Skip to main content

Read our report on six communities’ experiences with pandemic funding and programs, which provides valuable lessons learned to improve federal emergency response programs.

<Case Study>

Our risk model identifies misrepresentations in pension plan assistance applications.

The PACE Built a Predictive Risk Model: $94B Special Financial Assistance Program, 120 Day Review Approval Timeframe +Third-party Risks. PACE Predictive Risk Model PACE Data Scientists built a Predictive Risk Model using Power BI Visualization Tools

The American Rescue Plan (ARP) Act of 2021 appropriated about $94 billion to the Pension Benefit Guaranty Corporation’s (PBGC) new Special Financial Assistance (SFA) program.  

Multiemployer pension plans in financial trouble can apply to the PBGC SFA program for financial assistance. Under program rules, PBGC has 120 days to process applications. The agency reviews the applications only for reasonableness based on information submitted by SFA applicants.  

Given the short application approval time and third-party risk associated with ARP rules, PBGC’s OIG sought the PACE’s assistance. They needed to determine whether a plan should be eligible for financial assistance and how much funding it should receive.  

Our Solution: 

Our data scientists ingested 12+ years of annual pension plan reports, financial statements, and other data to establish trends, detect anomalies, and identify misrepresentation of facts in the application process. We developed 10 risk metrics, ranked by priority.  

We leveraged analytic techniques including outlier detection "Outlier" or "anomaly" detection is the process of detecting data points that dramatically differ from the norm. and regression analysis Regression analysis is a statistical method used to examine the relationship between two or more variables of interest. to detect plan attributes that were “out of the norm” when compared to historic code drivers and other similar plans. We developed a predictive risk model and pattern discovery techniques to highlight abnormal behaviors in SFA applications.  

The solution also included an executive level dashboard with a risk profile for each plan. 

Why this Matters: 

According to PBGC Inspector General Nicholas J. Novak:
"Using estimated risk scores and data within the dashboard, our users can now identify potential high risk or fraudulent applications as they come in. This will greatly improve efficiency. In the short run, the model will mostly be used to identity questionable SFA applications and to retrieve historical...information. In the long run, the model will be used to identify potential issues for plans that received SFA. Another objective is to use the dashboard for our research and analysis [of] all multiemployer plan issues."


Page last modified: 11/06/2023
Thank you for your feedback!
Would you tell us more? Feedback
Was this page helpful?