A casual glance through the Proceedings of past annual meetings of the American Economic Association reveals that in almost every year during the past twenty years, president-elects of the AEA have devoted at least one session to an examination of issues concerning the international monetary system. Prominent on the agenda has been the question of reform. How should the international monetary system be reformed so as to function more effectively? premise underlying this question is that the international monetary system has failed and that it must be reformed by an institutional change. In what follows I present some skeptical notes on both the verdict on the failure of the system and on some proposals for reform, especially the target-zones proposal. To set the stage, it is worth noting that one of the main sources of disenchantments with the present monetary system has been the unpredictability of exchange There has been nothing more confusing than reading through the ex post journalistic explanations offered for the day-to-day changes in the U.S. dollar. For example, over the past few years we were told that The dollar fell because the money supply grew faster than expected-thereby generating inflationary expectations, but on another occasion we were told that The dollar rose because the money supply grew faster than expectedthereby generating expectations that the Fed is likely to tighten up and raise interest rates. On another date we were told that The dollar fell since the budget deficit exceeded previous forecasts-thereby generating inflationary expectations on the belief that the Fed will have to monetize the deficit, but, on another occasion we were told that The dollar rose since the budget deficit exceeded previous forecasts-thereby generating expectations that government borrowing needs will drive up interest rates since the Fed is unlikely to give up its firm stance. On yet another day we were told that The dollar fell since oil prices fell-thereby hurting Mexico and other debtriddenoil-producing countries whose bad fortune may bring about the collapse of important U.S. banks, but, on another occasion we were told that The dollar rose since oil prices fell-thereby helping the debt-ridden oil-consuming countries whose improved fortune will help the vulnerable position of important U.S. banks. More recently the dollar changed again, and this time the explanation was a bit more sophisticated: The dollar changed because the extent of the revision of the estimated GNP growth rate was smaller than the expected revision of previous forecasts of these estimates. One cannot but sympathize with the difficulties shared by newspaper reporters and financial analysts who feel obligated to come up with daily explanations for daily fluctuations of exchange rates, and one can only imagine the deep frustration that yielded the recent headline in the International Herald Tribune according to which The dollar rose on no news. dismal performance of short-term forecasting does not reflect a lack of effort. Rather, it is an intrinsic characteristic of efficient asset markets. Difficulties in forecasting short-term indices of stock markets (like the Dow-Jones index) do not call however, for a reform of the way stock markets operate. For similar reasons one should not assess the performance of the international monetary system on the basis of short-term forecastability of exchange This does *International Monetary Fund, Washington, D.C. 20431, University of Chicago, and National Bureau of Economic Research. research reported here is part of the NBER's research programs in International Studies and Economic Fluctuations. Any opinions expressed are my own and not those of the National Bureau of Economic Research. 205

A major difficulty in incorporating the role of inflationary expectations in empirical work has been the lack of an observable variable measuring expectations. Thus, for example, in analyzing the demand for money during hyperinflation, Phillip Cagan in his classic contribution constructed a time-series of expected inflation using a specific transformation of the time-series of the actual rates of inflation. There are two conceptual difficulties with such an approach: first, the choice of the specific transformation used to generate the series of expectations is to a large extent arbitrary; and second it assumes that expectations about future prices are based only on past and present prices. Recent empirical work that was stimulated by Cagan's pioneering study elaborated on some aspects of the estimation procedures (see Thomas Sargent and Neil Wallace; Sargent 1977; Joseph Bisignano; Paul Evans; Rodney Jacobs 1975; Mohsin Khan 1975), and the functional form (see Robert Barro 1970; Benjamin Eden 1976). The ongrowing literature concerning expectations (for example, John Muth 1961; Robert E. Lucas) has led to an examination of the conditions under which the adaptive expectations process is rational in the sense of Muth (1961). See Sargent and Wallace (1973); Sargent 1977); Benjamin Friedman (1 975a); Michael Mussa. In this paper I propose a direct measure of expectations which is then incorporated in the analysis of the demand for money during the German hyperinflation. The major virtue of the proposed direct measure is that it is not derived from a specific mechanistic formula, but rather, it reflects the expectations of economic agents as manifested in market prices. The direct measure is based on data from the forward market for foreign exchange. The plan of the paper is as follows: Section I describes the direct measure of expectations and provides evidence on the efficiency of the foreign exchange market. Section II incorporates these expectations in estimating the demand for money. The issues that are discussed in that section involve the proper functional form, the proper price deflator, the stability of the demand for money during the various phases of the hyperinflation, possible lags of adjustment and the resultant estimates of short-run and longrun demand functions, and the role of price variability and uncertainty in the specification of the demand for money. Section III deals with the issue of inflationary finance and the money supply process. In this context I examine the interrelationships between money and prices and discuss some aspects of causality by analyzing the time-series properties of money and prices. Section IV contains some concluding remarks. *University of Chicago and Tel-Aviv University. I am indebted to John Bilson and Rolf Banz for suggestions and efficient research assistance. In revising the paper I have benefited from numerous suggestions by Robert Barro, Phillip Cagan, Kenneth Clements, Rudiger Dornbusch, Paul Evans, Stanley Fischer, Benjamin Friedman, Milton Friedman, John Gould, Zvi Griliches, Arnold Harberger, Albert Hart, James Heckman, Edi Karni, Mohsin Khan, David Laidler, Edward Lazear, Robert Lucas, Huston McCulloch, Merton Miller, Franco Modigliani, Michael Parkin, Aris Protopapadakis, Thomas Sargent, Jose Scheinkman, Larry Sjaastad, Jerome Stein, Lester Telser, and Arnold Zellner. Financial support was provided by a grant from the Ford Foundation.

This paper analyzes the key issues and lessons from the experience with flexible exchange rates during the 1970s. It analyzes the efficiency of the foreign-exchange market and the volatility of exchange rates, as well as the relationships between exchange rates and interest rates. A key distinction is made between anticipated and unanticipated events, and it is shown that the key factor affecting exchange rates has been "news." The analysis then proceeds to analyze the relationship between exchange rates and prices. The deviations from purchasing power parities are being interpreted in terms of the modern asset-market approach to the exchange rate.

This paper analyzes the key issues and lessons from the experience with flexible exchange rates during the 1970s. It analyzes the efficiency of the foreign-exchange market and the volatility of exchange rates, as well as the relationships between exchange rates and interest rates. A key distinction is made between anticipated and unanticipated events, and it is shown that the key factor affecting exchange rates has been "news." The analysis then proceeds to analyze the relationship between exchange rates and prices. The deviations from purchasing power parities are being interpreted in terms of the modern asset-market approach to the exchange rate.