Equilibrium Exchange Rates in Transition Countries: Evidence from - PDF document

A preliminary version, 11 th November 2002. Please do not cite. Equilibrium Exchange Rates in Transition Countries: Evidence from Dynamic Heterogeneous Panel Models Byung-Yeon Kim *+ (University of Essex) and Iikka Korhonen (BOFIT) Abstract : We use a dynamic heterogeneous panel model, namely, pooled mean group estimator, to estimate real equilibrium exchange rates for advanced transition countries. Our method is based on out-of-sample estimations using middle-income and high-income countries. We find that in all the five transition countries, the Czech Republic, Hungary, Poland, Slovakia, and Slovenia, exchange rates have converged to real equilibrium exchange rates expressed in the US dollars in recent years. However, we find that the currencies of the countries are substantially overvalued when real effective exchange rates are used. Keywords : exchange rates, transition economies, dynamic heterogeneous panel estimations JEL Classification: C33, F31, P27 * B-Y Kim, Department of Economics, University of Essex, Wivenhoe Park, Colchester CO4 3SQ, United Kingdom. Tel: 44-1206-872777. Fax: 44-1206-872724. E-mail: bykim@essex.ac.uk Iikka Korhonen, Bank of Finland, Institute for Economies in Transition (BOFIT), PO-Box 160, Helsinki, FIN-00101, Finland. E-mail: iikka.korhonen@bof.fi + We would like to thank for valuable comments participants of an internal BOFIT seminar in September 2001 and of the BOFIT workshop in April 2002. We would especially like to thank Kari Heimonen and Zsolt Darvas for their comments. Usual disclaimer applies. Part of this research was undertaken while B-Y Kim was visiting BOFIT in 2001. B-Y Kim gratefully acknowledges the excellent research environment offered by BOFIT. 1

1. Introduction Following very high inflation in the early period of transition toward a market economy, stabilisation has already been achieved for most of the advanced transition countries. Annual inflation rates in these countries are in single digits, although still higher than in the European Union (EU). Now new challenges await them. Within a few years, these countries are likely to become EU members. Membership in the EU obviously provides new opportunities and benefits for these countries, but it also imposes a number of obligations. One of these obligations is to become a member of the monetary union, that is, to adopt the euro. In order to satisfy the criterion on exchange rate stability for adopting the single currency, a candidate country must be a member of the Exchange Rate Mechanism 2 (ERM2), which allows the domestic currency to fluctuate within the band of � 15% around the central parity, for at least two years without a devaluation of the central parity. Some questions naturally arise in these circumstances. What is the appropriate level of the exchange rate? How can the current exchange rate be evaluated against this appropriate level? Joining first the ERM2 and then euro at an overvalued level of currency suggests loss of competitiveness and subsequently a slower convergence of real incomes towards the EU level. A slow convergence induces cost not only for the transition economics but also for the existing member of the EU through various funds supporting the real convergence within the EU. For example, in 2001 the per capita GDP (calculated with purchasing power adjusted exchange rates) in the accession countries of the Central and Eastern Europe varies from 25% (Romania) to 69% (Slovenia) of the EU average (Eurostat, 2002). In the largest accession country, 2

Poland, per capita GDP is 39% of the EU average. Moreover, an overvalued currency is more likely to suffer from a speculative attack. On the other hand, joining the ERM2 with an undervalued exchange rate will result in inflationary pressures. As the exchange rate is fixed, the expected real appreciation of the currency can only take place through higher inflation. This means an increase in the probability of failing the Maastricht convergence criterion on inflation at the ERM2 stage, therefore jeopardising a quick entry into the common currency. Estimating an equilibrium exchange rate in a transition country is not an easy task. Main problems include the brief history of the transition period together with difficulties associated with using data from the socialism period. The socialism period is of little use to estimate an equilibrium exchange rate because prices in a socialist economy failed to reflect an underlying market mechanism. Using data from the transition period, which has lasted for about ten years, is not expected to provide reliable estimation results of an equilibrium exchange rate, because of its short time profile. Furthermore, exchange rates during initial transition years were affected largely by non-conventional factors such as a sharp increase in demand for foreign goods and assets, high expected inflation and the tendency of the authorities to set initial exchange rates at a sharply undervalued level (Halpern and Wyplosz, 1996; Coricelli and Jazbec, 2001). This suggests that a benchmark value for the real exchange rate is misleading particularly for an early period of the transition if data from transition countries are used to estimate equilibrium exchange rates. A more promising approach appears to use an out-of-sample estimation. Several existing studies use the samples of non-transition economies to estimate equilibrium exchange rates in transition countries (Halpern and Wyplosz, 1996; Krajnyák and Zettelmeyer, 1998, Begg et al., 1999). They use dollar wages in a 3

sample of (mainly) non-transition countries as proxy for the real exchange rate, and regress dollar wages on a set of variables. Based on the coefficients estimated from the regressions, equilibrium dollar wages are computed for transition countries. All studies find that the initial undervaluation of the real exchange rates was followed by considerable real appreciation, although the currencies of the most transition countries were still undervalued at the end of the sample period. Unfortunately their estimations do not go beyond 1997. This paper analyses an equilibrium exchange rate for five advanced transition countries, namely Poland, Hungary, Czech Republic, Slovenia and Slovakia. We improve on the previous studies by exploiting a time-series and panel dimension of the data set. More specifically, we use a dynamic heterogeneous non-stationary model, Pooled Mean Group (PMG) estimator developed by Pesaran et al. (1996, 1999), and apply this model to real exchange rates against the US dollar of 29 middle- and high-income countries from 1975 to 1999. We also use the sample of 19 middle- and high-income countries from 1980 to 1999 to estimate real effective exchange rates. This model allows us to take previous overlooked estimation problems into account: non-stationarity of the data, heterogeneity across countries, dynamics, and differentiation between long-run and short-run properties. Based on PMG estimation results, we calculate equilibrium exchange rates for the sample countries, and then apply the resulting coefficients to compute the equilibrium exchange rates for the transition countries. These equilibrium exchange rates are, in turn, compared with the actual exchange rates, allowing us to assess the degree of over- or undervaluation. We also perform robustness checks on our results. The paper is structured as follows. Section 2 reviews literature on exchange rates in transition economies. Section 3 discusses our methodology. Following a brief 4

discussion on our data set and model, Section 4 presents estimation results of long-run equilibrium exchange rates, and evaluates misalignment of the actual exchange rates in transition countries. Section 5 checks robustness of our results with some alternative estimations of equilibrium exchange rates. Section 6 concludes. 2. Real exchange rates in transition economies Exchange rate has been one of the central policy tools for many transition economies to stabilise their economy. Transition countries have adopted diverse exchange rate regimes. Some have opted for a very hard version of peg, i.e. currency board, while others have adopted a more conventional fixed exchange rate regime or managed floating. Most of the countries have changed course along the way in response to different economic circumstances. Despite of the diversity of exchange rate regimes, these countries in general share a common trend of exchange rate movements. Usually, a sharp initial nominal and real depreciation of currency was followed by real appreciation as domestic inflation exceeded subsequent nominal depreciation and foreign inflation over the course of transition. In this section we first review briefly the relevant literature on the determinants of equilibrium real exchange rates, and then focus on studies concerning real exchange rates in transition countries. 2.1 Determinants of equilibrium real exchange rate Real exchange rate (RER ) is generally defined as the nominal exchange rate adjusted for price level differences between countries. More formally, denote RER t as the real exchange rate (in period t ), E t as the nominal exchange rate (in units of foreign 5