Optimal currency hedging : horizon matters

Investors have long debated what fraction, if any, of their portfolio's currency exposure they should hedge. Although the answers cover a broad range, often with dubious rationale, most informed investors agree that the solution should be based on mean-variance optimization, deployed either to maximize expected utility for cases in which the investor has non-zero expectations for the mean currency returns, or to minimize risk when the means are assumed to equal 0. This approach presents a serious challenge, however, because it depends on how currencies co-vary with each other and with the underlying portfolio, and these covariances themselves vary significantly with the return interval used to estimate them. The authors show that monthly covariances produce unreliable results for horizons that are longer than one month