Can a Test Tell If You're a Good Entrepreneur?
How psychometrics are helping loan officers weed out the bad risks from the good.
Bankers around the world know there are profits to be reaped by making loans to promising small businesses that fall just short of traditional definitions of “creditworthy.” Ever since Nobel Prize winner Muhammad Yunus’ Grameen Bank pioneered "microfinance" by making tiny loans to single mothers in Bangladesh, development policymakers also have believed that getting credit to small businesses—those too large for Grameen-style microloans but still lacking collateral or credit history—is not only possible, but the key to helping a nation’s economic growth. So how to figure out who’s a good risk for a loan—and who isn’t?
Now, some banks think a simple, 40-minute questionnaire for entrepreneurs can tell them a lot about whether a company is worth the risk. Two public policy researchers at Harvard, Bailey Klinger and Asim Khwaja, first adapted the questionnaire from psychometric evaluations— a tool psychologists, child custody lawyers, executive recruiters, retail employers, and even the NFL have used since the concept emerged from academia in 1918, when the U.S. Army started mentally testing potential doughboys.
Psychometric evaluations typically include brief, open-ended questions related to ethics, attitudes, skills, and intelligence. In Klinger and Khwaja’s model for testing how likely an entrepreneur is to make good on a loan, that means questions like “Do you like to attend parties?,” “Do you enjoy taking things apart to see how they work?” or “Do you believe that luck is a big part of success?” as well as cognitive problems like “Look at this sequence of numbers for five seconds, then try to recite as many digits as you can.”
The researchers launched pilot programs in South Africa, Kenya, Rwanda, Colombia, and Peru in 2008, and spent two years administering evaluations and following the success rate of test-takers. They discovered an unusually high correlation in the answers submitted by successful entrepreneurs who were also good borrowers.
“[Psychometrics] aren’t perfect; everyone who does very well on [the] SAT [test] doesn’t end up with a 4.0 in college,” says Brian Hoffman, professor of psychology at the University of Georgia. “But it’s a cost-effective predictor.” That is, psychometrics appear to be a cheaper, more accurate way to predict human behavior in many fields that have been using them to evaluate people.
The Harvard team’s innovative application of psychometric tools earned a stream of grants and plaudits, and eventually a prize for innovation in development finance sponsored by the G-20. Boosting growth among small and medium-sized enterprises, from second-hand-clothing sellers to light manufacturers, has long been seen by development experts as crucial for job creation and ultimately poverty alleviation in low-income countries. In countries like Yunus’s home of Bangladesh, the economy is dominated by a few large, established corporations or industries, and innumerable, and informal, micro-enterprises. Meanwhile, mid-sized operations average only 15 percent of low-income countries’ GDP, compared to roughly 50 percent of high-income nations’ GDP. Research finds that neither the really big nor the really small businesses have the growth potential—or unmet credit needs—of those mid-sized operations with 10 to 250 employees. But lenders tend to see this segment, generally lacking collateral or extensive financial records in the developing world, as too risky a bet. They aren’t the charity cases that foundations and policymakers itch to assist, either.
“These business owners might be poor, but not poorest of poor. Not the kinds you can put on a glossy brochure to pull people’s heart strings about helping them make money,” says Klinger.
To expand the use of their questionnaire, Kwhaja and Klinger created a company called the Entrepreneurial Finance Lab (EFL) in 2010, and signed up one of the largest banks in Africa, Standard Bank of South Africa. The next step has been to send loan officers into bazaars, street markets, and main streets with laptops or cell phones loaded with the questionnaire.
The bank found that when an entrepreneur qualified for a loan using both the bank’s traditional credit scoring methods and EFL’s questionnaire, delinquency rates were 2.8 percent. Applicants who qualified for a loan based only on the banks’ traditional credit scoring methods had a delinquency rate of 14.5 percent. And in 677 of 913 applications, where the bank said no but psychometrics said yes (and therefore a loan was granted), the delinquency rate was 9.7 percent. Psychometric evaluations were helping Standard access, albeit on a small scale, a huge new pool of fairly successful borrowers that it otherwise would have turned down. Those 677 businesses now had access to growth capital, enabling them to hire new employees.
“The whole point of the development finance business in the last 15 or 20 years was to get around [the credit] problem,” according to Nicholas Hope, director of Stanford’s Center for International Development and a veteran of World Bank programs in China and Indonesia. He was referring to a problem known as the "missing middle,” in which small and medium-sized enterprises were overlooked by lending institutions and charitable organizations.
Roughly 45,000 applicants have answered some version of EFL’s questions as part of their loan application process. Banks like Standard in South Africa but also others such as Mexico’s BBVA Bancomer and EFL’s newest client in Indonesia, Bank BPTA (pending formal central bank approval, though they’re already using the tool more widely than any other bank), have made $156 million in loans—generally from $800 to $100,000—to entrepreneurs based at least partly on their psychometric responses.
The loans are going out at a pace of roughly $8 million a month, according to Klinger. In early November, the Inter Development Bank in Washington, a sort of World Bank for Latin America, announced $25 million in loan guarantees for Latin American banks that incorporate psychometrics in lending decisions for micro- to medium-sized enterprises. Some banks are even considering using psychometrics to re-evaluate terms on loans they have already made, Klinger said.
This rapid rate of adoption globally could do a lot to spur growth in a huge, long-neglected segment of small businesses. But, with large corporations using sophisticated new tools to engage borrowers with little institutional support or financial savvy, there are potential pitfalls. Critics of psychometrics in hiring have claimed that the tests can reinforce gender and racial stereotypes. Seth Kaplan, professor of psychology at the University of Georgia, described the dynamic this way: “A test is a sample of behavior, and anytime you’re trying to make inferences about future behavior, it’s always kind of a deterministic model.”
Klinger agreed that, in the lending context, it would be ethically dubious to judge creditworthiness based on latent personality traits but argued that this mischaracterizes the method being applied by EFL.
“We walk a very careful line,” he says. “We work with some of the top people in the field, but we are not psychologists. We’re not measuring whether somebody is smart or has a certain level of ambition, but asking some questions from that context, and relating those answers to statistical history of default rates. It may sound like a fine-print legal distinction, but it’s a substantive difference.”
Jared Miller, Latin America director for Klinger’s company, summed it this way: “We don’t measure who [entrepreneurs] are really, we measure who they present themselves to be.”
Critics of micro-finance also contend that the high interest rates that for-profit players like HSBC offered borrowers canceled out the potential to lower poverty rates. Francisco Rojo, project manager for the IDB’s Multilateral Investment Fund, which focuses on smaller-scale, market-based development projects for the IDB and has helped EFL enter the Latin American market, allows, “We are not setting out criteria in terms of interest rates; banks decide that.” In his view, though, that would likely prove unnecessary. “The idea is, if they charge very high interest rates, and they see it successful with profit, then they will begin to lend to that same segment and then they will compete and interest rates will get lower.” EFL executives echoed this point.
Given the rapid rate at which banks are adopting tools from psychometrics to expand the reach of their loan officers, and small businesses’ estimated $2 trillion in unmet credit needs in the developing world, we’ll find out soon whether Rojo’s prediction about interest rates comes to pass.