The forward discount puzzle and market efficiency

We argue that the forward discount puzzle is primarily a statistical phenomenon and that statistical rejections of Uncovered Interest Parity do not necessarily constitute valid rejections of market efficiency. We find by using a Taylor expansion a theoretical negative bias in existing regressions of UIP. We propose two alternative tests for market efficiency, one of which is designed to measure the degree of market inefficiency. Our results from these tests indicate that for all four of the bilateral dollar parities studied the foreign exchange market is efficient despite decisive clear rejections of UIP using the conventional regression approach.