The Nevada Division of Social Services (DSS) has chosen data and AI leader SAS to help the state reduce error rates in its Supplemental Nutrition Assistance Program (SNAP) program. New federal accuracy requirements mean states with rates above six percent will be on the hook for billions of dollars they didn’t plan for. Nevada DSS will use a social benefits solution, SAS Payment Integrity for Food Assistance, which can not only calculate a more accurate SNAP error rate; it can also identify opportunities to reduce errors caused by human mistakes or deliberate fraud and abuse.
While Nevada DSS has historically had SNAP payment error rates well below the national average, the new six percent threshold put the state at risk of a $52 million budgetary penalty. SAS will analyze the department’s SNAP benefit determinations and provide monthly risk scoring for which determinations are most likely to be in error. Nevada DSS can triage cases most urgently in need of review based on the likelihood of error and the dollar amount at risk of error.
“We have been working diligently to implement plans to reduce that error rate and come in under six percent,” said Kelly Cantrelle, Deputy Administrator over Program and Field Operations at the Nevada DSS. “We are launching technology initiatives that can help reduce utilizing data analytics, along with continuous income evaluation to let us know when income goes over the SNAP threshold. We hope that these two initiatives could potentially lower the SNAP payment error rate by up to two percent, which would bring us under that six percent error rate.”
With just the initial results from the first two months of deploying the SAS solution, NV DSS case workers investigating cases have already identified $130,000 in monthly overpayments.
“It was a ‘prove it’ moment that was quite encouraging,” said John Maynard, a principal solutions architect at SAS and former program integrity lead for Ohio Medicaid. “Nevada has done such a great job keeping their error rate down, that once we have more data in our solution, I believe the state will reach its goal. By assisting Nevada in lowering SNAP payment error rates, SAS will help ensure that families eligible for food assistance will get the full amount of the benefits they qualify for, while also reducing the likelihood of families owing the government money due to overpayment.”
Following the initial implementation, the SAS solution will be automated and integrated into NV DSS’s monthly workflow. This will allow leadership to receive, prioritize and act on automated risk scores while simultaneously identifying suspicious patterns indicative of fraudulent activity.
As the SAS SNAP AI model pipeline ingests more monthly data, the model being trained will become even more precise at distinguishing between unintentional administrative errors and intentional fraud. This continuous model learning and training cycle will enable NV DSS to continuously enhance error and fraud detection capabilities, helping solidify its SNAP payment error rate well below the federal threshold.
