Peso Problems, Bubbles, and Risk in the Empirical Assessment of Exchange-Rate Behavior

Abstract

One of the most puzzling aspects of the post-1973 floating exchange rate system has been the apparently inefficient predictive performance of forward exchange rates. This paper explores some aspects of each of three leading explanations of forward-rate behavior. The paper first develops a simple rational-expectations model of the "peso problem" that generates some key empirical regularities of the foreign exchange market: seemingly predictable and conditionally heteroskedastic forward forecast errors, along with possible directional misprediction by the forward premium. The implications of bubbles for tests of forward-rate predictive efficiency are discussed next. It is argued that the existence of bubbles is extremely difficult (if not impossible) to establish empirically. Even though some types of bubble would distort standard tests on the relation between spot and forward exchange rates, it seems unlikely that there bubbles have been an important factor. Finally, the paper examines foreign-exchange asset pricing under risk aversion and suggests that a convincing account of forward-rate behavior should also help explain the results found in testing other asset-pricing theories, such as the expectations theory of the interest-rate term structure.