Title: Dynamic copula factor model and its application to financial risk assessment
Authors: Ziyi Wang - The Hong Kong University of Science and Technology (Hong Kong) [presenting]
Mike So - The Hong Kong University of Science and Technology (Hong Kong)
Abstract: Correlation analysis has been an important component of financial risk analysis. However, the nonlinear dependence among financial returns and time-varying features haven not been fully captured by existing models. By incorporating market factors under the CAPM model, we propose a new cross-sectional vine copula factor model to better capture the dependence among financial returns. Vine decomposition is applied to estimate conditional dependence by expressing a high-dimensional distribution by linking the financial returns to the market factors and linking the market factors by copula functions. With the modeling of the marginal distribution of returns using a GARCH-t structure, the proposed model can capture non-linear and non-monotonic dependence while accounting for heteroscedasticity in financial returns. The computation burden due to high-dimensionality now concerns only the number of market factors, regardless of the dimension of financial returns. Simulation study is performed to illustrate that our methodology works in high-dimensional situations. An empirical study with multiple financial time series is also conducted to illustrate this new model.