International Diversification and Return Predictability: Optimal Dynamic Asset Allocation

Abstract

There is growing evidence that global economic indicators are capable of predicting international stock returns, and hence the use of such indicators should provide economic benefits in international asset allocation. The goal of this paper is to construct dynamically efficient strategies that optimally utilize this predictability, and study their performance. We find that there are significant in-sample gains from international diversification for a US investor utilizing term spread and the lagged world index as predictive instruments. More importantly, we show that these gains can be realized out-of-sample. We focus on two out-of-sample periods, capturing the emergence and subsequent collapse of the 'dot.com' bubble. Our dynamically optimal strategies beat the benchmark by a wide margin, and avoid any losses during the period where all country indices had negative returns. Unlike the commonly used myopically efficient strategies, our dynamically optimal strategies do not require extreme long and/or short positions and also incur considerably lower transaction costs. Our strategies thus could have been feasibly implemented by long-only portfolio managers.