A gravity analysis of international stock market linkages

Abstract

The last decade has witnessed a marked improvement in information technology. Such an improvement has reduced the information cost for market participants. Thus, whether the influence of geographic factors on international financial linkage is still significant nowadays is an important question yet to be addressed. This article develops a gravity model of international financial linkages. Using the panel data of bilateral cross-country stock market correlations of 23 countries, it is found that the correlations are negatively associated with the Great Circular Distance (GCD) between the financial centres of these countries and positively associated with the duration of overlapping trading hours among stock exchanges and the colonial links between countries. However, whether the countries share a common border or language does not affect the stock market correlations.

Acknowledgements

We thank Tiffany Wong for able assistance. All errors are ours.

Notes

¹For more studies on the comparison of stock market movements, one is referred to Gallagher et al. (1997) and Taylor and Tonks (1989).

²The gravity model has long been used to explain the international trade of goods (Bergstrand, 1985). Bilateral gross aggregate trade flows are usually explained by the following specification: , where M _ijk is the dollar flow of good or factor k from country i to country j, Y _i and Y _j are incomes in i and j, N _i and N _j are population in i and j and d _ij is the distance between countries (regions) i and j. U _ijk is a log normally distributed error term with E.

³The border variable indicates whether there is a common physical border between two particular countries. Countries connected by a bridge or tunnel (e.g. France and the United Kingdom) or those separated by a narrow strait (e.g. Malaysia and Singapore) are considered to have a common border.

⁴We have five countries that use English as their main languages. Spain and Mexico share a common language, and Portugal and Brazil also share a common language.

⁵We use GDP per capita from the International Financial Statistics (IFS) to classify the countries. The variable is used to analyse whether the extent of development of the country's economy affects the stock market correlations.

⁶The result is in line with French and Poterba (1991) who suggested that investors hold a large portion of their equities in terms of domestic assets. Such a home bias occurs because investors feel more comfortable when investing in domestic assets they are familiar with. Accordingly, the contagion effect between neighbouring countries is stronger than that between distant countries.

⁷This is because markets with common trading hours react to information simultaneously, and the simultaneous changes in stock prices lead to an increase in correlation.

⁸This is partly due to the launch of the Euro, which eliminates the currency risks within the Euro zone and removes some existing barriers to cross-border investments.