王文周、施黎蒙、Xiaoqian Zhu.2016. Operational risk aggregation measurement based on business lines correlation: A mutual information approach. Discrete Dynamics in Nature and Society, Volume 2016

Abstract:
      The correlation is more complicated than the traditional linear dependence. While Pearson correlation coefficient is constructed by the hypothesis of a linear association, mutual information (MI) is a powerful alternative identification to capture both linear and nonlinear correlation. Based on MI, the generalized correlation coefficient can be obtained and can reflect generalized dependence. This paper evaluates the correlation between business lines by MI. With normal copula, the operational risk capital is calculated. The result shows the correlation between business lines is nonlinear which is undetectable for the linear coefficient and require more operational risk capital than linear correlation.