A0829
Title: Dynamic spatial panel data models with copulas and model validation
Authors: Hideatsu Tsukahara - Seijo University (Japan) [presenting]
Abstract: The purpose is to consider a dynamic spatial panel data model with common factors as an extension of multifactor models in financial econometrics. It incorporates explanatory variables in the classical linear regression manner and utilizes a spatial weight matrix as a means to express network dependence. The underlying disturbance distribution is assumed to be Gaussian. By fitting a skew-t copula to the residuals, one can check the Gaussian assumption as well as whether the asymmetry in dependence structure and tail dependence have been captured by the variables already in the model. The aim is to propose estimation and testing methods for unknown parameters in the model, and investigate their properties. Some simulation results and empirical applications to real data are given.
