The Federal Housing Administration’s outdated computer systems were partially to blame for nearly 10,000 improperly insured mortgages, a blunder that could end up costing $41 million in federal money.
An internal investigation in March revealed that in 2016, the FHA signed off on 9,507 mortgages — with a total balance of $1.9 billion — for borrowers who didn’t actually qualify for government-backed loans, either because they had unpaid federal debts or delinquent child-support payments.
Now a new analysis from the Urban Institute, a non-partisan Washington think tank, pegs the cost of those erroneous loans at $41 million based on overall FHA claims data and the significantly higher delinquency rates for the flagged mortgages.
“This is just one example of the economic consequences of relying on an outdated system,” Urban Institute researchers Karan Kaul and Edward Golding wrote.
Ancient information technology infrastructure at the FHA and the Department of Housing and Urban Development has long been identified as a significant problem for borrowers and lenders alike. The government tracks mortgages using software based on COBOL, a programming language that dates back to the late 1950s, and officials have found it increasingly difficult to hire programmers who know how the system works.
In the case of the inappropriately approved mortgages, a database first developed in 1987 was to blame. When officials from HUD’s Office of the Inspector General (OIG) cross-checked the Credit Alert Interactive Voice Response System (CAIVRS) against the Treasury Department’s Do Not Pay system, they found a significant portion of delinquent borrowers that HUD’s system did not identify.
The Urban Institute researchers note that HUD and FHA have agreed to make the necessary changes, but that officials are waiting on a $30 million cash influx from Congress.
“With adequate resources and systems, the FHA might have avoided insuring the ineligible loans and the $41 million in estimated losses for taxpayers,” Kaul and Golding concluded. “The $30 million would have more than paid for itself.”
Written by Alex Spanko