scieee AI-readable full text Open interactive document viewer

DO trade and technology reduce asymmetries? Evidence from manufacturing industries in the EU

Rovelli, Riccardo,Paci, Raffaele

Abstract

EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.

Full text

Rovelli, Riccardo; Paci, Raffaele Working Paper DO trade and technology reduce asymmetries? Evidence from manufacturing industries in the EU Quaderni - Working Paper DSE, No. 301 Provided in Cooperation with: University of Bologna, Department of Economics Suggested Citation: Rovelli, Riccardo; Paci, Raffaele (1997) : DO trade and technology reduce asymmetries? Evidence from manufacturing industries in the EU, Quaderni - Working Paper DSE, No. 301, Alma Mater Studiorum - Università di Bologna, Dipartimento di Scienze Economiche (DSE), Bologna, https://doi.org/10.6092/unibo/amsacta/5007 This Version is available at: https://hdl.handle.net/10419/159144 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc/3.0/ 1 Do trade and technology reduce asymmetries? Evidence from manufacturing industries in the EU Raffaele Paci * Riccardo Rovelli ** This version: November 1997 Abstract Are asymmetric shocks to output less important for industries which are more open to trade and more technology-intensive? Our results, obtained from a correlation analysis between growth rates of value added in thirteen manufacturing industries in eleven European countries between 1979 and 1990, clearly support the hypothesis. This finding suggests that policies which promote trade and technological innovation may help to decrease the importance of the asymmetric components of the business cycle within European countries. JEL classification:E32; F15; F33 _________________________ * Università di Cagliari and CRENoS. E-mail: [email protected] ** Università di Bologna. E-mail: [email protected] Previous versions of this paper have appeared as “Contributi di ricerca Crenos, 96/5”, and as IGIER W.P. 1996, no.109. Acknowledgments. We thank Crenos (Cagliari) and Igier (Milano) for research support. Rodolfo Helg and Paolo Manasse commented extensively on the first draft of this paper. We also thank for comments participants at the Eastern Economic Association 1997 Annual Conference and at the Bank of Italy-Sadiba 1997 Conference. Wim Deblauwe provided excellent research assistance. All errors are our own. 1. Introduction Many recent analyses of the benefits and costs of joining EMU have focused on the importance of asymmetries in the business cycle among potential members of the union. Many researchers have found these asymmetries to be quite relevant for the EU as a whole. In our view, however, the implications of these results have sometimes been over-emphasized. While business cycle asymmetries are clearly a negative entry in the balance of benefits and costs resulting from the participation to EMU, it should also be clear that they can be taken care of, for instance by an appropriate design of fiscal policy or by incentives to increase factor mobility within the union. Furthermore, what has been often neglected is that such asymmetries are not exogenous to the institutional context of the economy, and that they are likely to be influenced by the decision of European countries to participate in the EU and in EMU. In this paper we inquire in particular whether an increased openness to trade within European industries (which is the predicted outcome of the establishment of one market and of the adoption of a single currency) and a higher degree of technological intensity are correlated with a reduced importance of business cycle asymmetries. For this purpose, we estimate and interpret a statistical model of the evolution of, and correlation between, growth rates of value added of thirteen manufacturing industries in eleven European countries, between 1979 and 1990. We document to what extent growth rates in each “local industry” are correlated at the country or at the industry level, and whether the relative importance of these correlations is related to some specific characteristics of each local industry - and in particular to the importance of international trade and to the degree of technological intensity. To motivate intuitively our line of research consider, as an example, two industries in the same country, the first one of which sells in foreign countries a higher share of its output. Is the growth rate of output in the first industry likely to be relatively less correlated with growth rates in other industries in the same country? Many economists would probably agree that, ceteris paribus, this is likely to be so: the first industry, probably, is relatively more exposed to the international business cycle, and less exposed to the domestic component of the business cycle1. But then, as barriers to trade in the second industry are reduced, the degree of asymmetry will also be reduced. In this view, contrary to the analysis à la Mundell (1961), the existence of symmetry between countries is no longer a prerequisite, but will instead become the outcome of a process of full economic integration. The paper is organized in the following way. In section 2 we review the literature and, in section 3, we discuss the hypotheses to be tested. In section 4 we briefly describe the data and evaluate the results of the analysis. Section 5 concludes. 2. The evidence on asymmetric shocks, and their relevance to EMU In this section we briefly review and discuss the literature on two related issues. First, we evaluate the evidence documenting the relevance of asymmetries in the business cycles of different countries. Second, we discuss to what extent this evidence has implications for the desirability of monetary unification. 2.1 Review of the literature Typically, the literature on international real business cycles finds that country-specific sources of variation dominate over industry-specific sources. This points to the prevailing of asymmetric cyclical fluctuations. See Backus, Kehoe and Kidland (1992) and Costello (1993). Other papers look at the correlation properties of stock market returns, and they also find that returns are more strongly correlated along the country rather than the industry dimension. See Drummen and Zimmerman (1992) and Heston and Rouwenhorst (1994). Parallel to this, another literature has been developing in the recent years. Its main concern is to assess the relevance of business cycle asymmetries, viewed as an obstacle to the establishment of a currency area within a group of countries. Most of this literature is related to the debate on EMU. The starting point has been the paper by Bayoumi and Eichengreen (1993). They found that, identifying Germany as the anchor area , four other countries join Germany in the core of the EU: Belgium, Denmark, France and Netherlands. This core is defined by a high correlation of supply shocks, and to a lesser extent also of demand shocks, within this group of countries. In a similar vein, Bini Smaghi and Vori (1993), using 1The answer to this question would probably be the same, whether the international business cycle is mainly driven by innovations on the demand or on the supply side. disaggregate data, find that for the six founding members of the EC, 60% of the explained variance of manufacturing output is explained by sector-specific shocks alone. For 10 EC countries, this percentage is 23% (for 1976-1990). This again points to the existence of a core of countries with a considerable degree of symmetry. Using a different empirical model, Helg, Manasse, Monacelli and Rovelli (1995; henceforth HMMR) found that, for all EU countries, within-country correlations of output growth dominate over within-industry correlations. Similarly, after modeling output innovations as residuals from a cointegrated VAR, they used principal components analysis to show that more variance of output innovations is explained at the country, rather than the industry level. This points to a potentially high level of asymmetry. However, they also found that, within a subset of the EU countries, country principal components are strongly correlated with each other. On this basis, they identify a core of “symmetric” countries, which include in the inner layer Belgium, Germany and the Netherlands, followed by Denmark and France. Whereas the papers referred to above examine data for the manufacturing sectors only, Bayoumi and Prasad (1995) define 8 sectors, which add up - in terms of valued added - to GDP. They also distinguish between long and short run disturbances. They found that in the long run (average growth rates) region-specific disturbances are dominant in Europe, negligible in the US 2. From the point of view of the business cycle, they also found that short-term fluctuations in Europe tend to be in general more “idiosyncratic”. In the Manufacturing, Transport and Trade sectors, in particular, regional shocks (even if they explain less variance than sectoral shocks) are in relative terms considerably more important than in the US, as they account for 20-28% of total variance. Thus the literature on EMU points to the fact that, as far as output or value added is concerned, there is a core of countries with a higher degree of symmetry (country-specific innovations are of relatively low importance), but that if we extend the analysis to include all 2 Significantly, however, region-specific disturbances are not important in Europe as far as employment growth is concerned. In another paper which looks primarily at the determinants of employment, Marimon and Zilibotti (1996) also find that almost 80% of the long-run employment growth differentials across EU countries and industries is accounted by sectoral effects, and only 20% by country effects. However, the latter figure rises to 47% over the business cycle frequency. Thus, it seems that the determinants of output and of employment shocks may be of different origin. Although it would be interesting to examine in some detail and possibly explain this difference, this is not the purpose of this paper. the countries in the EU (during the Eighties, eleven countries) then country or idiosyncratic factors gain importance. 2.2 Is asymmetry an obstacle to monetary unification? In general, the evidence which we have reviewed shows the importance of country factors (at least outside a core of EU countries): that is, of shocks which are not positively correlated, and thus asymmetric, between countries. What are the implications of this finding for the debate on monetary unification? The existence of asymmetric shocks between countries, which are about to enter a process of monetary unification, is clearly a potential source of costs and frictions alongside that process. However, there are at least three reasons why we should not overstate the significance of these findings: • In a Mundellian world (with only two completely specialized countries, and elastic aggregate supply (or equivalently, with exploitable Phillips curve tradeoffs), exchange rate policy is an appropriate therapy against asymmetric shocks: if a positive shock in Germany affects France negatively, and Germany won’t react to it, France might want to devalue. But in a world of many countries it is unclear whether the exchange rate is such a good policy instrument: as France devalues vis à vis Germany, it will at the same time affect the exchange rate towards all the other countries in the world, which might otherwise be undesirable (See Melitz, 1994).3 • Similarly, in a world of two, incompletely specialized countries, or with different regions within each country, it is not clear whether an “asymmetric” shock (between the two countries) is also “symmetric” inside one country. That is, because industries are not homogeneously located within a country, different regions might be affected differently by the shock. If this is the case (that is, if an existing country is not an optimal currency area), then it is not necessarily true that either country will want to use the exchange rate as a policy instrument against that shock (See Bofinger, 1994, and Melitz, 1994). At a different level of analysis, if policy makers within a group of “asymmetric” countries are able and willing to commit to long run policy choices, they will weight the short 3 Bayoumi and Eichengreen (1997) analyze the determinants of bilateral nominal exchange rate variability within a group of 21, mostly European, countries. They find differences in growth rates to be an important explanatory variable, whereas the size of bilateral exports is negatively related to exchange rate variability. run benefits of using the exchange rate as a tool of stabilization policy against the long run benefits of committing to exchange rate (and possibly, monetary) stability and (if the conclusions of this paper are correct) the may also anticipate that, as exchange rate stability fosters trade and economic integration between countries, the cyclical behavior of industrial sectors will become more positively correlated among those countries. This implication has been examined in particular by Frankel and Rose (1996). In a paper which is motivated by the same considerations discussed in this paper, they examine bilateral trade and real activity data for 21 countries and find that “closer international trade links result in more closely correlated business cycles across countries.” (p.2) The purpose of this paper is thus to examine the validity of the last argument mentioned above. For instance, we shall document to what extent the magnitude of asymmetries is lower in more open industries. If this is so, then this would establish a prima facie argument in favor of asymmetries being endogenous to the process of trade integration. These issues will be more thoroughly discussed in the following section. 3. Trade, innovation and asymmetry In this section we discuss how the extent to which one country is exposed to output shocks of an (a)symmetric nature (relatively to other countries) may be influenced by economic variables such as the degree of trade openness and of technological intensity. To start with, we need to give a more precise definition of symmetry. We define “symmetric” a situation where the role of domestic factors (demand and supply conditions in other industries; level and composition of final demand; labor market conditions) in explaining the cyclical behavior of an industry or a sector does not dominate (or is low) relatively to the role attributable to international factors. This implies that business cycles should be contemporaneously correlated across countries. We define “asymmetric” the opposite situation, in which “country” variables matter considerably. 3.1 The role of trade openness Assume an economy which, initially, is closed to trade (of goods, services and know how). It is then natural to assume that demand or supply shocks will be independently distributed of, and thus asymmetric to, those affecting other countries. As this economy begins to trade with other countries, then demand shocks affecting those countries might affect exporting industries. As different industries in different countries use the same imported inputs (say, oil), then supply shocks in the oil industry will affect all those industries world wide which use that input. These effects introduce elements of symmetry. This introduces the first conjecture. Conjecture 1. For a given technology, sectors which are more open are more likely to be affected by demand and supply shocks which are symmetric across countries. It is useful in this context to clarify our notion of openness. First, notice that available data measure, for each manufacturing industry, exports and imports of both intermediate and final goods. Thus, for instance, imports include both intermediate goods used within the same industry but purchased abroad and also goods sold for final purchase, and possibly competitive, to some extent, with domestic production. From modern approaches to international trade we know that, if industry structure is imperfectly competitive, this will lead to intra-industry trade and the pattern of specialization will also be more similar across countries. Other approaches emphasize instead the tendency of trade to produce (at the limit, complete) specialization. This is so both in the more traditional approach, based on comparative advantages, and in the more recent literature, based on agglomeration economies (Krugman, 1991). It would seem plausible to assume that more symmetry should follow as long as more intra-industry trade develops. In general, empirical trade literature agrees on the fact that intra-industry trade has indeed been a characteristic feature of international trade in the postwar period. This confirms the plausibility of the hypothesis that more trade should produce more symmetry. In the next section we further examine the relationship between intra industry-trade and business cycle symmetry. 3.2 Intra-industry trade and business cycles symmetry To understand more clearly the relationship between openness and symmetry, consider the following examples. Assume that local industry A1 (industry 1 in country A) sells its output in equal shares in countries A, B, C. However, local industries B1 and C1 only produce for the domestic market. The discussion in section 3.1 suggests that we should relate symmetry to trade openness. But in the context of this example the relationship between symmetry and trade openness is not an obvious one. By construction, the sales of industry A1 will not be more correlated to the evolution of (say) aggregate demand in A than in B or C; thus, according to our definition of symmetry as reduced influence of domestic factors, industry A1 is subject to symmetric shocks, while industries B1 and C1 are not (since they are only influenced by local demand conditions). But all three industries are, by definition, open. It might seem more plausible to argue that, if trade is unidirectional (in either direction), it does not necessarily make an industry more symmetric (industry A1 trades and is symmetric; industry B1 trades and is not symmetric). Thus, Conjecture 1 must be taken to imply that trade openness of a particular local industry is a necessary, but need not be a sufficient requisite for symmetry. The example above might also be taken to imply that exports, more than trade openness, are a relevant indicator of symmetry. However, this conclusion (although it is confirmed by our particular example) is not necessarily true. Consider the following example. Industry A1 is as before, and industries B1 and C1 again produce for the domestic market only, but using imported input from A1. In this case, and to the extent that a technological shock hits A1's exports of intermediate products to the other countries, then we might observe a symmetric (and also industry-wide) shock. In this example, symmetry is related not so much to exports vs. imports, but to the fact that, even if unidirectional, trade is of the intra-industry rather than of the inter-industry type. While this is probably a better way to characterize our conjecture, unfortunately in our data we cannot distinguish between inter and intra industry trade. Despite this limitation, we think it may nevertheless be interesting to characterize empirically to what extent the relevance of country and/or of industry-wide variables is associated with the degree of industry openness.4 3.3 Country versus industry factors Until now, we have defined symmetry as implying a reduced role played by country factors (once factors affecting the international business cycle are taken into account). But should we also define as more symmetric a situation where each local industry is more strongly correlated, across the business cycle, with the corresponding local industries in other countries? It may be tempting to answer positively, and to some extent this has been the accepted answer in the literature. See Stockman (1988) and HMMR. For instance, the latter argue: "industry-wide shocks may naturally spread across countries, hence they embody an element of symmetry" (p.1030). We agree with this notion, but we also note that it is not equivalent to a "necessary" condition for symmetry. To understand, let us consider a 4 Our conjecture is also supported, at the aggregate level, by the recent findings of Frankel and Rose (1996). Using thirty years of aggregate data for industrialized countries, they find that countries Examining the results in some detail, the first row in Table 4 reports the explained variance of growth rates of value added in the whole sample (1573 observations over the years 1980-1990). The R2 is 0.27 when only the COUNTRY and TIME dummies are included, 0.35 when also the INDUSTRY dummies are added. On the basis of the argument above, we expect that the variance attributed to COUNTRY effects should increase for the less open, and less technological intensive local industries, and decrease for the more open and more technological intensive local industries. [Insert Table 4 here] SPLIT 1. (Ranking of all local industries according to trade openness). The R2 obtained from inclusion of COUNTRY and TIME effects only decreases as we move to the more open group of industries. In the fourth group, the exclusion of the COUNTRY effects (when the TIME and INDUSTRY effect are included) cannot be statistically rejected at the 1% significance level. SPLIT 2.A. (Ranking of all local industries according to technological intensity). Regressions in the four groups show that R2 due to COUNTRY effects decreases from the first two (less technological intensive) to the second two groups (although in the fourth it is higher than in the third). SPLIT 2.B. (Ranking of Euro-wide industries according to technological intensity). The R2 explained by COUNTRY effects decreases systematically from the low to the high TEC industries. SPLIT 4. (Sample split on the basis of the four clusters defined in Table 2). We observe the lowest R2 for CLUSTER 2 (characterized by high openness) and also CLUSTER 4 (characterized by high technological intensity). And again in these two sub-samples (and only in these) the COUNTRY dummies are not statistically significant. Summing up, with the partial exception of Split 2.A, all the results confirm that the COUNTRY effects are much less relevant in the more open and technologically advanced industries. We have also examined the regression looking for separate effects of the industry dummies and testing for their exclusion. As for the correlation analysis of section 4.1, there is no systematic change in the effect of the dummies across subsamples, and the exclusion of these dummies is generally accepted (data not reported, available on request). 4.4 Summing up the empirical findings 1. In reference to the questions and conjectures raised in section 3, we have found the following answers. • The extent to which growth rates in manufacturing industries in Europe are characterized by country-specific (asymmetric) effects is inversely related to the level of trade openness. This may be due to intra-industry trade spreading shocks across countries (conjecture 1), and also to technology innovations being “imported” through trade (conjecture 2). • A reduced role of country-specific factors is not in general associated with an increased role of industry factors, as measured by either the correlation or the regression analysis. So the short answer to question 1 is: no. • Industries with a higher volume of R&D expenditures exhibit a higher degree of symmetric behavior. So the answer to question 2 is affirmative. 2. 5. Conclusions and policy implications In this paper we have shown that the importance of country effects on growth rates in the manufacturing industry is inversely related to the levels of trade openness and of technological intensity. Is there any policy relevant implication of these findings? In the literature on currency areas, the fact that a group of countries is exposed to asymmetric, country-specific business cycle disturbances is considered as an obstacle to the adoption of a common currency (or to fixing the exchange rate). While this is (not the only factor to be taken into account, but) certainly correct in a static framework, the existence of some asymmetries may be related to a lack sufficient integration within those countries, and thus the decision to establish a currency area (or a monetary union) may validate itself in the process of time, as increasing integration, fostered by the union, will eventually lead to a more harmonized cyclical behavior across the participating countries. In the empirical analysis we have identified two factors that may act as a vehicle of further integration between countries: trade and technology. Thus, we should expect to find a more symmetric behavior (across different countries) for those “local industries” which are more open to trade and which require greater expenditures in technology. Since both factors - trade and technology diffusion within a group of countries - are likely to be strengthened as transactions costs are reduced and factor mobility is increased, as would be the case in a full monetary union, the broad policy implications of our findings are clear. However, a word of caution is required in at least four respects, which would also warrant an extension of our research perspective. (i) First, our analysis has not shed any light on the time-scale of the process of integration fostered by the growth of trade and the diffusion of technology. The longer is this process, the greater may be the short-run costs of a monetary union. (ii) Second, while our findings support the notion that such a monetary union might be beneficial in reducing business cycle asymmetries, would we obtain the same results if we had used as a starting point for the empirical analysis regions rather than local industries? Also, would the same conclusions also apply had we considered a more heterogeneous group of countries (in terms of their initial level of industrial development)? Or what alternative policies and solutions might instead be adopted in such cases? Clearly our analysis does not shed any light in this respect. (iii) Third, it would be desirable to know and understand more about the industry dynamics which lead to greater integration through trade, growth and technology diffusion. For instance, are multinational companies better at promoting integration? And which other policies, if any, should be activated to complement the adoption of a monetary union at the macro level? (iv) Fourth, to what extent would we have reached the same conclusions, if we had extended our analysis to other branches of economic activity besides the manufacturing industries? As the contribution of manufacturing to the generation of value added is generally decreasing in developed economies, this question is clearly an important one. However, our ability to explore these issues is constrained by an unavoidable trade-off between searching for (a)symmetries at a sufficient level of disaggregation, and taking into account the whole spectrum of economic activity in each country. We suspect that, if we took a more aggregate perspective, reducing the disaggregation in manufacturing to include other broad sectors (other industries, agriculture and services), then the relative weight of country versus sector effects would be increased. But this may be the subject of future research. REFERENCES Backus D.K., Kehoe P.J. and Kydland F.E. (1992), International real business cycles, Journal of Political Economy, 100, 745-775. Bayoumi T. and Eichengreen B. (1993), Shocking aspects of European monetary unification, in Giavazzi F. and Torres F. (eds), The transition to economic and monetary union in Europe, Cambridge (UK): Cambridge University Press. Bayoumi T. and Eichengreen B. (1997), Ever close to heaven? An optimum-currency-area index for European countries, European Economic Review, 41, 761-770. Bayoumi T. and Prasad E. (1995), Currency unions, economic fluctuations and adjustment: some empirical evidence, Discussion Paper Series, n. 1172, London: CEPR. Bini Smaghi L. and Vori S. (1993), Rating the EC as an optimal currency area, Temi di discussione no. 187, Rome: Banca d’Italia. Bofinger P. (1994) Is Europe an Optimum Currency Area?, Discussion Paper Series, n. 915, London: CEPR. Coe D.T. and Helpman E. (1995), International R&D spillovers, European Economic Review, 39, 859-888. Costello D. (1993), A cross-country, cross-industry comparison of productivity growth, Journal of Political Economy, 101, 207-222. Drummen M. and Zimmerman H. (1992), The structure of European stock returns, Financial Analysts Journal, July/Aug., 15-26. Frankel J.A. and Rose A.K. (1996) The endogeneity of the Optimum Currency Area criteria, Working Paper 5700, Cambridge (MA): NBER. Helg R., Manasse P., Monacelli T. and Rovelli R. (1995), How much (a)symmetry in Europe? Evidence from industrial sectors, European Economic Review, 39, 1017-1041. Heston S. and Rouwenhorst K.G. (1994), Does industrial structure explain the benefits of international diversification?, Journal of Financial Economics, 36, 3-27. Krugman P.(1991), Geography and trade, Cambridge: MIT Press. Marimon R. and Zilibotti F.(1996), Actual versus Virtual Employment in Europe: Is Spain Different?, unpublished. Melitz J. (1994), A multilateral approach to the theory of optimal currency areas, unpublished. Mundell R. (1961), A theory of optimum currency areas, American Economic Review, 51, 657-665. Stockman A.C. (1988), Sectoral and national aggregate disturbances to industrial output in seven European countries, Journal of Monetary Economics, 21, 387-409. Figure 1. Growth rates correlations 1.A Averaged by country be uk po dk gr ir nl fr es de it 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 country 1.B Averaged by industry 12 6 5 810 9 4 11 2 1 3 13 7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 country Notes: Data on the horizontal and vertical axis of Figure 1A (1B) come from the last rows (columns) of Tables A.2 and A.3 respectively. See text for details. Table 1 Cluster analysis of countries (Split 3, 1979-90) Cluster Members Main characteristics Cluster Centers (normalized var.) label TOP TEC BAL ∆TOP ∆TEC ∆BAL SMALL COUNTRIES BE, DK, IR, NL Small, open, low tech 1.12 -0.53 0.03 -0.50 -0.52 -0.80 LARGE COUNTRIES DE, FR, IT, UK Large, closed, high tech -0.85 0.71 0.53 -0.48 -0.46 1.12 SOUTHERN COUNTRIES ES, GR, PO Southern, with increasing openness and tech -0.35 -0.24 -0.75 1.31 1.31 -0.43 Table 2 Cluster analysis of local industries (Split 4, 1979-90) Cluster Number Main characteristics Cluster Centers (normalized var.) label of cases TOP TEC BAL CLU-1 55 Low openness, deficit -0.30 -0.34 -0.74 CLU-2 13 Very high openness, high deficit, high tech 1.80 -0.07 -1.17 CLU-3 53 Very low openness, high surplus -0.85 -0.31 0.90 CLU-4 22 Very high technological intensity, openness -0.35 2.354 0.22 Notes to Tables 1 and 2: Clustering method: Mean linkage between groups. See Table A.0 for the definition of the variables. Table 3 Growth rate correlations in selected sub-samples Within Country Within Industry Whole sample 0.29 0.16 TOP 1 (low) 0.36 0.18 Split 1 TOP 2 0.32 0.16 Trade openness TOP 3 0.28 0.16 TOP 4 (high) 0.24 0.16 Split 2.ATEC 1 (low) 0.35 0.21 Industry-level TEC 2 0.33 0.14 technological TEC 3 0.25 0.11 intensity TEC 4 (high) 0.27 0.18 Split 2.BEURO-TEC 1 (low) IND. 1,2,3,4 0.33 n.a Europe-wide EURO-TEC 2 IND. 6,7,8,9,10 0.32 n.a. technological intensity EURO-TEC 3 (high) IND. 5,11,12,13 0.22 n.a. Split 3 SMALL 0.24 n.a Countries LARGE 0.38 n.a. SOUTHERN 0.24 n.a. Split 4 CLUSTER-1 0.30 0.14 Local industry CLUSTER-2 0.19 0.14 characteristics CLUSTER-3 0.29 0.17 CLUSTER-4 0.30 0.19 Notes. See text and Table A.0 in the Appendix. Table 4 Regressions with Time, Country and Industry Effects Adjusted R2 with different regressors F-tests Sample No. Obsv. Time Time, COUNTRY Time, COUNTRY, INDUSTRY Excl. of COUNTRY All 1573 .100 .270 .345 R TOP 1 (low)398 .142 .341 .426 R Split 1 TOP 2 396 .109 .348 .337 R TOP 3 396 .109 .233 .435 R TOP 4 (high)385 .062 .173 .184 A TEC 1 (low)396 .109 .347 .413 R Split 2.ATEC 2 396 .109 .313 .409 R TEC 3 396 .076 .161 .292 (A) TEC 4 (high)385 .154 .294 .357 R EURO-TEC 1 (low) IND. 1,2,3,4 484 .100 .358 .473 R Split 2.BEURO-TEC 2 IND. 6,7,8,9,10 605 .174 .328 .378 R EURO-TEC 3 (high) IND. 5,11,12,13 484 .054 .184 .251 R CLUSTER-1 605 .078 .287 .375 R Split 4 CLUSTER-2 143 .067 n.s. n.s. A CLUSTER-3 583 .137 .303 .347 R CLUSTER-4 242 .145 .250 .275 A Notes: For the definition of the samples, see text and Table A.0 in the Appendix. The dependent variable is the growth rate of value added (GVA) divided by the standard deviation (in the time dimension) of GVA of local industry. Annual data, 1980-1990. All the dummies are time-indexed. The F-tests evaluate the exclusion of the COUNTRY dummies: R = reject exclusion at 1% level of significance; (A) = do not reject exclusion at 1%, reject at 5% level; A= reject at 1% level. APPENDIX TABLE A.0 DATA AND SOURCES COUNTRIES INDUSTRIES BE Belgium (trade includes Luxembourg) 1Food, Beverages, Tobacco DE Germany (Federal Republic) 2Textiles, Apparel, Leather DK Denmark 3Wood Products, Furniture ES Spain 4Paper, Paper Products, Printing FR France 5Chemicals GR Greece 6Rubber and Plastic IR Ireland 7Non-Metallic Mineral Products IT Italy 8Basic Metal Industries NL Netherlands 9Metal Products PO Portugal 10 Non-Electrical Machinery UK United Kingdom 11 Electrical Machinery 12 Transport Equipment 13 Office and Precision Instruments VARIABLES LABEL DEFINITION PERIOD SOURCE RD business enterprise expenditure in research and development 1981-90 Ocse: Science and Technology Indicators (Stan) VA value added 1979-90 Ocse: Stan; Industrial Stat.; Eurostat Xtotal exports (Europe and RoW) 1979-90 Volimex Mtotal imports (Europe and RoW) 1979-90 Volimex TEC, ∆TEC technological intensity average, %change (RD/VA) * 100 TOP, ∆TOP trade openness average, %change (X+M)/VA BAL, ∆BAL trade performance average, %change (X-M)/(X+M) GVA average % growth rate of value added, 1985 constant prices, sectoral deflator average Table A.1 Averages by Country and by Industry for the variables in the data set (1979-90) Country GVA TOP TEC BAL BE Belgium 0.12 4.96 4.66 -0.01 DE Germany F.R. 0.94 1.27 3.82 0.05 DK Denmark 1.30 3.43 3.77 -0.10 ES Spain 0.00 1.19 1.70 0.01 FR France 1.03 1.32 5.14 -0.03 GR Greece 0.90 3.29 0.72 -0.47 IR Ireland 3.46 3.93 1.29 -0.15 IT Italy 3.20 1.19 2.87 0.09 NL Netherlands 1.46 4.37 2.02 -0.09 PO Portugal 2.30 2.68 0.60 -0.13 UK United Kingdom 1.69 1.71 5.65 -0.14 Industry GVA TOP TEC BAL 1Food 1.88 1.42 0.65 0.05 2Textiles, Apparel -0.83 2.81 0.45 0.03 3Wood, Furniture 0.11 1.46 0.26 -0.16 4Paper, Printing 2.14 1.07 0.28 -0.23 5Chemicals 2.83 2.28 6.18 -0.08 6Rubber, Plastic 3.53 2.16 1.99 -0.06 7Non-Metallic Mineral Prod. 0.26 1.17 1.06 0.07 8Basic Metal Industries 0.69 4.90 1.53 -0.14 9Metal Products 0.49 1.03 1.14 0.06 10 Non-Electrical Machinery 1.31 3.74 2.85 -0.10 11 Electrical Machinery 2.28 2.32 9.14 -0.16 12 Transport Equipment 1.35 3.66 6.07 -0.15 13 Office, Precision instr. 3.36 6.67 6.51 -0.27 Note: see Table A.0.