Optimal look-back period for adequate and less procyclical credit capital forecasts

Abstract

This study presents an approach to determining the optimal look-back period for adequate and less procyclical credit capital requirement forecast of U.S. commercial banking system under the regulatory framework. The approach conducts rolling window estimations together with back-tests and procyclicality checks for various sizes of sample windows and portfolio aggregation methods. The empirical results first show that capital requirement forecasts from shorter-term samples are more adequate but more procyclical compared to those from longer-term samples. Second, the forecasts based on stand-alone method are more adequate than those taking diversification into account. Third, tail dependence is inevitable for diversification methods to satisfy the adequacy. Finally, based on these findings, the sample of two business cycles, twice that of the Basel Committee's minimum requirement of the historical observation period for internal rating grade, provides micro-prudentially the most adequate and macro-prudentially less procyclical credit capital forecasts in the U.S. commercial banking system.