Agencies turn to technology for help reducing payment errors

In the past few years, 14,000 felons, both fugitive and jailed, raked in $230 million in federal benefits they were not entitled to receive. Twenty thousand dead Americans earned $180 million. In 2009 alone, the federal government made $110 billion in improper payments--that's nearly double the amount taxpayers will end up shelling out for the massive financial bailout, according to the latest estimates from the Congressional Budget Office.

The Office of Management and Budget breaks down the numbers like this: One-third of improper payments can be explained by poor documentation that makes it impossible to verify whether they were accurate, and another third result from failure to confirm individuals are eligible to receive the payments in the first place. The rest boil down to simple program errors, or people duping the system.

"Just as important as the choice of what strategies to pursue is how we pursue them," Jeffrey Zients, federal chief performance officer and now OMB interim director, said in a memo that outlined a series of goals for improving IT performance. "Where efforts are off-track and a team is not making the necessary midcourse corrections, we will work with them to get efforts back on track. Where progress is being made and breakthroughs achieved, we will celebrate success and work to spread best practices for achieving results across government. Where progress toward a goal shared by multiple agencies requires interagency coordination, or where agencies face similar problems that could benefit from cross-agency attention, we will facilitate those efforts."

Among the agencies that likely will fall under the microscope is the Internal Revenue Service. More than a quarter of the $48.1 billion in payments made under the Earned Income Tax Credit for low- to moderate-income workers were incorrect, according to the research firm INPUT. Refunds went to recipients for taxes they didn't owe, even though data on workers' income and tax contributions are readily available on the W-2 forms that employers file.

"Agencies should be able to access more seamlessly information [that would help] prevent these clearly ineligible beneficiaries from receiving payments," says Danny Werfel, controller of OMB's Office of Federal Financial Management. "That requires administrative steps that allow existing data sets to be accessed and linked, and additional data sources created when necessary. We do it, but we need to do it more."

In July, President Obama signed the 2010 Improper Payments Elimination and Recovery Act, saying it would reduce waste and fraud by $50 billion by 2012. The law requires agencies to conduct recovery audits for programs that spend $1 million or more annually, review programs susceptible to significant payment errors every three years, and plan corrective actions for preventing future waste.

The White House also launched PaymentAccuracy.gov, a public website to track progress in reducing improper payments, and established the Do Not Pay List for agencies to verify individuals' or contractors' eligibility before making payments. Agencies were instructed to establish prepayment controls that systematically confirm eligibility of payment beneficiaries with existing databases, including the Social Security Administration's Death Master File and the General Services Administration's Excluded Parties List System that tracks companies barred from receiving federal contracts. At the same time, OMB is leading an initiative to integrate relevant databases into a central portal agencies can check prior to making payments.

"There's an element of cultural change that has to take place," says Frank Blaul, senior vice president of government services at Equifax. The consumer credit reporting agency is among the companies providing input for a white paper OMB is producing in partnership with the industry group TechAmerica. The document will lay out a phased approach to eliminating fraud, waste and abuse in government programs.

"Some agencies have better defined processes than others, but there is no silver bullet," Blaul says. "There needs to be an evaluation of current systems so improvements can be made, implementation of technology to help mitigate errors, and establishment of policy to ensure these programs are monitored."

Technology, Werfel agrees, plays a critical role in enabling access and integration of data, "so agencies can make smart, informed judgments before payment are made." But no single technology solution will address all risk of fraud or waste in federal programs. Agencies will have to customize data mining and analytics software to extract patterns from information and to flag anomalies, he says.

The Health and Human Services Department, whose Medicare program accounted for $24.1 billion in improper payments in 2009, according to a June report from the Government Accountability Office, plans to install data analysis tools to root out fraudulent payments in that program, as well as Medicaid and children's health insurance programs. Under a rule introduced in September, the Centers for Medicare and Medicaid Services will deploy computer applications that alert administrators to suspicious activities in the systems that process requests, including unusual patterns in billing and applying for services.

"The proposed regulations [provide] important new tools to help us move from a 'pay-and-chase' approach," which identifies unscrupulous acts after the government has issued a check, "to one that makes it harder to commit fraud in the first place," CMS Administrator Donald Berwick said during a conference call with reporters in September.

The agency will buy more sophisticated analytical tools to screen applications to enroll in the program and to identify patterns in phone calls to the toll-free Medicare hot line from beneficiaries who flag possible problems in the program. CMS also is seeking public comment on the rule's proposal to collect fingerprints from health care providers and suppliers, which would be checked against law enforcement databases.

Technology solutions depend largely on individual programs, all of which face different challenges in closing loopholes or addressing flaws in their systems. For the Agriculture Department that means finding a way to prevent school employees from ringing up the cost of meals incorrectly, which led to $1.6 billion in improper payments through the School Lunch Program in 2009, INPUT reported. That objective is distinct from HHS identifying individuals attempting to defraud the health insurance system, or the IRS flagging errors taxpayers make in filing their own returns.

"For the most part, these capabilities are developed as situations arise and are tacked onto the systems that are currently in place," says John Reece, an independent IT consultant and former chief information officer at the IRS. "When we did modernization, we did a thorough re-examination of fraud and nonpayment and late payment of taxes. Those modernized systems reflected our best thinking then, but they've been updated since and will be again. The key is to put in place end-to-end systematic controls that can identify [incidents] of fraud within the program that need to be addressed."

The simplest solutions often have the biggest impact. Electronic benefits transfer cards, which replaced paper checks for distributing Food Stamps to low-income people, have reduced improper payments at the Agriculture Department from 38 cents per dollar to 1 cent per dollar since 1999, according to INPUT.

GSA had a markedly different result when it deployed charge cards governmentwide for small and routine purchases. The SmartPay cards were intended to reduce fraud and waste, but they actually did the opposite. The program has been the subject of numerous congressional hearings and Government Accountability Office reports that revealed employees used the cards to purchase personal items.

"The amount of fraud that slipped through as people tried to control cash and reconcile paper forms was one of the many reasons that drove us to use credit cards in the first place," says Alan Balutis, director and distinguished fellow with Cisco Business Solutions Group.

Balutis was chief information officer at the Commerce Department when it first began testing the use of charge cards in the mid-1990s. Software linked to the cards can identify potential fraud and waste--flagging charges for certain vendors or products deemed off limits, for example, or notifying administrators when employees exceed caps on expenditures.

"The problem then and now is the lack of any desire by managers to actually examine the data and take action when the system begins to note certain irregularities," he says. "The failure has never been with the card or the card technology, which has only gotten more powerful and exact over the years; the failure is that no one is paying attention. Do we need some device like the ones in Rube Goldberg cartoons, where smoke bellows out and lights start flashing whenever something abnormal is detected?"

Though automated processes can go a long way in identifying anomalies in financial systems, the human element is crucial to targeting the right data and taking steps to address problems once they're identified.

"We've got terabytes of data coming in, but they're only as power- ful as the intelligence behind the questions that we ask them to answer," Werfel says. "[Agencies] need to take a multidisciplinary approach to formulating these queries that includes the right combination of people who understand how fraudsters might take advantage of the complexities of the program.

That's all about analytics and human intelligence, as much as technology. It's where I see the greatest promise for development."

Technology change breeds operational change, which breeds cultural change, Balutis adds, noting agencies have to consider all three if they want to see any true return on investment.

"Sometimes we turn to technology as if it is going to be the panacea, without taking a look at some of the other aspects of implementation," Balutis says. "How do you deal now with the vast array of information you're going to get, and the speed with which it's going to occur? To some extent, it takes considerable rethinking of processes and operations."