Forecasting World Stock Returns and Improved Asset Allocation

There is little evidence in the literature on whether predictability of stock returns leads to improved asset allocation and performance (Handa and Tiwari 2006). Handa and Tiwari (2006) found fixed results for forecasting 1-month-ahead results in the USA for 1954–2002 period; the past-returns model worked well from 1974 to 1988 and poorly from 1959 to 1973 and 1989 to 2002. There are mixed academic results for many financial tests. In this report, we show that it is possible to improve performance of a naive “60/40” model of equity and debt to a “60/40” model with Global Timing (GT). We create a Global Timing signal based on the 12-month moving average of the differential between the LIBOR rate and the All World Country (ACW) index. If the predicted return signal, the differential of the 12-month average returns on the ACW, exceeds LIBOR by a statistically significant difference (one standard deviation), then a “buy” signal is created. If the predicted return signal is less than −1.645, one standard deviation, then a “sell” decision is made. A neutral position exists in the signal and no change is made.