Data-mining saves federal crop insurance program $456 million, director says.

The agency overseeing the federal crop insurance program has saved taxpayers $456 million since 2001 by using data-mining to weed out fraudulent claims before they are paid, the agency's director recently told a U.S. House of Representatives subcommittee.

Eldon Gould, administrator of the U.S. Department of Agriculture's Risk Management Agency, detailed the now 5-year-old data-mining program before a hearing of the House Agriculture Committee's Farm Commodities and Risk Management Subcommittee.

Essentially, the program culls claims data and looks for patterns among producers, agents and claims adjusters, seeking patterns such as "producers whose crop insurance losses are not consistent with those of their neighbors," Gould said. Another red flag is when a combination of producers, agents and loss adjusters appear in a number of questionable claims.

When questioned by ranking subcommittee member Rep. Bob Etheridge, D-N.C., as to how the RMA calculated the $456 million figure, Gould said the agency was able to use the data-mining "to find out producers that were found to be filing fraudulent claims ... Those were claims that were never paid"

Those savings represent a return of $23 for every dollar spent on the data-mining program, Gould told the panel. Heading off bad claims at the front of the pipeline also means that RMA doesn't have to devote resources to administrative sanctions or costly lengthy criminal investigations, Gould said.

Also, once people are caught, they tend not to file fraudulent claims in the future, Gould said: "Our analysis shows that the change in claims behavior ... persists for several years, resulting in overall savings that are even higher over a longer-term period."

RMA, through the Federal Crop Insurance Corp., oversees and administers the country's crop insurance program under the Federal Crop Insurance Act. Crop insurance is offered through 16 private crop insurance companies, with RMA providing reinsurance and premium subsidies and reimbursing insurers for their overhead costs. RMA also monitors the crop insurance system for fraud, waste and other abuses.

Congress gave RMA five years' worth of funding for the data-mining program beginning in 2000, funding that expired in 2005 and since has been extended via an annual appropriation. RMA asked for $3.6 million in its fiscal 2007 budget to continue the program. The agricultural appropriations bill that passed the House has that amount earmarked for the RMA's data-mining efforts; the funding bill is now in the hands of the Senate Appropriations Committee.

The panel also heard about another tool used to stem losses: Multispectral imagery, which photographs crops using other-than-visible parts of the light spectrum, which can show crop losses much more accurately than a loss adjuster making estimations on the ground.

In one test case, 30 adjusters were brought to a farm in Iowa and estimated, on average, that losses in one field were 70%. A multispectral analysis, however, showed those losses actually were closer to 35%.