The first step is to answer the question of what “real-time” actually means. Depending on the program and industry speed of response, I’ve heard answers that varied from milliseconds for the biggest banks in processing credit card charges to 24 hours for some government programs. A better description of what most government agencies are trying to achieve isn't real-time fraud detection, but near real-time, or pre-payment. The concern is all around decisions before the money goes out the door, and timeliness depends on how quickly benefits are paid based on an eligibility determination or a specific billing or claim.
From a technology point, pre-payment fraud detection is definitely alive and kicking across many industries. The fraud-fighting and anti-money laundering technologies of the SAS® Enterprise Fraud and Financial Crimes Framework are in place in banks across the world processing transactions nearly instantaneously, utilizing analytics to detect anomalies and predictive models based on past cases to flag suspect people and transactions. Some large government organizations are moving rapidly towards pre-payment approaches, both here in the United States and abroad.
So, if the technology isn’t the sticking point, what is? Data, people and processes. Let’s start with data. Key information for making eligibility and pre-payment decisions in programs often involves matching files from different state and federal agencies. The timeliness of death matches, incarceration data, and information on benefits an applicant receives from other programs may only come in once a week, month or quarter. A decision that needs to be made in the next 48 hours can’t be held up for a month while waiting for the latest incarceration data. The best opportunity is to match from the last report, which will still catch many attempts at fraud, make the determination, and keep matching against future reports.
People is another challenge for government. The book and movie Moneyball showed how analytics could be applied to baseball to help a team with a small payroll beat the big guys. It is an inspiring story built on real events, but I kept thinking that if this was about government, the Athletics would have only been able to field six players, and still try to beat a team with all nine positions covered. If an organization truly doesn’t have the staff to process the information from the cases that are flagged pre-payment, then the detection solution isn’t utilized and errors and fraud continue unchecked.
The key here is changing processes to ensure that there are even a few people dedicated to that triage process. With enough information pulled together from different sources, those determinations are much faster than they were previously. When handled correctly on the front end, fewer cases are opened for payment, as well as fewer investigations and overpayments down the line. Those improvements in total time across the life of all claims and claimants offset the small investment in pre-payment triage, bringing life to this approach.
Enough about my thoughts. I want to hear from the government staff and managers out there. Do you believe it is possible to move fraud detection up front to become fraud prevention? What’s holding you back? Who has made some interesting strides in this area?