You are here

Sniffing out Fraud

States tap technology to protect welfare benefits and rein in improper payments.

States tap technology to protect welfare benefits and rein in improper payments.

When Tom Rachel of the Anoka County Attorney’s Office investigates welfare fraud, most of the information he needs is at his fingertips. He simply logs into a Web application that pulls data from different agency databases, giving him a central place to dig up proof that people are scamming the state.

Last year, for example, his office received a tip that a couple with four children who were receiving state-subsidized medical insurance was anything but needy. Investigators used the Minnesota Department of Human Services’ Program Integrity Network (PIN) to review the couple’s public-assistance case files and compare their claims of economic hardship with other data, such as their employment, income and car-registration records.

The electronic trail led investigators to the truth: The couple owned a $579,000 house, and they and two of their businesses had 28 cars registered in their names, including five Mercedes, a BMW and a Hummer, recalls Rachel. In the old days, investigators would have been hard-pressed to compile such data so quickly.

“[The system] helps us gather information and build our case without having to call all over the place and convince everyone that we are who we are, and get them to e-mail or fax the information to us,” says Rachel, a 35-year veteran investigator for the county. “It saves us so much time. There’s no way we could do this job without it.”

The couple was charged with stealing about $35,000 worth of state medical benefits, but their case is just the tip of the iceberg. In all, law enforcement and the state’s Human Services employees have used the PIN system to identify more than $12.2 million in overpayment and cost-avoidance in 2006. That’s money currently being recovered or benefits that were stopped because investigators discovered that welfare recipients were not eligible.

From Here, There and Everywhere

Like Minnesota, many states are using information technology to help investigators uncover welfare fraud, but they’re also building in safeguards to prevent fraud from happening in the first place. Their mission is to protect benefits for low-income adults and families who need help the most, whether for child care, health care, food or general assistance.

The technical approaches vary widely from state to state. Minnesota uses a Web interface to a data warehouse that connects disparate sources of data on the back end. Louisiana and Oklahoma combine business-intelligence tools with mapping features. Other states, such as Indiana and Kentucky, build on legacy investments by enhancing them with new technology. Kentucky, for example, is investing in a new Web application, while Indiana is deploying a service-oriented architecture and new workflow management and document management systems.

Feds Aid Fight

The U.S. Department of Health and Human Services’ Administration for Children and Families has partnered with 44 states to combat welfare fraud. Together, they will analyze the Social Security numbers of everyone who receives public assistance to make sure people are not receiving benefits from more than one state.

Meanwhile, Arizona, California, New York and Texas use fingerprint-imaging systems as part of the application process to stop people from receiving benefits from multiple counties in their states, says Richard Nawrot, chair of the intergovernmental affairs committee for the United Council on Welfare Fraud.

Many states regularly use computer-matching programs to check state and federal databases (such as birth, death and prison records), as well as credit reports and employment verifiers, to make sure people are qualified for benefits, Nawrot says. State governments also build internal controls into their IT systems to ensure that state workers don’t commit fraud, such as requiring multiple levels of approval before applicants receive benefits, he adds.

“Most states do a great deal of front-end detection to prevent fraud because it’s harder to chase it down after the fact,” says Nawrot, who is also director of New York’s Office of Temporary and Disability Assistance Program Integrity.

The Minnesota Department of Human Services has developed one of the most comprehensive IT systems to fight and eradicate fraud. The key is its data warehouse.

The department built the data warehouse in 1996 to analyze health-related data and provide reports to department heads and state leaders. In 2001, IT decided to take advantage of all that data and created the first iteration of PIN, to allow county fraud investigators to conduct their research and allow child-care caseworkers to make sure welfare recipients were, in fact, qualified to receive assistance.

First, the department let PIN users tap into Human Services data, such as case files for Medicaid, child care and food stamp programs. Then they began connecting to other state agencies, for such things as new-hire information from the Department of Employment and Economic Development, car-registration data from the Department of Public Safety and boat registrations from the Department of Natural Resources. The department recently connected to credit bureaus so investigators can run credit checks, and to the Social Security Administration to validate Social Security numbers.

“PIN made it easier for investigators to look for information without having to spend hours trying to figure out how each different computer-system interface works,” says applications architect Nina Terhaar, who manages the data warehouse.

The data warehouse, which holds 50 GB of data, is essentially a set of databases. It connects to other state-agency databases using the Open Database Connectivity and Object Linking and Embedding standards and makes copies of their data in real-time nightly, weekly or monthly batches, Terhaar says. It’s a four-node Unix server system, so if one set of servers goes down, another takes over, reducing downtime.

On the front end, Terhaar used Microsoft’s Visual Studio software-development tools — specifically ASP.Net — to develop the Web application. She also relies on Microsoft SQL Server databases to house a case-management system that stores information on open investigations.

Johanna Berg, left, and Nina Terhaar of the Minnesota Department of Human Services supply the technology used to chase welfare scammers.

“This application underscores what happens when a business with a need teams up with technology,” says Johanna Berg, CIO of the Department of Human Services. “The result is an award-winning effort that helps decrease fraud and saves significant taxpayer dollars.”

Thwarting Food Stamp Fraud

Louisiana’s Department of Social Services is just beginning to allow its employees to examine its various databases to find scofflaws. Its most unique effort combines business-intelligence and data-warehousing tools with geographic information systems (GIS) technology to catch people who misuse their food stamps.

Some people sell their food-stamp benefits to dishonest retailers for cash, often at half the actual value, says Raymond Pease, director of the fraud and recovery section of the Office of Family Support in Louisiana’s DSS.

Big Bucks for Baby-Sitting

Last year Minnesota investigators discovered a couple who had under-reported their income in order to receive $106,310 in child-care assistance payments.

The state has done away with actual food stamps and now uses Electronic Benefit Transfer (EBT) cards, which are like debit cards. Because the transactions are electronic, the state knows when and where people purchase their food. When a food-stamp recipient bypasses the local grocery store and drives all the way to the other side of town to shop for food, says Pease, it raises a red flag for investigators, who then probe further for fraudulent activities.

Hurricane Katrina wiped out a major food-stamp fraud investigation the state was conducting in 2005. The state is only now restarting its investigation. “Our goal is to mine all the available data and significantly decrease the amount of time that it takes to reveal fraud,” Pease says.

Oklahoma, which implemented a similar system in 2005, is already reaping the benefits. The state’s business- intelligence and mapping tools generate monthly reports that highlight suspicious activities, such as recipients who use their EBT cards out of state for several months, or convenience stores that have a higher number of EBT transactions than the norm, says Michael Fairless, inspector general of investigations at the Oklahoma Department of Human Services’ Office of the Inspector General.

“Sometimes the transactions are legitimate — for instance, purchases from a specialty retailer that sells prepackaged meat or fish in bulk — but it’s given us clues to true food-stamp fraud and it’s helped us track down criminals,” Fairless says.

Because of budget constraints, some states can’t build completely new anti-fraud systems. Kentucky and Indiana, for example, have created new features or applications on top of existing mainframe applications.

Kentucky’s Cabinet for Health and Family Services is building a new Web-based application that will aggregate state and federal agency databases, giving investigators and caseworkers a single Web site to verify data, such as wages and driver’s licenses, to help determine whether people are qualified for benefits, says Kathy Frye, the cabinet’s deputy CIO.

Until funding is available to replace the eligibility system built in 1993, Kentucky will continue to find creative ways to use it as its main fraud-fighting tool, Frye says. The system includes built-in safeguards to weed out internal fraud, such as storing audit trails and checking to make sure benefits weren’t mailed to employees’ addresses or to the local-office address.

During the past two years, the IT department has made enhancements by programming new interfaces that allow the application to check several state and federal agency databases, including prison and death records, to ensure the state stops payment of benefits to the recently imprisoned or deceased. The application then automatically matches the records against its case files, says information systems manager Greg Whitt. If problems are spotted, caseworkers are immediately notified. In the six months since the state began reviewing death records, it has discovered 717 dead people who were receiving benefits.

“It’s important that the processes are automated,” says Deputy Commissioner Mark Cornett. “If we had to manually pore through reports, there would be no way to do it.”

Indiana overhauled its technology to bolster its fight against welfare fraud and improve customer service. New document management and workflow management systems digitize all welfare recipients’ records and help speed up fraud investigations, says Zach Main, director of the Division of Family Resources in Indiana’s Family and Social Service Administration.

When Main arrived in 2005, he discovered that the technology and policies used to run the state’s public assistance programs were full of inefficiencies and the potential for fraud. Caseworkers managed every case from beginning to end with no oversight. The state’s policies for approving benefits to needy residents were also full of gray areas, he found.

In response, Main developed a set of clear-cut management policies and business processes, and with the governor and state legislature’s support, the state awarded a 10-year, $1.16-billion contract to IBM to oversee its public-assistance programs. IBM is taking the state’s existing mainframe system and data, surrounding it with different technology components, and tying it all together with a service-oriented architecture, says John Lyons, an IBM employee who is serving as the deputy program executive for Indiana’s project.

Pass the Aspirin, Please

With a new document-management system, the state is scanning welfare recipients’ paper records and migrating their existing data stored in the mainframe system into a central repository that will let fraud investigators quickly find the records they need, Lyons says.

“In the old days, it was a real headache trying to find anything because everything was on paper. Trying to collect all the data into one investigation file was daunting,” Lyons says. “Now with the touch of a button, you can instantaneously see every piece of correspondence and transaction [for each welfare recipient].”

The workflow management system has built in all the new business practices and policies that Main developed to manage the benefit-approval process. One new guideline, for example, requires multiple approvals before an applicant receives benefits. IBM is also building two fraud-detection applications as part of the workflow management system. One is a data broker that will allow the state’s caseworkers to check people’s applications against other state or federal agency records to verify information, such as Social Security numbers, Main says.

A second application will use data-warehousing tools to analyze the state’s welfare-recipient data and seek out suspicious activity, such as employers or specific streets that have an abnormally high number of welfare recipients, Lyons says.

The state began a pilot of the new system last fall in 12 counties. “We will get rid of fraud institutionally because caseworkers will no longer control cases from beginning to end,” Main says. “Should our new procedures fail to root out the fraud, then our technology, like external data brokers, will go a long way toward reducing it.”

I.T. Aid

Tips for building and running a fraud detection and prevention system

  • Build a data warehouse in easy-to-swallow bites. Connect your department data first, then connect other agency databases.
  • If you’re developing a Web interface, choose an architecture that matches the rest of your IT infrastructure. Java, .Net or PHP: Hypertext Preprocessor should work.
  • Deploy a multinode system, so if one group of servers fails, another group automatically takes over.
  • Ensure data integrity by checking more than one identifying piece of data, such as names against Social Security numbers.
  • Safeguard the data. Use encryption to protect file transfers and put proper controls in place.
Jan 14 2008

Comments