Fitch: New Loss Data Provides Context for Expected Fannie Mae Deals
Fannie Mae enhanced its single family residential loan-level historical dataset on July 22 by adding loan-level loss data. In doing so, Fannie Mae increased transparency to the market in anticipation of an actual loss credit offering in fourth quarter-2015. To date, all CAS risk-sharing transactions have passed losses on defaulted loans to investors using a pre-determined loan loss severity schedule.
Fitch conducted an analysis of Fannie Mae's historical loss data and contrasted it with the fixed severity schedules used in CAS transactions to date. For 60-80 LTV loans, observed loss severities closely matched the severity schedule at all default levels. For 81-97 LTV loans, the observed severities were generally in line with the schedule for defaults above 10%, while at the lower default range observed severities were modestly lower than the schedule. In this comparison, historical loss severity averages included defaulted loans that subsequently cured and prepaid, or liquidated without a loss.
While actual and CAS scheduled severities are similar, the credit enhancement required for actual loss risk-sharing transactions may differ from the CE seen in CAS to date. 'Credit enhancement requirements for actual loss transactions will be driven by the particular credit, leverage and mortgage insurance profile of the pool, and may be higher or lower than the CE in existing fixed-severity transactions,' said Director Sean Nelson.
The report also compares Fannie Mae's loss data to Freddie Mac's loan-level historical loss data. For liquidated loans originated between 2003 and 2006 loss severities are very similar between the two GSEs. However, among loans originated before 2003 and after 2006, the data suggests that severities on Fannie Mae loans appear to be lower than those of Freddie Mac.
The weighted average loan attributes and geographic distribution of the properties are very similar between the two data sets, and Fannie Mae's apparent lower loss severities seem to be due to a combination of lower expenses, higher net sales proceeds, and higher mortgage insurance recoveries.
'The data suggests differences between Fannie and Freddie loss severities among loans with similar profiles, and points to certain drivers.' said Nelson. 'However, there may be subtle compositional differences between the two data sets that influence the severities.'
Due to differences in portfolio composition there is some variance in how the agencies selected their historical sample sets, which may result in differences between the two datasets that are not easily identifiable in the data, such as underwriting guidelines and layered collateral risk attributes. If present, such differences may affect observed loss levels in an otherwise 'apples-to-apples' comparison.
Комментарии