Had we made adjustments for country-specific tax, accounting, and regulatory differences, the performance of this strategy might be improved significantly. Predictability within Global Equity Markets Using Valuation Measures It is interesting to compare the predictability across country stock markets to the predictability in stocks within countries using the same exact measure. For every developed stock market, we create a within-country long/short portfolio using B/P for individual stocks, capitalization-weighted long positions in the highest third of companies, and capitalization-weighted short positions in the lowest third.13 This results in 23 within-country equity long/short portfolios. We then equal weight these country long-short portfolios to derive a global stock-selection value effect. Not surprisingly, the stock-selection effect also works within countries, with a much larger statistical significance than the country-selection effect. Table 25.1 reports these results. The average spread in return on stocks within countries is 15.8 percent, and the volatility of the global stock-selection portfolio is 10.4 percent, resulting in an IR of 1.52 and a probability value on the stock-selection effect of 0.00 percent. While the value effect within equity markets is large, it is not as large as we might expect given the higher number of assets used in the stock-selection strategy. In the country-selection long/short portfolios, our average opportunity set was 18 countries, whereas the breadth of opportunities is 8,352 in the stock-selection portfolios. The breadth in the stock-selection portfolios is 464 times larger! According to Grinold's Law, all else being equal, the IR on the stock-selection strategy should be larger by a factor of 22, which is the square root of the ratio of breadths in the two strategies. In fact, the ratio is only three. This means that what the country-selection strategy loses in breadth, it mostly makes up for in forecasting accuracy. Recently, Peter Hopkins and Hayes Miller substantiated this finding when reviewing the performance attribution on global equity managers: The managers in their data set experienced far more success in forecasting countries than in forecasting stocks.14 The diversification in country indexes probably drives the significantly higher forecast accuracy in country selection relative to stock selection. Whereas a forecasting model in an individual stock will be notoriously uncertain, the residual risks from individual stock models wash out in a broad portfolio, resulting in more precise forecasts. In addition, Grinold's (1989) breadth is technically the number of independent 12The information ratio multiplied by the square root of the number of years in the sample (n) is distributed student-^ with n degrees of freedom. We show p-values from a one-tailed test for the hypothesis that the mean return to the strategy is positive. 13Whereas an equal-weight portfolio of countries is quite feasible, and equal-weight portfolio of 7,500 global stocks is not. We use capitalization-weighted portfolios here to be more representative of the opportunity set. 14Hopkins, Peter J. B., and C. Hayes Miller, 2001, "Country, Sector, and Company Factors in Global Equity Portfolios," AIMR Research Foundation Monograph.