The Office of Federal Housing Enterprise Oversight said late yesterday that it is considering a change on its assumption of losses during foreclosure, a change that could have dramatic effect on the formulas used to calculate risk-based capital requirements for both Fannie Mae and Freddie Mac. OFHEO said current loss severity equations "understate losses on defaulted single-family conventional and government-guaranteed loans." From the press statment:
OFHEO proposes two changes to the current rule. The first change is proposed because certain loss severity equations result in the Enterprises recording profits instead of losses on foreclosed mortgages during the calculation of the risk-based capital requirement. Unaltered, the current loss severity equations overestimate Enterprise recoveries for defaulted government-guaranteed and low loan-to-value (LTV) loans. The results generated by the current loss severity equations are not consistent with the Risk-Based Capital Regulation and result in significant reductions in the risk-based capital requirements of the Enterprises. The second change is proposed because the current treatment of FHA insurance associated with single-family loans with an LTV below 78 percent is inconsistent with current law. The proposed changes in the NPR would correct these deficiencies.
The most stunning part of this to me is the admission that current calculations surrounding loss severity are at best outdated -- and at worst, wrong. Here's the full proposal, and I have to admit it made the econometrician in me feel all warm and fuzzy. Essentially, the problem in layman's terms is that the RBC algorithm currently used employs a term that calculates recovery proceeds based on a) a factor for mean estimate of recovery proceeds and b) an estimate of loan-to-value based on mean prices that may be very inaccurate for specific areas. In other words, co-opting the NAR here -- can you believe it? -- all real estate is local and the OFHEO is finding that its global formulas are creating some problems in the current housing environment. The bottom line here appears to be that defaults are up on even low LTV loans in the GSEs' portfolios (because defaults are accelerating in general and because so many more firsts have become "low LTV" thanks to the housing boom), and that some areas of the country are experiencing much more severe price corrections than other areas of the country -- both affect the risk-based capital calculation and make it possible for a GSE to record phantom gains on foreclosure. (I should stress that this is only a phantom gain, the result of a problem in how the equation is constructed and not a reflection of actual profit or loss.) That being said, there is nothing phantom about how the proposed changes will affect the GSEs - the proposed changes could mean a drastic increase in risk-based capital requirements for each GSE. The changes would have had the effect of increasing capital requirements at Fannie Mae by $7.5 to $9.8 billion in the fourth quarter of 2006; Freddie would have seen its requirements increase by $4.5 to $5.4 billion in the same time frame.