Estimating the capital charge for external fraud risk： A clinical study

The main objective of this paper is to estimate the economic capital for covering the external fraud risk within a financial institution. This is a kind of operational risk which is due to acts of a type intended to defraud, misappropriate property or circumvent the law, by a third party. From the methodological point of view, we apply the Loss Distribution Approach （LDA）, based on the Internal Operational Loss Database （IOLD） provided by a Spanish Saving Bank. More specifically, we asses the potential impact of the severity distribution on the Capital at Risk （CAR）. In absence of normality, we try to adjust the Lognormal, Weibull and Exponential functions when modelling the severity of losses. As a result, we find a high divergence in terms of capital charge depending on the statistical model selected. In consequence, in order to obtain a realistic model, we highlight the relevance of the goodness of fit between the empirical and the theoretical distribution.