Seo-96 FIGURE 11.8 Beta of Currency against Hedged Assets (G-7 Portfolio, U.S. Perspective) both positive and negative (and, on occasion, extremely so). Second, while the beta coefficients vary depending on the time, the central tendency seems to be zero. To explore this issue further, we broke the sample into several discrete pieces and calculated the beta coefficients and correlations for each segment. Table 11.4 illustrates that over long periods of time (e.g., 10 or 20 years), the beta coefficients and correlation levels are close to zero. Over shorter time intervals, however, both elements can deviate dramatically from zero. The data in Table 11.4 have two important implications for portfolio strategy. First, given that the correlation figures are quite low over long time horizons, little is lost by assuming, for strategic asset allocation purposes, that the correlation between currency and asset excess returns is zero. Second, the observation that correlations change sign, and for prolonged periods, can be viewed as providing a rationale for active currency management. How does the assumption that the long-term correlation between currency and asset excess returns is zero affect implied currency returns? Figure 11.9 plots the relationship between the implied currency return and the currency hedge ratio. Similar to the results shown in Figure 11.7, the implied return to currency decreases as the percentage of currency hedged increases. As currency's marginal contribution to risk decreases, the implied return associated with open currency positions also decreases (under the assumption of optimality). In contrast with TABLE 11.4 Beta Coefficients by Period   Period Beta Correlation T-Statistic* Past 10 years -0.02 -0.05 -2.52 Past 20 years -0.01 -0.03 -1.87 1/1/78-12/31/85 0.05 0.16 7.00 1/1/86-12/31/88 -0.06 -0.16 ^.43 1/1/89-12/31/93 0.05 0.10 3.65 1/1/94-12/31/97 -0.09 -0.24 -8.03 *The t-statistic is a measure of statistical significance.