The Obama administration is deploying a new weapon in its battle against improper payments. Analytics software is helping federal agencies find errors and flag potential waste, fraud and abuse before payments are made.
The opportunity for savings is huge. The Office of Management and Budget estimates agencies lost $98 billion in improper payments in fiscal 2009. "That is such an unacceptable, high number of improper payments at a time when we have an extremely tight budgetary environment," says OMB Controller Danny Werfel. "We have called on agencies to not only tighten their belts but to use more innovative ways . . . to improve the efficiency of government."
To get ahead of the curve, some agencies are turning to predictive analytics software to sift through reams of data and identify payments that warrant additional scrutiny. The software is being applied to benefits programs, grants and contracting, among other areas.
Senior federal managers are under pressure to reduce erroneous payments. In November 2009, the president issued an executive order that directed OMB to identify programs with the most improper payments and to establish targets for curbing losses. Agencies were encouraged to employ cutting-edge software to address the problem.
"The use of data mining and other business intelligence tools to look across programs and find errors has been powerful . . . and we want to make sure that agencies are investing in those tools and looking for opportunities to leverage them," Werfel says.
One leader in using predictive analytics software is the Centers for Medicare and Medicaid Services. CMS uses the software to find questionable payments before they are made.
"CMS uses a variety of prepayment analytic software to identify payments with characteristics that are outside the norm, such as duplicate payments, overpayments in general and a large number of payments to one provider. This helps CMS establish a profile of the type of payments that should trigger a risk or a more intensified review," Werfel says.
Since 2008, the Defense Department has been using business activity monitoring software to inspect financial transactions and flag high-risk payments for extra review before they go out. Department officials say the software identified more than $300 million in potential improper payments in the first year. The Recovery Board is using business analytics software to review information on recipients of 2009 American Recovery and Reinvestment Act funds in search of risky payments that have already been made. The software searches data the Recovery Board has collected as well as other publicly available sources.
Agencies are seeing a significant return on their investments in analytics software. "It makes agencies smarter about where to spend those oversight resources, where to scrutinize payments and then actually in the cases where it's finding an error, in preventing [the payment] from going out the door or finding the error afterwards so we can collect on that error," Werfel says. "Both are extremely helpful."
Software from such vendors as SAS Institute Inc., SPSS Inc. and Information Builders allows agencies to use statistical modeling, data mining and forecasting to predict trends rather than analyzing them after the fact.
"What agencies are trying to do is create more sophisticated models to identify where potential fraud might be and predict where it might happen before it happens," says Dan Vesset, program vice president for business analytics at market research firm IDC. "It is a hot area."
One application involves auditing tax returns to find the ones that are likely to recover the most revenue. "Both on the federal and state levels, we see use of these tools for determining good audit targets," says Bill Haffey, predictive analytics strategist for the public sector at SPSS. "They're looking at audited returns to find which returns might be more lucrative in determining auditing resources."
Agencies including the Social Security Administration, the Internal Revenue Service and the Agriculture Department are turning to analytics software to detect fraud. Los Angeles County, for example, used SAS software to find a day care scam. The software can detect a recipient of state or federal child care funds who claims to support 10 or 15 children but actually lives in a one-bedroom apartment. It also can uncover linkages from the recipient to others involved in the fraud, including people claiming to have children on the roster at the phony day care center.
"One of the most sophisticated technologies that SAS has is social network analysis, where we are uncovering those linkages that you can't see with the naked eye," says Karen Knowles, general manager of SAS Federal. "We're identifying linkages between participants, payers and providers in everything from health care to day care to food stamps."
As agencies push to make operations more transparent and provide more information to the public, they should make better use of that data through predictive analytics, says Molly O'Neill, the former chief information officer at the Environmental Protection Agency and now a vice president at the consulting firm CGI Federal.
"Now that the data is going to be available from all the federal agencies, we can actually start grabbing it and using it for real decision-making, targeting and projections," she says. "We can move from displaying it on dashboards . . . to using it for things people care about."
O'Neill says agencies can gain more from predictive analytics than reducing improper payments. The software also can help agencies model scenarios and forecast future needs for funding and resources. "What we really need on the planning side is predictability," she says. "Then we can scale up and scale down based on indicators."
Carolyn Duffy Marsan is a high-tech business reporter based in Indianapolis who has covered the federal IT market since 1987.