|
Abstract:
|
The use of copulas for modeling dependency has been a classical topic in financial research. What make the task of modeling market dependency challenging are the complex financial networks along with the time-varying nature. When copulas are applied to financial time series, they are usually assumed to be static over time. However, in the literature of empirical finance, it has been a stylized fact that correlations observed under ordinary market conditions differ substantially from those observed in hectic periods.
In this project, we propose a class of time-varying full-range dependence copulas for modeling market dependency. Our models can not only capture the change in the degree of dependence over time, but also the varying type of dependence structure (e.g., symmetric versus asymmetric dependence, and tail dependence versus independence). Real financial data will be used to demonstrate the usefulness of our proposed model.
|
