Risk Factors, Copula Dependence and Risk Sensitivity of a Large Portfolio.

Summary In this paper, we propose a flexible tool to estimate the risk sensitivity of a high-dimensional portfolio composed of different classes of assets, especially in extreme risk circumstances. We build a so-called Cvine Risk Factors Model (CRFM), which is a non-linear version of a risk factor model in a copula framework. Our tool allows us to decompose the risk of any asset and any portfolio into specific risk directions depending on the context. As an application, we compare the sensitivity of different types of portfolios to extreme risks. We also give an example of a view-type analysis as usually performed by portfolio managers who examine what their portfolio becomes under specific circumstances: here we examine the case of a low inflation context. These analyses allow us to detect changes in the diversification opportunities over time.