The great statistician John Tukey once quipped “The best thing about being a statistician is that you get to play in everyone’s backyard.” The broader meaning of this quote is that computational methods that work in one discipline have (frequently unexpected) applications in other disciplines. For example, in medical research, a common technique used to mitigate issues which may occur when patient treatment cannot be ethically randomized, is called “propensity score matching”1.
While practitioners of consumer credit risk modeling are rarely interested in the kinds of experimental design controls needed in medical research, there is an application of this process which has the potential to create value in a financial context. Specifically, we can use this process to match accounts with known financial performance (e.g., repayment) to accounts under consideration, where future performance needs to be predicted and the risk associated with these accounts needs to be quantified. In other words, propensity score matching is an alternative to traditional account scoring and has shown promise in predicting payment performance over time.
Using historical account-level data from our FAAzE data platform, FLOCK Specialty Finance is piloting a wide range of data science and computational methodologies like propensity score matching to enhance and broaden the way we think about pricing and risk. Integrating evolved analytical approaches like propensity score matching into our risk modeling approach is part of FLOCK’s commitment to our investors and to our clients…to be more than a transaction.
For more information about FLOCK Specialty Finance, contact Jennifer Lewis Priestley, CDO (jpriestley@flockfinance.com)
[1] journals.lww.com/ccmjournal/Citation/2023/03000/Outcomes_and_Adverse_Effects_of_Baricitinib_Versus.1.aspx
