With the Corker-Warner bill, we have a little more insight into what private-public risk-sharing in a post-GSE mortgage market might look like. But it's debatable whether the private sector would or would not bear a sufficient share of the risk.
Such uncertainty requires us to look at the assumptions behind this latest proposal to put taxpayer funds at risk on mortgages that go bad in the future. How can we be sure that what has been proposed is actuarially fair from an insurance perspective?
The story starts deep in the bill's language, where it requires private capital to take the hit on at least the first 10% of losses experienced in a security, leaving the new Mortgage Insurance Fund on the hook for losses in excess of that amount. There's not a lot of detail on exactly where that 10% figure comes from.
It all comes down to the likelihood of realizing certain scenarios for credit losses over time and stipulating a threshold based on extremely rare events. The problem is that establishing the probability of an exceptionally bad event isn't all that easy, as we found during the mortgage meltdown.
Before any hint of a problem became apparent, most industry experts severely underestimated the likelihood of a national home price decline of the magnitude observed during the crisis. Nothing has changed all that much in our ability to assess what risk practitioners call "tail risk." That's unfortunate by itself, but becomes even more problematic down the road for taxpayers who will be on the hook for any crisis.
At first glance, a 10% credit support provided by the private sector in a senior-subordination structure a classic type of mortgage structured financing being repurposed this year under a credit enhancement pilot with the GSEs seems like adequate protection.
Taking Fannie Mae's worst origination year (2007) as a benchmark, applying a 50% loss severity rate against a cumulative default rate on this book of 15% (the actual rate is currently about 11.5% but it hasn't yet peaked), you end up with a 7.5% lifetime loss rate on Fannie's worst book of business under the largest decline in home prices since the Great Depression. Based on this simple example, 10% credit protection seems like a good deal for taxpayers, particularly since the credit quality of the continuing business is comparatively pristine and therefore much less likely to default.
Reinforcing the 10% credit support number is an analysis conducted by Andrew Davidson and Co. on behalf of the Bipartisan Policy Center Housing Commission (which ultimately recommended a public-private risk-sharing scheme similar to Corker-Warner). As the basis for the analysis, ADCo ran Freddie Mac loans originated in 2012 through 20 different housing market scenarios with varying probabilities of occurrence. In a standard senior-sub structure, credit losses are absorbed according to a schedule where the first losses are taken by the junior bondholders. Once those are exhausted, the mezzanine bondholders take losses above that of the junior bondholders up to the amount of bonds held by the mezzanine credit investors. Only after the junior and mezzanine bondholders have taken all of their losses do the senior bondholders take a hit.
In the analysis of Freddie loans, under a scenario with no mortgage insurance, the subordinated tranches would suffer losses as high as 10% of the pool only 0.5% of the time, with a loss severity of 44%. In other words, 99.5% of the time the holders of senior bonds would not experience any losses. Again, this makes a 10% threshold for federal taxpayers seem very conservative. However, that loss scenario assumes that home prices fall 25%.
There's the rub. Including the Great Depression along with the more recent mortgage crisis, we have two scenarios where home prices declined more than 30%. Over a period of 80 years or more, that makes the likelihood of an event that would result in home prices falling more than 30% five times more likely than the relatively timid extreme scenario in the Bipartisan Policy Center's analysis. While we saw earlier that GSE lifetime losses from the latest crisis are likely to turn out well under 10%, the point is that a rigorous analysis should be required as part of any legislation instead of a hard number at this point based on incomplete facts.
It could very well be that the 10% figure in the proposal is too high on an actuarial basis. Or it could be too low. Given what's at stake, it seems somewhat cavalier for Congress to be hardwiring a credit loss threshold without requiring a proper due diligence be conducted first.
Clifford Rossi is the Executive-in-Residence and Tyser Teaching Fellow at the Robert H. Smith School of Business at the University of Maryland.