LA County closes in on fraud rings with the SAS® Fraud Framework
Los Angeles (LA) County has the largest population of any county in the nation, with approximately 27 percent of California's residents. If it were a country, LA County would be among the top 20 in the world as measured by GDP. The county's budget is approximately $23 billion dollars, which goes toward providing social, health-related and law enforcement services.
Customer Success Video
Check out this video to learn more about Los Angeles County and its successes with SAS.
View Video(Runtime: 5 mins., 14 secs.)
You have questions; our customers have answers. Check out this video Q&A.
Director of Research
(Requires Windows Media Player 6.4.7 or higher)
The Chief Executive Office is responsible for establishing and administering the budget for the county's services and programs, which includes monitoring and measuring them for efficiency and cost-effectiveness.
"Using the results of the POC we did projections and cost-savings analysis and concluded that the county could expect to see a return on investment of somewhere between $7 million to $30 million annually," says Manuel Moreno, Director of Research, Chief Executive Office. "The POC showed a number of advantages in using the technology, not only related to costs, but how we prioritize investigations. With social network analysis and predictive analytics, we can identify benefit recipients and service providers and predict those most likely to engage in fraud and where there might be a potentially large loss of funds."
Moreno says common fraud cases are characterized by criminals filing false claims of employment, which usually include declaring employees who don't exist. In some cases, businesses are set up by the heads of fraud rings. These individuals are in collusion with recipients, who falsely declare that their children are attending nonexistent child care programs. Sometimes, criminals declare work schedules that are either false or shorter than the amount of time claimed.
The POC demonstrated that by using predictive models and peer group analysis to detect behavioral anomalies in the utilization of child care services, fraud risk scores could be developed to help decrease the number of false positive cases assigned to investigators. The POC also allowed the team to map potential fraud rings.
"One of the things we showed is the system's ability to utilize social networks to detect if individuals are likely to commit fraud, based on their fraud risk score," explains Moreno. "We established a network, for example, consisting of participant and provider nodes and illustrated their relationships. We looked at whether this small network fit into the larger scheme of networks, in which participants are in collusion with other child care providers. With the technology, we identified strong central nodes. In one case, we had a child care provider servicing many nodes of participants engaging in collusion activities."
"We needed a unified approach to get a handle on the fraud problem," Moreno continues. "We wanted a solution that provided data integration, as well as a powerful analytical tool workbench. To perform sophisticated predictive modeling we needed historical data, and for that we had to integrate many external and internal data sources, such as the state of California's Employment Development Department data and business license data from Los Angeles County."
Based on metrics from the POC, Moreno and his team concluded that data integration, predictive capabilities, data mining tools and the insight provided by the SAS solution can be used effectively to detect fraud before it occurs.
"We calculated that the accuracy rate of fraud rings identified by the social network analysis solution to be, with reliability, 85 percent."
Copyright © SAS Institute Inc. All Rights Reserved.
Los Angeles County
LA County faced a 40 percent growth in fraud activity related to child care services provided to CalWORK's participants in the Stage 1 Child Care Program
Expected return on investment is between $7 million and $30 million annually. An 85 percent accuracy in identification of suspected fraud rings, and the ability to integrate multiple internal and external data sources to apply sophisticated predictive analytics
“We calculated that the accuracy rate of fraud rings identified by the social network analysis solution to be, with reliability, 85 percent.”
Manuel Moreno, PhD
Director of Research