Porównanie bayesowskich modeli Copula-AR(1)-GARCH(1,1) z asymetrycznością rozkładów warunkowych
Comparision Bayesian Copula-AR(1)-GARCH(1,1) Models with Asymmetric Conditional Distribution

Summary/Abstract: The main aim of the paper is to formally assess the relative explanatory power of competing bivariate Copula-AR-GARCH models with symmetric and skewed Student t distributions on the example of data from the The Warsaw Stock Exchange. The subject of comparison were 22 Copula-AR(1)-GARCH (1,1) models, which differed in assumptions on the copula and the occurrence of skewness in marginal distributions. In the context of the models under consideration, Monte Carlo Important Sampling methods were used to estimate the characteristics of a posteriori distribution and the marginal density of the observation matrix. For analysing empirical data, a posteriori models turned out to be ones more likely to have symmetrical conditional t-Student distributions. For the logarithmic daily growth rates of the two sub-indicies of the stock index WIG, the highest a posteriori probability was obtained by the Clayton-Gumbel copula model. The use of the skewed Student's t-distribution did not improve the explanatory power of the Copula-GARCH models.

• Details
• Contents