"There are two big takeaways. First, that numerical ability heavily correlates to mortgage default, even controlling for a lot of other things," said Stephan Meier, one of the report's authors. He is a behavioral economics expert and professor at Columbia Business School. "Second, defaults were not driven by the mortgage choices people make. Their mistakes come somewhere else."
The results, "Numerical Ability Predicts Mortgage Default," were published Monday in the Proceedings of the National Academy of Sciences.
More From NBC:
Courtcalls for more scrutiny in affirmative action
Kerryto Russia: 'Do the right thing' on Snowden
Expert:Snowden may be carrying more info to leak
Meier and co-authors Kris Gerari and Lorenz Goette examined actual mortgage payment streams obtained from the Federal Reserve, and then contacted mortgage holders to assess their math skills. Each was asked a series of five questions (see them below) and then assigned into four "buckets." Four of those questions test the ability to perform simple division or calculate percentages, like this: "A shop is selling all items at half price. Before the sale, a sofa costs $300.How much will it cost in the sale?"
Roughly 1 in 7 test-takers couldn't answer even two such questions correctly, and landed in the lowest bucket. That group was four times more likely to be in foreclosure than consumers who landed in the top bucket by answering all five questions correctly.
The fifth question tested ability to compute compound interest. Only 1 in 8 test-takers scored in this highest group, and of those, only 5 percent were in foreclosure.
The researchers went to great pains to control for numerous outside factors, such as income, overall education level, IQ, and type of mortgage, Meier said. Math skills had the most pronounced effect on likelihood that consumers would run into trouble. For example, homeowners with no college degree but high math skills performed better than college graduates with poor math skills, he said.
(Read More: Mortgage Apps Tumble as Rates Soar)
"This is very specific to numerical ability," Meier said.
The study does not offer additional insight into why math skills can predict mortgage trouble. Other studies have shown that those with poor math skills save less for the future, so that might suggest the group is "more vulnerable to income shocks," Meier said. Or they might make less informed choices about credit cards or insurance.
But math skills turned out to be a much better predictor of default than type of mortgage, he said.
"Even if you gave this group a plain vanilla, 30-year fixed mortgage, this group would still have difficulties," Meier said. "I'm not saying (exotic) mortgages are good...the results suggest the problems are outside the mortgage."
Americans' trouble with math is well-documented. U.S. students' math skills frequently grade among the poorest in the developed world (in one recent study, the U.S. ranked 31st).
Meanwhile, the most recent U.S. Department of Education's National Assessment of Adult Literacy showed that consumers are terrible at solving real-world math problems, such as calculating tips or comparing prices in grocery stores. For example, only 42 percent of U.S. adults could pick out two items on a restaurant menu, add them and calculate a tip. In a result that neatly parallels Meier's study, only 13 percent were deemed "proficient," and only 1 in 5 could calculate mortgage interest.
(Read More: Rising Home Prices Are 'Unsustainable')
But the Department of Education results are more than 10 years old, showing little progress has been made in shoring up Americans' basic math skills
The phenomenon was identified even longer ago, in 1988, by mathematician John Allen Paulos in a book called "Innumercy" — the author's invented term for mathematical illiteracy. In the book, Paulos argues that being a math dummy in America is not frowned upon, like illiteracy — in fact, it can be socially desirable. People often joke about their inability to balance a checkbook, he said, to the delight of friends.