The demand for skills and labour costs in partner countries

Economics of Transition Volume 19(3) 2011, 611–637 DOI: 10.1111/j.1468-0351.2011.00412.x The demand for skills and labour costs in partner countries Evidence from the enlarged EU1 Alessia Lo Turco* and Aleksandra Parteka** *Department of Economics, Polytechnic University of Marche, Ancona, Italy. E-mail: a.loturco@ univpm.it **Department of Economics, Gdansk University of Technology, Gdansk, Poland. E-mail: [email protected] Abstract We analyse the consequences of trade integration in Europe (1995–2005) detecting how the labour costs in partner countries affect the domestic demand for high- and low-skilled labour in ‘Old’ (EU-15) and five ‘New’ EU member states. In general, independently of the skill level of workers, the results suggest complementarity between domestic and foreign labour. However, when we take into account the typology of sectors, the demand for the high skilled in low skill intensive sectors in ‘New’ EU members is boosted by the increase of the average labour Received: November 5, 2009; Acceptance: December 9, 2010 1 A previous version of the paper was circulated under the title: EU enlargement, economic interdependence and the labour markets in old and new member states. We thank the participants of the conference organized by the National Bank of Poland (Warsaw, 2008), PUE-PIEC Workshop (Università Roma Tre, Rome, 2009), the SMYE Conference (Istanbul, 2009), Aarhus Business School Departmental Seminar (Aarhus, 2009), and EEA Conference (Barcelona, 2009) for valuable comments and suggestions. We also thank Stefano Staffolani for his help. Financial support received from the Italian Ministry of Education, University and Research (Scientific Research Programs of National Relevance 2007 on European Union policies, economic and trade integration processes and WTO negotiation) is gratefully acknowledged. Aleksandra Parteka acknowledges the financial support from the Foundation for Polish Science. The two authors contributed equally to this paper. Any remaining errors are ours.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA 612 Lo Turco and Parteka cost in ‘Old’ EU members. Thus in low skill intensive sectors, the high skilled in ‘New’ member countries can substitute for employment in EU-15 countries. JEL classifications: F15, F16, J31. Keywords: Labour markets, trade, EU integration. 1. Introduction The effects of trade integration on the labour market are among the most debated consequences of globalization. In particular, the widespread feeling that increases in the cost of labour in developed countries can be offset through imports from low labour cost countries implies that the foreign labour force can become a substitute for the domestic labour force. For Europe, in particular, the 1990s brought the intensification of links between Eastern and Western European countries and the general opening up toward economies previously hidden behind the Iron Curtain. Increasing economic integration, initiated by trade agreements in the mid-1990s and completed by the enlargements in 2004 and 2007,2 has boosted the intensification of mobility of production factors and trade across Europe, especially in the form of outward processing trade (OPT). Wage differentials have influenced the location of separate phases of production processes, and by the 1990s, ‘New’ member states (NMS) (mainly Central and Eastern European countries, CEECs) were already important hosts of outsourcing practices for the EU-15 (Baldone et al., 2001). The importance of the processing trade in CEECs rose considerably in the 1990s – between 1988 and 1999 outward processing exports to (imports from) CEECs increased by approximately 12.4 (17.1) percent annually (Egger and Egger, 2005a). Now, the CEECs’ total exports to the EU are linked strongly to the fragmentation of production (de Benedictis and Tajoli, 2008). Following the political worries about the implications of trade integration with transition and developing economies resulting from possible damage to low skilled labour in developed countries, the empirical literature is concerned mainly with the labour markets in advanced countries. Research has focused particularly on the effects of imported intermediate inputs on the structure and/or level of demand for labour. At the same time, with a few exceptions3 little empirical research has been dedicated to the trade–labour markets interactions at the industry level, especially for the ‘New’ EU members. Within this framework, we focus on the EU case and try to answer a simple and straightforward question: does the cost of labour in partner countries affect the 2 Ten countries (Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovenia and Slovakia) joined the EU in 2004 while Bulgaria and Romania acceded in 2007. 3 See Section 2.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development The Demand for Skills and Labour Costs in Partner Countries 613 demand for domestic high and low skilled labour? Several features distinguish our work from existing contributions. The recent increased accessibility of detailed sector-level labour statistics for separate EU countries (also NMS) allows us to shed new light on the interaction mechanisms between the labour markets of ‘Old’ and ‘New’ members. Based on data availability we focus on EU-15 and NMS-5 economies (the five being the Czech Republic, Hungary, Poland, Slovenia and Slovakia), analysing the interactions between the domestic and partners’ sector labour market conditions. In this respect our work extends existing research on the labour market effects of European trade integration because it considers ‘Old’ and ‘New’ members together, allowing also for some heterogeneity across the subgroups. Furthermore, taking into account that a part of trade integration in the enlarged EU also concerns some service sectors, we extend the traditional focus of the empirical analysis beyond manufacturing and include business services among the sectors exposed to international competition. Moreover, an important novelty lies in the fact that the demand for labour is assumed to be affected not only by its own price and other domestic input prices but also by labour costs in partner countries: ceteris paribus, if an increase in foreign wages positively/negatively affects the domestic demand for labour, we interpret this as a hint of substitutability/complementarity between home and foreign labour inputs. The key idea is that deep trade integration can bring about fragmentation of production. If this is the case, the domestic demand for labour is also related to the cost of labour abroad. We consider employment by skill category and, contrary to the traditional manual/non-manual worker dichotomy, we define skills according to the workers’ education level. This allows for an interpretation of the results in terms of the relationship between the skill upgrading of a sector in one country and the skill upgrading in the same sector abroad, and therefore in terms of convergence/ divergence of skill structures of industries across Europe. The rest of the paper is organized as follows: in the next section, we review the theory and the empirical literature presenting alternative views on the labour market effects of increased trade integration. In Section 3 we describe the data and present descriptive statistics on trade intensification between EU-15 and NMS in the years 1995–2005, along with a description of changes in employment structure in these two groups of countries. In Section 4 we focus on revealing the degree of ‘Old’–’New’ members’ sectoral interdependency. To this aim an empirical model of sector labour demand for the high- and low-skilled is estimated. Subsequently, we present the elasticities of labour demand with respect to domestic and foreign wage conditions. 2. Literature review The main issue raised by our analysis is the interdependence of labour markets within the EU manifested through the impact of wage conditions in partner  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 614 Lo Turco and Parteka countries on the domestic demand for labour. Whether domestic and foreign labour input, in terms of labour in general as well as its different types (that is skilled, unskilled), are complements or substitutes is a question that theory has addressed in several ways, but only indirectly. On the other hand, in the burgeoning literature on the role of globalization (Feenstra, 1998; Hummels et al., 2001; Krugman, 2008) and the increasing inequality between skilled and unskilled workers, the final effect of trade in intermediates on the wages of the unskilled depends very much on the initial hypothesis of the model. Assuming a single final good, Feenstra and Hanson (1996, 1999, 2003) show that trade in the low skill intensive parts of the production process reduces the relative demand for and wages of unskilled workers in advanced countries. These unskilled workers are replaced by skilled workers in developing countries. In a specific factor model Kohler (2001) reinforces this view and shows that when FDI takes place together with outsourcing, labour always loses and when arm’s length transactions are the only possibility, the intensity of the fragment outsourced is relevant for the final impact on wages. On the contrary, Arndt (1997) shows that, within a framework with two final goods, the more labour intensive parts of production may be sent to labour abundant countries, but wages and employment may increase in the labour intensive sectors of advanced countries because of regained competitiveness. In this respect an increase in employment/wages abroad need not be related to a reduction at home. The new paradigm of the unbundling of tasks around the world (Grossman and Rossi-Hansberg, 2006, 2008) recalls this theoretical suggestion, finding that low skilled labour need not be a loser from globalization because of increased productivity following the unbundling. However, when the prices of final goods change, this conclusion can be challenged, especially if the unbundling does not concern the offshoring of a single type of task but instead a complete bundle of tasks involving different types of labour (Kohler, 2004, 2009). It is evident that the literature on the trade–wage nexus has focused mainly on the labour markets in advanced countries. The main concern has been whether low skilled workers in these countries (for example in the UK and USA) are negatively or positively affected by the fragmentation of the production process. Furthermore, it should be noted that so far the empirical evidence on the relationship between delocation of production phases and the composition of employment in European countries has been in the form of country-specific studies.4 Among the great number of works dealing with this issue in general, there is only a limited number of contributions implicitly concerning the recent experience of European integration. In general, in the literature on European integration issues, so far little emphasis has been put on a parallel assessment of the importance of trade intensification between EU-15 and NMS for employment structures in both groups of countries. 4 The only exception is the work by Hijzen and Swaim (2007) that explores the relationship between overall employment and offshoring for 17 high-income OECD countries.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development The Demand for Skills and Labour Costs in Partner Countries 615 In reality, Central Europe is already well integrated into EU-based networks (Kaminski and Ng, 2005); therefore, the relative demands for different types of labour (in terms of skill content) in ‘Old’ and ‘New’ member states are likely to be strongly interdependent. Egger and Egger (2003) argue that the increase in outsourcing to Central and Eastern Europe and the former Soviet Union has shifted manufacturing employment in Austria considerably in favour of high-skilled labour while a moderate increase in the skill premium took place.5 Another example is the study by Helg and Tajoli (2005) who analyse the experience of Italy and Germany and find that the increase of the skilled-to-unskilled labour ratio in Italy was caused by international fragmentation of production (IFP) while in Germany IFP appears to have no influence on changes in the relative demand for skilled labour. Geishecker (2006) finds that outsourcing to Central and Eastern Europe reduced the relative demand for unskilled workers in Germany. As far as labour market developments in NMS are concerned, within the EU-focused research there is only limited evidence of the impact of trade integration on changes in average wages in NMS from Central and Eastern Europe (Egger, 2006; Egger and Egger, 2002, 2005b; Egger and Pfaffermayer, 2004). However, since European data availability has been limited to recent years, there is very little evidence on the effects of trade on the evolution of wages of different skill categories of labour in NMS, especially using disaggregated data and not limited to country-specific studies. Egger and Stehrer (2003) analyse specific developments of the wage bill between non-manual and manual workers in the Czech Republic, Hungary, and Poland in 14 manufacturing sectors, finding a negative impact of intermediate goods trade on the skilled to unskilled wage bill ratio.6 A noticeable gap in the research in this field can now be challenged with the recent data we describe below. 3. Data description In order to address directly the relationship between the economic structures of EU-15 countries and NMS resulting from trade integration, we use disaggregated trade and industrial statistics for former EU-15 member countries (the ‘Old’ members group)7 and five out of ten NMS which joined the EU in 2004 (the Czech 5 Egger and Kreickemeier (2008) provide a theoretical basis for this result in a model where the interactions between relative factor endowments and the skill intensity of the domestic production is explored in a setting with imperfect labour markets. 6 Also Esposito and Stehrer (2009) focus on different skill categories; however they analyse the sector bias of the skill biased technological change (SBTC) hypothesis, which appears to play an important role in rising skill premiums in Hungary and Poland, but is not confirmed in the Czech Republic. 7 Depending on the year, the statistics for Belgium (BEL) and Luxembourg (LUX) are available for both countries separately or aggregated together; therefore we aggregated the data that were reported separately for BEL and LUX, treating them jointly throughout the analysis (BLX).  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 616 Lo Turco and Parteka Republic, Hungary, Poland, Slovenia and Slovakia, from now on called NMS-5 and included in the ‘New’ group).8 Unfortunately, detailed industrial statistics (especially those needed for calculation of skill specific wages) are not available yet for the remaining NMS. Growth and Productivity Accounts9 is our primary source for data on the labour markets in EU countries (employment levels in each sector, sector specific skill intensity, that is, the share of hours worked by workers with different skill levels10). These variables are needed for the calculation of medium wages (labour compensation of different categories of workers employed within each sector and their time of work), as well as for sector specific value added and intermediate input price indices. Statistics reported in national currencies were recalculated into euros using bilateral exchange rates from Eurostat. Trade statistics (the volume of bilateral exports and imports within the same sector between NMS and EU-15 members and the volume of total trade with all world partners) were obtained from the UN Comtrade Database (United Nations Statistics Division, 2008) through the WITS retrieval system11 which allowed us to obtain recalculated series of trade data following the industry list consistent with NACE division (a basic classification of the industrial statistics). We focus on manufacturing and business services and, to match trade and industrial statistics at the sectoral level, we reorganized the original data and aggregated all available statistics into 13 sectors.12 Complete labour market data for NMS 8 EU-20 is composed of EU-15 and NMS-5. Throughout the study the following abbreviations have been adopted: EU-15 (‘Old’) – Austria (AUT), Belgium and Luxembourg (BLX), Denmark (DNK), Spain (ESP), Finland (FIN), France (FRA), Germany (GER), Greece (GRC), Ireland (IRL), Italy (ITA), Netherlands (NLD), Portugal (PRT), Sweden (SWE), United Kingdom (UK). NMS-5 (‘New’) – the Czech Republic (CZE), Hungary (HUN), Poland (POL), Slovakia (SVK), Slovenia (SVN). 9 We used the data from the latest release EUKLEMS Database, 2008 (http://www.euklems.net). All the series in the EUKLEMS database were created based on statistics provided by National Statistical Institutes (NSIs), but a particular emphasis has been put on the harmonization of the basic data, ensuring cross sample comparability. Since data by labour types (according to the skill level) are not part of standard statistics reported by NSIs, EUKLEMS uses survey data as background sources (See Timmer et al., 2007 and EUKLEMS, 2007 for details). 10 Skills are defined on the basis of educational level. We use statistics originally classified according to the international ISCO classification into workers with high skills (HS – higher/tertiary education), medium skills (MS – secondary education) and low skills (LS – basic education) which are combined into two skill groups (h = HS, l = MS + LS) for this study. See Section 3.2. 11 World Integrated Trade Solutions (http://www.wits.worldbank.org). 12 Sectors A–M included into the analysis have been divided into two groups according to their skill typology from EU KLEMS. Low skill intensive sectors: A. Food, beverages and tobacco; B. Textiles, leather and footwear; C. Wood and products of wood and cork; D. Pulp, paper, printing and publishing; F. Rubber and plastics products; G. Other non-metallic mineral products; H. Basic metals and fabricated metal products; L. Manufacturing, nec; recycling. High skill intensive sectors: E. Chemicals and chemical products; I. Machinery, nec; J. Electrical and optical equipment; K. Transport equipment; M. Renting of machinery and equipment, other business services. We eliminate agriculture, mining, and public services from the analysis in order to focus on manufacturing and IT services.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development The Demand for Skills and Labour Costs in Partner Countries 617 are not available prior before 1995; therefore our analysis covers the period 1995– 2005 which is an important decade for the observation of the increased interdependence within an integrating Europe after the Europe Agreements. 3.1 Changes in trade relations between ‘Old’ and ‘New’ member states The progressing economic integration in Europe has intensified trade relations between Western Europe and countries which joined the EU in 2004 and 2007. Our main interest is to focus on the transmission mechanism via trade from partner countries; therefore we concentrate on import flows which were coming to ‘New’ partner countries from EU-15 and vice versa.13 Table 1 presents the first insight into the dynamics and significance of trade flows between NMS-5 and EU-15. The first set of columns presents the importance of EU-15 countries as a source of imports for NMS-5, the next three columns present analogous figures measuring the importance of imports from NMS-5 for EU-15 countries and the final two columns show the normalized trade balance (where positive value is a sign of being a net exporter). In 2005, depending on the sector, imports from EU-15 amounted to between 47.7 percent to 75.3 percent (the maximum was reached in the renting of machinery and equipment and other business services sector) of total imports reported by NMS-5. However, between 1995 and 2005 the share of import flows from EU-15 countries had diminished in most sectors as a percentage of total imports reported by NMS-5. The relative importance of imports from EU-15 in trade structures in ‘New’ countries increased considerably in such sectors as ‘Wood, products of wood and cork’, ‘Chemicals and chemical product’, ‘Basic metals and fabricated metal products’ and ‘Renting of machinery and equipment and other business services’. On the other hand, if we consider the importance of NMS as partners for EU-15, the shares of import flows from NMS-5 as a percentage of total imports reported by EU-15 are quite low (in 2005 up to 8.1 percent of total imports directed to EU-15 from the world), but since 1995 NMS-5 as importers have gained importance in the overall EU-15 structure of imports in all but three sectors (in one of these ‘Textiles, leather and footwear’, the drop in the share of imports from CEECs may reflect increasing competition from low labour cost Asian countries). Importantly, NMS-5 improved their position as a source of EU-15 imports in more advanced sectors such as ‘Electrical and optical equipment’, ‘Transport equipment’ and ‘Renting of machinery and equipment and other services’. In the case of these sectors the share of import flows from NMS-5 as a percentage of total EU-15 imports more than doubled between 1995 and 2005. Therefore, it is clear that, independently of the sector 13 For Western European partners, trade with NMS-5 represents the bulk of trade flows coming from/sent to NMS in general; therefore, we present statistics concerning countries from our panel (NMS-5) and analogous figures referring to all 12 NMS are available on request.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 618 Table 1. Trade patterns: share of import flows from EU-15 countries to NMS-12 and NMS-5 (and vice versa) and normalized trade balance (by sector) Imports from EU-15 to NMS-5 (% total NMS-5 imports) Notes: NMS-5: CZE, HUN, POL, SVK, SVN. EXPfromNMS5toEU15 IMPtoNMS5fromEU15  100% *Calculated as:EXP þIMP fromNMS5toEU15 toNMS5fromEU15 Source: Own elaboration with UN Comtrade data. NTB (NMS-5 vs. EU-15)* 1995 2005 D 1995 2005 D 1995 2005 57.3 72.7 50.6 72.0 62.6 75.6 69.2 59.2 79.9 63.7 76.2 67.9 66.2 56.2 58.9 52.8 70.6 68.6 75.0 62.1 64.5 72.7 47.7 71.1 52.4 75.3 )2.0 )18.9 4.3 )2.0 9.6 )0.8 )10.3 9.0 )9.0 )25.1 )6.7 )22.8 13.7 1.4 5.2 8.6 1.8 1.5 2.3 6.0 4.4 2.4 1.6 2.1 5.3 0.6 2.9 3.5 7.7 4.2 1.4 6.4 6.3 5.4 6.0 6.0 6.8 8.1 2.1 102.9 )32.2 )10.5 141.6 )3.3 177.3 5.2 22.2 150.8 281.8 218.0 51.7 228.9 )9.7 15.5 69.3 )33.5 )35.9 )29.8 14.8 20.0 )41.5 )23.9 )9.5 36.3 )37.3 )2.2 2.6 42.8 )15.3 )48.9 )16.8 2.7 )2.4 )7.7 18.9 19.1 52.2 )15.5 Lo Turco and Parteka  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development A. Food, beverages and tobacco B. Textiles, leather and footwear C. Wood and products of wood and cork D. Pulp, paper, printing and publishing E. Chemicals and chemical products F. Rubber and plastics products G. Other non-metallic mineral products H. Basic metals and metal products I. Machinery, nec J. Electrical and optical equipment K. Transport equipment L. Manufacturing, nec; recycling M. Renting of machinery and equipment, other services Imports from NMS-5 to EU-15 (% total EU-15 imports) The Demand for Skills and Labour Costs in Partner Countries 619 taken into consideration, trade with EU-15 is still much more important for NMS than trade with NMS is for the EU-15. A very large proportion of total imports to NMS come from EU-15 countries while the reverse is not true.14 Analysing the changes in sector specific normalized trade balances in trade flows between ‘New’ member states and the EU-15 (last two columns of Table 1), we can confirm that the NMS-5 as a group still tends to occupy the position of a net exporter in sectors requiring rather low skilled labour, such as ‘Textiles, leather and footwear’ and ‘Wood and products of wood and cork’, but are net importers of food products, pulp and paper, chemicals, rubber products, machinery, and business services. However, between 1995 and 2005, NMS-5 managed to pass from the position of net importer to net exporter in such advanced sectors as ‘Electrical and optical equipment’ and ‘Transport equipment’. This may be caused by increasing FDI activity in these fields (for example car factories located in CEECs). 3.2 Changes in employment patterns in ‘Old’ and ‘New’ members Having seen the major characteristics concerning trade patterns within the enlarged EU, we now turn toward the presentation of sectoral patterns of employment in EU-15 and NMS-5 countries (complete labour statistics for the remaining NMS are unavailable). Importantly, the information on sector specific skill content permits us to trace the dynamics of skill structures in ‘Old’ and ‘New’ member states. We use the information on the share of hours worked in single sectors by persons with high, medium, and low skills,15 where skills in EUKLEMS are defined according to the worker’s education level – workers with a tertiary education degree or more are classified as high skilled (HS), with a secondary school degree as medium skilled (MS), and with a primary school degree or less as low skilled (LS). The information contained within the EUKLEMS is consistent over time for each country, but it might differ across countries. Therefore, full comparability is assumed only across the bachelor degree educational level (EUKLEMS, 2007). In order to preserve comparability and provide easier legibility of the results, throughout the whole analysis we consider two types of workers – high skilled (h) and less/low skilled (l): h = HS are workers with tertiary education and l = MS + LS are workers with secondary education or less. 14 A similar pattern is confirmed in export shares – the predominant share of NMS exports is directed to the EU-15. For example, in 2005 up to 78.7 percent of total exports from the NMS-12 textiles, leather and footwear sector and 72.6 percent of NMS-5 transport equipment exports were sent to the EU-15 market. In comparison, in 2005 the great majority of EU-15 exports (approximately 90 percent) was still directed to non-NMS markets. Detailed data are available on request. 15 In general, EUKLEMS distinguishes between employees and persons engaged – the difference between the two are the self-employed and family workers; therefore the discrepancy may be considerable in sectors with a large proportion of self-employed such as agriculture or retailing. However, specific labour statistics by skill groups are available only for persons engaged.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 620 Lo Turco and Parteka In Table 2 we present sector specific high skill content (the share of hours worked by highly skilled persons engaged – share(Lh) – in the EU-15 and NMS-5 in 1995 and 2005. Additionally, we present percentage changes between 1995 and 2005 in overall employment in single sectors (measured by the change in total hours worked, DL) and in the number of hours worked by highly skilled workers, DLh. How important is highly skilled labour in the employment structures in ‘Old; and ‘New’ members? On average by sector, in NMS-5 in 1995 10 percent of total hours were worked by highly skilled workers with a tertiary education; in EU-15 the figure was 11.6 percent; in 2005 the corresponding shares had risen to 16.5 percent and 18.2 percent, a sign of movement toward a larger share of highly educated workers (an overall skill upgrading of employment structures) in both groups. Between 1995 and 2005 share(Lh) increased in all sectors both in the EU-15 and in NMS-5, indicating a more intensive use of skills in the tradable economy. However, skill patterns are very different across sectors. Not surprisingly, the biggest proportion of hours worked (approximately one-third in EU-15 and one-fourth in NMS-5) by employees with a tertiary education level is in the services sector. A much lower proportion of time worked by people with the highest educational levels is typical for more traditional, labour intensive activities. Interestingly, both in 1995 and 2005, the NMS-5, compared to the EU-15, employed more higher-skilled workers in almost all sectors but not in those typically defined as ‘high-skill’, such as ‘Electrical and optical equipment’ and ‘Transport equipment’, indicating the within-sector differences in tasks performed in ‘Old’ and ‘New’ member states. As far as changes in total hours worked are considered (DL), between 1995 and 2005 in EU-15 the business services sector expanded noticeably (by more than 50 percent), but all other sectors, except ‘Basic metals and fabricated metal products’, registered a drop in the number of hours worked (it was most pronounced, by 31 percent, in ‘Textiles, leather and footwear’). In NMS-5 the picture is more complicated – among sectors that contracted we may find both traditional ones, such as ‘Food, beverages and tobacco’ or ‘Textiles, leather and footwear’ (hours worked fell by more than 40 percent), but also the more advanced chemical and machinery manufacturing sectors. Apart from service sectors, where the number of hours worked in 2005 was 66 percent higher than in 1995, employment in NMS-5 grew most in ‘Rubber and plastics products’ and in ‘Electrical and optical equipment’ sectors. In brief, while ‘Old’ members were moving labour from manufacturing to services, ‘New’ members were still experiencing labour force movements within manufacturing. The general tendency of skill upgrading is confirmed when we look at changes in the number (not share) of hours worked in each sector: Share(Lh) grew in almost all cases, even in sectors where the overall number of hours worked decreased. It means that employment structures of both ‘Old’ and ‘New’ EU countries employ highly educated labour more and more intensively.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development ‘Old’ (EU-15) Share(Lh) ‘New’ (NMS-5) Share(Lh) DL DLh DL DLh 1995 2005 1995–2005 1995–2005 1995 2005 1995–2005 1995–2005 A. Food, beverages and tobacco B. Textiles, leather and footwear C. Wood and products of wood and cork D. Pulp, paper, printing and publishing E. Chemicals and chemical products F. Rubber and plastics products G. Other non-metallic mineral products H. Basic metals and fabricated metal products I. Machinery, nec J. Electrical and optical equipment K. Transport equipment L. Manufacturing, nec; recycling M. Renting of machinery and equipment, other business services 2.7 3.0 5.6 7.0 8.2 6.4 6.0 5.9 6.2 10.1 9.4 4.4 23.6 6.0 4.4 8.8 10.6 12.2 9.1 9.0 8.5 8.7 13.2 12.9 6.7 30.2 )5.4 )31.4 )10.0 )11.1 )10.2 )1.6 )9.9 1.4 )2.2 )12.6 )1.9 )9.4 54.3 50.2 4.5 40.6 33.4 33.4 39.2 32.8 40.3 39.9 19.3 44.0 54.5 99.0 4.2 4.2 8.7 8.8 9.2 8.9 8.7 8.5 7.8 7.9 8.0 4.1 30.3 6.3 6.2 12.8 13.2 13.4 13.0 12.8 12.4 10.2 9.9 10.3 6.3 37.7 )15.8 )43.9 7.1 2.6 )22.2 41.7 )18.7 )5.1 )28.3 33.2 15.8 10.6 66.6 26.6 )20.2 53.4 51.3 11.3 103.4 16.5 34.0 )7.9 62.9 45.0 66.8 113.2 Average 11.6 18.2 18.9 63.3 10.1 16.5 13.2 52.7 The Demand for Skills and Labour Costs in Partner Countries  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development Table 2. Employment patterns: high skill content of employment structure and changes in hours worked, by sector, in % Notes: Weighted averages (by sector size) across countries within each group. NMS-5: CZE, HUN, POL, SVK, SVN. Source: Own elaboration with EUKLEMS data. 621 622 Lo Turco and Parteka 4. Modelling the domestic labour demand response to foreign labour costs 4.1 Empirical model In order to assess the degree of complementarity/substitutability between domestic and foreign workers in the EU we focus on the response of domestic employment in a sector with respect to overall average labour costs in the same sector in partner countries. We take into account the heterogeneity of tasks performed and disentangle different responses of domestic demand for skilled and unskilled labour. In order to allow for a flexible technology, we adopt a non-homothetic translog cost function. Deriving the log of the cost function C with respect to the log of input prices Px, we obtain cost share equations of the following form (Berndt, 1991, p. 470): @ ln C Px @C Px  Xx  ; ¼ ¼ Sx ¼ @ ln Px C @Px C ð1Þ P where Sx represents the cost share for each input x and x PxXx = C. Then, assuming a production technology with inputs such as: intermediate inputs16 (mat) and labour (L) – high and low skilled – it is possible to derive the conditional demand for high and low skilled labour as follows (i denotes countries, j = sectors and t = time, but apart from the error term we omit i and j subscripts to provide easier legibility of the formulas):17 ~ht ¼ ah S ~ht1 þ S l X  bhk  k¼h ~lt1 þ ~lt ¼ al S S l X k¼h wkt  pmat  blk  wkt  pmat  þ chy  yt þ Old X    dzhL  wpzLt þ kt þ eijt ; ð2aÞ z¼New  þ cly  yt þ Old X    dzlL  wpzLt þ kt þ eijt ; ð2bÞ z¼New where Sh and Sl respectively measure the cost shares of high and low skilled labour;  represents the deviation from the individual time mean to allow for industrycountry specific unobservable fixed effects and for any time invariant source of endogeneity. The dependent variable lag is included to control for the persistence of the labour cost shares. Lack of data on the capital stock for the NMS-5 represents a limitation for which we control, in a first step, by the within transformation of the 16 Intermediates include energy, material and service inputs. The original cost function then describes the total variable cost. 17 The exact detailed derivation of the final equations is based on Berndt (1991, pp. 469–476).  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 623 The Demand for Skills and Labour Costs in Partner Countries variables. The log of domestic hourly wages of skilled and unskilled labour is represented by wk (for each k = h, l); y is the log of real output and pmat represents the log of unit price of intermediate inputs. The latter appears in the denominator because the equation for the conditional demand of materials needs to be dropped from the system to avoid linear dependency among the left hand side variables and the singularity of the system variance–covariance matrix. However, we estimate a model (2) with the Maximum Likelihood estimator which allows for the invariance of the parameter estimates to the choice of the equation to delete.18 Finally, indexing with q the R partners in the EU and ranking the ‘New’ partners (NMS-5) from 1 to p and the ‘Old’ ones (EU-15) from p + 1 to R,19 wpzL , for z = Old, New, represents the log of the average labour cost in partner countries (WPL), measured as (time subscripts omitted): WPNew Lij WPOld Lij ¼ ¼ Pp importiqj  wageqj ; Pp q¼1 importiqj q¼1 PR q¼pþ1 importiqj  wageqj PR q¼pþ1 importiqj : ð3aÞ ð3bÞ For each country i and for partners classified as z ¼ Old, New; WPzLij is obtained as the weighted average of partners’ wage, wageqj, in the same sector j, with weights equal to country i’s imports from partner q in the same sector j. Such a 18 Actually, the presence of the lagged dependent variable in the model above would call for a GMM estimation technique. However, GMM estimators of dynamic panel data display poor finite sample properties and the bias of GMM becomes quite severe when the number of instruments is large relative to the number of groups in the panel. On the other hand, although the maximum likelihood estimator is generally consistent, the ML principle has not been any help due to the incidental parameter problem. Furthermore, the conditional ML for the AR(1) panel data model with FE equals the LSDV (Least Squares Dummy Variable) estimator, which is inconsistent under traditional large N and fixed T asymptotics (Nickell, 1981) but it has been shown that the bias is reduced when T grows large (Alvarez and Arellano, 2003; Hahn and Kuersteiner, 2002). In our case the use of GMM is not advisable since the NMS-5 group is very small and we also mean to explore the response of the demand for labour to foreign wages in the sub-samples of high and low skill intensive sectors, for which the cross-section dimension shrinks further for the NMS-5 and EU-15. Then, although our time span covers 11 years we stick to the Within ML to save the property of invariance of the results with respect to the equation deleted from the system. To determine whether the endogeneity of the transformed lagged dependent variable affects the consistency of the estimates of the remaining coefficients, for the overall sub-samples of NMS-5 and ‘Old’ members we compare the ML estimates of the system coefficients with their LSDVC (corrected LSDV) ones (Kiviet, 1995) which, under strict exogeneity of the remaining regressors, perform substantially well compared to GMM and IV in general in panel structures like ours (Bruno, 2005a,b). 19 The definition of partner countries adopted here refers to partners in the EU in our restricted sample composed of 20 countries, therefore every ’New’ member state has four ’New’ partners and 15 ‘Old’ partners, while every ‘Old’ member state has five ‘New’ partners and 14 ‘Old’ partners.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 624 Lo Turco and Parteka weighting scheme allows us to consider trade-based interactions between labour markets at home and abroad – foreign wage conditions in partner q can matter as long as trade is present and we assign major importance to the evolution of wage conditions in partner countries from which imports are particularly intense. Similar approaches in the empirical literature have concerned mainly the elasticity of substitution of labour between domestic parents and foreign affiliates, estimated with both firm- and sector-level US data (Brainard and Riker, 1997; Lawrence, 1994; Slaughter, 2000). In our case, the use of sector level data allows for a more general approach not limited to the multinationals’ activity. Following Feenstra (2004)20, the foreign labour cost is modelled as exogenous technology shocks affecting the sector cost function and the demand for skills, possibly in a non-neutral way so it is treated as any other structural variable. Although, due to data limits we cannot consider wages from trading partners outside the EU, any other source of foreign competition is meant to be captured by the within transformation of the variables and by the inclusion of time fixed effects. Allowing for common time shocks across the countries we control for the influence of globalization on our countries’ labour markets. Finally, the own and cross price elasticities are calculated as follows: enn ¼ enm ¼ bnn þ S2n  Sn ; n ¼ h; l; Sn bnm þ Sn Sm ; n; m ¼ h; l and Sn ð4aÞ n 6¼ m; ð4bÞ so that, for example, ehl denotes the elasticity of the demand for skilled labour with respect to the wage of the low skilled, and so on. The elasticities of demand for high and low skilled labour with respect to real output (y) and foreign wages (wpzL ) are obtained as: eny ¼ eznm ¼ cny Sn ; n ¼ h; l; dznm ; n ¼ h; l and Sn z ¼ New, Old: ð5aÞ ð5bÞ 4.2 Results Summary statistics of model 2 estimates are shown in Table A1 in the Appendix. Table 3 shows the elasticities of skilled and unskilled labour (Lh and Ll) with 20 Chapter 4.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 625 The Demand for Skills and Labour Costs in Partner Countries Table 3. Elasticities of high and low skilled labour with respect to output and wages ‘New’(NMS-5) wh wl y wpOld L wpNew L ‘Old’ (EU-15) Lh Ll Lh Ll )0.15 [0.15] )0.68*** [0.15] )0.11 [0.1] )0.28*** [0.05] )0.12*** [0.04] )0.19*** [0.04] )0.21*** [0.05] )0.07 [0.06] )0.11*** [0.03] )0.22*** [0.02] )0.35*** [0.04] )0.28*** [0.04] 0.04 [0.03] )0.05*** [0.02] )0.13*** [0.02] )0.06*** [0.01] )0.41*** [0.02] 0.04*** [0.02] )0.04*** [0.01] )0.13*** [0.01] Source: Own calculations. Note: *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. Standard errors are given in brackets. respect to domestic wages (wh and wl) and foreign labour costs (wpOld and wpNew ), L L calculated according to the formulas in Equation (4). The table should be read as a matrix with columns denoting the type of labour demand and rows referring to the price of the factor of interest and output; the cell on the intersection between the two contains the estimated elasticity. The first two columns contain elasticities calculated over the sub-sample of domestic labour in ‘New’ members and the remaining ones refer to the estimated elasticities for labour in ‘Old’ members. All the results include common time effects. The coefficient estimates for the elasticities in Table 3 are shown in Table A2 in the appendix, along with the Breusch–Pagan test for independence between the two equations and the test for strict exogeneity of the regressors for the LSDVC (corrected Least Square Dummy Variable) estimator.21 We failed to reject the null of strict exogeneity for the right hand side variables; therefore, once controlled for time invariant unobservables, no other source of endogeneity is at work for our right hand variables. Furthermore, as shown by comparing the coefficient estimates obtained with WG-ML (Within Group Maximum Likelihood) and LSDVC, the endogeneity of the lagged dependent variable does not affect the estimate of the remaining coefficients, so we continue with the system’s Within ML estimator. In general, the regularity conditions implied by the theory are respected since domestic price elasticities (Lh with respect to wh and Ll with respect to wl) are nega21 The test is performed according to Wooldridge (2002, p. 285).  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 626 Lo Turco and Parteka tive. They also are negative for material inputs as shown in Table A3 in the Appendix. Furthermore, the average prediction for the share of skilled and unskilled labour is positive as shown in Table A4 in the Appendix, along with the test for the symmetry cross-equation restrictions. The negative domestic cross price elasticities of Lh with respect to wl and Ll with respect to wh in Table 3 reveal a certain degree of complementarity between domestic low and high skilled labour. Turning to the elasticities of domestic labour with respect to foreign wages, the general result we obtain is the fact that independently of the skill level there are complementarity relations between domestic and foreign labour force in our sample of EU countries. NMS-5 domestic labour with tertiary education completed (Lh in Table 3, first column) is a complement to the labour force in partner countries, both those belonging to the EU-15 and NMS-5 group (negative and significant New ). Similarly, the second column reveals that elasticities between Lh and wpOld L wpL the foreign labour force from both ‘Old’ and ‘New’ partner countries is complementary to domestic low skilled labour from NMS-5, Ll. When looking at the third and fourth columns, referring to domestic high and low skill domestic labour in EU-15 countries, it turns out that the internal demand both for more and less educated workers in EU-15 countries diminishes when wages rise in partner countries. Therefore, in EU-20 an increase in the average wage in partner countries is related to a decrease in the average labour share ‘at home’. This is true in general, but the highest estimated elasticity is for the response of demand in the NMS-5 for high skilled labour with respect to average wage changes in ‘Old’ members – 1 percent rise in wages in EU-15 is linked to a 0.28 percent drop in domestic demand for high skill workers in NMS-5. Also the response of the domestic demand for low skilled workers in NMS-5 countries to changes in wage conditions in other NMS-5 economies is relatively strong (elasticity equal to )0.22). We conduct another check estimating the empirical model (2) on subgroups of sectors; we distinguish between low and high skill-intensive sectors according to the taxonomy adopted within the EUKLEMS database.22 In Table 4 we focus on the elasticities with respect to foreign wages for ease of presentation; the same elasticities for material inputs are shown in Table A5 in the appendix. The first two columns refer to high skill-intensive sectors and the second pair to the low skill-intensive sectors. Two panels contain the results obtained within the subgroups of domestic labour from NMS-5 and ‘Old’ members. The results give us the whole set of possible interactions across sectors and countries, but given our interest in the interdependency between ‘Old’ and ‘New’ members, we first focus on cross elasticities between workers from these two groups of countries. In most cases cross-border labour complementarity is confirmed (negative elasticities), but the important difference with the previous table mainly concerns low skill-intensive (more traditional) sectors. In such sectors wage increases in partners from ‘Old’ group is associated with the rise in the demand 22 See footnote 12 for the list of sectors classified as high skill-intensive.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 627 The Demand for Skills and Labour Costs in Partner Countries Table 4. Elasticities of high and low skilled labour with respect to wages by sectors typology High skill intensive sectors ‘New’ (NMS-5) wh wl y wpOld L wpNew L ‘Old’ (EU-15) wh wl y wpOld L wpNew L Low skill intensive sectors Lh Ll Lh Ll )0.647*** [0.193] )0.26 [0.190] )0.05 [0.044] )0.458*** [0.116] )0.409*** [0.075] )0.109 [0.079] )0.305*** [0.087] )0.133*** [0.030] )0.206*** [0.070] )0.267*** [0.046] )0.011 [0.199] )0.718*** [0.197] )0.208*** [0.054] 0.422** [0.175] )0.174** [0.071] )0.138*** [0.038] )0.240*** [0.057] )0.278*** [0.031] 0.043 [0.100] )0.042 [0.040] )0.219*** [0.049] )0.386*** [0.045] )0.148*** [0.028] )0.004 [0.043] )0.069*** [0.020] )0.135*** [0.016] )0.322*** [0.029] )0.189*** [0.018] 0.028 [0.025] )0.078*** [0.012] )0.597*** [0.046] 0.103** [0.052] )0.155*** [0.030] 0.135** [0.055] )0.003 [0.024] 0.016** [0.008] )0.479*** [0.019] )0.109*** [0.013] 0.049** [0.022] )0.012 [0.010] Source: Own calculations. Note: *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. Standard errors are given in brackets. for highly skilled labour in NMS-5, thus there may be a sign of possible competition between high skilled workers in ‘New’ EU member states and workers from EU-15, but only within less skill-intensive sectors. On the other hand, independent of the typology of sectors, we did not find any signs that NMS-5 workers could pose a threat to the high or low skilled domestic EU-15 labour force (workers employed in EU-15 are not significantly affected by NMS-5 wage conditions). EU-15 workers employed in less skill-intensive sectors should not especially fear competition with NMS-5, since they could also be substituted by workers from other EU-15 countries.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 628 Lo Turco and Parteka Table 5. Overall growth (1995–2005) ‘New’ (NMS-5) High skill intensive sectors Low skill intensive sectors ‘Old’ (EU-15) High skill intensive sectors Low skill intensive sectors Sh Sl wpOld L wpNew L 0.03 0.22 )0.25 )0.10 0.31 0.20 0.63 0.59 0.31 0.48 )0.16 )0.10 0.36 0.26 0.58 0.59 Source: Own calculations. Trying to quantify the importance of these elasticities and determine the strength of cross-border labour interdependency, we report in Table 5 the overall growth of cost shares and foreign wages between 1995 and 2005. Wages in partner countries grew in all cases, but more in partners belonging to the ‘New’ group. Both in NMS-5 and in EU-15, shares of high skill labour grew (especially in less skill-intensive sectors) while shares of low skilled labour declined (especially in more advanced sectors). We can combine these growth rates with the estimated elasticities from Table 4. For example, in the case of low skill-intensive sectors (typically perceived as more vulnerable to foreign competition) the increase in labour cost in the ‘Old’ member partners could have accounted for approximately 8 percentage points (=0.20 · 0.422) of the overall increase in the demand for high skilled employees in NMS-5 in the same sectors. 4.3 Robustness checks In order to check the robustness of the results, we also tested a final specification by including in model (2) average wage in the rest of the economy, wo, and a measure of capital services, cap (due to data availability in this case the number of NMS drops to three and the group of ‘Old’ members does not include Greece). The main findings from the previous tables are confirmed.23 Then, a further check was accomplished. To determine whether our results are driven by the measure of average labour cost adopted (3), we first substituted it with the average labour cost for the high and the low-skilled in partner countries (wpzh and wpzl respectively with z = Old, New). Next, we examined whether the significance of our elasticities is driven by the use of imports as a weight of the importance of partners’ labour markets in the domestic economy. To do so, we calculated the unweighted average labour cost and another measure of weighted labour cost where, in the spirit of Frankel and Romer (1999), we substituted imports 23 Detailed results available from the authors upon request.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development The Demand for Skills and Labour Costs in Partner Countries 629 from the partners with their geographical prediction.24 Then, we checked whether our results hold in general (in other word when considering all the countries in the larger EU as potential competitors regardless of their trade relations with the country under analysis), or when measuring the extent of potential competition by their actual distance from their own market. Consequently, we compared the benchmark results with those obtained when we used no weighting scheme or with imports predicted by distance as a weight for the average labour cost in the formulas in Equation (3). Our results in general remain robust, aside from the response of labour in ‘Old’ members which is not significant when the weight is removed or substituted. Furthermore, when average partner wages wpOld are substituted for the average wage of the high and wpNew L L and low skilled an interesting result appears: the demand for the highly skilled workers employed in NMS-5 low skill intensive sectors is stimulated particularly by the increase in the wages of the low skilled workers in ‘Old’ member states. In fact, the elasticity of NMS-5 low skilled labour with respect to the average wage of the ‘Old’ members’ low skilled labour (wpOld l ) is 0.42 while the elasticity with respect to the wage of the EU-15 high skilled labour (wpOld h ) is 0.28. This could mean that the high skilled labour in NMS-5 is a better substitute for low-skilled labour in the EU-15 than for highly skilled labour in the EU-15 in ‘Old’ members’ low skill-intensive sectors.25 5. Summary of the findings and conclusions This paper has focused on the interdependence between labour markets in ‘Old’ and ‘New’ member states of the integrating European Union. The evidence of the recent boost in trade between the two groups of European countries, together with the overall skill upgrading registered in the employment structures in ‘Old’ and ‘New’ member states, suggests that the two phenomena may be related through the thoroughly documented process of production fragmentation. With respect to the previous limited evidence on the trade–labour nexus in the EU, the increased accessibility of detailed sector-level data on labour markets in separate EU member countries (EU-15 and selected NMS, namely, Poland, Hungary, the Czech Republic, Slovakia and Slovenia) allowed us to shed new light on the interactions between labour markets in ‘Old’ and ‘New’ member states. We add to the existing empirical literature considering ‘Old’ and ‘New’ EU members at the same time in a homogeneous empirical setting, including the business service sector in the analysis and 24 We ran a regression of bilateral imports on bilateral distances (from CEPII) and a full set of pair, reporter, and partner fixed effects, along with time dummies and their interaction with reporter and partner dummies and with the distance. Ceteris paribus, the interaction of time dummies with the distance is meant to represent the ease of improved communication and transportation infrastructure through time. 25 Detailed results available from the authors upon request.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 630 Lo Turco and Parteka covering the 1995–2005 time span, which is convenient for observing the effects of EU progressive enlargement. Finally, our main contribution concerns the detection of the effect of the average evolution of labour costs in partner countries on the domestic demand for skills. From the estimates of the conditional labour demand for the high and low skilled, augmented by the inclusion of a measure of average labour cost in partner countries, our work leads to plausible results. Unsurprisingly the labour markets in ‘Old’ EU members are much less affected by foreign labour costs than the demand for skills in ‘New’ EU members. For this reason, our main results concern the latter group and imply, in general, that labour in NMS-5 is complementary with respect to labour from partners, both from the ‘Old’ group (EU-15) and from other NMS-5. This finding confirms Kaminski and Ng’s (2005) evidence of the NMS-5 as locations interconnected by different phases of a complex EU-based production chain. However, when looking for possible heterogeneous effects across different sectors, the demand for high skilled labour employed within low skill-intensive sectors in NMS-5 is boosted by an increase in average labour costs of ‘Old’ member partners, especially average low skilled labour costs. Such a result suggests that the substitutability of NMS-5 high skilled labour for (low skilled) labour in the EU-15 countries seems to characterize the period under analysis and contributes to interpreting the skill upgrading of manufacturing in NMS-5 in line with trade based explanations (Feenstra and Hanson, 1996). In conclusion, our analysis shows that EU enlargement has fostered the skill upgrading of the productive structure in less skill intensive activities in NMS-5 countries possibly leading to convergence with respect to manufacturing structures in the ‘Old’ EU members. On the other hand, our study does not confirm the fears that the enlargement process could cause severe adjustments in the labour markets in ‘Old’ EU members considered as a homogeneous group. References Alvarez, J. and Arellano, M. (2003). ‘The time series and cross-section asymptotics of dynamic panel data estimators’, Econometrica, 71(4), pp. 1121–1159. Arndt, S. W. (1997). ‘Globalization and the open economy’, North American Journal of Economics and Finance, 8, pp. 71–79. Baldone, S., Sdogati, F. and Tajoli, L. (2001). ‘Patterns and determinants of international fragmentation of production: Evidence from outward processing trade between EU and Central Eastern European Countries’, Weltwirtschaftliches Archiv, 1371, pp. 80–104. de Benedictis, L. and Tajoli, L. (2008). ‘Similarity in trade structures, integration and catching-up’, Economics of Transition, 162, pp. 165–182. Berndt, E. R. (1991). The Practice of Econometrics, Reading, MA: Addison-Wesley. Brainard, S. and Riker, D. (1997). ‘Are U.S. multinationals exporting U.S. jobs?’ NBER Working Paper No. 5958, Cambridge, MA: The National Bureau of Economic Research. Available at: http://www.nber.org/papers/w5958.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development The Demand for Skills and Labour Costs in Partner Countries 631 Bruno, G. F. S. (2005a). ‘Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models’, Economics Letters, 87, pp. 361–366. Bruno, G. F. S. (2005b). ‘Estimation and inference in dynamic unbalanced panel-data models with a small number of individuals’, The Stata Journal, 54, pp. 473–500. CEPII (Centre d’Etudes Prospectives et d’Informations Internationales) (2006). ‘Bilateral distance data’ (dist_cepii), Available at: http://www.cepii.fr. Last accessed 20 March 2008. Egger, P. (2006). ‘Intermediate goods trade and international wage convergence in Central Europe’, Empirica, 33, pp. 181–192. Egger, H. and Egger, P. (2002). ‘How international outsourcing drives up Eastern wages’, Weltwirtschaftliches Archiv, 1381, pp. 83–96. Egger, H. and Egger, P. (2003). ‘Outsourcing and skill specific employment in a small economy: Austria and the fall of the iron curtain’, Oxford Economic Papers, 55, pp. 625–643. Egger, H. and Egger, P. (2005a). ‘The determinants of EU processing trade’, The World Economy, 282, pp. 147–168. Egger, H. and Egger, P. (2005b). ‘Labour market effects of outsourcing under industrial interdependence’, International Review of Economics and Finance, 14, pp. 349–363. Egger, H. and Kreickemeier, U. (2008). ‘International fragmentation: Boon or bane for domestic employment?’ European Economic Review, 52, pp. 116–132. Egger, P. and Pfaffermayer, M. (2004). ‘Two dimensions of convergence: National and international wage adjustments effects of cross-border outsourcing in Europe’, Review of International Economics, 125, pp. 833–843. Egger, P. and Stehrer, R. (2003). ‘International outsourcing and the skill-specific wage bill in Eastern Europe’, The World Economy, 261, pp. 61–72. Esposito, P. and Stehrer, R. (2009). ‘The sector bias of skill-biased technical change and the rising skill premium in transition economies’, Empirica, 36(3), pp. 351–364. EUKLEMS (2007). The EUKLEMS Growth and Productivity Accounts: Version 1.0; Part I: Methodology, prepared by Timmer M., van Moergastel T., Stuivenwold E., Ypma G., Mahony M., Kangasniemi M., on behalf of the EU KLEMS consortium, Available at: http:// www.euklems.net/data/EUKLEMS_Growth_and_Productivity_Accounts_Part_I_Method ology.pdf. Last accessed 15 February 2008. EUKLEMS Database (2008). The EUKLEMS Growth and Productivity Accounts, University of Groningen and University of Birmingham, Available at: http://www.euklems.net. Last accessed 10 October 2008. Feenstra, R. C. (1998). ‘Integration of trade and disintegration of production in the global economy’, Journal of Economic Perspectives, 124, pp. 31–50. Feenstra, R. C. (2004) Advanced International Trade. Theory and Evidence, Princeton, NJ: Princeton University Press. Feenstra, R. C. and Hanson, G. H. (1996) ‘Foreign investment, outsourcing and relative wages’, in Feenstra, R. C., Grossman, G. and Irwin, D. (eds), The Political Economy of Trade Policy: Papers in Honour of Jagdish Bhagwati, Cambridge, MA: MIT Press. Feenstra, R. C. and Hanson, G. H. (1999). ‘The impact of outsourcing and high-technology capital on wages: Estimates for the United States, 1979–1990’, Quarterly Journal of Economics, 114, pp. 907–940. Feenstra, R. C. and Hanson, G. H. (2003) ‘Global production sharing and rising inequality: A survey of trade and wages’, in Kwan Choi, E. and Harrigan, J. (eds), Handbook of International Trade, London: Basil Blackwell.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 632 Lo Turco and Parteka Frankel, J. A. and Romer, D. (1999). ‘Does trade cause growth?’ American Economic Review, 893, pp. 379–399. Geishecker, I. (2006). ‘Does outsourcing to Central and Eastern Europe really threaten manual workers’jobs in Germany?’ The World Economy, 295, pp. 559–583. Grossman, G. M. and Rossi-Hansberg, E. (2006). ‘The rise of offshoring: It’s not wine for cloth anymore’, Proceedings, Federal Reserve Bank of Kansas City, 2006, pp. 59–102. Grossman, G. M. and Rossi-Hansberg, E. (2008). ‘Trading tasks: A simple theory of offshoring’, American Economic Review, 985, pp. 1978–1997. Hahn, J. and Kuersteiner, G. (2002). ‘Asymptotically unbiased inference for a dynamic panel model with fixed effects when both n and T are large’, Econometrica, 70, pp. 1639– 1657. Helg, R. and Tajoli, L. (2005). ‘Patterns of international fragmentation of production and the relative demand for labor’, North American Journal of Economics and Finance, 16, pp. 233– 254. Hijzen, A. and Swaim, P. (2007). ‘Does offshoring reduce industry employment?’ National Institute Economic Review, 201(1), pp. 86–93. Hummels, D., Ishii, J. and Yi, K. (2001). ‘The nature and growth of vertical specialization in world trade’, Journal of International Economics, 54(1), pp. 75–96. Kaminski, B. and Ng, F. (2005). ‘Production disintegration and integration of Central Europe into global markets’, International Journal of Economics and Finance, 14, pp. 377–390. Kiviet, J. F. (1995). ‘On bias, inconsistency, and efficiency of various estimators in dynamic panel data models’, Journal of Econometrics, 68, pp. 53–78. Kohler, W. (2001). ‘A specific factors view on outsourcing’, North American Journal of Economics and Finance, 12, pp. 31–53. Kohler, W. (2004). ‘International outsourcing and the factor prices with multistage production’, The Economic Journal, 114, pp. 166–185. Kohler, W. (2009). ‘Offshoring: Why do stories differ?’ European Community Studies Association of Austria Publication Series, 12(1), pp. 17–49. Krugman, P. (2008). ‘Trade and wages, reconsidered’, Brookings Papers on Economic Activity, 1, pp. 103–154. Lawrence, R. Z. (1994). ‘Trade, multinationals and labor’, NBER Working Paper No. 4836, Cambridge, MA: The National Bureau of Economic Research, Available at: http:// www.nber.org/papers/w4836.pdf. Nickell, S. (1981). ‘Biases in dynamic models with fixed effects’, Econometrica, 49, pp. 1417– 1426. Slaughter, M. J. (2000). ‘Production transfer within multinational enterprises and American wages’, Journal of International Economics, 50(2), pp. 449–472. Timmer, M. P., O’Mahony, M. and van Ark, B. (2007). ‘Growth and productivity accounts from EUKLEMS: An overview’, National Institute Economic Review, 200, pp. 64–78. United Nations Statistics Division (2008). Commodity Trade Statistics Database UNComtrade, available through World Integrated Trade Solutions (http://www.wits.worldbank.org/ witsweb). Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data, Cambridge, MA: MIT Press.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 633 The Demand for Skills and Labour Costs in Partner Countries Appendix Table A1. Summary statistics model 2 Variable Mean SD Min Max Observations 0.05 0.06 0.06 0.01 0.00 0.00 )0.04 0.51 0.48 0.17 N = 2,717 n = 247 T = 11 )0.78 1.29 1.29 0.12 )3.03 )2.79 )1.38 3.34 2.79 )0.17 N = 2,717 n = 247 T = 11 0.22 0.07 0.07 0.02 0.04 0.07 0.11 0.55 0.47 0.30 N = 2,717 n = 247 T = 11 )1.42 1.27 1.27 0.12 )3.68 )3.51 )2.00 2.45 1.87 )0.82 N = 2,717 n = 247 T = 11 4.81 0.29 0.23 0.18 3.91 4.18 3.16 7.20 6.26 5.75 N = 2717 n = 247 T = 11 )1.05 0.27 0.25 0.11 )1.73 )1.63 )1.52 )0.12 )0.45 )0.52 N = 2,717 n = 247 T = 11 )1.64 0.27 0.25 0.10 )2.31 )2.15 )2.08 )0.62 )1.10 )1.06 N = 2,717 n = 247 T = 11 )3.06 0.30 0.21 0.22 )4.60 )3.64 )4.22 )2.10 )2.70 )2.32 N = 2,717 n = 247 T = 11 Sh Overall Between Within wh Overall Between Within Sl Overall Between Within wl Overall Between Within y Overall Between Within wpOld h Overall Between Within wpOld l Overall Between Within wpNew h Overall Between Within  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 634 Lo Turco and Parteka Table A1. (cont) Summary statistics model 2 Variable Mean SD Min Max Observations )3.81 0.32 0.23 0.23 )5.34 )4.42 )4.95 )2.68 )3.32 )2.98 N = 2,717 n = 247 T = 11 )0.83 1.28 1.28 0.13 )2.54 )2.40 )1.36 2.75 2.47 )0.35 N = 2,717 n = 247 T = 11 )1.46 1.26 1.26 0.14 )3.45 )3.30 )1.98 1.79 1.49 )0.96 N = 2,717 n = 247 T = 11 wpNew l Overall Between Within who Overall Between Within wlo Overall Between Within Source: Own calculations. Table A2. Coefficients estimates for elasticities in Table 3 ‘New’ (NMS-5) ah bhh bhl chy dOld dNew al blh ‘Old’ (EU-15) WG-ML LSDVC WG-ML LSDVC 0.53 0.00 0.04 0.00 )0.04 0.00 )0.01 0.00 )0.01 0.00 )0.01 0.00 0.37 0.00 )0.04 0.00 0.65 0.08 0.03 0.01 )0.04 0.01 )0.01 0.00 )0.01 0.01 )0.01 0.01 0.48 0.03 )0.04 0.01 0.73 0.00 0.03 0.00 )0.03 0.00 )0.01 0.00 0.00 0.00 0.00 0.00 0.51 0.00 )0.03 0.00 0.80 0.02 0.03 0.00 )0.02 0.00 )0.01 0.00 0.00 0.01 0.00 0.00 0.55 0.03 )0.02 0.00  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 635 The Demand for Skills and Labour Costs in Partner Countries Table A2. (cont) Coefficients estimates for elasticities in Table 3 ‘New’ (NMS-5) bll cly dOld dNew N T n ‘Old’ (EU-15) WG-ML LSDVC WG-ML LSDVC 0.11 0.00 )0.04 0.00 )0.01 0.00 )0.02 0.00 650 10 65 0.10 0.01 )0.03 0.00 )0.01 0.01 )0.02 0.01 650 10 65 0.08 0.00 )0.03 0.00 0.01 0.00 )0.01 0.00 1,820 10 182 0.07 0.01 )0.03 0.00 0.01 0.01 )0.01 0.00 1,820 10 182 Breusch–Pagan test of independence v2(1) 3.66 P-value 0.055 26.6 0.00 F-test of strict exogeneity of all the regressors 1.32 0.22 P-value 1.26 0.27 Source: Own calculations. Table A3. Calculated elasticities for material inputs ‘New’ (NMS-5) pmat wph wpl y wpOld L ‘Old’ (EU-15) Smat Smat )0.239*** [0.020] 0.050*** [0.003] 0.086*** [0.007] 0.034*** [0.003] 0.023 [0.015] )0.270*** [0.009] 0.045*** [0.002] 0.152*** [0.005] 0.027*** [0.002] )0.017*** [0.006]  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 636 Lo Turco and Parteka Table A3. (cont) Calculated elasticities for material inputs ‘New’ (NMS-5) ‘Old’ (EU-15) Smat Smat 0.042*** [0.008] 650 0.017*** [0.003] 1,820 wpNew L Observations Source: Own calculations. Note: *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. Standard errors are given in brackets. Table A4. Actual and predicted cost shares of high and low skilled labour Variable Mean SD Min Max ‘New’ (NMS-5) S~h S~h S~l S~l Observations 650 650 650 650 Cross equation restriction [Sh | wl = Sl | wh] 0.05 0.05 0.17 0.17 0.05 0.05 0.05 0.05 0.00 0.01 0.03 0.04 0.31 0.33 0.33 0.35 ‘Old’ (EU-15) S~h S~h S~l S~l v2(1) = 1.23 0.05 0.06 0.05 0.06 0.23 0.07 0.23 0.07 1,820 1,820 1,820 1,820 Cross equation restriction [Sh | wl = Sl | wh] v2(1) = 0.97 Prob. v2 = 0.27 0.00 0.49 0.00 0.51 0.05 0.53 0.06 0.55 Prob. v2 = 0.32 Source: Own calculations. Table A5. Calculated elasticities for material inputs II Smat ‘New’ (NMS-5) pmat wph Smat ‘Old’ (EU-15) High skill intensive sectors Low skill intensive sectors High skill intensive sectors Low skill intensive sectors )0.229*** [0.035] 0.076*** [0.006] )0.203*** [0.021] 0.033*** [0.003] )0.332*** [0.016] 0.064*** [0.005] )0.190*** [0.010] 0.025*** [0.002]  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development 637 The Demand for Skills and Labour Costs in Partner Countries Table A5. (cont) Calculated elasticities for material inputs II Smat ‘New’ (NMS-5) wpl y wpOld L wpNew L Observations Smat ‘Old’ (EU-15) High skill intensive sectors Low skill intensive sectors High skill intensive sectors Low skill intensive sectors 0.083*** [0.009] 0.019*** [0.004] 0.080*** [0.018] 0.088*** [0.012] 250 0.089*** [0.010] 0.046*** [0.005] )0.029 [0.024] 0.018* [0.010] 400 0.139*** [0.009] 0.037*** [0.003] )0.008 [0.009] 0.031*** [0.004] 700 0.152*** [0.006] 0.023*** [0.003] )0.023*** [0.008] 0.004 [0.004] 1,120 Source: Own calculations. Note: *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. Standard errors are given in brackets.  2011 The Authors Economics of Transition  2011 The European Bank for Reconstruction and Development