Financial co-movement and correlation: evidence from 33 international stock market indices

The analysis of financial market co-movement is an important issue for both policy makers and portfolio managers, for example, in terms of policy coordination and portfolio diversification. This paper presents evidence based on a data set of 33 daily international stock market indices. Initially, using established cointegration and multivariate GARCH frameworks, we report results that suggest correlations with the US have not in general exhibited an upward trend. The main exception to this is the G7 economies, although even here the correlations declined over the last part of the sample. On a regional basis there is stronger evidence of rising correlations, notably in Europe, although again this evidence is not ubiquitous. Further, we also implement the recently developed non-parametric, model-free and realised variance methodology to generate correlation coefficients. This method overcomes deficiencies in both the cointegration and GARCH methods. The results here largely support those of the GARCH analysis. Finally, we use the time-varying correlation methods to form international portfolios and compare their performance to that of an equally weighted portfolio. Results suggest that accounting for time-variation is particularly beneficial for the larger markets but more marginal for the smaller markets. Our results thus suggest that there remains room for portfolio managers to obtain diversification benefits, while policy makers may need to take in account possible adjustment costs of coordinated action.