September 2010

Moderator: Welcome to the Federal Reserve Bank of Atlanta's EconSouth Now podcast. Today, we're joined by Kris Gerardi, an economist in the research division here at the Atlanta Fed. He'll be speaking today about research he and two colleagues did on the connection between financial literacy and the subprime mortgage crisis.

Thank you for joining us today, Kris. It's always a pleasure to talk to you.

Kris Gerardi: Thanks, Tom. It's great to be here.

Moderator: Kris, what led you and your colleagues to suspect a relationship between an individual's level of financial literacy and his or her likelihood of becoming delinquent on a subprime mortgage?

Gerardi: Well, I think there were many commentators and analysts that suspected such a relationship, but there really was no direct empirical evidence. For example, Robert Shiller's book The Subprime Solution talks about such a possibility, and Ed Gramlich's book on the subprime market [Subprime Mortgages: America's Latest Boom and Bust] discusses how the subprime market was characterized by poorer, lower-educated, and minority borrowers who were unlikely to fully comprehend the complicated mortgage products that they were using.

In addition, claims of predatory lending were rampant throughout the crisis, and it seemed unlikely that such behavior could take place if all borrowers fully understood their mortgage contracts. Now, let me be clear, by no means do we find any direct evidence of predatory behavior in our study, nor was that the point of the study, but it did seem to us that a necessary condition for predatory lending would be an inability to comprehend mortgage terms.

Moderator: Right, interesting. Kris, in general, how would you describe the relationship between financial literacy and the subprime mortgage delinquency?

Gerardi: Well, we found that the ability to perform simple mathematical calculations, which we refer to as numerical ability, is negatively correlated with the incidence and extent of mortgage delinquency. That is, the lower a borrower's numerical ability and the poorer that borrower scored on the questions that we asked, the more likely it was that the borrower had missed mortgage payments and experienced foreclosure.

One of the interesting findings of the study is that the relationship between mortgage outcomes and financial literacy doesn't seem to be explained by the choice of mortgage characteristics. So, one of the big advantages of our data set was that it allowed us to see detailed mortgage characteristics. So we were able to control for the fact that borrowers with lower numerical ability were more likely to put less money down, for example, and to have a larger fraction of their income taken up by mortgage payments. In addition, we could control for borrowers' credit scores and whether they chose a mortgage with a fixed payment or a loan with an interest rate that would reset, or rise after a set period of time. But, controlling for these and other borrower and mortgage characteristics doesn't weaken the relationship that we found between financial literacy and mortgage outcomes.

Moderator: I see. In your research, Kris, you cite a specific aspect of financial literacy as playing a role in mortgage repayment behavior, and that is numerical ability. Can you describe how you determined numerical ability and its role in borrowers' repayment behavior?

Gerardi: Sure. So, to measure numerical ability, we asked the borrowers in our survey to answer five simple math problems. Four of the questions were relatively simple and only involved an understanding of basic mathematical concepts like addition and subtraction and multiplication and division. The fifth question was a little more difficult and involved the compounding of interest over multiple time periods. We then used the answers to these five questions to form an index of numerical ability.

In addition to these questions, we also asked two questions about inflation to measure economic literacy and a question that has been shown to be highly correlated with IQ measures, in order to elicit cognitive ability. Unlike the numerical ability index, these two measures weren't correlated with mortgage repayment behavior

Moderator: Interesting. My next question, I guess, is two questions. What methods did you use to gather data, and what types of questions did you ask people?

Gerardi: Well, we randomly chose a sample of mortgage borrowers from our administrative mortgage data and conducted a phone survey. The survey took about 20 minutes to get through, and the questions were designed to measure financial literacy, mortgage characteristics, and socio-demographic characteristics—like age, gender, race, education, and income. Our final sample included about 350 borrowers, all of whom had obtained subprime mortgages in either 2006 or 2007 in the states of Massachusetts, Connecticut, and Rhode Island.

Moderator: You mentioned in your paper that you conducted your survey in the summer of 2008. How did you control for the faltering labor markets during that time in determining delinquency in making mortgage payments?

Gerardi: Well, since we knew the property addresses of the borrowers in our data set, we were able to merge town-level unemployment rates calculated by the Bureau of Labor Statistics into our data set. And this allowed us to control for the crash in labor markets from the onset of the recession.

In addition, our data set allowed us to calculate our own set of house price indexes at the town level, and, thus, we were able to control for the effect of the house price crash on mortgage outcomes, which was especially important, as previous research has found that the fall in house prices that began across the U.S. in late 2006 was the most important contributor to the mortgage market crash and subsequent foreclosure crisis.

Moderator: Right. Kris, did you observe statistically significant differences between borrowers who were new to the mortgage market versus those who had taken out mortgages before?

Gerardi: Tom, that's a good question. One of the unique aspects of our data is that it allows us to track a borrower from the time that they purchase their home and to observe all of the mortgages that they take out on their property. Thus, we knew how many mortgages, if any, the borrower had taken out before the specific subprime mortgage that we focused on in the interview, and this allowed us to control for any prior experience in mortgage markets that a household may have previously obtained. This is important, as one might expect previous experience to be important insofar as the borrower may have learned about aspects of the mortgage contracts, like interest rate resets, that a new borrower may struggle with understanding. Surprisingly, though, prior experience in mortgage markets was not related to mortgage outcomes, at least in our data set.

Moderator: That's very interesting. Kris, my final question to you is, what implications do your findings have? Should we conclude that financial education should receive greater emphasis, or does that oversimplify a broader challenge in informing people about mortgage-borrowing decisions?

Gerardi: While we believe the results of our study should raise some concerns about the role of financial literacy and mathematical ability in the success of mortgage outcomes and the sustainability of homeownership, we're a little reluctant to make any broad policy proposals.

First, since we conducted a survey, we weren't able to randomize through a controlled experiment, and as a result, there's always the possibility that some other unobserved factor may be causally responsible for the variation in mortgage delinquency in our data and also correlated with our measure of financial literacy. Now, even if the causality does run from poor financial literacy to poor mortgage outcomes, I would hesitate to propose simple policies, such as a mandatory financial education course, for prospective homeowners, as it is likely that financial literacy would also affect other aspects of life. Perhaps a more effective policy would be to begin teaching economic and financial problem-solving to students in high school or even middle school. After all, the vast majority of children in this country will end up purchasing a home and obtaining a mortgage and investing in some type of financial asset such as stocks or bonds when they become adults.

In any case, further research on this topic is necessary, especially for other segments of the mortgage market and other areas of the country, and we hope our study may provide the motivation for such research to occur.

Moderator: Kris, thank you so much for your time. It's always a pleasure to discuss your research with you.

Gerardi: Thanks a lot, Tom.

Moderator: Again, we've been speaking today with Kris Gerardi, an economist at the Atlanta Fed. This concludes our EconSouth Now podcast on financial literacy and the subprime mortgage crisis. For more information, please see the third quarter 2010 edition of EconSouth. On our website,, you can read Kris's working paper about this topic. Thanks for listening, and please return for more podcasts. If you have comments, please send us e-mail at