Style Management in Equity Country Allocation

Strategies that entailed country selection based on relative strength (momentum) posted significant market risk–adjusted returns over the past 30 years, but relative-value strategies based on book value of equity to market value of equity did not. Because these two fixed-style strategies are negatively correlated, using them for style diversification and for style timing (rotation) is potentially rewarding. In the study described here, style diversification enhanced return and lowered risk but style timing provided consistent risk-adjusted performance that was superior to the performance of fixed-style strategies or style diversification. Many practitioners exhibit a consistent and strong value or momentum bias. This phenomenon is true irrespective of whether the point of view is security selection, sector selection, or country allocation. Both a value strategy and a momentum strategy can provide good average performance over a long period of time (in fact, they have done so over the past 30 years), but the strategies suffer from relatively long, nonoverlapping periods of underperformance. These periods can prove fatal for many portfolio managers.The correlation between value and momentum strategies is negative and fairly strong, so diversification (combining the uncorrelated strategies to reduce the risk involved in picking only one) and style timing (rotating between the two strategies) are two approaches that come to mind as useful for portfolio managers trying to avoid the fatal periods. A timing strategy is based on a criterion supported by psychological evidence that prior gains and losses affect subsequent risk-taking behavior. Timing is an attempt to switch portfolio exposure toward the potentially more rewarding strategy. A combination of the diversification and timing strategies can be thought of as “style management.” It would allow managers to maintain an exposure to both styles while alternately tilting toward the potentially more rewarding one.We studied the implementation and performance of these three strategies on a global basis through the use of country indexes. The countries included in the study are the 18 largest markets in the world; the sample period is January 1975 through August 2003, for a total of 344 months. The relative-value signal was based on country price-to-book ratios, and the relative-strength signal was based on country one-year previous prices.Compared with the fixed-style strategies, style diversification, based on an equally weighted combination of the relative-value signal and the relative-strength signal, enhanced return and lowered risk in the period studied. The timing strategy, however, provided consistent risk-adjusted performance superior to the performance of the diversification strategy and of the fixed-style strategies alone. In turn, the style management strategy posted even better risk-adjusted returns than the fixed-style strategies, the diversification strategy, and the timing strategy.In robustness tests, we found that transaction costs do not appear to undermine the statistical significance of reported results. We also examined how exposure to risk factors would alter the performance of the three strategies. To do so, we used a four-factor pricing model, with the risk factors being the equally weighted world market benchmark excess return over the U.S. risk-free rate, a size (liquidity) factor, the relative-value factor, and the relative-strength factor. After controlling for these four risk factors, we found that both the timing and management strategies can provide positive and significant risk-adjusted returns.In further tests, we measured how good each model was at selecting good countries (rather than timing the right style at a given time) and how good it was at using the right style at the right time. The performance attribution was carried out through the addition of a squared term for each previously mentioned factor in the model. As expected, we found that the abnormal performance from the timing model derives exclusively from its timing ability whereas the diversification model provides significant selection ability. The management model provides balance in terms of timing and selection.The models in this study seem promising for implementing stylediversification and timing strategies in country allocation. Further research could examine sector allocation as well as other conditional criteria. Managing the style bias in country allocation appears to be an important source of alpha for global portfolio managers. Furthermore, our study somewhat confirms the assumption advanced by behavioralists that after prior losses, people tend to adopt a more defensive (value-oriented) decision-making process whereas after prior gains, they tend to be less risk averse and open to the influence of momentum.