It nevertheless attempts to do a “multidimensional analysis” which evidently gives heavy weight to FICO (how “multidimensional” can an analysis be on a low doc/no doc loan?). As Tom Adams notes: FICO is not a terribly accurate predictor of performance for a subprime mortgage – FICO scoring only became widespread after 1999 and there was very little analysis available for its long term predictive ability for much larger loan balances than autos or credit cards. However, despite lack of sufficient data, years of true underwriting standards, such as debt to income, LTV, months of reserve, payment stubs and tax returns, were abandoned in favor of using FICO as the underwriting tool. As a note – subprime auto lenders used detailed scorecards for their borrowers, but FICO was not the primary component of their scoring. Other factors were deemed more important. These transactions have held up much better than subprime mortgages and they were on depreciating assets (which is why, of course, lenders had to be more careful). Also, the authors get the analysis of purchase vs. refinance wrong. With prime mortgages, purchase loans tend to perform better. However, subprime loans were traditionally refinance loans (usually cash out). In the early days, subprime lenders were fairly careful about who they lent to – their target market was people who had been in their homes for at least 5 years, had stable jobs and work history (and were able to document their income with tax returns etc, even if self employed), had equity and had encountered a financial difficulty – thus, the need for more cash. In the mid-2000s, the new target market became first time home buyers who had very little credit history (thus, making FICO score an even worse predictor, since it was based on “thin files”), had no money for down payments (thus, piggy back seconds) and a short job history. Basically, purchases were bad in subprime lending and had generally been avoided in the early days. Moreover, many lenders gamed labels like subprime and alt A. We had to create our own definitions. Also, many lenders provided poor information regarding documentation – each lender had its own marketing names for their documentation programs,so there was no uniform standard of what “limited documentation” looked like. This also presented an opportunity for lenders to game the system (and lenders like Countrywide were the worst abusers). DTI was also very inconsistent.
This serves to illustrate that many conventional analyses even now tend to miss the dramatic deterioration in underwriting standards. The use of low and no doc loans rose rapidly from 2004 onward, and these pools were particularly favored by the subprime shorts. Moreover, we now know how some aspects of the underwriting were abused, for instance, by the use of inflated appraisals, so analyzing historical data will not provide a full measure of the fall in underwriting standards. Yet digging into the comfortable narrative of the subprime shorts as heros, or at least harmless, would reveal yet another viper’s nest of bad practices and abuses. The officialdom seems determined to push onward with its “look forward, not back” stance, which means the perps will be able to engage in similar types of looting when the opportunity next presents itself.