Despite recent reports to the contrary, we find that even recently-the 1991-2000 period-the country factor still dominates industry influences. This conclusion is robust to different test formats although the relative magnitude of the two sources of variation changes widely. One factor affecting the degree of country-factor dominance is the presence or absence of smallcap stocks in the sample: small-caps have an above average variability (after controlling for industry and country effects) and are also less sensitive to their global industry index than large-caps. Another factor that matters is the country coverage (especially the presence of emerging markets) and the level of industry aggregation (NACE 3 versus 4, for example). Methodology matters too. Heston and Rouwenhorst (1994) rank the world, country, and industry factors on the basis of their own variance, but this ranking may miss the ranking on the basis of stock-return variance explained if exposures are dissimilarly distributed across factors. Finding that the assumption of similar exposures is, in general, not realistic, we incorporate the distributions of the exposures into the assessment of the relative importance of country v industry factors, taking care to purge out the variability due to estimation error. By this metric, the dominance of the country factor becomes unassailable.
展开▼
