The Impact of Higher Order Moments on Market Risk Assessment

Abstract

For a long time, the normality assumption has been used extensively in the literature for its simplicity. However, recently there is new evidence that the properties of the normal densities are no more accurate for dealing with stylized facts such as the return time series of financial assets which deviate obviously from the normal shape. In order to overcome the misleading inferences of the normal density when assessing the market risk, we have to take into account the higher moments (higher order than the variance) of the returns’ distributions. According to basic finance theory, investors are looking for stability and mean return (asymmetry) and trying to avoid return's variance (kurtosis). In this context, we try to make risky measures more robust. So we implemented a new measure of risk based on four moments of the return time series and also accounts for the existence of extreme events. The purpose of this work is to test the accuracy of this new multi-moments risky measure on the French financial market especially in the period of crises. The results show that the first four moments of the return time series seem to be still insufficient to predict accurately the probability of crisis occurrences.