DOCUMENTOS DE ECONOMIA Y FINANZAS INTERNACIONALES

Transcription

1 DOCUMENTOS DE ECONOMIA Y FINANZAS INTERNACIONALES Trade tyes with develoed and develoing countries what can we learn from Sanish data? Juliette Milgram Baleix Ana I. Moro-Egido Aril 2005 DEFI 05/06 Asociación Esañola de Economía y Finanzas Internacionales htt:// ISSN Las oiniones contenidas en los Documentos de la Serie DEFI, reflean exclusivamente las de los autores y no necesariamente las de FEDEA. The oinions in the DEFI Series are the resonsibility of the authors an therefore, do not necessarily coincide with those of the FEDEA.

2 Documentos de Economía y Finanzas Internacionales (DEFI) TRADE TYPES WITH DEVELOPED AND DEVELOPING COUNTRIES WHAT CAN WE LEARN FROM SPANISH DATA? Autores: MILGRAM BALEIX Juliette & MORO-EGIDO Ana I.* Abstract In this aer, we investigate the nature of Sanish intra-industry trade and find that intraindustry trade with CEEC, Asian and Mediterranean countries has increased considerably since the middle of the Nineties. The second aim of the aer is to study if the comarative advantage argument also exlains vertical intra-.industry trade between different income countries. According to OLS estimations, technological differences do increase DVIIT while hysical caital differences decreases it. Results obtained alying Heckman method suort the idea that differences in hysical caital reduce the robability of IIT to occur but the level of vertical and horizontal IIT is better exlained by the roximity of artners, the similarity in develoment level and size of market than by differences in hysical caital endowments. The variables considered, mostly country-secific do have the same imact on vertical and horizontal IIT with emergent countries. * Universidad de Granada Facultad de Ciencias Emresariales y Económicas Deartamento de Teoría e Historia Económica, Camus de la Cartua GRANADA SPAIN. Tel: ; Fax: ; milgram@ugr.es. JEL Classification Numbers: F13, F17

3 1. Introduction An imortant feature of international trade is the raid growth of intra-industry trade and esecially trade of vertical differentiated roducts. Emirical and theoretical research has focussed in last decades in the determinants of vertical intra-industry trade. According to this literature, countries secialise in a quality segment deending on their comarative advantage and, as for inter-industry trade, differences in factor endowments would enhance vertical intra-industry trade. Thus, this tye of two-way trade develos more and more between high income countries and emergent countries but little studies have focused on the determinants of intra-industry trade in this case, in general because these flows were negligible until the second art of the Nineties. The urose of this aer is to examine Sanish intra-industry trade for the eriod and to study if the comarative advantage exlanation allows for a good understanding of vertical intra-industry-trade between unequal artners. This aer is organized as follows: the next section resents the theoretical framework of the emirical model tested further on. Section 3 briefly describes the level of intra-industry trade for Sain. Section 4 resents the emirical model and section 5, the econometric results. Finally, the last section summarises the main conclusions. 2. Theoretical framework Production under increasing returns to scale, together with the hyothesis of consumer's references for varieties ustify that similar roducts could be both exorted and imorted (Krugman, 1979; Lancaster 1980; Helman, 1981). These arguments also exlain why intraindustry trade generally takes lace among similar and rich countries. Helman and Krugman (1985) rovide the first exlanation for intra-industry trade (IIT) between unequal artners by adding differences in endowments to the revious exlanations. Since differentiated roducts are suosed to be more caital-intensive, the volume of intra-industry trade will be larger, the larger the intensity in caital relative to labour of the trading artners. And because a larger market allows for economies of scale to occur, similar and large markets will result also in more intra-industry trade. Finally, the more different their caital-labour ratios, the lower will be intra-industry trade. But roducts can rather be differentiated horizontally or vertically. In the first case, roducts differ in their design, colour or another attribute but not intrinsically. In the second case, roducts are differentiated by their quality. According to Falvey (1981) and Falvey and Kierzkowski (1987), vertically differentiated roducts are the fruit of different roduction function: low quality roducts may be labour-intensive while high quality roducts might be caital-intensive. Consequently, differences in factor endowments are exected to enhance vertical IIT. In this sense, the exlanation of vertical intra-industry trade would be in essence the one of the Heckscher-Ohlin model. Vertical differentiated roducts can also be roduced by more or less qualified emloyees (Gabszewicz et al. 1997). More in line with Ricardian models, countries can also have a comarative advantage in a quality segment due to technology differences (Flam and Helman, 1987) or differences in research and develoment exenditures (Gabszewicz et al. 1981). 3. Measurement of intra-industry trade. Levels of IIT between Sain and 188 countries were calculated for the eriod at the 8-digit level of disaggregation of the EU s Combined Nomenclature (CN). Data were obtained from COMEXT, Eurostat s database. Then roducts categories were concorded to the 15 industries of the NACE Clio R 25 classification. We use the Adusted Grubel-Lloyd Index (1975), following Greenaway and Milner (1983). We define volume of intra-industry trade (IIT) between Sain and country for each 8-digit 2

4 roduct as the overla between Sanish exorts X and imorts M. For each industry k, IIT is obtained as the sum of IIT volume at the roduct level: IIT = IIT = 2 min( X ; M ) Adusted Grubel-Lloyd Index (AGL) is therefore the share of IIT in total volume of trade: 2 min( X ; M ) IITk k AGLk = * 100 = *100 X + M X + M k k k This measure of IIT allows for both geograhic and industry aggregation (k can either be the total or any level of a classification). Even if estimations are erformed at the country level, we also calculate the intensity of IIT by regions (reg) in order to summarize the most stylized facts in grahs. The index is calculated as follows: IIT 2 min( X ; M ) AGL reg k = reg X k reg k + M k *100 = reg k reg k X + M *100 Abd-el-Rahman (1986) assumes that differences in unit value calculated er tonne reflect differences in qualities and Greenaway et al. (1994) and Fontagné et al. (1987) use this methodology to disentangle between vertical and horizontal intra-industry trade. Therefore, if the exort and imort unit values differ less than ±α ercent, roducts are considered similar or horizontally differentiated. Otherwise, that is if unit values of exort and imort differ substantially, this flow is considered as trade of vertically differentiated roducts. Unit values of exort UV(X) and imort UV(M) are calculated at the most desagregated level for each overla bilateral flow. Then, IIT of vertical differentiated roducts (DVIIT) and IIT of horizontal differentiated roducts (DHIIT) are obtained as follows: DHIIT UV ( X ) / UV ( M ) 1 α ; 1+α k k IIT = k if [ ] IIT = DVIIT if UV ( X ) / UV ( M ) [ 1 α ; 1+α ] k, where the arameter α is an arbitrarily fixed threshold (usually equal to 0.15 or 0.25). Nillson (1999) also suggested another measure in the case of develoing countries for which the level of IIT could be concentrated in relatively few grous of roducts while the total trade might be small. In this case, the AGL which is unscaled in nature since it doesn t reflect the absolute level of IIT can conduce to biased estimations. Nillson suggested using the average level of intra-industry trade er roduct traded as an alternative measure. This measure could facilitate the comarisons of the extent of IIT between small and large countries and can easily be comuted for different level of aggregation. Table I dislays the correlation coefficient between all measures. The volume of IIT is highly correlated with IIT volume er roduct traded (96%) while the correlation with each of these variables with the AGL index is much lower (resectively 78% and 82%). As a consequence, using IIT volume or IIT volume er roduct traded as exlained variable in the estimations should lead to the same results while exlaining the AGL index could introduce some imortant differences in the results. Turning to the value of the arameter α that should be used, vertical intra-industry trade volume when a difference in unit values of more or less 15% is used is correlated at 99% with the measure of vertical IIT when a margin of 25% is used. Similar results are obtained for Horizontal intra-industry trade, indicating that the choice of one of these two values for α, although arbitrarily, should not have substantial effects on the estimations results. So we will use the IIT volume and a margin of 25%. k 3

5 3 Levels of Sanish intra-industry trade Grah I summarizes the imortance of each tye of trade between Sain and the different regions. The regions we consider are EU, OECD, ACP (African-Caribbean Pacific Countries), Latina America, Develoing Asian countries, NIC (Asian New Industrialised Countries), CEEC (Central and Eastern Euroean), NIS (New Indeendent States, former URSS), MNA (Mediterranean and Northern African Countries), Middle Eastern countries. Intra-industry trade (IIT) reresents a great roortion of the Sanish trade with the EU but also reresents more and more of its trade with OECD, CEEC, NIC of Asia and MNA countries. In 2000, IIT reresented 33% of total volume of trade with the EU, 13,4% with the others countries of OCDE, 13% with CEEC's, 6,2 with NIC of Asia and 4,9 with MNA. For all regions, IIT consists mainly in vertical differentiated roducts excet with CEEC's and EU members for which each tye of trade reresent the same weight. This fact imlies that quality is the main concern for cometitiveness. Manufactured roducts are more differentiated in nature but the industry with highest level of IIT differs from one region to another. All these facts are suorted by data resented in Table II for Grah II dislays the level of IIT trade between Sain and 188 countries and their GDP er caita for two years: 1996 and It is clear that low income countries have a relatively low level of IIT trade with Sain. Thus, the relation between these two variables is not so straightforward if countries with an intermediate or high income er caita are considered. 4 Hyothesis of the model It is generally assumed in the literature that IIT, like total volume of trade, is ositively correlated with averages of country size (roxied by GNP) and negatively correlated with the trade barriers that may exist among them (such as, for examle, transortation costs reresented by the geograhical distance). It is also assumed that IIT is negatively related to differences in factor endowments (roxied by differences in er caita GDP) and ositively with caital intensity (roxied by average er caita GDP). When the different natures of IIT are taken into account, some of these variables are likely to have a different imact and some additional factors should be added. We discuss below the hyothesis for each variables: Differences in er caita income between Sain and the trading artner (DifGDPc ): This variable is the most imortant since its influence is not clearly established. In line with the model develoed by Falvey and Kierkowski (1987), differences in factor endowments would enhance trade of vertical differentiated roducts (comarative advantage exlanation). The effect of this variable should be ositive on two-way trade of horizontal differentiated roducts. Average er caita income of Sain and the trading artner (AVGDPc ): Since demand for variety is assumed to increase with income, both tye of IIT are assumed to be ositively related with average er caita income. Differences in GDP between Sain and the trading artner (DifGDP ): A great difference in economic size reflects both differences in demand and in suly sizes and is suosed to reduce any kind of IIT. Average GDP of Sain and the trading artner (AvGDP ): The greater the economic size used as a measure of market size. In the line of the Linder hyothesis, external markets can be considered as an extension of the internal market and local demand stimulates the innovation of roducts. In the context of the Chamberlin hyothesis of the reference for varieties of consumers is high, a large market indicates a more diverse demand for differentiated goods. Economic size also reflects the suly otential and therefore the exort otential in general, but more likely for differentiated goods since the roduction of these goods are under increasing returns to scale. The average economic size is therefore exected to increase the volume of trade. 4

6 Differences in government exenditures between Sain and the trading artner (difg): This variable is used as a roxy of ublic services. The idea is that ublic services contributes to increase the efficiency of rivate firms and articularly in the case of horizontally differentiated goods since cometitiveness is not only based on rices but on others attributes, quality of inuts, services quality, etc...we exect a ositive sign for HIIT while the effect on VIIT is undetermined. Average government exenditures of Sain and the trading artner (G): Again, we also include the average value of government exenditure since the same differences can occur for different levels of exenditure. This variable should have a ositive imact on IIT whatever the definition is used. Distance (dist,): the geograhical distance (in km) between the Sanish caital (Madrid) and the caital of country is a roxy for transortation costs that is suosed to reduce any kind of trade. GINI index of the trading artner (GINI ): The higher the Gini index, the more egalitarian is the income distribution in the trading artner. The redicted sign is negative for VIIT since a more unequal distribution of income will favour the demand for a larger sectrum of quality roducts and ositive for HIIT since demand for horizontal differentiated roducts is a maor concern for medium and high incomes households. Trading blocks: We include a dummy that takes the value 1 if the trading artner belongs to the Euroean Union and 0 if not. Since trade barriers should reduce any kind of trade we exect that trading artners that maintains lower tariffs and non-.tariffs barriers should face higher levels of any kind of trade. Since the EU (and thus Sain) grants a referential access to develoing countries deending on their develoment level and their region of origin, we seek to cature the effect of tariffs and non-tariffs barriers by the mean of dummies for the 11 main regions considered by the common trade olicy. Thus, due to multicollinearity roblems (these grous resent similar GDP already taken into account or are in the same region of the world as reflected by the distance) all these dummies could not be taken into account in the estimation. Common language: we also introduce a dummy that catures the effect of idiomatic differences that takes the value 1 if Sanish is the official language of the trading artner and 0 if not. Number of flows considered: Some industries roduce more differentiated goods than others and some of them roduce more horizontal than vertical goods. In order to account for these country-industries secificities, and following Aturuane (1999) we introduce the number of flows registered at the 8 digit level in each industry between Sain and country. Three alternative secifications (model 3, 4, 5) have been tested in order to better cature the effect of demand, and endowments like hysical caital, human caital and technology caital. Thus, GDP er caita not only reflect a suly side henomenon but also an income effect. Besides, we also relace this variable by more direct roxies for factor endowment differences, namely hysical caital er caita (INV), technological caital er caita (R&D) and Human caital (exenditure in education). We also introduce differences in consumer exenditure in order to reflect the demand side henomenon (see Table III for a list of variables). 5. Econometric results In the model, the exlained variable is alternatively total intra-industry trade volume of trade (IIT k ) and vertical (DVIIT k ) and horizontal (DHIIT k ) intra-industry trade in order to identify differences in theirs determinants, where reresents the Sanish trade artner and k the industry according to the NACE CLIOR25 nomenclature. The model is tested in its 5

7 logarithmic form. We first resent the results for the cross-section analysis for year 1996 estimating by OLS method and thereafter the results of the Heckman rocedure Cross-section analysis with OLS method The overall icture for the OLS estimations (Tables IV.A-C) is the robustness of the results with the maority of them significant at the 1 ercent level and a high level adusted R- squared (ranging from 0.65 for DVIIT to 0.83 for total IIT). In general, variables show the exected signs or are not significant. Concerning the variables which we guess that have the same influence on any kind of trade like DifGDP and AvGDP, they show the exected signs, namely negative and ositive. Sharing the same border seems to increase any kind of trade also, Sanish intra-industry trade is higher with the EU members as exected and Dummies for industries are in general significant and ositive for IIT and VIIT but in more cases negative and significant for DHIIT. The imact of the distance is always negative and very significant. Whereas not resented here, estimations erformed for the volume of inter-industry-trade shows that the effect of distance is higher in this case than for IIT in general reflecting that transortation costs matter more for homogeneous roducts where rices are the imortant issue. More surrising is that distance seems to have a worse effect on DHIIT than on DVIIT. These results can be exlained by the fact that the nearest countries of Sain are also the countries that share the same tastes for goods and so, the countries that are more willing to engage in DHIIT with Sain. When GDP er caita is substituted by others roxies for endowments (model 3, 4, 5), the effect of distance increases drastically reflecting the fact that furthest countries are also those with lower income er caita or which income er caita differ much from Sain. Turning to the most imortant variable, DifGDPc, it affects negatively any kind of IIT, reflecting that the more similar the endowments the higher the level of IIT. This result is not surrising and in line with Nillson (1999) considering unequal artners as in this study. This result is more surrising for vertical differentiated roducts since imorts and exorts of similar roducts differentiated by their quality is exected to increase with difference in factor endowments while the effect should be the oosite for horizontal differentiated roducts. The works of Mora (2002) for intra-eu trade and of Blanes y Martin (2000) for the Sanish case show that vertical differentiated roducts can mainly be exlained by the comarative advantage exlication in the line of the H-O and Ricardian models. Though, differences in human caital or technological caital have a more evident ositive effect on DVIIT than hysical caital differences. In Blanes and Martin (2000) differences in GDP er caita have a negative influence on any kind of intra-industry trade both when OECD trading artners are considered or when a larger samle of both develoed and develoing countries are considered. Creso and Fontoura (2004) observed that the traditional determinants of trade ought to exlain the Portuguese volume of vertical intra-industry trade when considering a samle of 46 countries. Thus, the authors also include the interaction between the Gini variable and er caita income differences. We are scetical about this multilicative variable since its exected influence is not clear and we are unable to determine if the negative sign it dislays goes in the sense of the comarative advantage theory or, on the oosite, gives arguments against it. Conclusions from model 1 and 2 show that differences in er caita income are against the comarative advantage exlanation of DVIIT. To confirm this conclusion we aim at verifying if GDP er caita is a good roxy for caital difference or if a more refined measure of endowment could shed some light on the vertical exlanation (model 3, 4 and 5). Differences 1 Panel data estimations of each tye of IIT and tobit estimations on anel data for the AGL index have also been driven. Results didn't really differ and were omitted due to sace requirements. Results are available on request. 6

8 in consumer exenditure do not dislay the exected sign since they affect ositively any kind of IIT, while a higher average in consumer exenditure results in less intra-industry trade. We guess that this unexected results is due to the fact that inter-industry trade is the most excellent trade between Sain and countries that are really different from a demand size ersective. This intuition is artly confirmed by the fact that, if we use consumer exenditure er caita instead of total exenditure in consumer goods, the variables aear not significant. On the other side, models 3, 4 and 5 cature better the effect of suly factors. For any kind of IIT, differences in investment exenditure have a negative influence and we do not observe differences between the influences of this variable on vertical or horizontal differentiated goods. Differences in research and develoment exenditure increase IIT and DV IIT while this effect is less clear in the case of DH IIT (the coefficients are not always significant). Influence of technology differences aear to be confirmed here. The differences in human caital factor endowments measured by differences in sending on education er caita have a negative effect on each tye of intra-industry trade. The differences on government exenditures do not have a stable influence on any kind of intra-industry flows. In sum, results from the OLS estimations do not confirm the imact of human caital differences on DV or DH IIT, certainly due to the bad quality of the roxy used. The most imortant result is that technological differences unambiguously enhance intra-industry trade while differences in hysical caital have a significant negative effect on both tye of intraindustry trade. Without forgetting the imortant outcome deduced from models 3, 4 and 5, we will now ay more attention to model 1 since we seek to find more evidence in favour or in oosition of the comarative advantage exlanation of vertical intra-industry trade as far as differences in hysical caital is concerned (model1). More consistent results have been searched considering the heterogeneity of countries. In Table V, we resent the estimation of model 1 for EU countries on one art, and other selected countries namely countries from OECD, Asia, Latin America, MNA and CEEC. Interesting results are found. In this case, differences in GDP er caita still influence negatively DVIIT for others countries (and thus overall IIT) but are not significant for HIIT and neither exlain significantly vertical intra-industry trade with the EU. Another imortant feature is that a high GINI index enhances any kind of IIT in the case of others countries while reduce IIT in the case of the EU. These results confirm our initial intuition that determinants of IIT differs slightly if we consider whether similar countries or unequal artners Cross-section estimation with Heckman method One roblem of the estimations roosed in the revious section is that we don't take into account the flows that resent a zero value since we use the logarithmic transformation of IIT volume. We argue that we should take into account the determinants of overall IIT when considering the difference between determinants of vertical whether horizontal IIT. That is, exlaining variables (both country and industry-secific effects) may exlain the fact that countries engage in IIT but could influence in a different way the fact that these two-way flows concerned horizontal or vertical differentiated roducts. Since these shares are very low in some case and we guess that determinants of IIT could differ strongly if countries share secific characteristics, we finally use the Heckman estimation method that allows to into accounting this selection effect. It consists in estimating searately a robit equation and an ordinary least squares equation. Thus rior to the estimation of equation above for DV IIT and DH IIT, we must define a binary variable D i according to the following scheme: D k 1 = 0 if f IIT IIT k k 0. = 0 7

9 This variable takes the value 1 if total intra-industry trade exists and 0 in the oosite case. Using D i as the deendent variable, we estimate the robit equation: Pr( D = 1 / z ) = F( β' z ) k k k The statistically significant variables will contribute to exlaining exclusively the robability of aearance of intra-industry trade flows. The next ste consists of studying the determinants of DV IIT and DH IIT, rovided that IIT i is different from zero. We should take into account the bias introduced due to the elimination of the zero observations from the samle when secifying the model, that is E( DVIIT / ITT 0 ) = δ ' z + E( u / ITT 0 ) k k k k k Where z k reresents the set of exlanatory variables defined in the revious section for the industry k and country and the second term of the right-hand side of the equation is roortional to the inverse of the Mills ratio. The estimation of this inverse is obtained from the robit model estimated in the first stage of the analysis. Estimations have been driven for two different sets of countries: all countries and excluding EU and countries with lowest income what finally means MNA, CEEC. NIC of ASIA and OCDE countries. Table V dislays the results for both robit estimation (selection equation) and the OLS estimation. Let look first at the results of the robit estimations that exlain the robability for IIT to occur. Since we use the same variables (model 1) as for the OLS estimation for ositive IIT volume, results of the robit estimation in table VI are directly comarable with model 1 in table IV.A. All the variables resent the exected sign or are not significant indicating that the variables used exlain better the level of IIT volume than the reason why some bilateral flow consist in inter or intra-industry trade. It is articularly the case for differences in market size (GDP) that is not significant and the average income er caita which is significant and has a ositive sign only when EU countries are included. The results also suggest that IIT trade is more likely, the higher the aggregate level of income, as exected. The indicator of differences in factor endowments has a significant negative sign which imlies that some degree of similarity in endowments is a condition for IIT to aear. With regard to the remaining variables, nor distance nor the GINI index have a significant imact on the robability to engage in IIT. Let turn now to the second stage of the estimation, where the level of vertical or horizontal intra-industry trade is exlained. Here, different atterns are achieved since difference in market size has a negative and significant imact on the volume of IIT trade (regardless to its nature) but difference in factor endowment doesn't have a significant effect in determining the level of IIT. Thus, distance, GINI and the dummies for contiguity and EU membershi do resent the exected signs while they fail to exlain significantly the robability of IIT to occur. 6. Conclusions In this aer, we have investigated the nature of Sanish intra-industry trade and found that intra-industry trade with CEEC, Asian and Mediterranean countries has increased considerably since the middle of the Nineties. The second aim of the aer was to determine how comarative advantage affects DVIIT. According to OLS estimations, technological differences do increase DVIIT while hysical caital differences decreases it. Results obtain with the Heckman method suort the idea that differences in hysical caital do lay a role for IIT to occur but the level of vertical and horizontal IIT is better exlained by the roximity of artners, the similarity in develoment level and size of market. Furthermore, our results suggest that the variables considered, mostly country-secific do have the same imact on vertical and horizontal IIT with emergent countries. 8

10 7. References Abd-el-Rahman, K..S. (1986): Réexamen de la définition y mesure des échanges croisés de roduits similaires entre les nations, Revue Économique, nº 1, París. Aturuane,C., Dankov, S., Hoekman, B. (1999) Horizontal and vertical intra-industry trade between Eastern Euroe and the Euroean Union, Weltwirtschaftliches Archiv, vol. 135(1), Blanes, J. V., Martin, C. (2000): The Nature and Causes of Intra-Industry Trade: Back to the Comarative Advantage Exlanation? The Case of Sain, Weltwirtschaftliches Archiv, vol. 136 (3), Creso, N. and Fontoura, P., (2004), Intraindustry Trade by Tyes: Eaht can we learn from Portuguese Data?, Review of World Economics, Vol 140(I), Diaz Mora C., (2002) The role of Comarative Adventage in Trade within Industries: A Panel Data Aroach for Euroean Union, Wetwirtschaftliches Archiv, Vol.138 (2), Durkin J.T., Kryier M. (2000) Difference in GDP er caita and the Share of Intraintraindustry Trade: The Role of Vertically Differentiated Trade, Review of International Economics, 8(4), november. Falvey, R.E. (1981): "Commercial Policy and Intra-Industry Trade", Journal of International Economics, vol. 11, Falvey R.E. and Kierzkowski, H. (1987): "Product Quality, Intra-Industry Trade and (Im)Perfect Cometition", en Kierzkowski H. (ed.), Protection and Cometition in International Trade, Clarendon Press, Oxford, ca. 11, Flam H., Helman E., (1987), Vertical Product Differentiation and North-South Trade, American Economic Review, 77(5), Fontagne L., Freudenberg M. (1997): Intra-industry trade: methodological issues reconsidered, CEPII Working Paer, nº 97-01, Paris: CEPII. Fontagne L., Freudenberg M., Peridy N. (1998): Commerce international et structures de marché: une vérification emirique, Economie y Statistiques nº 135, , Gabszewicz J., Shaked A., Sutton J.(1981) International Trade in Differentiated Products International Economic Review, 22, 3, october, Greenaway, D., Hine, R. C. and Milner, C. (1994): "Country-Secific Factors and the Pattern of Horizontal and Vertical Intra-Industry Trade in UK", Weltwirtschaftliches Archiv, vol. 130 (1), Greenway, D., Hine, R. C. and Milner C. (1995): Vertical and Horizontal Intra-Industry Trade: a Cross-Industry Analysis for the United Kingdom, The Economic Journal, 105, Grubel, H.G., Lloyd P.J. (1975): Intra-Industry Trade, London: The Macmillan Press Ltd. Helman, E. (1981) "International Trade in the Presence of Product Differentiation, Economies of Scale and Monoolistic Cometition", Journal of International Economics, 11(3), Helman, E. and Krugman P.R. (1985): Market Structure and Foreign Trade: Increasing Returns, Imerfect Cometition and International Economy, Cambridge (Mass.): MIT Press. Krugman, P. R. (1980): "Scale Economies, Product Differentiation, and the Pattern of Trade", American Economic Review, 70, Lancaster, K. (1980) "Intra-industry trade under Perfect Monoolistic Cometition", Journal of International Economics, 10(2),

11 Martin A., Orts (2001): "A two-stage analysis of monoolistic cometition models of intraindustry trade", Investigaciones Económicas, 25(2), Martin J.A., Orts V. (2002): Vertical Secialization and Intra-Industry Trade: The role of Factor Endowments, Weltwirtschaftliches Archiv. 138(2), Nilsson, L. (1999) Two-way Trade between Unequal Partners: The EU and the Develoing Countries, Weltwirtschaftliches Archiv. 135(1), Aendix of Tables and Grahs GRAPH I: Adusted Grubel-Lloyd Index by regions (Vertical and Horizontal Intra-industry trade, ). EU Share of IIT OCDE % of total volume of trade Year Horizontal IIT Vertical IIT Grahs by codreg2 Dev. Asia Share of IIT NIC Asia % of total volume of trade CEEC MNA Year Horizontal IIT Vertical IIT Grahs by codreg2 10

12 % of total volume of trade Share of IIT ACP Latin Am NIS Middle East Year Horizontal IIT Vertical IIT Grahs by codreg2 Source: See Table I GRAPH II: Sanish Intra-industry trade with 188 countries and GDP er caita of the trading artner (1996 and 2000). IIT with Sain and GDP er caita 1996 IIT with Sain and GDP er caita 2000 Share of IIT (%) Share of IIT (%) GDP er caita, PPP (current international $) GDP er caita, PPP (current international $) Source: See Table I 11

13 Table I: List of variables Variable definition Source Trade volume Comext, Eustostat database CGDP GDP, PPP (current international $) WDI database RGDP GDP, PPP (constant 1995 international $) WDI database Distance Distance in km between caitals CEPII database CConsum Final consumtion exenditure (current LCU) WDI database rconsum. Final consumtion exenditure (constant LCU) WDI database CInvest Gross caital formation (current LCU) WDI database RInvest Gross caital formation (constant LCU) WDI database CGovern. General government final consumtion exenditure (current LCU) WDI database RGovern.. General government final consumtion exenditure (constant LCU) WDI database R & D Research and develoment exenditure (% of GDP) WDI database Education Public sending on education, total (% of GDP) WDI database Gini Gini Index WDI database # flows Number of 8.digit roducts traded Comext, Eustostat database Contiguity 1 if artners share the same border CEPII database CodeEU 1 if artner belongs from the EU(15) CEPII database 12

14 Table II: Correlations between alternative measures of trade tyes (Sanish trade, 1996). Year 1996 #obs: 188 IIT volume DVIIT volume DHIIT volume DVIIT volume DHIIT volume IIT Volume IIT AGL For all roducts (α=.25) (α=.25) (α=.15) (α=.15) /nº of roducts traded IIT volume DVIIT volume (α=.25) DHIIT volume (α=.25) DVIIT volume (α=.15) DHIIT volume (α=.15) IIT Volume / nº of roducts traded IIT AGL DVIIT AGL (α=.25) DHIIT AGL (α=.25) DVIIT AGL (α=.15) DHIIT AGL (α=.15) DVIIT AGL (α=.25) DHIIT AGL (α=.25) DVIIT AGL (α=.15) DHIIT AGL (α=.15) Notes: IIT: volume of intra-industry trade; DVIIT: IIT of vertical differentiated roducts; DHIIT: IIT of horizontal differentiated roducts; AGL: Adusted Grubel-Lloyd Index

15 Table III.A: Adusted Grubel Lloyd Index by industries, Sain s trade with lower income countries (% of total volume of trade) ACP Dev, Asia Latin Am MNA NACE CLIO R25 IIT DH DV IIT DH DV IIT DH DV IIT DH DV 01 Agricultural, forestry and fishery roducts 0,1 0,0 0,1 2,8 1,4 1,4 1,6 0,1 1,5 0,9 0,5 0,3 06 Fuel and ower roducts 0,0 0,0 0,0 0,1 0,0 0,1 0,3 0,2 0,2 0,1 0,0 0,1 13 Ferrous and non-ferrous ores and metals 0,3 0,0 0,3 4,1 1,9 2,2 0,7 0,1 0,6 1,3 0,3 1,1 15 Non-metallic minerals and minerals roducts 0,1 0,0 0,1 3,7 0,5 3,2 3,9 1,1 2,8 3,1 1,0 2,2 17 Chemical roducts 1,2 1,1 0,1 5,2 2,5 2,7 6,7 3,3 3,3 5,0 2,0 3,0 19 Metal roducts 0,4 0,0 0,4 5,8 2,4 3,4 3,2 0,1 3,1 4,1 0,2 3,9 21 Agricultural and industrial machinery 0,6 0,1 0,5 6,1 1,3 4,7 5,3 0,6 4,6 3,4 0,6 2,8 23 Office and data rocessing machines, recision and otical instruments 1,3 0,3 1,0 15,2 8,9 6,3 4,7 0,5 4,3 6,5 1,2 5,3 25 Electrical goods 0,2 0,1 0,2 11,7 8,2 3,5 9,6 1,1 8,5 6,5 1,1 5,4 28 Transort Equiment 21,1 20,4 0,7 3,6 1,2 2,4 17,4 2,5 14,9 7,4 5,1 2,4 36 Food, Beverages, tobacco 0,3 0,1 0,2 0,8 0,2 0,6 2,9 1,2 1,7 1,3 0,4 0,9 42 Textiles and clothing, leather and footwear 1,1 0,0 1,0 13,8 4,1 9,7 0,9 0,1 0,8 4,4 0,9 3,4 47 Paer and rinting roducts 0,1 0,0 0,1 1,5 0,4 1,0 3,0 0,3 2,7 5,1 0,8 4,3 48 Others manufactured roducts 0,4 0,1 0,3 12,8 2,0 10,7 4,3 0,5 3,8 9,5 1,2 8,3 49 Rubber and lastic roducts 0,4 0,0 0,4 22,0 10,1 12,0 3,8 2,5 1,4 7,8 2,3 5,5 Total 2,1 1,9 0,2 4,9 1,8 3,1 3,5 0,7 2,8 3,3 1,1 2,3 14

16 Table III.B: Adusted Grubel Lloyd Index by industries, Sain s trade with intermediate and high income countries (% of total volume of trade) EU OCDE CEEC NIC Asia NACE CLIO R25 IIT DH DV IIT DH DV IIT DH DV IIT DH DV 01 Agricultural, forestry and fishery roducts 9,5 3,9 5,6 0,9 0,1 0,8 0,5 0,2 0,3 0,3 0,1 0,2 06 Fuel and ower roducts 19,6 12,3 7,3 0,5 0,3 0,2 0,1 0,0 0,0 0,3 0,0 0,3 13 Ferrous and non-ferrous ores and metals 27,1 19,4 7,7 3,2 1,6 1,6 0,7 0,5 0,3 2,9 2,0 0,9 15 Non-metallic minerals and minerals roducts 21,8 5,6 16,2 3,4 1,0 2,3 6,2 0,7 5,6 2,1 0,2 1,9 17 Chemical roducts 36,0 14,9 21,1 17,9 2,2 15,6 6,8 1,1 5,8 6,6 3,2 3,4 19 Metal roducts 37,5 13,6 24,0 17,5 2,6 14,8 12,0 4,3 7,7 6,5 0,9 5,6 21 Agricultural and industrial machinery 26,5 10,1 16,4 17,8 3,5 14,3 13,6 2,5 11,2 6,8 0,9 5,9 23 Office and data rocessing machines, recision and otical instruments 21,2 4,7 16,5 12,3 2,0 10,3 12,7 6,3 6,3 7,4 0,4 7,0 25 Electrical goods 36,8 11,6 25,2 21,0 3,4 17,6 8,7 3,1 5,6 9,5 2,1 7,4 28 Transort Equiment 45,6 31,5 14,1 20,2 6,7 13,4 23,8 16,6 7,2 4,6 0,2 4,5 36 Food, Beverages, tobacco 15,9 7,0 8,9 2,7 0,8 1,9 1,2 0,3 0,9 3,1 0,1 3,0 42 Textiles and clothing, leather and footwear 32,4 7,9 24,6 9,1 2,2 6,9 8,0 3,0 5,0 6,3 0,4 5,9 47 Paer and rinting roducts 29,1 9,2 19,9 15,0 3,7 11,3 9,8 4,0 5,8 6,4 0,9 5,5 48 Others manufactured roducts 40,4 12,2 28,2 23,9 2,4 21,5 16,1 3,3 12,8 11,3 1,6 9,8 49 Rubber and lastic roducts 48,4 31,0 17,4 21,3 12,6 8,7 24,1 14,0 10,0 22,0 10,1 12,0 8,8 3,0 5,8 Total 33,0 17,1 16,0 13,4 3,1 10,3 13,0 6,7 6,3 4,9 1,8 3,1 6,2 0,9 5,3 15

17 TABLE IV.A : OLS Estimation of Total Volume of Intra-industry trade TOTAL ITT MODELO1 MODELO 2 MODELO 3 MODELO 4 MODELO 5 Intercet *** *** *** (4.464) (4.288) (11.93) (3.802) (1.73) Dif CGDP *** (0.081) *** (0.075) *** (0.524) Avr. CGDP *** (0.163) *** (0.153) *** (0.466) Dif CGDP c * (0.136) ** (0.107) Avr. CGDP c *** (0.33) *** (0.278) Distance *** *** *** *** *** (0.097) (0.090) (0.303) (0.351) (0.174) Dif Consum *** (0.342) (0.394) *** (0.315) Dif Invest *** (0.182) *** (0.207) *** (0.200) Dif Invest Lag ** (0.333) (0.327) Dif Govern *** (0.630) Avr Consum *** (1.005) *** (1.189) (0.630) Avr Invest *** (0.753) *** (0.839) Avr Invest.Lag *** (0.247) Avr Govern *** (0.744) * (0.893) (0.599) R & D *** (0.117) *** (0.159) * (0.111) Education ** (0.132) ** (0.219) Education Lag (0.207) Gini * (0.009) *** (0.019) (0.20) *** (0.016) # flows *** *** *** *** (0.0004) (0.0004) (0.0006) (0.0006) (0.0005) Contiguity *** (0.354) *** (0.364) (0.483) *** (0.522) *** (0.498) CodeEU *** (0.204) *** (0.202) *** (0.348) *** (0.399) *** (0.274) Dummy Sector Sig (a) Sig (a) Sig Sig (a) Sig (b) # of obs Ad. R-squared (*) siginificative at 10%, (**) significative at 5% and (***) significative at 1% (a) all sectors are significative ositive at 1% excet sector 6 which is negative at level 10% (b) all sectors are significative ositive excet sector 6 and sector 42 Source: See Table I

18 TABLE IV.B: OLS Estimation of Vertical Intra-industry trade DV ITT MODELO1 MODELO 2 MODELO 3 MODELO 4 MODELO 5 Intercet *** *** *** (4.562) (4.351) (12.18) (3.866) (1.675) Dif CGDP *** (0.081) *** (0.075) *** (0.534) Avr. CGDP *** (0.169) *** (0.156) *** (0.475) Dif CGDP c ** (0.139) (0.108) Avr. CGDP c *** (0.327) *** (0.284) Distance *** *** *** *** *** (0.100) (0.093) (0.310) (0.356) (0.168) Dif Consum *** (0.351) (0.400) ** (0.303) Dif Invest *** (0.188) *** (0.213) Dif Invest Lag *** (0.192) Dif Govern * (0.339) (0.331) *** (0.235) Avr Consum *** (1.028) *** (1.206) (0.607) Avr Invest *** (0.772) *** (0.912) Avr Invest.Lag *** (0.238) Avr Govern *** (0.760) (0.851) (0.567) R & D *** (0.135) *** (0.160) * (0.106) Education (0.192) ** (0.222) Education Lag (0.200) Gini ** (0.009) *** (0.019) (0.020) *** (0.015) # flows *** *** *** *** (0.0004) (0.0004) (0.0006) (0.0006) (0.0005) Contiguity *** (0.360) *** (0.366) (0.492) *** (0.528) *** (0.478) CodeEU *** (0.209) *** (0.206) *** (0.359) *** (0.408) *** (0.264) Dummy Sector Sig (a) Sig (b) Sig Sig (a) Sig (a) # of obs Ad. R-squared (*) siginificative at 10%, (**) significative at 5% and (***) significative at 1% (a) all sectors are significative ositive at 1%, only sector 36 at 5% and sector 42 at 10% (b) all sectors are significative ositive at 1%, only sector 13 and 36 at 10% Source: See Table I 17

19 TABLE IV.C: OLS Estimation of Horizontal Intra-industry trade DH ITT MODELO1 MODELO 2 MODELO 3 MODELO 4 MODELO 5 Intercet *** (5.067) Dif CGDP *** (0.095) Avr. CGDP *** (0.189) Dif CGDP c ** (0.157) Avr. CGDP c *** (0.368) Distance *** *** (4.780) *** *** (18.05) ** (0.087) (0.939) *** *** (0.177) (0.777) (0.124) *** (0.316) *** *** (0.118) (0.108) (0.550) Dif Consum (0.492) Dif Invest *** *** (5.959) *** (0.526) ** (0.504) ** (0.314) (2.195) *** (0.212) (0.392) (0.340) Dif Invest Lag (0.251) Dif Govern *** (0.510) (0.406) (0.293) Avr Consum * * ** (1.668) (1.848) (0.806) Avr Invest *** ** (1.470) (1.628) Avr Invest.Lag *** (0.313) Avr Govern ** * *** (1.060) (1.111) (0.774) R & D * *** (0.175) (0.200) (0.131) Education ** (0.244) (0.264) Education Lag (0.274) Gini *** *** *** (0.011) (0.033) (0.028) (0.020) # flows *** *** *** *** (0.0005) (0.0005) (0.0008) (0.0008) (0.0006) Contiguity *** (0.381) *** (0.379) (0.68º) *** (0.618) *** (0.569) CodeEU *** (0.231) *** (0.223) *** (0.515) *** (0.558) *** (0.330) Dummy Sector Sig (a) Sig (b) Sig (c) Sig (b) Sig (b) # of obs Ad. R-squared (*) siginificative at 10%, (**) significative at 5% and (***) significative at 1% (a) all sectors are significative ositive excet sectors 6, 15, 36 Y 42 (b) all sectors are significative ositive excet sectors 6, 15, 19, 36 Y 42 (cb) all sectors are significative ositive excet sectors 15, 19, Y 47 Source: See Table I 18

20 TABLE V: OLS Estimation for MODEL 1 of total IIT, DVIIT, DHIIT : EU and selected emergent countries TOTAL ITT DV ITT DH ITT EUROPEAN OTHER EUROPEAN OTHER EUROPEA OTHER MODEL 1 UNION COUNTRIES UNION COUNTRIES N UNION COUNTRIES Intercet *** *** *** (17.51) (5.64) (16.25) (6.00) (26.21) (6.50) Dif CGDP *** *** *** *** *** (0.393) (0.079) (0.365) (0.083) (0.586) (0.093) Avr. CGDP *** *** *** *** *** *** (0.294) (0.191) (0.367) (0.204) (0.594) (0.221) Dif CGDP c * ** ** (0.480) (0.185) (0.448) (0.197) (0.730) (0.223) Avr. CGDP c * *** *** ** *** (1.922) (0.363) (1.783) (0.386) (2.908) (0.416) Distance *** *** *** *** *** *** (0.341) (0.123) (0.316) (0.131) (0.503) (0.149) Gini *** *** *** (0.030) (0.009) (0.028) (0.010) (0.045) (0.012) # flows *** *** *** (0.0006) (0.0008) (0.0005) (0.0009) (0.0009) (0.001) Contiguity *** *** (0.336) (0.364) (0.312) (0.312) Dummy Sector Sig (a) Sig (d) Sig (b) Sig (e) Sig (c) Sig (f) # of obs Ad. R-squared (*) siginificative at 10%, (**) significative at 5% and (***) significative at 1% (a) all sectors are significative ositive at 1% excet sector 15 at level 10%, and sector 6 which is negative (b) all sectors are significative ositive excet sector 15 (c) only sectors 6,13, 17 21, 28 and 49 are significatives, (d) all sectors are significative ositive excet sector 36 and 42 (e) all sectors are significative ositive excet sector 13, 36 and 42 (e) all sectors are significative ositive excet sector 6 and 42 Source: See Table I 19

21 ALL REGIONS SOME TABLE VI: Estimation (HECKMAN S METHOD) for MODEL 1 of DVIIT and DHIIT : all countries and selected emergent countries. DV ITT DH IIT MODEL 1 ALL REGIONS SOME COUNTRIES (d) COUNTRIES (d) Intercet *** *** *** *** (4.721) Dif CGDP *** (0.084) Avr. CGDP *** (0.170) Dif CGDP c (0.141) Avr. CGDP c *** (0.388) Distance *** (0.097) Gini *** (0.008) # flows *** (0.0004) Contiguity *** (0.363) Dummy EU *** Dummy Sector SELECTION EQUATION Intercet *** (23.06) Dif CGDP (0.299) Avr. CGDP *** (0.572) Dif CGDP c * (0.281) Avr. CGDP c ** (0.648) Distance (0.099) Gini (0.006) # flows *** (0.002) Contiguity Dummy EU (0.517) Dummy Sector (5.656) *** (0.087) *** (0.217) ( ) *** (0.366) *** (0.115) (0.009) *** (0.0009) (5.141) ** (0.095) *** (0.189) (0.155) *** (0.374) *** (0.115) *** (0.010) *** (0.0005) *** (0.374) *** (0.224) (6.049) (0.094) *** (0.230) (0.175) *** (0.386) *** (0.131) (0.303) *** (0.0010) (0.208) Sig (a) Sig (c) Sig (b) Sig (a) *** (4.217) (0.531) *** (1.446) * (0.299) (0.727) ( ) ** (0.010) *** (0.002) *** (30.60) (0.396) *** (0.777) ** (0.301) *** (0.701) (0.124) (0.009) *** (0.002) *** (4.721) ** (0.661) *** (1.922) (0.349) (0.843) * (0.163) (0.151) *** (0.0035) (0.570) Sig Sig (a) Sig (a) # of obs # of censored obs # of uncensored obs (*) siginificative at 10%, (**) significative at 5% and (***) significative at 1% (a) all sectors are significative ositive at 1% excet sector 15 at level 10% excet 6,13,21,36,42 (b) all sectors are significative ositive excet sector 6,15,36,42 (c) only sectors 6,36 and 42 (d) MNA, CEEC. NIC of ASIA and OCDE countries Source: See Table I 20