Liar's Loans

How misaligned incentives between brokers, banks and borrowers encouraged widespread falsification on mortgage applications.
August 31, 2009
Print this page

During normal times, the rate of mortgage delinquency is between 1 and 3 percent. The average delinquency rate on home loans is just over 10 percent at this stage in the credit crisis. At a handful of banks, the rate is about 25 percent.

While many explanations have been offered up to explain these disproportionately high delinquency rates, professor Wei Jiang believes the best way to understand what went wrong is to look closely at micro-level data. Jiang worked with Ashlyn Nelson of Indiana University and Edward Vytlacil of Yale University to examine the documentation of more than 700,000 loans issued by one bank that recorded a 25 percent delinquency rate.

The bank provided data that included every piece of information recorded at loan origination — including address, reported income, length of time in job and other similar data that lenders typically use to assess the credit-worthiness of mortgage applicants, a large field of data that allowed the researchers to determine how the bank made each loan position.

The researchers categorized each of the mortgages as either high-documentation (hi-doc) or low-documentation (low-doc), depending on the amount of supporting data accompanying the loan application. Across both categories, Jiang and coresearchers found that the greatest delinquency rates came from the broker channel — loans originated by brokers were 50 percent more likely to be delinquent than loans originated by the bank. They also confirmed that brokers approached lower quality borrowers — people with lower income, lower credit scores and who live in poor neighborhoods.

But Jiang and her coresearchers found that while low-doc borrowers had higher rates of delinquency, the loan applicants actually looked better on paper than high-doc borrowers. “Low-doc borrowers had slightly better credit scores, slightly better incomes — slightly better everything,” Jiang says. “But we estimate that low-doc applicants exaggerated their income by about 20 percent on average.” A look at documentation alone without considering the point of origin would not reveal a true sense of default risk, but a look at origination alone could provide a good sense of which loans are at higher risk of default.

Jiang says the problem with the low-doc loans is not necessarily a visibly lower lending standard. Rather, it’s the lack of verification of the numbers provided in loan applications. “Is there undisclosed debt? Is the job stable? Neither the bank nor its brokers held the application data to any standard verification test.”

The highly imperfect result of such laxity reflects a common economics problem. “When you delegate a job to someone else,” Jiang says, “that person doesn’t take your full interest to heart.” The interests of the bank and the broker are misaligned: the banker will eventually take on the loans as assets, but the broker’s incentive stops once the deal is closed and a commission is earned.

Consistent with this, Jiang found that correspondent brokers — those who contract through a single bank and thus have an interest in maintaining a good relationship with the bank — are far less likely than noncorrespondent brokers to bring in loans that later result in high delinquency rates.

A second pair of misaligned interests exists between banks and borrowers, who are not always motivated to provide accurate information to the bank. Jiang found that low-doc borrowers whose applications were not verified tended to exaggerate their income by about 20 percent: hence the higher the reported income the greater the likelihood of delinquency. (In a full-doc loan the reverse is true: the lower the reported — and verified — income, the greater the likelihood of default.)

The bank also applied lower standards to mortgages that were more likely to be securitized. “People argue that a bank cares more about broker lending standards since the bank gets stuck with the loan permanently,” says Jiang. “But if a bank can sell the loan to the secondary market three weeks after it’s made, then the bank cares less about standards.”

In some sense, securitization makes brokers out of banks. And during the period of high securitization, in 2005-06, when the worst loans — with the highest default rates — were issued, lower standards were applied to mortgages, not only by brokers but also by the bank itself. But buyers of securitized mortgages cherry-picked only the better loans, and banks eventually found themselves stuck with many of the most risky loans.

The industry standard is typically half-bank, half broker-originated loans, but the bank Jiang studied was 90 percent broker-based and 10 percent bank-based — explaining why the delinquency rates for the bank in Jiang’s study were so high above the industry norm. There are a handful of other banks whose mortgage default rates hover at around 25 percent. Jiang suggests the same underlying numbers might be found in these other banks, because it is likely that they mostly relied on noncorrespondent brokers.

Given the problems associated with the broker channel, why would banks use brokers at all?

“These banks all had more than 50 percent annual growth for three or four years,” Jiang says. “You cannot build branch staff that quickly; the only way banks could achieve that growth rate was to use the mortgage broker channel to hone in on marginal borrowers.”


Wei Jiang is associate professor of finance and economics at Columbia Business School.

Wei Jiang

Wei Jiang is Arthur F. Burns Professor of Free and Competitive Enterprise in the Finance and Economics Division, and the Vice Dean (for Curriculum and Instruction) at Columbia Business School.  She is also a Scholar-in-Residence at Columbia Law School, a Senior Fellow at the Program on Corporate Governance at Harvard Law School, and a Research Associate of the NBER—Law...

View full profilePersonal Website

Read the Research

Wei Jiang, Ashlyn Aiko Nelson, Edward Vytlacil

"Liar's Loan? Effects of Origination Channel and Information Falsification on Mortgage Delinquency"


Download PDF View abstract/citation