Where Has All The Money Gone?
July 15, 2008
It will take years to sift through the mortgage market's wreckage before we know all the details behind its collapse. But public policy can't wait that long to draw some conclusions about the role played by low down payment home loans to people of modest means. This is of particular importance to federal bank regulators, who are responsible for enforcing the Community Reinvestment Act (CRA), and to whoever turns out to be the regulator for Fannie Mae and Freddie Mac with authority over their housing goals.
Plenty of commentators have suggested in the last year that more aggressive lending to people with lower credit profiles and little or no money to put down are drivers of the crisis. But at least two sources of information suggest that while these borrowers' performance has been poorer lately than in prior periods, it was other types of lending to different types of borrowers that is putting the hurt on lenders, investors and guarantors.
Fannie Mae built one of the largest portfolios of community lending assets in the country in the last 20 years. Starting with its Community Home Buyers© products offering the first standardized 5 percent and later 3 percent down payment products, the company expanded and tweaked these products until it incorporated its My Community Mortgage© product into its Desktop Underwriter© automated underwriting software in 2006. From 1996 until 2006 I oversaw much of the development work on these products.
In 1998, Fannie Mae joined with the Ford Foundation and the Self Help Ventures Fund to launch the Community Access Program, an owner-occupant mortgage lending partnership aimed specifically at low-and moderate income, low wealth households. Ford also funded a separate ongoing research effort at the Center for Community Capital at the University of North Carolina to provide a rigorous evaluation as the program evolved.
In a presentation prepared for a Neighborworks America symposium in Cincinnati in May, 2008, the Center's former director Michael Stegman summarized some of the ongoing research's findings on performance of these loans. Comparing their performance from the beginning of the program in 1998 through September, 2007, the UNC research shows that the Community Access loans experienced 90+ day delinquencies (where the borrower fails to pay for at least three months) at rates that were higher than prime fixed rate mortgages, but significantly lower than FHA-insured loans, subprime fixed rate loans and subprime ARM loans. Compared to subprime loans, Community Access loans were significantly slower to show the first 90+ day delinquencies than either fixed or adjustable subprime loans, and in later years of the program significantly slower than prime, fixed rate loans. This performance held pretty steady until 2006, when these trends showed a marked deterioration, with the 90 day delinquency rate rising much faster than it had for earlier "vintages," but still significantly better than subprime loans from the same period.
It was assumed at the time these loans were underwritten that they would not perform as well as prime loans, and they did not. Pricing decisions were made on those assumptions, and they seem to have been borne out. What the analysis also shows, however, is that well underwritten loans to people of modest means with low down payments far outperformed subprime loans issued during the same periods.
More Data
The other interesting source of information about loan performance is Fannie Mae's quarterly investor information summaries, the latest of which was released in May, 2008. One table in particular, titled "Fannie Mae Credit Profile by Key Product Features (page 24 in the linked file), offers some tantalizing insights into Fannie Mae's losses.
The table does not allow a direct evaluation of the company's community lending products. It only offers partial slices of data. So, for instance, the table shows that the company has a total single family credit book of $2.606 trillion, and within that a $128.1 billion credit exposure to loans with credit scores below 620 (the usual cut off before moving into subprime borrower territory, and the cut off for any loans under the My Community Mortgage© product). This was 4.9 percent of the total credit exposure, but accounted for 14 percent of the credit losses in the first quarter of 2008. That's a multiple of slightly less than 3...not good, but certainly not enough to blow them up with such a relatively small base.
Similarly, the table shows that Fannie had $258.6 billion in loans with down payments of less than 10 percent. That was 10.3 percent of the credit book, but accounted for 17.4 percent of the first quarter credit losses. That's a multiple of about 1.7.
Loans with credit scores below 620 and less than 10 percent down accounted for $30 billion, or 1.2 percent of the credit book, but 6 percent of first quarter credit losses. That's a multiple of 5. This is not too surprising when you combine crummy credit histories with low down payments.
So far the story seems to be that loans at the "tail ends" of the credit spectrum are doing more poorly than their share of the total would suggest.
Liar, Liar, Loan's on Fire!
But the table also accounts for Alt-A and subprime loans. The latter made up a very small piece of the credit book, only $8 billion, or 0.3 percent of the book, and 1.4 percent of the losses. Around a multiple of 4, but a very small nominal amount.
Alt-A, on the other hand, accounted for $310.5 billion, or 11.2 percent of the total credit book of $2.6 trillion, but....wait for it....42.7 percent of first quarter credit losses. That's a multiple of nearly 4 on a helluva base. Compare this to the low downpayment loans - 10 percent of the credit book, but only 17.4 percent of the losses.
Alt-A loans were supposedly made to people with good credit but with special flexibilities, like income that was reported but not verified, or no stated assets, and so on. They seldom had mortgage insurance (only 40 percent did, according to this table, compared to 92.7 percent of those with less than 10 percent down), which means Fannie Mae is much more exposed to losses from these loans. They also tended to be much higher balance loans, often were accompanied by separate second mortgages from other lenders that actually drove up the overall LTV, and were concentrated in states with rapidly escalating and now falling home prices.
These loans have become known in the industry as "liar loans." As in, lied about income, lied about assets. Bankers forgot a key principle of the Reagan era - trust, but verify. The cratering performance of these loans is one result.
Fannie Mae's table doesn't analyze the loans by exclusive category; many loans fit into more than one. Low credit score and low down payment loan numbers are partially or fully included in the Alt-A numbers, and vice versa, so comparing the ratios is not totally apples to apples.
A Little Knowledge...
These numbers are tantalizing, but ultimately frustrating, because they still do not allow a reliable analysis of how the products most specifically targeted to low wealth, low income borrowers are performing and what share of Fannie Mae's losses they account for.
Likewise, there is no public information available to analyze the performance of billions of dollars of specialized loan products that regulated banks put on their books to help them comply with Community Reinvestment Act (CRA) requirements.
My colleague and former Fannie Mae-er Ellen Seidman has disposed handily of suggestions that CRA is the root cause of the mortgage market's meltdown. I support her analysis 100 percent. But all of the contextual facts still do not answer the very important question of how the loans made to satisfy CRA requirements - or for Fannie Mae, their legislative housing goals - actually are performing and what lessons regulators, lenders and advocates should be learning from the last 20 years' experience.
There are no published data from banks about their loan performance, just as there is scarce product line information from Fannie Mae and Freddie Mac. It would be in the regulators' and the public's best interests to find a way to get this information from the lending community in order to shape the regulatory environment based on actual facts rather than self-serving or uninformed assertions. OFHEO could do this analysis on Fannie and Freddie's books. The OCC and the Fed could do it for regulated banks.
For the long term, what's puzzling and sobering is that having invested so much time and effort in these initiatives, regulators and industry have so little information about the performance of these loans. Fixing that gap would be a good goal for the next Administration.