INTRODUCTION TO ‘‘EXCHANGE RATES IN EUROPE AND AUSTRALASIA: FUNDAMENTAL DETERMINANTS, ADJUSTMENTS AND POLICY IMPLICATIONS’’ JEROME L. STEIN* Brown University
GUAY C. LIM University of Melbourne The papers in this issue are concerned with the behaviour of exchange rates – their fundamental determinants, adjustment processes and policy implications. The authors combine theory with empirical evidence, test hypotheses as well as show the relevance of their analysis for policy. This Introduction addresses several questions. What are the contributions of the articles to the economics of exchange rates? Are they significant in increasing our understanding of the current issues and in addressing questions of policy? How can one explain the movements in the European and the Asian currencies? What can we expect to happen to the euro with the enlargement of the euro area? Do the papers provide frameworks to guide empirical research? In this Introduction, we highlight aspects of each paper that addresses these issues.
I.
Framework and Overview
The papers in this issue may be assessed collectively, and with respect to each other, with the aid of Figure 1 which is a diagrammatic representation of equation (1) below N e ðtÞ ¼ Re ðtÞ½P ðtÞ=P ðtÞ:
ð1aÞ
ne ðtÞ ¼ re ðtÞ vðtÞ:
ð1bÞ
N e ðtÞ is defined as the equilibrium nominal exchange rate (expressed as units of foreign currency per unit of domestic currency, so a rise is an appreciation of the domestic currency). It is the product of the equilibrium real exchange rate and relative prices. The logarithm of the equilibrium nominal exchange rate is equal to the log equilibrium real exchange rate Re ðtÞ less the log ratio of domestic to foreign prices P ðtÞ=P ðtÞ. An equilibrium value is denoted by the superscript ‘‘e’’. Lower case letters in equation (1b) denote the logs of the variables in equation (1a), and v is the log of relative prices. Figure 1 graphs the log of the actual nominal exchange rate nðtÞ against the log of the ratio of domestic to foreign GDP price deflators vðtÞ, for two equilibrium log real exchange rates re (1), and re (2). The log equilibrium real exchange rate is the intercept and the slope is )1. The figure shows clearly that the equilibrium nominal exchange rate is a family of lines. * Correspondence: Jerome L. Stein, Division of Applied Mathematics, Box F, Brown University, Providence RI 02912. Email:
[email protected]. Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
330
DECEMBER
AUSTRALIAN ECONOMIC PAPERS
n(t) I
E5|I(t)
II R[Z(1)]
n(1)
E1
E3 R[Z(2)] v(1)
v(t)
v(4)
v(2) n(2)
I
E2 E4 II
Figure 1. Adjustments of the nominal exchange rates and relative prices to a change in the equilibrium real exchange rate
The equilibrium real exchange rate should be sustainable, with the following characteristics. (C1) At that rate, there is internal balance where the rate of capacity utilisation is at its stationary mean.1 (C2) There is also external balance such that, on average, there are no changes in reserves or speculative capital flows. (C3) There is portfolio balance, that is investors are content to hold their existing portfolios of foreign and domestic assets at the existing domestic rates of interest. (C4) The ratio of external net liabilities to GDP converges to an endogenous sustainable constant. In general, the variable re ðtÞ is also the mean of a distribution, conditional upon a vector ZðtÞ of real fundamentals that vary over time. We will write this as re ðtÞ ¼ R½ZðtÞ. Many of the papers in this issue are concerned with the modelling and testing of this proposition and they are discussed in Section II. The papers include: Duval who examined whether the equilibrium real exchange rate is a constant (as suggested by the purchasing power parity) and Gylfason who discusses the proposition that the real exchange rate always ‘floats’. This section also includes the paper by Detken, Dieppe, Henry, Marin and Smets which contains an evaluation of various models of the real exchange rate as well as papers by MacDonald and 1 This is a clearer and less ambiguous definition of internal balance than requiring that the unemployment rate be at its NAIRU. The reason is that the NAIRU is not objectively measurable and has not been constant. On the other hand the rate of capacity utilisation is stationary.
Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
331
Maseo-Fernandez, Osbat and Schnatz on the BEER approach to estimating the relationship, re ðtÞ ¼ R½ZðtÞ. We also draw attention to the NATREX approach which has been applied by many papers in this volume. However most of the papers also consider the policy implications of an equilibrium exchange rate that evolves over time. Again, Figure 1 may be used to frame the analysis. Consider initially an equilibrium real exchange rate of R½Zð1Þ which is conditional upon vector Zð1Þ of real fundamentals. In this case, the (log) equilibrium nominal exchange rate is line I-I and if (log) relative prices were v(1), the initial equilibrium will be at E1 with log nominal exchange rate n(1). Let there be a change in the fundamental determinants from Z(1) to Z(2) which depreciates the equilibrium real exchange rate to R½Zð2Þ. The new equilibrium nominal exchange rate must now lie on line II-II below line I-I. Thus, at point E1, the combination fvð1Þ; nð1Þg is not in equilibrium with R½Zð2Þ. Equilibrium can be restored if the log nominal exchange rate nðtÞ converges to line II-II. The convergence could occur by a depreciation of the exchange rate from point E1 to point E2, if relative prices were fixed at v(1). Alternatively, if the nominal exchange rate was fixed at n(1), equilibrium could be restored by a movement from E1 to E3, by a decline in relative prices from v(1) to v(2). Of course, equilibrium could also be established by any other nominal exchange rate and relative price combination on the II-II line. With this framework in mind, Section III draws attention to the papers by Gaspar-Issing and Friedman on the role of monetary policy in an economy with an evolving real exchange rate. This section also highlights four other papers that have applied the NATREX model to generate the equilibrium real exchange rate and which have then discussed the policy implications of their analysis. The papers include that by Stein on the Euro and the issues of enlargement, Fischer-Sauernheimer on the DM and the lessons of history; Federici-Gandolfo on the Italian lira and its misalignment, and Rajan-Siregar on the relative performance of the exchange rate system in Hong Kong and Singapore. Figure 1 also illustrates the many possible interactions between adjustments of the nominal exchange rate, relative prices and the real exchange rate as shocks hit an economy. For example, an economy may be subjected to shocks, such as experienced during the Asian currency crises, without affecting the fundamentals Z(1). In this case, adjustments in the nominal exchange rate and in the relative prices may be represented as movements around the I-I line. If on the other hand the shock also caused a change in the fundamental determinants to Z(2), then the change in the nominal rate nðtÞ would also include a shift from the I-I line to the II-II line. Section IV considers the two remaining papers that examine the behaviour of nominal exchange rates. The paper by Lim models the adjustment path of nðtÞ as an interaction of fundamental and portfolio behaviour. The paper by Goss-Avsar considers behaviour in the yen market. Concluding remarks are contained in the final Section V.
II.
Fundamental Determinants of the Real Equilibrium Exchange Rate
Perhaps the first question that should be addressed is whether the real exchange rate is a constant as implied by the hypothesis of purchasing power parity (PPP) or that it changes over time as shown by shifts in the intercept in Figure 1. Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
332
AUSTRALIAN ECONOMIC PAPERS
DECEMBER
There have been numerous studies to show that the real exchange rate is not a constant and that PPP does not hold. In this volume, Duval’s study of the synthetic euro, over the sample period 1970:1–1999:4, concludes the following. (1) The PPP is not a satisfactory concept in the medium to longer run. The extreme length of deviations from PPP implies that they cannot be explained solely by monetary disturbances combined with nominal price rigidities. (2) In the case of developed countries, the Balassa-Samuelson hypothesis is only weakly supported, if at all.2 The presence of non-traded goods explains only a very small portion of the long run behaviour of the real exchange rate. Duval uses direct measures of relative price non-traded/traded goods, based upon sectoral price deflators, rather than the ratio of aggregate price indices the CPI/WPI, to test the Balassa-Samuelson. (3) In most cases, relative PPP is also rejected for traded goods. Any credible theory for the long run equilibrium real exchange rate should also embody a theory for the equilibrium real exchange rate for traded goods. The paper by Gylfason also contends that the equilibrium real exchange rate is not a constant, but it always floats: the intercept is not a constant, but varies over time. Furthermore, that although there is an ‘equilibrium’ real exchange rate Re ðtÞ to which the actual real exchange rate converges, the convergence to equilibrium would always take time, because there are stock adjustments, lags in the underlying functions and price and wage rigidities. Convergence may also be non-monotonic because of the lag structure involving the interaction between the real exchange rate and the underlying import and export functions. Gylfason also shows that there have been long and persistent periods of overvaluations (for example in Africa and in Latin America) when some of the conditions (C1)–(C4) above were violated. These overvaluations reduced the trade balance, increased the foreign debt, and caused declines in the rate of investment, economic activity and growth. Moreover, the distortions that accompany the overvaluations lowered efficiency and total factor productivity. The main implication of these two analyses is that we need models which focus on the real fundamental determinants of the equilibrium/sustainable real exchange rate. Most of the papers in this issue are concerned with estimating just such a relationship. Detken, Dieppe, Henry, Marin and Smets [hereafter DDHMS] carefully examined the literature concerning the determination of exchange rates, in order to evaluate the explanatory powers of the various models and hypotheses.3 DDHMS expose both the statistical uncertainty surrounding the respective estimates as well as general model uncertainty. They discarded those models that were either non-operational, in the sense that the crucial variables were not objectively measurable, or whose structural equations have been shown to be inconsistent with the evidence. They also examined various econometric methodologies ranging from the reduced form time series approach which does not impose theoretical structure on the specifications, to the structural VAR approach to decompose the real exchange rate into its permanent and temporary components. They also used a NATREX approach and a small-scale macroeconomic model for the euro area, as exposed in Fagan, Henry, and Mestre (2001). The models were evaluated using the same data set over the same sample period 1973:1–2000:4, and the aim is to compare the results of the various approaches and ascertain the more robust determinants of the equilibrium exchange rate of a synthetic euro EU-11 relative to the trading partners US, UK, Japan, Switzerland. 2 The Bundesbank (1995) reached the same conclusions, concerning the Balassa/Samuelson effect for the D-Mark. 3 See also Stein and Paladino (1997) for an evaluation of the models used in international finance.
Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
333
The empirically oriented approach that is not model-specific and directed at estimating the behavioural equilibrium real exchange rate (BEER) usually implies a smaller degree of under- or overvaluation. On the other hand the equilibrium concept in a BEER model is very different than in more structural approaches, like the Natural Real Exchange Rate (NATREX) framework or the area-wide model. All these approaches are positive, and not normative, economics because there is no value judgement implicit in the derived equilibrium real exchange rate. There is no implication that exchange rates should be managed. The focus upon the equilibrium trajectory, in both BEER and NATREX, has been motivated by several factors. Perhaps the most important rationale is that short-term approaches rely very heavily upon anticipations and consequently the explanatory power of short-term models tends to be unsatisfactory. Hence, it is more fruitful to focus research on the equilibrium (long run) real exchange rate and, if required, these fundamental relationships may then be used to underpin studies of the short-term movements in exchange rate.
a) The BEER approach The BEER approach developed by Clark and MacDonald (1999) can be summarised by the following equations. SðZðtÞÞ IðRe ðtÞ; ZðtÞÞÞ ¼ CAðRe ðtÞ; ZðtÞÞ
ð2aÞ
Re ðtÞ ¼ RðZðtÞÞ
ð2bÞ
logRe ðtÞ logRðtÞ ¼ a½r ðtÞ rðtÞ
ð3Þ
logRðtÞ ¼ RðZðtÞÞ þ a½rðtÞ r ðtÞ:
ð4Þ
The equilibrium real exchange rate Re ðtÞ in equation (2a) equilibrates saving SðZðtÞÞ less investment IðRe ðtÞ, ZðtÞÞ to the current account CAðRe ðtÞ; ZðtÞÞ, where ZðtÞ denotes a vector of (exogenous) explanatory variables, determined eclectically from economic theory. Equation (2a) is solved for the equilibrium real exchange rate, here written as equation (2b). This is the longer-run cointegrating equation which is estimated to generate the BEER. In the medium run, the uncovered real interest rate parity (UIRP) equation (3) is added. The real exchange rate RðtÞ converges to its longer run equilibrium Re ðtÞ in proportion to the real long-term foreign ðr Þ less domestic (r) interest rate differential. The real exchange rate in equation (4) combines the systematic component determining the log of the equilibrium real exchange rate, R½ZðtÞ, with the real interest rate differential, ðr r Þ. Makrydakis, de Lima, Claessens and Kramer (2000) describe BEER as follows: ‘The BEER, . . . attempts to explain the actual behavior of the real exchange rate in terms of a set of relevant explanatory variables, the so-called ‘fundamentals’. The fundamental exchange rate determinants are selected according to what economic theory prescribes as variables that have a role to play over the medium to short-term …. [In BEER] . . . the underlying theoretical model does not have to be specified. The exchange rate equilibrium path is then estimated by quantifying the impact of the ‘fundamentals’ on the exchange rate through econometric estimation of the resultant reduced form.’ Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
334
AUSTRALIAN ECONOMIC PAPERS
DECEMBER
In this volume, the BEER approach is adopted by Maseo-Fernandez, Osbat and Schnatz to determine the equilibrium real value of the synthetic euro and by MacDonald for the New Zealand Dollar. Duval also includes aspects of BEER in his model. Maseo-Fernandez, Osbat and Schnatz undertake a meticulous empirical analysis of the medium-term determinants of the synthetic Euro EU/12 partners, using quarterly data 1975:1– 2000:4. The econometric methodology is the VECM. The candidate fundamental variables are measures of productivity, net foreign assets, measures of fiscal variables, the real price of oil (an inverse measure of the terms of trade), and interest rate differentials. They find a cointegrating equation which tracks the real value of the synthetic euro thus showing that the equilibrium real exchange rate, the intercept in Figure 1, is systematically determined by real variables. MacDonald also estimates the long-run effective exchange rate of the New Zealand dollar using the BEER approach and the variables tested are similar to those already mentioned. An interesting aspect of this study is the inclusion of an output gap variable. By setting the gap to zero, and letting the interest are differential equal a risk premium, a long term equilibrium real exchange rate was derived. Perhaps the most important advantage of the BEER approach is that it considers only variables that are empirically significant. However, results tend to be sensitive to which variables are/are not included in the regression and the results are also consistent with a range of different exchange rate models. Hence, there are questions of interpretation which restricts its use for framing policy issues and analysis.
b) The NATREX approach The equations of the NATREX approach contains the following key equations. SðX ðtÞ; ZðtÞÞ IðRe ðtÞ; X ðtÞ; ZðtÞÞÞ ¼ CAðRe ðtÞ; X ðtÞ; ZðtÞÞ:
ð5Þ
dF ðtÞ=dt ¼ CAðRe ðtÞ; X ðtÞ; ZðtÞÞ:
ð6Þ
dY ðtÞ=dt ¼ bðtÞIðRe ðtÞ; X ðtÞ; ZðtÞÞ
ð7Þ
Re ðtÞ ¼ RðZðtÞÞ
ð8Þ
The NATREX model of the equilibrium real exchange rate starts with equation (5) above, evaluated when the rate of capacity utilisation is at its stationary mean and when real longterm real rates of interest between the home and world have been equalised. The vector of endogenous variables X ðtÞ consists of the ratio of net liabilities to GDP denoted by F ðtÞ, and the growth rate of GDP. Vector ZðtÞ contains exogenous fundamental variables. These are typically measures of social – public plus private – consumption/GDP, productivity and the terms of trade. The model also includes two dynamic equations. Equation (6) states that the rate of change of net foreign liabilities is the negative of the current account. Equation (7) states that the growth rate is directly related to the rate of investment/GDP. The coefficient bðtÞ is the productivity of investment, which is an exogenous variable included in vector ZðtÞ. The NATREX model concerns a sustainable rate, which has characteristics (C1)–(C4) above. It is a framework for analysing the longer run determinants of the real exchange rate, namely it provides the theory for the fundamental determinants of the equilibrium real exchange rate, Re ðtÞ (or equivalently in log terms, the intercept in Figure 1). Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
335
Detken and Marin-Martinez (2001) describe NATREX as follows: ‘The NATREX approach is based upon a rigorous stock-flow interaction in a structural macroeconomic growth model. The approach tries to link the real exchange rate . . . to a set of fundamental variables explaining saving . . . investment . . . and the current account . . . Its particularity with respect to the standard macroeconomic balance approach is the distinction made between the medium-run equilibrium and the long-run equilibrium where in addition net foreign debt . . . and the capital stock . . . have each reached their steady state levels’.
In this volume, the NATREX approach is adopted by DDHMS, Duval, and Stein for the synthetic Euro, by Fischer-Sauernheimer for the D-Mark, by Federici-Gandolfo for the Italian Lira, and by Rajan-Siregar for the exchange rates of Hong Kong and Singapore.4 The synthetic Euro is a weighted average of the component currencies relative to their main trading partners. The papers by DDHMS and by Duval include their structural models of the NATREX approach to explain the equilibrium real value of a synthetic Euro. All coefficients have the signs hypothesised by the NATREX approach, and are significant. After estimating the coefficients in the structural equations, they solve the model for the equilibrium real exchange rate. The main driving forces underlying the equilibrium real exchange rate are the ratio of social – public plus private – consumption/GDP, the growth of productivity and the terms of trade. They compare the actual values of the synthetic euro with the estimates of the medium and longer run NATREX-generated value and find strong support for their versions of a NATREX based approach. Both papers have expanded the approach in interesting ways, for their particular case study, but their suggestions have broader applications.
III.
Exchange Rates and Policy
a) Monetary policy Gaspar-Issing discuss the role of the exchange rate in the context of the European Central Bank (ECB) policy. The ECB announced a stability oriented monetary policy strategy with several characteristics. First, the primary goal is to keep price inflation in the medium term below 2 per cent p.a. Price stability provides an important contribution to sustainable growth and efficiency in resource allocation. Second, a prominent role is assigned to the growth of M3 as the main systematic determinant of the price level, but the ECB also uses other instruments and indicators. Third, the exchange rate target is not considered as an appropriate target because it may be inconsistent with the goal of price stability. In terms of Figure 1, suppose that the foreign price is growing at an annual rate of 2 per cent so that the desired relative price ratio is constant at v(1). The ECB employs monetary policy to keep the relative prices at v(1). The exchange rate of the Euro is neither a target nor an objective. Insofar as there are changes in fundamentals ZðtÞ, the nominal exchange rate of the Euro would vary vertically, between nð1Þ and nð2Þ. If the ECB tried to maintain the nominal exchange rate of the Euro at nð1Þ then, as the equilibrium real exchange rate varied between R½Zð1Þ and R½Zð2Þ, the ECB would lose control over the relative prices which would vary between v(1) and v(2). 4 The study of the French Franc/D-Mark by Crouhy-Veyrac and Saint Marc (1995), where they use the NATREX model, obtains results consistent with the studies of the Euro in this issue of the Australian Economic Papers.
Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
336
AUSTRALIAN ECONOMIC PAPERS
DECEMBER
Friedman, in his Comment of the Gaspar-Issing paper, stresses the medium term macroeconomic issues that are underemphasised in the Gaspar-Issing paper. Friedman points out that their paper leans heavily upon the well known idealised Neo-Classical world of monetary neutrality, and that their model has nothing to say about short run dynamics. Gaspar-Issing conclude that: ‘Standard monetary neutrality propositions are among the few results that a prudent central banker can get comfort from’. Friedman responds that, taken literally, their sentence means that central bankers are irrelevant. He argues that central bankers are important insofar as money is not neutral and does have real effects in the medium run. Neutrality propositions give little guidance to effective central bankers. In terms of Figure 1, suppose that the economy is at point E1 when the equilibrium real exchange rate changes to R½Zð2Þ. The nominal exchange rate is now overvalued. The neutrality argument is that it is a matter of indifference whether equilibrium is restored by a depreciation of the exchange rate to point E2 or by a decline in relative prices to point vð2Þ, or some linear combination that would place the economy on line II-II. Friedman and also Gylfason cannot accept the neutrality argument as a guide to policy. A depreciation of the nominal exchange rate from nð1Þ to nð2Þ has very different economic effects from a relative price decline from vð1Þ to vð2Þ. Friedman brought up this issue within the context of the Euro area. There is an implicit equilibrium real exchange rate among the Euro area countries. Changes in the implicit real exchange rate between any pair of member countries, which in a different system would show up as changes in nominal exchange rates, are just as much a concern within the Euro system. They must be adapted to. But under the Euro system, adjustments between member countries must occur from E1 to E3 via changes in relative prices and wages. The medium run adjustment entails changes in output, employment and social tensions, because wages and prices are not perfectly flexible. Adjusting to a change in the equilibrium rate via a change in nominal prices and wages is far more difficult to achieve than adjusting via a change in the nominal exchange rate. Friedman’s argument has great relevance for the current plans to enlarge the Euro area with countries from Central and Eastern Europe. Should these countries enter the Euro area?
b) Euro and the issue of enlargement The Euro Area (EMU) is in the process of expanding its membership. The new members, the enlargement countries, are mainly the Central and Eastern European Countries. Enlargement will affect both the existing countries and the accession countries. The latter are and will be undergoing structural changes in moving from a centrally planned to a market economy. It is difficult to predict their development and they will enter the EMU at different points in time. In order to answer the question: what will be the effect of enlargement of the Euro area upon the equilibrium nominal value of the Euro, we must know what have been the determinants of the equilibrium real value of a synthetic Euro. A valid theory concerning the actual real value of the Euro, whose birth occurred only a few years ago, should be able to explain the real value of the synthetic Euro over the period of floating rates. The advent of the ECB can be expected to change monetary policy and relative prices, but monetary policy should not affect the longer-run equilibrium real value of the Euro. It is impossible to make a prediction concerning the policies that will be in force in the enlarged Euro area, and no one knows at what exchange rates the countries will enter. There are various possible scenarios. Stein uses the NATREX model to evaluate the effect of enlargement upon the equilibrium, defined as a sustainable, value of the real Euro under different scenarios. Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
337
Stein first specifies a NATREX model to explain the underlying theoretical structure linking the ‘fundamentals’ ZðtÞ to the equilibrium real value of a Euro. He then evaluates the explanatory power of the model and finally generates impulse-response functions to show how the Euro converges to its equilibrium value under different scenarios. There are two basic scenarios of enlargement within the context of the NATREX model. Each scenario concerns different elements in the vector ZðtÞ of control variables. In scenario I, there is a rise in social consumption to benefit the current generation. For example, there are increases in government expenditures that raise social consumption of goods and services – to ameliorate the condition of the present generation. This scenario is described by a parametric decrease in the saving. In scenario II, there are policies to stimulate growth to benefit the consumption of future generations. These policies consist of liberalisation of economies, rises in investment and policies to lower marginal costs of producing tradable goods. These policies are aimed at improving the welfare of future generations. Scenario II is described by parametric increases in the current account function and the productivity of investment. The traditional models, such as Mundell-Fleming, argue that decreases in social saving – an expansionary fiscal policy – would appreciate the real exchange rate. This indeed occurs. The decline in saving appreciates the real exchange rate, which leads to a current account deficit and restores macroeconomic balance. The NATREX model refers to this as a medium run effect, because it is evaluated at given values of debt/GDP and growth rates. However, the decline in social saving leads to a current account deficit which raises the level of F ðtÞ the debt. The rise in the debt decreases the current account by increasing the interest and dividend payments to foreigners. The longer run effect of an expansionary fiscal policy is to depreciate the real exchange rate below its initial level. In scenario I, the initial appreciation of the equilibrium real value of Euro will be more than offset by its subsequent decline. In scenario II, the change in vector ZðtÞ will be associated with increases in both the investment function and the competitiveness of the economy. The real exchange rate will first appreciate with a capital inflow. In the longer run, when the productive investment makes the economy more competitive, saving will rise more than investment, and there will be current account surpluses. The decline in the net foreign debt increases the current account function and the equilibrium real exchange rate will appreciate further. In scenario II the real exchange rate appreciates in the medium and then further to the longer run. According to the NATREX model, the net effect of enlargement depends upon the relative importance of the two scenarios.
c) A historical analysis of the D-Mark Fischer and Sauernheimer ask what have been the driving fundamental factors behind the movements in the real value of the D-Mark? And what are their specific impacts? For this purpose, they specify a NATREX model to capture the characteristics of the German economy. Their estimated cointegrating equation verifies that the fundamental determinants of the equilibrium real exchange rate are the relative social consumption/GDP, total factor productivity, relative materials prices and the real long-term interest rate differential with the US. The signs of the coefficients correspond with the medium run effects. Their analysis then allows for a detailed assessment of the periods of over and undervaluation of the actual real effective exchange rate of the D-Mark relative to their estimated NATREX value. The authors examine the sub-periods with appreciating and depreciating values of the D-Mark. Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
338
AUSTRALIAN ECONOMIC PAPERS
DECEMBER
Fischer and Sauernheimer separate the roles of the external fundamentals – the US real long-term interest rate and relative materials prices – and internal fundamentals – relative social consumption and total factor productivity. From 1990:4 to 1993:3, the combined effects of three fundamentals – total factor productivity, social consumption and prices of imported materials – led to an unprecedented 35 per cent appreciation of the NATREX. The market rate lagged behind, so that the D-Mark was undervalued before the crises periods. At the end of 1993, the D-Mark was undervalued by 5 per cent which means that other currencies were overvalued by 5 per cent. As Friedman points out above, responding to a 5 per cent overvaluation by depreciating the nominal exchange rate is very different from responding by deflating relative prices. Insofar as the other European countries’ currencies were not flexible, an exchange crisis could be expected. The Fischer-Sauernheimer analysis of this crisis period – 1992/93 – is an important study because unlike the ‘event studies’ papers, it offers a view of the crisis in a broader context which shows more clearly the factors which led to the crisis. From 1993:3 on the NATREX value of the D-Mark declined. The factors responsible for the decline were varying US real long-term interest rates and falling German productivity. After 1995, misalignment disappeared. In the last year of its existence, the actual real value of the D-Mark was approximately equal to its equilibrium value as measured by the NATREX.
d) The lira and misalignments Federici and Gandolfo introduce two innovations in their analysis of the equilibrium value of the Italian lira. The first is the way they evaluate their model and the second is the way they specify and estimate their structural equations. There are several competing definitions of the equilibrium real exchange rate and each definition may be used as a benchmark from which to evaluate the misalignment of the actual real exchange rate. Federici-Gandolfo suggest that a good strategy to test one’s model of the equilibrium real exchange rate is to simulate that model and see how well its predicted under and overvaluations match with ‘known’ periods of misalignment. Their adaptation of the NATREX framework into a structural model of the Italian Lira also includes innovative features such as the introduction of endogenous growth and the application of continuous time econometrics. Their structural equations are carefully specified and contain behavioural equations such as that investment responds positively to the marginal product of capital less real long-term rate of interest and that consumption is positively related to net worth (capital less foreign debt). The Marshall-Lerner conditions are also satisfied. All of the parameters of their structural equations have the signs hypothesised by the NATREX approach and are significant. The predictive performance of the equilibrium real exchange rate is based upon an out of sample analysis. The authors consider three sub periods, including the 1992 currency crisis. The out of sample estimates indicate that, since the 1980’s, there was a cumulative overvaluation of the lira. The crisis was triggered by both the cumulative overvaluation and the liberalisation of capital movements. They show that estimated misalignment from the model is consistent with the views expressed earlier by the Governors of the Bank of Italy. In the 1970s, official policy was to depreciate the exchange rate to increase the trade balance, rather than to target inflation. This is consistent with the Federici-Gandolfo estimates that the lira was undervalued. In the 1980’s the official policy was to slow down inflation using the exchange rate as a tool. There was a real overvaluation. Governor Ciampi (1988) acknowledges an overvaluation. ‘The Central Bank chose to deploy the exchange rate in the fight against inflation in the Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
339
belief that the balance of payments adjustment had to be achieved through alterations in both the conditions of production and demand and the criteria for setting prices and wages’. The official view that the lira was overvalued is consistent with their out of sample estimates.
e) Currency board or flexible exchange rate An important issue for small open economies is the choice of an exchange rate system. Is it more efficient to have a hard peg with a currency board or a more flexible exchange rate? The article by Rajan and Siregar addresses this issue by comparing Hong Kong and Singapore. Hong Kong has a Currency Board Authority (CBA) and Singapore has a soft peg with a monitoring band arrangement. Neither the central parity nor the band is completely fixed. Instead they are periodically reviewed so as to ensure that they are consistent with economic fundamentals and market conditions. Hong Kong and Singapore are similar small open economies, exposed to similar external shocks. A comparison of these two cases over the period 1983–2000 is effectively an examination of adjustments, as shown in Figure 1, from E1 to E3 for the currency board system, and from E1 to E2 for a flexible exchange rate system. Rajan-Siregar examined whether the Hong Kong rules based CBA performed better than the Singaporean more flexible and discretionary exchange rate system by looking at the degree of misalignment of the two currencies. The first step in estimating misalignment is to determine what happened to the equilibrium real exchange rate R½ZðtÞ. Exchange rate overvaluation is very costly, and has been at the heart of most recent currency crises. Rajan and Siregar applied the NATREX approach to estimate the equilibrium real exchange rate R½ZðtÞ and hence to assess misalignment. The fundamentals ZðtÞ that they find to be significant are real government spending, productivity, and terms of trade. Productivity is the dominant variable for Singapore and terms of trade for Hong Kong. Overall, neither currency appears to have experienced severe misalignment prior to 1997. A contrasting trend emerges during the post 1997 financial crisis. Since 1996, the Hong Kong dollar was overvalued, whereas the Singapore dollar has been close to its equilibrium value. Singapore has been far more successful in weathering the speculative attacks and effects of the ‘contagious fallout’ during the 1997–98 crisis. Singapore has, since the 1990’s, consistently outperformed Hong Kong both in terms of experiencing faster growth rates and lower inflation. The differences were particularly stark in the midst of the regional financial crisis. Hong Kong GDP fell 5 per cent and Singapore was one of the few economies in East Asia to have staved off financial contraction. Against this background, they conclude that the more flexible intermediate exchange rate regime in Singapore has outperformed the Hong Kong Currency Board Authority in general, and in particular in the last decade when a series of external shocks hit the region impacting upon both economies.
I V.
Adjustments of the Nominal Rate
a) Interactions of fundamental and portfolio behaviour Lim is concerned with the behaviour of the nominal exchange rate as it adjusts from a point of equilibrium say E1 to another point of equilibrium along II-II. The model is described in the following equations Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
340
AUSTRALIAN ECONOMIC PAPERS
DECEMBER
E½rðtÞ ¼ rðt 1Þ
ð9aÞ
E½nðtÞ ¼ nðt 1Þ þ E½p ðtÞ pðtÞ
ð9bÞ
E½nðtÞ ¼ nðt 1Þ þ E½i ðtÞ iðtÞ þ hðtÞ
ð10Þ
nðtÞ nðt 1Þ ¼ wðtÞE½p ðtÞ pðtÞ þ ð1 wðtÞÞE½i ðtÞ iðtÞ þ hðtÞ þ uðtÞ
ð11Þ
Equation (9a) is an alternative way to describe the shifts in the equilibrium real exchange rate, where rðtÞ is the log of the real exchange rate. It follows from the fact that the log real exchange rate can be modelled as a non-stationary variable which in turn can be described as a random walk process, here shown in expectation form. Writing equation (9a) in its equivalent nominal form as equation (9b) gives Lim an expression she describes as the fundamental behaviour of the expected nominal exchange rate. Thus, in terms of Figure 1, while most of the papers mentioned in the earlier Sections, estimate the determinants of the intercept term (re ðtÞ), Lim uses equation (9b) to capture the shift from a point on I-I to a point on II-II, for example a point like E1 to a point like E4 (corresponding to a change in relative prices from vð1Þ to vð4Þ, that is the inflation differentials [pðtÞ p ðtÞ). The expected change in the nominal rate is also driven by the uncovered interest parity relationship, here written as equation (10), where hðtÞ is a risk term and i, i are the domestic and international nominal interest rates respectively. The risk premium is modelled as a GARCH process. This equation reflects portfolio behaviour and it may be shown in Figure 1, as say a change from E1 to say E5|I(t) where I(t) represents the informational set containing the interest-differentials and risk which cause, in this example, an expected appreciation of the exchange rate. Lim is interested in modelling the interaction of the fundamental and portfolio behaviour (the pull towards E4 versus the pull towards E5|I(t)) that drives changes (small and large) in the nominal exchange rate. She models the (log) change in the nominal rate, nðtÞ nðt 1Þ, as a weighted average of the inflation differentials and the interest differentials plus risk, shown in equation (11), where uðtÞ is an error term. The weight wðtÞ is endogenously determined and depends on the relative magnitude and size of the fundamental and portfolio errors. The empirical analysis is devoted to an analysis of the currencies of Australia and the ASEAN3 countries of Malaysia, Indonesia and Thailand. The sample period is 1987:01– 2000:12, and contains the pre crisis, Asian currency crisis and post-crisis sub-periods. Her study shows the extent to which changes in the Australian dollar, Malaysian ringgit, Indonesian rupiah and Thai bhat reflected adjustments towards fundamental values (i.e., adjustments towards E4) or adjustments towards portfolio values (i.e., adjustments towards E5jIðtÞ).
b) Liquidity and volume None of the models of the real exchange rate discussed above can forecast short term movements in exchange rates. The reason is that, in the short run, exchange rates behave as speculative prices, where expectations and responses to ‘news’ are critical. The final paper in the volume, by Goss and Avsar reminds the reader that short term exchange rate movements Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
2002
EXCHANGE RATES IN EUROPE AND AUSTRALASIA
341
are determined in an environment where liquidity and volatility are important, and these factors can influence the size of the forum in which trading takes place. They study the relation between liquidity and trading volume, and between volume and volatility for the US dollar/Japanese yen futures contract, which has been trading since 1972. The paper is motivated by a desire to discover whether there is a tendency to increasing returns to liquidity with market development for futures markets. They show that liquidity varies directly with market development, and that this relationship provides a major incentive for mergers among exchanges. They investigate the relationship between liquidity and volume, volume and volatility, and between liquidity and the speculation ratio.
V.
Concluding Remarks
There are indeed fundamental determinants of the equilibrium real exchange rate ZðtÞ, but the content of ZðtÞ is not uncontroversial and can vary over time and with the modelling strategy adopted. This tends to introduce quite some uncertainty for any attempt to derive ‘the’ equilibrium exchange rate. This conclusion is supported by analysis based on both BEER and NATREX approaches. The implication of this conclusion for adjustment processes in the economy, especially in an environment characterised by inflation-targeting monetary policies (as is currently practiced by numerous countries) deserves further studies. We hope this issue of the Australian Economic Papers will serve as a catalyst for more research into the fundamental determinants of the real exchange rate, and its implications for policy and adjustment processes.
References Clark, P. and MacDonald, R. 1999, ‘Exchange Rates and Economic Fundamentals’, in R. MacDonald and J. L. Stein (eds), Equilibrium Exchange Rates, Kluwer Academic Publishers, Amsterdam. Crouhy-Veyrac, L. and Saint Marc, M. 1995, ‘The Natural Real Exchange Rate between the French Franc and the Deutschmark’, in J. L. Stein, P. R. Allen et al., Fundamental Determinants of Exchange Rates. Oxford University Press, Oxford. Detken, C. and Marin-Martinez 2001, The Effective Equilibrium Exchange Rate since the 70’s: a Structural NATREX Estimation, European Central Bank, Frankfurt. Deutsche Bundesbank, Monthly Report/Monatsbericht 1995, Overall Determinants of the trends in the Real External Value of the Deutsche Mark, August, 17–37. Fagan, G., Henry, J. and Mestre, R. 2001, ‘An Area-wide Model (AWM) for the Euro Area’ ECB Working Paper No. 42, European Central Bank. Makrydakis, S., de Lima, P., Claessens, J. and Kramer, M. 2000, The Real Effective Exchange Rate of the Euro and Economic Fundamentals: a BEER Perspective, European Central Bank, Frankfurt. Stein, J. L. and Paladino, G. 1997, ‘Recent Developments in International Finance: A Guide to Research’, Journal of Banking and Finance, vol. 21, pp. 1685–1720.
Blackwell Publishing Ltd/University of Adelaide and Flinders University of South Australia 2002.
