|
FI’s key deliverable was a series of econometric models that predict default and prepayment on RD business loan guarantees at the loan-level.
FI began by conducting a detailed analysis of B&I program regulations, operating procedures, and policy history. We obtained historical transaction and position data from RD’s systems, assessed it for quality using a series of automated tests, and converted raw data into datasets structured to support analysis. We used these datasets to estimate econometric equations that correlate default and prepayment activity with macroeconomic conditions, and with loan, borrower, and lender characteristics. After testing more than one hundred alternative equations, we selected final equations that balance efficiency and explanatory power.
From our experience supporting Federal credit agencies, we understood the inter-organizational dynamics of implementing model changes. As RD’s adoption of the model is subject to approval by OMB, we sought to engage OMB early in the process and solicit their input on major methodological decisions. Also, as modeling results feed into RD’s financial statement audit, we ensured that we maintained an audit trail, effective documentation, and a transparent analytical approach.
FI helped RD gain new insights and capabilities to manage its multi-billion dollar business loan guarantee portfolio. USDA better understands program performance and risk drivers, and can make better-informed policy decisions in support of its mission.