A Closer Look at GSE Credit Losses

Click here for a printer-friendly version of this article

The news the bailout of the GSEs is far from over has caught a number of observers by surprise. The TARP capital injections have been largely repaid. Many apparently expected that the same would hold for the Fannie and Freddie bailout. Rather than reporting news of continued losses matter-of-factly, articles in the New York Times and Wall Street Journal seem to be written from the perspective of someone wondering how on earth the ongoing bailout could possibly cost so much.

Some suspect foul play, believing that the scale of the losses are evidence of a conspiracy where Fannie and Freddie have been required by conniving regulators to buy bad loans from banks eager to transfer credit losses to taxpayers. This line of reasoning is easily refuted, however, by a closer look at the GSE financial reports from the past decade. Amazing as it may seem, the losses generated by Fannie and Freddie come from the asset-backed securities (ABS) and loans acquired during the “boom” of 2005-2007. Over 93% of losses are accounted for by securities and loans the GSEs already owned or guaranteed as of December 31, 2007. The remaining 6% of losses come from loans originated in 2008 and 2009 to fund new mortgages (or refis), not stinker loans acquired from banks.

Through the end of 2009, Fannie and Freddie required $125.9 billion in Treasury capital injections to maintain a positive net worth. By June, this total will increase to $144.8 billion. Fannie Mae lost $13 billion in the first quarter of 2010 and asked Treasury for an additional $8.4 billion. Freddie Mac lost $8 billion during the quarter and will need an additional $10.5 billion from Treasury to get back to zero. In 2007 and 2008, most of the losses came from fair value losses on subprime ABS. Although Fannie and Freddie generally did not buy subprime or Alt-A loans directly in significant quantities prior to 2005, they were the largest purchasers of subprime securities issued by other firms. Indeed, in its 2006 Annual Report (page 16), Fannie Mae lamented the decline in subprime origination that year: “because subprime mortgages tended to meet many of the HUD goals and subgoals,” the decline in subprime origination “has further limited our ability to meet these goals.” With their holdings of AAA subprime ABS worth 47 cents on the dollar at the end of 2009, Fannie and Freddie incurred huge losses on these securities that were reflected in 2008 financial statements.

However, since the start of 2009, the losses have shifted from market value losses on securities to credit losses on loans Fannie and Freddie purchased or guaranteed. Starting in 2006, Fannie and Freddie started to purchase Alt-A and other higher risk loans in large quantities to win back market share. According to Freddie Mac’s 2007 Annual Report, it purchased $227 billion in interest-only and Alt-A mortgages in 2006 and 2007 and retained exposure to $154 billion in Alt-A mortgages at the end of 2007 (page 94). In addition, Freddie Mac added $37 billion in negative amortizing mortgages or “option ARMs” during these two years (footnote 2 on page 102). Fannie Mae was a much larger purchaser of Alt-A mortgage loans, buying $350 billion from 2005 to 2007 (calculated from percentage of business volume estimates on page 129) and retained total Alt-A exposure of $344 billion (page 23) at the end of 2007.

The problem is that 2006 and 2007 were also the years house prices were at their highest. So by launching an aggressive campaign to win back market share in these years, Fannie and Freddie were acquiring enormous exposure to the subsequent declines in house prices. When house prices remain constant, the loss incurred on a defaulted loan with a 90% loan-to-value (LTV) ratio is unlikely to be more than 30%. But if the house price falls by 40% – as has been the case in much of the country where mortgage originations soared during 2004-2007 – the loss more than doubles to 66%.1

According to their most recent filings, Fannie Mae has $153 billion in bad mortgages (calculated from page 7). Freddie Mac has $115.5 billion (page 15). Roughly $97 billion of Fannie’s $153 billion in bad loans are concentrated in 2006 and 2007 (see serious delinquency rates by year on page 7). About $48 billion of Freddie’s loans are concentrated in these two years (page 16). To get a sense of how much the GSEs are likely to lose on these mortgages, look at Page 142 of Freddie Mac’s most recent quarterly filing (see table below). The “weighted average collateral severity” provides the expected losses for each category of loan. For example, losses on fixed rate Alt-A loans made in 2007 has been 58%. This means that the value of defaulted Alt-A loans from this year is about 42 cents on the dollar (1-0.58). Defaulted subprime loans from 2007 are worth only 34 cents (1-0.66). With defaulted 2006-2007 loans worth 40 cents each, Fannie and Freddie’s losses from just these two years could exceed $150 billion. And this assumes that no new loans from these two years default – a highly unlikely scenario to say the least.

The table also provides the cumulative default rates by loan and vintage. About 63% of variable rate Alt-A loans originated in 2006 and 2007 have already defaulted. When multiplied by the 57% loss severities, this means that every dollar Fannie and Freddie invested in adjustable rate Alt-A mortgages during these two years is now worth 64 cents, on average. This stunning evaporation of wealth is the source of their cumulative losses.

Today, Freddie Mac estimates that its assets are worth $59 billion less than its liabilities (page 53). This is the present value of expected future losses. Using a 5% discount rate assumption, this would translate to $95 billion in additional losses on current mortgages if such losses are announced at a rate of $5 billion per quarter over 5 years. Fannie Mae estimates that its assets are worth $145 billion less than its liabilities (page 51). The losses here would total $232 billion per quarter over 5 years at a quarterly loss rate of $12 billion. In total, the losses still embedded on Fannie and Freddie’s balance sheets could equate to over $325 billion over 5 years in nominal terms.

It is not surprising that people have responded to such extraordinary losses by wondering about a “backdoor bailout,” but close observers of the GSEs do not find these losses terribly surprising. The GSE model coupled a fundamental economic problem – the under-pricing of insurance – with a fundamental finance problem – a total lack of diversification – and added in a political problem – meager capital requirements that could not be raised because of the perception that tougher regulation was “anti-housing.” To get a sense of how silly this arrangement was, consider Freddie Mac’s 2004 Annual Report (page 21), where it explains that its guarantee fees had fallen to 17.5 basis points in 2004. This means that Freddie Mac received an average of $350 per year to insure a $200,000 mortgage. Assuming a 30% loss on defaulted mortgages, the break-even default rate would be 0.58%. With insurance premiums this low, Fannie and Freddie would have suffered enormous losses if the mortgage default rate rose to 1%!

The Senate decided to side-step the issue of what to do about Fannie and Freddie when it cleared the financial regulation bill. But this approach is not sustainable. Democrats that wish to avoid reform (for now) need Republican buy-in for this strategy. Otherwise Congress is laying the groundwork for a crisis in 2011 when credit losses continue to run at a combined $5 billion per quarter and the mounting public backlash pushes them to abruptly cut off funding. This could then lead to a crisis for the holders of more than $5 trillion in Fannie and Freddie obligations that have become completely dependent on taxpayer support.

To try and jumpstart this process, e21 contributors Donald Marron and Phill Swagel recently authored a new reform proposal for Fannie and Freddie. Their plan, first published at Economics21 aims to “protect taxpayers and the overall economy from the systemic risk posed by the former GSE model, while ensuring that financing remains available for housing even in periods of credit market strains.”

Let’s hope the reform discussion gets going – and the U.S. finally gets ahead of the GSE disaster.

Footnotes

1 Assume a 40% transaction cost associated with foreclosure. With constant prices, the collateral value declines from 1 to 0.6 and the loss is 30% (0.6-0.9 mortgage). With a 40% house decline, the same 40% transaction costs imposes a 66% loss (0.34-0.9 mortgage).