Optimal Close-to-Home Biases in Asset Allocation

Abstract

This article studies optimal portfolio decisions with (long-term) liabilities for small-open economy based investors, including the optimality of currency hedging (Walker (2008a)). The representative investor is based in Chile, but results are likely to hold more generally. The problem is set up as in Sharpe and Tint (1990) and Hoevenaars, Molenaar, Schotman and Steenkamp (2007). Hedging the liabilities and the consumption currency may imply optimal close-to-home biases, defined as overweighting asset classes which are highly correlated with local ones. The implementation challenges include: developing a methodology to estimate expected returns in local (real) currency; estimating the covariance matrix allowing for serial and crossed-serial correlations; and checking the results’ robustness using a resampling methodology. The findings are: i) portfolios always have optimal close-to-home biases, beyond the investment in local fixed income to hedge liabilities; ii) currency hedging reduces investment in close-to-home asset classes, iii) but has ambiguous effects on welfare – which is detected with the resampling methodology; iv) currency hedged long-term US bonds are useful for hedging local interest rate risk; and v) liabilities give access to high risk-return portfolios, not affecting otherwise the overall shape of the efficient regions. This article should be useful to investors based in small open economies, including pension funds, insurance companies, sovereign wealth funds and Central Banks.