When Good Incentives Lead to Bad Decisions
New research by Associate Professor Shawn A. Cole, Martin Kanz, and Leora Klapper explores how various compensation incentives affect lending decisions among bank loan officers. They find that incentives have the power to change not only how we make decisions, but how we perceive reality.
Among the culprits contributing to the recent financial crisis were bank loan officers who approved mortgage loans that were doomed to fail. Many of these frontline workers were motivated by bonuses and other incentives to approve quantity over quality. Critics decried their voracity. But new research suggests there was something at work beyond simple greed, setting the stage for deeper exploration of how incentives shape not only what we do, but also how we perceive reality.
"The question of incentives is fundamental to economics."
"The question of incentives is fundamental to economics," says Shawn A. Cole, an associate professor in the Finance Unit at Harvard Business School. "A frequent criticism of bankers in the recent crisis is that they took a lot of incentive pay in the years leading up to the crisis. And when their firms lost a lot of money, bonuses could not readily be reclaimed by investors in the firm who lost money."
In a new research paper, coauthored with World Bank economists Martin Kanz and Leora Klapper, Cole explores how various performance incentives affect lending decisions among bank loan officers. Not surprisingly, the authors found that loan officers were more judicious about issuing loans when their bonus incentives were tied to whether the loans performed well.
More surprisingly, they found that incentives actually have the power to distort loan officers' perceptions of how a loan will perform.
An experimental field study
In an experimental field study, the researchers set out to test the efficacy of three distinct incentive schemes for employees in charge of assessing risk and issuing loans: the origination bonus (in which officers are rewarded a commission for every loan issued); the low-powered incentive (in which officers receive small rewards for loans that perform well and small penalties for loans that perform badly); and the high-powered incentive (in which officers receive big rewards for issuing loans that perform well, but big penalties for loans that default).
To begin, the researchers collected a random sample of actual applications from entrepreneurs seeking commercial loans for the first time. They acquired the loan files from a large commercial lender in India, a country challenged by a dearth of verifiable financial data. As is common in emerging markets, many residents have no personal identification documentation, let alone a credit score. Loan officers often end up relying on qualitative information and intuition, all the while knowing that these loans may be vital to the nation's economy.
"Small businesses in emerging markets are thought to be among the large engines of employment and economic growth," Cole explains. "But they often find it hard to get access to financing because they lack the large amounts of collateral that banks typically require. So we were interested in studying this small business segment, which is difficult to lend to but could potentially have a large impact."
The researchers then recruited 209 experienced loan officers from several leading private—and public-sector Indian commercial banks to participate in up to 15 loan assessment sessions. Participants were asked to evaluate six applications per session, rating applications on personal, business, management, and financial risk before granting or denying the loan.
Because these applications had already been processed in real life, the researchers knew whether each one had been accepted for a loan, and which of the accepted loans had eventually defaulted. But for the purposes of the experiment, participants had no information other than the data in the original applications. In this way, the researchers could determine whether they made good decisions without the benefit of hindsight.
In addition to the quality of the decision, the researchers measured the participants' screening effort, their personal assessment of an application's riskiness, and whether the participants decided to issue the loan.
At the start of each session, before evaluating the applications, participants were assigned to one of the aforementioned incentive schemes, such that some incentives came with bigger rewards or penalties than others.
The researchers found that incentive schemes based solely on origination resulted in a 16 percent increase in lending compared with a high-powered incentive scheme, and reduced profitability by 5 percent. Meanwhile, the promise of high-powered incentives increased the detection of bad applications by 11 percent, thus boosting the profits per originated loans by 3 percent.
"We found that when the stakes were higher, [participants] exerted more effort and made better decisions than when the stakes were lower or when they were simply incentivized to originate loans," Cole explains. "It wasn't just that they were more conservative, but that they were doing a better job of ferreting out bad loans."
Before deciding whether to approve an application for a loan, participants were told to rate the application on a scale of 0 to 100. Loan approvals were not contingent on a minimum score. "This approach is common practice for consumer and small-enterprise loans for which the bank records internal ratings but does not use predictive credit scoring in the approval process," the authors explain in their paper, Incentivizing Calculated Risk-Taking: Evidence from an Experiment with Commercial Bank Loan Officers.
Cole says he was especially surprised to find that the nature of the incentive actually changed participants' perceptions of whether an application was a risky proposition or a safe bet. When evaluating identical applications, participants receiving an origination bonus issued higher ratings than those receiving incentives that were tied to a loan's success. It wasn't simply that receiving an origination bonus increased the likelihood that an officer would approve an application for a loan; it was that the bonus made him or her truly believe that the application was more worthy.
"When you look at the portfolio of loans made by loan officers who were paid an origination bonus, it actually looks less risky than a portfolio of the exact same set of loans made by officers who were penalized when the loans defaulted, because their ratings were distorted," Cole explains.
To Cole, that finding is significant because it suggests that irresponsible lending practices are based not just on greed, but also on inadvertent delusion.
"When the stakes were higher, [participants] exerted more effort and made better decisions than when the stakes were lower or when they were simply incentivized to originate loans."
"It's informative for thinking about the financial crisis and how we got there," he says. "A cynical view is that mortgage originators in the United States made a lot of loans that they knew for sure were going to fail, but they issued them anyway because they just wanted to collect their bonuses. And certainly that happened sometimes. But based on our research, it also could be that people's true judgment was distorted because of their incentives."
The research has implications not only for banks, but also for any firm that offers incentives to its employees.
"When thinking about compensation policies, it's important to realize the various effects they could have," Cole says. "It's possible that setting up an incentive scheme somehow conveys information to your employees not just about what types of things you want them to do, but also about what you want them to believe."
Deferred compensation dilemma
While those findings suggest that high-powered incentives are more effective than origination bonuses, there are drawbacks inherent in the delayed nature of performance-based bonuses, the researchers discovered. The paper notes that any incentive tied to a loan's performance must be paid at least several months after the loan is issued, simply because it takes a while to determine whether the borrower is going to pay the loan back.
For the sake of the experiment, the researchers promised immediate compensation to some participants and deferred compensation to others, regardless of the incentive scheme. Unfortunately, the researchers found that a deferral in compensation led to lazier application screening by the participating loan officers. In other words, delayed compensation attenuated the positive effects of the high-powered incentives.
"While our inclination might be to make incentive payments conditional on performance, that may come at a cost because deferred compensation is less able to motivate loan officers than immediate compensation," Cole says. "A lot of literature shows that people in a wide variety of settings overvalue today's money versus money in 90 days. People are impatient and, many would argue, irrationally so."
In a follow-up line of research, Cole is studying the link between a firm's compensation programs and the personalities of its employees.
"We're conducting a project to see how the incentive schemes you choose affect the type of workers you get," he explains. "If you have an incentive scheme that rewards performance a lot you may get individuals who like risk, whereas if you have a more conservative incentive scheme you may get individuals who are more conservative. And depending on the goals of your organization, you may want to attract different types of people."