The Kroll Bond Rating Agency has issued an update of its residential mortgage-backed securities model methodology, Residential Mortgage Default and Loss Model. Before the financial crisis, ratings models seemed to be very reliable, data-driven models of probity and caution. We have since learned that different mortgage vintages (the year of origination) can behave very differently and ratings models could be based on simplistic assumptions. Hopefully, the updated Kroll model does not suffer from those flaws, although their key takeaways seem pretty basic to me:

• Underwriting standards are the fundamental determinant of mortgage quality.
• Negative home equity creates a major incentive for borrower default, resulting in substantial credit loss.
• Credit scores continue to have value as a relative indicator of risk.
• Inflation of real home prices above the long-term mean is unsustainable and represents increased credit risk. (4-5)

Kroll’s update does include some interesting revisions, including,

Reduced default expectations for purchase loans. It has long been observed that purchase loans generally have lower default risk than refinancings, all else being equal. This is attributed to the fact that a purchase represents an actual arms-length transaction which yields a more accurate view of a home’s value than an equivalent refinancing transaction. However the pre-crisis mortgage vintages showed high levels of default associated with purchase mortgages. This was largely due to the practice of extending credit to first time homebuyers, often on very favorable terms despite these borrowers having little credit history or poor credit history. This poor performance by purchase loans was reflected in the historical data regression analysis used to develop the RMBS model.

Based on analysis focused on both jumbo and conforming prime mortgages, KBRA has found that, for these loans, the traditional benefits of purchase loans remain well established, and we have adjusted the model ‘s treatment of purchase loans to reflect lower default expectations relative to equivalent refinancing mortgages. This revision is effective with the publication of this report.

*     *     *

Penalty for high debt-to-income (DTI) loans. While the KBRA RMBS model does not contain a specific risk parameter based on DTI, it is our opinion that very high DTI loans can bear significant incremental risk. When we began to encounter newly originated loans with back-end DTIs in excess of 45%, we assigned an additional default penalty to such loans. This has been documented in presale reports for those rated RMBS backed by loans with high DTIs. (3)

Time will tell if Kroll got it right . . ..