Is World Stock Market Co-Movement Changing?

Abstract

We examine world stock market comovement for 23 countries over 30-years from 1981 to 2010 -- a period that includes major changes in international trade institutions, capital market regulations, and the advent of the internet era. Our analysis is based on a generalized factor model that allows for country, regional, and world factors. We show that the commonly used world Fama French factor model with the world value-weighted portfolio as a proxy for the world factor, has biased results. We develop a new methodology, the Weighted-Generalized Canonical Correlation (W-GCC) method, to identify the presence of a world factor that explains variation in the stock price movements across all countries. We show in simulation studies that our methodology is able to identify the country factor even when the world factor is weak in some countries. We test for changes in comovement allowing for a structural change. We find that stock price comovements with respect to the world factor are stable through the mid 1990s and then starts increasing from the mid 1990s through 2010. The world factor we identify is highly correlated with the world factor estimated using an out-of-sample data set of ten newly emerged countries and comovement of these country returns also increase from 2000 to 2010. The Chinese B stock market shows a dramatic increase in comovement over the time period but not the Chinese A stock market, suggesting that comovement is more likely driven by investment cash flows. Our results have important implications for the benefits of international portfolio diversification strategies and the convergence of the cost of capital across countries.