Modelos avançados para risco operacional: uma análise empírica da abordagem de distribuição de perdas

Abstract

Operational risk is a fascinating issue and has been experiencing a rising relevance both in academy and in financial and capital markets. This research aims to provide an in-depth assessment of the loss distribution approach (LDA) appropriateness as a tool for measuring operational risk capital requirement for a Brazilian large-sized financial institution. Operational loss data within a five-year span were collected, which allowed the measurement of capital requirement for the eight event types of operational risk occurred in the Retail Banking business line. Twenty-four distinct scenarios have been prepared from several changes in model parameters, such as imposing different thresholds levels, the data adjustment by inflation index, and the possibility of grouping data caused by a common operational loss event observed at the same day. Research results suggest that parameters settings are a critical feature for the process and can substantially change operational risk capital requirements. Results still point to the need to determine frequency and severity distributions for each combination of business line-loss event type, since theres no a priori appropriate distribution that should be used in the calculation of operational risk capital requirement.