Residual Income Approach to Equity Country Selection

The predictive power for country selection of expected returns estimated through the residual income model is examined through analysis of 19 developed-country indices for 1988–2005. Zero-investment strategies based on a ranking or optimization methodology—expected returns and conditional country risk estimates—posted significant positive performance over various holding periods. Risk-adjusted returns remained significant after control for four world risk factors—market, size, the book-to-market ratio, and momentum—constructed through a country stratification methodology based on stock constituents. The results were robust to various long-term growth estimates and to different country-universe subsamples and remained robust after transaction costs were taken into account.Ample evidence exists that country versions of company characteristics have a certain degree of predictive power for future country returns. Although these variables do allow ranking strategies, however, they do not allow optimization strategies. We favor the use of expected returns over the use of these variables for selecting countries in a global equity strategy.We examined the suitability of using expected returns derived from the residual income model (RIM) in global equity country selection. This model has the advantage of relying on analysts’ forward-looking information, and it allowed us to translate accounting numbers computed within different country-specific accounting standards into a harmonious measure of expected return. The model provides a reliable and consistent measure of the implicit expected rates of return among countries.According to the RIM, the intrinsic value of the market index is equal to the sum of the book value of the market index plus the present value of abnormal earnings. Abnormal earnings are equal to the difference between forecast earnings and normal earnings, where normal earnings represent the charge for the cost of capital. We inferred the implicit expected return (discount rate) from the equality between the intrinsic value and the market price of the index.Our sample comprised 19 developed countries and spanned the period 1988–2005. We first created zero-investment portfolios by following a ranking strategy in which the long portfolio contained the most attractive countries with respect to expected returns obtained via the RIM and the short portfolio contained the least attractive countries. We then examined an optimization strategy based on conditional risk estimates in addition to the implicit expected return estimates.Both strategies posted significantly positive excess returns over the period. The ranking strategy provided an annual average raw return of 8.3 percent with an annualized standard deviation of 8.6 percent over one-month holding periods. The optimization strategy provided a slightly lower annual average raw return (8.1 percent) but also a lower annualized standard deviation (7.2 percent). Although a one-month horizon produced the best results, three-, six-, and twelve-month holding periods all produced significantly positive returns.The strategies’ results held when we controlled for four world risk factors—market, size, the book-to-market ratio, and momentum. The alpha coefficients dropped only slightly from those of the raw returns.In further tests, we assessed the robustness of the strategies to various parameters. First, we tested other proxies for the estimate of the abnormal long-term growth rate in the RIM. In our original calculations, we assumed that abnormal earnings grew at the long-term expected inflation rate as measured by the historical inflation rate observed over the previous year. In the robustness tests, we used short-term interest rates and consensus inflation forecasts as proxies for the long-term growth rate of earnings. Returns from the strategies remained positive and significant for both these proxies.We also showed the robustness of the strategies to various universe subsamples—the developed countries ex the United States, the subsample composed of the MSCI Europe/Australasia/Far East Index ex Japan, and the 12 largest and most liquid markets.Finally, we tested the results’ robustness to transaction costs. Conservative transaction costs did not undermine the statistical and economical significance of the results.Considering our results, the use of the residual income model for country allocation seems promising.We documented basic zero-investment strategies for applying the model, but there are many ways of implementing such strategies. For example, an investor could tilt global equity portfolios toward countries with higher expected returns and avoid countries with lower expected returns. All variants of the basic zero-investment strategies would add value over a passive benchmark.