Operational risk corresponds to losses in banks that are not generated by market and credit risks. The materials presented in this page have been written between 2000 and 2003 when I was in charge with Antoine Frachot of designing the internal model of Crédit Lyonnais. During this period, we have tried to establish a standard model to compute operational risk based on the loss distribution approach. Most of the papers have been written to promote this model and to influence the regulation authorities to accept it. Today, this model is largely used by banks to compute the capital for operational risks.
This paper is a synthesis of the different works we have done at Credit Lyonnais and at ITWG-OR (Industry Technical Working Group - Operational Risk). It explains how to implement LDA in practice, from the estimation of severity and frequency distributions to scenario analysis. This paper has been published in The Basel Handbook: A Guide for Financial Practitioners edited by Michael Ong. The working paper is available at SSRN.
In January 2001, the Basel Committee on Banking Supervision publishes the second consultative paper for Basel II and proposes to use a method called IMA to measure the capital. Using the loss database of Crédit Lyonnais, it appears that IMA is not robust and could not be calibrated. We also decided to write this paper to promote the loss distribution approach. This paper has had a great influence on the computation of operational risk with internal models. In September 2001, six months after the diffusion of this paper, the Basel Committee on Banking Supervision replaces IMA by AMA and LDA. The paper is available at SSRN. A summary of this research has been published in the French journal La Revue Banque.
This research is another example of papers that have been written to influence the regulation. In April 2003, the Basel Committee on Banking Supervision suggests to sum all the capital charges by business lines and risk types to define the regulatory capital requirement. This proposition is equivalent to assume a perfect correlation between aggregated losses. In this paper, we demonstrate that correlations between aggregated losses are necessary small even with conservative assumptions. This paper has been published in the Operational Risk & Regulation journal. It has been reprinted in the book Operational Risk: Practical Approaches edited By Ellen Davis. The working paper is available at SSRN. A summary of this research has been published in the French journal La Revue Banque.
In this paper, we propose a model to solve some data heterogeneity and scaling issues in operational risk. This paper has been published in the Operational Risk & Regulation journal. The working paper is available at SSRN. A summary of this research has been published in the French journal La Revue Banque.
In this paper, we propose several techniques to aggregate heterogeneous operational risk losses. The paper is available at SSRN.
It is the first paper that analyses the threshold problem when we collect operational risk losses. The paper is available at SSRN.
Back to the main page.