Are Default Rate Time Series Stationary? : A Practical Approach for Banking Experts

Szigel, Gábor and Gyűrűs, Boldizsár István (2023) Are Default Rate Time Series Stationary? : A Practical Approach for Banking Experts. FINANCIAL AND ECONOMIC REVIEW, 22 (4). pp. 107-135. ISSN 2415-9271

Preview

Text fer-22-4-st4-szigel-gyurus.pdf Download (1MB) | Preview

Abstract

As the IFRS 9 accounting standard requires banks to recognise impairments based on a forward-looking expected loss concept, banks must estimate the quantitative relationship between default rates and macroeconomic indicators (GDP, unemployment, etc.). In such models, the stationarity of the (usually short) default rate time series is often the most critical issue. In this article, we provide practical advice for banking experts on how (under which circumstances) they can still use short default rate time series in OLS regressions even if those fail regular stationarity tests. We argue that if margin of conservativism is requested for the underlying default rate projections, then applying (seemingly) non-stationary default rate time series in OLS models might not necessarily be problematic.

Actions (login required)

Edit Item