TESTING FOR A STRUCTURAL BREAK IN DYNAMIC CONDITIONAL CORRELATION MODELS

In this paper we propose an approach for testing a structural break in dynamic conditional correlation models that is based on the extension of ordinary model equation with differential parameters, combined with dummy variables. This enables us to directly observe the difference in individual parameter or in a combination of parameters before and after the break and offers us a lot of testing opportunities. We make an extensive Monte Carlo simulation experiment of the proposed methodology for both the DCC and the ADCC multivariate GARCH models which indicate good small sample properties. In the empirical part we perform the test on the daily data of some European stock indices where we find a significant structural break in both the long-run mean and the dynamic part of the models after the introduction of the currency euro.