We've all heard of her. A little blond girl with a penchant for trespassing, and very finicky preferences in mattresses and breakfast food. Unfortunately, Goldilocks isn't just part of a fairy tale, she is alive, well, and causing havoc throughout government. If you haven't made the connection yet, "bear" with me. Government is at its best when it is open, transparent and accountable to its citizens. Believe me, I spent nearly 20 years fighting fraud and abuse in workers' compensation by employers, claimants and medical providers for Washington State. However, it is all too easy to criticize staff and management in government agencies for either being "too hard" or "too soft" on the issue of fraud, particularly at the individual case level. After living it for so long, I began calling this the Goldilocks Complex.
I regularly heard the same stakeholders that were clamoring for an enhanced fraud program with more tools, staff and a better approach to rooting out fraud stepping up to argue why we should go easy on a member of their organization when we were taking a tough line on compliance. Fraud was always about "that other guy, not me". When we caught someone working while drawing workers' compensation benefits, or an employer that was stashing hordes of cash while claiming they were out of business, the supporters went away, but most cases aren't so cut and dry, and the evidence rarely includes video of them planning the crime.
So how does a well-intentioned government executive thread this delicate path? The answer is analytics.
Is the bed too hard?
By using analytics to analyze facts, figures and behavior, not only is it possible to determine the cases with the highes potential likelihood of fraud and abuse, but also the factors that cause them to score high, and even estimate the potential value. That information arms staff and management. The cases with the highest scores and dollars at stake go the traditional route of an audit. Others may be handled with lower-level intervention, such as phone calls and letters detailing the issues identified and addressing through voluntary compliance. A request to a doctor may be to just adjust billings for the last three months. The doctor pays less than in a full audit, and no penalties, while the agency achieves compliance with few resources spent, and shows they are watching. Vigilance alone goes a long way towards preventing fraud. At the same time, it shows that you and your agency aren't just heartless bureaucrats, with only one solution to every issue of incorrect billing or potentially abusive behavior. Utilize the tool that is appropriate to the size of the issue, and let analytics lead the way.
Is the bed too soft?
Being able to explain the cases you don't investigate or take action on can sometimes be as critical as the ones that you do. Just ask anyone that is tasked with child protection. Utilizing analytics and good predictive models to decide which claims look like they are right on track and shouldn't have time spent on them is a fantastic way to balance use of resources, eliminate false positives and spend time on true government fraud. Being able to explain why you made those choices helps eliminate claims that you simply don't care about fraud and aren't truly investigating or taking action.
Getting the bed just right
Analytics can uncover deep information about a medical provider and their shady billing practices, or show that a claimant is clearly no longer injured and should go back to work. Better yet, it can draw the connections between a business you are starting to look at, and others. It can show how they are outliers compared to their peers, but oddly connected with other individuals that have already been found commiting fraud or abuse.
If you are commiting resources to a full investigation or audit, make sure it is on the right case, with the best opportunity for return.
The moral of the story
My program improved dramatically by ensuring that we spent resources on not just taking action on cases of fraud and abuse, but on rooting them out. Good detection is key. We starting with data mining and combining data sets from different agencies, then moved further down the path with the SAS Fraud Framework, utilizing anomaly detection and predictive modeling techniques at the core of an analytics approach to find the cases we should spend time on, and the ones we shouldn't. For the ones in the middle, we began using novel approaches that achieved compliance with very few resources spent, which will always receive support from elected officials. It doesn't matter where you are today. Start down that path, and at its conclusion, your bed will be just right.