Cash-flow Criteria for European RMBS Transactions
A typical European RMBS cash flow transaction consists of a number of rated notes that differ in seniority with respect to interest and principal payments from the underlying mortgage portfolio, in so-called senior/subordinated structures. There is usually a first-loss fund provided by the originator of the assets underneath the rated notes, often called the reserve fund. This is used to cover both interest shortfalls and principal losses arising in the transaction. A liquidity facility might also be incorporated, which is used to bridge timing mismatches that can occur between the asset cash flows and the required liability payments. The transaction might also include specific structural features designed to minimize the issuer’s exposure to external economic factors (e.g., interest rate hedges).
There are many variants to the generalized case described above. Structures tend to vary depending on the underlying collateral (e.g., prime RMBS transactions tend to differ structurally from nonconforming RMBS transactions), and across different countries (e.g., U.K. prime RMBS transactions differ structurally from Spanish or Italian prime RMBS transactions). This is generally for practical reasons. For example, U.K. prime mortgage originators tend to have very large portfolios, and have used “master trust” type structures primarily as a tool to reduce the costs of multiple securitizations over time. In contrast, Spanish and Italian transactions typically swap the entire asset cash flows to receive principal plus a fixed spread, primarily because the underlying mortgage loans tend to have quite variable interest rates, reset dates, and fixed periods.
The ability to accurately model the specific features of each transaction is crucial. Models that only create an approximate cash flow stream do not allow for an accurate assessment of the transaction’s strengths and weaknesses, as they may overlook transaction features critical to the assessment of the ability of the assets to repay rated debt over different stress scenarios. The procedure that Standard & Poor’s has adopted in order to ensure the accuracy of the cash flows is outlined below.
Until recently, Standard & Poor’s has reviewed cash flow models submitted by the arranger/issuer to ensure they replicate the transaction structure and incorporate Standard & Poor’s required stress scenarios. Now though, Standard & Poor’s develops cash flow models internally.
This approach retains the advantage that the exact transaction structure is replicated, but there are further efficiency and transparency benefits. Direct comparisons of the cash flow mechanics between transactions are likely to be more efficient, as they will no longer require the review of cash flow models developed by different arrangers. In addition, due to the bespoke nature of each cash flow model, the innovations and variations seen in the European market are incorporated. On this basis, arrangers/issuers are no longer always required to submit their own cash flow models to Standard & Poor’s for RMBS transactions.
Standard & Poor’s stresses the transaction cash flows to test both the credit and liquidity support provided by the assets, subordinated tranches, cash reserve, and any external sources (such as a liquidity facility). Stresses to the cash flows are implemented at all relevant rating levels.
For example, a transaction that incorporates ‘AAA’, ‘A’, and ‘BBB’ tranches of notes will be subjected to three separate sets of cash flow stresses. In the ‘AAA’ stresses, all ‘AAA’ notes must pay full and timely principal and interest, but this will not necessarily be the case for the ‘A’ or ‘BBB’ tranches, as they are subordinated in the priority of payments. In the ‘A’ case, all ‘AAA’ and ‘A’ notes must receive full and timely principal and interest, but not necessarily so for the ‘BBB’ tranche, as it is subordinated to both ‘AAA’ and ‘A’.
For each loan in the pool, Standard & Poor’s estimates the likelihood that the borrower will default on their mortgage payments (the foreclosure frequency), and the amount of loss upon the subsequent sale of the property (expressed as a percentage of the outstanding loan). The total mortgage balance assumed to default, and the total amount of this defaulted balance that is not recovered for the entire pool is determined by calculating the weighted-average foreclosure frequency (WAFF) and the weighted-average loss severity (WALS; note that 1 minus WALS is the assumed recovery rate). The WAFF and WALS estimates increase as the required rating level increases, because the higher the rating required on the notes, the higher the level of mortgage default and loss severity they should be capable of withstanding.
This credit analysis focuses on the characteristics of the loans and the associated borrowers. Given the variability in mortgage lending and borrower behavior across Europe, Standard & Poor’s applies country-specific criteria in its assessment of the WAFF and the WALS for a portfolio. As a consequence, the assumed percentage of defaults and the amount of recoveries can differ substantially across jurisdictions. Readers should consult the individual credit criteria associated with each country for detailed descriptions of the criteria for sizing defaults and recoveries, available to subscribers of RatingsDirect, Standard & Poor’s Web-based credit analysis system, at www.ratingsdirect.com. They can also be found on Standard & Poor’s Web site at www.standardandpoors.com.
As described above, the WAFF at each rating level specifies the total balance of the mortgage loans assumed to default over the life of the transaction. Standard & Poor’s assumes that these defaults occur over a three-year recession. Standard & Poor’s will assess the impact of the timing of this recession on the ability to repay the liabilities, and chooses the recession start period based on this assessment. Although the recession normally starts in the first month of the transaction, the ‘AAA’ recession is usually delayed by 12 months. The WAFF is applied to the principal balance outstanding at the start of the recession (for example, in a ‘AAA’ scenario the WAFF is applied to the balance at the beginning of month 13). Defaults are assumed to occur periodically in amounts calculated as a percentage of the WAFF. The timing of defaults generally follows two paths, referred to here as “fast” and “slow” defaults. These timings are shown in table 1.
Whitepapers & Resources
Standard & Poor’s assumes that the recovery of proceeds from the foreclosure and sale of repossessed properties occurs 18 months after a payment default in U.K. transactions (i.e., if a default occurs in month one, then recovery proceeds are received in month 19). The value of recoveries will be equal to the defaulted amount less the WALS. The time taken to repossess and sell a property can vary widely across the European countries, primarily because the legal procedures required before a lender can repossess and sell a property differ across jurisdictions (see table 2). Standard & Poor’s will therefore adjust the foreclosure period for each country to account for this.
Note that the WALS used in a cash flow model will always be based on principal loss, including costs. Standard & Poor’s assumes no recovery of any interest accrued on the mortgage loans during the foreclosure period. In addition, after the WAFF is applied to the balance of the mortgages, the asset balance is likely to be lower than that on the liabilities (a notable exception is when a transaction relies on overcollateralization). The interest reduction created by the defaulted mortgages during the foreclosure period will need to be covered by other structural mechanisms in the transaction (e.g., excess spread).
The liquidity stress that results from short-term delinquencies, i.e., those mortgages that cease to pay for a period of time but then recover and become current with respect to both interest and principal, is also modeled. To simulate the effect of delinquencies, a proportion of interest receipts equal to one-third of the WAFF is assumed to be delayed. This applies for the first 18 months of the recession and full recovery of delinquent interest is assumed to occur after a period of 18 months. Thus, if in month five of the recession the total collateral interest expected to be received is £1 million and the WAFF is 30%, £100,000 of interest (one-third of the WAFF) will be delayed until month 23.
Three different interest rate scenarios – rising, falling, and stable – are modeled using both high and low prepayment assumptions. Interest rates always start from the rate experienced at the time of modeling. For example, in the rising interest rate scenario, LIBOR (or EURIBOR) rises by 2% per month to a ceiling of 18% (12%), where it remains for the rest of the transaction’s life. Where there is a longer than average foreclosure period (e.g., Italy or Greece) the effect of high interest rates over the life of the transaction is unduly stressful, and the interest rate is allowed to ramp down after three to four years. For falling interest rates, interest rates fall by 2% per month to a floor of 2%, where they remain for the rest of the transaction’s life. For stable interest rates, the interest rate is held at the current level throughout the life of the transaction. Note that in the ‘AAA’ scenario the interest rate increase will not begin until month 13. Also note that interest rate scenarios will be revised if there is sufficient evidence to warrant it.
Transactions are stressed according to two prepayment assumptions, high and low. These rates of prepayment are differentiated by country of origin, as shown in table 3.
Prepayment rates are assumed to be static throughout the life of the transaction and are applied monthly to the decreasing mortgage balance. Standard & Poor’s reserves the right to increase the high prepayment assumption in the event that historical prepayment rates are at high levels, or the transaction is particularly sensitive to high prepayments (e.g., the transaction relies heavily on excess spread).
It should be noted that in a ‘AAA’ scenario an expected prepayment rate (e.g., 8% to 15%) will be modeled before the recession for the first year of the transaction. This is applied for both the low and high prepayment scenarios, to ensure that the WAFF is applied to a consistent asset balance in month 13 (the ‘AAA’ scenario recession start month).
In combination, the default timings, interest rates, and prepayment rates described above give rise to 12 different scenarios, summarized in table 4.
Unless the transaction has the benefit of a GIC with an appropriately rated entity, Standard & Poor’s assumes that the transaction will suffer from a lower margin on reinvested redemption proceeds and other cash held in the vehicle than the margin being received on the underlying assets. If proceeds are received and reinvested throughout the quarter, and the long-term rating on the GIC account provider is lower than that on the rated notes being subjected to the stress, then the reinvestment rate is assumed to be the index less a rating-dependent margin, with a floor of 2%. The rating-dependent margin is a multiple of the contractual margin. The multiple used for this calculation varies from one at the ‘A’ level to five at the ‘AAA’ level.
Mortgage payments from borrowers are typically paid by direct debit into a collection account, transferred to a transaction account in the name of the issuer, and finally credited to the GIC account. The degree to which an insolvency of the originator would affect the cash flow from the assets depends, therefore, on the collection account’s characteristics.
The amount at risk depends on the timing of payments from borrowers and the frequency with which these funds are transferred to the transaction account. If all borrowers pay on the same day of the month, then even with daily sweeping of the collection account, up to one month’s cash flow from the assets is potentially at risk.
The collection account is often not in the name of the issuer, as most originators do not want to ask borrowers to change their direct debit instructions as a result of securitization. Under English law, if the issuer has been granted the benefit of a properly executed declaration of trust over the collection account, then insolvency of the originator should not result in a loss of funds, but should only involve a simple delay. This risk will need to be modeled appropriately for each transaction, but normally results in the U.K. in a delay of one month’s cash flow for three months over an interest payment date.
In other European countries, insolvency of the originator is more likely to result in a loss of funds, the amount of which depends on the frequency of the transfer of money from the collection to the transaction account. This amount is generally modeled as a loss of interest and principal in the first month of the recession.
All the issuer’s foreseeable expenses should be modeled (e.g., mortgage administration fees, trustee fees, standby servicer fees, cash/note administration fees, etc.). These expenses should also include any tax liability the issuer may have. These fees are either a fixed amount each year, or are sized as a percentage of the outstanding mortgage loans (or a combination of both).
Standard & Poor’s normally requires a schedule of these expenses to be provided. In addition to foreseeable expenses, the model should contain amounts sized for contingent expenses, such as the need for the trustee to register legal title to the mortgages in the event of insolvency of the originator. This amount can vary from £150,000 to £300,000, depending on the size of the transaction, and can be modeled either as a separate contingency reserve or as a “haircut” to the reserve fund.
In general, notes are not written down as losses are experienced on the assets. Instead, principal losses experienced on the mortgage pool are recorded in a principal deficiency ledger, which tracks the extent to which the principal balance of liabilities exceeds that of the assets. At each rating level, Standard & Poor’s requires that principal deficiencies do not exceed the existing subordination.
For example, in a transaction with £100 million ‘AAA’ senior notes, £9 million ‘A’ junior notes, and a £1 million reserve fund, the principal deficiency at any point in time should not exceed £10 million in the ‘AAA’ runs and £1 million in the ‘A’ runs. If there is insufficient income to fund the principal deficiency, however, Standard & Poor’s considers the risk to a transaction to be low if the principal deficiency is remedied within a short period of time using excess spread.
Basis risk occurs when the value of the interest rate index used to determine the interest payments received from the assets differs from that of the liabilities.
This can occur when assets and liabilities are linked to different indexes (e.g., mortgages are linked to one-month LIBOR, liabilities to three-month EURIBOR), or both are linked to the same index but set on a different date (mortgage interest rate set on the first of the month, liability interest rate on the 20th, say). Here, there is the risk that the index for the assets falls below that of the liabilities, such that asset interest payments are insufficient to make the required payments to the liabilities.
Where this risk is not hedged, Standard & Poor’s will assess the historical performance of the indexes in question, and calculate the difference over a certain time (e.g., 20 days in the above example) that has been experienced historically. The average difference between the indexes is then calculated, assuming that in periods where the index for the mortgages has been higher than that for the liabilities, the difference between the two is assumed to be zero. This average is then subtracted every month from the asset margin.
In addition, two spikes in the liability interest rate index are also modeled. The height of each spike is determined as the maximum difference between the two indices, and occurs at the beginning of the first two years of the transaction.
In summary, Standard & Poor’s now assess most European RMBS cash flow transactions using internally created bespoke models, and replicate and stress the exact structure of each transaction. As the European RMBS market evolves and matures, Standard & Poor’s will assess the adequacy of the current criteria and further develop it as necessary.