An OCA study in Europe An empirical investigation of the EU countries conditions for qualifying for the Economic and Monetary Union

Transcription

1 M.Sc. thesis in Business Administration (Finance and International Business) Author: Lasse Gavnholt Jygert Advisor: Philipp Schröder An OCA study in Europe An empirical investigation of the EU countries conditions for qualifying for the Economic and Monetary Union The Aarhus School of Business April 2008

2 Abstract This paper aims at studying the conditions of the EU countries with respect the optimum currency area (OCA) criteria. We run two studies in which we use Germany and the euro area as the benchmark respectively. The purpose of the former is to be able to compare the results with previous studies that find a core group and two periphery groups of which the core group is the most converged. We are able to confirm these findings to some extent, but with changes as the Southern European countries have improved significantly. In the second study we use the euro area as a benchmark implying that the results give a picture of the actual situation. We identify four groups of countries with different characteristics, but they are less clear which makes it difficult to make explicit conclusions. By ranking the countries we find however a clear trend in which the EMU countries have dominating positions due to their symmetry in the business cycle. Finally, we also compare the results with those of the Maastricht criteria and we find that such a comparison is subject to a great dispersion implying that we cannot observe any convergence between them. Since official documents on EMU tend to focus only on the potential benefits, academics have an obligation to point out various clouds that may go with the silver lining. Paul Krugman (1993)

3 Table of content Page 1 Introduction Brief history and technicalities of the EMU Research questions and methodology Data Limitations OCA: Theoretical aspects and empirical evidence Classical theory and monetary policy Introduction to OCA theory The OCA criteria Symmetry in the business cycle Openness ratio Wage flexibility Insurance schemes The classical theory OCA theory continued: Implications and empirical evidence Economic integration Empirical evidence OCA critique Benefits Trade effects Sum up Maastricht criteria Implications and propositions Fulfilment of the criteria Sum up Empirical analysis Variables Output disturbances Trade intensity Trade structure Wage flexibility Interpretation of variables Methodology and model Cluster analysis Distance measure Cluster method Critique of model Empirical results Ranking Cluster analysis EU27 and the euro area Results A note on trade theory Cluster analysis

4 4.5.3 Ranking by classification Implications Denmark and the EMU U The Danish OCA: Germany, France and the Netherlands Comparison of criteria: Maastricht versus OCA Shortcomings The variables Extent of the study and suggestions for the future Conclusions

5 1 Introduction Since the establishment of Coal and Steel union in 1951, the European Union (EU), and Europe, has experienced an ongoing integration at a high speed in many aspects regarding legal, political and economic issues. Especially the economic integration in the EU has been remarkable with the single market and the creation of the Economic and Monetary Union (EMU). Intra-regional trade and economic growth has increased at impressive levels implying a high standard of living in most EU15 countries. With the enlarged EU, a new group of countries will hopefully undergo (and many are already) the same development. As part of the economic integration a monetary cooperation was started almost three decades ago with the final result being the EMU formally introduced in Along with the establishment of the EMU came the Maastricht criteria, a set of economic variables with regard to macroeconomic stability that were supposed to act as fulfilment criteria for joining the EMU. These criteria have been widely discussed in various media with the consequence of less focus on the Optimum Currency Area (OCA) criteria developed by the OCA theory. This theory aims at suggesting the extent to which a country (or region) have the right economic conditions for entering a currency area with less emphasis on stability per se and with more focus on flexibility and real economic convergence. Many empirical studies have been made in the name of the OCA theory in the context of the EMU. They have shown somewhat similar results that the EU overall has become more integrated concerning trade intensity and intra-industry trade (IIT) among others and thereby supporting the EMU. They have also proved that regions have become more divergent meaning that countries are not structurally uniform implying that regions from different countries show more similarities in some cases. Most empirical studies investigating the role of OCA in the EU reach a certain consensus in the way that they find a core of countries, Germany and its neighbours, that seem to be in a better position for being part of the EMU, and two periphery groups representing the Southern and Northern Europe respectively whose conditions for being in the EMU are less optimistic. Many of these studies were conducted throughout the 90s prior to the introduction of the EMU in 1999 with the - 3 -

6 intention of providing an indicator of the suitability of entrance to the EMU. The choices of variables are overall similar implying that comparability of the studies is feasible. It has now been some years since the majority of the studies referred to in this paper were conducted and therefore it may be interesting to pursue a similar study based on more recent data, however as many countries have already entered the EMU the purpose will rather concern current conditions than entrance conditions for these countries. This is what we will do in this paper in which we will study the conditions for all 27 EU countries with respect to the EMU regardless of their present status. Furthermore, we will assess them similarly by the Maastricht criteria and compare these two types of criteria. 1.1 Brief history and technicalities of the EMU In this section we will briefly present the history of the EMU along with some technical aspects. This is in order to give the reader a better overview of the EMU (history) and thus a better understanding of the analysis that we will later conduct. As we briefly mentioned in the introduction a monetary cooperation was started in 1979, known as the European Monetary System (EMS). EMS was the beginning of the EMU introducing the composite currency, the European Currency Unit (Ecu) and the Exchange Rate Mechanism (ERM) imposing bands of bilateral fluctuations of 2.25% lasting until 1993 (Zervoyianni et al. 2006). Hereafter the fluctuation bands were raised to 15% which is also the size of the band acting in the ERM II that was introduced in 1999 prior to the EMU. The EMU was formally introduced in 1999, but without notes and coins. The exchange rates were however locked from hereon. Denmark has a special cooperation within the ERM II with a 2.25% fluctuation band as it has not joined the EMU by referendum. Greece acted in ERM II at 15% bands as it was not ready before that and it also joined the real EMU in Today (2008) there are 15 EMU members including the EU15 countries (except for Denmark, Sweden and UK) and Slovenia, Malta and Cyprus. In 2009 Slovakia will join. In addition to this the Baltic countries participate in the ERM II (European Commission 2007). For the outstanding group (except for Denmark, Sweden and UK) it is rather a question of not yet fulfilling the Maastricht - 4 -

7 criteria rather than a political issue as in Denmark. Figure 1 provides an overview of the 27 member states status with regard to the EMU. Figure 1: Member states status concerning the EMU Source: European Commission (2006 & 2007) It appears that the cooperation has been dynamic considering its short history and it is indeed an ongoing process. It seems also remarkable that Slovenia has already joined as Slovakia will join already in 2009 considering its recent past. Their path will most likely be followed by the other new member states. 1.2 Research questions and methodology As we mentioned in the introduction, we endeavour applying the classical OCA variables in our analysis with more recent data however. By using newer data we may be able to overcome (some of) the Lucas-critique suggested in various articles (by Stanley & Rose (1998), Artis (2003), Mann- Quirici (2005)) since our data would be expected to reflect the introduction of the EMU for the member countries

8 Using former studies by Bayoumi and Eichengreen, (B&E), (1997) and Artis and Zhang (2001) among others, as a benchmark we wish to investigate if their findings still hold by applying the same variables, but with more recent data. We also wish to extent the analysis to include the 12 new member countries. However, we do not have a benchmark for this analysis. We ask five questions as following: The Maastricht criteria: 1) How do the 27 EU countries comply with the Maastricht criteria? The OCA framework: 2) Can we confirm previous studies, by using Germany as benchmark and 14 sample countries, regarding creation of groups with different conditions concerning the EMU? 3) By extending the sample countries to include all 27 EU countries and using the euro area as the benchmark, are we able to identify homogenous groups with respect to the OCA criteria? 4) By ranking the countries according to the OCA criteria classified by importance, can we then identify a trend regarding OCA criteria or the countries relation to the EMU? And finally we ask: 5) To what extent do the Maastricht criteria comply with the OCA theory based on our findings? We will apply different methodologies according to the analysis. Regarding the Maastricht criteria we will make a simple interpretation of the results. Concerning the OCA criteria we will use cluster analysis for identification of homogenous groups and for ranking the countries we will make simple calculations of the standardized values. 1.3 Data We have extracted all our data from the eurostat database covering the period of and we apply annual series. We use aggregate GDP data in order to construct the variable regarding output disturbances and for deriving the real output we use the consumer price index (CPI) as deflator. Similarly, we use aggregate export and import data for construction of the variable regarding trade - 6 -

9 intensity and for trade structure we use product series. As a proxy for wages we use the total labour costs that will construct the variable on wage rigidity together with the real output series. For the Maastricht criteria we use the inflation, interest rate, public debt and budget deficit for the Maastricht criteria as already defined in eurostat. 1.4 Limitations The content and scope of this paper are naturally subject to limitations. The paper has two overall purposes; I) To present the OCA theory and implications along with the implied critique and II) to conduct a quantitative analysis based on the OCA theory in order to investigate the 27 EU countries conditions with respect to the EMU. Our focus is therefore the EU however we make references to studies from the US. Furthermore, we wish only to study the economic conditions represented by our data, thus we abstract from any political arguing

10 2 OCA: Theoretical aspects and empirical evidence In this chapter, we will focus on the OCA theory first developed by the theorists Mundell (1961), McKinnon (1963), Kenen (1969) on to more recent literature. In the first section, we will take a textbook approach in order to explain the basic mechanisms that define OCA theory. Later in the chapter we discuss the empirical studies that have been made, mainly in Europe. Finally, we will present critique of the empirical evidence and some implications of the OCA per se and the theory derived suggested by various OCA protagonists. 2.1 Classical theory and monetary policy Classical macroeconomic theory suggests that countries can alleviate asymmetric shocks by using monetary policies, among others, as a tool to restore equilibrium. This can be presented as in figure 2 by using the traditional two-country textbook model exhibiting aggregate demand and supply in Germany and France, two major European economies. The vertical axis shows the pricing level and the horizontal equivalent represents the aggregate output. With inspiration from De Grauwe (2005), we show how a demand shift (thus asymmetric demand shock) affects France and Germany. The supply curve is held constant assuming that wages and input prices remain unchanged. Figure 2: A demand shift Source: Based on De Grauwe (2005) - 8 -

11 Let us assume that the demand shift causes demand to decrease in France and increase in Germany. This implies an output decrease in France and an increase in unemployment. The demand curve in France takes a downward shift from D F to the dotted line. In Germany the opposite case implies the demand curve to shift upward from D G to the dotted line similarly. Before the introduction of the EMU in 1999, such a demand shift could have been overcome by the use of monetary policy 1. There are two ways in which monetary policies can be applied. For a country that operates in a flexible exchange rate system it can either lower its interest rate or devalue its currency against the other country. We show this in figure 3 in which the demand curve shifts back upwards (downwards) for France (Germany) leading to a restoration of equilibrium. Figure 3: Restoring equilibrium with monetary policies Source: Based on De Grauwe (2005) Lowering the interest rate in France (and possibly an increase of interest rate in Germany in order to prevent overheating of the economy) would stimulate borrowing and thereby aggregate demand implying a restoration of output and unemployment. In addition to this, the French franc would be subject to a depreciation in the short run due to an increased money supply and a more attractive German money deposit (but an appreciation (restoration) in the long run - see Box 2.1 below) which would make French products cheaper and thus further boost aggregate demand. The contrary would 1 Indeed other factors can (and do) substitute the role of monetary policy. We will elaborate on these in the next section

12 be the case in Germany. The alternative, a devaluation of the franc, would also boost German demand for French products and thereby restore the French output. Source: Author s own creation with formulas taken from Krugman and Obstfeld (2000) 2 Our two-country example has assumed monetary independency allowing France and Germany to pursue monetary policies in order to mitigate a demand shock by boosting and dampening demand respectively. This adjustment has been possible without causing inflationary pressures on Germany, exactly because of the monetary independency. If Germany and France had been part of a monetary union (which is the case), adjusting by the use of monetary policies might have a flipside effect which is the inflationary pressure that a monetary expansion of the common currency would imply. This is the core issue and dilemma that Mundell (1961) addresses in his discussion about optimum currency areas. 2 We use nominal terms, alternatively the interest rate parity can be constructed using real terms. It would then look as follows: t,t+ 1 t rff E t [ q FF/DM ] = t,t+ 1 t 1+ rdm q FF/DM letting r represent the real interest rate and q denoting the real exchange rate. We should also stress that the interest rate parity is based on the assumption that the relative purchasing power parity (PPP) holds, thus in our France-Germany context we arrive at: t t π FF E FF/DM = t t 1+ π E DM FF/DM

13 2.2 Introduction to OCA theory In the previous section we saw how two countries could counteract an asymmetric shock by the use of monetary policies. In a monetary union, these tools cease to exist separately, but there are other ways through which such a disturbance can be alleviated and equilibrium restored. We will in the following discuss the factors and their qualities The OCA criteria The OCA criteria are the factors that the OCA literature requires to be present for an area in order to form an OCA. The variables (quantity, content and importance) have been subject to an ongoing discussion by scholars (Mundell (1961), McKinnon (1963), Kenen (1969), Krugman (1993), Stanley and Rose (1998), Fidrmuc (2004) among others). The rationale behind the OCA criteria is that these (combined criteria or separate criterion) are necessary factors for an area such as the US or Euroland in order to constitute an OCA. If some of these are not fulfilled, according to the theory, the nations are better off keeping their own currency roughly speaking 3. As above mentioned, there exists no general agreement among scholars as to the scope and content of the criteria however there seems to be a consensus about a set of variables that can be presented as in figure 4. Figure 4: The OCA criteria Source: Author s own representation, but based on various OCA literature 3 These nations would then again be subject to a discussion about whether or not they form an OCA across their regions. We will elaborate on this issue first addressed by Mundell (1961)

14 These variables have often been used in empirical studies as we shall see later and this should support their validity. Figure 4 is an attempt to present them in a simplified and categorized way capturing their quality. The representation of variables and their construction may however be subject to critique that can only be directed at the author. The criteria have different qualities in the sense that some of them are desirable prerequisites (symmetric output shocks) whereas others are adjustment tools such as factor mobility and wage flexibility Symmetry in the business cycle Symmetry in the business cycle is defined as a positive co-movement between the two countries output. This implies that the extent to which output disturbances 4 vary is rather small. The existence of highly correlated business cycles implies then that the currency area can pursue monetary policies with common interest as we saw earlier in the chapter. Monetary expansion in the case of asymmetric shocks will also be able to solve the output problem, but at the expense of inflationary pressure in the fortunate country (i.e. Germany in our example). Thus, the criterion of symmetric shocks is a crucial aspect in terms of monetary policy as a tool for adjusting output disturbances without causing inflationary pressure Openness ratio The openness ratio concerns the share of bilateral trade (often quantified as the sum of exports and imports or only exports) relative to a benchmark (often GDP). The rationale is that when two (or more) countries trade more, they develop an economic interdependency leading to tighter business cycle activity between them. This implies that they become less exposed to asymmetric shocks assuming that this mechanism holds. Thus, for countries with a symmetric business cycle their level of trade intensity seems irrelevant as other factors apparently determine the business cycle. The issue was dealt with by McKinnon (1963) in which he discusses the level of tradables to nontradables 5 goods in relation to the openness of an economy. He claims that specialization of exportables leads to more openness of a country because the domestic consumption of a few goods 4 We do not distinguish between demand and supply disturbances, but it may concern either. 5 We define tradables as goods and services that can be imported or exported (thus traded) and non-tradables as those that cannot. A typical textbook example is a haircut. McKinnon (1963) defines tradables as 1) exportables, which are those goods produced domestically and, in part exported ; 2) importables, which are both produced domestically and import

15 or services is limited, thus a large amount has to be traded. This is the scenario proposed by Krugman (1993) in which he claims that integration leads to industrial specialization that will aggravate the conditions for an OCA because specialization implies asymmetry through industryspecific shocks (more on this in section 2.4.1). This OCA criterion has been subject to a discussion of the endogeneity of the OCA criteria. This is based on the Lucas-critique (Lucas 1973) that generally criticizes empirical studies exactly because they are empirical and thus do not represent the future. Critiques (e.g. Stanley and Rose 1998) claim that joining the EMU would boost the trade after the period measured, thus changing the trade fulfilment of the criteria. Product diversification In our diagrammatic version of the OCA criteria we have added two extra variables on the openness property, IIT and product diversification. The latter is discussed in Kenen (1969) pointing out the importance of a diversified product mix. He claims that having a diversified product mix may alleviate external shocks (see in detail in section 2.3). Intra-industry trade The IIT variable is based on evidence from Fidrmuc (2004) in which he tests the link between IIT trade and business cycles. Using data from OECD countries in the 1990s he finds that bilateral IIT induces the convergence of business cycles. This implies that ITT is crucial in regard to asymmetric shocks instead of trade intensity per se. Fidrmuc thus differ from Kenen in the way that Fidrmuc finds that IIT can alleviate asymmetric shocks whereas Kenen suggests that product diversification can have the same effect, but on shocks in general (not only asymmetric). Thus, if two countries form an OCA, have little IIT, but the exports of the countries are highly diversified, they may experience asymmetric shocks, but the shocks may only have little effect on them due to the diversification. A symmetric shock however may be overcome by use monetary policies Factor mobility We define factor mobility as the mobility of labour and capital mobility. Capital mobility refers to the flexibility to which capital can be transferred to one area from another (see also McKinnon s theory in section 2.3). In this case, capital mobility would prevent potential losses by moving the

16 factors and thereby stimulating further profit in the fortunate country in our example. Unlike labour mobility, it would not necessarily imply large changes in the demographics (see below) with regard to aggregate output, but may alter the per capita income side depending on the movement in labour. But as structural obstacles often prevail, a complete mobility of labour is very unlikely to be the case, in particular in a European context. Labour mobility refers to the ability and flexibility of workers to move from one country (region) to another. In our two-country example from section 2.1, labour mobility would imply that French workers migrate to Germany. They would thereby solve the unemployment problem in France and supply the excess demand for labour in Germany without causing inflationary pressure. Such a situation would, however, change the demographics of the region dramatically. The migration would imply that the former aggregate output level in France was not restored, but increasing the German aggregate output level. On the per capita side, it would not necessarily change anything, but on an aggregate level, it would cause the German economy to grow in total size at the expense of a shrinking French economy. This scenario is presented by Krugman (1993) in which he compares the states of New York and California and finds that such mechanism actually exists, however at a slow pace (see section 2.4.1) Wage flexibility We define wage flexibility as the extent to which wages can fluctuate. Flexible wages induce the option to adjust wage rates according to economic activity and unemployment levels. In our twocountry example from before, having wage flexibility would accordingly smooth the process and enable the two countries to arrive at a new equilibrium as illustrated in figure 5. The mechanism is that unemployed workers in France will lower their wage claims implying lower costs of output and a downward shift in the supply curve. On the contrary, wages will increase in Germany due to the excess demand for labour causing an upward shift in the supply curve. We arrive then at a new equilibrium with output as before, but with a cost of higher price level in Germany

17 Figure 5: Restoring equilibrium with wage flexibility Source: Based on De Grauwe (2005) Insurance schemes Insurance schemes can occur through fiscal transfers or (completely) integrated financial markets. The former concerns the existence of a fiscal policy at federal level that can mitigate asymmetric shocks by fiscal transfers 6. The EU does have regional funds that are distributed to less fortunate regions as a long-term strategy, but in the short term there is no tool to adjust. This may occur at a national level through encouragement of economic activity in poor regions or through unemployment benefits. The size and purpose of the EU budget are also critical factors as it is too small and does not apply to short-term adjustments (De Grauwe 2005). The other insurance scheme is based on a risk-sharing approach. Assuming that financial markets are fully (or at least highly) integrated implies that a situation as in section 2.1 would affect both countries. This is because residents in France hold assets in Germany and vice versa. The output loss experienced in France affects thereby the residents of Germany. De Grauwe (2005) stresses that the role of integrated financial markets may not be sufficient because of the differences in income and private portfolios. Thus, the poor French workers who may already be those most exposed to the demand shift may not hold German assets (or any assets at all). It is thus more likely that it will mitigate the effect for the already well-off French residents. 6 Regarding the EU today, this can occur through domestic governmental policies

18 The OCA criteria are thus factors with different qualities that can be derived from demographics, macroeconomic conditions and direct to political decisions. They need not all of them be fulfilled, but the since fully completion of either one of them seems less likely, the existence of many of them should be most desired. We will in the next section present the founders behind the OCA criteria and this may give us a better understanding of the underlying OCA theory. We have now discussed the OCA criteria in the context of monetary policy in which we have seen the implications of not having independent monetary policies. We have also seen some adjustment tools and other conditions with the purpose of dealing with the implications. In the next section we will take a critical look at the OCA criteria in order to understand their applicability. 2.3 The classical theory It is notoriously acknowledged that Mundell (1961) gave birth to the OCA theory and thereby started an academic debate on the issue. Mundell presented the scenario similar to what we presented in section 2.1 in which a country is subject to a demand shift favouring the other country. He suggests that equilibrium can only be restored through monetary policies while accepting an inflationary pressure in the fortunate region. This holds under the assumption that the government pursues full employment, implying that monetary policy in such case causes an inflationary bias in the multiregional economy. Mundell uses the example of a world with two countries, the US and Canada with separate currencies, and two regions, East and West. In the case of a demand shift that favours West implying an increase in unemployment in East and inflationary pressure in West, there is a dilemma between inflation and unemployment. Inflation can only be escaped at the expense of unemployment, argues Mundell, and vice versa. Under regional East-West currencies the dilemma could be avoided and the problems overcome due to the symmetry within each region. This has created the basis for the first OCA criterion that we saw earlier. Mundell does mention the mitigating role of factor mobility in the context of Europe, but concludes that this is non-existing. McKinnon (1963) took another approach in his analysis of a country s openness stressing the relationship of tradables and non-tradables. In accordance with Mundell (1961) he suggests that the optimality of OCA is defined by the ability of the OCA to solve the objectives of full employment,

19 inflation, but also includes the balance of payments issue. McKinnon further argues that moving from closed economies to open economies, flexible exchange rates become less effective as a control device for external balance and more damaging to internal price-level stability. He suggests that for a country with a low tradables-to-non-tradables ratio, it may be an idea to peg the proportion of non-tradables. In order to improve the trade-balance the exchange rate must be altered by changing the domestic price of the tradables. A currency devaluation would thereby cause domestic prices of tradables to rise, but the effect on the general price index would be less. McKinnon also discusses the aspect of factor mobility. In the situation of a demand shift favouring region A over region B, he stresses the need for factor mobility. If A-type industries can be extended to or developed in region B, the need for factor mobility may not be great. In addition to this, monetary policies through individual currencies can further alleviate the shock. In the case where A-type industries are not transferable as in the above case, factor mobility is then essential in order to prevent a decrease in the unit income of potentially mobile factors of production in region B. It is important again to stress the break-up of factors into labour and capital. Thus, factor mobility may take place either through transfer of capital or by movements of labour. Kenen (1969) developed the theory by Mundell (1961) and McKinnon (1963) by claiming other factors important in the OCA context. He criticized Mundell for his use of terminology with respect to regions as regions in Mundell s terminology are not to be found on a map, but rather functional areas in terms production, technology and demand, thus subject to the same kind of disturbances. This definition though makes sense in the case of the single product region. This simplification, Kenen argues, may lead to powerful results, that may not have been obtained using real geography. Another adjustment to Mundell s analysis is the relation between production and labour. Kenen stresses that in Mundell s definition of regions, labour must be homogenous (in order to satisfy the interregional labour mobility criterion) implying that OCAs have to be small 7. 7 Mundell does stress that the OCA may not be too small to be subject to speculative attacks

20 Using Mundell s two-region example, Kenen, takes an alternative approach concerning the balance of payments and income. He suggests that the increase in demand for the West region s product 8 could lead to a rise in investment in the region, implying higher income and increase in imports causing a deficit in the balance of payments. An increase in import would mean an increase in export of East, thus offsetting the imbalance of income. This situation would then be a natural adjustment process. Kenen also questions the effects of factor mobility asking if it can restore the trade balance between the regions. Kenen advocates the use and utility of fiscal policies for depressed regions arguing that the existence of such can alleviate their situation. He mentions debt issue and federal fiscal transfer as the main tools claiming that regions can more easily borrow (or lend for that matter) in the national capital markets than countries can borrow abroad. He furthermore mentions the existence of federal fiscal transfers in a number of countries, including the US. Kenen also points to the importance of well-diversified economies in terms of the product mix. But he stresses that diversification cannot mitigate all disturbances implying that factor mobility also plays a role in the case of high diversification. Kenen defines two kinds of disturbances; dependent and independent. In the case of the latter, the effect will be small because each export product is quite different from the rest, leaving export earnings stable. But factor mobility may be low. If the export products are close substitutes, the external disturbances may not be independent, but contrary to the other case, factor mobility may be high. Labour and capital may be easily transferable between industries thus serving as an alleviating tool. This may be regarded to be in contrast with Mundell s point about small OCAs. Kenen is more cautious in his analysis of capital formation in the sense that he doesn t regard (domestic) investment as (sufficiently 9 ) mobile necessarily (to some extent contradicting his own argument that close substitutes will increase factor mobility depending on the degree 10 ). He suggests that much will depend on capital intensity in the industries as well as investors 8 And we stress here the explicit use of the singular conjugation of product. 9 By sufficiently we mean to the extent that it can restore equilibrium. 10 We may interpret it with regard to labour only, but it may be necessary with some degree of capital too. Thus, we may choose to interpret his argument as following: Close substitutes increase factor mobility, but in particular with regard to labour, whereas the effect on capital mobility is less certain

21 expectations in terms of the duration of the disturbance. Finally, but not least, Kenen makes an important caveat saying that his analysis does not apply to fluctuations in business cycles since these tend to effect all, but not one, products. 2.4 OCA theory continued: Implications and empirical evidence As we have seen so far, there are many factors and different academic views affecting the OCA theory. In spite of the different analyses and views, it is fair to argue that there is some kind of consensus about the OCA criteria, leaving out the application and importance of them. Having recognized the OCA criteria, in this section we thus turn to the empirical studies of the OCA criteria, some aiming at quantifying the fit of OCA in a European context, some analysing the relation between the variables. However, first we will present some suggested implications of economic integration that is a rather relevant and disputed issue. This issue may belong to the section regarding critique (of the criterion concerning trade intensity) but we choose to elaborate on it separately as it requires deeper explanation and analysis Economic integration The consequences of further economic integration, with the EMU as a culmination, have been widely discussed, but without arriving at a consensus as we just mentioned. We will in the following present the different views on the implied mechanisms. We will concentrate on two opposite views presented by Krugman (1993) and the European Commission (1990). Krugman s view on economic integration has been considered, by some scholars, a pessimist view because he, based on US evidence, believes that the EMU will lead to more asymmetric shocks and thus more asymmetry in the business cycle. Krugman argues that increased economic integration (EMU), likely to imply lower transaction costs, will lead to specialization on a regional (or country-wise 11 ) basis, thus making the EU more likely to experience asymmetric shocks. In order to understand Krugman s argument we may use figure 6 which is based on Krugman (1993). It is a simple illustration of two regions supply curves 11 We should stress that clustering of industries has the same implications whether it concerns regions or countries regarding (lack of) monetary policies. However, governmental intervention is more feasible for countries than regions covering countries

22 in a given industry, CC and C*C*. In this example, demand is assumed to be completely inelastic meaning that total industry output is given by 00, the sum of 0Q and Q0* in the regions respectively. The example also entails that competition is perfect, only subject to location-specific external economies. Krugman then asks us to imagine a situation in which the two regions are self-sufficient yielding figure 6. Figure 6: Supply curves and cost advantage Source: Based on Krugman (1993) The region producing at c has seemingly a cost advantage of c*-c with respect to the other. Krugman s point is that if the transaction costs 12 exceed c*-c, the geographically dispersed equilibrium may be stable. Or put differently, if transaction costs 13 exceed the cost savings between the regions it makes no sense to export from one to the other. But if the opposite holds, it may imply that the region with a cost advantage will produce for the other region, i.e. specialization, and this can be extended to other industries. This leads us to his main conclusion that the EMU 12 Krugman extends his analysis to a matter of setting up a plant in another region subject to a fixed cost versus transaction costs. We leave our example as in the text since it sufficiently demonstrates the point. 13 In terms of the EMU these could be with respect to changing currencies, but in general economic it extends to others such as tariffs etc

23 implying a reduction of transaction costs will cause the EU to become specialized causing regional or national concentrations of industries. The European Commission (1990) has an opposite view on the mechanism following the EMU. It presents evidence suggesting that IIT is already prevalent within the EU implying little specialization. This is supported later in Fidrmuc (2004) who also reports high levels of IIT within EU15 and moderate levels of IIT between the CEECs and EU15. High IIT intensity is less relevant at baseline following Krugman s scenario implying that it will change though. Several decades of economic integration in the EU however may serve as a good indicator of the current IIT levels. Former work by the commission finds a negative correlation between trade barriers and symmetric shocks implying that an elimination of trade barriers will lead to more symmetry among shocks due to more intra-industry trade. It believes thus that intra-industry trade is encouraged by economies of scale and product differentiation factors which are accordingly obstructed by trade barriers. Thus, a removal of these, e.g. through EMU, will encourage further intra-industry trade (European Commission 1990). Krugman stresses the obvious costs and benefits from industry concentration claiming that the latter creates better conditions for benefiting from external economies 14. On the other hand he recognizes the exposure to which less diversified economies will become more subject, quantified by technology and demand shocks. In the presence of high factor mobility it will create divergence in long-term growth rates. Krugman presents an example of a region with immobile factors, but assuming wage flexibility 15 in the case of an external demand shock. He suggests that the wage flexibility and immobile factors would create a mean-reversion mechanism in which wages and factor costs decline as a result of the recessionary state, thus attracting other investment and industries due to its lower costs. This is similar to the mechanism presented in section We interpret this as external economies of scale (contrary to internal economies of scale conditions from which a company can benefit in terms of e.g. infrastructure implying lower transportation costs). External are those from which a whole industry can benefit, but in the case of concentration there may be even better prospects for the company(ies) in question exactly because of the concentration. 15 It may be argued that this condition should more explicitly be emphasized

24 He also measures regional concentration of industries in four big regions 16 in the US and compares them to the four big European countries, Germany, France, Italy and Spain. His model is an industry-concentration index: s i * i s i where s i is the share of industry i in total manufacturing employment in a region/country and s i * is the equivalent in another region/country. Krugman then finds a higher concentration in the US than in Europe. His analogy is that due to higher integration, the US is more specialized and this will apply to Europe too in case of the EMU. Earlier we mentioned that Krugman suggested that his scenario would create long-term growth divergence. This is based on empirical evidence ( ) from two US states with historically similar growth in per capita income, California and New York. He finds that for these regions labour mobility is the adjusting mechanism to asymmetric shocks. He reports that in California the employment rate rose relative to that of New York by 86% in the period examined implying that unemployment in one New York caused workers to migrate to California and thereby establishing the natural rate of employment given the conditions. In brief, equilibrium is established by migration (not wage flexibility or monetary policies) restoring unemployment and per capita income, but on an aggregate basis the growth rates are diverging. However, this mechanism is a rather slow transition and may take years. Krugman concludes that this example is the unfortunate case for the US because it doesn t solve structural problems in the unfortunate regions. He also mentions that this mechanism has its flaws, among others immobile workers and refers to the Mezzogiorno 17 problem. He also believes that labour mobility may be a problem in the EU. This point is supported by the European Commission (1990) that finds little mobility in the EU compared to the US and believes that overall it is not very feasible due to cultural (e.g. language) barriers, but also social security systems. 16 These are made up by Krugman for the purpose and consist of states. The regions are North-East, Mid-West, South and West. 17 The Mezzogiorno (in English: Mid-day) problem refers to the strong division of economic activity in Italy with the Southern part being far behind the Northern part. Mezzogiorno is an ancient term referring to the Southern part of Italy starting below Lazio

25 We have now seen two different views on the mechanism between economic integration and asymmetric shocks. Krugman s theory is indeed a useful contribution supplemented by evidence from the US. The opposite view is also arguable and it enjoys empirical support from the EU itself Empirical evidence This section deals with empirical studies explaining and investigating the conditions and use of the OCA theory. We will discuss these in this section and use them as an inspiration in our empirical part. De Grauwe & Vanhaverbeke, (DG&V), (1991) investigate the case of Europe by using national and regional data 18 in the period They test real exchange rate movements, output and employment growth, labour mobility and unemployment. They find that real exchange rate variability is twice as high in nations compared to regions. They also find that real exchange rate variability is lower in EMS-countries than non-ems countries. Studying the interregional labour mobility, they find a much lower mobility in the Southern European countries (Italy and Spain) than in the North (France, UK, Western Germany, Netherlands, and Denmark). This is surprising in the light of the high regional divergence in per capita income in the South compared to Northern Europe which should accordingly encourage migration however this may be the exact cause. This is the problem that Krugman (1993) referred to discussing the Mezzogiorno issue of Italy. Some of the difference between labour mobility, they claim, should also be found in the sizes of the country. E.g. moving from Catalonia to Andalucia is a bigger change than moving inside Denmark which is smaller than these regions, situated in the opposite parts of Spain, regarding population. They also deal with potential aggregation bias, i.e. differences in regional size, since the regions are of more or less the same size in the study. This however does not apply to the smaller countries. Studying divergences in output growth, DG&V find that in the long run regions (within nations) tend to have a higher divergence than countries implying that economic development is more 18 The national data come from nine European industrialized countries plus the US and Japan and the regional data come from five major European countries

26 unequal within countries than between countries 19. Finally, they measure growth in employment and relate it to unemployment rates. They find higher long-run divergence in employment in regions than countries. In particular Spain and Italy show high divergence and this may explain their large variations in unemployment rates regionally. This may also support the previous evidence on immobility in labour and thereby the divergence in output growth. Additionally, DG&V study the correlation between real exchange rate variations and output growth. They find that the correlation is stronger and more significant at regional level than national level (short-run and long-run) suggesting that even though real exchange variability is lower regionally than nationally, it plays indeed a significant role in the adjustment process. Accordingly, the future Europe with the EMU can be regarded bilaterally; one side supported by the fact that the degree of national disturbances is low and thereby implying a decline in asymmetric shocks as Europe becomes more integrated. This view, however, disregards the regional differences by considering aggregates and this may be subject to critique considering the sizes of countries and regions. This leads us to other view claiming that asymmetric shocks will not decline regionally in an integrated Europe where borders (not from the OCA terminology) tend to become less important with regard to the economic development. B&E (1993) report higher variability in real exchange rates in Canada than in the EU and suggest that this may be a good proxy for the EMU. That is, since Canada is running a successful monetary union the EU should be able to do so accordingly. They also measure asymmetric disturbances in the EU and US. They find that asymmetric shocks occur in the EU and two groups can be identified experiencing the shocks differently. The core group consisting of Germany, France, Belgium, the Netherlands and Denmark experience smaller shocks with higher correlations between them. The opposite was the case for the periphery countries, the UK, Italy, Spain, Portugal, Ireland and Greece. Using principal component analysis, they find that more of the variation with regards to inflation and output growth in the US can be explained than in the EU. They also find that the US states are more inter-correlated than the EU countries are. Finally, the report faster adjustment in the US states than the EU countries, and they believe that it may reflect higher factor mobility. 19 In the short run, the picture is more blurred and there is no clear trend

27 Fatás (1997) studies business cycles quantified by growth in employment in EU over time during the EMS. He finds increased correlations between regions and the EU and declining correlations between regions and their country suggesting that the EU is becoming more integrated. However, the correlations between regions and countries are still greater than those between regions and the EU in the recent period. In Italy, however, there is almost no difference 20. Overall, Fatás reports increased correlations between countries and EU in his sample period. B&E (1997) create an OCA index based on the classical OCA criteria with the purpose of quantifying European countries suitability for the EMU. They estimate a regression equation in which the dependent variable reflects variation in bilateral nominal exchange rates, and the independent variables represent output variation, trade structure of exports, trade intensity and size of the economies respectively. The intuition is that the independent variables explain better the suitability for the EMU than exchange rates (those quantifying the fit) do. They run the regression against Germany which is used as a proxy (benchmark) for the EMU-area due to its economic influence. Their results (extrapolates for 1987 and 1995) show three groups of countries; A converged group for which the EMU is indeed suitable, a converging group moving in the right direction with respect to convergence and other countries showing little convergence. The first group consists of Austria, Belgium, the Netherlands, Ireland and Switzerland. The second group is constituted by UK, Denmark, Finland, Norway and France as the surprising aspect of not being in the first. The outsider group includes Italy, Greece, Portugal and Spain. They also show bilateral relationship for other countries indicating that Italy and Spain show convergence vis-à-vis France and the same goes for Finland and Sweden. The implication of this is that these countries may rather form a currency area with each other. Artis and Zhang (2001) study the OCA conditions in Europe using cluster analysis also with a Germany benchmark. They use six variables regarding synchronization of the business cycle, volatility of the real exchange rate, synchronization of the real interest rate, openness to trade, convergence of inflation and labour market flexibility proxied by a measure of employment 20 Fatás also finds that the prosperous Italian region Lombardia is closer correlated with Germany than Southern Italy. Thus, from this perspective, he claims, an OCA with Northern Italy and Germany is more obvious than an OCA in Italy

28 protection legislation. The variables inflation and real interest rate thus share the Maastricht criteria. Their findings support those of B&E (1997) in which they identify three clusters (groups), however with the exception of France as part of the core group. The core group which consists of France, Austria, Belgium and the Netherlands, and a northern and southern periphery group with the Scandinavian countries plus UK and Ireland in the former and Spain, Italy, Greece and Portugal in the latter. The features of the core group include a high business cycle correlation and high trade intensity with Germany. The northern periphery group characterized by a less synchronized business cycle and less protection in the labour market whereas the southern group is subject to dispersion of inflation and interest rate and high protection in the labour markets. Bénassy and Révil (2000) use the same approach as B&E (1997) in their attempt to create an equivalent index for the CEECs in comparing the EMU and USD. However, they leave out the trade-variable arguing that small countries are more open (i.e. trade more) than large countries and this condition is thus accounted for in the SIZE-variable 21, but adding a dummy accounting for specific CEEC behaviour. They find that the CEECs have paid too much attention to the USD over the EMU. The EMU should accordingly be more dominant in their currency baskets than past and present show. This has changed for many CEECs since the time of the article however it simply supports the CEECs path towards the EMU. The empirical evidence shows somewhat unanimous results concerning the EU s development with respect to economic integration and its candidacy as an OCA. We have seen that regions show greater divergence between each other than do countries and even though they experience lower real exchange rate variability than countries do, it seems that it shows still an important role because of its close correlation with output. The empirical evidence shows to some extent a consensus in terms of clustering of OCA. B&E (1997) and Artis and Zhang (2001) agree that there is an OCA core consisting of Germany, the Benelux, Austria and sometimes France, Denmark and Ireland. In the same way they find that there are two periphery groups (perhaps to different extents) representing the Northern and Southern Europe. 21 However, as B&E (1997) point out, this does not always hold, e.g. comparing Germany (a large and open country) with Spain (a smaller and more closed country)

29 2.4.3 OCA critique Frankel and Rose (1998) take a critical approach with respect to empirical studies of Europe as an OCA. First they measure the relationship between trade intensity and business cycles. They find a strong positive relationship between trade intensity and business cycle activity confirming what they term the OCA endogeneity, that is, a change in one variable (e.g. trade) will lead to closer business cycle activity. This supports DG&V (1991) and Fatás (1997) in the sense that a more integrated EU has converged business cycle activity. Their critique is not based on this particular relationship, but is a mere Lucas-critique of various OCA criteria which criticises econometric models for their insufficient forecasting of effects of policy changes due to lack of microfoundations. In our context, some countries (e.g. the less converged groups as identified by B&E) might not look like obvious candidates for the EMU at the time of the study, but may look better in the future as integration (trade) apparently drives business cycle convergence. This critique can also be applied to the case of the Maastricht criteria or stability pact subjects that we will discuss later. Fidrmuc (2004) develops further on the OCA endogeneity suggested by Frankel and Rose (1998). Fidrmuc investigates the trade intensity-business cycle activity relationship and confirms Frankel and Rose s (1998) findings, but he expands it and finds that this is with respect to IIT. It is then trade structure quantified by IIT that drives business cycle activity and not trade intensity per se that does 22. This factor may have been underestimated due to the fact that EU intra-trade is dominated by IIT. In the context of the accession countries it may turn out crucial what relationship is more dominant. However, Fidrmuc reports relative (to within EU15) high shares of IIT between EU15 and the CEECs. Mann-Quirici (2005) also criticizes the traditional empirical OCA studies that we presented above claiming that they reflect little evidence of the future in accordance with the Lucas-critique as suggested by Stanley and Rose (1998) in the OCA context. He suggests that the US is a better proxy than EU itself (empirically) for investigating an OCA in EU. His hypothesis is that a policy change that eliminate one margin (variable) for macroeconomic adjustment (namely: the nominal exchange rate) should encourage adjustment on others, thereby bringing back equilibrium. Mann-Quirici 22 We should perhaps stress that in order for IIT to become prevalent, a certain level of trade is necessary. That is, if bilateral trade is non-existing so becomes the trade structure

30 tests the role of real wages as adjustment tool in the absence of monetary polices and argues that it is a better adjustment tool (compared to e.g. labour mobility and fiscal transfers) because it is faster than labour mobility which becomes sticky once migration patterns are established. He finds a positive relationship between wages and output in the US and he claims that a similar development will take place in European countries after the introduction of the EMU as a consequence of the need for flexibility. Finally he suggests that fiscal transfers (domestically) may work as primary stabilizers in the short run which is supported, he claims, by the large GDP contribution in the public sector. Artis (2003) evaluates the OCA theory with respect to the Maastricht criteria. He distinguishes between political requirements and economic rationality with reference to the Maastricht criteria and OCA criteria respectively. He mentions that EMU membership is often assumed desirable without regarding the OCA conditions, but focusing only on the Maastricht criteria. Artis also comments on the Lucas-critique and suggests that idiosyncratic shocks, e.g. in UK, are due to idiosyncratic policies, in the sense that pursuing independent monetary policies implies independent shocks. This above critique suggests that one needs to be cautious about the interpretation of the empirical studies made about. Even though many of the empirical studies show similar results in their grouping of countries with respect to the OCA criteria, they may prove useless if the input-variables do not represent the right conditions or assumptions. One should also be wary of the mechanisms implied by further integration with regard to Krugman s (1993) analysis. A counterargument may be that specialization does not come from one day to another like economic integration is an ongoing dynamic process tracking decades of European history as suggested by Berger and Nitsch (2005) Benefits As the OCA theory mostly concerns how to cope with the costs implied by a common currency, we will in this section discuss the benefits and a few of the costs not yet mentioned. We will not make a cost-benefit consideration as we only make our conclusions from the OCA theory, but the benefits 23 See their analysis under trade effects

31 should of course be kept in mind. Some of the costs we did not discuss are following De Grauwe (2005) a large one-time fixed cost associated with the conversion between currencies, price increases in several EMU-countries as a result of the EMU and this must be regarded as a (onetime) cost from a consumer point of view and as we already mentioned the loss of monetary policies that can cause dilemmas with regard to inflation, output and unemployment unless some of the OCA criteria are fulfilled. Considering the benefits, De Grauwe mentions the elimination of transaction costs to be a large benefit, in spite of its relative size to GDP, and indirectly he points to the issue of price transparency for consumers, however, this is not quite the case reporting a great dispersion of prices between the EMU-countries, even for tradables. He also suggests that the elimination of uncertainty regarding exchange rates will lead to a welfare gain in a world populated by risk-adverse individuals. Or put differently, business opportunities that might have been rejected due to uncertainty about exchange rates and thereby revenue may in an EMU be exploited causing output to increase. The European Commission (1990) suggests that the macroeconomic situation in a union of one currency will be stabilized with respect to inflation, leading to a lower cost of capital, and output variability. Based on simulation studies it finds that the more integrated a monetary union, the more benefits from stability with regard to the macroeconomic situation will be the result. In an assessment of the first five years of the EMU the European Commission (2005) reports a decline in inflation and budget deficits and a historical low level of interest rates. Bayoumi, Eichengreen and Von Hagen, (B&E&VH), (1997), discuss the costs and benefits and find overall that it is indeed difficult to quantify these questioning some of the mechanisms assumed in cost-benefit analyses such as the relationship between trade and output. Following their view, the elimination of exchange rate risk in the long-run 24 as a result of the EMU will encourage trade. But to what extent trade will lead to output growth is more uncertain they claim, and convincing analysis is just not possible due to the lack of knowledge on the subject. They also find that labour mobility as a stabilizing tool is less prevalent in the EU than in the US and in this respect less suitable as an OCA variable supporting DG&V (1991) whose study was 24 Short-run exchange rate risk is hedgable by the use of derivatives

32 only regionally and would perhaps be expected even less suitable considering migration between nations. B&E&VH study the extent to which fiscal policies can be applied (not transfers, but debt issuance) and they suggest that it may be easier for sub-central governments in federations than subcentral governments in unitary states to issue debt even though they advocate the opposite as subcentral governments in federations have larger budgetary power than those of unitary states. De Grauwe and Schnabl, (DG&S), (2004), also evaluate the costs and benefits of the EMU in the light of the CEECs suggesting that trade will be positively affected by the elimination of exchange rate volatility which will likely also decrease the risk-premium and interest causing a boost in the economy. They assume thereby a positive link between trade and output as questioned by B&E&VH. They also mention integration of financial markets between EU15 and the CEECs as a mean of insurance against asymmetric shocks. As we have seen there several aspects of the benefit, and cost, side to be considered and some of these are subject to disagreement about their scope and implications. However, there seems to a consensus about the EMU s effect on macroeconomic stability along with the savings from transaction costs 25. There also seems to be a consensus about the obscurity of the relation between economic integration and (a)symmetry of shocks. Furthermore there is an agreement that certain conditions, e.g. factor mobility and wage flexibility, may alleviate asymmetric shocks, but the degree of their effect (in particular in the EU) may still be subject to discussion Trade effects As we have mentioned several times so far in this chapter, an increase in trade is an inevitable consequence from a further integrated market and the EMU. There seems to be a consensus among the authors used in this chapter about this issue. Even though there are many other positive effects from the EMU which we have discussed just above it seems appropriate to devote a minor section to trade in which we will briefly present empirical evidence. Using a gravity model Berger and Nitsch (2005) investigate whether there has been an increase in trade. They use historical data from of 22 industrialized countries. Thus, they are able to 25 We do not deliberately mention trade and welfare effects since there is doubt about the degree of their effect and inter-relation. We do discuss trade effects more profound in the following section

33 adjust for time trends. Taking a short-run approach, they find that, ceteris paribus, trade has increased 21% more between EMU countries than between non-emu countries over the last decade. But, when accounting for the trend in integration over time, the effect disappears. They argue that this performance is rather a culmination of past economic integration prior to the introduction of the EMU. Therefore other countries considering joining the EMU (or any currency union) should be cautious about expectations unless there has been taken proper account of the dynamics underlying the European institutions. This would apply to the CEECs. Rose and Stanley (2005) study trade effects using meta-regression analysis in which they evaluate 34 previous studies. They report overall positive trade effects, some very large, in the interval 30-90%. Adjusting for publication bias they arrive at a lower estimate, 20-80%, but still indeed very large. They acknowledge that their analysis has weaknesses in the sense, that in case of a general systematic bias in the studies, it won t be able to distinguish it from an authentic empirical effect. They believe, however, that their results are valid, but their extent might be limited. 2.5 Sum up This chapter introduced us to the OCA theory and its implications. We discussed the (measurable) variables that constitute the OCA criteria including business cycle convergence, trade, factor mobility, wage flexibility and insurance schemes. We learned that there is a general consensus about these however disagreement about their scope and importance still remains. We have seen that there are different views on economic integration and thereby the adoption of a common currency with regard to business cycle symmetry. Krugman (1993) believed that evidence from the US could be applied to Europe implying that the EMU would lead to more geographical specialization and causing asymmetric shocks. He also found that asymmetric shocks in the US were overcome by labour mobility. This was however in contrast to Mann-Quirici (2005) 26 who reported wage flexibility to be the prevailing mitigation factor in the US. As a contrast to Krugman, the European Commission advocated that the EMU would encourage IIT and thereby reduce the member countries exposure to asymmetric disturbances. 26 It is however notable that Mann-Quirici does not include labour mobility in his study because it is not (as) feasible for measurement (as wage flexibility)

34 We have seen a number of empirical studies with the purpose of identifying the conditions for a common currency in Europe by applying the OCA theory. DG&V (1991) reported higher divergence between regions within countries than between countries concerning output growth which could imply a more integrated Europe where national frontiers are less present. This was supported by Fatás (1997) who reported increased correlations between regions and the EU and declining between regions and their country by using employment growth. DG&V also found substantially higher labour mobility in the Northern countries than in the Southern countries. This advocates an OCA in the Northern countries, but argues against an equivalent including the Southern countries. B&E (1997) and Artis & Zhang (2001) conducted studies applying multiple OCA variables. Their findings were similar in the sense that they identified groups of countries according to their performance on the OCA variables. Both of the studies found a core group consisting of Germany, the Benelux countries, Austria and Switzerland (and Ireland/France) depending on the study. In addition to these Bénassy & Révil found that the euro should be the dominating currency for the CEECs (not included in previous studies) with regard to currency pegs, baskets and the like compared to the USD

35 3 Maastricht criteria So far we have discussed the OCA theory and empirical studies. We have seen that there are certain criteria that must be fulfilled for a number of countries (regions) in order to constitute an OCA according to the theory and these criteria did not hold for all the EMU countries. This section is concentrated around the Maastricht (or convergence) criteria derived from the Maastricht treaty of 1992, criteria that need to be satisfied for a country to enter the EMU. We will in this section briefly present the criteria and discuss their implications. The criteria can be split into a set of formal criteria measuring macroeconomic harmonization and convergence and a set of less clearly defined criteria concerning real convergence regarding income level, balance of payments and additional price and cost indices. The latter are therefore difficult to measure and we will in the following concentrate on the formal criteria. The formal criteria are following the Maastricht Treaty with protocols (Eurotreaties 1992): Government budgetary position: The budget deficit 12 months prior the date of assessment should not exceed the reference value which is a constant 3% of GDP. Public debt: Public debt 12 months prior to the date of assessment should not exceed the reference value which is a constant 60% of GDP. Inflation: Annual inflation 12 months prior to the date of assessment should not exceed the reference value defined as the average inflation rate of the three EU member states (not just EMU members) with the lowest inflation by more then 1.5 percentage points. However, in practice if countries experience a negative inflation they are not used as reference country. Interest rate: The nominal long-term interest rate 12 months prior to the date of assessment should not exceed the reference value defined as the average nominal long-term interest rate of the three best performers with regard to inflation by more then 2 percentage points. Longterm interest is defined as long-term government bonds or comparable securities taking into account differences in national definitions. Stability of exchange rate: The currency has to participate in the ERM II for at least two years and in this time respect the normal fluctuation margins without severe tensions and bilateral devaluation against any other member state on its own initiative This criterion is left out in our evaluation because not all 27 member countries have participated in the ERM II

36 As it can be seen the criteria concerning debt and budget are constant ratios, but variable in the sense that the GDP changes from time to time. The inflation and interest rate criteria are variable with a fixed margin and these are based on the macroeconomic situation in the EU represented by the two indicators. It should perhaps also be mentioned that violation of the criteria can still lead to accession to the EMU as exceptions. Evaluation may also take into account the other criteria on real convergence, and this has indeed been the case with some of the current EMU members. The flexibility of the performance implying a liberal interpretation of the criteria on deficit and debt 28, argues Thygesen (1993), made it also easier for the member countries to accept. As a continuation of convergence after the introduction of the EMU, the stability and growth pact has been established. It is a continuation of the budget and deficit criteria including a surveillance of the performance of countries. It allows the budget deficit to exceed the 3% if it is considered exceptional meaning that it is due to unfavourable economic conditions such as natural disasters or economic downturn. The pact also imposes the possibility of fining countries in case of excessive deficits. De Grauwe (2005) suggests that one can interpret the pact concerning public debt as it forces countries not only to stay below the 60%, but also to bring down the debt to zero. He also criticises their inflexibility suggesting that they should more reflect economic cycles and conditions. This issue has also been recognized within the European Commission. 3.1 Implications and propositions Thygesen (1993) discusses the convergence criteria and presents EMU indicators based on 1991 data. The indicator is subjectively weighted, emphasizing inflation and public debt, and comprises budget deficit, public debt, inflation and unemployment. The results show that Germany, France, the Netherlands and Denmark represent the core of top-performers followed by the UK. Portugal, Belgium, Spain, Ireland and Italy define the bottom whereas Greece is in a situation of divergence all by herself Thygesen finds. Lavrac (2004) compares the ten CEECs (excl. Romania and Bulgaria) conditions for entering the EMU with respect to the current (of 2004) 12 members equivalent at a similar point in time (five 28 The following quotation made the budgetary criteria less objectionable argues Thygesen (1993): If the deficit ratio has declined substantially and continuously to a level close to the reference value, or if the ratio only exceeds this value exceptionally and temporarily, performance could still be regarded as acceptable. A similar formulation counts for the debt ratio, but with no mentioning of being close to the reference value

37 years) prior to the introduction of the EMU. He studies four of the criteria 29 ; fiscal deficit, public debt, inflation and the long-term interest rate. The data show that the CEECs are not far from fulfilling the criteria with average values of all the countries meeting three of the four criteria. Two countries, Latvia and Lithuania, already met the criteria. The variable lacking is the fiscal deficit where the average value is well above the allowed 3% and with many countries not qualifying. Considering the data for the 12 EMU member countries in 1994, their performance was rather ambiguous. Only two of them, Germany and Luxembourg, met the criteria, but the four southern countries, Spain, Italy, Greece and Portugal, did not even meet any of them. Lavrac therefore concludes that the CEECs are at least as qualified as were the 12 already members at a similar point in time prior to entry. He mentions the real convergence though, of e.g. GDP, which may have been different for the four southern countries and CEECs. Mundell (1997) also discusses the EMU in the light of the Maastricht criteria with respect to its stabilizing effect. He suggests that the benefit of stabilization in some countries that do not meet the Maastricht requirements, e.g. Greece and Portugal, is underestimated. He thereby advocates the flexibility of not meeting the criteria, also for political reasons, thus allowing e.g. Greece to enter EMU is the best way to obtain macroeconomic stabilization in the country. Babetski et al. (2004) investigate the CEECs performance concerning real convergence. They find that the CEECs show convergence concerning relative GDP level and business cycle activity, however at different rates of pace. This is in accordance with Lavrac (2004) and their most recent performance as we shall see in the next section. 3.2 Fulfilment of the criteria In this section we investigate the status of all the 27 EU countries regarding the Maastricht criteria as of 2006 as shown in table 1. This is not a fair comparison as the EMU-countries should not be 29 The criterion on exchange rate variability is left out because the CEECs were neither EU, nor EMS II members at the time of writing the article

38 evaluated by these criteria, but by the growth and stability pact that concerns the budget deficit and public debt with the same reference values. Table 1: The 27 EU countries performance on Maastricht criteria Source: Author s own creation Note: Shaded areas show fulfilment of the Maastricht criteria. Thus, the inflation and interest rates are not relevant for these countries, however it seems that, measured by the Maastricht criteria, only Spain and Greece exceed the limit. Furthermore, we are aware that the criteria should be measured 12 months prior to assessment which is impossible as some of the countries are not planning entrance in the near future 30. Thus, we use it as a hypothetical situation where the countries are being assessed simultaneously having in mind their different status. 30 For the already members, older data could have been used prior to their entrance

39 Overall, it is striking that the CEECs fulfil most of the criteria considering their different status regarding the EMU and their economic conditions. Concerning the public debt it is striking to see the sizes in Italy and Greece and it should be noted that Belgium in spite of the still high public debt has decreased it significantly from 1994 with Lavrac reporting 136% of GDP. Cyprus, Malta and Slovakia all perform well which is also required since they are prone to enter the EMU in 2008 and 2009 respectively. The Baltics perform well too perhaps reflecting their participation in the ERM II. 3.3 Sum up We have now discussed the Maastricht criteria and their implications. We saw that they concern macroeconomic stability and after entrance to the EMU there is a continuation through the growth and stability pact regarding public debt and budget deficit as the interest rate and inflation should be kept at a desired pace by the common monetary policy. We also saw that many of the CEECs are close to qualifying for the formal criteria if they were to be assessed and that some of incumbent EMU countries like Italy and Greece still were struggling with the fulfilment of public debt. However, the criteria (incl. growth and stability pact) are flexible implying that EMU countries undergoing an unfortunate economic cycle may exceed the debt/deficit in order to stimulate the economy by an expansionary fiscal policy

40 4 Empirical analysis In this section we examine the EU countries conditions with respect to the EMU by applying the OCA criteria 31. We should stress that we study each country s conditions and not the EU as a whole. We do therefore not aim at identifying a European OCA. Instead we try to assess to what extent a country fulfils the OCA criteria with regard to the euro area and in our cluster analysis we may identify groups with similar characteristics. On total we conduct four minor studies (summarized in table 2) in which we 1) rank the countries based on the mean of the standardized values of the criteria 2) and 3) identify homogeneous groups and 4) rank the countries based on the classification of the variables prior to the cluster analysis. For number 1 and 2 Germany is the benchmark and for 3 and 4 the euro area is applied as benchmark. Table 2: Overview of empirical analyses Source: Author s own creation We have extracted all the data from Eurostat in order to limit data-based errors in the study. We use annual 10-year series covering reflecting the most recent available data, and thus covering most of the EMU period. 4.1 Variables We have based our variables on the OCA literature and in particular the previous studies by B&E (1997) and Artis and Zhang (2001) as these two will be used for comparison in 1) and 2). Some of their variables are however subject to insufficiencies of data 32 or economic arguing. We thus arrive at four variables including disturbances in output, bilateral trade, intra-industry trade and flexibility of wages. Since these are not (exactly) the same as our benchmark studies we stress that this is per se problematic when comparing results. The purpose of comparison is however just to make a rough benchmarking and we believe that it is not crucial for our study. We choose output disturbances as a 31 By the OCA criteria we mean those four that we have chosen. These will be presented later in this chapter. 32 Due to the period in question there are obviously no nominal exchange rates as used by B&E. Other variables, e.g. on migration, have not been possible to obtain for all the countries

41 proxy for the extent of business cycle symmetry which is considered the crucial variable regarding the OCA framework. We also include the flexibility of wages, quantified by the correlation between the real output and the total labour costs, in order to measure the extent to which adjustment can happen as proposed by Mann-Quirici, since the inter-regional labour mobility is limited in Europe as already discussed. Finally, we use two trade variables regarding the level of bilateral trade and the type of trade in order to measure the exposure to which symmetric business cycles can be induced following the theory. We would also have liked to include a variable regarding fiscal insurance schemes, but this was rather difficult as there is little literature on this in terms construction and empirical evidence Output disturbances Following B&E (1997) we measure output disturbances relative to the benchmark 33 as the standard deviation of the difference in the logarithm of real output between country i and the benchmark. More formally it can be presented as in equation (1) N N 2 σ ΔY = ( ΔYi ΔY ) where j = benchmark (1) i, j j Source: Based on B&E (1997) And ΔY i and ΔY j represent the year to year difference of the change in real log output in country i and the benchmark. This variable is supposed to reflect (a)symmetry in the business cycle between the sample country and the benchmark. Thus, if this variable displays a small value, it signifies that the concerned country has a rather symmetric business cycle with the benchmark and vice versa for a high value Trade intensity Trade intensity is supposed to reflect the openness to economic integration and thus exposure to the common business cycle of a country as measured by the sum of trade to GDP. Following Bénassy & Révil (2000) it is defined as the total sum of export and import of country i to the currency area 33 That is either Germany or the euro area. 34 Note that throughout the paper we adjust for N and not N-1 as we regard our sample as the population

42 relative to a benchmark. A common used benchmark is the GDP which we will also apply. The variable is therefore constructed as following: Export ij + Importij 100 ( ), where j = benchmark (2) GDP Source: Based on Benassy & Révil (2000) i And export ij and import ij denote the total sum of export and import respectively between country i and the benchmark. We divide it with the GDP in order to get a relative measure that tells us the importance of trade in relation to the economy. Thus, the more trade accounts of the economy, the more we should expect the business cycle of the country and the benchmark to converge. This variable was empirically tested by Stanley and Rose (1998) who found a positive relation between bilateral trade and output growth Trade structure As we discussed in chapter 2, this variable was proven by Fidrmuc (2004) who reported a positive relation between IIT (and not only trade) and economic activity meaning that a similar trade structure thus reduces the probability of asymmetric (industry-specific) shocks. Following B&E (1997) we define similar trade as small differences in the sum of trade (export and import) for different product groups relative to a benchmark (total export or import). Large differences thereby imply dissimilarities in the trade structure. More formally it can be shown by the following expression: 1 k k exp ortij import ( total k exp ort import ij k ij total ij 2 ) where j = benchmark (3) Source: Based on B&E (1997) The left-hand side within the parentheses represents the share of bilateral exports of product k 35 in proportion to total bilateral exports from country i to the benchmark. The right-hand side within the 35 We use the main SITC groups from Eurostat; food and live animals; beverages and tobacco; crude materials, inedible except fuels; mineral fuels, lubricants and related materials; animal and vegetable oil, fats and waxes; chemicals and

43 parentheses measures the share of bilateral imports of product k in proportion to total bilateral imports from the benchmark to country i. We take the square of the sum of differences and the square root of that in order to obtain only positive values. We adjust for the number of products by multiplying with 1/K and we thereby calculate the average share for all the products Wage flexibility Following DG&V (1991) we construct a variable reflecting the flexibility of wages by regarding the correlation between the labour costs and real output growth. The equation for calculating the correlation is shown in equation (4) in which we take the correlation of the year-to-year change in the labour costs and the real output growth. σ LC, Y ρ LC, Y = (4) σ σ LC Y Source: Watsham and Parramore (2005) Where the correlation, ρ LC, Y, is equal to the covariance, σ R, Y, divided by the standard deviations multiplied by each other and the covariance can be expressed as: ( Yi Y ) ( LCi LC) σ R, Y (5) N Source: Watsham and Parramore (2005) Where Y i and LC i represent the year-to-year changes in our period of the real output and the labour costs respectively and N is the number of observations. The standard deviation is found as in equation (1). 4.2 Interpretation of variables In this section we will briefly discuss the variables with emphasis on how to interpret them. The purpose of this is to explicitly ensure the meaning of the results. The qualities of the variables are shown in figure 8 in which we try to present them while displaying their interrelations. related products, N.E.S.; manufactured good s classified chiefly by material; machinery and transport equipment; miscellaneous manufactured articles; commodities and transactions not classified elsewhere

44 Figure 8: The mechanism of the OCA criteria Source: Author s own creation As we already discussed, the output disturbance variable is supposed to reflect the extent of (a)symmetry in the business cycle. A small value indicates a more symmetric business cycle with the benchmark. If this criterion is fulfilled, the importance of the other variables reduces as they are supposed to either encourage convergence in the business cycle (trade and trade structure) or alleviate asymmetries (adjusting wages). As economies continuously change we cannot be sure that a symmetric business cycle will remain constant for countries experiencing this implying that fulfilment of the other variables is still desirable. Overall it can be said that this is the core criterion as the purpose of the other criteria is connected to this one as depicted in figure 8. The wage flexibility is an adjustment tool with the purpose of alleviating the effects of output disturbances. Flexible wages only and factor mobility in general cannot be considered sufficient for fulfilment as it does not necessarily solve structural problems 36. The trade intensity shows to what extent the bilateral trade influences the economy and thus how integrated the country is with the currency area. We disregard the theories of Krugman (1993) and the EC (1990) implying that we define high trade intensity, ceteris paribus, as a positive factor. Where the trade intensity measured the exposure to output disturbances, the trade structure aims at measuring the probability of disturbances through the (dis)similarities in the trade. Large 36 As we saw in section it also had a cost of causing inflationary pressure in the high growth country (region)

45 differences should therefore, ceteris paribus, imply more disturbances through industry-specific shocks. 4.3 Methodology and model We apply two measures in which we 1) standardize the results from the four criteria and 2) conduct a cluster analysis with the purpose of identifying homogenous groups with respect to the OCA criteria. Standardization is done by using the formula given below: X X STV = (6) σ Source: Afifi et al. (2004) Where X is a country s value for a given variable and X is the average of all the countries and σ is the corresponding standard deviation. The sum of the standardized values is the thus zero. The variables for which a low value is preferable, i.e. output disturbance and intra-industry trade, the standardized values are multiplied by -1. In the first study in which Germany is used as benchmark, we take the average of the standardized variables disregarding any weight-measure and base our ranking on this. The purpose of this is to obtain a relative OCA scale for the countries which is also comparable to the B&E (1997) study Cluster analysis As above mentioned, cluster analysis is an econometric approach whose purpose is to identify homogenous groups based on various data. There are many different methodologies and approaches, however we will not discuss these, but present our choice of methodology. We base our choice on Everitt (2001) who reports that the group average method is overall the best for all kinds of data and Artis and Zhang (2001) who apply the same methodology in their study. We will in the following briefly present our choice of distance measure and cluster method Distance measure The most commonly used distance measure is the Euclidian distance (Afifi et al. 2004) and following Artis and Zhang (2001) we will use this. It can be written as follows:

46 d ij = p k = 1 2 ( x x ) (7) ik jk Source: Blunch (1999) Where x ik is the value of the k th variable on object i. In our OCA context k would represent one of the four OCA criteria and i and j would represent countries Cluster method We use the group average clustering which is an agglomerative procedure implying that the starting point is N clusters, i.e. each observation is its own cluster. In equation (8), we show how the dissimilarity coefficient, ( d( ω j, ωk ), between two clusters ω j andω, is calculated. k 1 d( ω j, ωk ) = d( j, k) (8) ω ω j k j ω j k = ω k Source: Artis and Zhang (2001) Where ω j and ω k denote the number of objects in the clusters, ω j andω k, respectively. In order to get a better understanding of the methodology we show a simple example of the method in box Critique of model As is the case with most economic models, their advantages are followed by shortcomings that we will briefly discuss here. The model does not adjust for any weight of each variable 37 which may be appropriate however there is not enough empirical support for doing that. Artis and Zhang (2001) are subject to the same problem. On the other hand, the purpose of this procedure is to identify similar groups of countries with regard to all the OCA criteria and weighting based on empirical evidence may not be applicable to every country. Finally, the outcome of the model is simply an identification of homogeneous groups (weighted or not) from which we cannot necessarily make explicit interpretations concerning their fulfilment of the OCA criteria. 37 Which is the case in B&E (1997) in which the derived coefficients are based on the nominal exchange rates from their regression model

47 - 45 -

48 4.4 Empirical results In the following sections we will present and discuss the results from our analysis. In order to give an overview of the variables that we deal with, we start by presenting the criteria and these are shown in table 3. Table 3: OCA criteria Source: Author s own creation It seems overall that the EMU countries and Denmark have more symmetric business cycles as they display smaller values in the output disturbances compared to outsiders Sweden and UK. The largest traders are the small neighbouring countries surrounding Germany including Austria, Belgium, Denmark, Netherlands and Luxembourg. Denmark distinguishes itself from these countries by having a trade structure above average, i.e. less similar than average. This is contrary to the UK that has a rather similar trade structure, but as before mentioned a diverging business cycle Ranking I this section we will present the ranking based on the average of the standardized values from table 3. The result is presented in figure 9 in which we show the ranking in a descending order. The grouping is made arbitrarily but based on the differences in value. We identify four groups even