:: By Megan Lisagor

:: Photography by Dan Dry

Investigations ::

As April tax deadlines drew near, free preparation services were there for the taking. Nonprofits such as the Center for Economic Progress would file low-income earners’ returns and open bank accounts to deposit their refunds, but potential clients regularly passed up the offer. Marianne Bertrand, professor of economics in the Graduate School of Business, has a cheap solution for bringing more people on board: apply psychology. She believes that marketing can be used to sell the tax-savings program and others like it.

Bertrand, who has degrees from the Université Libre de Bruxelles and Harvard, looks beyond standard economic incentives. “Instead of pricing let’s talk about other factors,” she says from her fifth-floor office. “I’ve been thinking for a while about the decision-making of the poor, how to use psychological measures to alter their behavior.” Psychology studies suggest that certain cues can influence a person’s choice. She guessed that the same effect would be true among consumers in a financial setting.

To field-test the theory, Bertrand and a team of researchers, including Dean Karlan, MBA’97, MPP’97, did direct marketing with a difference. Measuring borrowing in a loan campaign, they analyzed the effects of both pricing and psychological cues. With the two sets of data, “we’re able to make a direct comparison,” says Karlan, an assistant professor of economics and international affairs at Princeton. “That’s the key thing this paper does.” First the team had to find a lender willing to participate. South Africa provided an ideal testing ground because its consumer-credit market includes a large for-profit industry extending small, high-interest cash loans to employed individuals—millions of working poor.

Partnered with one of the country’s biggest micro-lenders, the researchers then sent direct-mail solicitations to 57,703 clients, who had borrowed in the past, offering a new loan with monthly interest rates ranging from 3.25 percent to 11.75 percent. In addition to the pricing variables, the team introduced a number of marketing variables. They alternated the examples of repayment and the deadlines for acceptance from two to eight weeks, compared competitors’ rates, and ran photos of a man or a woman. They also offered a promotional giveaway, provided a follow-up phone call, and used five different phrases to state how the cash could be spent. “We’re trying to ask questions that aren’t just about font size,” Karlan explains. “That’s something the direct-mail solicitation guys don’t do.”

Both sets of variables, pricing and marketing, were randomized. For instance, some clients saw only one example of repayment for a given loan term and size while others received multiple possibilities. Psychology studies “suggest that people often are tempted to defer [a] decision when a compelling reason for choice is not readily available and the decision feels hard,” the team writes in the working paper “How Much Does Psychology Matter? A Field Experiment in the Consumer Credit Market,” in contrast to standard economics in which having more options shouldn’t affect borrowing. Meanwhile, they added a photo because research shows that people’s decisions may be altered by a subject who appears attractive or similar to themselves.

To measure the effects, the team recorded loan take-up. Using the response at each interest rate as a benchmark, they calculated which marketing manipulations garnered more acceptances. Men, for instance, were more likely to take the offer when the letter contained a female picture. Next the researchers estimated the “interest-rate equivalent” of a manipulation—how large a price change would have been needed to achieve the same result. “Consistent with standard economics, the interest rate significantly affected loan take-up,” they report. “Inconsistent with standard economics, though, the psychological cues also significantly affected take-up,” particularly when people were getting a lesser deal. In fact, any “one manipulation has the same effect as a one to two percentage-point change in the monthly interest rate”—regardless of a customer’s education or wealth.

Still, “not everything works,” Bertrand notes. Comparing the offer to competitors’ rates did nothing for borrowing. The other manipulations, she says, were “much more context-specific.” For instance, the effect of the promotional giveaway was not statistically significant overall, but among more frequent borrowers it made a big difference. “As a whole, these findings suggest that standard economic models may be missing very important drivers of choice,” the team concludes. “The development of...richer models may be necessary in order to reach an accurate description of economic behavior.”

Such behavioral models, they say, have applications outside the business world. “Many of the insights gained in this paper are also relevant for the design of more socially oriented programs,” the researchers write. “Through increased focus on the marketing of their programs, governmental agencies may achieve broader participation without having to rely solely on greater financial incentives.” Bertrand, who studies discrimination issues as well, is interested in how initiatives such as food stamps and retirement-savings plans “are presented to people.” She recently analyzed participant behavior in a Center for Economic Progress program that opens bank accounts for low-income earners and is brainstorming ideas for boosting the nonprofit’s annual tax-savings effort. “We can really apply this to a policy environment,” she argues, conceding that the public sector changes more slowly than the private domain.