Standard & Poor's Ratings Services recently modified its assumptions to analyze cash flows for U.S. residential mortgage-backed securities (RMBS) that rely on excess interest as a form of credit enhancement, with the exception of those U.S. RMBS transactions that are backed solely by home equity line of credit (HELOC) collateral.
The enhancements to our U.S. RMBS cash flow analysis include, among other things, the ability to test an expanded number of cash flow scenarios, including multiple collateral prepayment and default assumptions. The revisions also include user-defined variables that govern the likelihood of credit support release and priority of principal payments. This article describes these changes, in addition to the potential effect on transaction structures.
This article also addresses the "payment structure and cash flow mechanics" principle discussed in "Principles-Based Rating Methodology For Global Structured Finance Securities," published May 29, 2007, on RatingsDirect and Standard & Poor's Web site at www.standardandpoors.com.
Revised Cash Flow Assumptions Summary
The revised cash flow assumptions, which were implemented March 4, 2008 (see "Standard & Poor’s Enhances U.S. Residential Mortgage Cash Flow Model"), and are now available to the market in SPIRE 3.1, released April 7, 2008, provide what Standard & Poor's believes is a transparent method of evaluating proposed securitization structures using a wide range of user-defined variables.
The key aspects of the revised cash flow assumptions include the following:
The implementation of a user-defined credit enhancement multiple (CEM), which is the factor by which the senior enhancement must increase for step-down to occur, as well as the amount of credit enhancement that each class must maintain following step-down in the absence of a trigger event. Varying the CEM may affect the timing of step-down and principal distribution amount (PDA) definitions;
The implementation of user-defined delinquency triggers and cumulative loss triggers;
The implementation of a user-defined overcollateralization (O/C) floor; and
The introduction of back-ended default curve assumptions in addition to Standard & Poor's standard default curve assumptions.
The revised cash flow assumptions are intended to provide ratings stability to those classes rated by Standard & Poor's. They are based on recent observations of transactions that passed their step-down dates without the occurrence of trigger events and released credit support that could have otherwise been used to absorb actual and projected losses to rated classes. Standard & Poor's believes that the enhancements to our assumptions address the variability of the timing of defaults and losses demonstrated by the actual performance of recent vintages (see "Implemented Revisions").
We previously published our revised approach to calculating the O/C floor, which was intended to mitigate the premature release of credit support (see "S&P Analyzes the Impact of Higher Overcollateralization Floor Guidelines On U.S. RMBS Transactions," published Dec. 20, 2007). The revised assumptions discussed in this article build upon the analysis provided in the aforementioned article by addressing the potential release of credit support at and after step-down under a number of scenarios, through both the release of O/C and the payment of principal to the subordinate classes.
Background
It is necessary to understand the transaction definitions that govern the amount of credit support and principal payment priorities in order to comprehend the potential effect of our revised assumptions and the testing we describe below in "Implemented Revisions." We provide the typical definitions issuers use in these transactions and our observations in the following section.
Definitions and their potential effects on transaction performance
Transactions that rely on excess interest as a form of credit enhancement typically include definitions for the step-down date, senior enhancement percentage (SEP), delinquency trigger event, principal distribution amount, O/C target, and cumulative loss trigger event based on common assumptions. We created a hypothetical transaction to illustrate how Standard & Poor's believes the definitions interact and how we believe changes to these definitions may affect a transaction's performance. We refer to the common multiple (traditionally 2x) in the definitions as the CEM. Standard & Poor's evaluates transactions based on a CEM determined by the issuer.
Step-down date
The step-down date will be the earlier of:
The first distribution date on which the aggregate certificate principal balance of the class A certificates has been reduced to zero; and
The later to occur of the distribution date occurring three years from the closing date and the first distribution date on which the SEP is greater than or equal to approximately double (a multiple of 2x) the initial SEP.
SEP
On any distribution date, the SEP will be equal to a fraction, the numerator of which is the sum of the aggregate certificate principal balance of the class M (subordinate) certificates and the O/C amount, and the denominator of which is the aggregate stated principal balance of the mortgage loans after giving effect to distributions to be made on that distribution date.
The multiple used in the step-down date definition typically has been two, such that the SEP is required to double for step-down to occur. However, when prepayments result in a SEP exceeding a doubled original SEP in transactions by the step-down date, a higher multiple utilized for step-down may better protect the rated classes should losses occur post step-down, as more credit support would be retained.
Delinquency trigger event
A delinquency trigger event is in effect with respect to any distribution date if the three-month average of the 60-plus day delinquency percentage, as determined on that distribution date and the immediately preceding two distribution dates, equals or exceeds XX.XX% ("delinquency trigger threshold") of the SEP.
In the typical definition above, the delinquency trigger threshold is a percentage of the SEP, as opposed to a percentage of the outstanding pool balance. Traditionally, this delinquency trigger threshold was based on the expectation that the SEP would have doubled from its original percentage by the step-down date.
Prior to step-down, the SEP continues to increase as only the senior classes receive principal payments. Since the delinquency trigger threshold as defined by the issuer is held constant, the percentage increase in the SEP results in a like increase to the level of delinquencies, as a percentage of the outstanding pool balance, needed to cause a trigger event. Once the senior certificates have been completely paid off (SEP at 100%), the percentage of delinquent loans in the pool that would cause a delinquency trigger event equals the delinquency trigger threshold (48% in the example in chart 1).
Chart 1
Chart 1 is based on a hypothetical transaction used in the "Scenario testing" section, and shows the interaction of the SEP increase and the percentage of delinquencies required to cause a trigger event under three different voluntary terminal constant prepayment rate (CPR) environments. The three environments correspond to CPR assumptions of 15%, 25%, and Standard & Poor's standard CPR assumptions for Alternative-A (Alt-A) and subprime collateral, which are generally 33% for fixed-rate collateral and 45% for adjustable-rate collateral. The 'A' rating stress environment was used for all three CPR assumptions. Table 1 provides the hypothetical pool's loss coverage and loan product composition.
Table 1
Hypothetical Transaction Loss Coverage And Loan Product Composition (%)
Loss coverage
FF
LS
LC
AAA
30.3
65.2
19.8
AA+
27.7
61.4
17.0
AA
25.3
57.1
14.4
AA-
23.1
55.6
12.8
A+
21.1
53.1
11.2
A
19.3
51.6
10.0
A-
17.7
50.2
8.9
BBB+
16.2
49.3
8.0
BBB
14.9
48.4
7.2
BBB-*
13.6
47.6
6.5
Loan products
ARMs
Fixed
CES
77.23
19.78
2.99
*Lowest rated class in hypothetical structure. ARM—Adjustable-rate mortgage. CES—Closed-end second liens. FF—Foreclosure frequency. LS—Loss severity LC—Loss coverage.
As chart 1 shows, the speed at which the delinquency trigger threshold is reached is driven in part by the prepayment speed. Increasing the CPR assumption from 15% to 25% causes the delinquency trigger threshold of 48% to be reached in month 72, which is 43 months earlier than under the 15% CPR assumption scenario for the hypothetical transaction. Under Standard & Poor's standard CPR assumptions, the delinquency trigger threshold would be reached in month 52.
By month 37, which is the typical step-down date if senior classes remain outstanding, the SEP has doubled under both the 25% and Standard & Poor's CPR assumption scenarios. On that distribution date, the 60-plus day delinquencies, as a percentage of the outstanding pool balance, that would be needed to cause a delinquency trigger event are 19.9% and 29.8% for the 25% and Standard & Poor's CPR assumptions, respectively, assuming a delinquency trigger threshold of 48% of the SEP. It should be noted that under a 15% CPR the doubling of the SEP needed for step-down does not occur until period 45, at which point 16% of the outstanding pool balance would need to be 60-plus days delinquent to cause a delinquency trigger event.
The delinquency trigger threshold may be adjusted based on the issuer's expectations of the 60-plus day delinquencies and SEP at step-down. For example, assume the issuer intends for a delinquency trigger event to occur at step-down if at least 10% of the outstanding pool balance is delinquent. If, at closing, the SEP is 20% and the expectation is that the SEP will double (a multiple of 2x) to 40% at step-down, the delinquency trigger threshold would be set at 25% of the SEP (delinquency trigger threshold x SEP = pool delinquencies needed to cause trigger event).
However, if instead the SEP grows to 60% (a multiple of 3x original SEP), 15% in total 60-plus day delinquencies would cause a trigger event using the 25% delinquency trigger threshold. An issuer can revise the threshold to reach the projected delinquency rate under the issuer's expected default and voluntary terminal CPR assumptions. In this example, the delinquency trigger threshold of 16.7% could be used so that at step-down, assuming the SEP tripled to 60%, the 60-plus day delinquencies needed to cause a trigger event would again be 10% of the outstanding pool balance.
Class M-1 PDA
The class M-1 PDA on any distribution date is the excess, if any, of:
The sum of the aggregate class certificate balance of the class A certificates (after taking into account the distribution of the class A PDA for that distribution date) and the class certificate balance of the class M-1 certificates immediately before that distribution date; over
The lesser of approximately XX.XX% of the aggregate stated principal balance of the mortgage loans for that distribution date and the excess, if any, of the aggregate stated principal balance of the mortgage loans for that distribution date over $___ million (the O/C floor).
The percentage provided in the PDA formula is traditionally set such that after step-down, each class is paid principal to maintain double (a multiple of 2x) its original credit enhancement, assuming the O/C floor has not been reached.
Standard & Poor's believes that revisions to the enhancement percentages as defined by the issuer for both the step-down date and the PDAs should result in changes to principal distribution priorities after step-down. For example, a higher required enhancement percentage will result in a reduction in the "bottom-up" principal payments to more subordinate certificates after step-down, which we described in our previous article detailing the O/C floor analysis (see "S&P Analyzes the Impact of Higher Overcollateralization Floor Guidelines On U.S. RMBS Transactions").
Required O/C amount (O/C target)
The required O/C amount (O/C target) is XX.XX% of the aggregate stated principal balance of the mortgage loans as of the transaction's closing date if the distribution date is before the step-down date. If the distribution date is on or after the step-down date, the required O/C amount is the greater of: 2x XX.XX% of the then-current aggregate stated principal balance of the mortgage loans as of the end of the related due period and the O/C floor. However, if a trigger event is in effect, the required O/C amount will be an amount equal to the required O/C amount for the immediately preceding distribution date.
In most transactions, the O/C target is defined as in the typical definition above to allow a reduction at step-down in the required O/C amount from its initial target amount to double (a multiple of 2x) the initial percentage as a function of the outstanding, rather than original, collateral balance, as indicated in the definition above. Any reduction in the O/C target amount is limited by the O/C floor. In our prior article regarding the O/C floor, we describe in detail the effect changes to the O/C floor may have on a transaction's structure.
Cumulative loss trigger event
A cumulative loss trigger event occurs if, for any distribution date, the amount of realized losses incurred since the transaction's closing date through the last day of the related due period divided by the aggregate stated principal balance of the mortgage loans as of the transaction's closing date exceeds the applicable percentages for such distribution date as provided in the example in table 2.
Table 2
Cumulative Loss Trigger Event Percentages
Distribution dates occurring in month
Loss (%)
25-36
1.400 for the first month, plus an additional one-twelfth of 1.700 for each month, thereafter
37-48
3.100 for the first month, plus an additional one-twelfth of 1.650 for each month ,thereafter
49-60
4.750 for the first month, plus an additional one-twelfth of 1.400 for each month, thereafter
61-72
6.150 for the first month, plus an additional one-twelfth of 0.700 for each month, thereafter
73 and thereafter
6.850
In the definition above, the level of realized losses needed to result in a cumulative loss trigger event increases each month over a six-year period, ultimately reaching a terminal value of 6.85%. When actual timing of losses lags the timing assumed in setting the cumulative loss trigger thresholds, the cumulative loss trigger may not be effective in retaining credit support to absorb future losses. In such a scenario, tighter cumulative loss trigger thresholds would more likely result in a trigger event and retention of credit support. The effect on the hypothetical transaction is described in the "Implemented Revisions" section.
Implemented Revisions
We implemented the following revisions to our cash flow assumptions March 4, 2008.
Back-ended default curve
We have incorporated back-ended default curve assumptions into our ratings analysis to simulate scenarios where a majority of defaults occur after the first potential step-down date. For example, 53.50% of the defaults occur after period 36 under our unseasoned back-ended default curve, compared with 41.25% for the unseasoned subprime and Alt-A default curve. By modeling triggers and the release of credit support under the back-ended default curve, Standard & Poor's tests for the sufficiency of credit support to absorb projected losses over the transaction's life.
As with the standard default curve, the back-ended default curves (seasoned and unseasoned, as appropriate) are applied to all rating categories. Table 3 shows our back-ended default curve assumptions.
Table 3
Back-Ended Default Curve Assumptions (%)
% defaults (original bal. + prefunding)
Month
< 12 mos. seasoned
> 12 mos. seasoned
1
2.25
6.25
6
4.00
4.50
12
4.50
11.25
18
11.25
8.50
24
8.50
8.25
30
8.25
7.75
36
7.75
11.05
42
11.05
11.05
48
11.05
9.80
54
9.80
7.80
60
7.80
7.50
66
7.50
6.30
72
6.30
0.00
Expansion of the number of cash flow scenarios
Our revised cash flow assumptions incorporate additional variables for each rating category, including the evaluation of user-defined delinquency and cumulative loss triggers, CEM, O/C floor, as well as a back-ended default curve. Each scenario is tested under our three projected interest rate vectors.
Additionally, our cash flow analysis may include two voluntary prepayment assumptions for each scenario. The structure is first run with our published voluntary prepayment assumptions, which stress the adequacy of excess spread in a higher voluntary prepayment environment. If that prepayment scenario results in an insufficient balance of outstanding collateral to support the full realization of expected losses throughout the default and loss curves, an additional scenario is also run, which reduces our voluntary prepayment assumptions to support the full realization of expected losses and to stress the adequacy of overall credit support.
Chart 2 shows that with the introduction of the evaluation of these additional variables, each rating category may be run in up to 18 scenarios.
Scenario testing
Tables 4-8 describe the potential relative changes that issuers can expect to the transaction structure, given the number of variables that may affect proposed structures under our revised assumptions. The results in the tables are based on a hypothetical transaction with a size of $1 billion, which has been tranched into a single senior certificate and nine subordinate certificates, spanning each rating category from 'AA+' to 'BBB-' (see table 1 for loss coverage and product composition information). The scenario testing results cover the following four potential factors that the issuer can modify:
CEM (applied to the SEP and PDA definitions);
Delinquency trigger;
O/C floor; and
Cumulative loss trigger.
Each table provides the senior and subordinate class sizes, O/C target, O/C floor, CEM, and O/C release percentage at step-down, if applicable. For ease of comparison, the O/C release percentage is specific to the 'A' rating scenario under the back-loaded default curve, rather than the highest rating category scenario for which O/C was released as a result of step-down. Additionally, for examples where step-down occurs, the highest rating level at which step-down occurs is provided.
As a starting point, table 4 illustrates a base scenario in which the capital structure passes our cash flow assumptions presuming that a trigger event is always in effect and there is no release of credit support on or after the step-down date.
Table 4
Base Scenario: Assuming Step-Down Does Not Occur
(%)
Senior class size
81.45
Subordinate class size
14.10
Initial and target O/C
4.45
O/C floor
2.60
O/C—Overcollateralization.
Table 5 shows the effect on the hypothetical transaction when varying the CEM. As discussed in the "Definitions And Their Potential Effects On Transaction Performance" section, the CEM impacts the step-down date, PDA, O/C target, and SEP on which the delinquency trigger is set.
Table 5
Variable Credit Enhancement Multiples
2.0
2.5
3.0
Senior class size (%)
83.30
82.15
81.85
Subordinate class size (%)
5.65
9.70
11.45
Initial and target O/C (%)
11.05
8.15
6.70
O/C floor (%)
2.60
2.60
2.60
Terminal cumulative loss trigger (%)
6.25
6.25
6.25
Delinquency trigger threshold (%)
48.00
48.00
48.00
O/C release at 'A' rating category (% of original bal.)
4.80
2.23
0.80
Highest rating category to experience O/C release
AA-
AA-
AA-
O/C—Overcollateralization.
As the CEM is increased (holding the O/C floor, delinquency trigger threshold, and cumulative loss triggers constant), the capital structure realizes benefits in the form of a larger aggregate subordinate class size and a lower initial and target O/C. The higher CEM required by the issuer for step-down and principal distributions results in a reduction in the O/C amount released and smaller principal payments being distributed to the lower-rated classes following step-down, increasing the available credit support.
The effect of the CEM on the percentage of 60-plus day delinquencies needed to cause a trigger event adjusts in step with each change in the CEM. Therefore, in addition to the credit support retention caused by the CEM's effect on step-down and PDA distributions, the likelihood that a delinquency trigger event occurs increases proportionately with each CEM increase as delinquency levels needed to cause a trigger event are reduced. This further restriction on credit support release contributes to a lower O/C release amount.
Table 6 shows the effect of an adjustment to the delinquency trigger threshold (holding the CEM, cumulative loss trigger, and O/C floor constant) on the capital structure. As the delinquency trigger was tightened, the amount of O/C release was reduced. The reduction in the initial and target O/C is a consequence of the credit support retention as the tighter delinquency trigger is more likely to restrict the ability to step-down.
Table 6
Variable Delinquency Trigger Thresholds (As A % Of SEP)
48.00
38.00
29.00
Senior class size (%)
83.30
82.80
82.50
Subordinate class size (%)
5.65
7.25
8.35
Initial and target O/C (%)
11.05
9.95
9.15
O/C floor (%)
2.60
2.60
2.60
CEM
2
2
2
Terminal cumulative loss trigger (%)
6.25
6.25
6.25
O/C release at 'A' rating category (% of original bal.)
In table 7, the O/C floor is varied, ranging from 50 basis points to 350 basis points. The effect on the capital structure was discussed in our previous article detailing the O/C floor analysis (see "S&P Analyzes the Impact of Higher Overcollateralization Floor Guidelines On U.S. RMBS Transactions"). Increasing the O/C floor results in a reduction to the O/C release amount. As with increases to the CEM or a tightening of the delinquency trigger, the retention of credit support allows for a lower target O/C as a result of restricting the amount of credit support released.
Table 7
Variable O/C Floor (% Of Original Bal.)
0.50
2.60
3.50
Senior class size (%)
83.50
83.30
82.50
Subordinate class size (%)
4.75
5.65
8.45
Initial and target O/C (%)
11.75
11.05
9.05
CEM
2
2
2
Terminal cumulative loss trigger (%)
6.25
6.25
6.25
Delinquency trigger threshold (%)
48.00
48.00
48.00
O/C release at 'A' rating category (% of original bal.)
Table 8 shows the effect of tightening the hypothetical transaction's cumulative loss triggers. We note that although reference is made to a reduction in the final or "terminal" cumulative loss trigger value, each incremental percentage in the cumulative loss trigger definition was similarly reduced in our hypothetical transaction.
Table 8
Variable Terminal Cumulative Loss Percentages
6.25
4.00
3.25
Senior class size (%)
83.30
82.95
81.45
Subordinate class size (%)
5.65
6.80
14.10
Initial and target O/C (%)
11.05
10.25
4.45
O/C floor (%)
2.60
2.60
2.60
CEM
2
2
2
Delinquency trigger threshold (%)
48.00
48.00
48.00
O/C release at 'A' rating category (% of original bal.)
As with a tightening of the delinquency trigger, note that a lower cumulative loss trigger percentage may restrict step-down and prevent credit support release under various stressed scenarios. In our hypothetical example, lowering the terminal cumulative loss percentage to 4.00% from 6.25% results in a lower O/C target and higher aggregate subordinate class size as a result of the additional credit enhancement retained. Once the terminal cumulative loss percentage is reduced to 3.25%, the transaction does not step-down in any scenario, resulting in a reduction of the initial O/C to 4.25% and an aggregate subordinate class size of 14.10%. Note that the prevention of step-down results in our base case scenario (see table 4). Because structures must survive the timing and magnitude of defaults and losses under the back-ended curve as well as the standard default curve, a significant reduction in the terminal cumulative loss trigger percentage is needed to prevent an O/C release under our cash flow assumptions.
To summarize these results, the initial and target O/C amounts as defined by the issuer may be substantially higher in the "triggers evaluated" scenarios which may allow step-down, the associated release of credit support and subordinate principal payments, versus the base scenario in which the trigger events are assumed to occur, and that maintain available credit support and sequential principal payments. Within each scenario, as a variable factor was altered in a manner that reduced the likelihood, magnitude, or size of projected credit support release, the O/C requirements were reduced due to either a greater retention of the initial credit support or a reduction in the amount of principal paid to the lower-rated subordinate classes on or after the step-down date. As different scenarios impact various levels of the capital structure, each rated class must survive up to 18 paths at its associated rating stress.
The examples depicted in the "Scenario testing" section represent a limited sampling of the myriad variations that the issuer can select, depending on preference and the potential effect that varying those structural components may have on the capital structure. Standard & Poor's will review proposed transaction structures and assess credit enhancement based on the structural features the issuer proposes.
Related Publications
The following articles are available on RatingsDirect. The criteria articles are also available on Standard & Poor's Web site at www.standardandpoors.com. In the left navigation bar, select Ratings and then Policies, Criteria & Definitions. The article is located under the Criteria tab.
"Principles-Based Rating Methodology For Global Structured Finance Securities," published May 29, 2007;
"Standard & Poor's Enhances U.S. Residential Mortgage Cash Flow Model," published March 4, 2008; and
"S&P Analyzes the Impact of Higher Overcollateralization Floor Guidelines On U.S. RMBS Transactions," published Dec. 20, 2007.
These criteria represent the specific application of fundamental principles that define credit risk and ratings opinions. Their use is determined by issuer-specific or issue-specific attributes as well as Standard & Poor's assessment of the credit and, if applicable, structural risks for a given issuer or issue rating. Methodology and assumptions may change from time to time as a result of market and economic conditions, issue-specific or issuer-specific factors, or new empirical evidence that would affect our credit judgment.
Analytic services provided by Standard & Poor's Ratings Services (Ratings Services) are the result of separate activities designed to preserve the independence and objectivity of ratings opinions. The credit ratings and observations contained herein are solely statements of opinion and not statements of fact or recommendations to purchase, hold, or sell any securities or make any other investment decisions. Accordingly, any user of the information contained herein should not rely on any credit rating or other opinion contained herein in making any investment decision. Ratings are based on information received by Ratings Services. Other divisions of Standard & Poor's may have information that is not available to Ratings Services. Standard & Poor's has established policies and procedures to maintain the confidentiality of non-public information received during the ratings process.
Ratings Services receives compensation for its ratings. Such compensation is normally paid either by the issuers of such securities or third parties participating in marketing the securities. While Standard & Poor's reserves the right to disseminate the rating, it receives no payment for doing so, except for subscriptions to its publications. Additional information about our ratings fees is available at www.standardandpoors.com/usratingsfees.
