تنوع پرتفوی بین المللی:بازبینی اثرات نرخ ارز، صنعت و کشور

We examine the relative importance of country, industry, world market and currency risk factors for international stock returns. Our approach focuses on testing the mean-variance efficiency of the various factor portfolios. An unconditional analysis does not show significant differences between country, industry and world portfolios, nor any role for currency risk factors. However, when we allow expected returns, volatilities and correlations to vary over time, we find that equity returns are mainly driven by global industry and currency risk factors. We propose a novel test to evaluate the relative benefits of alternative investment strategies and find that including currencies is critical to take full advantage of the diversification benefits afforded by international markets.

One of the core questions in international finance is which factors drive international equity returns. A substantial amount of research focuses on the comparison of country versus industry factors. This was first investigated by Lessard (1974). Renewed attention to the issue and a voluminous literature was sparked by the work of Roll, 1992 and Heston and Rouwenhorst, 1994 and Griffin and Karolyi (1998). Today, more than three decades later, the country-industry debate is still ongoing.1 A closely related issue is financial market integration. In fully integrated markets, only global risks are priced (e.g., Solnik, 1974a, Sercu, 1980 and Adler and Dumas, 1983) while in segmented markets only local risks are priced.2 Several recent papers provide evidence that national equity markets are becoming more integrated within the world market (e.g., De Jong and De Roon, 2005, Carrieri et al., 2007 and Pukthuanthong and Roll, 2009). This suggests that international equity returns are increasingly driven by global rather than by local factors. Further, currency risk factors could play an important role for global stock returns. Several international asset pricing models show that, in equilibrium, when currency risk induces deviations from purchasing power parity, investors require to be compensated for bearing exchange rate risk (e.g., Stulz, 1981 and Dumas and Solnik, 1995). De Santis and Gerard (1998) and Lustig and Verdelhan (2007) report empirical evidence of a premium for currency risk.3 This paper investigates to what extent international stock returns are driven by country, industry and currency risk factors. In addition, we consider as benchmark models the world Capital Asset Pricing Model that includes the world market return and the International CAPM that includes several currency returns as well. A common approach for comparing country and industry effects is based on a factor model with country and industry dummy variables (Heston and Rouwenhorst, 1994 and Griffin and Karolyi, 1998). This model assumes a unit exposure to the global market shock for all assets, which may lead to biases in comparing country and industry factors (Baele and Inghelbrecht, 2009 and Bekaert et al., 2009). We propose an alternative approach and a novel test to provide new insights into the role of currency, country and industry factors in driving equity returns in the seven largest world economies over the February 1975 to June 2011 period. We proceed in two steps. First, we conduct spanning tests to investigate whether a subset of the portfolios or factors under consideration is mean-variance efficient when tested against the remainder of the assets. If for instance, the Adler and Dumas (1983) version of the international CAPM is the appropriate model to describe equilibrium returns in international markets, the world and currency factor portfolios would span all other asset returns. In a second step we propose a new test to investigate the relative benefits of alternative international diversification strategies based on country, global industry and currency factor portfolios. We perform all our tests both unconditionally and conditionally, as well as both unconstrained and with reasonable limits on short sales. Our unconditional tests do not detect significant differences between country, industry and world portfolios. Spanning is only marginally rejected for the ICAPM portfolios with respect to global industry returns. Moreover, the unconditional analysis provides no evidence that currency risk factors matter; currency returns are spanned by country, industry and world returns. In contrast, when we allow for time-varying means and (co)variances using returns on managed portfolios, we can identify which factors dominate. Our conditional results show that international equity returns are primarily driven by industry and currency risk factors. While country returns are spanned by global industry returns, industry returns are not spanned by country returns. This corresponds to a tangency portfolio that is linear in the returns on industry portfolios only. Furthermore, the world market portfolio is efficient with respect to country returns, but not with respect to global industry returns. The dominance of global industry factors over country factors is in line with the seven developed countries in our sample being among the most integrated equity markets in the world. Finally, we find that currency returns significantly improve the mean-variance efficiency of country, industry and world market portfolio returns. These findings have important implications for cross-border investment strategies. Using our new test, we find that passive country- and industry-based equity strategies deliver indistinguishable Sharpe ratios. When portfolios are rebalanced every month, industry-based strategies lead to significantly higher Sharpe ratios. However, this outperformance critically depends on the ability to take short positions. Further, while including currencies in passive strategies does not deliver significant benefits, in actively rebalanced portfolios currencies are essential for achieving optimal portfolio performance. Strikingly, adding currency deposits to long-only managed country or industry portfolios doubles their annual Sharpe ratios. Cross-border investors take implicit currency positions and investing in currency deposits may hedge against currency risk (e.g., Campbell et al., 2010). At the same time, currency deposits can also be attractive investments themselves, due to their non-zero expected returns.4 We allow for both hedging and speculative benefits by jointly optimizing over international equities and currency deposits. We perform a battery of robustness tests for the comparison of country and industry factors. We include three additional developed markets leading to the same number of countries as industries, we exclude the IT industries, and we change the sample by excluding each country and industry one by one. Our spanning test results are robust: unconditional tests show no significant difference, while conditional tests confirm the dominance of global industry factors. However, while unrestricted dynamic industry-based strategies still lead to slightly higher Sharpe ratios than country-based strategies, the difference is smaller and loses statistical significance when excluding an industry or when adding three additional countries. In addition, we perform our analysis in local currency denominated country and industry returns. Currency risk may confound the comparison of country and industry factors, as a country return only depends on one currency, while a global industry return depends on a basket of currencies. We find that our results on the relative importance of country and industry effects are robust when using local currency returns. However, this does not imply that currency risk is irrelevant. As discussed above, when including separate currency risk factors in our conditional analysis, we find that they have an important impact on international equity returns. Finally, we examine whether the gains from active portfolio strategies compared to passive investment strategies are sustainable given transaction costs. We find that the level of transaction costs that would eliminate any gains from dynamic strategies is very high (on average 145 basis points per month for a one-way trip), which suggests that the gains from dynamic strategies are, at least to some extent, practically feasible. The paper proceeds as follows. Section 2 develops the empirical framework, introduces our new test and shows the link with spanning tests. Section 3 describes the data. Section 4 reports our unconditional results, while Section 5 discusses the conditional analysis. Section 6 discusses a number of robustness tests and Section 7 concludes. Appendix A shows the derivation of our Sharpe ratio test and Appendix B discusses the size and power of the test.

Although the benefits of international diversification arising from the relatively low level of correlations among national equity markets are now well documented (e.g., Solnik, 1974b, Elton and Gruber, 1992 and De Santis and Gerard, 1997), the issue of which factors drive international equity returns remains controversial. This paper proposes a different approach and a new test to provide fresh insights into the country-industry debate that has been ongoing in the international finance literature for over three decades. First, we test which set of factor portfolios is mean-variance efficient and spans the other portfolios: countries, industries or world market and currency factors. Second, we rigorously examine and propose a new test to evaluate the efficiency gains and diversification benefits of different international portfolio strategies. We focus on the seven largest world economies between 1975 and 2011. Our unconditional analysis does not allow us to infer whether country or industry factors dominate: country returns span industry returns and vice versa. Currency risk factors do not play an important role here. However, when we allow for predictable time variation in expected returns, volatilities and correlations using conditioning information, we can identify which factors are most important. Our conditional analysis shows that international equity returns are mainly driven by industry and currency risk factors. The dominance of global industry factors over country factors is in correspondence with a relatively high degree of market integration. In a second stage, we analyze the implications of these findings for international diversification strategies. We find that actively rebalanced industry-based strategies outperform country-based strategies, but only if short selling is allowed. With short sales constraints, or when using passive portfolio strategies, country- and industry-based diversification strategies deliver similar Sharpe ratios. Finally, we find that active currency investing adds substantial diversification and return enhancement benefits to cross-border portfolios. Whether this outperformance is due to the hedging or speculative benefits of currency positions is left for future research.