EITM: Empirical and Econometric Research on Exchange Rates and Present Value Models

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

The proposal consists of two sections of roughly equal magnitude. Both sections consider linear models, focusing on present value models, and with special reference to exchange rates. It is well known that if a series y is the expected present discounted value of a series x, then y is a random walk if x is a random walk. Much of the research develops empirical and econometric implications of a new result: if the discount factor used in discounting future x's is very near 1, then in a precise sense y will behave very much like a random walk as long as x follows any I(1) process. The first section describes empirical research on exchange rates. One aim of this research is to evaluate whether the generally accepted random walk in exchange rates is substantially attributable to a discount factor that is near 1. This will be accomplished in part by studying a wide range of linear exchange rate models, in part by specifically comparing the "large discount factor" explanation with one that looks instead to a random walk in observable or unobservable fundamentals. The research will also study the economic determinants of exchange rates, under assumptions that do not necessarily entail a discount factor near 1, and in environments in which the exchange rate is not expressed as a present value. The second section concerns theoretical econometric and empirical work. Generalizations and extensions of the implications of the large discount factor will be considered. An alternative asymptotic approximation that turns on the discount factor being near 1 will be developed. This theory has promise to deliver insight into the behavior of asset prices. The theory will be used in development, evaluation and application of tests for random walks, including autocorrelations and long-horizon cross-correlations. It will be applied as well to tests of forward rate unbiasedness. The econometric research will also develop asymptotically chi-squared statistics for testing the random walk hypothesis, under assumptions that do not necessarily entail a discount factor near 1. Broader impacts resulting from the research Understanding the behavior of exchange rates, and, more generally, asset prices, is central to our understanding of the economy. The research has direct bearing on the answers to questions such as: Should we look to asset prices as forecasters of the real economy? What are the sources of movements in exchange rates and other asset prices? How do exchange rates and other asset prices respond to policy-induced movements in interest rates? What constitutes a good policy response to fluctuations in exchange rates and other asset prices?