Case Study

Challenge: The Federal Emergency Management Agency (FEMA) administers programs that provide monetary assistance to people and entities impacted by disasters. These assistance programs are hindered by fraudulent requests for funding, so FEMA contracted the Summit team to assist in strengthening the agency’s existing fraud detection infrastructure through accurately depicting the process flow of application data and finding additional fraud identifiers. Summit’s role was to augment FEMA’s current practices to improve fraud detection capabilities.

Solution: The Summit team met with key FEMA stakeholders to better understand the nodes an application moves through prior to being closed (i.e., denied funding). As part of the conversations, the team received other potential fraud identifiers that inspectors see during their operations. The Summit team then designed a logistic regression to estimate the likelihood that an application for assistance is fraudulent. To create the model, Summit leveraged the preexisting fraud detection methodology used by FEMA then enhanced it by conducting an evaluation and identifying other applicant attributes that were correlated with closed applications. By leveraging FEMA’s existing fraud detection infrastructure, the Summit team was able to produce a model that combined FEMA’s institutional knowledge with the statistical analysis and econometric modeling expertise of the Summit team. 

Result: FEMA implemented the model Summit produced in October 2022. The model has expanded FEMA’s ability to detect fraudulent funding requests, thus ensuring that assistance is received by genuine applicants at a higher frequency. Furthermore, Summit’s investigation of the current fraud infrastructure within the organization allows FEMA to identify and close gaps efficiently so that the impact of disasters can be mitigated more effectively.