Laurie Goodman of Amherst Securities and formerly of UBS, has come out with a damning report, which estimates that the total losses at Fannie and Freddie could be as high as a mindblowing $448 billion. Keep in mind that so far the government has injected $112 billion into the nationalized entities. Yet if this estimate is correct, another $336 billion will have to be funneled from taxpayers. This money will have to come from new debt issuance and is certain to add to the multi trillion budget deficit. Also, putting the banker tax in perspective, the number is nearly three time greater than the $120 billion expected to be collected over a period of many years, and causing so little ruckus on Wall Street and so much posturing by the President.
We assume that all of the nonperforming and re-performing loans default, as do half of the performing loans with negative equity. Our rationale: once a borrower is 60+ days delinquent (our definition of non-performing), he has a very low cure rate—well under 10%. And the reperforming loans have been re-defaulting at an unbelievably high rate. So, for example, prime PLS re-performers have redefaulted at the rate of 10.9% per month over the past 3 months or 75% per annum. It is not unreasonable, as a first approximation, to assume that they eventually default. The hardest category to analyze are performing loans with negative equity. According to our numbers in Exhibit 13, over the past 3 months, those loans have experienced a default transition rate of 18% per annum and a prepayment rate of 10% per annum. If these numbers remain constant over the life of the mortgages, it suggests 64% of the pool will default, 36% will prepay. (Intuitively, a mortgage will either prepay or default, a tiny percentage pay on schedule to maturity. The % expected to default is [defaults/(defaults + prepays)], which is 18/28 or 64% for this loan cohort.) We believe this puts an upper bound on defaults, as it assumes no housing recovery over the life of the mortgages.
The top section of Exhibit 14 shows our loss estimation calculations assuming 100% of the non-performer and re-performers will default, 75% of the performing borrowers with >120 LTV will default and 25% of the performing borrowers with 100-120 LTV will eventually default. (Since 50% of the performing loans with negative equity have an LTV of 100-120 and the other 50% are >120 LTV, our assumption is the equivalent of a 50% default on the basket of performing loans with negative equity.) We further assume a 50% severity on defaulting loans, in line with the experience of PLS prime loans, corrected for loan size. Thus, we estimate losses of 9.59% on the GSE guarantee business. Interestingly, these numbers were very close to an OFHEO estimate that 9.4% losses were experienced over a 10-year period on the 1983 and 1984 loans made in the oil states (Arkansas, Louisiana, Mississippi, and Oklahoma) during the oil bust in the 1980s. (Texas was excluded, as its economy was more diverse. Houston was hit hard, Dallas was not.) This 9.4% overall loss rate reflects a 11.3% default rate with 61% severity on Freddie, an 18.5% default rate at 66% severity on Fannie.
Putting these estimates to actual numbers yields the following:
The Fannie guarantee business is currently $2.81 trillion, suggesting losses of $270 billion. The Freddie guaranty business is currently $1.86 trillion, suggesting losses of $178 billion. Thus, losses should total $448 billion. Note that we have used the same default rate and severity for Fannie and Freddie. In reality, the Fannie guarantee business has a higher percentage of risk layered loans than the guarantee business at Freddie. Consequently, these loss estimates are probably proportionately too low for Fannie, too high for Freddie. These gross losses will be distributed across four categories – write downs already taken by Fannie and Freddie and reflected in their loan loss provisions, future credit losses to be taken by Fannie and Freddie, losses absorbed by mortgage insurers, and losses absorbed by originators through put backs. Fannie’s loan loss reserves total $66 billion, $57 billion for MBS guaranty losses, $9 billion for loan losses. Freddie’s loan loss reserves total $30 billion, $29 billion for PC guaranty losses, $1 billion for loan losses. The remaining $352 million of losses will show up across the other three categories (Fannie and Freddie future losses, mortgage insurers, and originator put backs) over time.
Obviously these numbers are very assumption dependent, as illustrated in the bottom section of Exhibit 14. If we lower the probability of default to 15% for the performing loans with an LTV of 100-120, and 65% on performing loans with an LTV of >120, the blended default rate on negative equity borrowers is 40% [15%*(0.5) + 65%*(0.5)] and aggregate losses drop to 7.7% or $358 billion.
If Laurie is correct, not only will numerous additional episodes of housing-focused Quantitative Easing be needed, but such a deterioration, once it is acknowledged by the auditors who obviously have no interest in indicating how bad the situation is, would have significant political repercussions which would reverberate from the FHA all the way to the White House. Another implication is that the shadow inventory is likely to continue remaining unsold so as not to blow the bottom out of the supply side in the market, likely confirming that banks will continue to pretend the vast majority of real estate assets on their balance sheets are unimpaired when in actuality they deserve a haircut of 20%,30% or more.