Dynamic currency hedging for international stock portfolios

© 2012 Dr. Wei ZhangAs world financial markets become increasingly integrated and cross-border equity investment continues to grow, the issue of currency exposure takes on increasing importance. Research in the area of currency risk management is likely to benefit a broad range of market participants such as fund managers, multinational corporations and governments (in managing their foreign reserves). The need to install proper currency risk management schemes for foreign investments can also arise from the large exchange rate fluctuations we have evidenced for numerous currencies since the start of the floating-exchange-rate regime, especially during crises times such as the Asian financial crisis in 1997 and the recent global financial crisis of 2008. The literature in currency hedging lacks consensus about how currency risk should be managed for an internationally diversified stock portfolio. The purpose and contribution of this thesis is to address a number of important gaps in the existing literature by examining the following three research questions: i) can conditional hedging outperform static hedging and unconditional hedging on an ex-ante basis? ii) can joint optimization outperform the overlay strategy out-of-sample when portfolio construction and currency hedging are based on conditional estimates of the mean, variance and covariance? and iii) will the performance of a conditional hedging strategy improve when the international asset portfolio is diversified over more currencies? The thesis presents empirical tests of a number of competing currency hedging strategies under two hedging objectives: portfolio risk-minimization and mean-variance optimization both in sample and out of sample. Our focus is on investigating the performance of a dynamic (conditional) hedging strategy relative to an unconditional hedging strategy and three static strategies namely no hedge, half hedge and full hedge. A dynamic conditional correlation multivariate GARCH model is employed to estimate time-varying covariance among stock market returns and currency returns. The conditional covariance is then used in the estimation of conditional hedge ratios. The study considers 7 developed economies over the period January 2002 to April 2010, and estimates daily conditional hedge ratios for portfolios of various stock market combinations from both US and Australian investors’ perspective. The in-sample period is January 2002-December 2005, and the out-of-sample period is January 2006-April 2010. For the purpose of risk management, hedging is implemented via a currency overlay, i.e. hedging positions are determined for a given international stock portfolio. Hedging result comparisons with those of traditional static hedging and unconditional hedging show significant improvement for conditional hedging in terms of risk reduction within sample. For a US investor, diversification benefits, in terms of risk reduction, tend to be maximized when about 50%-60% of the portfolio is allocated to foreign equities. Conditional hedging proves to consistently reduce portfolio risk for various levels of foreign investments. Out of sample, conditional hedging continues to dominate other hedging strategies in terms of risk reduction. The result is especially strong during the recent global financial crisis when hedging is needed the most. The dominance of conditional hedging over other strategies is in many cases statistically significant. Also, the Australian dollar and Canadian dollar are shown to be effective at cross hedging the portfolio containing all seven countries both in- and out-of-sample. Campbell et al. (2010) suggest that currency hedging demands for a given global portfolio are the same regardless of the currency base. We show that the same hedging strategies implemented from an Australian perspective (rather than a US perspective) generate comparable results within sample. However, out of sample, all hedging strategies except for the conditional hedging strategy add risk for an Australian investor, despite the hedged portfolio risk being similar to that for a US investor under each hedging strategy. This is because the unhedged portfolio risk from the Australian perspective is much smaller than that from the US perspective during the out-of-sample period. The result highlights the impact of base currency on hedging performance. For the purpose of mean-variance optimization, we compare jointly optimized portfolios (in which conditional asset positions and currency positions are determined simultaneously) with currency overlay portfolios (in which conditional currency positions are optimized given the asset position) as well as portfolios that are unhedged, half hedged and fully hedged. The result shows that the joint optimization dominates the overlay strategy which in turn dominates the static hedges within sample. However, no strategy consistently outperforms other strategies out of sample. In conclusion, for a risk-minimizing investor, the conditional hedging strategy dominates the unconditional hedging strategy and the static hedging strategies both in and out of sample. The dominance is also found for a mean-variance investor within sample, but the result cannot be extended out of sample. Moreover, we show that AUD and CAD are effective at cross hedging a portfolio invested in the seven stock markets. Also, hedging effectiveness for the same stock portfolio varies across different base currencies