«Roman Horváth and Luboš Komárek No 665 WARWICK ECONOMIC RESEARCH PAPERS DEPARTMENT OF ECONOMICS Optimum Currency Area Theory: A Framework for ...»
OPTIMUM CURRENCY AREA INDICES:
EVIDENCE FROM THE 1990s
WARWICK ECONOMIC RESEARCH PAPERS
DEPARTMENT OF ECONOMICS
Optimum Currency Area Theory: A Framework for
Discussion about Monetary Integration
In this paper the authors calculate OCA-indexes for industrial countries in an effort to estimate the benefit-cost ratio of adopting a common currency. The results correspond to the estimation of Bayoumi and Eichengreen (1997b) and show that the ranking of the economies suitable to form a monetary union stays the same in the 1980s as well as in the 1990s. In other words, the economies, which were structurally close to each other in the 1980s, remain close in the 1990s and the opposite is valid for the structurally different economies. This empirical estimation also does not provide evidence for views, which emphasise the seemingly striking difference between the core and the periphery of the European Union. The authors perform also an estimation of the same index by including the Czech Republic and find no support for the view that the economy of the Czech Republic could possibly structurally differ more than * Roman Horvath – Graduate student at Central European University, Hungary and Faculty of Social Sciences, Charles University, The Czech Republic, e-mail: firstname.lastname@example.org. Lubos Komarek – Advisor to Bank Board Member, The Czech National Bank, Prague, Assistant Professor at University of Economics Prague, The Czech Republic and a former student at the University of Warwick, e-mail: email@example.com. We note that everything contained in this paper represents their own views and should not be construed as representing those of the Czech National Bank. However, all errors and omissions remain entirely the fault of the authors. We are grateful to Jarko Fidrmuc (Oesterreichische Nationalbank), Jan Frait (The Czech National Bank) and Julius Horvath (Central European University) for helpful comments. The Czech National Bank research program has supported research staying behind the paper. We thank Eszter Timar (Central European University) for language advice.
the EMU member countries between each other. Then they conclude that if the EMU is sustainable, the accession of the Czech economy should not change it.
E32, F42, E42, F33
Optimum currency area – EMU – monetary policy – convergence – core and periphery 1. INTRODUCTION The optimum currency area (OCA) theory arises from the debates about the exchange rate regimes and adjustment under balance of payments disequilibria. Mundell (1961) in his seminal work on OCA theory challenged Friedman’s (1953) view on floating exchange rate regime as means to the adjustment under balance of payments disequilibria due to exogenous shocks. Mundell (1961) in his model of an asymmetric shift in demand of two countries stressed that optimum currency area can differ from the actual currency area. Such difference could cause inability of the floating exchange rate regime to cushion the shock and bring the countries back to equilibrium. That is why Mundell (1961) offered some non-exchange rates means for adjustment, as labor mobility, nominal flexibility and fiscal transfers. Later, Ingram (In: Kawai, 1987), McKinnon (1963) and Kenen (1969) extended the list of non-exchange rate means for adjustment by considering financial integration, openness and national product diversification.1 According to McKinnon (2000) Mundell presents a neo-Keynesian model, still in a belief that it is possible to eliminate the effect of shocks by national monetary and fiscal policies. There is also another neo-Keynesian relic: Mundell’s implicit assumption of the downward sloping and stable Phillips curve.
However, there are two other later articles of Mundell (1973a, 1973b), which bring completely different argumentation concerning the optimum currency area. This is his monetarist view on the subject: if countries can adopt common currency without substantial change in their purchasing parities, then they gain better allocation of the capital, since they will get rid of the uncertainty in the exchange rates. Foreign reserves will have to increase less than proportionally with the size of the economy. Then, under asymmetric shocks in countries with common currency, there will not be the decline in output, because the costs of absorbing the shocks would be effectively spread in time. The existence of two Mundell models - early and recent - explains why he is heavily quoted both by proponents and skeptics of the European Monetary Union (EMU).
This paper instead of looking for the crucial economic characteristics to determine where the (illusionary) borders for exchange rates should be drawn we concentrate on benefits and costs of the common currency. That means that we assume that by definition no single country fulfills completely the attributes to make it an optimal member of a monetary union.
Moreover, it is a relevant question – in this context – how much the trade integration matters. There are at least two views on this issue. First, countries can benefit from higher trade integration because it leads to more effective allocation of resources. With higher trade integration there will be further synchronization of national business cycles because trade among European countries is typically intra-industry trade based on economies of scale and This search for optimum currency area characteristics is not exhausted by these examples. For survey see Horvath (2001).
imperfect competition. Thus, it will not lead to higher specialization and – above all – it will not lead to higher possibility of asymmetric shock occurrence.
Proponents of the second view2 argue that higher trade integration will lead to higher specialization. Because of the economies of scale, higher integration will lead to the regional concentration of the industrial activity. Thus asymmetric shocks are more likely to occur in the future (since the output is less diversified) and this will bring extra costs to monetary union.
Frankel and Rose (1998a, b) show that the higher the trade integration, the higher the correlation of business cycle among countries. 3 Furthermore, they emphasize that business cycle and trade integration are inter-related and endogenous processes to establishing a currency union. Thus, they demonstrate that countries may fulfill the OCA criteria ex post, although they did not fulfill them ex ante. Monetary union entry raises trade linkages among the countries and this causes the business cycle to be more symmetric among the participants of the union. 4 The arguments of Frankel and Rose (1998a,b) lead to a conclusion that the costs of implementing common currency are relatively low. However, there are some doubts on the validity of the endogenous OCA criteria. In a theoretical model Hallett and Piscitelli (2001) show that the validity of endogenous OCA hypothesis is uncertain and dependent to a large extent on the structural convergence in the beginning phase of the monetary union. Without the sufficient structural convergence, implementing common currency would cause greater divergence.
Maybe the interesting question is not the search for the optimal exchange rate regime, but the search for the optimal variability of the exchange rate. Bayoumi and Eichengreen (1997a,b and 1998) suggest an approach for modelling exchange rate variability, which takes into account the multiple interdependency of the economies. This paper follows the line, which Bayoumi and Eichengreen begin. Thus, the purpose of this paper is to estimate to what degree the exchange rate variability may be explained by the traditional OCA criteria, as defined in the classical OCA literature in the 1960s. Also, this paper attempts to determine the so-called OCA-indexes, which for given pairs of countries assess the benefit-cost ratio for adopting a common currency.
The paper is structured in the following manner. In section 2 we provide the methodology of estimation, in section 3 we present the results. Finally, we summarise and conclude.
Krugman (1993). De Grauwe (1997) discusses the limitations of Krugman’s view. He shows that Krugman assumes that the regional concentration of industry will not cross the borders of the member countries, while borders will be less relevant in influencing the shape of these concentration effects. If so, then asymmetric shock will not be country specific and floating exchange rate could not be used to deal with asymmetric shocks anyway. In addition there will be trade creation among the monetary union countries.
See Rodrick (2000) for a critique of econometric methods used by Frankel and Rose (1998a, 1998b).
According to Fidrmuc (2001) the intensity of intra-industry trade is another variable with positive impact on the synchronization of business cycle.
2. METHODOLOGY Countries experiencing symmetric shocks or high trade linkages tend to have stable exchange rates. In other words the more the OCA criteria among the countries are fulfilled, the lower should be the exchange rates variability among considered countries. Under this
assumption we estimate the equation:
SD(eij) = a + b1SD(?yi-?yj) + b2DISSIMij + b3TRADEij + b4SIZEij (1) SD(eij) measures the volatility of bilateral nominal exchange rates, SD(?yi-?yj) captures the asymmetric shocks at national level, TRADEij is the proxy for intensity of trade linkages, DISSIMij assesses the asymmetric shocks at industrial level and SIZEij measure the size of the economy and assess utility from maintaining own currency. 5 The proxies are computed as follows: SD(eij) is the standard deviation of the change in the logarithm of the bilateral exchange rate between countries i and j on monthly basis, SD(?yi-?yj) is the standard deviation of the difference in the logarithm of real output between i and j, DISSIMij is the sum of the absolute differences in the shares of agricultural, mineral, and manufacturing trade in total merchandise trade, TRADEij is the mean of the ratio of bilateral exports to domestic GDP for the given two countries, and SIZEij is the mean of the logarithm of the two GDPs measured in U.S. dollars.
The data sample contains twenty-one industrial countries for the period from 1989 to
1998. These are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Great Britain, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, and the USA. For convenience we label these data as representing the 1990s. When calculating variable SD(eij) we used the data from IFS-IMF, the data for SD(?yi-?yj) were calculated from World Bank, TRADEij was calculated using the data from Directions of Trade – IMF and World Bank, variable DISSIMij was calculated with the use of the data from Monthly Statistics of Foreign Trade-OECD and SIZEij from the World Bank data. When putting together the data matrix we follow the advice of Bayoumi and Eichengreen (1997a,b and 1998), this allows me to compare the results for different time periods.
Since we are interested whether the exchange rate variability is explicable by traditional OCA criteria, we consider the variables with the impact across the borders in all the equations. Bayoumi and Eichengreen (1997a,b and 1998) find little evidence that more open economy tends to fix its currency. But since the openness is also one of the traditional OCA criteria, we include the proxy for openness, too.6 This means that we estimate the
The lower the size the lower the relative utility of maintaining its own currency. SIZEij can possibly capture the effect of adjustment costs, too. The bigger the countries are in economic terms, the higher the costs of transition to adopting of common currency.
The proxy was calculated as an arithmetic mean to the i-th and j-th country ratio of trade (export + import) to its GDP.
SD(eij) = a + b1SD(?yi-?yj) + b2DISSIMij + b3TRADEij + b4OPENij (2).
The analysis takes into account all the relationships between each of the economies.
There is a pair of countries in each row of the data matrix. Given 20 industrial countries we obtain 190 observation.7 The expected signs of explanatory variables are as follows: the exchange rate volatility is expected to depend positively on business cycle, dissimilarity in the commodity structure of export, and negatively on the trade linkages. The expected sign of the openness is theoretically indeterminate.8 We are aware that there is a possibility that the independent variable influences the dependent variable, i.e. there is a potential influence of exchange rates variability on growth or volume of trade. However, taking the standard deviation of output and volume of bilateral trade considerably reduces this influence.
The relationship of the first country with second one is the same as the second one with the first. That’s why the number of observation equals 20!/18!2!. Since the data for calculation of the variable DISSIM were not available for Greece except the year 1997, we finally excluded Greece from the analysis. At first, we took the data for the year 1997 as an average measure of Greece’s DISSIMij for the period 1989-1998, but the tests on outliers using studentized residuals showed that many observations on Greece are outliers even at p-value lower than 0.01. If this was caused by the lack of the data or for another reason is uncertain.
See Isard (1995).
One can expect that the OCA criteria will explain less of the exchange rates variability in the 1990s than e.g. in the 1980s due to the advances in monetary integration in the EU and also due to EMS financial crisis in 1992-1993.
Note: Volatility in this table means the standard deviation of the change in the logarithm of the bilateral exchange rate between countries i and j; based on monthly data.