Intra Industry Trade and shock asymmetry in a monetary union - A study of trade evolution in the Euro Zone

Contents

List of Abbreviations

List of Figures

List of Tables

1 Introduction

2 Optimum Currency Area criteria
2.1 The “classic” OCA theory
2.2 Are OCA criteria endogenous?

3 Intra-industry trade
3.1 An interplay between theory and empirics
3.2 Trade in vertically differentiated products - an ambiguous phenomenon

4 Optimum Currency Areas and Intra-industry trade
4.1 Trade pattern evolution endogeneity of OCA criteria - the ex ante view
4.2 The ex post view

5 Methodology

5.1 The Data Source: COMEXT Database on european trade
5.2 Decomposing trade flows

6 Results

7 Conclusion

References

A figures

B Source Code
B.1 R

List of Abbreviations and Variables

illustration not visible in this excerpt

List of Figures

1 Grubel and LLoyd decomposition of Inter- and Intra-industry trade. Source: Fontagné and Freudenberg (1997)

2 Measuring Intra-industry trade in horizontal and vertical differentiation: Green- away et al. versus Fontagné and Freudenberg. Source: Fontagné and Freuden- berg (1997)

3 Total bilateral trade within the EA-15 from 1999-2011 by trade type

4 Trade shares in total bilateral trade within the EA-15

5 Total bilateral trade by country and trade shares, 1999-2000

6 Trade shares in biateral trade by country

7 HIIT and VIIT shares in bilateral relationships within the EA-15

8 Market structure, differentiation of product and the determinants of trade. Source: Fontagné and Freudenberg (1997)

9 Trade shares in total EU-12 bilateral trade

10 Trade shares in bilateral trade by country

11 Trade shares in trade with the EA-15 from 1999-2011

List of Tables

1 10 most and least important bilateral relationships of HIIT

2 Share of trade types in total intra EU trade TOP and least important HIIT

1 Introduction

The optimality of the Euro Zone as a currency union has frequently been discussed in academics and the public over the course of European economic and monetary integration. Thetopic is generally addressed using criteria that emerge from the early literature on optimalcurrency areas. Important issues in this literature are trade integration (McKinnon, 1963),product diversification (Kenen, 1969) and similarities in business cycles (Frankel and Rose,1998). The fulfillment of these criteria reduces the impact or likelihood of asymmetric shocksthat often constitute the benchmark scenario in the literature. This is essential in assessing the desirability of a monetary union since countries cannot adjust via the exchange rateinstrument. The criteria of trade integration, business cycle co movements and diversifiedeconomies are clearly of great importance for the Euro zone since it has encountered a significant rise in trade volumes over the last decades. All the more as strong empirical evidencehas been found for a positive correlation between trade integration and business cycle similarity and the endogeneity reinforcement through further economic and monetary integration(Frankel and Rose, 1997).

Since trade is an important issue for currency areas, use can be made of theoretical andempirical findings in trade theory and economic geography. This paper does so in examiningthe evolution of trade patterns within the Euro zone. Theory associates intra-industry tradewith specialization in varieties and similarities between countries in terms of factor contentand economic size. High shares of intra-industry trade in bilateral trade therefore point to aneconomic diversification and business cycle similarity. Those countries are thus more likelyto verify OCA criteria (Fontagne and Freudenberg, 1999). Concerning the EMU ambiguouspredictions have been made prior to its creation. Whereas Krugman (1993) predicts the creation of dissimilarities and thus a reinforcement of shock asymmetries in the monetary union,Fontagne and Freudenberg (1997) as well as Frankel and Rose (1997) provide indication forthe opposite scenario.

In their 1997 paper Fontagne and Freudenberg examine trade pattern evolution within the EMU using highly disaggregated trade data from the period 1980-1994 and decomposing bilateral trade flows in three different types namely Inter-industry trade, HIIT and VIIT. They particularly highlight the importance of making the distinction between the latter two types by pointing out the differences in their respective determinants and consequences. This is essential when concluding on the fulfillment of OCA criteria.

This paper sets out to analyze the bilateral trade flows within the Euro zone for the period of 1999-2011 in order to identify recent trends in the evolution of trade patterns and to comparethem to the predictions that have been made prior to EMU creation. The empirical approachis adopted from Fontagne and Freudenberg (1997). Section 2 briefly reviews the basics ofOCA theory and the discusses the endogeneity of certain criteria. Section 3 provides anoverview of theoretical and empirical developments in trade theory concerning measurementand interpretation of different trade types. Section 4 creates the link between the two fieldsof research. Section 5 describes the methodology and the a discussion of the results is donein Section 6. Conclusions are drawn in Section 7. I used the statistical programming language R for calculations with the graphics package ggplot2 for visualizations. All results arereproducible and the source code is provided in the appendix. All links to web pages givenin the paper were verified December 11, 2012.

2 Optimum Currency Area criteria

2.1 The “classic” OCA theory

The desirability of entry to a currency union depends on the relative size of the benefits andcosts of joining. The foundations of OCA theory go back to the 1960s and mainly to threeseminal contributions by Mundell (1961), McKinnon (1963) and Kenen (1969). The criteriathat emerged from this early literature still form the main building block of the theory andaccordingly are frequently applied in assessing the desirability of a currency union.In his 1961 paper Mundell uses a Keynesian two-country framework, which assumes stickywages and prices and no capital mobility. He shows the consequences for the two countries offoregoing the smoothing effect of the exchange rate adjustment tool due to irrevocably fixedexchange rates. He provides the example of a demand shift from Country 1 to Country 2. Ina framework with sticky prices and wages the nominal exchange rate is fixed as well whichmeans that only a transfer of labor from country 1 to country 2 can clear labor marketsand provide adjustment in case of an adverse shock in one country (Mundell, 1961). Thisframework deeply influenced subsequent research in the field to the extent that the ability tocope with asymmetric shocks or to reduce their occurrence lie at the heart of the majority ofOCA criteria.

Mundell’s main argument in favor of joining a currency union are greater savings in the transaction costs and in the risk emerging from exchange rate uncertainty. Those should foster trade and investments. Thus, countries more integrated in international trade are more likely to form a Currency Union (Mundell, 1961).

McKinnon (1963) and Kenen (1969) further extend the theory taking parameters such as country size and structure into account (Kenen, 2000).

McKinnon (1963) considers a small open economy and identifies it not to be an OCA since nominal exchange rate adjustment largely effect domestic price levels and reduce the usefulness of the domestic currency as a unit of account. He concludes by stating that an OCA must have a minimum body of non-tradable goods to guarantee for the stability of the domestic purchasing power. McKinnon’s criterion often is interpreted and referred to as the openness criterion suggesting that the absolute degree of trade integration is a criteria for optimum currency union (see Frankel and Rose (1997)), however it is the relative degree of trade integration that should be considered.

Kenen’s (1969) paper deals with a number of issues including the need for a fiscal transfersystem to smoothen the consequences of shocks. However, it is mainly known for the criterion of domestic diversification. He points out that sector specific shocks in a diversified economy have smaller impact on the economy as a whole in terms of the required subsequent change in its real exchange rate. In addition, in case of independence and centered distribution of sector specific shocks the law of large numbers can be applied and as a result the aggregate exports will be more stable (Kenen, 1969).

When assessing the desirability of a currency union it is important not to consider an isolated criterion since this can lead to contradictions. Taking the Kenen (1969) criterion as an example and strictly applying it iteratively the most diversified economy would be the whole world. By contrast, referring to Mundell (1961) perfect intra-zone factor mobility could be achieved by taking each production sector as an OCA.

In subsequent decades interest in OCA theory declined and did non revive until the late 1980s. As it did, however, the focus was on measurement since the creation of EMU provided an important real world experiment. Hence, instead of reworking the original framework the classic theory was applied in empirical studies (Kenen, 2000). This is why the above criteria are still of major importance.

Although the bulk of literature has been of an empirical nature, some efforts have been made towards reworking the classic theory and adding new criteria¹. A criterion that is particularly often cited in empirical publications on the EMU is the correlation of business cycles. It is considered to reduce risk of asymmetric shocks Frankel and Rose (1997) which can be questioned (see Section 2.2).

Building up on the empirical work this paper focuses on the following three criteria: Productdiversification, openness and Business cycle correlation. Openness here refers to the degreeof trade integration with respect to the rest of the economy. It is related to the McKinnoncriterion but also to the benefits of participating in a monetary union mentioned above. Thechoice of this criteria is motivated with their link to intra-industry trade (see Section 4). Inaddition, the latter two are often considered to be endogenous. This matter will be discussedin the next section.

2.2 Are OCA criteria endogenous?

As mentioned in the preceding section the case of EMU has frequently been assessed underdifferent angles of OCA theory. A bulk of empirical research focused on the measurement of shock asymmetries within the EMU. A prominent example are the findings of Bayoumiand Eichengreen (1992) revealing significantly more idiosyncratic shocks for 11 Europeancountries when compared to those of US regions. They thus predict difficulties in operatinga European monetary union. Frankel and Rose (1998) question this view by testing twohypotheses. First, they suggest that trade integration and business cycle correlation arepositively correlated. They empirically investigate this relationship analyzing thirty years ofdata from 21 OECD countries and measuring business cycles with four separate indicatorsfor economic activity. Their findings indeed reveal a strongly positive effect of greater tradeintegration on the correlation of economic activity. Their second hypothesis is that tradeintegration, being subject to macroeconomic policy changes grows endogenously over timewith increasing economic integration. With respect to EMU this means that the abandon ofexchange rate risk and the reduction in adjustment costs should foster both trade integrationand, being correlated to that, business cycle similarity. Therefore, to paraphrase the authors,EMU could be justified ex post Frankel and Rose (1998).The findings of Fatas (1997) supportthis result revealing dramatic increases in Business cycle correlation with the Euro 12 and inparticular the “Southern Countries” such as Spain, Italy and Portugal. This contrasts withthe expectation of a two speed Europe.

The findings of Frankel & Rose strongly influenced research in subsequent years. Yet, theirresults can be questioned. Kenen (2000) challenges the way Frankel and Rose (1998) interprettheir results. He agrees that business cycle correlation unambiguously increases with “thethickness of the channels” (Kenen, 2000) i.e. the bilateral trade volume. But, he also showsthat this does not necessarily result in less pronounced asymmetric shocks (see Kenen (2000)).Furthermore, from a theoretical point of view, economic integration has an ambiguous effecton business cycle correlations. In a scenario opposed to that described by Frankel and Rose(1998) countries specialize in products over the course of increasing trade integration. Ifsector specific shocks predominate, Business cycles will thus become more idiosyncratic. Forthe EMU such phenomena has for instance been put forward by Krugman (1993). In his “TheLessons of Massachusetts for the EMU” he argues that completion of the European CommonMarket in combination with the implementation of EMU would promote agglomeration andspecialization thus reinforcing the asymmetry of shocks. This occurs due to the “reductionin the costs of doing business across space” (Krugman and Venables, 1996). This can due tothe abolition of tariffs but also and more relevant for this paper savings in exchange rate risk.Krugman and Venables (1996) provide a stylized theoretical model that assumes factors to be immobile between countries. In this framework, returns to scale outweigh the decreasing transportation costs at some point, fostering agglomeration. This, in turn, increases returns to scale and so on.

The migration of an industry to another country is costly if labor markets are unable to react,and will in the long term lead to changes in production structures due to factor redistribution(Beaulande, 2004). This kind of specialization can be observed in the United States, wherehighly specialized regions such as the silicon valley or the Detroit automobile industry exist. Itcan be argued that the cultural differences between the European countries and the existenceof a multitude of languages are factors that will hamper this development (Krugman andVenables, 1996).

3 Intra-industry trade

First research on Intra-industry trade as opposed to Inter-industry trade was done in the 1960s. In the following decades there has been a steady interplay between refined empirical findings on Intra-industry trade and developments in trade theory. The following section provides a brief chronological overview of the literature on the topic.

3.1 An interplay between theory and empirics

Early literature on international trade in the 1950’s suggests that an increase in trade volume is associated to specialization of countries in specific industries according to comparativeadvantages. This can bring about potentially sizable adjustment costs associated with unemployment and displacement of factors towards comparatively advantaged countries. On thepolitical level, this invoked a certain fear of a two speed Europe, with a core specializing inhigh value added industries and a periphery Fontagne and Freudenberg (1999).

However as the European economic integration was driven forward and trade volumes increased, empirical studies did not confirm this hypothesis. On the contrary, rising figures for Import and Export within the same industry were observed and thus questioned the comparative advantage theory of trade. This was the first time that Intra-industry trade occurred as an empirical phenomenon and it was not until the 1970’s that a theoretical explanation for this phenomenon was delivered (Fontagne and Freudenberg, 1997).

Using an imperfect competition framework this new international trade theory argued thatdue to internal economies of scale and utility gains that arise from an increased variety ofproducts, producers in different countries tend to specialize in varieties and trade betweencountries occurs within the same product category Fontagne and Freudenberg (2002). Hence,this theory considers trade to be a function of the similarities in taste, factor endowments,economic size and specialization. In a seminal contribution Helpman and Krugman (1985)provide a synthesis of those two theories. As a result, a new theoretical framework emergedassociating inter-industry trade with comparative advantages and the scale economy modelwith intra industry trade in similar products².

3.2 Trade in vertically differentiated products - an ambiguous phenomenon

However, this integrated equilibrium model of international trade was still missing some important points. First, Finger (1975), analyzing US trade data observed higher variationsin factor content within product groups than among the groups. This means that even ifhigh shares of IIT are observed this trade flows can reflect differences in factor content andproduction patterns (Fontagne and Freudenberg, 2002). Torstensson (1996) identified humancapital to be of particular importance. This challenges the assumption of IIT being directlyrelated to the scale economy model which assumes a net factor content of close to zero. Differences in factor content are positively related to differences in the quality of the exchangedproducts ?. For the purpose of interpretation it is therefore necessary to distinguish betweentwo types of Intra-industry trade: trade in products that differ in terms of quality (verticallydifferentiated) and trade in products differentiated by variety (horizontally differentiated).The determinants of these two types are not the same. Horizontally differentiated (similar) products are sold at the same price and are perfect substitutes. They can therefore beassociated with the scale economy model of international trade. In contrast, vertically differentiation reflects a specialization along the quality range (Fontagne and Freudenberg, 1997).As already mentioned, this specialization is based on differences in factor endowments (?),fixed costs in R&D (Gabszewicz et al., 1981) or qualification of the labor force (Gabszewiczand Turrini, 1997) resulting in a net factor content different from zero. In other words, of twocountries specializing on different stages of the quality range produce with different production functions. According to empirical findings by Cabral et al. (2006) vertical IIT involvesindeed similar net exchanges of labor of different skill levels to that of inter-industry trade.Consequently, adjustment process similar to those of inter-industry specialization are likelyto set in. This is in contrast to the assumptions of the IIT smooth adjustment hypothesiswhich has assumed adjustment costs of European economic integration to be low.

Fontagne and Freudenberg (1997) classify VIIT as an intermediate case between comparative advantage and the traditional view of IIT. That means that economic distance and comparative advantage is no longer exclusively associated to inter-industry trade but also matters for IIT in vertically differentiated products.

Not only for Intra-industry trade more than one model exists to explain the empirical phenomenon. As described in Section 2.2 external economies of scale can lead to agglomerationeffects. In this case Inter-industry trade shares will also increase (Fontagne and Freudenberg,2002). For the discussion of optimum currency areas, however this distinction is not that im- portant since the consequences (adjustment costs through migration of production factors) remain the same as in the comparative advantage scenario.

A summary of the different theoretical determinants of the different trade types is providedin Figure 8. It is important to keep the multitude of determinants in mind when makingconclusions on empirical findings on trade pattern evolution. As discussed for the case ofVIIT and HIIT the consequences do not always remain the same if the underlying theoreticalframework changes. But identifying the determinants and consequences is essential to creating a link between trade theory and the theory of optimal currency areas as the next sectionwill explain.

4 Optimum Currency Areas and Intra-industry trade

4.1 Trade pattern evolution endogeneity of OCA criteria - the ex ante view

The touching point between IIT and trade integration are the three criteria highlighted in section 2.2, namely product diversification, trade integration and business cycle correlation. As already mentioned earlier it is important to identify the determinants of the different trade patterns (see Section 3). This section creates the link using the interpretation of Fontagne and Freudenberg (1997). VIIT is therefore considered as an intermediate case that is further to be analyzed for the problem at hand. Fontagne and Freudenberg (1997) tend to interpret VIIT similar to HIIT, but for no explicitly stated reason.

HIIT being less ambiguous in its determinants and consequences can easier be linked to OCAcriteria. First, high HIIT shares point to a diversified production. We consider a bilateraltrade relationship. A high proportion of HIIT means that in both countries a high proportionof industries engage in exports. Production in both countries is thus diversified and accordingthe Kenen argument the creation of a currency union would be feasible (Kenen (1969) andFontagne and Freudenberg (1999)). Note however, that the Kenen criterion of a diversifiedeconomy can also be verified if Intra-industry shares are low in case of low trade integration.The relative importance of trade shares can also be revealing with respect to Business cyclescorrelations and potential specialization and agglomeration processes. Under the “Krugmanscenario” cross-border HIIT trade shares should decrease as a consequence of production redistribution whereas in the “Frankel & Rose scenario” they should increase over time. Thisis because sector specific shocks will affect both countries in the same way if trade in thissector occurs in both directions. Note, that the importance of HIIT for the latter two criteriadepends on the degree of openness (Beaulande, 2004). The importance of trade in terms ofGDP should therefore also be included in the discussion. Coming back to VIIT one couldargue that it is more likely to verify the Kenen argument than the Business Cycles argumentsince in general, the high and low quality sector of a product group are not affected by thesame shocks Fontagne et al. (2006).

In their paper ”‘Endogenous symmetry of shocks in a monetary union”’, Fontagne andFreudenberg (1999) discuss the endogeneity of OCA criteria by analyzing trade pattern evolution within the EMU and provide an ex ante view on EMU creation. They state two principalarguments on why the European Union will endogenously foster shock symmetry.

- Different effect of FX rate variations on trade patterns. Theoretically, the demandelasticity in similar products is comparatively high given the possibility to purchaseperfect substitutes. Thus, the effect of exchange rate changes should be more pronounced for HIIT than for the other two types. By contrast, if comparative advantagesare large enough variations should have little effect on inter-industry trade. Regressingthe different trade types on exchange rate variability and a number of other independent determinants of trade Fontagne and Freudenberg (1999) find negative correlationswith HIIT and VIIT the absolute value of the coefficient for HIIT being twice as high.They state that according to their results fears of increasing divergence between EMUmembers is no longer justified, nevertheless the interpretation is ambiguous since, asmentioned, VIIT reflects a specialization along the quality spectrum, and is not necessarily connected to OCA criteria.

This is partly in line with their findings on the evolution of trade patterns within theEU-12 for a period from 1980-99 (see Fontagne and Freudenberg (2002)): With the exception of Spain and Portugal HIIT shares remain constant over the whole period. Thisdoes not indicate an endogeneity of OCA criteria considering the fact that Europeaneconomic integration was promoted at the same time on the political level. Still though,in most countries Vertical IIT shares rise at the expense of One-way trade, with theexception of Greece and Ireland (see 10. This can be interpreted as counter evidence forthe specialization of countries as Fontagne and Freudenberg (2002) do. Nevertheless,in summary, the results are ambiguous.

- Reduced uncertainty in location choices. As firms don’t any longer need take into account exchange rate changes in their location choices their tendency to locate in a sectoral agglomeration decreases (Fontagne and Freudenberg, 1999). This argument is in contrast to the above cited predictions made by theoretical models such as the one provided Krugman and Venables (1996) (see Section 2.2).

4.2 The ex post view

Having introduced a bulk of literature that deals with the subject from an ex ante perspective,i.e. before the Euro has actually been introduced I will provide a brief overview of the ex ante literature before getting to the methodology and the presentation of my results.

A first example of research that questions the assumption of endogeneity is provided byFirdmuc (2004). He includes IIT in analysis of Frankel and Rose (1998) and no longer finds a relation between trade and business cycle correlation but between IIT and the latter. Inother words, the particular structure of trade matters for business cycle correlation. We mightimplicitly already have figured this out in the preceding section but how is it then possiblethat Frankel and Rose (1998) find such strong evidence for positive correlation of businesscycle similarity and trade integration? Here, a look at the results obtained by Fontagne andFreudenberg (2002) mentioned in the preceding section might provide some explanation. Figure 9 shows the evolution of trade patterns for the period of 1980-1999. Total intra-industrytrade is strongly increasing and so where total trade volumes. If we assume in this case VIITto be more associated to symmetric shocks than to the opposite, regressing Business cyclecorrelation on trade integration reveals positive coefficients, which were obtained by Frankeland Rose (1998).

Nevertheless, empirical support for the endogeneity hypothesis has also recently been foundfor instance by Mendonca et al. (2011) who use a beta regression approach of business cyclecorrelation on trade intensity with EU data from the period of 1967- 2003. However, theyalso find a negative marginal effect which became more negative till the end of the periodof analysis. This suggest that trade loses its determining importance in Business Cycle correlation. Lee (2012) equally finds increasing BC co movements within the European Unionusing wavelet analysis. However, they are not significantly stronger than with other European Countries. This is in line with Papageorgiou et al. (2010) who confirms an increasedsynchronization up to 1999 but does not find evidence for any increase since then until 2009.In addition his analysis reveals an increase in the number of clusters which supports the“Krugman Scenario”.

In summary, we have two different types of concurring scenarios: the “Frankel & Rose scenario” of Intra-industry specialization in varieties within the countries and the “Krugman scenario” of inter-industry specialization.

This paper tries to obtain indications on whether for the Euro zone any of those scenarios set in. It does so in building on the publications of Fontagné and Freudenberg applying their methodology on trade data of the period from 1999-2011. The following section provides a description for the data used and explains the methodology.

[...]

¹ For an overview of OCA criteria and a critical discussion see Tavlas (1993).

² Helpman and Krugman (1985)(p. 131): To use a terminology that has been widely accepted, we can have a HeckscherOhlin view of interindustry specialization but a scale economy view of intraindustry trade.