LA County unravels a web of child care fraud
- By Paul McCloskey
- Oct 19, 2015
When Los Angeles County began to explore data mining to help track fraud in its California Work Opportunity and Responsibility to Kids Child Care Program, it was daunted by the complexity of the scams it faced.
The county’s Department of Public Social Services (DPSS) began to see signs of escalating collusion among providers and recipients in the program, which helps low-income families pay for child care services so parents can go to work.
“The collusion occurs between the care providers and the recipient, and then it expands like a web,” said Michael Sylvester II, assistant director of the county’s Bureau of Contract and Technical Services.
In a typical fraud scenario, an unlicensed provider takes care of many children in a home setting. Kickbacks might flow from a provider to a care recipient or from provider to provider across large and small networks.
Another common scheme is for a provider to receive CalWORKs money when kids are actually in free after-school child care or at home. “That person is really not providing child care,” Sylvester said. “Instead, they are taking the child care fees and divvying them up in many different ways.”
More and more of the collusion is driven by fraud rings. Identifying the size and scope of those networks provided an incentive to increase the county’s investment in an anti-fraud data-mining solution.
“The rings can get pretty big,” Sylvester said, adding that the county’s original goal was to find out how far the criminal networks extended. To do so, officials launched the Data Mining Solution (DMS) for Child Care Welfare Fraud Detection.
DPSS drew on the SAS Fraud Framework for Government to build a platform that ties together data mining, social network analysis and rules management applications. The county’s data is maintained in a secure cloud hosted by the company’s Advanced Analytics Lab for State and Local Government.
The social networking analytics have helped the county drill down into program data and begin to identify fraud networks. Via a dashboard, investigators can pull up data on providers, families using the providers, the work locations and the phone numbers of people tied to those locations.
“You start to see how the linkages spawn off,” Sylvester said. “This lets our investigators really drill into the data, open it up and find out, my goodness, there are 10 others associated with that beneficiary.”
The county also developed a triage unit to analyze alerts and funnel suspected fraud cases to investigators. The triggers are activated by business rules showing potential fraudulent activity, such as a person reporting income paid in cash or an unusually long distance between a beneficiary’s home and the child care site.
So far, the system has had a positive financial impact, said Sylvester, who estimates that the the project has cut time to prosecute fraud investigations by 18 months and generated millions in annual cost savings. It has also led to hundreds of fraud referrals to county caseworkers, leading to cost avoidance savings throughout the department.
Looking a few years down the road, Sylvester predicts that data mining and other tools will continue to improve investigators’ ability to find new data sources that will help them stay in front of their casework. Some of those sources might eventually include external public- and private-sector data.
“Realizing you have huge data stores with information in them that can be pulled out and put together, it’s incumbent on all of us to find ways to put them in the budget and enable investigators who are really working hard to catch perpetrators,” Sylvester said.
Paul McCloskey is senior editor of GCN. A former editor-in-chief of both GCN and FCW, McCloskey was part of Federal Computer Week's founding editorial staff.