Research Bears Out Subprime Pricing

Mr. Pennington-Cross, a senior economist at the Office of Federal Housing Enterprise Oversight, authored a working paper to address loss distributions between subprime and prime mortgage loans. The following is an excerpt from the working paper, and the full version is available on OFHEO's website (www.ofheo.gov). OFHEO said the conclusions of its working papers are those of the author and do not imply concurrence by other OFHEO staff.

Research has shown that borrowers of subprime mortgages often default more frequently than do borrowers of prime mortgages. In addition, borrowers of subprime mortgages also appear to react differently to changing interest rates and other economic conditions than do other borrowers. Few studies have explored the link between the likelihood of prepayment and default and the corresponding losses associated with defaults.

When mortgage borrowers default, they generally impose losses on lenders, mortgage insurers, mortgage-based security holders and others. The losses differ by entity and each prices its services accordingly. In addition to setting prices these entities also use non-price credit rationing techniques (credit scores, loan-to-value ratios, income verification, etc.) to limit their exposure to risks. As a result, each institution bears an accompanying distribution of potential losses that varies by borrower and lender and is priced accordingly.

Due to larger variations in lending standards ("flexible lending") and the more precarious financial condition of subprime borrowers it is more likely that subprime lending has both higher loss rates and more variability in loss rates. If priced appropriately, however, no shortage in the supply of subprime loans should result.

Probability distributions of loss rates are not directly observed, but they may be estimated. Using a sample of Fannie Mae and Freddie Mac 30-year, fixed-rate prime and subprime loans, this study estimates probability density functions of loss rates for these two types of mortgages.

The results show that the mean of the subprime loss distribution is 5.15 times higher than the mean of the prime loss distribution when private mortgage insurance is used and 6.0 times higher when PMI is not used. In addition, PMI reduces expected default losses by more than 85% for both prime and subprime mortgages.

One additional feature available from the distributions of losses is that these distributions permit the determination of the economic capital requirements associated with various risk tolerances. Economic capital is defined as the amount of capital that is required to cover unexpected losses in a portfolio. For instance, the results of the simulations can be used to derive a 0.08% economic capital/asset offset for prime mortgages and a 1.14% economic capital/asset for subprime mortgages without mortgage insurance.

Loans in which the contract rate on the mortgage was at least 100 basis points higher than the average rate reported in the Freddie Mac Primary Mortgage Market Survey at origination were identified as subprime. This study added to the empirical literature by analyzing loan defaults and prepayments during the period 1995 through 1999. It included prime and subprime loans in the sample obtained from Fannie Mae and Freddie Mac. The availability of credit scores limited the maximum age of the loan to five years. This analysis does not reflect the entire subprime market, but rather those loans purchased by Fannie Mae or Freddie Mac.

In general, the results are consistent with the premium paid by the borrower. Interestingly, "B" rated loans have a slightly lower mean expected loss rate than "B+" rated loans. This may be an indication that the increase in the required downpayment reduced losses more than the decrease in credit history quality increased losses.

The estimated loss severity model shows that the amount recovered from the sale of the foreclosed property is sensitive to the loan-to-value ratio at foreclosure, state level foreclosure laws, and whether the loan is prime or subprime.

If no PMI is used to mitigate loss exposure, the typical (median) loss on the synthetic subprime portfolio is approximately 5.23 times higher than the synthetic prime portfolio. The results also show that risk-sharing agreements based on downpayment requirements can substantially reduce expected losses. For instance, the simulations show that the introduction of PMI reduces median losses for synthetic prime and subprime portfolios by over 85%. Estimates of economic capital requirements are much lower than current regulatory requirements, which may help to explain the high rate of securitization.

Lastly, an analysis using a representative loan type simulation revealed that expected mean loan losses generally follows pricing patterns in the subprime market, despite the fact that the highest risk classification loans require larger downpayments to mitigate the higher default rates.

Copyright 2003 Thomson Media Inc. All Rights Reserved. http://www.thomsonmedia.com http://www.mortgageservicingnews.com