Impairment Forecasts with Stage Migration

Abstract

With the introduction of expected credit loss based impairment methodology banks stress testing programs need to include forecasts of stressed impairment losses as an important component in the firmwide stress testing programs. The forecasts of stressed impairment losses are comprised of incurred loss up to the forecasting horizon of the stress test, and, the regular expected credit loss after the stress test horizon. In this paper we analyze the effect of IFRS 9 stage transfer on the stressed impairment forecasts. For a sample state transition model, portfolio and stage transfer rules it is shown that the stage transfer can have significant effects on the staged impairment forecast compared to the IFRS 9 actuals (current expected credit loss). Specifically, stage 1 dominated initial portfolios can have significantly increased stressed impairments forecasts compared to the IFRS 9 actuals impairment calculation under stressed scenario. This reflects their downside in terms of stage transfer potential to stage 2 up until forecast time. Recently downgraded or lower quality portfolios in stage 2 can have the opposite effect of decreased stressed impairments forecasts compared to the IFRS 9 actuals impairment calculation under stressed scenario. This reflects their upside in terms of stage transfer potential to stage 1 up until forecast time. A lifetime measurement only approach for all assets (as in US CECL) ensures a direct link between actuals stressed impairment amounts and stressed impairment forecast amounts. Their difference can be attributed to the forecast new business assumptions and natural decay due to run-off. Taking into account stage transfer in impairment forecasts we find it is more natural to analyze the stressed impairment forecast relative to a baseline scenario impairment forecast benchmark. Such an approach focus more specifically on the macroeconomic stress effect and its impact.