افزایش در حرکت مشترک در بازارهای سهام ملی: یکپارچه سازی بازار و یا حباب تکنولوژی اطلاعات؟

A stylized fact in the portfolio diversification literature is that diversifying across countries is more effective than diversifying across industries in terms of risk reduction. But with the rise in comovement across national stock markets since the mid-1990s, this no longer appears to be true. We explore if this change is driven by global integration and therefore likely to be permanent, or if it is a temporary phenomenon associated with the recent stock market bubble. Our results point to the latter hypothesis. In the aftermath of the bubble, diversifying across countries may therefore still be effective in reducing portfolio risk.

One of the most pronounced empirical regularities in international equity markets has been the low degree of correlation of returns across national stock markets. This empirical regularity has broken down in recent years. As Fig. 1 shows, the correlation coefficient of US stock returns with equity returns in other developed countries has risen from a relatively stable level of around 0.4 from the mid-1980s through the mid-1990s to close to 0.9 more recently. There are several possible explanations for this. First, there may have been a decline in home bias in the portfolio holdings of investors. As a result, the marginal investor in German equities may no longer be German, so that country-specific investor sentiment now plays a smaller role in national equity markets. Second, firms may be becoming more diversified across countries in their sales and financing. As a result, companies around the world may be more exposed than before to the global business cycle, causing national stock markets to move together more. Third, it is possible that the rise in comovement since the mid-1990s is simply a temporary phenomenon associated with the recent stock market bubble.2 For the portfolio diversification literature, the question of whether the rise in synchronization across national equity markets is driven by fundamentals, and therefore likely to be permanent, or if it is linked to the recent stock market bubble, and thus more likely to be temporary, is of great interest. This is because recent studies have challenged the widely held view that variation in international stock returns is due mainly to country-specific effects. For example, Heston and Rouwenhorst, 1994 and Heston and Rouwenhorst, 1995 show that country-specific sources of return variation are dominant even in geographically concentrated and economically integrated regions such as Western Europe. Griffin and Karolyi (1998) find that this is even more so for a broader sample that includes emerging markets, where global industry factors explain only around four percent of the variation in national stock markets.3 However, against the backdrop of the dramatic rise in comovement across national stock markets in the late-1990s, recent additions to the diversification literature have found that global industry effects are becoming more important. Baca et al. (2000) report that the importance of global industry factors in explaining international return variation increased towards the late-1990s. Cavaglia et al. (2000) show that industry factors surpassed country effects in importance in the late-1990s, concluding that diversification across industries may now provide greater risk reduction than diversification across countries. This result matches that in L'Her et al. (2002) who find that global industry effects surpassed country effects in importance in 1999–2000. These findings dovetail with a growing conviction in the investment community and in the financial press that globalization and the new economy are raising the importance of global industry effects, at the expense of country-specific factors. Against this background, we make several advances over the literature. We construct a new dataset that covers virtually the entire global stock market in capitalization terms and find—for this more comprehensive dataset—that industry effects have gone from less than half as important as country effects in the mid-1990s to almost twice as important in recent years. This shift is primarily driven by a dramatic rise in magnitude of industry effects, with country effects roughly stable since the mid-1990s. But what is the rise in global industry effects capturing? Is it a reflection of greater economic and financial integration across countries, in which case the rise in sector effects is likely to be permanent, or is it simply a temporary phenomenon associated with the recent stock market bubble?4 We explore this question in two ways. First, we examine the breadth of the recent rise in sector effects, by exploring the evolution over time of country and industry effects outside of the technology, media and telecommunications (TMT) sectors. Our focus on TMT derives from the fact that these sectors have been identified in financial circles as being central to the recent stock market bubble.5 We find that, outside of TMT, there is no significant rise in the absolute or relative importance of global industry effects since the mid-1980s, something we find hard to reconcile with the notion that their rise is capturing greater economic and financial integration. For one thing, this is because there is no a priori reason to think that greater integration should be confined to a narrow set of sectors. For another, the greater volatility of TMT stocks is perhaps a temporary phenomenon associated with the recent stock market bubble. Beyond TMT, we find no systematic increase in the importance of industry effects. Instead, we observe that the ratio of industry to country effects follows a U-shape pattern from the mid-1980s to the late-1990s, a cyclical pattern whereby global industry effects become temporarily more important in relative and absolute terms around periods of stock market distress, such as October 1987 and March 2000. We view this cyclical pattern as further evidence that the recent increase in industry effects is temporary. But what if the recent rise in TMT industry effects reflects the fact that these sectors are more international than their “old economy” counterparts, something for which there is at least anecdotal evidence?6 We address this question in two ways. First, we follow Griffin and Karolyi (1998) who distinguish between traded and non-traded goods sectors. Consistent with their paper, we find that country-specific factors explain a larger proportion of the variation of stock returns for non-traded than for traded goods industries. However, we find that the relative importance of global industry effects rises faster towards the end of the sample for non-traded than for traded goods sectors, something we find hard to reconcile with the hypothesis that the recent rise in industry effects is driven by globalization. Second, we collect firm-level balance sheet data from Worldscope on how integrated the firms in our sample are into the global economy, using the percentage of total sales that are generated by foreign subsidiaries and the fraction of total assets that are held by these subsidiaries as proxy measures for international integration. Compared to the sector-level traded versus non-traded goods classification, this approach reduces the potential for measurement error because traded goods firms may after all not export and non-traded goods companies may sell their services abroad. Ranking our sample by the share of international sales and assets that firms have, we find that global industry effects explain more variation than country effects for stock returns of firms in the top quartile (internationally diversified) than in the bottom quartile (not internationally diversified). However, the ranking also shows that the relative importance of industry effects has risen by more since the mid-1990s for the bottom than for the top quartile of firms. Again, consistent with our results for the sector-level traded versus non-traded goods classification, we find that industry effects towards the end of the sample have become more important for firms that are less global—more evidence against the hypothesis that the recent rise in sector effects is driven by greater global integration. We address the same question using a slightly different approach. One way in which the recent literature has rationalized the growing importance of global industry effects is that they are capturing the increasingly global nature of certain industries. We investigate this hypothesis explicitly by using the foreign sales and asset data to sort the firms in our sample into deciles and creating decile dummies that proxy for the extent to which firms operate globally. We then augment the standard regression in the literature—a cross-sectional regression of international stock returns on country and global industry dummies—with these decile dummies, which we call international diversification effects below. In the event, we find that they explain virtually none of the recent rise in industry effects. Our attention has so far focused on the changing importance over time of the global industry effects. We also investigate the variation over time in the importance of country-specific effects—for our global sample they have declined significantly since the mid-1980s. We explore, with a special focus on emerging markets, if the dates for the opening of financial markets in Bekaert et al. (2002) explain variation over time in the importance of these country effects. We find that the dates for financial liberalization in Bekaert et al. (2002) match up roughly with declines in the importance of country-specific effects in many emerging markets during the mid-1990s. However, these declines were reversed during the Asian crisis and country effects for emerging markets remain much larger today than for mature markets, as recently established by Serra (2000). In summary, we find that there is little indication that a rise in global integration at the firm level is driving the recent rise in industry effects. In the aftermath of the bubble, diversifying across countries may therefore still be effective in reducing portfolio risk. The remainder of this paper is organized as follows. Section 2 discusses the data, while Section 3 reviews our empirical approach. Section 4 presents the results, and Section 5 concludes.

The degree of comovement across national equity markets has increased dramatically since the mid-1990s. In this context, Baca et al. (2000) report that the importance of global industry factors in explaining international return variation increased towards the late-1990s. Cavaglia et al. (2000) show that industry factors surpassed country effects in importance in the late-1990s, concluding that diversification across industries may now provide greater risk reduction than diversification across countries. This result matches that in L'Her et al. (2002) who find that global industry effects surpassed country effects in importance in 1999–2000. Since portfolio managers have traditionally followed a top-down approach, first choosing countries in which to invest and then selecting the best securities in each market, the question if the rise in industry relative to country effects is permanent or not is of great importance. Should they change the way the make portfolio decisions, or is the recent rise in sector effects a temporary phenomenon linked to the stock market bubble? We find that, beyond the TMT sectors at the heart of the recent stock market bubble, there is no evidence that industry effects have significantly outgrown country factors in importance. We also study if the recent rise in the importance of TMT industry effects is due to the fact that these firms are more international than their “old economy” counterparts. Our results do not, however, support this hypothesis. We find that the importance of global industry effects rises faster for non-traded than for traded goods sectors towards the end of the sample, something we find hard to reconcile with the notion that the recent rise in sector effects is driven by greater international integration at the firm level. In addition, there is little difference towards the end of the sample in the relative importance of industry effects across traded goods sectors (excluding IT) and non-traded goods sectors. Using firm-level information on the degree of international exposure of individual companies (international sales or assets), we find that the rise in industry relative to country effects since the mid-1990s has been more pronounced for less international firms. Again, this evidence goes against the hypothesis that the recent rise in industry effects is driven by global integration at the firm level. Finally, we use the international sales and asset data to construct international diversification effects and find that they explain almost none of the recent rise in industry effects. Taken together, we see this evidence as suggestive that the recent rise in sector effects is a temporary phenomenon associated with the TMT bubble. For portfolio managers this suggests that in the aftermath of the IT bubble the “old” strategy of diversifying across countries rather than industries may still have merit in terms of reducing portfolio risk.