Dino Martellato Università Ca Foscari di Venezia

Size: px
Start display at page:

Download "Dino Martellato Università Ca Foscari di Venezia"

Transcription

1 UNIVERSITA CA FOSCARI DI VENEZIA OPENNESS, DEPENDENCY AND PRODUCTIVE EFFICIENCY IN NINE EUROPEAN COUNTRIES Dino Martellato Università Ca Foscari di Venezia Miguel Angel Tarancon Moran Universidad de Castilla La Mancha, Toledo DIPARTIMENTO DI SCIENZE ECONOMICHE, UNIVERSITÀ CA FOSCARI DI VENEZIA N.d.L June 2005

2 OPENNESS, DEPENDENCY AND PRODUCTIVE EFFICIENCY IN NINE EUROPEAN COUNTRIES by Dino Martellato, Università Ca Foscari di Venezia, and Miguel Angel Tarancon Moran Universidad de Castilla La Mancha, Toledo Abstract Among the many factors exlaining the disaointing economic erformance of the EU there is, in articular, inefficiency in roduction and it is likely that efficiency is lower in the sectors which are less exort-oriented and where firms are more rotected against cometition. If this is true, the firms in less exort-oriented sectors ass their inefficiency to the firms in those that are more exort-oriented. This means that country cometitiveness is likely to increase when efficiency is increased in the non-exorting sectors or when interdeendency between sectors decreases in the country itself. With this in mind, we look into the structure of three large Euroean countries (Germany, Italy and Sain) and six small countries (Austria, Belgium, Denmark, Finland, The Netherlands and Sweden). Using the 995 comarable inut-outut tables we first isolate the more exort-oriented sectors from the others and then measure the degree of deendency of exort-oriented sectors on the others, in each country. We use an index of deendency which considers both direct and indirect linkages between sectors. Thereafter, we focus roductive efficiency in the non exort-oriented sectors and we roose a very simle index of efficiency which does not require knowledge of the best ractice technology, as is usual in the available literature. Instead of searching for the most efficient or best ractice technology, we look at the best country. This leads us to rank the nine countries according the degree of their efficiency on the basis of domestic and total inut coefficients. We find some regularities: Austria always comes first and Italy always eighth. By using total inut coefficients, differences between the nine countries aear always to be larger than those obtained with domestic coefficients. The efficiency index aears to be ositively correlated to the average growth rate in the eriod. Key words: oenness, deendency, efficiency, structure JEL codes: D57, O52, P5 2

3 INTRODUCTION The disaointing economic erformance of the Euro Zone, and notably of the larger founding members of the EU, troubles the whole Union. Nobody knows if erformance would have been better without EMU i.e. Economic and Monetary Union, but with its creation and enlargement, Euroe has certainly been ursuing both a defensive strategy in the world arena and a strategy of change as a trigger for growth. Globalization and structural change are deemed to catalyse the ositive forces in the economy and seed u growth. For both the layman and the educated economist, an ever larger and more integrated EU is not very different from globalization. Both imly higher mobility of caital, labour, goods and services. One of the main features of globalization obs migration intrigues both the layman and the educated economist. For the former it is a source of fear or hoe; for the latter, an oortunity to increase efficiency in roduction. For the economist, stronger integration can be equated with more efficiency and more efficiency with growth. From this oint of view, the economic erformance of many Euroean countries does not unambiguously rove the assertion that more integrated markets always increase efficiency and bring about better economic results, at least in Euroe and at least so far. Indeed, market integration, enlargement and globalization are having mixed results in Euroe. Thus the double equation integration equals efficiency and efficiency equals growth is not always true; most of the accession countries have erformed well, while the six founding countries have had a disaointing erformance, at least on average. In this aer we deal with oenness, deendency and efficiency by comaring nine different Euroean countries in 995. Three of them (Germany, Italy and Sain) are large; the other six (Austria, Belgium, Denmark, Finland, The Netherlands and Sweden) are smaller. As trade is an obvious examle of a means by which different economies integrate, we look at the degree of oenness of different sectors and make a distinction between sectors which are exort oriented and sectors which are more inward oriented (Section 2.). Then we consider the deendency which exists between the two grous of sectors (Section 2.) and we examine the degree of roductive efficiency in the non-exorting sectors (Section 3.). This allows us to rank countries by quantifying efficiency in the different sectors, countries and technologies by detecting for each country the best and the worst technology (Section 3.). We conclude by regressing the average growth rate in the eriod to the deendency and efficiency in 995. There aears to be a relation between efficiency and erformance. Although the analysis is very reliminary and artial, it can shed light on how the roductive structure affects economic erformance in a articular year. In order to say something about the relation between changes in deendency and efficiency and the growth erformance we should extend the analysis diachronically.. OPENNESS AND DEPENDENCY ACROSS SECTORS To understand whether a articular sector is exort oriented or not it is commonlace to see whether its exorts are large enough relative to the added value of the sector and to the national exorts. To measure the relative degree of oenness in a articular sector, therefore, we first comare the ratio between exorts and added value in the sector and the economy as a whole. That is: 3

4 Oenness : O = n i= n ex av i= ex av i i, for all s. where: ex are the exorts in the sector (country ), and av is the added value in P P the same sector. The index shows whether the sector is more ( O > ) or less oen ( < ) than the national average. To establish whether a certain country is sufficiently secialized in the same sector, we must comare its sector s share to the average in all countries. ex Secialization : S = m n i= m = n = i= ex ex i ex i, for all s. The S indicates whether the country is more secialized ( S > ) or not ( < ). Ultimately, a sector is considered to be exort oriented it its exorts weigh enough in comarison to the exorts of other sectors and countries, i.e. if: O > and S >. The sectors fulfilling the double criterion are indicated in Table, and their exorts account for a share which goes from 42% (Denmark and Sain) to 68% (Germany and Finland) of the national exorts, as shown in Table 2. O S 4

5 Table /A. Oenness and Deendency es (I): Austria, Belgium, Denmark. Austria Belgium Denmark Exorting es Deendency es Exorting es Deendency es Exorting es Deendency es Oenness Secialization Oenness Secialization Oenness Secialization a Agriculture, hunting and forestry 0,53 0,63 0,57 0,57 0,80,76 2 b Fishing 0,00 0,00 0,74 0, 2,68 3,89 0,9 0,02 0,02 NO 3 ca Mining and quarrying of energy roducing materials 0,48 0, 0,00 0,00 2,27 3,06 0,35 0,57 0,20 NO 4 cb Mining and quarrying excet energy roducing materials 0,96 0,68 2,0,27,4,6,63 YES,3 0,45 5 da Manufacture of food roducts; beverages and tobacco 0,76 0,48,5,26,80,45 2,60 YES 2,76 2,87,80,08,94 YES 6 db Manufacture of textiles and textile roducts 4,,66 0,84,4 0,95 NO? 2,56,64 0,82,7 0,96 NO? 3,3 0,87 7 dc Manufacture of leather and leather roducts 4,83,54,05 0,62 0,65 NO? 2,77 0,28 2,45 0,37 8 dd Manufacture of wood and wood roducts 2,08,67,32 0,87,5 YES?,29 0,35 2,4 0,79 9 de Manufacture of ul, aer and aer roducts; ublishing and rinting 2,40,7,04,24,28 YES 0,92 0,36 0,77 0,3 0 df Manufacture of coke, refined etroleum roducts and nuclear fuel 0,66 0,28 2,,79 0,93,08,00 YES? 2,49 0,89 dg Manufacture of chemicals, chemical roducts and man-made fibres 4,6 0,78 2,85,64 0,80 0,79 0,63 NO 3,80 0,86 2 dh Manufacture of rubber and lastic roducts 3,55,38,04 0,87 0,9 NO? 2,38, 0,82 0,60 0,50 NO 2,55 0,96 3 di Manufacture of other non-metallic mineral roducts,55,59,42,22,74 YES,49,35,4,22,39 YES,45 0,99 4 d Manufacture of basic metals and fabricated metal roducts 2,74,35 0,90 0,84 0,76 NO 2,38,37 0,66,7 0,78 YES?,74 0,5 5 dk Manufacture of machinery and equiment n.e.c. 4,05,08 0,82 0,64 0,53 NO 2,27 0,40 3,4 0,92 6 dl Manufacture of electrical and otical equiment 3,78,2 0,72 0,92 0,67 NO 2,3 0,44 2,96 0,65 7 dm Manufacture of transort equiment 5,26 0,83 3,0,0 0,75,29 0,96 YES? 2,49 0,2 8 dn Manufacturing n.e.c. 2,02,7 0,84 0,38 0,32 NO,70 0,64 3,47,99,3,0,43 YES 9 e Electricity, gas and water suly 0,9 3,4 0,09 0,92 0,26 2,56 20 f Construction 0,05 4, 0,05 3,8 0,00 0,22 2 g Wholesale and retail trade; reair of vehicles, ersonal and hous. goods 0,54 0,92 0,75,9 0,7,2 22 h Hotels and restaurants 0,0 0,4 0,46 4,55 0,88 0,95 0,83 NO 0,07 0,68 23 i Transort, storage and communication 0,74 0,59,00 0,93,79,93 0,64,5 0,73 NO? 24 Financial intermediation 0,27,85 0,32,96 0,06 0,32 25 k Real estate, renting and business activities 0,37,24 0,39,56 0,4 0,49 26 l Public administration and defence; comulsory social security 0,00 0,00 0,00 0,00 0,04,67 27 m Education 0,00 0,00 0,00 0,6 0,02 5,00 28 n Health and social work 0,00 0,00 0,00 5,9 0,00 0,00 29 o Other community, social, ersonal service activities 0,0,0 0,6,3 0,04 0,39 30 Activities of households 0,00 0,00 0,00 0,00 0,00 0,00 Source: our elaborations on EUROSTAT data. 5

6 Table /B. Oenness and Deendency es (II): Finland, Germany, Italy. Finland Germany Italy Exorting es Deendency es Exorting es Deendency es Exorting es Deendency es Oenness Secialization Oenness Secialization Oenness Secialization a Agriculture, hunting and forestry 0,23 0,54 0,68 0,56 0,56 0,75 2 b Fishing 0,8 0, 3,54 0,2 0,68 0,3 3 ca Mining and quarrying of energy roducing materials 0,2 0,04 0,29 0,30 0,09 0,04 4 cb Mining and quarrying excet energy roducing materials 0,89 0,63 0,8 0,4 0,84 0,34 5 da Manufacture of food roducts; beverages and tobacco 0,55 0,38,0 0,59 0,84 0,59 6 db Manufacture of textiles and textile roducts 2,2 0,50 2,66 0,8 2,47 3,3 0,54 0,54 0,29 NO 7 dc Manufacture of leather and leather roducts 2,32 0,38 2,65 0,36 3,62 5,62,02 0,4 0,42 NO? 8 dd Manufacture of wood and wood roducts 3,02 3,32 2,4 0,83,77 YES? 0,56 0,2 0,59 0,25 9 de Manufacture of ul, aer and aer roducts; ublishing and rinting 2,52 4,26,05 0,96,0 YES?,56 0,62 0,87 0,34 0 df Manufacture of coke, refined etroleum roducts and nuclear fuel,33 0,98 0,86 0,38 0,56 0,42 dg Manufacture of chemicals, chemical roducts and man-made fibres 2,02 0,5 4,04,36,42 0,58 0,82 NO? 2, 0,7 2 dh Manufacture of rubber and lastic roducts,93 0,70 2,20,7 0,89 0,32 0,29 NO 2,0,8,07,02,09 YES 3 di Manufacture of other non-metallic mineral roducts,7 0,63,00 0,88,77 2,06,02,8,20 YES 4 d Manufacture of basic metals and fabricated metal roducts,90,6 0,65 0,62 0,40 NO 2,4,2,04 0,58 0,6 NO?,46 0,94 5 dk Manufacture of machinery and equiment n.e.c. 3,28,5 0,62 0,44 0,27 NO 3,57,3 0,86 0,20 0,7 NO 4,25,47,2,08,3 YES 6 dl Manufacture of electrical and otical equiment 3,5,56 0,54 0,78 0,43 NO 2,7,08,04 0,70 0,73 NO? 2,85 0,83 7 dm Manufacture of transort equiment 2,76 0,34 4,5,55 0,76 0,74 0,56 NO 3,60 0,72 8 dn Manufacturing n.e.c. 2,43 0,65,53 0,58 3,2 2,3,39,7,62 YES 9 e Electricity, gas and water suly 0,02 0,29 0,05 0,49 0,0 0,4 20 f Construction 0,00 0,00 0,00 0,39 0,05 2,79 2 g Wholesale and retail trade; reair of vehicles, ersonal and hous. goods 0,5 0,7 0,57 0,80 0,49 0,78 22 h Hotels and restaurants 0,00 0,05 0,34 3,34 0,00 0,00 23 i Transort, storage and communication 0,85 0,79,04 0,7,9,04 0,76,3 0,86 NO? 24 Financial intermediation 0,09 0,38 0,0 0,59 0,39 2,42 25 k Real estate, renting and business activities 0,26 0,82 0,8 0,82 0,6 0,53 26 l Public administration and defence; comulsory social security 0,05,58 0,02 0,84 0,00 0,00 27 m Education 0,00 0,00 0,00 0,00 0,00 0,00 28 n Health and social work 0,03 42,64 0,00 0,00 0,00 0,00 29 o Other community, social, ersonal service activities 0,08 0,7 0,,35 0,06 0,54 30 Activities of households 0,00 0,00 0,00 0,00 0,00 0,00 Source: our elaborations on EUROSTAT data. 6

7 Table /C. Oenness and Deendency es (III): The Netherlands, Sain, Sweden. The Netherlands Sain Sweden Exorting es Deendency es Exorting es Deendency es Exorting es Deendency es Oenness Secialization Oenness Secialization Oenness Secialization a Agriculture, hunting and forestry,58 3,36,9 0,02 0,02 NO,68 3,63,9,26,49 YES 0,24 0,3 2 b Fishing 2,44,2,09 0,63 0,69 NO? 0,75 0,93 2,84 0,59 3 ca Mining and quarrying of energy roducing materials,62 6,60 0,5,0 0,56 NO? 0,02 0,0 0,4 0,04 4 cb Mining and quarrying excet energy roducing materials,4 0,38,46 0,93 2,5,95,29,04,35 YES 5 da Manufacture of food roducts; beverages and tobacco 2,32 2,43,08 0,03 0,03 NO? 0,90 0,95 0,52 0,26 6 db Manufacture of textiles and textile roducts 2,20 0,5,80,06 0,75 0,58 0,44 NO 3,23 0,4 7 dc Manufacture of leather and leather roducts,57 0,3 2,40 2,32 0,8 0,6 0,3 NO 3,25 0,9 8 dd Manufacture of wood and wood roducts 0,84 0,5 0,84 0,3 2,53 2,00,77,0,95 YES 9 de Manufacture of ul, aer and aer roducts; ublishing and rinting,3 0,55,5 0,44 2,06,84,04,7,22 YES 0 df Manufacture of coke, refined etroleum roducts and nuclear fuel 2,49 2,46 0,48 0,38 0,8 NO,95,42 0,70,25 0,88 NO? 2,39,8 0,45,0 0,50 NO? dg Manufacture of chemicals, chemical roducts and man-made fibres 3,8,67,9 0,87,03 YES? 2,83 0,88 3,35 0,78 2 dh Manufacture of rubber and lastic roducts 2,52 0,79 2,36,08 0,8 0,88 0,70 NO 2,54 0,86 3 di Manufacture of other non-metallic mineral roducts,2 0,63,63,94,7,23 2, YES,43 0,59 4 d Manufacture of basic metals and fabricated metal roducts,87 0,67,89 0,9 2,04,6 0,7,03 0,74 NO? 5 dk Manufacture of machinery and equiment n.e.c.,8 0,29 3,38 0,49 3, 0,98 6 dl Manufacture of electrical and otical equiment 2,5 0,65 3,75 0,70 3,53,33 0,95,06,0 YES? 7 dm Manufacture of transort equiment 2,56 0,38 5,58,90,04,0,05 YES 3,23,9 0,78,07 0,83 NO? 8 dn Manufacturing n.e.c. 0,75 0,33,5 0,66 2,3 0,64 9 e Electricity, gas and water suly 0,07 0,95 0,0 0,3 0,08 0,88 20 f Construction 0,05 3,82 0,00 0, 0,00 0,00 2 g Wholesale and retail trade; reair of vehicles, ersonal and hous. goods 0,87,34 0,64 0,84 0,88,06 22 h Hotels and restaurants 0,0 0,0 0,00 0,00 0,00 0,00 23 i Transort, storage and communication,67,36,47,20,76 YES,7 0,88 0,74 0,72 24 Financial intermediation 0,09 0,52 0,9,02 0,22 0,97 25 k Real estate, renting and business activities 0,35,2 0,44,25 0,27,33 26 l Public administration and defence; comulsory social security 0,03,48 0,00 0,00 0,04,36 27 m Education 0,02 2,92 0,00 0,00 0,00 0,00 28 n Health and social work 0,00 0,00 0,00 0,00 0,00 0,00 29 o Other community, social, ersonal service activities 0,35 4,08 0, 0,93 0,06 0,6 30 Activities of households 0,00 0,00 0,00 0,00 0,00 0,00 Source: our elaborations on EUROSTAT data. 7

8 Table 2 Exorts in the exorting sectors in relation to the total (%) Number of sectors Austria 0 59 Belgium 9 6 Denmark 5 42 Finland 5 68 Germany 6 68 Italy 7 5 Netherlands 7 55 Sain 7 42 Sweden 7 54 % To measure the deendency of exorting sectors on the remaining sectors one could use many criteria. We oted for a air of criteria aimed at assessing deendency both in direct and indirect terms. More recisely, we assume that deendency can be related to the coefficient size because the larger the weight in the buying sector costs, the stronger the effect of any change in the efficiency of the sulier sector. If we assume the number of non-exorting sectors in the concerned country to be d ( n d being the number of the exorting ones), size-deendency can be defined as the ratio between the sum of sector inuts from non-exorting sectors and the average sum of similar inuts for all exorting sectors; that is: Deendency : C C k k = d i= n d aik, for all k n d. d a = i= i n d The exorting-sector k is deendent on the non-exorting sectors if the condition > holds, for obvious reasons. The results for all exorting sectors in each country are shown in the third column of its section (Table ). The measure based on the size of technical coefficients shows an imortant limitation. It does not take into account the indirect effects that the sales from i to rovide. In order to avoid this limitation, an elasticity analysis can be alied. The elasticity obviously gives the effects of the changes in the elements of an inverse matrix caused by a change in an element of the original inut matrix. This results can be seen in Sherman - Morrison (950) and Maas (980), among others. According to Maas, the elasticity of the roduction of a sector i with resect to a change in the coefficient a kl is aroximated by the following exression: dxi ε xi x a = da, or : bik akl xl ε i kl x a = i kl xl ( a kl b lk ) x i axl where, b ik is an element of the Leontief inverse, and where it has been assumed that : da Δa = a with 0. xl kl kl 8

9 Thus, in order to calculate the deendency of an exorting sector on the non-exorter sector i, we will also consider the elasticity of its roduction with resect to the urchases from sector i, because only in such a way can the indirect connections be contemlated. The basic remise for incororating indirect effects is that these are greater if any given increment in the inut coefficient involves a larger change in the outut. The formula above, in our case, a i simlifies: ε b iai x a = i ( aib i ). That elasticity is a measure of deendency of the exorter on the non-exorter sector i. Now, continuing to assume that the first d sectors are the non-exorting ones, we can comute the average of the corresonding elasticities: d b a i i i= ( a b ) d i i ε x a = i d and use as the reference value the following average elasticity of deendency on nonexorting sectors: n b iai ( ) ε * i= aib i x = ai n with i. The reason for excluding from the average elasticity is that it is usually high, but it does ε x a ii not indicates real deendency on the rest of the economy. Therefore, to avoid underestimating the relative deendency on the non-exorters, we decided to exclude it. The index of deendency based on the elasticities is therefore: d ε x ai Deendency : E =. * ε x ai The results for all exorting sectors in each country are shown in the fourth column of its section (Table ). The elasticity-based index takes into account the direct and indirect relations between sellers and buyers. While the size-deendency index neglects indirect linkages, the elasticitydeendence index focuses only on otential deendency. To simultaneously cature the two asects we roose the following combined index: index : D = C E. Ultimately, the exorting-sector can be considered as deendent on non-exorting sectors if the combined index is: >, and vice versa. The results of the combined index for D all exorting sectors in each country are shown in the fifth column of its section (Table ). The sixth column in each section of Table summarizes our assessment. The YES s or a NO s indicate unambiguous cases of strong deendency and, resectively, weak deendency. A YES? and a NO? indicate more uncertain cases of deendency of a articular exorting sector on non-exorting sectors. 2. PRODUCTIVE EFFICIENCY 9

10 Efficiency is a key concet in the economics of consumtion and roduction. In the latter case it is commonlace to make a distinction between technical efficiency and economic efficiency, on the one side, and between static and dynamic efficiency, on the other. It is ossible, furthermore, to make a distinction between efficiency in one sector or industry and efficiency in the national economy as a whole. Efficiency in one secific industry is tantamount to minimal inut quantities or minimal inut cost, deending on whether one is considering technical or economic efficiency. If the selected industry roduces one single outut, efficiency is obtained when no inut can be reduced without decreasing outut. When we come to efficiency in the whole economy, many technologies and many oututs have to be considered simultaneously, which means that one has to understand how the choice between different outut combinations translates into the choice between different sets of technologies. The issue was resolved more than fifty years ago. Koomans (95), Dorfman, Samuelson and Solow (958), and others soon found the conditions according to which the choice between different technologies can be decouled from that concerning final outut comosition. From then on, roductive efficiency has been considered in relation to the use of a given amount of a single rimary inut (e.g. labour) or given amounts of a few rimary inuts to which constant relative rices are attached. In the famous substitution (or, non-substitution) theorem, it is shown that efficient technologies are invariant with reference to the final demand comosition, but only when oint roduction and economies of scale are absent and in the static case. Later on, Morishima (965) extended the theorem to the dynamic case by showing that efficient technologies are also invariant with regard to rices, rovided that the economy is Walrasian and either the interest rate or the wage rate is given. In the real world, such ideal conditions are not fulfilled and it is likely, therefore, that inut coefficients change with the final demand comosition. The roblem of assessing the degree of efficiency in a roductive system seems to require the determination of a benchmark set of inut coefficients. How is it ossible to find such a benchmark if the inut coefficients change with the final demand? One ossibility is to comare actual inut costs to ideal inut costs and see whether the given final demand bundle can be efficiently satisfied or not. Assuming that roduction is not on the frontier, the aroach requires the definition of the cost structure which is efficient in relation to a given final demand bundle. This requires the identification of the final demand bundle worth considering and the associated efficient vector of inuts, i.e. the efficient vector of quantities and the comanion vector of rices, a formidable task indeed. Cella and Pica (200) have followed this line by using intermediate inuts rices reflecting minimum marginal costs of roduction in each sector, i.e. shadow rices. In a study of five OECD countries they found that sector inefficiencies are largely due to imorted goods. Zofio and Prieto (2005) tried to evaluate best ractice technical efficiency using data enveloment analysis, i.e. mathematical rogramming. The technique makes it ossible to extract a benchmark unitary frontier from the techniques actually in use in the different economies and thus assess otential efficiency gains in the form of inut requirement reduction. In actual economies, i.e. in economies where technologies are not invariant with reference to the final demand comosition, the roblem of searching for the most efficient inuts vector cannot be solved once and for all. The most efficient combination of inuts is, in other words, conditional on the given final demand vector, and any change in the latter is likely to imly a change in the former. It is, indeed, well known that even at the lower scale of roduction, i.e. at the lan level, built-in flexibility often is referable to mere static efficiency. Fuss and Mc Fadden (978), in articular, considered the roblem of lant design when the firm roduces a single outut but faces considerable variability or real uncertainty regarding relative inut rices. They concluded that when firms face such inter-temoral variation or uncertainty on inut rices, the concet of static efficiency is inadequate and ossibly irrelevant. We susect, furthermore, that in a world oulated by thousands of firms roducing thousands of roducts and facing considerable uncertainty regarding not only relative rices of inuts and oututs but also demand comosition, the ideal of best ractice technology cannot be disoined from a articular demand vector. No technology, therefore, could be deemed efficient 0

11 in general as, at best, it can be considered efficient for a articular final demand bundle and for a articular system of relative rices. As inut coefficients bear uon efficiency in roduction, any change in the final demand comosition and successively in the inut coefficients is translated into changes in efficiency, ossibly making inefficient what initially was assumed to be efficient. It can be added that as the structural change not only requires a lot of time, but also tends to be rocrastinated, even the initial choice could have been inefficient. In conclusion, we assume not only that efficiency is not reached everywhere, but also that it is not the same across countries. To roceed, we need a definition of efficiency which is viable for emirical analysis but takes into consideration the relation between technologies and final demand. In an exchange economy, inefficient sellers make buyers ay more and this translates into lower rofits for the buyer and higher rices for the buyer customers. In an interdeendent system, in other words, roducers and sellers ass their efficiency or, deending on the circumstances, inefficiency on the buyers. As any given bundle of final goods and services usually requires the use of a certain amount of intermediate goods and services, the ursuit of efficiency requires that, for that bundle, the cost of those inuts be minimized. In a closed inut-outut system, we can assume that efficiency is obtained when a given demand vector is delivered with the minimal difference between final and total demand. Efficiency, therefore, is maximal for any final vector f, when: x = ( I Α) f = f, or when: Α = 0, and x = f, where the symbol Α refers to the matrix of total or technical inut coefficients. In this articular, and extreme case, efficiency is indeendent from the final demand vector f. Efficiency is lowered by any increase in the difference between the vectors f and x or by any increase in the multiliers; it being understood that, in this instance, the difference x i i i i f > 0 (for any alternativei ), reflects both f and the associated A. In the oen economy one should take into consideration intermediate imorts. This means that efficiency can be attained not only by roerly using domestic and imorted goods and services, but also by deciding rationally what to roduce and what to imort. It can be assumed that comanies seek to reduce costs by imorting intermediate goods and services, articularly in the exorting sectors. In the other sectors, imorts are limited or even imossible and the only ossibility of increasing efficiency is to change the technologies. Our task, therefore, is to see whether efficiency can be increased by combining the final demand vector with domestic inut coefficients different from the actual ones. In a system of countries such as the EU, it is ossible to calculate for any final demand vector in country s, the required total outut vector which would ideally be required by combining the given final vector f s with the matrix of domestic inut coefficients Ar (with Ai Αi ) of any other country r. The hyothetical total roduction vector in country s, with the domestic coefficients of country r, can then obviously be indicated as: x = ( I A ) f, for all r, s S, r s. rs r s We could argue that efficiency in the exorting is not the same as in the non-exorting sectors and argue that roducers face more cometition in the exorting sectors than in the nonexorting ones. Pushing this argument to the limit, one could add that in the exorting sectors, market cometition and market contestability automatically allow only efficient roducers to survive. In the non-exorting sectors, this outcome is less likely as inefficient roducers can safely rocrastinate change. We argue, therefore, that inefficiency is deeer in the non-exorting sectors and thus we are articularly interested in detecting inefficiency within them as the deendency of exorting sectors on non-exorting sectors comounds inefficiency in the former.

12 Assuming, as usual, that the n sectors can be divided into n d exorting sectors and d non-exorting sectors, efficiency in the non-exorting sectors, country s, can immediately be comuted by calculating total roduction in the non-exorting sectors, that is, and comaring this value with the amount of actual total roduction in the same sectors: x. On this base, we could argue that country following condition holds, and vice versa: i d irs i d x < x, iss i d xirs i d s is inefficient in its non-exorting sectors when the We are then rovided with a very simle way of measuring efficiency across the different countries. We assume that the most efficient country in the non-exorting sectors is the one which is able to minimize the result across all other countries by ideally transferring to them its own coefficients. By the same token, the less efficient country is the one which would always rofit from a reduction in the total roduction in its non-exorting sectors if it were ideally using the inut coefficients which are statistically observed in the other countries. As there are two varieties of deendency, one stemming from the use of domestically roduced intermediates and the other stemming from the correct use of domestic and imorted intermediates, in the next section we will consider both. iss 3. EFFICIENCY IN THE NINE EUROPEAN COUNTRIES By alying the first criterion ust defined, we could comare each one of the nine Euroean countries on the base of the 995 inut-outut data. The following table reorts ercentage differences in the total roduction in the different countries that result from the combination of different final demand vectors with different domestic inut coefficients. Along a given column, the final demand vector is the same, while along a given row, the inut matrix is the same. Our findings are easy to interret and show a clear attern which can be seen in Table 3 (roduction differences in the non-exorting sectors) and in Table 4 (column and row averages). Table 3 Efficiency in the non-exorting sectors, domestic coefficients 995. Austria Belgium Denmark Finland Germany Italy Netherlands Sain Sweden Austria 0-7,7 -,63-7,90-5,9-8,94-2,72-6,9-7,62 Belgium 2,27 0 6,33,85 4,6-0,89 4,27 4,49 0,76 Denmark 3,96-3,45 0-4,92-2,90-6,80 -,94-2,7-4,6 Finland 7,38 0,59 7,27 0,85 0,44 3,8 4,28-0,58 Germany 8,08 0,96 7,3-0,02 0 0,9 7,6 3,07 0,0 Italy 7,6-2,3 8,7,32 0,6 0 5,3,46 0,95 Netherlands 4,50-4,57,8-4,9-2,30-4,83 0-2,07-4,46 Sain 6,29-4,38 4,86-0,72-0,60-2,30 2,00 0-2,28 Sweden 7,8-0,63 4,08 0,3,8-2,67 2,90 0,67 0 Some countries, notably Austria, Denmark and the Netherlands, are clearly more efficient than others because they would always need to raise their roduction in the nonexorting sectors when using other countries coefficients (Table 4, column two). At the same 2

13 time, the other countries could always reduce their roduction in their non-exorting sectors when not using the coefficients of Austria, Denmark and the Netherlands (Table 4, column three) The difference between the two averages (Table 4, fourth column) clearly indicates the overall roductive efficiency in the non-exorting sectors of different countries. In this resect, Austria, Denmark and the Netherlands aear to be the most efficient in such sectors. Sain is in a slightly different osition as it benefits from its own technology, but it does not offer, on average, an equal advantage to the other countries. The remaining five countries (Germany, Sweden, Finland, Italy and Belgium) are in an entirely different situation: they would ideally benefit from the use of other countries technology while other countries would almost always increase their roduction thus decreasing their efficiency, if they were allowed to use the technology of those five countries. The difference between the corresonding averages makes it ossible to rank the nine countries (Table 4, last column) according their overall efficiency index in the non-exorting sectors. A oint to be underlined is that each country has its own set of non-exorting and exorting sectors, as indicated in the third, fourth and fifth rows of Table, and that the roosed index is based on the secific economic structure of the country. A different and equally interesting oint to remark is that it is ossible to find for any country (i.e. and more reciselu for its final demand bundle) the best and the worst set of domestic coefficients. The best alternative, i.e. the alternative with the highest efficiency in the non-exorting sectors, is obtained by using the Austrian domestic coefficients, everywhere. The worst result, i.e. the outcome yielding the lowest efficiency level in the non-exorting sectors, is obtained in some cases with the Belgian technology (Austria, Finland, Germany and Sain), in some others with the Italian technologies (Denmark, the Netherlands and Sweden) and in last two cases (Italy and Belgium) with the technologies available in Finland (see the figures underlined in Table 3). Table 4 Ranking based on domestic roductive efficiency Row averages Column averages Difference Rank Austria -5,48 6,38,87 Belgium 3,69-2,39-6,08 9 Denmark -2,60 4,23 6,83 2 Finland 2,78 -,67-4,45 6 Germany 3,03-0,44-4,70 7 Italy 2,58-2,87-5,45 8 Netherlands -,94 2,36 4,30 3 Sain 0,32 0,25-0,06 4 Sweden,52 -,97-3,49 5 The average real GDP growth rate from 990 to 2000 obviously reflects many factors and circumstances, however, it could be interesting to see whether the observed differences in efficiency and deendence are able to account for a decent fraction of the observed growth differences across sace. In a very simle regression analysis, it aears that the average real growth rate over the eriod is better exlained by differences in the efficiency index, than by differences in the degree of deendency or by the combined efficiencydeendency index. U to this oint we have used domestic inut coefficients, which means that the efficiency we have dealt with is the one concerning the domestically roduced intermediates. In order to assess efficiency according the use of intermediates from any geograhical origin we must use technical or total inut coefficients thus obtaining a different criterion for measuring efficiency. By redoing the analysis with total inut coefficients we obtain the results in Table 5. Along a row, the technical coefficients are constant and combined with different final vectors, while along a column, the final demand is given and the technical coefficients change. 3

14 Table 5 Efficiency in the non-exorting sectors, technical coefficients 995. Austria Belgium Denmark Finland Germany Italy Netherlands Sain Sweden Austria 0-47,96-88,24-7,6-93,67-88,2-70,02-78,52-93,82 Belgium 3, ,44 0,7-92,22-84,29-53,6-72,23-92,47 Denmark 2,44-48,49 0-3,26-93,68-85,36-60,63-78,83-93,89 Finland 25,4-37,52-83, ,5-85,79-56,73-74,29-9,77 Germany -5,93-5,56-87,66-4, ,2-70,8-80,65-94,3 Italy 7,49-39,24-86,62 -,28-92, ,48-76,69-93,08 Netherlands,26-4,26-86,67 6,44-92,55-88,3 0-74,87-92,43 Sain 5,48-38,7-85,78-4,26-92,89-86,67-62, ,9 Sweden 0,48-4,83-84,43 7,43-92,68-85,52-6,45-75,87 0 The differences across countries based on the efficiency in the use of all intermediate goods and services are larger than before, as can be seen from Table 6, which gives the results. Table 6 The ranking based on technical roductive efficiency Row averages Column averages Difference Rank Austria -63,06 7,8 70,87 Belgium -5,43-38,5 2,92 4 Denmark -5,3-76,2-24,9 7 Finland -55,2 -,72 53,48 2 Germany -54,73-82,52 53,0 3 Italy -49,64-76,9-27,26 8 Netherlands -50,9-55,49-4,58 5 Sain -50, ,07 6 Sweden,52-82,77-35,67 9 As before, we can calculate an efficiency index by taking the difference between the corresonding averages. A oint that is worth noting is that the absolute size of the differences between aearing in the third column of Table 6 increases over the corresonding entries in Table 4 a result that allows us to argue that by using imorted intermediate goods and services, all the nine countries are able to increase their efficiency. While Austria and Italy are able to retain their rank, resectively st and 8 th, the other countries scramble their ositions, although the differences narrow. Also in this case, the average growth rate over the eriod is well exlained by differences in the degree of deendency. 4. Data Source The whole exercise resented in this article is based on the inut-outut tables (IOT) of nine Euroean countries (Austria, Belgium, Denmark, Finland, Germany, Italy, The Netherlands, Sain and Sweden) for 995 which are freely downloadable from the Eurostat Database, website: htt://euroa.eu.int/comm/eurostat/. The flows refer both to domestic and total transactions and are at current rices. The IOT are aggregated to 30 different activity branches which are listed in the following table: 4

15 a Agriculture, hunting and forestry 2 b Fishing 3 ca Mining and quarrying of energy roducing materials 4 cb Mining and quarrying excet energy roducing materials 5 da Manufacture of food roducts; beverages and tobacco 6 db Manufacture of textiles and textile roducts 7 dc Manufacture of leather and leather roducts 8 dd Manufacture of wood and wood roducts 9 de Manufacture of ul, aer and aer roducts; ublishing and rinting 0 df Manufacture of coke, refined etroleum roducts and nuclear fuel dg Manufacture of chemicals, chemical roducts and man-made fibres 2 dh Manufacture of rubber and lastic roducts 3 di Manufacture of other non-metallic mineral roducts 4 d Manufacture of basic metals and fabricated metal roducts 5 dk Manufacture of machinery and equiment n.e.c. 6 dl Manufacture of electrical and otical equiment 7 dm Manufacture of transort equiment 8 dn Manufacturing n.e.c. 9 e Electricity, gas and water suly 20 f Construction 2 g Wholesale and retail trade; reair of motor vehicles, motorcycles and ersonal and household goods 22 h Hotels and restaurants 23 i Transort, storage and communication 24 Financial intermediation 25 k Real estate, renting and business activities 26 l Public administration and defence; comulsory social security 27 m Education 28 n Health and social work 29 o Other community, social, ersonal service activities 30 Activities of households SUMMARY In this aer we have comared the roductive structure in 995 of nine Euroean countries by focusing on oenness and deendency of exort-oriented sectors on non-exort-oriented sectors. We have also made a cross-country comarison on the base of efficiency in the lessexorting sectors using both domestic and total inut coefficients. First of all, we classified the thirty branches of economic activity according the degree of their oenness. Across sectors, the index aears to be rather smoothly distributed, but it is ossible to divide the exorting sectors from non-exorting sectors as most of the exorts are concentrated in ust a few branches. Their number ranges from five to ten, according the country, and their exorts account for 42% - 68% of the total. We then measured the degree of deendency of exorting on non-exorting sectors. In doing this we considered both direct and indirect linkages. Then we addressed the efficiency issue and tried to measure it in the non-exorting sectors. The simle efficiency index is based both on domestic and technical inut coefficients. The ranking of the nine countries changes, but Austria always comes first and Italy always comes eighth. The index shows large differences across countries in the efficiency of the non-exorting sectors. These differences are articularly large when technical coefficients are used and they are correlated with the average growth rate in the ten-year eriod

16 References. Cella, G., Pica, G. (200) Inefficiency Sillovers in Five OECD Countries: An Interindustry Analysis; Economic Systems Analysis, 3 (4), Dorfman, R., P. Samuelson, R. Solow, (958) Linear Programming and Economic Analysis; New York, McGraw Hill Book Co. Fuss, M., McFadden, D. (978) Flexibility versus Efficiency in Ex Ante Plant Design; in: M. Fuss, D. McFadden Production Economics: A Dual Aroach to Theory and Alications, vol. Production Economics; Amsterdam, North Holland P. Co.. Koomans, T. (95) An Analysis of Production as the Efficient Combination of Activities, in: T. Koomans (ed.) Activity Analysis of Production and Allocation, Cowles Commission for Research in Economics, Monograh 3; New York, John Wiley and Sons Inc.. Maas, S. (980) Die Reagibilität von Prognosen mittels Inut-Outut-Modellen auf Fehler im Datenmaterial (Berlin, Athenäum-Verlag), cited in Schnabl, H. (2003) The ECA-method for Identifying Sensitive Reactions within an IO Context, Economic Systems Research, 5, Morishima, M. (965) Equilibrium, Stability and Growth: A MultiSectoral Analisis. Oxford, Clarendon Press. Sherman, J., Morrison, W. J. (950) Adustment of an inverse matrix corresonding to a change in one element of a given matrix, Annals of Mathematical Statistics, 2. Zofio, J.L,. Prieto, A.M. (2005) Measuring Productive Efficiency in Inut-Outut Models by Means of Data Enveloment Analysis. Manuscrit, Universidad Autónoma de Madrid and IRNA-CSIC, Salamanca. 6