Do people behave the same in a laboratory experiment as they do in real life? No one knows, and that presents a problem for researchers.
“If I give you $10 to play in a lab experiment about your attitude towards financial risk, it’s not clear that your decision in the lab will reflect the attitude that you have when you make a bigger investment with your own money in real life,” says Professor Enrichetta Ravina, whose behavioral economics research often focuses on consumer and household finance. “Lab experiments designed to measure the risk aversion of individuals have produced results that suggest very low risk aversion. Is the same true for real life financial decisions?”
Theoretical models present another option. “We can write a theoretical model reflecting the number of people who invest in stocks and bonds and how they react when prices change and extract a number that represents their risk aversion, even if I can’t ask those people directly.” But this comes with its own limitations: it is difficult to observe all the information required by theoretical models to generate accurate predictions.
Ravina, and Professors Daniel Paravisini and Veronica Rappoport, turned to the Web and an online company to solve the problem of how best to measure the risk-return tradeoffs that individual investors make and, in particular, to get a sense of the relationship between risk attitude and wealth.
Lending Club is a peer-to-peer, membership-based microlending site whose members can apply for or grant loans to other members at rates typically lower than what a bank would offer. Lenders are presented with a Web page containing a sliding control that allows them to choose their risk-return tradeoff. The slider starts at the far left, representing a pool of the lowest risk and lowest return rate loans an investor can make (to borrowers with the highest credit scores). Investors willing to take on additional risk in order to earn a higher rate of return move the slider to the right to indicate the rate of risk they find acceptable.
Investors can make some changes to the suggested portfolio, for example, changing the amount of money they put into each of the pooled loans. Once these two decisions — risk-return and portfolio allocation — are reached, the loan is made.
The detailed data provided by Lending Club (which included demographic information like zip codes but did not include private or identifying information such as names or exact addresses) allowed the researchers access to all the information lenders had when making their decisions about risk-return rates and portfolio allocation. Zip codes and other demographic data allowed the researchers to estimate the home prices and median incomes of the lenders.
Combining the Lending Club data with demographic data from matching zip codes, the researchers estimated each individual’s risk aversion. “A very risk-averse person might score a 10, for example; a risk-loving person might score a one,” Ravina says. The researchers then looked at how these risk-aversion scores correlated with gender, income, debt, credit scores and other demographic characteristics.
On average, participants’ risk aversion was 2.85, with a median of 1.62, which suggests that the behavior subjects display in lab experiments carries over in the real world.
The researchers found that people are less willing to take risks when their wealth declines. They approximated investors’ wealth by looking at housing prices in the investor’s zip code between October 2007 and April 2008 — the period of large fluctuations in housing prices, including steep declines. That allowed the researchers to see how the Lending Club pool reacted to large changes in prices — how a big shock changes risk attitude.
“We found that the average investor’s risk aversion almost doubled, from about 2.85 to 5.2, when she experienced a 20 percent decline in home prices — the median decline in the sample — in her zip code,” Ravina says.
In general the researchers’ findings about risk aversion were consistent with expectations: they found a lot of variety, but overall, younger investors (those under 42) were 17 percent less risk-averse than older investors; married investors were more risk averse than unmarried investors.
Female investors at Lending Club, though, have on average 15 percent lower risk aversion than male investors — a finding inverse to what other studies suggest, which is that women are typically more risk-averse than men. The researchers believe this is probably because women who participate in Lending Club, who at 17 percent of participants comprise a minority of investors on the site, are more likely to be the financial decision makers in their household than women on average (in most two-person households men are the primary financial decision makers).
Overall, says Ravina, using their model to analyze behavior yields a more accurate sense of the choices investors make and their willingness to take on risk. “We look,” she says, “at real investments in a real-life setting.”
Daniel Paravisini is associate professor of finance and economics at Columbia Business School.
Veronica Rappoport is assistant professor of finance and economics at Columbia Business School.
Enrichetta Ravina is assistant professor of finance and economics at Columbia Business School.