More than 30 percent of the Federal Housing Administration loans originated from 2007 through 2009 will be delinquent within five years and the agency's data underestimate that risk, according to a study by the Federal Reserve Bank of New York and New York University.

The research published this month used loan information from data provider CoreLogic Inc. to track FHA-insured mortgages and predict default rates based on borrower characteristics.

The FHA, part of the Department of Housing and Urban Development, miscalculates risk because it counts refinanced mortgages as successful loan terminations, even though the same borrowers are refinancing into new mortgages backed by the government insurer, according to the paper. The researchers computed the risk of default by linking all of the loans connected to each borrower.

"Having such a very large fraction of the people who borrow from you become delinquent could never be regarded as good public policy," said Andrew Caplin, a professor of economics at New York University and one of the study's authors.

Within five years, fewer than 15 percent of FHA borrowers with loans originated between 2007 and 2009 will have paid off their mortgages either by selling their homes or through other means, the paper said.

Latest Critique
The study is the latest in a series of critiques by Caplin and others of the way the FHA tracks its financial health. The agency, which took on more loans as private insurers left the market in the aftermath of the 2008 financial crisis, guarantees about $1.1 trillion in home loans.

HUD spokeswoman Tiffany Thomas Smith declined to comment in detail because the agency has not yet reviewed the study. The FHA has defended its financial performance data and stressed that the quality of its loans is improving.

"Credit quality of loans the agency has insured over the past two years is the highest it has ever been," the agency said in a fact sheet posted on its website in March. "Early- payment default rates and current-period serious delinquency rates are a fraction of those seen in earlier books at the same point in their seasoning."

Caplin said he is urging the agency to change its model for computing default risk.

"In the current method, we're not building a future," he said. "If you can't even look at the data right, how can you possibly design the housing-finance institutions of the future?"

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.