The international division of labour, local resources and engineering workers: Eastern Europe in the global networks of the semiconductor Industry.

The decades-long internationalisation of the semiconductor industry has been characterised by dynamics that do not easily fit into logics of centrality and peripherality, pointing to the fact that complementary divisions of labour are not the end of the process but mark its beginning. This article revolves around two case studies of design centres in Eastern Europe, focusing on the work of engineers as well as the upgrading of processes at the design centre level. This locally-centred and historical perspective makes it possible to depict internationalisation and integration into the global design networks of the semiconductor industry as a complex dialectical interaction between the global and the local.


The complexity of internationalisation
The electronics industry in general and the semiconductor sector in particular have been in the vanguard of internationalisation and industrial restructuring towards vertical specialization (Lüthje, Hürtgen, Pawlicki & Sproll, 2012).Labour-intensive parts of semiconductor manufacturing were offshored to South-east Asia for the first time in the late 1960s, with more technology-intensive wafer manufacturing following in the early 1980s (Angel, 1994).While some semiconductor companies began to build up product development capabilities in developing markets in the mid-1980s, the industry-wide internationalisation of innovation started in the late 1990s.This long history of internationalisation of manufacturing and development renders the semiconductor sector very interesting for studies focused on the development of the international division of labour.Regions hitherto regarded as peripheral are integrated into the global networks of development as a new phase of internationalisation is unfolding.With some of the development centres having existed for many years, it is possible to analyse their function within company-wide networks after the initial phases of establishment (that offer only limited insights) have been successfully completed.The specific industrial history of Eastern Europe renders it especially interesting for the analysis of global integration.Questions regarding the development of the international division of labour cannot be answered any more by reference to simple dichotomies.The international division of labour is increasingly complex, with new locations, new functional hierarchies and a highly dynamic development.Traditional concepts of centrality and peripherality are challenged by empirical evidence, as are assumptions about the importance of new locations and the technical and organisational capabilities of local engineers.This article will offer empirical insights into these dynamics, mainly focusing on the drivers of this development.Global company strategies driven by economic pressures, global market shifts and industry-wide restructuring are important factors in this process (Ernst & Lüthje, 2003;Lüthje & Pawlicki, 2009;Pawlicki, 2010).However, local characteristics such as historical resources, industrial structure, educational and research institutions and labour markets, as well as the stability of the local labour force, are important in the process of integrating particular locations into the semiconductor industry.These local factors not only enable integration by providing the necessary resources, but are also able to transform global company strategies, leading to complex interrelations between the global and the local, where functional roles are not easily determined by sector-wide internationalisation processes.
The Global Design Network and Global Production Network approaches provide an overarching framework for understanding the dynamics of the international division of labour as an increasingly complex process, changing the way that the concepts of 'centre' and 'periphery' need to be understood, as upgrading takes place.In order to gauge the direction and scale of this development, this article will use two major tools.The first of these, a typology of knowledge work, is a proxy to estimate the complexity of the knowledge work performed at a particular location.The second is a classification tool that allows the results to be classified according to the functional role the entire development centre is playing within intra-organisational development networks.

Dynamics of the international division of labour -shifting assumptions
The dynamics of globalisation do not just entail a continuous integration of previously neglected geopolitical regions into the system of capitalistic production and consumption; they also increasingly change the way in which the international division of labour is organised.The world is not becoming flat; but the relative positions in the hierarchical organisation of production are shifting according both to global company strategies of value creation and to local characteristics, driven by historical dynamics and regulatory decisions.
In their work on the new international division of labour, Fröbel, Heinrichs & Kreye (1980) were able to show how a complementary division of labour started to evolve in the 1970s, as companies from central countries built up manufacturing operations in peripheral locations, mostly based on simple cost considerations.This, already traditional, view of the evolving international division of labour was once again used to analyse the integration of Eastern Europe into the system of capitalist production after 1989 (Berger et al, 2001).However the positions and functions of many production locations in this region shifted quite quickly, based both on existing historical local resources and on fundamental changes in the overall organisation of industry (Lüthje et al, 2012).Fröbel, Heinrichs & Kreye's analysis (1980) was based on the verticallyintegrated company with only simple outsourcing relations.The last three decades have seen dynamics of vertical specialisation that have fundamentally restructured how capitalist production is organised.Starting in the electronics industry, the idea of outsourcing central parts of production has spread throughout the economy (Sturgeon, 1999;Lüthje, Hürtgen, Pawlicki & Sproll, 2012).The Global Commodity Chains (GCC) approach (Gereffi, 1994) formed the basiss for an analysis of such complex production processes and their organisation across both company and geographical boundaries.The GCC approach has focused its analysis on the governance structures of particular GCCs, i.e. the power relations between the various firms that constitute a GCC.The Global Value Chains (GVC) approach (Gereffi et al, 2005;Sturgeon, 2009) reworked the initial GCC framework to allow for a differentiated analysis of the governance structures based on more economically-driven models of sectoral economics.However, as Taylor (2012) remarks, the GCC/GVC analysis lost the geographical dimension along the way, through its heavy focus on sectoral dynamics and governance forms in relation to lead firms.
One of the major achievements of the GCC and GVC approaches was to provide a framework in which processes of upgrading could be analysed, both at the firm and the country level.Particularly important are the learning processes and knowledge flows necessary for entering the value chains of globally operating companies and taking place within these chains.Technological and organisational adaptations to set standards are forcing new entrants to focus on upgrading investments which are often bolstered by organisational and financial assistance from lead companies.Similarly, within intra-organisational networks, new locations constantly need to work to develop their capabilities, often trying to use locally-available resources.The examples of Taiwan and South Korea show how countries are able to shift their relative position within the international division of labour through a specific form of integration into the global capitalist production system (Ernst, 2001).At the sectoral level, the electronics industry has shifted from the initially simple position of contract manufacturing to the provision of complex manufacturing and design services.This is illustrated by the move of some of these companies towards brand name businesses, exemplifying functional upgrading processes paralleled by shifts in their relative positions in the value chain (Sturgeon & Lee, 2005;Lüthje & Pawlicki, 2009).In recent times the debates around upgrading have been extended to encompass issues of work and labour, where economic upgrading and social upgrading are increasingly analysed as interdependent processes (Barrientos et al, 2011;Knorriga & Pegler, 2006).Some scholars point to problems for workers resulting from economic upgrading (Palpacuer, 2008).
This article follows these debates to some extent, discussing the impact of upgrading on shifts in work categories.Here, upgrading is linked to the mostly positive effects for engineers, who constitute a highly qualified and relatively privileged group of workers.However, this situation can be reversed when changes in engineering work take place in central locations, driven by upgrading in peripheral locations (Kämpf, 2008;Mahadevan, 2007).
Because the GVC and GCC approaches lack the aspect of local dynamics (Bair, 2005(Bair, , 2009)), Henderson et al (2002) have developed the concept of 'Global Production Networks' (GPN) to enable a more dialectical analysis of the process of integrating locations into global production systems.GPN analysis broadens its perspective to include a variety of actors and relationships, incorporating, most importantly for this article, the labour market as a fundamental driver of development (Coe et al, 2008;Coe & Jordhus-Lier, 2010).When international companies touch down in a particular location, either through the establishment of supplier relations or through their own operations, they drive local upgrading processes, but simultaneously need to adapt to local regulatory and institutional frameworks, as well as existing historical resources and barriers.One of the major resources and barriers is the local labour market.The stage of development of the labour force, coupled with the local level of wages, strongly influences the initial phase of global integration.However, the future development of a location is also dependent on other local factors, such as local product market size, industrial policy strategies and evolving local industrial structure.China represents the best example of how a huge local market and stringent industrial policies can be used to facilitate upgrading of the local manufacturing locations of foreign as well as Chinese companies (Pawlicki, 2010).An understanding of the integration of local characteristics, or place-related 'situatedness' (Coe et al, 2008), is necessary for an analysis of the upgrading dynamics of particular locations which avoids a top-down approach to the development of the international division of labour.
Despite its focus on inter-firm networks, the GPN approach also calls for a more detailed understanding of the firm (Coe et al, 2008).The black box of the firm has to be opened up to start to understand the intra-firm relationships and the way internal structures and relationships are connected with upgrading processes.Although intrafirm relations are hierarchical, firms can be conceived as networks within networks.Particular parts of the firm are not passive takers of orders but pursue their own goals and strategies.Upgrading shifts power relations as centres, particular design centres, gain independence and importance in specific areas, even obtaining a decisive influence over technology development within the firm.Localisation of increasingly complex work and managerial capabilities represent such shifts in the power relations within the intra-firm network.
These analyses focus on the manufacturing side of the production process, because the initial phase of globalisation was characterised by the offshoring of manufacturing.However, for several years, product development and design have also been undergoing offshoring and outsourcing (Brown & Linden 2009;2010;Ernst & Lüthje, 2003;Ernst, 2005).Innovation work is increasingly organised in Global Design Networks (GDNs) that span geographical and organisational boundaries.The sectoral dynamics of technology development paths, economic pressures and ongoing organisational restructuring enable innovation to be offshored, even making this a necessary precondition for the survival of the semiconductor industry (Lüthje & Pawlicki, 2009).
Motives for R&D offshoring and outsourcing can change over time as learning processes take place and trust relations are developed.Design locations that are initially low profile within intra-firm GDNs can develop strategic capabilities as central management changes its perception of local capabilities and also needs to retain engineering talent through increasingly more complex design projects.Some of the most important prerequisites and biggest issues in GDNs are interfaces that allow the exchange of knowledge across organisational boundaries as well as over distance.The ability to design, organise, formalise and manage knowledge interfaces is the main objective for successful GDNs, whilst knowledge sharing is the glue that keeps GDNs stable.Complex tasks such as chip design, with its highly dynamic technological development, are characterised by high levels of tacit knowledge.Tacit knowledge is a form of very personalised knowledge, making it spatially sticky.It is seen by many researchers as quite resistant to geographical dispersion (Pavitt, 1999).Hence, a major issue in interface design and management lies in structuring and formalising knowledge to make it easily exchangeable across organisational and geographical distances.Since this knowledge management is complex and expensive, the aim of achieving a highly modularised innovation process often faces major challenges.Although the division between tacit and codified knowledge and its equivalent learning processes at local and global levels respectively has been overcome for some time, and global links are recognised as fundamentally important for dynamic local learning in clusters (Bathelt, Malmberg & Maskell, 2004), specific localisation structures are still linked to different knowledge forms.For chip design, the fundamental division in knowledge forms is between digital and analogue chip design.While digital chip design is based on very fast knowledge depreciation and high levels of codifiability, analogue chip design requires high levels of experiential knowledge that does not allow for easy codification.Analogue chip design, as analysed in the following case studies, requires local knowledge generation and accumulation, while digital chip design calls for major global pipelines to access an ever-changing body of knowledge (Brown & Linden, 2010).
The case studies describe a process that has to some extent reverted to cluster learning as analysed by Barthelt et al (2004).The international link, or pipeline, exists from the beginning, since it is the point of departure for every new design centre in a GDN.With this in place, in the ensuing periods, intra-firm local buzz is generated through local functional upgrading.With the development of links that extend outside the design centre, the level of local embeddedness rises.
For years firms have been operating R&D facilities in regions that were at the centre of technology development in their respective industry, such as Silicon Valley for electronics.Such R&D centres were the firms' interface to these highly localised knowledge pools.The need to integrate market knowledge into product development is increasingly becoming a prominent driver for R&D offshoring, especially important if lead markets move abroad.As companies have to cope with the shortening of product life cycles and the expansion and diversification of their product range, time to market considerations are another facilitator for the international expansion of engineering operations.Kuemmerle (1997) proposed a typology of R&D sites in which they were broken down into home-base augmenting and home-base exploiting sites.Home-base augmenting R&D locations aim to tap into knowledge pools developing in specific high-tech regions, where universities and competing companies have established dense local knowledge networks.The output of home-base augmenting R&D operations is used worldwide throughout the whole firm in a range of applications for different countries.The knowledge flows are therefore very broad and directed towards other locations.Home-base exploiting sites, by contrast, are set up to support already existing manufacturing facilities or to adapt standard products to local market needs and help to commercialise them in foreign markets.These sites are dependent on knowledge and information from their central R&D organisation.Their location is driven by the geographical pattern of manufacturing, as well as important new markets.
The evolving complexity of the international division of labour calls for further development of this typology.Sachwald (2008) developed a third category of design centres that lies in between the home-base augmenting and the home-base exploiting sites.This category involves global development centres that are carrying out R&D tasks that can be easily separated and fed back into the parent company's innovation process (Sachwald, 2008).This includes back-office tasks, testing, implementation and documentation.Global development centres are the result of a growing modularisation of knowledge work.The fundamental characteristic of these R&D units is their quite visible detachment from their location, where neither a leading edge knowledge community, nor a considerable local product markets exists.The locational determinant for this type of design centres is cost efficiency and available local engineering talent, driving the location of such engineering units towards low-cost countries with excellent engineering talent.Such centres are neither augmenting nor exploiting the home base because they both use and produce valuable knowledge that can be applied worldwide in a variety of development projects.The local integration of such centres is quite low, compared to the high levels of local integration of home-base augmenting and exploiting locations which either have close relations to local knowledge communities or to local manufacturing, suppliers and markets.
Global development centres differentiate the typology of R&D units in a very important way because this category expresses specific structural developments that keep the internationalisation of innovation in flux.The introduction of this category helps to take into account the new level of internationalisation that is moving to a more systemic level, where global structures become fundamentally vital for innovation processes.
Nevertheless, the empirical data from the case studies presented here shows the limits of Sachwald's (2008) new category, a category that is rooted in the idea of the almost unlimited possibilities of modularisation.Processes of local integration, both as functional development and as increase of local embeddedness, call for a further development of the idea of global development centres.
The discussion of the evolving international division of labour is linked in this article with empirical evidence from field research on the work performed in the development centres in Eastern Europe.
This work-focused perspective makes it possible to formulate an analysis that goes beyond the officially-declared functional roles of particular locations, because specific work categories can be used as proxies to indicate capability levels.This seems necessary because both the GCC/GVC and the GDN approaches omit aspects of work almost entirely in their analysis (Selwyn, 2012).While Coe et al (2008) call for an expansion of the GPN analysis towards issues of labour, their focus is on union strategies and labour action.Since engineers are neither highly organised nor prone to overt labour action, the labour perspective has been shifted in this article towards a focus on the qualification of knowledge work located at a particular design centre.Such a qualification makes it possible to trace the dynamics of development within intra-firm networks linked to processes of upgrading, allowing us to open up the black box of the firm.Based on critical reflection on Gereffi's (2005) typology of work and Huws & Ramioul's (2006) analysis of service work in the global knowledge economy, a typology of knowledge work seems necessary to enable an analysis of the evolving international division of labour in the semiconductor industry.
Within the semiconductor development process four types of knowledge work can be distinguished.
First, there are architectural and research-related jobs involve the development and design of specific (internal) standards, knowledge resources and architectural decisions that guide the actions of engineers within the firm.Such positions are the interfaces between a range of different knowledge fields as well as organisational units, requiring high levels of technical knowledge and long experience with the technical aspects of the work.Traditionally, architectural and research positions have been located in core countries.
A second category concerns integrated positions that are situated within development projects that, locally, comprise all, or most of the functions required to implement the project in question.The local integration of chip design functions enables engineers both to develop extensive knowledge pools through contact with other engineers, and to realise broader possibilities for career development.Integrated projects also involve project management performed by local engineers, shifting the required skills increasingly towards managerial capabilities.As facilitators of upgrading, such projects are extremely important because they make it possible to develop high levels of capability in a specific location.
The third category comprises managerial positions relating mostly to the management of the particular design location, the program and the project team.It is of considerable importance whether local managerial posts are filled by expats or through the recruitment of local talent that not only has better local institutional and cultural knowledge but also has a higher incentive to promote local interests.
The fourth and final category concerns auxiliary work that is performed either when projects are not integrated and only specific tasks are offshored or when specific auxiliary functions, such as in-house software tool development, are located in low-cost locations.This is often based on a complementary division of labour that renders these positions precarious, because the required capabilities can also be found elsewhere.
The development of a form of internationalisation that goes beyond a complementary division of labour generates shifts in the types of knowledge work located at particular design centres, moving from auxiliary work towards more integrated and architecture-related jobs.Processes of functional upgrading drive these changes in the composition of local knowledge work, moving increasingly greater responsibility to the new locations.The spatial stickiness of knowledge (Pavitt, 1999) is a factor in this development, as the organisation of complex knowledge work is facilitated through localisation.Here, both local and global dynamics interlock.Local capabilities, such as engineering talent and labour market characteristics, converge with the global dynamics of technology (the requirement for relatively small design teams) and strategy (the increasing importance of cost competitiveness) to generate further internationalisation.

Methodological remarks
The two case studies presented here were carried out as part of my PhD research, financed though a scholarship from the Hans-Boeckler-Stiftung between 2009 and 2012.I visited these design centres to conduct semi-structured interviews with engineers and local managers.In addition to these interviews, I received guided tours of each of the design centres, which allowed me to gain a more detailed picture of the working atmosphere as well as the various local technical facilities.At the end of 2009 I visited all three of the Eastern European design centres of Midtronic, however, for the clarity of the argument, in this article I have focused only on the largeest and most mature one of these.My trip to Leadtech's Romanian design centre took place at the end of 2010.While visiting Romania I was invited to the National Institute for Research and Development in Microtechnologies (IMT-Bucharest), where I learned in detail about the long history of microelectronics in Romania from semiconductor researchers and was invited to visit their research production facilities.
More general information on the companies and the semiconductor industry is based on a long-term study of sources, mostly professional journals, company annual reports and US Securities and Exchange Commission (SEC) filings.My previous participation in the research project 'Innovation, Global Production and Work: Global Design Networks in the Semiconductor Industry and the Relocation of Science-based Work to China and East Asia, led by Dr. Boy Lüthje and Dr. Dieter Ernst and financed by the Volkswagen Foundation, allowed me to gain detailed background knowledge about the industry and, later, to access the existing data archives.

Midtronic -long history and central functions in the Czech Republic
Midtronic is a niche US chip company developing and manufacturing analogue chips, mostly discrete and power, as well as logic, chips.The company made its name as a specialist in operational cost cutting and, despite its move to higher valueadded products, it is still not perceived as a technology leader in the industry.It has manufacturing operations in Belgium, the Czech Republic, China, Japan, Malaysia, Thailand and the USA.Midtronic's GDN is comprised of technology and product development centres in Belgium, Canada, China, the Czech Republic, France, Germany, Ireland, Korea, Slovakia, Switzerland, Taiwan and the USA.
The Czech Republic has a long tradition in semiconductors that has been the basis of a relatively small but thriving and increasingly internationalised industry since 1989.Chip manufacturing operations and research and education infrastructure have been developed since the 1940s.Midtronic's biggest European design and manufacturing operations are located in a small, remote rural town in east Czechia, where semiconductor operations are the only local industry.The first investments were made by the company here in the early 1990s, through a joint venture with a local semiconductor company, at a time when Midtronic was establishing new international low-cost manufacturing operations. 1 In 1994, Midtronic founded its first Czech IC design centre (CDC) here, in line with its strategic program of cost reduction and to tap into the already-deteriorating local talent pool, initially employing twelve chip design engineers.Currently CDC employs 165 people (Interviews, 2009), which represents around 12.5 % of all its R&D staff (Form 10-K, 2010).
CDC started as an extended workbench, focusing on labour-intensive, low-tech manufacturing-related engineering work.However, because the company's local manufacturing operations focus on old process technologies, these close relations have gradually been abandoned.Most of the products currently developed at CDC use new process technologies that are manufactured in wafer fabs in Japan and the USA.CDC's current product focus is on solutions for computing and consumer systems, aimed at customers located mostly in Asia, where such systems are developed and manufactured.CDC has no local Czech customers.The centre is now based on integrated design operations comprising all the technical functions necessary for new product development, including IC design, layout, application engineering, product engineering and test engineering.CDC's engineers have developed the same technical proficiency as the engineers at other Midtronic sites, providing their technical expertise and driving developments in specific areas.They are company-leading experts in bi-polar technology, both in design and process technology.The technical know-how at CDC is used by the central sales and marketing department in its product planning process.Project management for local projects has been located at CDC for several years.Local project execution is quite autonomous, requiring only higher management functions from other sites, especially marketing and sales.It is only since 2007 that CDC has been managed completely by a local Czech centre manager.
Besides the integrated new product development capabilities, CDC also houses important technology development functions.The design system technology (DST) group provides models which are the most basic link between the process technology in the wafer fab and the building blocks used by IC design engineers in development centres.This is a critical function in the whole IC development and manufacturing process, enabling the organisation of globally-distributed work and constituting a prime example of manufacturing-related R&D work.The DST group at CDC is Midtronic's central modelling group, providing models for technologies used at all of the company's major and advanced manufacturing sites and supporting all of Midtronic's numerous technology development groups.The location of the DST group at CDC has been driven by the low cost of Czech engineers coupled with their technical proficiency and 1 These manufacturing operations were merged into Midtronic in 2003.
very low staff turnover, guaranteeing a continuing accumulation of deep technical expertise.
Although Midtronic competes with other multinational semiconductor companies on the Czech labour market for scarce analogue IC engineering talent, the competition is not too high.The company mostly recruits young university graduates with master's degrees, because the number of experienced engineers is constrained.To be able to find enough new engineers the company has recently been utilising two main means.First, Midtronic has extended the scope of its recruitment area to Poland, Hungary and Slovakia; and second, it has developed quite elaborate cooperation relations with local universities, dating back to the beginning of the Czech semiconductor industry.Together with a multinational EDA 2 provider, Midtronic has funded a laboratory for integrated circuit (IC) design classes to develop practical IC design skills with EDA equipment.Engineers from CDC regularly give lectures on the topics of electronic design and semiconductor technology at local universities.
CDC is characterised by highly stable engineering staff with very low turnover rates, in strong contrast with both its Asian and its US locations.Engineers with tenure of between five and nine years make up the biggest group at CDC.Unlike other regions, this quite conservative career approach does not have negative connotations in the Czech Republic.This stability is regarded as very positive by both local and higher management, guaranteeing reliability and a consistent level of quality, while the engineers' experience is constantly improved without the need for the company to fear that talent will drain away to competitors.Long-term planning of human resource development and a steady upgrading of the development centres' capabilities is enabled, whilst organisational costs are reduced.Despite this high stability, the motivation of engineers needs to be sustained by providing them with increasingly complex projects, thus driving functional and, especially, product upgrading still further.Midtronic's technological area of analogue and mixed-signal products facilitates this dynamic because it calls for high levels of experience and broad exchange of tacit knowledge, while project teams are quite small, comprising only a few engineers.This facilitates the company's aim of sourcing all necessary talent from a single site, which keeps its organisational costs low.This is supported by the still very competitive labour costs in the Czech Republic.During the last 15 years, salary levels have risen considerably, but the design centre manager estimates that they are still two to three times lower than those in the USA.
From the perspective of work categories, CDC has developed progressively from quite narrow auxiliary work in the beginning towards increasingly integrated work with additional research-related positions.This promotes steep learning curves for individual engineers, allowing them to develop not only deep but also quite broad knowledge.However, as marketing functions are organised both centrally and based on market proximity, it seems that on the management level functional integration has its limits for CDC, since local product market integration is missing.The 2 EDA is the acronym for Electronic Design Automation.EDA tools are complex software systems based on several computer-aided design (CAD) tools that help design engineers with the task of designing, synthesising and verifying the integrated circuits comprised of up several millions transistors.existence of research-related positions at CDC, in the DST department, indicates how the geographical patterns of innovation are shifting.The CDC has a focal role in fundamental aspects of technology development, driven by factors such as experience levels and cost considerations.However, the stability of the workforce seems to be the most important factor in location decisions.
CDC, once a prototypical home-base exploiting engineering centre linked to manufacturing operations, has developed into an integrated global development centre that does not easily fit into Sachwald's (2008) typology.Although quite detached from its location, since no local product markets exist, CDC is building up linkages to the local knowledge community, which it is also helping to develop.Providing highlyintegrated engineering work, the CDC is, at least locally, reversing the modularisation of R&D organisation.In the modelling and characterisation unit, CDC is also showing the characteristics of a home-base augmenting R&D centre, because it is providing the basic instruments used at Midtronic's R&D operations worldwide.

Leadtech -fast ramp-up in Romania
Leadtech is a German analogue and mixed-signal chip company developing and manufacturing semiconductor and system solutions aimed at the automotive industry and other markets, as well as for chip cards and security applications.Leadtech is one of the leading companies in these markets.The company employed around 26,500 people as of September 30, 2010.Around 5,700 people are employed in its worldwide R&D operations, with 50% employed in German R&D centres (Annual Report, 2010).Leadtech's manufacturing operations are in Austria, China, Germany, Hungary, Indonesia and Malaysia, while its R&D centres are in Austria, China, Germany, India, Italy, Romania, Singapore, Sweden, the UK and the USA.
Romania has a long history in the electronics and semiconductor industry, with its own IC development and wafer manufacturing operations developed during its socialist era.The remaining capabilities are mostly in educational and research areas, such as the National Institute for R&D in Microtechnologies (IMT) in Bucharest, which is an internationally renowned organisation cooperating in pan-European research projects.The first multinational IC development companies opened shop at the end of the 1990s in Bucharest.Initially these consisted mostly of small start-ups and specialist IC design houses focused on highly cost-driven business models, mainly in markets for analogue and radio frequency (RF) products.Many of these operations have changed ownership several times over the last ten years.Today, Romania, and especially Bucharest, is a well-established specialist location for IC design, hosting several design centres of large multinational semiconductor companies.
In 2005, Leadtech started its IC development operations in Bucharest, aiming to tap into the talent pool of experienced, and still cheap, IC engineers.The average net wage level at the Bucharest Development Centre (BuDC) for an engineer with three years experience is about three to four times the Romanian average wage.Neverthless, overall, costs for the company are still up to three times lower per employee than in locations in Germany.The company was initially able to hire engineers with several years of experience both in analogue IC design and in international development projects, due to the already developed labour market.In 2010, the BuDC employed around 200 people, giving it the status of a medium-sized development centre in the overall intra-organisational GDN of Leadtech.The BuDC focuses on automotive applications and chip cards.
The company's Austrian development operations initially led BuDC in all technical and managerial aspects.This situation has changed in recent years.However, although managerial positions have been increasingly staffed with local management talent, the design centre manager is still an expatriate from Austria.In its initial phase, BuDC served mostly as a support for German, Austrian and Italian development centres, along the line of a complementary division of labour.Strategically, Leadtech is trying to minimise the number of such multi-site projects because a high number of international interfaces limits efficiency and the speed of project completion, translating into higher development costs.Projects for the automotive markets make single-site project location possible, because their size is limited, generally not exceeding ten to twelve engineers.After an initial phase that lasted until around 2009, the development centre is now regarded as a fully-fledged design location, important for the overall business, focusing also on central projects for major customers.BuDC has been able to develop into a competence centre for one specific product group.
BuDC's work covers the whole value chain, from concept to product-ramp, It employs engineers for analogue and mixed-signal IC design, analogue IC layout, behavioural modelling, design flow support, test and product support and this enables it to put together locally-integrated development teams.The competence level of the engineers is regarded as 'almost as high' as in other locations.However, this has to be put into perspective, since the engineers in other locations have experience levels that sometimes exceed 25 years.Almost all of these technical functions were in place in Bucharest from the outset.Even the highly important function of concept engineering was initiated quite early at BuDC.The concept engineer is an architectural position with lead and control functions during the design process, reviewing the compliance of the design with the initial concept.Application engineering is not located at BuDC, becauses all products developed at the centre are for automotive system suppliers located outside Romania.At the managerial level, BuDC is still not fully locally managed.The manager of the design centres is an expat.However, for several years the program managers at BuDC have been locals.Program managers are positioned at the intersection of business line management and pure development.Together with marketing, which is not located in BuDC, they are responsible for one specific project from the initial idea right up until the production ramp-up.There are already international projects driven by program managers from BuDC.
Although the BuDC was, almost from the beginning, locally highly integrated, with a complete development cycle, the complexity of the development projects located lagged behind.The main focus of the development centre was on derivative projects.This situation, where technical capabilities exist but are only used for technically not very complex or sophisticated tasks, is changing.The learning phase was important for both sides -the engineers as well as the company.The engineers had to get accustomed to the company-specific development processes and technologies, while the company's management had to learn about the new capabilities at the BuDC, as well as building up enough trust to locate increasingly complex projects in Bucharest.
As in the Czech Republic, Romania's long history of semiconductor industry made it interesting for multinational companies.Although international companies started investing there later than in the Czech Republic, Romania was integrated quite successfully into the GDNs of the semiconductor industry.However, both of these countries lack a dynamic local product market for semiconductors, local electronic system companies, or a population of engineering talent comparable with those in China or India.As a result, the country can only exist as a small and quite specialised engineering location.The availability of experienced and cheap analogue engineering talent was initially one of the decisive reasons for Leadtech to invest in a development centre in Bucharest.However, recently the labour market has been increasingly drying up, as more semiconductor companies set up shop in Bucharest and universities shift their curricula away from IC design and towards software development, because multinational software companies offer higher salary levels.The workforce at the BuDC is quite stable, with turnover rates of around 8% per year.The manager of BuDC, perceives this level as extremely low, comparing it with design centres in Asia which he managed before taking up his position in Romania.However, when compared with other locations in Eastern Europe it is relatively high.This is the downside of a small but well-developed local labour market, where many other IC design companies exist and compete for talent.
The BuDC is an interesting example of a differentiated upgrading process driven by local and technical characteristics.The existence of a developed labour market and the relatively small size of development projects for the automotive market allowed functional upgrading to take place quite early in the history of the development centre.The long history of a Romanian semiconductor industry with both manufacturing operations and research institutions laid the foundations for an engineering workforce with solid technical knowledge and extensive experience.The relatively high turnover rate, combined with the gradual drying up of the labour market, has driven the relatively dynamic upgrading at BuDC.In this situation, BuDC management, as well as central Leadtech management, are pressed to locate increasingly complex and innovative development projects at the design centre, to feed the intrinsic technical interest of engineers and thereby retain their talent for the company.

Conclusions
The dynamics of change in the nature of engineering work described here provide important empirical evidence that the internationalisation of innovation is not driven by traditional logics of centrality and peripherality and that complementary divisions of labour are not the end of the process but mark its beginning.The case studies presented here show how internationalisation is moving through different phases locally.The extended workbench position of new R&D centres is normal during the initial phase, when both processes and resources within the centre as well as within the GDN have to be established and stabilised.This is followed by processes of upgrading that move the design centre towards a more integrated and independent position.The results of this upgrading and how it takes place are highly dependent on local characteristics.The speed of the process is driven by the dynamics of the labour market.The fast ramp-up and upgrading of the BuDC was determined in large measure by a higher staff turnover rate than at the CDC, as well as the existence of local engineering talent already experienced in international project work.Unwilling to act as a conduit for engineering talent for the local labour market, the management of the BuDC found itself under pressure to offer increasingly complex projects. 3As well as local labour market characteristics, other local factors can influence upgrading processes.An absence of local product markets can limit functional upgrading because requirements for the establishment of extensive field application engineering as well as sales and marketing departments are constrained.Both design centres therefore seem to be stuck in their roles as global development centres, without any possibility of fully securing their future through product innovations based on intimate market knowledge.
Global company strategies determine the results of local upgrading, especially in relation to work and functional roles.CDC's DST department illustrates the willingness of Midtronic to locate central research functions outside of its headquarters based on local expertise and cost advantages.Leadtech's strategy, on the other hand, seems to favour the localisation of functions related to product development, such as concept engineering and program management.While both of these locational decisions have resulted in a global development centre, the knowledge flows within the particular GDN are somewhat different, with CDC having a much more central position.Because the costs of organising international knowledge interfaces are very high, both the interests of companies and the particular characteristics of analogue chip design point to a preference for experience-based knowledge and relatively high levels of local integration, whenever possible.Functional upgrading makes it possible to reduce international knowledge interfaces to a minimum.The spatial stickiness of knowledge (Pavitt, 1999) can thus benefit peripheral locations, allowing them to secure their acquired functional roles.In this view, the spatial stickiness of knowledge does not impede its internationalisation but leads to a process of 'lumping' after new locations have been developed.The global is also breaking into the local at the university education level, where companies are investing heavily in the adoption of global standards.While helping to develop world-class education they simultaneously also push it towards their particular needs.This can result in overly narrow specialisation of engineers as well as of whole locations.
Education is an area where the interconnected relationship between the local and the global is observable.Global strategies based on a complementary division of labour and aiming at cost reduction through internationalisation are directed only towards utilising local engineering talent.The increasingly constrained local labour market both in the Czech Republic and in Romania drives companies to increase their local embeddedness and co-develop local capabilities through direct cooperation with universities.
3 However, turnover rates being too high may result in negative effects on upgrading, at least seen from a labour process perspective: constant changes in project team members require a highly standardised and controlled labour process to organise documentation and continuity within the development project (Mayer-Ahuja & Feuerstein, 2007).
For the theoretical discussion of functional roles of R&D centres it is evident that the expansion of Kuemmerle's (1993) typology was necessary, but the Global Development Centre category alone is not sufficient to enable the complexities of the empirical evidence to be fully grasped, because it is focused too much on the idea of a highly modularised organisation of knowledge work.To be able to analyse the main dynamics of internationalisation, the three part typology can be used as a starting point but there is also a need to focus on the transitional spaces between these three categories to achieve a full understanding of the dynamic nature of internationalisation.This points to the importance of a historical perspective on internationalisation (Henderson et al, 2002) which makes it possible to sketch its development path.Increasingly, locally-integrated global development centres must be viewed as the results of an ongoing development of the international division of labour, calling for further empirical and theoretical work on the typologies of R&D centres.
The growing complexity of internationalisation, as well as the shifting hierarchies within the international division of labour, will result in intensified competition between particular R&D locations.This can be especially important for locations that have until now been central, because their main competitive advantage, broad and extensive technical and organisational experience, is increasingly dwindling.Coupled with the currently much higher salary levels of engineers in developed economies, this situation could lead to major problems.However, as engineers in peripheral locations gain experience and broaden their functional and organisational capabilities, their salary levels start to rise, lowering the gap between them and their colleagues at the centre.This can result in wage explosions, especially in locations with highly dynamic labour markets such as India or China (Mayer-Ahuja & Feuerstein, 2007).The situation in Eastern Europe is somewhat different, because the labour markets are less turbulent and the absence of local product markets limits the overall scope for development.In this situation, it seems that the highly specialised small design centres in this region will not become major competitors in this internationalising industry.© Peter Pawlicki, 2012