Explaining the early years of the euro exchange rate: An episode of learning about a new central bank

https://doi.org/10.1016/j.euroecorev.2006.03.002Get rights and content

Abstract

Many observers were surprised by the depreciation of the euro after its launch in 1999. Handicapped by a short sample, explanations tended to appeal to anecdotes and lessons learned from the experiences of other currencies. Now sample sizes are just becoming large enough to permit reasonable empirical analyses. The model of this paper provides empirical support for the euro exchange rate to be affected by learning. By focusing on euro-area inflation as the key fundamental, the model is structured toward the dynamics of learning about ECB policy with regard to inflation. While a stated target inflation rate of 2 percent existed, it may be that market participants had to be convinced that the ECB would, indeed, generate low and stable inflation. With a prior distribution drawn from the pre-euro EMS experience and updating based upon the realized experience each month following the introduction of the euro, the evidence suggests that it was not until December of 1999 that the market assessed a greater than 50 percent probability that the inflation process had changed to a new regime. From this point on, trend depreciation of the euro ends and further increases in the probability of the new inflation process are associated with euro appreciation versus the US dollar, the British pound and the Japanese yen.

Introduction

The launch of the euro on January 1, 1999 was the most important international financial event since the end of World War II. This new currency was expected by many to garner immediate acceptance and challenge the role of the dollar as a vehicle currency. Nevertheless, in its infancy its role in the foreign exchange market has been less than what was expected.

As shown by Hartmann (1998), the nations that comprise the European Monetary Union make an economic unit at least as large as the U.S. . European Union (EU) GDP exceeds U.S. GDP, EU population exceeds U.S. population, EU exports surpass U.S. exports, and outstanding claims in total EU capital markets (bank assets, bonds and equities) are larger than those in the U.S. . All these indicators would lead us to think that the new currency would challenge the supremacy of the dollar as the most important currency in the world. Nevertheless, the triennial Bank for International Settlements (BIS) survey indicates that the dollar's share of foreign exchange market activity has risen while that of the euro compared to legacy European currencies has fallen. In 1998, the dollar entered on one side of 87 percent of foreign exchange transactions and the legacy euro currencies 53 percent1. In 2001, the dollar share rose to 90 percent while the euro's share was but 38 percent. In 2004, the dollar share was 89 percent while the euro's share was 37 percent. Further evidence is provided in Hau et al. (2002) who show that the average daily dollar/euro volume in foreign exchange trading is nine percent lower than the dollar/DM volume2. Moreover, they show that the trade volume of the euro with the yen and the Swiss franc decreased by 44 and 25 percent, respectively when compared with the mark. This decrease in volume is striking if we consider that the mark is only one of the legacy currencies in the monetary union.

Our focus is not on the volumes traded but on the prices. Most scholars and practitioners would consider a major issue to be an understanding of the determinants of the level of the exchange rate. Earlier explanations have included Sinn et al., 2001, Sinn et al., 2003, who explain the early weakness of the euro by arguing that holders of black market currency were afraid to convert their old European coins and black market notes into the euro in 2002, so they either spent them on goods and services, whereby the lower demand for money is associated with euro depreciation, or else they exchanged them for dollars prior to the appearance of euro currency. Alquist and Chinn (2001) say that the appreciation of the dollar after 1999 can be explained by U.S.–Euro area productivity differentials; however, the euro was also depreciating against the yen, so that this explanation alone cannot do. Our explanation emphasizes the role of the new central bank and the effect of a lack of European Central Bank (ECB) credibility on the exchange rate when the market is learning about the ECB policymaking process and addresses the initial euro depreciation as well as later appreciation against the dollar and pound.

Credibility of the European Central Bank, or the lack of it, has likely played a very important role in the determination of the price of the euro in its infancy. The ECB is not the central bank of one country. It covers the geographical area of 12 different countries, each with its own history, culture and economic background. In addition, the lack of historical performance creates uncertainty about the capability of this new institution in achieving low levels of inflation. Such characteristics initially increased the difficulty of accomplishing the principal mission of this central bank, price stability.

The introduction of a new central bank changed the inflation process in the euro area. Initially, the market had limited information about how committed the ECB was to maintaining low inflation. Even though the ECB stated that its primary objective, as laid down in the Maastricht Treaty, is to maintain price stability,3 rational agents need more than mere announcements to be convinced that the ECB is going to devote all its efforts to accomplish that goal. They will use all available past and current information to evaluate whether or not the ECB can achieve the target level of inflation.

We hypothesize that, beginning in 1999, the market was learning about ECB policymaking by observing the inflation rate in the euro area. Since then, agents are using this information to recognize how different the inflationary process is before and after the introduction of the ECB. We model this learning process using Bayesian updating to calculate the probability that the inflation process in the euro area has actually changed. The evolution of the probability of being in the new state, and consequently, the effects that learning has on the exchange rate, are the focus of the empirical work.

The paper is organized as follows: The next section presents a framework for modeling the exchange rate as a function of learning about the euro-area inflation process. Section 3 illustrates the empirical framework and the Bayesian estimation and inference methodology we adopt. Section 4 reports the empirical results from the estimation of the change point model. Then in Section 5, the estimated probability of a new inflation process is used as a variable to explain the dollar/euro, pound/euro, and yen/euro exchange rates. Finally, Section 6 offers a summary and conclusion.

Section snippets

Pre- and post-euro exchange rates

Before the introduction of the euro, the most important currency pair in the foreign exchange market was the mark/dollar. Denote the currencies as dollar (A) and mark (G). Here it is assumed that the market believes that exchange rates change because of the inflation differential between Europe and the U.S. and additional fundamental determinants which include interest differentials. Prior to the euro, the U.S.–Europe inflation differential for EMS currencies was greatly influenced by German

Empirical model and methodology

We model the inflation process in Europe as a change point model. Specifically, to account for the predictable persistence in inflation rates we rewrite the model in Eqs. (3), (4) asϕt=αt+βtϕt-1+ξt,where ξt is independently and identically distributed as a normal random variable with mean zero and variance σ2 and where the parameters θt≡(αt, βt) with t=1, …, n, are such thatθt={θ1iftτ1,θ2ifτ1<tτ2,θm+1ifτm<tn,where n is the sample size. This nonlinear time series structure assumes that one

Empirical results

We estimate the change point model in Section 3 using monthly inflation data for the euro area. From Datastream International we collect the monthly series of percentage year-to-year changes in the Harmonized mean of Consumer Price Indices. The series runs from December 1998 through May 2005 for a total of 78 observations.

Learning effects on the euro exchange rate

To test the hypothesis that learning about the inflation process affected the euro exchange rate, we model the monthly change in the logarithm of the exchange rate (Δlog(E)) as the dependent variable, and the probability that euro-area inflation follows a new process (Prob) along with a constant term as the independent variables. We estimate the effect of learning about the ECB inflation process on the dollar/euro, pound/euro, and yen/euro exchange rates.8

Summary and conclusion

To understand the behavior of the foreign exchange value of the euro in its infancy, we suggest that the market was learning about the ECB ability to maintain low inflation in Europe. We model this learning using Bayesian estimation and calculate probabilities that reflect the market's beliefs about the ECB's low inflation commitment. The calculated probabilities indicate that it was not until December of 1999 that the market assessed a greater than 50 percent probability that the inflation

Acknowledgments

We thank seminar participants at the Econometric Society winter meetings, Arizona State University, and the universities of Bern, Frankfurt, and Tübingen for their comments on an earlier draft. In addition, helpful comments were received from Siddhartha Chib, two referees and the editor (Juergen von Hagen). We remain responsible for any remaining shortcomings.

References (11)

  • S. Chib

    Estimation and comparison of multiple change-point models

    Journal of Econometrics

    (1998)
  • Alquist, R., Chinn, M.D., 2001. The euro and the productivity puzzle: An alternative interpretation. Working Paper,...
  • S. Chib

    Marginal Likelihood from the Gibbs Output

    Journal of the American Statistical Association

    (1995)
  • Geweke, J., 1992. Efficient simulation from the multivariate Normal and Student-t distributions subject to linear...
  • P. Hartmann

    Currency Competition and the Foreign Exchange Markets: The Dollar, The Yen and the Euro

    (1998)
There are more references available in the full text version of this article.

Cited by (0)

View full text