One University professor claims he has identified a main culprit of the housing collapse at the center of the country’s current economic crisis.
In a study, Uday Rajan, an associate professor of finance at the Ross School of Business, found that the statistical models used by banks to assess which loan applicants may default on their loans played a paramount role in undermining many of this country’s financial institutions.
In determining which loan applicants would likely default on their loans, Rajan found that many of these models relied too heavily on ‘hard’ information (credit scores, etc.) as opposed to ‘soft’ information (situational factors in the applicant’s life).
Rajan’s study concluded that models tended to rely so heavily on hard information because it was thought to be more credible than soft information when risky loans were repackaged and sold to investors.
Soft information is derived from questions the banks pose to potential loan recipients relating to things like job security and the potential of major upcoming expenses.
Rajan, along with two colleagues, examined data on securitized subprime loans issued from 1997 to 2006 to determine the reason why many models failed to predict failure.
Securitization is the process by which banks package various assets — like risky loans, for example — into securities, which they then sell to investors.
The researchers found that as the rate of securitization increased, so did the reliance on hard information about loan applicants. This was because, Rajan said, the hard information is easier to convey to potential investors as the assets were packaged and sold, repackaged and resold, and so forth.
Without securitization, the lender is directly hurt if the borrower defaults. However, after the package is sold as a security to an investor, the bank or lender no longer bears the full risk of the applicant potentially defaulting on the loan. The responsibility then falls, at least in part, on the investor.
According to Rajan, lenders were more apt to collect soft information from borrowers before the unprecedented growth in securitization after the year 2000.
“A lender has an incentive to verify undocumented information, or soft information, about the borrower,” Rajan reported, “but the incentive to acquire soft information about borrowers is lost under securitization, since only hard data can be transmitted credibly to the investor.”
This created a situation in which high-risk applicants whose hard information may have seemed credible were able to obtain loans, even though their soft information showed they weren’t likely to pay the money back.
When investors purchased securities containing these high-risk loans, they were presumably unknowingly putting themselves at a great deal of risk, leading to what has now become the subprime mortgage crisis.
But the crisis is not entirely the fault of statistical models’ failures, Rajan said. Rather, it was the veiling effect that such models had on greater economic problems that truly got the economy in the mess it is today.
“Statistical models that relied on hard information did not directly cause bad loans to be made,” Rajan said. “However, a reliance on such models likely prevented investors from learning about the extent of the problem.”
Rajan said he hopes the study will have far-reaching implications for the financial sector as a whole. He said it will help to develop “more effective regulation of the financial markets and financial institutions” to prevent similar scenarios from playing out in the future.
Rajan has earned numerous rewards for his work in the financial sector including the GSAM award for best paper in the Review of Finance and the NYSE award for best paper on equity trading at the WFA meetings, according to his University biography.