The accelerating globalisation of business enterprise and the integration of capital markets has fundamentally altered the equity investment environment. The relative importance of the fundamental determinants of international equity security prices has changed. In a recent study, we found convincing evidence that the economic forces that drive the performance of the industry to which a firm belongs have become more important than those associated with the firm’s home country.*
In this article we use the risk model developed in our study to demonstrate that industry allocations are as important as country allocations in determining the risk and reward opportunities in global portfolios.
We believe this “industrial evolution” is more than a mere transient feature of the past several years’ infatuation with technology stocks. The decline in trade barriers resulting from the growing importance of the World Trade Organisation (WTO), the emergence of large trading blocks (the EC, Nafta and Asean) and increased economic policy coordination (in particular for Emu member countries) all point towards an irreversible pattern of economic integration.
This is well reflected in the declining dispersion in economic and political risk indicators across developed market countries. For instance in 1984, the difference in short-term interest rates across developed market countries was as high as 17%; by 1999, this difference had nearly halved. Over this same period, differences in various indicators of political risk as reported by the Political Risk Services Group have also markedly declined.
Today, whether within traditional industries such as oil and automobiles, or within growth industries such as telecommunication equipment and enterprise software, leading companies operate across national and geographic boundaries in product, capital and labour markets that are becoming increasingly global. Even traditionally local industries such as retailing and electric and water utilities are competing globally. Consolidation and rationalisation of enterprise activities is well reflected in the explosion of cross-border mergers and acquisitions as documented in Table 1. The increasing importance of cross-border mergers within the same industry suggests that the productive capital stock of industries is becoming more global. Thus, differences across company valuations are increasingly reflecting differing expectations of future earnings across global industries.
Our analysis examines returns of individual securities in the 21 countries of the MSCI World Developed Markets universe from January 1986– August 2000. We use the 36 Financial Times industry level national total return indices to measure the performance of portfolios of securities belonging to the same industry within a country.
We postulate that each security can be viewed as a portfolio of factor exposures: the world index, the country of domicile and the global industry to which the security belongs. Each security is assumed to have full membership in one country and one industry. Each country factor is constructed so that it has the same industry composition as the world index. Similarly, each industry factor has the same country composition as the world index. This construction enables us to control for differing industrial structures across countries and differing country compositions across industries. Thus, for instance, we are able to identify the true or “pure” factor returns to an investment in the Swiss market, controlling for the fact that it has a larger exposure to the financial and pharmaceutical industries than the rest of the world. Similarly, we are able to identify the true or “pure” factor returns to an investment in the software industry controlling for the fact that software manufacturers are largely domiciled in the US. The “pure” factor returns can then be used to examine the risk and reward properties of alternative baskets of securities: countries and industries.
Our model estimates enable us to measure the changing importance of industry and country factors in global sector funds. Consider a world information technology fund; it can be viewed as a portfolio of exposures to the world information technology industry and to a basket of country factors. In Table 2 we report the components of the volatility for selected global sectors in the world index over two periods – the past five years and the previous 10. The table shows the industry factor volatility, controlling for country composition, and the standard market cap weighted industry volatility. The ratio indicates the portion of an industry’s volatility arising from pure industry effects, without any country effects. For instance, from 1995–2000, industry effects accounted for less than half of the volatility of the real estate sector and similarly accounted for 89% of the volatility of the information technology industry. A comparison of the two periods suggests that industry effects have risen in importance relative to country effects.
Our analysis enables us to draw inferences about the relative merits of international diversification. During the most recent decade we find that the correlation of country factor returns has risen as countries’ economic performances have become increasingly coordinated; gains from diversifying across countries are diminishing. Conversely, industry factor correlations have been relatively stable over the recent decade; gains from diversifying across industries are stable. Clearly however, investors would be well served by diversifying both across industries and across countries. Figure 1 provides some supporting evidence – the average correlation of country and industry factor returns over time.
The benefits from diversifying across industries can be further illustrated by comparing the volatility of domestic country indexes to the volatility of portfolios that optimally combine various industry portfolios with each domestic country index so as to minimise risk. Table 3 reports our results for a selection of countries.
Consider the US market. Our model estimates for 1995–2000 suggest that the volatility of the US market was 14%; if one had optimally combined global industry portfolios with the US market index in order to minimise risk, the resulting portfolio would have had a volatility of 10%. The diversification benefits from global industries are however particularly important for investors domiciled in countries characterised by a highly concentrated industrial structure. Consider for instance a Finnish investor. From 1995–2000 the volatility of the Finnish market (dominated by telecoms) was 31%, but could have been reduced to 10% by global industry diversification (adding non-telecom industries to the domestic portfolio). Similar results hold for the Swiss market and the Dutch market. Combining the other country portfolios to these domestic country indexes would have provided substantial diversification and risk reduction. However, the evolving risk structure of global equity markets affords an additional margin of benefit to industry diversification.
In view of this industrial evolution of the equity market, plan sponsors may wish to reconsider the traditional separation of equity portfolios into domestic and foreign components. Today, a policy allocation to global equity may lead to more efficiently diversified portfolios than separate allocations to domestic and to foreign equities. The potential inefficiency of separate domestic and foreign allocations is most obvious when considering the unintended industry exposures that result from equity benchmarks that are biased towards an investor’s small home market. A Swiss home-biased benchmark produces a significant allocation to the pharmaceuticals industry, which represents more than 30% of the Swiss market but only 8% of the global market. Why should a Swiss investor maintain a large exposure to pharmaceuticals?
We find evidence suggesting that the riskiness of industry factors is relatively stable. It may thus be possible for an investor to gain exposure to world equities in a manner that minimises his total risk. Using our country and industry factor volatilities up to December 1998, we constructed a portfolio aimed at minimising risk; country and global industry weights were allowed to deviate from the world benchmark by 10%. The volatility of this portfolio over the 1999–2000 period was 10.6%. A similar lowest risk portfolio that was neutral on country positions relative to the world benchmark would have registered a slightly higher volatility of 11.9%. The world index registered a markedly higher volatility of 15.6%! This would suggest that there might be scope for innovation in benchmark construction. Our analysis suggests that when one considers global equities it is critical to understand the industry composition.
For many years asset managers have conducted their research and decision-making processes based on the well-documented dominance of country factors. Thus, a two step process was adopted. Country allocations were first obtained on the basis of the relative attractiveness of markets. Subsequently, local research teams managed security selection within each country. Our findings, which reflect the recent globalisation of enterprise activities along industry lines, suggest that industry allocations are as important as country allocations in determining the risk and reward opportunities of global portfolios. This simultaneous industry and country factor approach provides a better framework for incorporating bottom up security specific insights.
*Cavaglia, Stefano, Chris Brightman and Michael Aked, (2000) “On the Increasing Importance of Industry Factors: Implications for Global Portfolio Management”, Financial Analyst Journal.
Stefano Cavaglia is head of equity strategies, Thomas Madsen is head of global equities and Brian Singer is head of risk management at Brinson Partners in Chicago