<p class="MsoNormal" style="text-align: justify; margin: 0in 0.5in 0pt; mso-pagination: none;"><span style="font-family: Times New Roman;"><span style="font-size: 10pt; mso-bidi-font-weight: bold; mso-bidi-font-style: italic; mso-bidi-font-size: 12.0pt;">This paper examines the historical predictive power of future spot spread in estimating currency changes.<span style="mso-spacerun: yes;"> </span></span><span style="font-size: 10pt; mso-bidi-font-style: italic; mso-bidi-font-size: 9.5pt;">Currency futures and spot rates over the last two decades are examined.<span style="mso-spacerun: yes;"> </span>Results show that as forecast horizon of currency depreciation increases, the slope coefficients become less positive, first losing their significance, and eventually for 1-month regressions, becoming negative for the British pound, Swiss franc and Japanese yen (significantly negative for the yen) indicating risk premiums differ with forecast horizon.<span style="mso-spacerun: yes;"> </span>On the other hand, expectations hypothesis is validated when the forecast horizon is 1 day.<span style="mso-spacerun: yes;"> </span>These results hold for each decade separately, as well as the total sample.<span style="mso-spacerun: yes;"> </span>Comparison of early (1980s) and recent (1990s) periods reveals expectations hypothesis is validated in the recent period.<span style="mso-spacerun: yes;"> </span>This indicates the trend towards a more efficient market.<span style="mso-spacerun: yes;"> </span>This should not be very surprising with the introduction of round the clock electronic trading medium and reduction of transaction fees in futures markets.<span style="mso-spacerun: yes;"> </span>This also implies that the absolute value of the risk premium has decreased over the last two decades.<span style="mso-spacerun: yes;"> </span>The extreme case of forward premium puzzle in one-month forecasts diminishes in the 1990s.<span style="mso-spacerun: yes;"> </span>The results are robust to partitioning the sample period into four sub samples and separating the data based on maturity of futures contracts.</span></span><strong><span style="font-size: 10pt; mso-bidi-font-style: italic; mso-bidi-font-size: 12.0pt;"></span></strong></p>