We use point-in-time accounting data to estimate monthly out-of-sample fair values of over 25,000 stocks from 36 countries with a novel methodology. A simple trading strategy based on deviations from fair value yields statistically and economically significant risk-adjusted returns in most regions, especially the Asia Pacific. Differences in the signal’s monthly alphas of 40-70 basis points between emerging and developed markets contrast with findings of prior research about the relative efficiency of these two market types. Globally, pre-transaction-cost alphas, which are unrelated to known anomalies, are positively related to trading costs, but exceed country-specific institutional trading costs. Thus, global equity markets are inefficient, but are relatively less efficient in counties with quantifiable market frictions, particularly trading costs, that deter arbitrageurs. Joint with Mark Grinblatt.