Private credit originators are often required to employ qualitative overlays (also called post-model adjustments or overrides) to adjust credit risk model outputs when models alone cannot fully capture certain risks. These overlays are judgmental adjustments applied on top of quantitative model results – for example, increasing expected loss estimates or modifying risk ratings – to account for limitations like scarce data, model deficiencies, or new risk factors. Such overlays became especially prominent during unprecedented events (e.g. the COVID-19 pandemic), and remain an integral tool for addressing emerging risks that lack historical data.
When used with robust analysis and governance, investors should welcome overlays as a way to ensure credit risk is measured prudently. However, excessive reliance on judgmental overlays can mask model weaknesses, so non-banking institutions have the pressure to govern overlays carefully and ultimately improve their models .
This report examines guidance on overlays and industry practices among non-bank private debt originators. Particular focus is given to the potential use of generative AI for overlay governance. We provide examples of overlays for emerging risks like trade tariffs, technological disruption, and climate change.
