to also include the following:

• Original loan term if this is a relevant risk factor (lesser risk for shorter-term loans and greater risk if tenor greater than market average);
• Primary residence (i.e., not a second home);
• Residency status of the borrower;
• Not a "jumbo" property valuation (i.e., an unusually large property presenting more idiosyncratic risk); and
• Other local relevant market factors.

Pool Analysis Methodology: Adjustment Factors For Variations From The Archetype

30. For each characteristic and attribute that differ from the archetype at a pool, loan, or borrower level, the criteria apply a corresponding adjustment to the foreclosure frequency, determined based on the methodology described in the Global Approach For The Analysis Of Mortgage Default Risk – Calculating The Foreclosure Frequency Of A Mortgage Pool section. Depending on the risk profile of the characteristic or attribute relative to the corresponding archetypal trait, the adjustment increases, and sometimes decreases, the foreclosure frequency assumption for the individual loan or pool, as further described below in this section. The typical foreclosure frequency adjustments for each of the factors are summarized in table 2 below. These adjustments vary across jurisdictions based on the differentiating factors identified below, to reflect local market characteristics and/or performance.

Table 2

Foreclosure Frequency Adjustments
Credit Attribute	Typical FF adjustment factor	Reasons for potential variance
LTV	Type 1, 2, or 3	See below
Affordability: DTI	0.8x-1.8x	Less weighting in underwriting lead to tighter range
Affordability: income multiple	1.2x–1.5x for multiples between 3.5x and 5.0x (or greater)	Less weighting in underwriting lead to tighter range
Seasoning	0.75x–0.50x for seasoning from five to 10 years or more
Employment status: self-employed	1.1x-1.5x	Local employment law considerations
Employment status: pensioner/retired borrower	1.0x–1.5x	High pool concentrations in the absence of mitigating factors
Security: second lien	1.3x-1.7x	If data on primary lien is not available a higher multiple is applied
Investment properties (excluding loans underwritten to DSCR in markets we consider as limited recourse )	1.3x-1.7x	Low end of range applicable if underwritten to borrower and property generated income with full recourse
Investment properties (loans underwritten to DSCR in markets we consider as limited recourse markets)	3.0x–6.0x	Varies based on loan-level DSCR; other adjustment factors may not be applicable (e.g., occupancy, DTI, employment type)
Occupancy: second homes	1.1x-1.3x	High end of range applicable if performance data is limited
Loan purpose: cash out/equity withdrawal	1.2x
Product type: fixed (if non-archetypal)	0.8x
Product type: floating (if non-archetypal)	1.2x
Product type: payment shock feature (e.g., interest-only and teaser-rate products)	1.1x–2.0x; may be removed six months after initial payment shock	Atypical product types considered to pose greater payment shock (such as balloons or negative amortization loans) warrant a higher adjustment factor
Arrears status	2.5x for 30 days delinquent; 5.0x for 60 days delinquent; and 100% FF for 90 days delinquent	Days delinquent threshold depends on commencement of foreclosure proceedings in applicable jurisdiction
Nonresidential loans	1.5x–2.0x	Pool concentration not to exceed 40%
Nonresidential borrowers	2.0x	Pool concentration not to exceed 40%; not applicable to trust, SPE, or similar borrowers with recourse to an individual
Residency status	1.5x-2.5x	Employment volatility of sector, and risks to foreclosure
Credit history	Up to 4x	Differences in market practices
Geographic concentration	1.1x-1.25x	Low end of range used when applied to entire region; high end of range used when applied to excess concentration only
Originator adjustment	0.7x-1.3x (or more)
DSCR--Debt service coverage ratio. FF--Foreclosure frequency. DTI--Debt to income. SPE--Special-purpose entity.

Loan-to-value (LTV) of the loan – LTV curves

31. The criteria recognize that the LTV of a loan is a dominant factor in predicting future mortgage performance. The actual relationship, however, tends to be country-specific, depending on local factors (for example, whether lending is on a full- or limited-recourse basis). Based on available data, our criteria consider different LTV measures.

32. A loan's original LTV (OLTV) is calculated as the original balance of a loan (and any second-lien exposures, where applicable) divided by the original valuation of the mortgaged property. A loan's current LTV (CLTV) is the current balance of a loan, reflecting the actual (or projected) principal pay down (and any second-lien exposures, where applicable), divided by the indexed initial value or other updated value of the property.

33. The criteria use an LTV curve to make upward and downward adjustments to the foreclosure frequency of a loan, depending on whether its LTV is higher or lower than the LTV of the archetypal loan in the country.

34. The LTV of a loan is an indicator of the borrower's willingness to pay back the mortgage, given the equity at stake in the financed property.

35. The criteria establish a sliding scale of a borrower's willingness to pay, depending on market characteristics. For example, in a market where lending is typically on a limited-recourse basis, a borrower may have an incentive to hand in the keys if little-to-no equity exists in a mortgaged property or if the borrower is in negative equity due to adverse property price fluctuations. In that case, a higher LTV, all else being equal, is a marker of higher risk. Conversely, in a market where high LTV lending is primarily driven by tax incentives such as mortgage interest deductibility, a high LTV loan may not necessarily signal increased borrower risk. To reflect such market specificities, the LTV curve features a steeper or flatter slope, respectively, in each case.

36. The criteria typically use one of three different LTV curve types (encompassing both the curve and the associated LTV measures). The use of different LTV curves recognizes that jurisdictional factors influence risk differently and reflects available data analysis. We discuss each curve type below.

Type 1 LTV curve

37. Typically used where mortgage lending is generally on a limited-recourse basis, this curve has the steepest slope. For example, based on a neutral point (a factor of 1x) for an archetypal LTV of 75%, the adjustment factor is 3.00x at an LTV of 95%, and 3.95x at an LTV of 100%. This curve increases exponentially as LTV levels increase, capping out at LTVs well above 100%.

38. In a limited-recourse market, the CLTV of a loan is typically as meaningful a relevant risk differentiator as the original equity investment (OLTV). Assuming the loan was originally granted at an LTV of less than 100%, the propensity of a borrower to default when in negative equity depends on the degree to which the value of the property has decreased. In other words, the degree of negative equity is a driver of default on an exponential basis. Therefore, in markets we consider to be limited-recourse, the LTV curve is typically applied to a combined OLTV and CLTV.

39. Chart 2 below illustrates an example of a type 1 LTV curve, based on an archetypal (neutral) LTV level of 75%.

Type 2 LTV curve

40. This curve is typically used in markets where mortgage lending is on a full-recourse basis. The slope or shape of the curve is similar to that of the type 1 curve, meaning that the "growth rate," or relative increase of the adjustment factor for a given LTV increase, is generally the same. However, the slope of the curve (for example, assuming a neutral point at 75%) at higher LTV levels (typically around 98%) starts to flatten and ultimately levels off (typically around a maximum of 130%).

41. The slope of the curve at higher LTVs does not increase exponentially because, in full-recourse markets, we believe the risk of negative equity is less of a factor, as borrowers cannot walk away from the property and lenders can go after non-house assets to recoup loan amounts. Therefore, in our view, there is less differentiation for LTV levels above 100% than in limited-recourse markets. As a result, in markets we consider to be full-recourse, the combined CLTV and OLTV measure gives greater weight to a loan's OLTV than to its CLTV.

42. In some cases where updated property values are not available due to limitations in indices, the LTV measure used to adjust foreclosure frequency may be based on the current loan balance and the original property value.

43. Chart 3 below illustrates an example of a type 2 LTV curve, based on an archetypal (neutral) LTV level of 75%.

Type 3 LTV curve

44. This curve is typically featured in full-recourse markets only, where underwriting standards or structural features change the nature of LTV as a performance indicator, and typically merit a less stressful LTV curve. In some markets, higher LTV loans are not driven by an inability to fund the property with equity, but by, for example, an incentive to maximize tax advantages from interest deductibility. Therefore, on a relative basis, lending at high LTV is more commonly commanded by such incentives. In other markets, underwriting is more focused on capacity to repay on an ongoing basis, as measured by debt to income (DTI), and as a result, we place greater weight on DTI in our analysis.

45. For these reasons, the curve is flatter than the type 2 curve, with lower top-end multiples, typically capping out at an LTV above 100% based on our analysis of performance data.

46. As for the type 2 LTV curve, the combined CLTV and OLTV measure gives greater weight to a loan's OLTV than to its CLTV.

47. Chart 4 below illustrates an example of a type 3 LTV curve, based on an archetypal (neutral) LTV level of 85%.

48. For all LTV curves, the approach for setting the neutral point (that is, the LTV level for which the adjustment is 1x, [e.g., the archetypal LTV]), reflects local market conditions.

49. The determination of the neutral point is based on our experience of the market, findings from originator reviews, and expertise from our bank rating analyses. We typically set the neutral point at around 75%. In some markets, mortgage lending is typically on a low LTV basis (e.g., 55%); in these cases, the LTV curves are adjusted accordingly, such that higher adjustment factors apply to lower LTVs, thereby maintaining incremental changes in the adjustment factors from the neutral point across markets.

50. Finally, in markets where the underwriting approach incorporates the use of standardized credit scores, the risk adjustment for LTV may be stratified by the borrower's credit quality, as indicated by its credit score.

Affordability

51. The archetypal pool contains loans that have been assessed by the lender for affordability at origination based on standard market practice in a country.

52. To adjust for varying degrees of mortgage affordability, the criteria use one of two different approaches, depending on the practice in the market:

• Affordability is measured as a debt service coverage or DTI ratio, expressed as a percentage, equal to the monthly cost of servicing the borrower's debt plus other expenses divided by the monthly income.
• Affordability is measured more broadly as an income multiple, equal to the total borrowing divided by the annual income of a borrower.

53. If loan-level data is not available for measuring affordability, we make adjustments based on our assessment of the originator's practices relative to market standards via the originator adjustment at the pool level (see below).

Debt-to-income (DTI) affordability measures

54. At a DTI ratio of 35%, a neutral adjustment (1x) typically applies. Where affordability ratios for a given market only consider mortgage debt as opposed to "all-in" debt, the neutral point is typically lower.

55. We apply a sliding scale of adjustment factors for affordability. Credit (a multiple of below 1x) may be given for affordability better than the neutral point, whereas the adjustment factor increases for DTIs above the neutral point (typically capping out at DTI values of around 60%). In markets where we consider affordability to be a more significant driver of credit risks than LTV, and our LTV curves are flatter, we use a higher range of adjustments to allow for sufficient differentiation.

Income multiple

56. At an income multiple of 3.5x--meaning the debt of a borrower amounts to 3.5 years of income--the neutral adjustment factor (1x) typically applies. For income multiples higher or lower than the neutral point, we apply higher or lower adjustment factors. As the approach to affordability using an income multiple typically reflects underwriting that is more primarily based on LTV, we usually do not apply the income multiple adjustment factor when a loan is more than 18 months seasoned, except for very high multiples (typically greater than 5.0x).

Alternative affordability measures

57. Some jurisdictions measure affordability by means other than DTI or income multiples, such as a net-surplus ratio. Where loan-level information on such measures is provided, we typically adjust foreclosure frequencies based on jurisdiction specific considerations such as performance.

58. Where affordability information is not available at the loan level, or a common standard for reporting this information does not exist in a market (typically in mortgage markets where underwriting is less automated), we seek to identify the relative underwriting affordability standards of each originator. When we consider that a particular originator applies a weaker standard than the market, we make a portfolio-level adjustment to the foreclosure frequency of a pool through the originator adjustment. The size of the adjustment reflects the increased level of risk from the individual originator's underwriting practices.

Seasoning

59. We have observed globally that seasoned mortgage loans (e.g., more than five years from origination) generally perform better than newly originated loans. Continued performance over an extended period indicates that the borrower can afford to service the mortgage debt through ups and downs. We therefore apply a positive adjustment factor to the foreclosure frequency of loans that are seasoned more than five years. We typically only apply seasoning credit to loans that are currently performing. The start point for seasoning credit may vary from five years if there is strong supporting empirical evidence.

Employment status and income documentation

60. The archetypal borrower is typically salaried. Loans to self-employed borrowers therefore tend to attract an adjustment factor. In jurisdictions with limited employment protection, there is less differentiation in terms of job security between salaried and self-employed borrowers; therefore, the adjustment we apply is typically lower than would otherwise be the case.

61. In addition, in markets where borrowers have a degree of optionality around the level of documentation provided to lenders (income documentation), we typically apply an adjustment factor to reflect non-archetypal attributes. Depending on the type of documentation, the adjustment varies in size and duration. The criteria recognize that the source of information used to verify income and therefore affordability, as well as expectations for the number of months of documentation that is considered, can vary considerably across different markets, based on local practice.

62. In jurisdictions where mortgage pools include material concentrations of loans to pensioners (at origination), there also is an adjustment factor in the absence of mitigating factors. Similarly, in markets where underwriting to unemployed borrowers (at origination) is provided for, a significant adjustment typically applies in the absence of any mitigating factors.

Security

63. Sometimes a residential pool includes loans secured by second liens on a property, to which we typically apply an adjustment factor. The adjustment factor depends on the availability of information on the first lien. If information for the primary lien is not available, then an adjustment factor at the higher end of the range is applied.

Occupancy status

64. During periods of economic stress, a borrower is more likely to default on a mortgage secured by a second home or an investment property than on a home that is a primary residence.

65. Where lending on second or holiday homes is common in a market, we typically apply an adjustment factor.

66. Where the property is not owner occupied but is an investment property, the criteria apply an adjustment factor that varies depending on whether the loan has been underwritten to the borrower's income or to the debt service coverage ratio (DSCR) of the rental property.

67. In addition, in jurisdictions where the lender may not have ultimate recourse to the borrower's primary residence or other assets, loans underwritten solely to the DSCR typically attract a higher adjustment factor, but certain other adjustment factors may not be applicable (e.g., DTI, employment, and occupancy type).

Loan purpose

68. Generally, a loan used to borrow against the equity in a property (cash-out) has the highest default risk. Equity withdrawal is typically considered riskier than a home purchase or a fully re-underwritten refinancing.

69. Also, when a refinancing loan has not been fully re-underwritten, we typically apply a smaller adjustment factor than for cash-out loans to differentiate refinancing loans from conventional purchases, unless we have data that demonstrates no difference in performance exists during a stress period.

Loan product type

70. The specification for the archetypal loan product type in a given country reflects the longstanding traditional underwriting practice in such market.

71. When supporting data exist, the risk profile of standardized products is therefore likely to be reflected in the performance that informed our base default assumptions. However, when there is a specific characteristic not captured by data or for non-standard products in a market, we may make adjustments for varying loan product types.

72. A variation from the archetype for the interest rate product may introduce more (or less) risk. For example, in a floating-rate market, a fixed product for life therefore typically attracts a positive adjustment (i.e., less than 1.0x). Alternatively, a loan with a non-archetypal feature that could create a payment shock, or a floating-rate loan in a fixed-rate market, typically attracts a negative adjustment (i.e., greater than 1.0x) to foreclosure frequency.

73. The adjustment for other types of payment shock varies depending on the materiality of the potential shock. We may remove the adjustment factor after the expiry of the feature. Examples of products with payment shock features include loans with an initial interest-only period, a temporary promotional ("teaser") rate, or a fixed-to-float interest rate change.

74. The loan denomination may also introduce a mismatch risk, such as loans with balances that adjust with inflation where there is a risk that wage inflation does not track the relevant inflation index in which the loan is denominated.

75. Depending on specific local market characteristics, we adjust for other product types where they are prevalent. Examples include product types with atypical amortization schedules that present specific risks relative to the archetype, including interest-only and balloon loans.

76. A balloon loan has a significant portion of its principal paydown that is due as a lump sum, often not until late in the mortgage's life. Depending on the market practice, an interest-only product may refer to a loan whose principal amortization is deferred for several years, or to a loan whose repayment is actually only due on the final repayment date for the mortgage (interest-only for life). The sizing of the corresponding adjustment factor under the criteria reflects the available performance data as well as the degree of commoditization of the product in the local market. For example, depending on the market, short-term interest-only loans typically attract a higher adjustment factor compared to longer-term interest-only loans where the borrower has more time to separately build equity to proceed with the final repayment.

Loans in arrears

77. An adjustment factor is applied to loans in arrears based on when foreclosure proceedings are typically commenced. A default frequency of 100% in all rating scenarios is typically applied if the borrower is 90-180 or more days in arrears depending on the jurisdiction. A range of adjustment factors are applicable to loans in arrears less than 90-180 days depending on the jurisdiction (for example, 2.5x for loans that are 30 days' delinquent and 5.0x for loans that are 60 days' delinquent).

78. In addition, if a borrower has been in arrears but is now current, depending on the time elapsed since the arrears occurred, specific arrears adjustments may apply. Where loan-level data is not available, we account for pool-level arrears via an originator adjustment.

Re-performing and restructured/modified loans

79. A re-performing loan is one that is not currently seriously delinquent or in bankruptcy, but has been in arrears for three or more monthly payments (or, in some cases, where relevant, have been modified/restructured), in the recent past. Once a borrower has evidenced consistent performance since serious delinquency over an extended period of time, the loan may no longer be considered re-performing for the purposes of our analysis.

80. We typically apply adjustments for re-performing and restructured/modified loans, depending on the materiality of a pool's exposure to these loans. These adjustments typically take into account such considerations as the type of restructure/modification granted, the time elapsed since the restructure/modification, and/or performance subsequent to restructure/modification. The seasoning adjustment for these loans is generally based on the restructure/modification date for the remaining life of the loan.

Nonresidential loans/borrowers

81. In certain jurisdictions, mortgage pools include commercial or mixed-use properties in addition to residential homes. These properties present specific risks and we typically apply an adjustment factor.

82. In jurisdictions where borrowers may also be commercial entities (for example, the borrower is set up as a small- or medium-sized enterprise), we typically apply an adjustment factor. Such adjustments generally do not apply to loans made to trusts, special-purpose vehicles, and similar entities with direct recourse to an individual borrower.

Credit history

83. A borrower's past performance is a differentiating indicator of future mortgage performance. We therefore adjust foreclosure frequency assumptions to account for a borrower's adverse credit history.

84. The criteria recognize that the underwriting standards of different jurisdictions have evolved to capture this risk in a variety of ways, which is reflected in our analysis accordingly. For example, in some markets, we consider the number of prior credit events experienced by a borrower over a certain time horizon, as well as whether those credit events are more or less recent. Alternatively, when available in a market in a standardized fashion, we consider credit scores incorporating the borrower's payment history on past and current debt obligations.

85. As noted above in the discussion of LTV factors, sometimes a combined adjustment factor for different buckets of LTV and credit scores may be used. This is relevant in markets where data demonstrate the predictive power of combined LTV and credit score metrics.

86. In other jurisdictions, a specific record of past credit history may not exist in a systematic fashion. In that case, our analysis seeks to address the risk of lending to borrowers with negative credit histories by way of an originator adjustment reflecting the relative appetite of a lender for underwriting loans to more tarnished credits. This is because the archetypal loan is in all cases defined as one made to a borrower with a clean credit record.

Geographic concentration

87. The criteria address geographic concentration risks in a pool by applying adjustment factors to the foreclosure frequency for portfolio concentrations that exceed specific geographic thresholds (e.g., regions and/or counties). Geographic concentration risk emanates from the risk of a localized downturn in a given country.

88. Thresholds for defining concentration are typically based on a country's population distribution as a proxy for diversification. For example, a threshold of two times the population is typically used under these criteria, rounded as necessary. Alternatively, the approach may use statistical methods where available to more specifically capture concentration risks.

89. Adjustments for geographic concentration are typically applied where the threshold is exceeded.

Original loan term

90. In markets where there is a meaningful differentiation in loan terms based on observed performance, we may apply an adjustment for variations from the archetype, and if loan level data is not available, we account for the risk in the originator adjustment.

Residency status

91. We typically apply an adjustment factor where we have assessed there may be a likelihood of higher default risk posed by borrowers that are not citizens or permanent residents of the country in which the mortgaged property is located.

92. The adjustment factor applied is at the higher or lower end of the range depending on considerations such as nonresident borrowers' exposure to employment stresses relative to residents, as well as practical issues associated with foreclosing on borrowers who may reside outside the country.

Pool originator adjustment

93. Mortgage loan performance typically reflects the quality of a lender's origination and underwriting processes. Therefore, as part of our credit analysis, we assess mortgage originators in a given country based on the historical performance of their loans, when available, and our views of the quality of their loan origination and underwriting processes. We also assess the originator's policies and procedures, including how they have changed over time. Our archetypical pool is based on the assumption of prudent underwriting of the mortgage loans, and we reflect our assessment of an individual originator relative to the archetype accordingly.

94. The criteria apply a pool-level adjustment factor to reflect the observed historical performance of various pools from different originators or lenders within each jurisdiction. Therefore, this adjustment typically also reflects our projections of variations in future delinquencies/arrears relative to the archetype. The factor may vary from transaction to transaction for the same originator as relevant.

95. Also, servicers are not always able to report loan-level data on certain characteristics that we typically consider in our analysis, and, in some cases, the information received at closing is not updated for surveillance. The analysis of originator and servicer risk includes the assessment of the underwriting policies regarding such loan and borrower features (including data provision), and may result in an originator adjustment under these criteria. As such, factors that we include to determine the originator adjustment tend to be country specific.

Small-pool adjustment

96. For pools that--unlike the archetype--initially contain fewer than 250 loans, the criteria apply an adjustment as a multiple to the foreclosure frequency. The adjustments are typically those indicated in chart 5 below.

Loan-specific risks

97. Our research shows that when a pool is sufficiently granular, the risks attached to an individual loan are mitigated at the pool level. Nonetheless, the analysis focuses on any material concentration in a few loans, which may result in a pool-level adjustment.

Global Approach For The Analysis Of Loss Given Default - Calculating The Loss Severity Of A Mortgage Pool

Property valuations, valuation haircuts, and valuation indexation

98. The calculation of recovery proceeds from a property's sale reflects the property market's state at the time of analysis. Typically, we are provided with the original property valuation recorded at loan origination. We therefore index this valuation through the date of our analysis, based on updated index information available. The indexation method typically uses publicly available information on house prices in the relevant jurisdiction. However, depending on the circumstances, we may not fully reflect all changes. In that case, depending on the level of availability of house price data, we adjust the indexed value to incorporate our forward-looking view on house price volatility.

99. The indexation logic is determined on a jurisdictional basis considering the indices available and the performance history of the local market. The same indexation logic is used for the purpose of calculating the CLTV of a loan in order to calculate default frequency assumptions.

100. Where updated property valuations are provided, we consider the process by which such values were derived, and how they compare to the original indexed values, to determine the appropriate property valuation to be used in our analysis.

101. The criteria also typically calls for a country-specific reduction (or haircut) of the valuation on a property, for valuation methods that deviate from that of the archetype. The haircut is typically up to 15% for non-full appraisals. In addition, where specific property risks exist (such as the risk of an atypical decline in property value), the criteria provides that additional haircuts may be applied to a property valuation. The property valuation used in the analysis is first haircut, where applicable, and then indexed.

102. For jurisdictions where an index or other source of changes in house prices is not available or, in our view, too limited to perform the analysis described above, we adjust the market value decline assumptions used to determine loss severity, as described further below.

Assessment of property over/under-valuation

103. We typically quantify over- or undervaluation embedded in the property value used to calculate the estimated loss severity related to a loan, and reflect this over- or undervaluation in the calculation of market-value decline. To assess the degree of over- or undervaluation in the national or regional property market, we use various data sources that are available, including from the Organization for Economic Co-operation and Development (OECD), FHFA, or other national data providers; expert opinions; and independent research. Endogenous and exogenous factors influence changes in property prices, reflecting macroeconomic trends and variable supply and demand.

104. We generally estimate the level of over- or undervaluation of the relevant property market by comparing the prevailing house-price-to-income ratio against its long-term average. This assessment is informed by an analysis of national data on current price to income (disposable household income, or GDP/capita) compared to the long-term average price to income (PTI). The specific method employed to estimate the over- or undervaluation percentage used in our rating analyses in a given country reflects our qualitative views of the available quantitative data, as well as our medium-term expectations.

Market value decline assumptions and repossession market value decline calculations

105. We apply a market-value decline (MVD) stress to the indexed property valuation. To account for the effect of over- or undervaluation, the MVD of a repossessed property (Repo MVD) is calculated using the formula below, based on the rating-specific input variables shown in table 3 below. Absent any over- or undervaluation, the resulting Repo MVD is as shown in the last column.

106. Repo MVD = 1 – [1 – (Fixed MVD +/- percentage of over/undervaluation x over/undervaluation)] x (1 – FSD)

• Fixed MVD is a fixed recessionary MVD. It represents a rating-specific stress applied to a property value.
• FSD is a forced-sale discount factor. Among other factors, foreclosed properties may sell at a discount, due to the stigma that repossession creates. The FSD is larger at lower rating levels and smaller at higher rating levels. This is because, in a more severe recession, a greater proportion of all property transactions contributing to the overall index comes from distressed sales.
• Percentages of over- or undervaluation limit the reduction of the Repo MVD in an undervalued property market. We use a constant percentage of any undervaluation across all rating levels. By contrast, the adjustment for overvaluation is linked to a particular rating level, and is highest for a 'AAA' rating. This is because a 'AAA' issue rating typically reflects greater stability than other ratings, relative to a long-term trend in property values. The degree of over/undervaluation in a property market is only partly reflected in the Repo MVD formula to account for our view that property values during a housing cycle in part reflect the current valuation of the property market, while reverting to the long-term average in the longer run.
• The Repo MVD may be capped at 75% for a given market.

Table 3

Property Market Adjustments For Calculating Repossession Market-Value Decline Modeling Assumptions
Rating level	Fixed market-value decline (%)	Percentage of overvaluation added (%)	Percentage of undervaluation deducted (%)	Forced-sale discount (%)	Repossession market-value decline, absent over/undervaluation (%)
'AAA'	40	50	(20)	10	46
'AA'	36	43	(20)	11	43
'A'	28	36	(20)	12	37
'BBB'	23	30	(20)	13	33
'BB'	19	25	(20)	14	30
'B'	15	20	(20)	15	28

107. In certain circumstances, the Fixed MVDs and/or FSD for a given jurisdiction may be different from those indicated in table 3 at one or more rating levels based on historical observations. For example, historical peak-to-trough house price data may show declines in excess of those assumed under the related stress scenario for that time period, after taking into account any adjustments for over- or undervaluation. In these cases, we calibrate the Fixed MVD based on observed declines, taking into account the related macroeconomic stress. For example, if we observe a certain decline in a moderate economic downturn, proportionately, the 'AAA' Fixed MVD will be higher. In addition, where we don't have data on changes in house prices through an economic cycle but perceive high risk such as countries with very high MMAs, we may apply increased Fixed MVDs and/or FSDs.

Repo MVD adjustment factors

108. We apply additional adjustments to the Repo MVD to reflect the potential for greater house price volatility relative to that assumed for the archetypal pool.

109. For example, we typically adjust the Repo MVD for property values that exceed a country-specific jumbo valuation limit (defined at the relevant geographic level-–national, regional, city, etc.) on an increasing scale and using a function that typically caps the adjustment at 1.2x-1.3x. Properties with higher valuations could experience higher loss severities, due to their smaller and less-liquid market. The country-specific jumbo valuation limit is generally based on the prevailing average house price, also considering house price distribution. In determining whether a valuation is jumbo, the modelling typically looks at the fully indexed valuation of the property.

110. Where relevant, we may also adjust the Repo MVD for certain properties to address the potential increased volatility in prices on account of other features such as location, property type, housing sector or loan denomination. For example, we may adjust the Repo MVD for properties located in certain geographic locations to reflect the view that there are potential demand/supply pressures, and geographic and structural concentration risks that may lead to more volatility in resale values (e.g., high-density, inner city apartments in some countries). In these cases, the adjustment factor typically ranges from 1.25x to 1.45x.

111. Fundamental to our MVD approach is our view that we adjust our assumptions for housing market volatility as necessary. This is generally achieved via indexation and the over- or undervaluation assessment. For jurisdictions where an index or other source of changes in housing price indexes (HPIs) is not available, or has material limitations, we instead adjust the Repo MVD of all loans in a pool to account for the potential impact of indexation and over- or undervaluation by a multiple. We typically adjust the Repo MVD by a multiple within a range of 1.0x-1.5x. The multiple applied in a particular jurisdiction is based on our consideration of the following factors, which may change over time:

• The overall level of perceived risk in the market as reflected in the MMA. Typically, the higher the MMA, the higher the multiple applied.
• Our assessment of house price volatility in the market. The greater our expectations of volatility, the higher the multiple applied.
• Supply and demand dynamics such as trends in housing starts, mortgage credit, interest rates, the impact of government intervention, or any relevant market distorting factors.

Foreclosure costs

112. Foreclosure costs incorporate legal and other expenses associated with enforcing security on a mortgaged property. Our assumptions are generally based on country-specific data obtained through investor reports, discussion with servicers and legal counsel, or other available sources. Foreclosure cost assumptions are typically reflected as: (i) a fixed amount plus a percentage of the distressed property value (i.e., the property value obtained after the application of the Repo MVD at the relevant rating level); or (ii) a percentage of the loan balance, depending on local practices and data presentation.

113. Where relevant in the local market, costs associated with property taxes and home insurance costs are separately accounted for.

Accrued and unpaid interest

114. Unpaid interest on defaulted loans that accrues through to the point of foreclosure (foregone interest) is typically incorporated into a loan's loss severity in the following situations:

• The servicer is contractually obligated to advance delinquent borrower interest payments through liquidation unless such advance is deemed non-recoverable from anticipated loan proceeds; or
• The rating analysis does not use cash flow models, such as for certain synthetic transaction structures.

115. In the cases above, the interest rate used for the purpose of computing the interest accrual typically reflects the yield on the defaulted assets, and varies by product type and original, current, and maximum rates.

116. Where accrued interest is not included in the loss severity calculation, the impact on a transaction's credit enhancement is addressed in our cash flow analysis by modelling the corresponding negative carry, using stressed interest rate assumptions.

Foreclosure period/liquidation timeline

117. Foreclosure period assumptions, which represent the estimated time to repossess and sell a property upon a default, are specific to each country, and reflect the typical time necessary for judicial proceedings, where applicable, and any other likely delay. We estimate those liquidation timelines based on historical data reflecting current and stressed observations or by benchmarking to comparable jurisdictions where country specific data is not available.

118. Foreclosure period assumptions may differ within a country by region, property type, whether the security is a first or second lien, other applicable law, or other considerations, where relevant. For example, in certain jurisdictions, the regime applicable to investment properties differs from that of owner-occupied property when foreclosing upon them.

119. Foreclosure timelines are used under these criteria to evaluate the total amount of accrued and unpaid mortgage interest, as discussed in the above section. Corresponding assumptions are also used as an input in our cash flow analysis to model the duration of corresponding negative carry before recovery proceeds are received.

Loss severity floors

120. To address any potential idiosyncratic risk on pools with very low LTVs that may not otherwise be captured by our assumptions, we floor the pool-level loss severity at 2%. Where available historical performance data or other relevant considerations indicate a more granular approach is relevant, we may apply loss severity floors at the loan level. However, we may go below the floors for loans that carry mortgage insurance or guarantees.

Mortgage loan insurance and guarantees

121. If some or all of the mortgage loans in a portfolio include a mortgage loan insurance or guarantee, then our global mortgage insurance criteria apply (see "Methodology For Assessing Mortgage Insurance And Similar Guarantees And Supports In Structured And Public Sector Finance And Covered Bonds," published Dec. 7, 2014). This excludes country specific guarantee schemes.

Other loss coverage adjustments

122. Where relevant, our loss coverage analysis may also consider the impact of other market-specific considerations on mortgage defaults and severities. For example, we may increase loss coverage if third-party reviews of loan data indicate variations to our expectations, or if we consider the likelihood that an entity is able to execute on representation and warranty repurchase obligations to be low.

Minimum Credit Enhancement Levels And Other Considerations

Credit enhancement and subordination floors

123. The criteria establish rating-specific minimum credit enhancement levels applicable for assigning ratings at issuance of a new transaction. These are set at 4.00% for 'AAA' ratings and 0.35% for 'B' ratings, subject to interpolation for the levels in-between. These credit enhancement (CE) floors are based on our view that there are limits on the predictability of mortgage loan performance. The 4.00% minimum credit enhancement for 'AAA' corresponds to 25x leverage. We believe that leverage above that level creates vulnerabilities that are inconsistent with the degree of creditworthiness associated with a 'AAA' rating.

124. Moreover, the minimum credit enhancement levels can't be funded solely through "soft" credit enhancement (commonly called "excess spread"), and the minimum amount of "hard" credit enhancement supporting a 'AAA' rating backed by residential mortgage loans is 2.5% (2% for 'AA' and 1.5% for 'A'). Hard credit enhancement generally includes such sources as subordination (which may include the benefit of mortgage insurance), overcollateralization, letters of credit, and reserve funds.

125. The analysis of minimum credit enhancement levels considers the specificities of each transaction. For example, if a transaction is exposed to negative excess spread--that is, loan interest at a pool level is lower than the cost of funds--this is accounted for (i.e., deducted from available credit enhancement) in our determination of CE floors. Likewise, if a transaction benefits from overcollateralization, a reserve, or other form of support that is either dedicated to cover specific structural risks in a transaction, unrelated to the pool's credit quality, or is otherwise relied upon for that purpose in our rating analysis, then its amount is typically deducted from available credit enhancement when determining if relevant floors are met. Examples include a commingling reserve, or overcollateralization that supports asset and liability mismatches in a structure.

126. In addition, when assigning ratings at transaction issuance, where relevant in the absence of other mitigating factors, out-of-model credit considerations are made to determine subordination floors typically for structures that feature pro-rata payments from inception (see table 4).

Table 4

Subordination Floors For Pro Rata Structures
Maximum potential rating	Subordination floor parameters
'AAA'	Highest balance loan liquidated at the greater of 50% loss severity and the applicable ‘AAA’ loss severity, plus the higher of the next nine-largest loss exposures at the 'AAA' loss severity and 20%
'AA' category	Highest balance loan liquidated at the greater of 50% loss severity and the applicable 'AA' loss severity, plus the higher of the next seven-largest loss exposures at the 'AA' loss severity and 20%
'A' category	Highest balance loan liquidated at the greater of 50% loss severity and the applicable 'A' loss severity, plus the higher of the next five-largest loss exposures at the 'A' loss severity and 20%
'BBB' category	Highest balance loan liquidated at the greater of 50% loss severity and the applicable 'BBB' loss severity, plus the higher of the next three-largest loss exposures at the 'BBB' loss severity and 20%
'BB' category	Highest balance loan liquidated at the greater of 50% loss severity and the applicable 'BB' loss severity, plus the higher of the next largest loss exposure at the 'BB' loss severity and 20%
'B' category	Highest balance loan liquidated at the greater of 50% loss severity and the applicable 'B' loss severity.

Originator insolvency, commingling, and set-off

127. The analysis of commingling and/or set-off risks that can result from an originator's or servicer's insolvency may use modeling to produce estimates of any exposure, after consideration is made of any structural mitigants.

128. For example, our legal analysis may conclude that the insolvency of the collection accountholder could result in a loss of funds deposited in the account. This analysis considers the degree to which a collection accountholder's insolvency affects the cash flow from the residential loan pool, if the collection account is not in the name of the issuer. The amount at risk depends on the timing of scheduled payments from borrowers, the frequency of transfers into the transaction account, and the level of prepayments.

129. If the loan originator (or servicer) is also a deposit-taking institution, modeling considers the possibility that borrowers with deposit accounts may set off these amounts against their outstanding mortgage loans. For example, we may derive an estimate of the set-off exposure from the mortgagees' highest historical deposit account balances on the originator's or servicer's books.

130. The modeling approach may also capture set-off risks from any borrowers that are also employees of the originator. This is because amounts owed to them--such as unpaid salaries, pension benefits, and subsidies--could offset payments due on their mortgage loans. As a result, the modeling approach typically treats the full amount of any employee's mortgage loan balance as entirely set off and all set-off amounts as principal losses.

Surveillance

131. Our analytical approach to surveilling transactions backed by residential mortgages that fall within the scope of these criteria reflects our view of change in the credit risk inherent in a mortgage pool over time. In our surveillance, our focus is generally on a subset of variables that we consider to be the most likely to materially change, and which reflect the credit quality, credit enhancement, and other relevant factors that drive the determination of ratings when analyzed in conjunction with observed and expected performance.

132. Our surveillance analysis incorporates observed pool performance relative to our expectations and reflects our forward-looking view of performance relative to the credit enhancement available at a given rating level. In applying the methodology and assumptions described in these criteria, our surveillance analysis considers the collateral's performance trends and dynamic characteristics as well as significant shifts in the pool's collateral composition to determine if updates to our most recent assumptions are warranted.

133. At the time of issuance, a pool of mortgage loans typically has very little, if any, observed performance data available. As such, our new issuance analysis primarily infers future collateral performance from the loan and borrower characteristics at the time of origination. As a pool of mortgage loans seasons, and default and loss patterns begin to emerge, those patterns are considered in determining the assumptions we apply in our surveillance analysis. For example, in our surveillance, we analyze projected foreclosure frequency based on the original loan attributes as well as the observed payment history. Further, the performance history informs the qualitative factors that were assessed at the time of our initial rating and how relevant any initial adjustment remains given our outlook for the collateral pool at the time of surveillance. For example, we may neutralize the originator adjustment factor applied in our analysis over time as performance history evidences the quality of the originations in the pool.

Appendix I: Determining The MMA

MMA methodology: building blocks and scoring

134. Our MMA analysis in a given jurisdiction consists of an assessment on a six-point scale ranging from "very low risk" to "extremely high risk." The MMA scoring assesses the following three core components:

• Economic risk;
• Mortgage industry risk; and
• Lender recourse provisions.

135. We first combine our assessments of the relative economic and mortgage market risk factor scores and then overlay our assessment of lender recourse. The resulting MMA is used to calibrate foreclosure frequency assumptions for the archetypal pool in the country, within a given range at the 'AAA' rating level and at a minimum level for 'B' ratings. A review of relative performance data and their availability, combined with any other relevant factors, helps determine where within a set range the final 'AAA' assumptions fall for the jurisdiction in question.

136. Table 5 below shows the steps for the MMA methodology framework.

Table 5

MMA Methodology Framework
Factor	Sub-factors
Step 1: economic risk
A. Economic resilience	Economic structure and stability, macroeconomic policy flexibilty, and political risk
B. Economic imbalances	Expansionary phase, private-sector credit growth, residential real estate prices, equity prices, and current-account balance and external debt position; or correction phase and expected impact on the banking sector
C. Credit risk in the economy	Private-sector debt capacity and leverage, lending and underwriting standards, payment culture and rule of law
Step 2: mortgage industry risk
A. Institutional framework	Banking regulation and supervision, regulatory track record, and governance and transparency
B. Competitive dynamics	Risk appetite, industry stability, and market distortions
C. Market sensitivity	Unemployment variability and welfare support levels
Step 3: lender-recourse provisions
Assessment of lender recourse provisions (full vs. limited)	In practice or by law, lenders may predominantly only rely on the mortgaged property as collateral to repay the loan in the event a borrower defaults (we refer to this as limited recourse vs. full recourse in other jurisdictions)
Final step: 'AAA'
Determination of 'AAA' foreclosure frequency assumptions from applicable range/determination of 'B' case	Relative mortgage performance and data availability and other factors help determine where, within the 'AAA' range for a given MMA score, the 'AAA' foreclosure frequency assumption will be. Current and expected performance help determine the foreclosure frequency 'B' case.
MMA--Mortgage market assessment.

137. The assessments of economic risk and mortgage industry risk are each determined from the average of the risk scores assigned to the associated factors listed in table 5. Each factor is individually scored within a range from 1 to 6, with a score of "1" corresponding to "very low risk" and "6" corresponding to "extremely high risk," as shown in table 6 below.

Table 6

MMA Factor Scores
Relative risk assessment	Risk score
Very low risk	1
Low risk	2
Intermediate risk	3
High risk	4
Very high risk	5
Extremely high risk	6
MMA--Mortgage market assessment.

138. The average of the economic risk and mortgage industry risk scores then forms the initial MMA, before adjustments for consideration of the provisions for lender recourse in the applicable market. Each score corresponds to a relative risk assessment ranging from "very low risk" to "extremely high risk," again as shown in table 6 above.

139. Depending on our assessment of lenders' recourse to borrowers' assets besides the mortgaged property, this increases the country's MMA typically by at least one category (e.g., from a score of '2' to at least a score of '3', or higher) to account for limitations in recourse (by law or in practice). The result is the final MMA, which is used to calibrate foreclosure frequency assumptions.

Step 1 of MMA methodology: evaluating economic risk

140. For each jurisdiction, the MMA methodology considers the relative levels of "economic resilience," "economic imbalances," and "credit risk in the economy," which together capture the relevant economic risks that are relevant to assess mortgage risk.

• Our analysis assesses economic resilience by rank ordering the wealth of an economy, as primarily measured by GDP/capita; a wealthier economy has more capacity to absorb economic shocks. Where GDP is heavily influenced by atypical factors (an example may be significant oil production in a country), we may adjust the score to ensure that the rank ordering is still relevant, considering the population segment in the country's archetypal pool.
• We consider relative levels of economic imbalances, depending on whether an economy is expanding or correcting. For example, if the economy is expanding, we look at the pace of private sector credit growth against the pace of residential house price growth, which may signal a credit fueled house price bubble. If an economy is contracting, we assess the likely length and impact of the contraction.
• Our assessment also considers relative levels of credit risks in the economy by looking at private sector debt capacity and leverage, the standard of underwriting for residential mortgages, and payment culture and rule of law. Private sector debt capacity and leverage looks at private sector credit/GDP, scaled by GDP/Capita (i.e., higher leverage presents more risk in relatively less wealthy economies); adjustments are made if GDP understates or overstates the relative wealth of the archetypal residential mortgage borrower. We assess the relative levels of underwriting of originators and practices of servicers and also consider factors--such as the percentage of mortgage market lending over high (e.g., 80%) LTVs; the average indexed LTV of the overall mortgage market; the number of factors typically considered for mortgage credit assessments, including, for example, moderately conservative cash flow requirements and collateral values; product types; the scale of non-prime lending; and lender concentration risk--all as relevant to mortgage market risk. Where we do not view the underwriting standards as at least moderately conservative, we elevate our assessment of credit risks accordingly.
• Finally, we consider the payment culture and rule of law in the given jurisdiction, including creditors' rights, the predictability of the legal framework, and applicable bankruptcy law. For instance, we may look at external indicators such as the World Bank governance data, and, unless we view the payment culture and rule of law to be at least moderately strong, we elevate our assessments of credit risks accordingly.
• In order to determine each individual factor score contributing to the overall economic score in our MMA framework (see table 5), our analysis typically uses as a starting point the same factor scores as these are determined within our Banking Industry Country Risk Assessment (BICRA) analysis for the relevant country. This is because the same factors that generally affect credit risks in the banking sector are also typically relevant for the analysis of residential mortgage risks, and therefore our MMAs.
• However, if the drivers of the economic risk score are less relevant for the analysis of mortgage risk, we may make adjustments. For example, with regard to the assessment of "economic imbalances" and "credit risk in the economy," the MMA analysis of these factors typically excludes any BICRA adjustments for commercial real estate prices and/or sovereign government credit stress. This is because these components are typically less correlated with residential mortgage performance (e.g., commercial real estate market), and/or already accounted for elsewhere in our analysis (e.g., sovereign credit stress addressed through Rating Above the Sovereign criteria) and therefore less relevant for the MMA.
• Similarly, the assessment of each individual sub-factor from the BICRA analysis considers the relevance of each such factor as a driver of mortgage market risk in the jurisdiction, as opposed to banking sector risk. This is because some factors may influence economic risk in a way that is more or less relevant or pronounced for the analysis of mortgage risk. In doing so, our analysis reflects all relevant considerations made above, as they pertain to the analysis of residential collateral in the country. For a given country, this may therefore result in an assessment for an individual sub-factor of economic risk (see table 5) that differs from that made as part of our BICRA analysis in such country.

Step 2 of MMA methodology: evaluating mortgage industry risk

141. Our assessment of mortgage market industry risk for any jurisdiction focuses on three factors: the relative strength and stability of the market's "institutional framework," any impact stemming from the "competitive dynamics" of the local mortgage lending industry and finally the "market sensitivity" of expected borrower default rates to changes in macroeconomic conditions such as unemployment.

142. As with our assessment of economic risk in step 1, our approach recognizes that many of the factors that generally influence banking risks are generally equally important in assessing mortgage market industry risks. Therefore, our framework here again typically uses as a starting point the outputs of the application of our BICRA criteria in the relevant jurisdiction. For example, our assessment reflects the quality and effectiveness of bank regulation, the track record of authorities in managing financial sector turmoil, and the competitive environment of a country's banking industry, including the industry's risk appetite and its structure.

143. However, our MMA of industry risk also recognizes that default risks of residential mortgages are sensitive to unique features. Therefore, the mortgage industry risk score for a country also assesses sensitivity to changes in unemployment by looking at historical unemployment volatility and by assessing levels of social welfare support, the impact of different personal insolvency regimes, and differing social perceptions associated with defaulting on mortgage debt.

Institutional framework

144. We start with the assessment of the institutional framework score, as described in the relevant BICRA analysis for a given country.

145. Our analysis of the relative strength of the institutional framework primarily considers the current policy framework and regulations for lending, including an analysis of institutional initiatives and regulatory track record; it also focuses on the extent to which the authorities in a given jurisdiction effectively and transparently oversee prudential lending standards.

Competitive dynamics

146. The assessment of competitive dynamics considers the mortgage lending industry risk appetite, its stability, and the impact of market distortions.

147. As with the institutional framework, the scoring of competitive dynamics also starts with the assessment of banking industry risk appetite in the BICRA analysis for the relevant jurisdiction. We believe the same key factors typically influence residential mortgage underwriting and servicing.

148. However, the MMA does not always consider the same factors in relation to industry stability or market distortions, where the factors do not speak to mortgage industry risk. On the other hand, where relevant the MMA analysis reflects in our scoring, any risk in a country that more specifically pertains to the underwriting of mortgages. In addition, our assessment also considers the mortgage lending industry in its entirety, including non-bank financial institutions (NBFIs).

149. Risk appetite considers profitability, compensation culture, product mixes, and portfolio growth. Distortions typically include factors such as high government share in lending, the level of central bank control over mortgage rate setting or that of government-sponsored entities.

Market sensitivity

150. The initial market sensitivity score is determined by an assessment of the relative levels of unemployment variability and social welfare support in a country (see table 7). Typically, in countries with higher levels of welfare support, the impact on mortgage default risk is expected to be less, all else being equal, for a given change in unemployment, than in countries with lower levels.

Table 7

Assessing Market Sensitivity To Unemployment Change Based On Welfare Support
Unemployment change score	1	2	3	4	5	6
Welfare support score
1	1	2	2	3	3	4
2	2	2	3	3	4	5
3	2	3	3	4	5	6

151. Unemployment variability is typically assessed by looking at the relative volatility of historic unemployment data in a country as published by the World Bank, adding S&P Global Ratings' medium-term projections. In doing so, we typically do not include data that is reflective of a structural stage in the development of an economy that we believe is no longer relevant.

152. A score on the scale of '1' to '6' is assigned for the unemployment score based on the categorization in table 8, which we update over time as necessary to preserve a meaningful distribution.

Table 8

Scoring Unemployment Change
Standard deviation of historic unemployment data	<=1	<=2	<=3	<=4	<=5	>5
Unemployment change score	1	2	3	4	5	6

153. Similarly, social welfare support in a country is assessed as either "high," "medium," or "low," with corresponding scores of '1,' '2,' or '3.'

154. Typically, a country characterized as having a high level of support exhibits welfare spending that is above the OECD average as a percentage of GDP, and has relatively generous unemployment benefits. This typically includes countries with relatively high levels of tax. In addition to a high-level assessment of GDP percentage spent on welfare support, we review the characteristics of employment social insurance, including coverage and duration and, for example, the percentage of the last salary that is covered. This analysis considers other sources of funds that may become available to a borrower upon becoming unemployed, such as withdrawals from regulated funds otherwise not readily accessible, as well as other forms of support in a given jurisdiction, where relevant.

155. A country that spends below the OECD average of GDP on welfare and has limited coverage and duration of unemployment social insurance is typically assessed as "low." A country that neither characterizes as "high" or "low" is assessed as having a "medium" welfare support score.

156. In addition, in certain circumstances, the overall score for market sensitivity may be adjusted up or down by one point (for example, where we have seen evidence that employment levels displayed resilience in the face of GDP declines).

157. Finally, in some countries, we have observed other factors that contribute to positive performance such as a particularly strong social stigma attached to personal bankruptcy or a default on individual debt, high levels of family support, or household savings levels. All else equal, this may lead to a lower propensity to default. Certain legal frameworks also create a disincentive for borrowers to enter into personal bankruptcy (for example, the length of discharge periods), which can also reduce the propensity to default. Giving consideration to both of these circumstances may improve the overall score by one point. Conversely, in some jurisdictions, the very opposite may be true, and in these instances, we may worsen the overall score by one point.

Step 3 of MMA methodology: assessing lender recourse provisions

158. In some jurisdictions (in practice or by law), lenders predominantly only rely on the mortgaged property as collateral to repay the loan in the event a borrower defaults. As a result, borrowers may be less incentivized to maintain mortgage payments when house prices decline; we refer to this as limited recourse. We have observed that this has a significant impact on borrower behavior versus full-recourse markets and is therefore a key differentiating factor that we consider when rank ordering MMA. All else being equal, we expect a MMA for any given country to be at least one category weaker for limited- versus full-recourse markets.

Final step of MMA methodology: determining 'AAA' and 'B' foreclosure frequency

159. Finally, we consider relative mortgage performance, data availability, and any other pertinent factors to help determine where a country's 'AAA' foreclosure frequency assumption is within the 'AAA' range for the relevant MMA score. We also review current and expected performance to determine our 'B' foreclosure frequency case.

160. For jurisdictions where we have limited data, the weaker the MMA, the more likely it is that our foreclosure frequency assumption is at the higher end of the applicable range. If we have limited data, but good reason to believe that the product mix and economic situation are very similar to a peer where we do have access to more extensive data, then we may likely set the level corresponding to that peer.

161. When reviewing performance to help assess the relevant percentage in the 'AAA' foreclosure frequency range for a given country, we typically look at the following:

• Performance through economic cycles. At the least, we look at dynamic arrears evolution over time. Where available, we consider default data over time. For a given economic cycle, how does the performance of a given jurisdiction compare to others with the same or close MMAs? Where historical data indicates that relative performance has been strong, and we have evidence to believe it will continue to be so, we tend to fall on the lower end of the foreclosure frequency range.
• Expected lifetime performance for archetypal mortgage pools. If we don't have default data, we may approximate with an arrears roll rate analysis. If we can make an assessment of what lifetime performance for an archetypal pool may be over a given economic scenario, this helps to inform our foreclosure frequency assumptions within the range.
• We compare to other jurisdictions with similar MMA scores, but account for differences in product mix and underwriting standards when making this comparison. For example, if a given jurisdiction had performance that appears worse than its peers, but typically lends at higher LTVs, we need to consider our LTV curve to make sure we are not double counting the impact of a higher LTV market.

162. For the 'B' foreclosure frequency assumption, we typically do not assign a percentage lower than the limit set for the corresponding MMA. We may go higher, however, and we typically consider the following when making this assessment:

• We consider the base-case default and loss assumptions used in our bank rating analysis.
• We consider performance we are seeing today in mortgage pools, accounting for the product mix relative to the archetypal pool.
• We consider our near-term economic forecasts, and factors such as our expectations for mortgage lending rates, inflation and wage inflation, unemployment long term versus short term, house prices, availability of credit, and consumer leverage.

Appendix II - Country-Specific Criteria Application Considerations

Appendix IIa: Country-Specific Criteria Application Considerations--Japan

163. This appendix provides additional information regarding the application for Japan of these residential mortgage criteria and the cash flow criteria (see "Global Framework For Payment Structure And Cash Flow Analysis Of Structured Finance Securities," published Dec. 22, 2020). It is also intended to be read in conjunction with the relevant Sector and Industry Variables Report for Japan (the Japan SIVR; see "Sector And Industry Variables: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Japan)," published Dec. 16, 2022.) A SIVR is a publicly available criteria-related publication that describes sector, industry, asset class, or regional variables that we expect to periodically update to reflect our views on changing macroeconomic and market conditions. SIVRs are not criteria because they do not establish a methodological framework for determining credit ratings.

Details Of Further Considerations When Calculating The Foreclosure Frequency Of A Japanese Mortgage Pool

Originator adjustment

164. Specific examples of the factors considered in determining the originator adjustment for a pool of Japanese residential loans typically include but are not limited to:

• Transactions without updated loan-by-loan (LBL) data available during surveillance until the weighted average seasoning of loans reaches five years;
• Borrowers in arrears in the case of transactions without updated LBL data during surveillance;
• Apartment loan borrowers who are insured by group credit life insurance;
• Borrowers with no income verification;
• Loans with lower documentation standards;
• Loans with a full valuation (instead of a lender's valuation);
• Interest only (I/O) or balloon payment loans;
• Selection bias to reperforming and restructured/modified loans;
• Owner-occupied loans financing mixed properties with more than 50% commercial occupation;
• Commercial borrowers in owner-occupied or condominium investment loans;
• Borrowers with adverse credit histories;
• Loans with an initial term to maturity that is longer than the market standard;
• Borrowers who are not Japanese citizens or do not have permanent resident status in Japan;
• Any other observed risk characteristics that are not adjusted for at a loan-by-loan level;
• Significant exposure to single obligors; and
• Changing in underwriting guidelines for which impact is not yet visible in performance data.

Loan to value (LTV)

165. LTV is calculated in a three-stage process. Stage 1: The original LTV (OLTV) is calculated using the original loan balance and the property valuation at the time of loan origination. Stage 2: The current LTV (CLTV) is calculated using the loan balance as of the portfolio cut-off date and the current indexed property value. Stage 3: The LTV is calculated by using the specific weighting of the OLTV and CLTV as described in the Japan SIVR. The OLTV and CLTV are based on the full loan balances secured on the property, including loan parts outside the asset pool and prior- and second-ranking balances, if any.

166. A loan's CLTV is the current balance of a loan, reflecting the actual principal pay down, divided by the indexed initial value or other updated value of the property. For transactions without updated loan-by-loan (LBL) data available during surveillance, we use a hypothetical current loan balance reflecting the scheduled amortization to calculate the CLTV.

Loan affordability: Debt-to-income (DTI)

167. In calculating the ratio of DTI, we incorporate any secondary income such as rental income from investment properties or a guarantor's income where appropriate.

168. For apartment loan RMBS, the primary source of repayment is the rental income from the apartment properties, rather than the borrower's salaried income. Therefore, a debt service coverage ratio (DSCR) adjustment is applied in place of the DTI adjustment.

Borrower occupancy status (investment properties)

169. The investment property classification includes second homes, investments in condominium units for lease, and investments in apartment buildings for lease.

170. A typical borrower of an apartment loan is a landowner who builds a rental apartment building for investment and/or inheritance tax purposes. Unlike owner-occupied prime mortgage loans in Japan, most apartment loans in the country are extended on a nonrecourse basis.

171. We apply an adjustment factor based on an apartment loan's actual DSCR (see the chart titled "Adjustment For Apartment Loans By Debt Service Coverage Ratio" in the Japan SIVR).

172. The borrower's capacity to pay for apartment loans relies more on rental income from the property than earnings from his or her job. Thus, we do not apply adjustment factors based on DTI, employment status, or loan seasoning (as these are described above or below) for Japanese RMBS backed by apartment loans.

Employment status

173. We assume all loans in the Japanese mortgage market are income verified. If the pool includes loans with no income verification, an additional risk will be captured in the originator adjustment.

174. For self-employed borrowers, we apply an adjustment based on the tenure of self-employed borrowers, subject to availability of such data.

175. For apartment loan RMBS, the major source of repayment is a rental income from the apartment properties, not a borrower's stable salary income. Therefore, adjustment for employment status is not applicable.

Geographic concentration

176. If a pool has significant geographic concentration risk that we believe is not sufficiently captured, we may capture it using the originator adjustment.

Seasoning (adjustment factors for loan seasoning)

177. Adjustments for loan seasoning are provided in table 4 of the Japan SIVR. The seasoned term of a refinancing loan includes the years elapsed on the previous (original) loan (i.e., the seasoned period prior to refinancing).

178. For transactions without updated LBL data during surveillance, we do not apply a seasoning adjustment factor. Not doing so mitigates risk arising from the lack of information about loans that are still outstanding in the pool, as well as the current outstanding amount of each loan.

179. For apartment loan RMBS, the primary source of repayment is not a borrower's stable income, but rental income from the apartment properties. Repayment history does not necessarily mean an improvement in credit quality, in our view. Therefore, we do not apply the seasoning adjustment factor.

Loans in arrears

180. When only pool-level performance data are available and no adequate data are available about which loans are delinquent during surveillance, this adjustment is not applied. Instead, the risk related to pool-level actual arrears is captured in the originator adjustment.

Payment shock (Loan product type)

181. Adjustments are applied to loans considered to have a potential for an increase in monthly payments. Loans where those principal payments start within a year after loan origination are not treated as loans with payment shock or interest only (IO) loans, because this is one of the standard mortgage products in Japan.

Second-lien loans

182. We apply an adjustment factor for second-lien mortgage loans as indicated in table 4 of the Japan SIVR. We differentiate a variable by the other lien loan's data availability.

183. Borrowers in Japan who are eligible for mortgages can borrow 100% of the mortgaged property's purchase price in general, so second-lien loans are not common in the Japanese market. Borrowers who want to borrow two loans, one with a fixed interest rate and the other floating, or a part of their required funds from the public financial sector/employee loan program at a lower interest rate use second-lien loans. In those cases, we believe the foreclosure frequencies of such borrowers do not differ from borrowers who purchase houses using first-lien loans only.

Small pool adjustment for apartment loan pools

184. In the case of apartment loan RMBS, we apply a small-pool adjustment for apartment loan pools in this section rather than a small pool adjustment described above in our residential mortgage criteria.

185. We define a sufficiently diversified Japanese apartment loan RMBS as one that has an effective loan count of 40 or more at a transaction's closing. We analyze and measure the effective loan count using the Herfindahl Index (see chart below) to assess the diversity of apartment loan RMBS. Specifically, the inverse of the Herfindahl Index is often used to count the effective number of loans over which the pool balance is distributed. We do not assign ratings to apartment loan RMBS with effective loan counts of less than 40 at the transaction's closing under this methodology.

Group credit life insurance adjustment

186. We apply adjustments for group credit life insurance if: i) borrowers do not take out group credit life insurance, ii) borrowers have a right to terminate it at their sole discretion, or iii) a transaction agreement stipulates that the transaction party terminates the group credit life insurance in the event of the originator's insolvency.

187. We may make an adjustment to the variables shown in the Japan SIVR to better reflect the pool's composition and transaction characteristics. For example, we may lower the variable in case of high weighted average foreclosure frequency (WAFF) for a pool as well as in case of mixed pools where one portion of the loans are from borrowers who have taken out life insurance, and one portion is not. Conversely, we may raise the variable in the case of a low pool WAFF.

188. Traditional apartment loans are not insured by group credit life insurance because of their nonrecourse nature. We therefore do not apply this adjustment to apartment loans.

Details Of Further Considerations When Calculating The Loss Severity Of A Japanese Mortgage Pool

Forced-sale discount (FSD)

189. On a case-by-case basis, we may increase or decrease the forced sales discount where there is sufficient information to support it being either higher or lower than envisaged using the standard calculation.

Current loan to value

190. For the purpose of the weighted average loss severity (WALS), a loan's CLTV is the current balance of a loan, reflecting the actual principal pay down, divided by the indexed initial value or other updated value of the property.

191. For transactions without updated LBL during surveillance, we use the hypothetical current loan balance, reflecting the scheduled amortization, to calculate a loan's CLTV.

Property valuations

192. For owner-occupied residential loans and condominium investment loans, the valuation of the collateral property uses available data in the order below. As the level of down payments relative to the purchase price is a more relevant driver to assess the default risk of a borrower, we give priority to the purchase price over the lender's valuation.

• Full appraisal
• Purchase price
• Lender's valuation

193. For apartment loans, the valuation of the collateral property uses available data in the order below. This is because investors usually refer both values indicated by the income approach and the cost approach.

• Full appraisal
• Whichever is lower of the purchase price or lender's valuation

Valuation haircut

194. A standard valuation method in Japan is a purchase price or a lender's valuation because a full appraisal is not common in the Japanese mortgage market. Given this, we generally do not apply a haircut based on the valuation type. Where relevant, we may make an adjustment to better reflect the pool's characteristics. For example, we may revise up or down a lender's valuation if they employ less or more conservative valuation methods relative to market standards.

195. Full appraisals are generally more conservative than purchase prices or lenders' valuations. If the pool includes loans with full appraisals, the correspondingly more conservative underwriting compared with market standards will be captured in the originator adjustment.

Mortgage guarantees

196. Some residential loans in Japan are secured by a mortgage guarantor or insurer. In such cases, we apply the methodology outlined in "Methodology For Assessing Mortgage Insurance And Similar Guarantees And Supports In Structured And Public Sector Finance And Covered Bonds," published Dec. 7, 2014.

197. In cases where a mortgage guarantor or insurer does not have an issuer credit rating, we take into consideration the possibility of recoveries from collateral properties if the guarantor creates a mortgage right to secure its recourse rights against the obligor.

198. In cases where a transaction structure has been established to transfer the amount recovered from the mortgage, we will review the effectiveness of the structure. If the structure effectively allows recoveries from the mortgage to be transferred to creditors, we incorporate the entire estimated amount to be collected from the collateral properties into our analysis. Where such a structure has not been established, we incorporate half of the estimated amount to be collected from the collateral properties into our analysis in cases where the parent company of a guarantor is a regulated financial institution. In all other cases, we do not incorporate recoveries.

Details Of Further Considerations When Calculating Minimum Credit Enhancement Levels In Japanese RMBS Or Covered Bond Transactions

Pro-rata pay structure and tail risk in JHF RMBS

199. The government-related entity (GRE) Japan Housing Finance Agency's (JHF) RMBS employs a pro rata pay structure from inception without subordination floors. To mitigate unexpected event risk in the tail end, in addition to the credit enhancement floors outlined above in our "credit enhancement and subordination floors" section, we apply in our analysis:

• A 1.2x multiple to the foreclosure frequency assumptions during pro rata; and
• A minimum loss assumption at each rating category, based on a certain number of loans defaulting and liquidating as outlined in the table below.

Table 9

Rating Category Minimum Loss Assumptions For Transactions With Fewer Than 100 Loans
Maximum potential rating	Minimum loss assumptions
AAA	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'AAA' loss severity, plus the higher of the next nine largest loss exposures at the 'AAA' loss severity and 20%.
AA category	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'AA' loss severity, plus the higher of the next seven largest loss exposures at the 'AA' loss severity and 20%.
A category	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'A' loss severity, plus the higher of the next five largest loss exposures at the 'A' loss severity and 20%.
BBB category	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'BBB' loss severity, plus the higher of the next three largest loss exposures at the 'BBB' loss severity and 20%.
BB category	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'BB' loss severity, plus the higher of the next largest loss exposure at the 'BB' loss severity and 20%.
B category	Highest-balance loan liquidated at the greater of 50% loss severity and the applicable 'B' loss severity.

200. We view the unique structure of JHF RMBS (including the removal of default loans, four-month delinquent loans, and loan-term modified loans prior to a beneficiary certificate trigger event, a trigger to switch to sequential from pro rata upon an occurrence of trigger events), along with the application of the additional minimum loss assumptions and WAFF adjustments described herein, as providing sufficient mitigants as described in our "credit enhancement and subordination floors" section. Therefore, subordination floors described above in table 4 of our residential mortgage criteria are not applicable to JHF RMBS.

Pro rata pay structure and tail risk in non-JHF RMBS

201. In non-JHF RMBS, the subordination floors described above in our "credit enhancement and subordination floors" section generally mitigate tail risk associated with pro rata pay transactions from inception. However, to compensate for any unexpected risks, we would in addition apply a 1.2x multiple to the foreclosure frequency assumptions in our cash flow analysis for transactions without updated LBL data during surveillance that are pro rata or may become so in the future.

Additional credit enhancement floors for apartment loan RMBS

202. Apartment loan RMBS sometimes contain very large loans in the underlying pools. To ensure rating stability in case a larger loan defaults, we establish a credit enhancement floor for apartment loan RMBS as shown in the table below. Transactions in higher rating categories should withstand a certain number of large-loan defaults in the portfolio. Based on this table, credit enhancement for 'AAA' rated RMBS backed by apartment loans must be larger than the aggregate balance of the five largest loans. After taking into account loss severity (i.e., recovery assumptions from the defaulted loans under a 'AAA' stress scenario), the 'AAA' rated securities should withstand a net loss of the 10 largest loans. We can consider soft credit enhancement, such as excess spread, if appropriate.

Table 10

Number Of Largest Apartment Loans With Default Exposure By Rating Category
	AAA	AA category	A Category	BBB Category	BB Category	B Category
Gross loss	Top 5	Top 4	Top 3	Top 2	Top 1	N/A
Net loss	Top 10	Top 8	Top 6	Top 4	Top 3	Top 2
N/A--Not applicable.

203. In some cases, we may adjust gross loss or net loss thresholds for each rating category by changing the number of loans to be covered. For example, we may lower a threshold when the remaining life of the transaction is short or in cases where the threshold breach would in our view quickly resolve.

Details Of Further Considerations When Applying Our Cash Flow Criteria To Analysis Of Japanese RMBS Or Covered Bond Transactions

204. In our rating analysis, we also analyze a transaction's payment structure and cash flow mechanics. This analysis uses our own quantitative models to assess whether the cash flows from the assets suffice, at the applicable rating levels, for making timely payments of interest and ultimate payment of principal (i.e., by or before the legal maturity date).

205. In our cash flow analysis, we use as inputs the pool-level WAFF and WALS as described in previous sections above, to reflect credit stress at each rating level.

206. During modeling, cash flow stresses test the credit and liquidity support the assets need. Our models consider any available structural support, such as cash reserves, liquidity facilities, and hedging arrangements.

207. In our surveillance of existing ratings, cash flow modeling may show that under the 'B' stress a particular tranche will miss interest payments or fail to repay the principal by or before the final legal maturity date. If this is the case, then our initial assessment, assuming all factors remaining the same, may be to consider lowering the rating on those securities to 'B-' or lower.

208. Depending on our view of a transaction's immediate cash flow position, the rating could move into the 'CCC', 'CC', or 'C' rating category, consistent with our ratings definitions, and in accordance with "Criteria For Assigning 'CCC+', 'CCC', 'CCC-', And 'CC' Ratings," published Oct. 1, 2012.

Default timing

209. The cumulative amount of defaults for cash flow modeling is equal to the pool's WAFF, whereas the cumulative amount of recoveries is calculated as (1 – WALS).

210. For certain structure types (for example, master trust transactions), we may run additional analysis to test the sensitivity of the liability structures to different default timing curves. In some cases, we may make an adjustment to the default timing curves to better reflect the pool's composition and transaction structure. For example, we may shorten the default curves where the mortgage portfolio's average remaining term gets shorter.

211. Foreclosure period assumptions represent the estimated time to repossess and sell a property upon a default. They also reflect the typical time necessary for judicial proceedings and any other likely delay.

212. Cash flow modeling considers the negative carry resulting from interest due on the rated liabilities during the foreclosure period.

213. The loss severity estimates used in the cash flow modeling are based on the loan principal and assume no recovery of interest accrued on the mortgage loans during the foreclosure period.

Interest rate scenarios

214. If a transaction features an interest rate cap--either on the note coupon or through a cap agreement with an external counterparty--upward interest rate stress assumptions exceeding the cap level may be unduly beneficial for the transaction's cash flow projection. In such events, we may apply a different interest rate stress to test sensitivity of the ratings in the absence of the cap.

215. Specific structural features may involve using additional cash flow stresses, such as alternative interest rate patterns or different default timing curves. For example, if we consider a transaction to have unique characteristics that lead to different sensitivity to interest rate scenarios, we may apply an additional scenario as well.

Basis risk

216. We adjust the interest spread modeled between assets and liabilities by applying different spreads over the life of a transaction as a haircut to the margin. The size of the spreads typically depends on the distribution of historical differences among indices, using the rating-specific values corresponding to the percentiles shown in the table below.

Table 11

Basis Risk Percentile Stresses
Rating category	Percentile (%)
AAA	95
AA	90
A	65
BBB	50
BB	40
B	30

Prepayment scenarios

217. Residential loan prepayments vary the amount of excess spread available and this may affect the absolute level of defaults exhibited in a transaction.

218. When analyzing the payment structure and cash flow mechanics of Japanese RMBS, we typically test the transaction's ability to withstand high and low prepayment scenarios as set out in table 11 of the Japan SIVR.

219. We may raise the prepayment assumptions if a pool's historical prepayment rates were higher than historical averages or if a transaction were particularly sensitive to prepayment risk (e.g., excess spread notes). We may also reduce prepayment stress in situations where long-term historical data support lower prepayment rate assumptions for a specific loan product.

Modeling of reinvestment rates and spread compression

220. We typically do not assume any revenues associated with any cash or securities the issuer holds. This is because a general practice in Japan is for issuers to hold any cash on the noninterest bearing account.

221. Also, we typically do not assume any spread compressions on the loan pool in our cash flow analysis. This is because risk-based pricing is not common in standard mortgage underwriting practice in Japan. As a consequence, all loan spreads are very similar in each pool.

Originator insolvency, commingling, and set-off

222. The analysis of any commingling or setoff risks that can result from an originator's or servicer's insolvency follows the application of the counterparty criteria (see "Counterparty Risk Framework: Methodology And Assumptions," published March 8, 2019). The counterparty criteria determine the maximum supported rating based on available mitigation of such counterparty risks. Mitigants may include a counterparty's commitment to implement remedies upon a downgrade, or structural factors such as the coverage of the risk through credit enhancement, demonstrated through the modeling of any exposure. In the context of Japanese RMBS transactions, government-sponsored deposit insurance also often mitigates commingling and setoff risks. If we find that deposit insurance fully mitigates the risk, the rating will not be constrained under counterparty criteria. Where our analysis relies on government-sponsored deposit insurance to cover commingling or setoff risks, we analyze the resulting sensitivity to sovereign risk under our criteria for rating above the sovereign (see "Incorporating Sovereign Risk In Rating Structured Finance Securities: Methodology And Assumptions," published Jan. 30, 2019).

223. Regarding commingling risk, the degree to which a collection account holder's insolvency affects the cash flow from the assets in a mortgage loan pool depends on the collection account's characteristics, if the collection account is not in the name of the RMBS issuer.

224. The amount at risk depends on the timing of scheduled payments from borrowers, the frequency of transfers into the transaction account, and the level of prepayments.

225. If a Japanese RMBS transaction satisfies the below five conditions, we assume that commingling risk at an originator deposit-taking institution in Japan is mitigated by the Deposit Insurance Corp. of Japan (DICJ), a government-sponsored insurer:

• The servicer falls under the jurisdiction of the deposit insurance law in Japan;
• The loan agreement specifies that loan transfers are allowed (borrower's recognition);
• The transaction scheme specifies that the deposit-taking institution entrusts the loans with the special-purpose company and continues loan servicing operations as a servicer as dictated by an administrative agreement;
• The servicer appropriately manages proceeds by separating the collection of securitized loans from its own funds; and
• The payments are denominated in Japanese yen.

226. For an originator that is a deposit-taking institution in Japan, we consider setoff risk to be mitigated by the DICJ up to the maximum insured amounts of the borrower's deposit amounts.

Modeling of senior fees and expenses

227. The modeling of all an issuer's foreseeable expenses uses stressed costs to reflect the need to replace the initial service provider.

228. The most significant portion of senior fees and expenses is related to the trustee, the servicer, and the backup servicer, if involved. Estimation of transaction maintenance fees usually includes a certain increase from the amount stipulated in the transaction agreement to account for a potential replacement of the trustee or servicer. We also examine the size of the transaction, the degree to which the structure is typical, and the levels of trustee fees in the market. The estimate of maintenance fees reflects our findings.

229. To reflect the likely cost of replacing the initial servicer, we set servicing fees at the higher of the contractual rate and the estimated rates depending on the size of the transaction and the appointment of a backup servicer as in table 12 of the Japan SIVR.

230. We may apply different assumptions from those detailed in table 12 of the Japan SIVR. For example, we may lower the level for a repackaged transaction to reflect easier/simpler tasks of the trustee/servicer. On the other hand, we may raise the level for the exotic transaction to reflect a difficulty of the trustee/servicer's replacement.

Additional costs for group credit life insurance

231. For transactions in which group credit life insurance premium payments are included in the waterfall, the assumption incorporates an expected increase in premiums from current premiums. This is because the group credit life insurance premiums rise as the ages of those borrowers in the insured group rise. Premium payments also rise with the ages of insured borrowers, in line with the seasoning of the loan pools. We assume increased premium payments in our cash flow analysis as set out in table 9 of the Japan SIVR.

232. In some cases, we may adjust the level of additional costs for group credit life insurance to better reflect a pool's characteristics and composition. For example, there might be cases where borrowers enter into new types of insurance that include wider coverage than typical insurance. In such cases, insurance premiums tend to be higher than existing ones, and therefore we might need to revise upward our assumptions for insurance premiums. In addition, there might be cases where some borrowers in the pool enter into life insurance while others do not. In such cases, we may assume premium payments for policyholders only.

Fixed-rate convertible loans

233. Many lenders in Japan often offer fixed-rate convertible mortgage loans to borrowers. In this type of mortgage loan, borrowers can choose either a fixed or floating rate when the initial fixed-rate period expires. We typically assume borrowers of such loans will opt to repeat the 10-year fixed-rate period in our cash flow analysis.

Details Of Further Considerations In The Surveillance Of Japanese RMBS Or Covered Bond Transactions

234. The characteristics of a mortgage loan pool evolve over time. For instance, some loans in a pool will default or prepay, and other characteristics of the pool will change. As a result, features initially consistent with the archetypal pool may deviate from those of the archetype over a pool's life.

235. Ongoing surveillance of the revised characteristics of individual mortgage loans allows the assessment of risks in less homogenous pools.

236. Rating changes resulting from surveillance emanate from changes in the performance of a pool and the analysis of pool performance data and trends. Changes in the ratings on supporting counterparties and in the rating on the sovereign can also influence our ratings on the notes.

237. For a given Japanese RMBS portfolio with changed pool characteristics, the surveillance analysis would use the same methodology and assumptions described in this document.

238. If loan-by-loan data are available only at the time of a transaction's closing, the transaction is exposed to a risk of the undetected changes in its pool composition during the surveillance period. For example, the weighted-average LTV of a mortgage pool containing loans with a wide distribution of original LTVs at closing may rise because of prepayments of loans with lower-than-average LTVs. This leads to possible adverse selection risk as well as an associated risk coming from a lack of information about which loans are outstanding in the pool and outstanding loan balance for transactions lacking updated loan-by-loan data in the surveillance period. We use the approaches below to address this risk.

• We apply a hypothetical current loan balance reflecting the scheduled amortization as the current loan balance at the time of the surveillance.
• We do not apply seasoning adjustments.
• We apply additional originator adjustments until the weighted loan seasoning of the pool exceeds five years.
• We capture a stress related to the pool level arrear in the originator adjustment.

239. For apartment loan RMBS, we always receive updated loan-by-loan data during surveillance, considering lower loan counts in the pool compared with transactions backed by residential or condominium investment loans.

Appendix IIb - Country-Specific Criteria Application Considerations--Korea

240. This appendix provides additional information regarding the application for Korea of these residential mortgage criteria and the cash flow criteria (see "Global Framework For Payment Structure And Cash Flow Analysis Of Structured Finance Securities," published Dec. 22, 2020). It is also intended to be read in conjunction with the relevant Sector and Industry Variables Report for Korea (the Korea SIVR; see "Sector and Industry Variables: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Korea), " published June 2, 2023). A SIVR is a publicly available criteria-related publication that describes sector, industry, asset class, or regional variables that we expect to periodically update to reflect our views on changing macroeconomic and market conditions. SIVRs are not criteria because they do not establish a methodological framework for determining credit ratings.

Details Of Further Considerations When Calculating The Foreclosure Frequency Of A Korean Mortgage Pool

Originator adjustment

241. Specific examples of the factors considered in determining the originator adjustment for a pool of Korean residential loans typically include but are not limited to:

• Origination and underwriting quality;
• Servicing quality;
• Originator specific performance to address idiosyncratic risks that are not captured in other adjustments;
• If we believe originator's LTV calculation is materially different from market standards or we believe inclusion of second and subsequent-ranking balances will significantly increase LTV level;
• If we believe the originator's practices on DTI calculation or requirements are significantly different from market standards, or the performance data suggests additional negative or positive adjustment;
• If we believe additional risk is not sufficiently captured in loan product adjustments;
• Unavailable data on loan and borrower characteristics (for example, LTV, DTI, income verification, restructurings, defaults, prior mortgage arrears, and property valuation method);
• If originators can't provide data at loan level for employment status, second homes or investment properties underwritten to borrower income, we apply an originator adjustment (see table 4 in the Korea SIVR). We typically size this originator adjustment by multiplying the applicable adjustment by our assumption of the proportion of loans with the relevant characteristic in the pool. We determine this assumption based on pool-level data, if available, the originator's underwriting policies, or market-level data, if representative;
• Loans to borrowers with negative credit histories;
• Recent changes in product offering, credit score process or underwriting guidelines, where the impact of which is not yet visible in performance metrics;
• Positive or negative selection not captured in other adjustments;
• Scenarios where the performance of a pool that has been sold deviates from our expectations for pools from that originator and for which a neutral originator adjustment is assumed;
• Weak representations and warranties of the loans in the transaction documentation and pool audit results;
• Dynamic or revolving asset pools;
• Any transaction's specific geographical concentration not captured by regional concentrations;
• We may analyze data from the issuer/servicer for reperforming loans on re-default rates stratified by forbearance type to calibrate the originator adjustment for such transactions; and
• Any other observed risk characteristics that are not adjusted for at a loan-by-loan level.

242. For covered bonds, we may reduce the foreclosure frequency to reflect an issuer's willingness and ability to continue managing the cover pool. This assessment considers in particular the following factors:

• The existence in the transaction documentation of a periodic test of the pool's credit quality;
• The rating on the issuer;
• The importance of the program in the issuer's funding mix;
• The frequency of issuance from the program;
• The number of different covered bond programs that the issuer runs;
• Whether the issuer differentiates the way in which it manages the pools backing such covered bond; and
• The eligibility criteria for the cover pools.

243. The criteria set out that we consider the potential changes of credit risk over time. For RMBS transactions backed by a pool whose assets change (e.g., by virtue of loan substitutions, product switches, or similar, or revolving), in determining the pool's weighted-average foreclosure frequency (WAFF) and the weighted-average loss severity (WALS), we consider the potential increase of credit risk over time as a result of changes in pool composition. We assess possible deterioration in pool composition based on the transaction's documented asset-eligibility criteria, the history of the originator and, in particular, any observed changes in origination, underwriting, and related performance.

Loan to value (LTV)

244. LTV is calculated in a three-stage process. Stage 1: The original LTV (OLTV) is calculated using the original loan balance at the time of the latest advance, typically at loan origination and the property valuation at the time of that advance. Stage 2: The current LTV (CLTV) is calculated using the loan balance as of the portfolio cut-off date and the current indexed property value. The CLTV calculation incorporates any applicable valuation haircuts. Stage 3: The LTV is calculated by using the specific weighting of the OLTV and CLTV as described in the Korea SIVR. We may also consider the maximum drawable balance, further advance, and purchase price if applicable in our analysis.

245. The OLTV and CLTV are based on the full loan balances secured on the property, including loan parts outside the asset pool and prior-ranking balances. CLTV could reflect updated prior-ranking balances if data is available. Second and subsequent-ranking balances will also be included in both OLTV and CLTV if data is available. Prior-ranking balances could include but are not limited to super-priority rights for small leases and first-lien "chonsei," if any. Chonsei are a type of security deposit by tenants common in the Korean real estate market. Second and subsequent-ranking balances could include but are not limited to second-lien chonsei.

246. In the LTV calculation, mortgage loan insurance covering super-priority rights for small leases in excess of the LTV limit set by the regulators is not considered. Please refer to 'Mortgage loan insurance' below.

247. An originator adjustment may apply if the originator's LTV calculation is materially different from market standards or we believe the inclusion of second and subsequent-ranking balances will significantly increase the portfolio LTV levels.

Loan affordability: Debt-to-income (DTI)

248. DTI is defined as total principal and interest payment for all household loans over income (following the debt-service ratio (DSR) definition by local regulators). Secondary income and spousal income might be aggregated if this is in compliance with the regulators' guidelines. If spousal income is considered, spousal debt will also be considered.

249. If loan-level data is not available for measuring affordability, we make adjustments based on our assessment of the originator's practices relative to market standards via the originator adjustment at the pool level.

250. We may apply originator adjustments in addition to the DTI adjustment on a case-by-case basis, depending on the performance data provided, or if we believe the originator's practices are significantly different from market standards.

Seasoning (adjustment factors for loan seasoning)

251. If corresponding data are available, the criteria may consider the seasoning of the performing relationship between the borrower and the mortgage lender of an existing mortgage loan being refinanced, instead of the seasoning of the new loan that results from such refinancing. For instance, a loan may have been refinanced with a new loan that has the same term or a shorter term, and the loan balance and interest rate are the same or lower. In that case, seasoning could be based on the origination date of the original loan rather than the date of the refinancing.

Second and subsequent lien loans

252. We apply an adjustment factor for second and subsequent lien mortgage loans as indicated in table 4 of the Korea SIVR. We differentiate a variable by the other lien loan's data availability.

253. Mortgage loans with super-priority rights for small leases are not considered as second lien as long as there are no other prior-ranking balances. We typically aggregate the super-priority rights for small leases with the mortgage claim and treat them as a single loan in the analysis.

254. We may consider different adjustments on a case-by-case basis, depending on the servicers' underwriting policies, track record, and performance data provided.

255. To the extent a pool exhibits a concentration of third and subsequent lien loans, then we may apply a higher originator adjustment.

256. In all cases above, the loan purpose adjustment does not apply.

Borrower occupancy status

257. We apply adjustment factors for loans to finance investment properties. The investment property classification includes second homes, investment property with loans underwritten to borrower income, and investment property with loans underwritten to non-borrower income, such as the property's DSCR.

258. Typically, for investment properties with loans underwritten to non-borrower income, borrower-related adjustments such as DTI, self-certification and employment status do not apply. We may consider different adjustments on a case-by-case basis, depending on the performance data provided.

Loan product type (payment shock and interest Only)

259. Where a borrower may face payment shock from two separate sources, for example, a borrower with a fixed-to-float and incremental repayment loan, we do not apply the adjustment twice, but apply the highest adjustment once. However, for limited recourse loan products, we apply both the adjustment for limited recourse, and any further adjustment for payment shock and interest-only loans, where applicable.

260. We might apply additional adjustments based on product type through originator adjustments if we believe risks are not sufficiently captured in abovementioned adjustments.

Re-performing loans and restructured/modified loans

261. We may consider different adjustments on a case-by-case basis, depending on the servicers' restructure policies, track record, and performance data provided.

262. We typically define a reperforming loan as a loan that has been 90 or more days past due or restructured in the three years leading up to the analysis date and is current as of that date.

263. When a reperforming arrangement is made, it is typical that a full reassessment of the borrower's affordability capacity is made; we consider this akin to a re-underwriting of the loan. Accordingly, for pools classified as reperforming, we calculate potential future seasoning credit based on the date a loan was previously 90 or more days in arrears.

264. In addition to the original loan and borrower information provided, we may also consider updated data sourced through the restructuring process in our analysis of reperforming loans, where available, on a case-by-case basis.

265. In addition, as part of the analytical process, we analyze data from the issuer/servicer on re-default rates stratified by forbearance type. This analysis is used to calibrate the originator adjustment for such transactions.

Geographic concentration

266. If a pool has significant geographic concentration risk that we believe is not sufficiently captured, we may account for it using the originator adjustment.

Residency status

267. We would apply a lower adjustment if the originator or servicer provides evidence that the origination process is robust and the performance of loans to non-Korean citizens are comparable to Korean citizens.

Details Of Further Considerations When Calculating The Loss Severity Of A Korean Mortgage Pool

Valuation haircut

268. A standard valuation method in Korea is through a transaction-based database provided by Kookmin Bank, Ministry of Land, Infrastructure and Transport, Korean Real Estate Board, or other government-recognized entities, because a full appraisal is not common in the Korean mortgage market for certain property types. Sometimes, the valuation can also be provided by the tax administration. Considering this, we generally do not apply a haircut to these valuation types. We apply a certain haircut if the valuation is based on sales price or other valuation methods (see table 6 in the Korea SIVR). Where relevant, we may make an adjustment to better reflect the pool's characteristics. For example, we may revise up or down a lender's valuation if they employ less or more conservative valuation methods than market standards.

269. A valuation haircut might apply at pool level when loan level data is not available or might vary when an originator's valuation method significantly deviates from the industry standard.

Forced-sale discount (FSD)

270. On a case-by-case basis, we may increase or decrease the forced-sale discount where there is sufficient information to support it being either higher or lower than envisaged using the standard calculation.

Accrued and unpaid interest

271. Prior-ranking balances, typically superpriority rights or first-lien chonsei, will not accrue interest and will not result in unpaid interest to be incorporated into a loan's loss severity if a cash flow analysis is not performed.

Mortgage loan insurance

272. To determine the loss severity, no credit is given to mortgage loan insurance which covers super-priority rights for small leases in excess of the LTV limit set by the regulators. We currently don't have any historical data for claims-payout ratio to determine an assumption in accordance with the methodology outlined in "Methodology For Assessing Mortgage Insurance And Similar Guarantees And Supports In Structured And Public Sector Finance And Covered Bonds," published Dec. 7, 2014. We may consider giving credit to the mortgage loan insurance when more claims-payout ratio data is available to us.

Details Of Further Considerations When Calculating Minimum Credit Enhancement Levels In The Analysis Of A Korean RMBS Or Covered Bond Transaction

Table 12

Minimum Credit Enhancement
	AAA	AA	A	BBB	BB	B
CE Floor*	4.00	2.50	1.50	1.00	0.50	0.35
Hard CE Floor	2.50	2.00	1.50	n/a	n/a	n/a
*CE Floor is applied to final CE results in cash flow analysis, instead of credit losses from credit model. The floor is calculated as a percentage of pool balance.

Details Of Further Considerations When Applying The Cash Flow Criteria To The Analysis Of A Korean RMBS Or Covered Bond Transaction

273. In our rating analysis, we also assess a transaction's payment structure and cash flow mechanics. This analysis uses our own quantitative models to evaluate whether the cash flows from the assets suffice, at the applicable rating levels, for making timely payments of interest and ultimate payment of principal (i.e., by or before the legal maturity date).

274. In our cash flow analysis, we use the pool-level WAFF and WALS described in previous sections as inputs, to reflect credit stress at each rating level.

275. During modeling, cash flow stresses test the credit and liquidity support the assets need to make timely interest payments and final principal payments by their legal maturity dates. Our models consider any available structural support, such as cash reserves, liquidity facilities, and hedging arrangements.

276. For revolving stand-alone RMBS structures (i.e., structures backed by a pool whose assets change or revolve), the modeling approach aims to reflect the structure after the activation of any "early amortization" (or stop-substitution) triggers and applies cash flow stresses from this point. An early amortization trigger is an event or situation that halts the substitution of assets in a revolving loan pool. We do not apply this approach to pools supporting covered bonds, even though the assets in these pools may be substituted over time. This is because the starting assumption of the collateral analysis under our covered bond criteria is the default of the issuing bank, as a result of which we do not expect the cover pool to be actively managed (it would become a static pool).

277. This criteria appendix does not include the assumptions used to assess refinancing costs in covered bonds structured with an asset-liability mismatch (such as target asset spreads). Those are described in "Covered Bonds Criteria," published Dec. 9, 2014.

278. In our surveillance of existing ratings, cash flow modeling may show that under the 'B' stress, a particular tranche will miss interest payments or fail to repay the principal by or before the final legal maturity date. If this is the case, then our initial assessment, assuming all factors remain the same, may be to consider lowering the rating on those securities to 'B-' or lower.

279. Depending on our view of a transaction's immediate cash flow position, the rating could move into the 'CCC', 'CC', or 'C' rating category, consistent with our ratings definitions, and in accordance with "Criteria For Assigning 'CCC+', 'CCC', 'CCC-', And 'CC' Ratings," published Oct. 1, 2012.

Default timing

280. The cumulative amount of defaults for cash flow modeling is equal to the pool's WAFF, whereas the cumulative amount of recoveries is calculated as (1 minus WALS).

281. In some cases, we may make an adjustment to the default timing curves to better reflect the pool's composition and transaction structure. For example, we may shorten the default curves where the mortgage portfolio's average remaining term gets shorter.

282. Foreclosure period assumptions represent the estimated time to repossess and sell a property upon a default (see table 6 in the Korea SIVR). They also reflect the typical time necessary for judicial proceedings and any other likely delay.

283. Cash flow modeling considers the negative carry resulting from interest due on the rated liabilities during the foreclosure period.

284. The loss severity estimates used in the cash flow modeling are based on the loan principal and assume no recovery of interest accrued on the mortgage loans during the foreclosure period.

Interest rate scenarios

285. Specific structural features may involve using additional cash flow stresses, such as alternative interest rate patterns or different default-timing curves, among others.

Prepayment scenarios

286. Residential loan prepayments vary the amount of excess spread available, and this may affect the absolute level of defaults exhibited in a transaction.

287. When analyzing the payment structure and cash flow mechanics of Korean RMBS and covered bonds, we typically test the transaction's ability to withstand high and low prepayment scenarios as set out in the Korea SIVR.

288. We may raise the prepayment assumptions if a pool's historical prepayment rates were higher than historical averages or if a transaction was particularly sensitive to prepayment risk. We may also reduce prepayment stress in situations where long-term historical data support lower prepayment rate assumptions for a specific loan product.

Modeling of reinvestment rates and spread compression

289. We typically do not assume any revenues associated with any cash or securities the issuer holds unless there is an eligible guaranteed investment contract (GIC) in place.

290. On a case-by-case basis, we apply a spread compression assumption if we identify a risk of material decrease in spread over time. For example, borrowers paying higher interest rates might tend to prepay earlier or indicate higher default risk.

Originator insolvency, commingling, and set-off

291. The analysis of any commingling or setoff risks that can result from an originator's or servicer's insolvency follows the application of the counterparty criteria (see "Counterparty Risk Framework: Methodology And Assumptions," published March 8, 2019). The counterparty criteria determine the maximum supported rating based on available mitigation of such counterparty risks. Mitigants may include a counterparty's commitment to implement remedies upon a downgrade, or structural factors such as the coverage of the risk through credit enhancement, demonstrated through the modeling of any exposure.

292. Regarding commingling risk, the degree to which a collection account holder's insolvency affects the cash flow from the assets in a mortgage loan pool depends on the collection account's characteristics, if the collection account is not in the name of the RMBS issuer.

293. The amount at risk depends on the timing of scheduled payments from borrowers, the frequency of transfers into the transaction account, and the level of prepayments.

Modeling of senior fees and expenses

294. The modeling of an issuer's foreseeable expenses might use stressed costs to reflect the need to replace the initial service provider.

295. The most significant portion of senior fees and expenses is related to the servicer. Estimation of transaction servicing fees usually includes a certain increase from the amount stipulated in the transaction agreement to account for a potential replacement of the servicer. We also examine the degree to which the structure is typical, and the levels of servicing fees in the market. The estimate of servicing fees in table 7 of the Korea SIVR reflects our findings.

296. To reflect the likely cost of replacing the initial servicer, we set servicing fees at the higher of the contractual rate and the assumed rates in the table 7 of the Korea SIVR.

297. Other unexpected expenses may arise during the life of a rated security, such as when amending transaction documents, costs associated with perfection of title, and any other unexpected costs required for the orderly maintenance of the assets. In the absence of any structural mechanism, such as a reserve designated to cover substantial or full amount of the unexpected expenses, an expense cap, or if the cap is significantly lower than industry standard, we might assume extraordinary expenses of up to a certain amount in our cash flow analysis (see table 7 in the Korea SIVR).

298. We may apply different assumptions from those detailed in the Korea SIVR. For example, we may lower the level to reflect easier/simpler tasks of the servicer. On the other hand, we may raise the level for an exotic transaction to reflect a difficulty of the servicer's replacement.

Delinquency

299. We assume a delay of a proportion of scheduled interest and principal receipts in each of the first several months of a hypothetical recession, and set full recovery of the arrears to take place several months after the delinquency occurs (see table 7 in the Korea SIVR). The cash flow stress for delinquencies is independent of the arrears adjustment to the WAFF.

300. For pools that contain residential loans with an option to temporarily suspend the periodic payments (payment holiday loans), the criteria appendix includes a delay of a proportion of scheduled interest and principal receipts. In situations where there is the potential for payment holidays to be granted after a loan's inception or where payment holidays have been granted (e.g., due to governments' and banks' forbearance measures for households and small and midsize corporates), we may apply an additional stress in our cash flow analysis where relevant. In those instances, the criteria appendix includes a delay of a proportion of scheduled interest and principal receipts based on an estimate of the proportion of a pool that opts to take a payment holiday and the likely duration of the holiday. The likely duration will be assessed with reference to factors that may include, but are not limited to, relevant legislative frameworks, collateral credit quality, servicers' policies, and available servicer data on payment holidays granted.

Appendix III: Superseded Criteria Articles

301. These criteria fully superseded the following criteria articles as they became effective in each applicable market. Some markets require prior notification to and/or registration by the local regulator, where the criteria become effective when so notified by S&P Global Ratings and/or registered by the regulator. Alternatively, the criteria may become effective at a specified date:

• Methodology And Assumptions For Rating Japanese RMBS, Dec. 19, 2014 (superseded as of Jan. 6, 2023);
• Methodology And Assumptions: Assessing Pools Of European Residential Loans, Aug. 4, 2017;
• Methodology And Assumptions: Rating Mexican Residential Mortgage-Backed Securities, June 7, 2012;
• LEVELS Mexico Estimates The Risk Of Defaults And Recoveries For Mexican RMBS, Jan. 20, 2009;
• Securitization In Latin America: Rating Criteria For Argentine Mortgage-Backed Securities, Aug. 31, 2004; and
• Securitization In Latin America: Existing Assets: Asset-Specific Rating Criteria, Sept. 1, 2004.

Appendix IV: Jurisdictions That Are Out Of Scope Of These Criteria

302. The following jurisdictions are out of scope of these criteria, as they are analyzed according to separately published criteria:

• Australia is analyzed according to "Australian RMBS Rating Methodology And Assumptions," published Sept. 1, 2011;
• New Zealand is analyzed according to "New Zealand RMBS Rating Methodology And Assumptions," published Sept. 14, 2011; and
• The U.S. is analyzed according to "Methodology And Assumptions For Rating U.S. RMBS Issued 2009 And Later," published Feb. 22, 2018.

303. The following jurisdictions are out of scope of these criteria, as they are analyzed under a "Principles Of Credit Ratings" approach:

• China; and
• Singapore.

IMPACT ON OUTSTANDING RATINGS

304. The total outstanding portfolio of ratings that are subject to the expansion of scope of these global criteria to include Korea consists of three covered bond program ratings and 19 covered bond issuance ratings.

305. As to the anticipated impact of the expansion in scope of these global criteria to Korea, our testing indicated that there would be no rating impact on our Korean covered bond ratings.

306. This analysis is intended to serve as a broad, directional guide to the possible ratings impact of the criteria. Ultimately, actual ratings impact may vary, depending on the specifics and performance of the asset pool and structural features of a particular transaction or program.

REVISIONS AND UPDATES

307. This article was originally published on Jan. 25, 2019.

308. Changes introduced after original publication:

• On March 1, 2019, we republished this criteria article to make nonmaterial changes to criteria references.
• On March 1, 2019, we republished this criteria article to make nonmaterial changes to criteria references.
• On May 24, 2019, we republished this criteria article to make material changes, expanding the scope to include the U.K. and updating the impact section.
• On Sept. 6, 2019, we republished this criteria article to make material changes, expanding the scope to include Belgium, France, Ireland, and The Netherlands and updating the impact section.
• On July 10, 2020, we republished this criteria article to make material changes, expanding the scope to include Denmark, Finland, Norway, and Sweden and updating the impact section. We also made nonmaterial changes. Following the gradual expansion of the scope of the criteria, we streamlined the scope language to no longer list each country in scope, and have highlighted in appendix which jurisdictions are analyzed under separate criteria or a "Principles of Credit Ratings" approach, and remain out of scope of these criteria. We also clarified our definition of 'LTV curve type' in paragraph 36. Finally, we made changes to criteria references.
• On Jan. 8, 2021, we republished this criteria article to make material changes, expanding the scope to include Austria, Germany, Greece, Hungary, Italy, Portugal, and Spain and updating the impact section. We also made nonmaterial changes to update the contact information and the superseded criteria articles and the related criteria and research sections.
• On Dec. 15, 2021, we republished this criteria article to make nonmaterial changes to update criteria references.
• On May 12, 2022, we republished this criteria article to make a correction to the x-axis in chart 4. We also made nonmaterial changes by removing horizontal and vertical axis markers in chart 4, consistent with the layout for charts 2 and 3. In table 3 and paragraphs 107 and 112, we clarified that the determination of Repo MVD is at each rating level. In paragraph 6, we clarified a reference to the "Related Criteria And Research" section. Finally, we updated the "Related Criteria" section and the contact information.
• On Dec. 16, 2022, we republished this criteria article to make material changes, expanding the scope to include Japan. We added an appendix (Appendix IIa) that provides details of the application of the criteria in the Japanese market and is to be read in conjunction with the Japanese RMBS Sector And Industry Variables article. We also made nonmaterial changes to the text and format of the article, to reflect the introduction of Appendix IIa that governs the application of the criteria in Japan, as well as other editorial refinements to existing text. Finally, we made nonmaterial changes to update the contact information and the superseded criteria articles and the related research sections.
• On June 2, 2023, we republished this criteria article to make material changes, expanding the scope to include Korea. We added an appendix (Appendix IIb) that provides details of the application of the criteria in the Korean market and is to be read in conjunction with the Korean RMBS and covered bond Sector And Industry Variables article. We also made nonmaterial changes to the text and format of the article, to reflect the introduction of Appendix IIb, as well as other editorial refinements to existing text. Finally, we made nonmaterial changes to update the contact information and the related research sections.

RELATED CRITERIA AND RESEARCH

Related Criteria

• Banking Industry Country Risk Assessment Methodology And Assumptions, Dec. 9, 2021
• Environmental, Social, And Governance Principles In Credit Ratings, Oct. 10, 2021
• Global Framework For Payment Structure And Cash Flow Analysis Of Structured Finance Securities, Dec. 22, 2020
• Counterparty Risk Framework: Methodology And Assumptions, March 8, 2019
• Incorporating Sovereign Risk In Rating Structured Finance Securities: Methodology And Assumptions, Jan. 30, 2019
• Methodology For National And Regional Scale Credit Ratings, June 25, 2018
• Methodology And Assumptions For Rating U.S. RMBS Issued 2009 And Later, Feb. 22, 2018
• U.S. Residential Mortgage Operational Assessment Ranking Criteria, Feb. 22, 2018
• Sovereign Rating Methodology, Dec. 18, 2017
• Structured Finance: Asset Isolation And Special-Purpose Entity Methodology, March 29, 2017
• Covered Bond Ratings Framework: Methodology And Assumptions, June 30, 2015
• Covered Bonds Criteria, Dec. 9, 2014
• Methodology For Assessing Mortgage Insurance And Similar Guarantees And Supports In Structured And Public Sector Finance And Covered Bonds, Dec. 7, 2014
• Global Framework For Assessing Operational Risk In Structured Finance Transactions, Oct. 9, 2014
• Criteria For Assigning 'CCC+', 'CCC', 'CCC-', And 'CC' Ratings, Oct. 1, 2012
• Methodology For Applying RMBS Small Pool Adjustment Factor, May 24, 2012
• New Zealand RMBS Rating Methodology And Assumptions, Sept. 14, 2011
• Australian RMBS Rating Methodology And Assumptions, Sept. 1, 2011
• Principles Of Credit Ratings, Feb. 16, 2011

Related Research

• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Korea), June 2, 2023
• Sector And Industry Variables: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Korea), June 2, 2023
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Japan), Dec. 16, 2022
• Sector And Industry Variables: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Japan), Dec. 16, 2022
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Japan), July 29, 2022
• S&P Global Ratings Definitions
• Evolution Of The Methodologies Framework: Introducing Sector And Industry Variables Reports, Oct. 1, 2021.
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Spain, Portugal, Italy, And Greece), Jan. 8, 2021
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Austria, Germany, And Hungary), Jan. 8, 2021
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Spain, Portugal, Italy, And Greece), Oct. 30, 2020
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Austria, Germany, And Hungary), Oct. 20, 2020
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Sweden, Norway, Finland, And Denmark), July 10, 2020
• Request for Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Sweden, Norway, Finland, And Denmark), Nov. 6, 2019
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Ireland), Sept. 6, 2019
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (The Netherlands, Belgium, And France), Sept. 6, 2019
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (Ireland), July 3, 2019
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (The Netherlands, Belgium, And France), June 3, 2019
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (U.K.), May 24, 2019
• Request For Comment: Global Methodology And Assumptions: Assessing Pools Of Residential Loans (U.K.), March 12, 2019
• Guidance: Global Methodology And Assumptions: Assessing Pools Of Residential Loans, Jan. 25, 2019
• RFC Process Summary: Global Methodology And Assumptions: Assessing Pools Of Residential Loans, Jan. 25, 2019
• Request For Comment: Global Methodology And Assumptions: Assessing Pools of Residential Loan, Sept. 17, 2018

EMEA Analytical Contacts:	Elton Eakins, London + 44 20 7176 3698; elton.eakins@spglobal.com
	Alastair Bigley, London 44 (0) 207 176 3245; Alastair.Bigley@spglobal.com
LatAM Analytical Contacts:	Leandro C Albuquerque, Sao Paulo +55 (11) 3039-9729; leandro.albuquerque@spglobal.com
	Jose Coballasi, Mexico City + 52 55 5081 4414; jose.coballasi@spglobal.com
APAC Analytical Contacts:	Yuji Hashimoto, Tokyo + 81 3 4550 8275; yuji.hashimoto@spglobal.com
	Hiroshi Sonoda, Tokyo (81) 3-4550-8474; hiroshi.sonoda@spglobal.com
	Jerry Fang, Hong Kong + 852 2533 3518; jerry.fang@spglobal.com
	Yalan Tao, Hong Kong + 852 2532 8033; yalan.tao@spglobal.com
Methodology Contacts:	Herve-Pierre P Flammier, Paris (33) 1-4420-7338; herve-pierre.flammier@spglobal.com
	Andrew M Bowyer, CFA, London (44) 20-7176-3761; andrew.bowyer@spglobal.com
	Andrew O'Neill, CFA, London + 44 20 7176 3578; andrew.oneill@spglobal.com