Citigroup Global Markets is adding to its analysis of subprime mortgage securitizations to allow for better comparisons of credit performance, regardless of whether deals have mortgage insurance (MI).
Although deep MI mitigates losses, deals with that insurance often have less credit enhancement from other sources, such as overcollateralization, excess spread and subordination, said Ivan Gjaja, director in ABS research at Citi. By tracking credit enhancement levels, it is possible to compare losses between deals with and without MI, Gjaja said.
Citi added several new tables to its latest edition of the Home Equity Loan Lifecycle Overview (HELLO), dated Oct. 7. Gjaja said the new tables, including one that shows the average initial subordination and overcollateralization, allow investors to account for how deep MI affects the deals, at least indirectly.
"Deep MI is very common in the subprime mortgage market. So it's difficult to compare the collateral from one deal to another if the actual credit performance in one deal is distorted by insurance payments," Gjaja said. "What we did instead was to report the credit enhancement levels. So you can think of it as loss performance relative to the enhancement."
Deep MI - which provides coverage to mitigate the severity of losses on individual loans if they default - has become increasingly prevalent over the past few years. New issues with deep MI grew from 12% in third quarter 2000 to 41% in the second quarter of 2003, according to Citi's analysis of the deals rated by Standard & Poor's. Usage of deep MI peaked at 55% in third quarter 2001.
But as MI has become more prevalent, credit enhancement for new issues has declined. For second quarter 2003, credit enhancement was 14.2% for fixed-rate new issues, down from 20.2% in the first quarter of 2000. The highest level of credit enhancement was 21.7% in the fourth quarter of 2001, which was also the quarter with the lowest level of deep MI (12%). For adjustable-rate new issues, credit enhancement decreased from 24.7% in first quarter 2000 to 22.2% in second quarter 2003.
That is a concern for investors because there are risk factors that can reduce the effectiveness of MI, such as a greatly uneven distribution of loss severities across defaulted loans. Another possibility is that the insurer may not pay on the policy - for example, if there is incorrect or incomplete loan information, or the originator or servicer does not adhere to the guidelines - so the actual collection rates may be lower. Also, the insurer may become financially unstable and default on its policy.
The new charts in the HELLO report also help investors account for other issues that complicate deal comparisons, including collateral factors such as FICO scores and LTVs. Citi said average FICO scores on fixed-rate deals increased from 604 in first quarter 2000 to 629 in the second quarter of 2003, an indication of lower credit risk and therefore lower enhancement levels. Conversely, average LTVs were higher for 2002 deals, at 81%, than for 2000 and 2001 deals, at 77% and 78%, respectively - a trend that presumably led to increased enhancement levels.
When evaluating the performance of a securitization, investors may be better able to incorporate these collateral factors by relating losses to credit enhancement, Gjaja said.
"Basically, we think this is the correct way to compare deals with and without mortgage insurance," he said. "We think looking at the raw numbers without adjustments for credit enhancement is not a useful comparison."