ABSTRACT

In recent years, the occurrence of operational losses in financial institutions has increased the interest of academics and policy makers in operational risk. One of the main problems regarding the economic capital required to cover operational risk is the lack of sufficiently large databases. We present a set of models mixing internal and external data to predict both the severity and the frequency of operational losses. We show that, rather than a one-size-fits-all solution, there are several approaches, each presenting opportunities and limitations in the logical framework. Our findings offer useful insights for enhanced risk practice and prudential supervision.