When Stanford economist Jonathan Levin started studying subprime lending in 2004, he had no idea how hot the topic would become. His interest in the subject was piqued by a research proposal from an undergraduate student who had worked at a national auto retailer that issued car loans to low-income buyers. The company kept a massive database on its customers and their payment histories, and it was willing to share the data in a way that many lenders have since become reluctant to do. 

Speaking about it today, Levin’s voice still reflects the excitement he felt in seeing the numbers for the first time. “They provided a window into the behavior of a whole population of low-income Americans,” he says. Levin points out that as a general rule “the data visibility is not great on people with low incomes and bad credit histories.” His sample promised to illuminate a whole range of questions: “What does the demand for loans look like in this market? How do you explain the behavior of customers? And what steps can companies take to manage risk and stay profitable?” 

‘We’re trying to understand how contract terms, such as the required down payment in a financed purchase, affect the selection of customers and their later behavior.’

These questions fed directly into one of Levin’s broad research interests, which he describes as “building econometric models of demand and supply for markets where what’s being traded is a contract”—for example, a loan-repayment plan over a period of years. “This means modeling which kind of people choose which contract, and then how they behave. If you flip this around to the perspective of a firm or a regulator, we’re trying to understand how contract terms, such as the required down payment in a financed purchase, affect the selection of customers and their later behavior, and whether different contract designs lead to more or less efficient or profitable outcomes.” 

One of the clearest gleanings from Levin’s data on the used-car market is that subprime purchasers tend to “pay immediate attention to short-term cost and not much to the later cost,” Levin says. “Their decision of whether to make a car purchase depends less on what the long-term financial obligations are, and more on the here and now.” The data illustrate this strikingly—a $100 increase in down payment has the same effect on consumer demand as a $3,000 increase in overall car price. So when overall car prices go up, buyers tend simply to take out bigger loans. But when down-payment requirements rise, the immediate pressure causes many buyers either to take out a smaller loan—thus making a less expensive purchase—or else forgo the purchase. 

Lenders can take advantage of this dynamic to set down-payment levels that will prevent risky borrowers from taking out loans they can’t afford to repay. “In the case of the firm we’ve been studying,” Levin says, “it uses credit scoring to set the down-payment levels. Customers with a worse track record are required to make a larger payment; those who appear more likely to make payments can finance a larger share of the purchase.” 

Fortunately for the cause of efficient lending, Levin says, “credit scoring can do a very good job in terms of stratifying borrowers according to their default risk.” Additionally, advances in information technology have made it possible for companies to price risk “in much more sophisticated ways than was possible 20 years ago.” With better tools for collecting and analyzing consumer data, lenders are better able to identify risky borrowers and optimize downpayment levels (as well as interest rates) for individual customers. 

Of course, with all of these sophisticated tools available for issuing profitable subprime loans, one wonders how things went so drastically wrong on such a large scale in the subprime housing market. Levin has only recently begun taking a serious look at subprime lending in this specific context and he is cautious about extrapolating too far from the data that he has already analyzed. “I think it will take years to untangle everything that has happened,” he says. “Economists will be sifting through the data for a very long time.”

Riskier borrowers tend to demand larger loans, thus creating a double liability for unwary lenders.

But Levin’s research is already increasing understanding of the recent tribulations in the subprime housing market. For one thing, it seems that the irrational expectation of perpetually rising housing prices gave lenders a false sense of the overall default risk on mortgage loans. As a result, lenders may have been less than vigilant in imposing front-end barriers, such as substantial down-payment requirements, that would have screened out risky borrowers. 

Levin also points to securitization in the mortgage market as a possible trouble spot. In many cases, the originators of subprime housing loans simply chopped up the mortgages they had issued, packaged them in discrete bonds, and resold them to a diffuse network of investors who may not have been fully aware of the risk they were taking on. Though Levin hastens to add that he has not studied this closely, it is easy to imagine how disaster could follow from lenders insulating themselves from default risk in this way. 

Another contributing factor to the subprime collapse may have been the noble intentions of policymakers and public advocates who pressured lenders to issue subprime loans to facilitate homeownership for low-income families. There is a tension between this goal and the desire to keep default rates low. Today, some of the same people who habitually denounced yesterday’s overly exclusionary lending practices heap scorn on lenders for alleged “predatory” tactics of ensnarement and foreclosure. 

Levin is careful to note that his research does not directly address such large policy controversies. He emphasizes that market-design choices for companies and regulators alike often involve trade-offs. “There is typically no way to set up a market that dominates on every dimension,” he says, and the optimal arrangement can depend on competing values such as efficiency, distributive fairness, and revenue maximization.

Levin takes this temperate approach in the area of healthcare as well, where he and some colleagues have recently examined price inefficiencies caused by distortions in the health-insurance market. Health insurers have the capability to create increasingly sophisticated actuarial models (based on health histories and average costs related to known health conditions) for implementing risk-adjusted pricing of insurance premiums. The catch is that, due to federal tax incentives, most Americans get their health insurance through employer-provided plans, and federal law prohibits employers from charging different premiums on the basis of individual health characteristics. The result is a misallocation of health-insurance rates and resources, which causes a moderate but measurable “welfare loss” among insurees. 

Again, though, Levin is careful not to conclude too much from this observation. “It’s an interesting question as to what would happen if the law were changed,” he says. “It seems possible that most firms would not change their benefits policies,” since “it might be perceived as an issue of fairness.” He also points out that some of the price distortions in the health-insurance market come from natural informational problems, as many insurees refuse to divulge their private health-related habits and conditions. Further, Levin notes, there might even be some benefits to charging flat-rate premiums, since this practice tends to keep rates stable over the long term for a large pool of people who might otherwise face unpredictable yearly rate hikes. 

Beneath all the complexities, it is clear that much of what drives Levin is his excitement over the marriage of theory and practice that lies at the heart of the field of economics. This is what drew him to the discipline during his stint as a Fulbright scholar at the University of Oxford, where, after having taken only two economics classes as an undergraduate (he was an English and math major at Stanford), he was thrilled to see the “elegant mathematical models” of the academy being put into almost immediate practical use in government and the private sector. After getting his doctorate at MIT, he landed happily back at Stanford.

Anthony Dick is a former associate editor at National Review.