Dynamic Correlations Between the U.S. Equity Market and Australian Sector Indices and Their Regime Differences: Evidence from DCC-GARCH with Wavelet–Changepoint Segmentation

Abstract

This study observes the time-varying linkage relationship between the US stock market and the three industry indexes in Australia, and wants to see if this correlation has different laws at different stages. We use the daily index data from Yahoo Finance to calculate the logarithmic rate of return, and also use ADF and KPSS test to check the stationarity of the data. The division of stages is automatically determined by a data-driven method: firstly, an American volatility proxy index is decomposed by wavelet, and then combined with the detection of PELT mutation point, and finally two breakpoints are found, which divides the whole sample period into three stages. Then, we estimate a DCC-GARCH model to find out the dynamic correlation between the American market and each Australian industry. The results show that there has always been a positive linkage between them, and the ranking of strong and weak industries is very stable: Consumer Discretionary has the strongest connection with the American market, followed by Industrials, and Consumer Staples has the weakest correlation. On the whole, the average correlation in different stages does not change much, but the tail linkage and related volatility in the middle stage will become stronger, which means that the effect of diversification will become worse when the markets are more closely linked. We use the thick-tailed DCC model with multivariate T distribution innovation to test the robustness, and the results are consistent with the previous ones, and the information criterion is better than the Gaussian benchmark model.