The Sources and Aims of Innovation in Services Variety Between and Within Sectors

Transcription

1 The Sources and Aims of Innovation in Services Variety Between and Within Sectors Dr. Bruce S. Tether ESRC Centre for Research on Innovation and Competition(CRIC) The University of Manchester and UMIST International Conference on New Trends and Challenges of Science And Technological Innovation in a Critical Era October 30, 2003,Taipei, R.O.C. This paper has been accepted for publication and is forthcoming in the journal Economics of Innovation and New Technology, Vol. 16, No. 6, pp

2 The Sources and Aims of Innovation in Services: Variety Between and Within Sectors Abstract Services dominate economic activity, but remain under-researched by analysts of innovation and technological change. The early one size fits all theories of innovation in services have in recent years given way to an appreciation that services are diverse, not least in their innovation activities. This paper draws on recent empirical evidence from large-scale surveys undertaken in 13 western European countries, to investigate the extent and the sources of innovation in five services sectors. The analysis includes the extent to which services innovate, and amongst innovators the extent to which they engage in R&D and collaborative arrangements for innovation. The analysis supports the recent literature which emphasises significant differences between sectors in their pattern of innovation behaviour, but also highlights significant intra-sectoral differences in innovation behaviour. This intrasectoral variation deserves much fuller investigation in the future. Other CRIC Papers authored or co-authored by Bruce Tether (see: B. S. Tether (2001) Identifying Innovation, Innovators and Innovative Behaviours: A Critical Assessment of the Community Innovation Survey (CIS), CRIC Discussion Paper No. 48. B. S. Tether and J. S. Metcalfe (2001) Horndal at Heathrow? Co-production, Learning and Innovation Investigating the Processes of Runway Capacity Creation at Europe s Most Congested Airports, CRIC Discussion Paper No. 46. R. W. Coombs, M. Harvey and B. S. Tether (2001) Analysing Distributed Innovation Processes: A CRIC Position Paper, CRIC Discussion Paper No. 43. M. Harvey, A. McMeekin, S. Randles, D. Southerton, B. S. Tether and A. Warde (2001) Demand, Consumption and Innovation: A CRIC Position Paper, CRIC Discussion Paper No. 40. B. S. Tether (2000) Who Co-Operates For Innovation Within The Supply-Chain, And Why?, CRIC Discussion Paper No. 35. B. S. Tether, C. Hipp and Ian Miles (1999) Standardisation and Specialisation In Services: Evidence From Germany, CRIC Discussion Paper Number 30. B. S. Tether (1999) Small Firms, Innovation and Employment Creation in Britain and Europe: A Question of Expectations A CRIC Briefing Paper No 2. B. S Tether (1998) Small and Large Firms: Sources of Unequal Innovations? CRIC Discussion Paper No 11. B. S. Tether and S Massini (1998) Employment Creation in Small Technological and Design Innovators in the UK During the 1980s, CRIC Discussion Paper No

3 1.Introduction 1 In economics, services have long had a Cinderella status, being neglected and marginal (Miles, 2000, p. 371), but their continued expansion - whilst manufacturing contracts - has drawn increasing attention to services in recent years. Given that almost all of our understanding of innovation and of innovation processes at the micro level has been derived from studies of manufacturing (Gallouj and Weinstein, 1997), the growth of services raises questions about the adequacy of our understanding of innovation activities in service dominated economies, particularly as innovation is regarded as fundamental to the competitiveness of advanced economies (European Commission, 2000) This paper provides evidence on the extent, sources and aims of innovation in five service sectors. It does so by drawing on the second European Community Innovation surveys (CIS-2), which were carried out in 13 countries in For most of these countries, the CIS-2 was the first wide-scale and official inquiry into the nature of innovation in services; the first CIS having been confined to manufacturing in all countries except the Netherlands. The inclusion of (certain) services in the second CIS therefore provides a valuable opportunity to investigate innovation in services. The purpose of this paper is to highlight some of the main findings from these surveys, and to consider how this evidence contributes to a broader understanding of innovation and technological change within services. The paper is based on simple descriptive statistics using Europe wide averages. This is for two reasons. Firstly, most of the data were provided to the author in the form of national averages at the (mainly two digit) sectoral level, rather than as firm level micro-data. For most variables, these national averages display considerable variation, which is difficult to understand on a theoretical basis. Most probably, the differences reflect differences in interpretation (of the questions on the survey), rather than substantive differences between countries. To reduce the reporting task, the national averages have been converted to European averages using a method outlined in Appendix A. In using these simple statistics we stress the indicative rather than the exact nature of the results. It is more important to appreciate the patterns in the data, rather than the absolute levels of each phenomenon investigated. By using simple statistics, we make few assumptions about the quality of the data and homogeneity amongst the respondents in the various countries in their interpretation of the survey questions. Secondly, Eurostat has restricted the use of the data made available to the author. Only findings previously reported in a report to the European Commission (Tether et al., 2001) may be reproduced. This prevented the development and presentation of more sophisticated analyses. For these reasons, depth is sacrificed for breadth, and the paper covers a number of issues, which could not be examined in depth in a single paper. The paper is structured as follows. Section 2 provides a sketch or overview of some of the main views on innovation in services. Section 3 then introduces the data-source, the findings from which are discussed in Section 4. In particular, Section 1 This paper draws on the findings of a European Commission funded study on innovation in services (see Tether et al., 2001). I am grateful to the Commission for permission to pursue the academic publication of this work. 3-2

4 4 considers: the extent and sources of innovation in services; the extent of R&D amongst innovating service firms; expenditures on innovation activities; engagement in co-operative arrangements for innovation; the sources of information for innovation; and the aims of innovation. Finally, Section 5 provides some conclusions. 2 Innovation in Services: Changing Understandings Services received little attention from economists and scholars of (technological) innovation before the 1980s, with manufacturing, together with a few peculiar services, notably computer services and telecommunications, being seen as the principal source of new technologies and as the engine of economic growth. This neglect of services reflects the (still widely held) traditional view that services are essentially unprogressive, with little capacity to change, especially from within. Instead, in this view innovation in services is largely dependent upon adopting externally developed technologies that facilitate new service provision and/or enhance service productivity. Manufacturers and the few peculiar services are the source of these adopted technologies. Significantly, the vast majority of the technologies adopted are standard, off-the-shelf technologies, which are widely available and are used by services only in the manner intended by their producers. Thus there is little if any creativity on the part of services as users of technologies. Consequently, as the technologies are widely available, there is little to differentiate service providers by quality, and competition is based on price. The rate of progress in terms of quality and the price quality ratio is also essentially dependent on the rate of technological progress in the supply industries. As den Hertog (2000, p. 499) puts it: The dominant view of innovation in services portrays the process as supplier-dominated, with service firms being dependent on * their suppliers for innovative inputs. From this perspective, services are uninteresting with respect to innovation and technological change. Given the dominance, at least until recently, of this perspective, it is unsurprising that services other than the aforementioned peculiar computer services and telecommunications have attracted little attention from analysts of innovation, who have focused on the producers rather than the users of new technologies. Within innovation studies, this supplier-dominated perspective on services is closely associated with Pavitt s (1984) taxonomy of technological activities. There, Pavitt classified all private services (alongside traditional manufacturers) as supplier dominated (i.e., as essentially passive adopters of externally developed technologies). Subsequent research in this Pavitt tradition essentially involves the translation of an understanding of innovation derived from studies of manufacturing to a service context. Works include those by Pavitt himself (Pavitt et al., 1989), but also by Barras (1986, 1990), Evangelista (2000) and Miozzo and Soete (2001). These studies have recognised that not all services are supplier dominated, but that in contrast to the traditional view - services can be sources of new technologies, with some of these technologies being produced interactively between users and producers - see, for example, Miozzo and Soete s (2001) classification of specialised technology suppliers and science based services, and Evangelista s (2000) identification of science and * However, the views expressed in this paper are my own and do not necessarily reflect those of the European Commission 3-3

5 technology based services and of technical consultancy. In large part these developments stem from the recognition that some of the most obviously knowledge-based and technologically progressive activities, such as design and research and development, are services, and are classified as such when outsourced by manufacturers. However, within this trajectory of research, many if not most services retain the supplier dominated (or technology user ) characterisation. The traditional view of services is being more fundamentally challenged by another group of researchers, largely associated with Lille University in France (Gallouj and Weinstein, 1997; Gadrey and Gallouj, 1998; Djellal and Gallouj, 2000; Sundbo and Gallouj, 2000). This group, which we will label the Lille school, has specialised in the analysis of innovation in services, and argues that innovation in services is different in nature to (archetypal) innovation in manufacturing. By failing to recognise the specificities of services and their innovation activities, mainstream economic and innovation studies have overlooked both the important contributions of services to manufacturing (Gadrey et al., 1995) and some of the most important dimensions of innovation behaviour within services. Intangibility and 6 the interactive aspects of services are especially prominent amongst the peculiarities of services, and service innovation, emphasised by this group of researchers. Indeed, these features can be seen as the keys to understanding service activities (although, see Hill, 1999). Because services tend not to have an independent physical existence, service innovations can be invisible, and because services are interactive, often being co-produced by the provider and user acting together (with simultaneous production and consumption), the authorship of any innovation is often unclear. Furthermore, because service events are often unique, it is often difficult to differentiate between service variations and innovations. But this also highlights the flexibility of services - they are often constantly adapting and reforming their activities to provide solutions to changing and highly differentiated customer requirements. From this perspective, services are far from dull providers of standard activities, but are instead dynamic and fluid, constantly changing to meet customer requirements, and achieving this through creative combinations of hard (i.e., equipment, computer software, etc.) and soft (i.e., human skills, operating practices, etc.) technologies. That is, through mixing equipment with human knowledge and skills. Because it does not fixate on technological innovation in the narrow sense of the creation and distribution of embodied technologies 2 (Evangelista, 1999), the Lille school provides a markedly different view of services from the traditional view. But arguably it paints an exaggerated picture, or one that holds mainly for a group of rather exceptional services - knowledge intensive business services (KIBS), such as consultancy. The generalisability of the Lille school s characterisation of services and innovation in services is questionable, although it should be appreciated that the picture has perhaps been deliberately distorted because the researchers have sought to emphasise the differences between observed patterns of innovation in services (and especially in KIBS) and those expected by the received, supplier-dominated / embodied-technologies based view. In recent works, 2 Embodied technologies is the stock of technologically productive assets consisting of machinery, equipment, plant and operating systems (including both computer hardware and software). This contrasts with disembodied technologies which is knowledge embodied in people and codes (i.e., documents, plans, patents, etc.) (Evangelista, 1999, pp. 8-9) 3-4

6 Lille school scholars, such as Sundbo and Gallouj (2000) and den Hertog (2000), have highlighted the existence of multiple patterns of innovation in services that are not all highly interactional. There has therefore been some convergence between the ideas being developed by scholars working in the Pavitt tradition (e.g., Evangelista, Miozzo and Soete) and those from the Lille school. Both increasingly recognise the diversity of innovation behaviours in services. A third emerging and less defined view of innovation, which is applicable to services, draws on Austrian (Kirzner, 1997) and evolutionary economics (Metcalfe, 1998), and the competence based theory of the firm (e.g., Foss and Knudsen, 1996). This perspective sees competition and innovation as (related) processes, which are themselves related to the scope for strategic positioning. From this perspective, there is typically a wide variety of possible innovation trajectories within each particular sector (Tether et al., 2001), be it manufacturing (or product) or service based. Indeed, one important recent trend amongst manufacturers is greater competition through services, rather than through product quality and price (Howells, 2000). The extent to which firms exhibit variety in their behaviours is partially related to their competitive circumstances (which in turn is partially a result of their competitive actions). In an environment characterised by rapid technological change, such as in the early stages of an industry life cycle, firms are likely to have to innovate (at least in their outputs) to survive, because otherwise their offerings are likely to become obsolete and have no demand. At the other extreme, where there is a high degree of stability or closure in the nature of the outputs expected by the market, there may be little reward for (product) innovation. Process innovation may be possible, so long as this does not impact significantly on the nature of the products or services provided. Between these extremes there is greater scope for individual strategic action, as firms seek to move into and out of various market segments. This movement reflects the competitive process, as firms attempt to both shape their market and respond to the repositioning of rivals 3. In general, this view accepts that the adoption of embodied technologies developed outside the firm can be highly significant for service development, but considers that adoption is rarely as passive or as dependent as in the traditional view, even amongst the supplier-dominated firms of the Pavitt tradition. Instead, this view considers that the vast range of technologies available (and in development) creates an enormous number of possible configurations of use, and furthermore many combinations of hard and soft technologies are not immediately obvious, nor intended, by the technology providers. Even for firms engaged in very similar lines of business, the optimal technology set or configuration is rarely apparent, especially ex ante. Moreover, different uses and combinations of technologies have implications for the nature of the services provided. Small differences in the strategic positioning of firms can imply quite different patterns of investments in hard and soft technologies. Some strategy researchers, notably Porter (1980), argue that there are only a small number of effective strategies, but recent research on services has questioned this (see Uchupalan, 2000), suggesting a wide variety of mixed strategies are possible. Moreover, strategic positioning is dynamic, not static, 3 For example, British Airways has recently sought to focus more on high paying business class passengers, and has reduced its economy class capacity. This example itself shows how services can seek to combine markets.it does not make sense for British Airways to become an exclusively business class airline. 3-5

7 This view therefore sees services as much less passively dependent on manufactures for innovation and technological progress than in the traditional view. Importantly, innovation is seen not only in the creation of new technologies, but also in their creative use, which often reflects the interpretation of market knowledge (which is as significant as technological knowledge). But this perspective also considers that, in general, service providers have much greater control over the nature of their outputs and are less dependent on interactions and relationships with their customers than is suggested by the Lille school. For although the service provider may adapt the service to meet specific customer requirements, in general the provider can decide whether or not to participate in the activity and on the extent to which it is prepared to make accommodations. Often, only very minor (or frequently requested) customisations will be undertaken. This is reflected empirically in much more standardisation of service outputs (i.e., the provision of services that are unchanged between customers) than the Lille school s view would suggest (Tether et al., 2001). Moreover, rather than being co-produced, most services, including many that are relational, are essentially undertaken by the service provider with little input from the user, and the provider remains ultimately responsible for the nature of the service provided. The purpose of this brief review has been to sketch various perspectives on services and their innovation activities in order to provide a context for the empirical findings presented later in the paper. Although an exhaustive review of the literature has not been provided, it should be clear that the various strands of literature are increasingly appreciating the diversity that exists within services and their innovation activities (see Miles, 2001, for a review). To some extent, this variety follows from the residual nature of services, for the categorisation services contains almost everything that is not agriculture and extraction (i.e., primary production) or manufacturing (i.e., secondary production). Consequently, services cover a wide and highly diverse range of activities. The breadth of services, in conjunction with the diversity of their innovation activities, means it is difficult to generalise about what findings to expect from empirical research. An advantage of the old one-size-fits-all theories such the technologydependence / supplier-dominated view, is that these provide strong hypotheses about the behaviour we should expect, not least with respect to innovation. For example, as mere adopters of standard technologies, the supplier dominated view suggests services rarely if ever undertake R&D. But with the changing understanding of services and their innovation activities, these one-size-fits-all theories are increasingly seen as inadequate, or at best partial accounts. Other theories, such as the emerging strategic positioning perspective, present other problems, for without detailed information on the competitive circumstances of the firm this perspective provides little guidance as to what (range of) behaviours can be expected in empirical analyses. This situation would be improved if research were able to identify a small number of widely engaged strategies, although difficulties would arise if firms are found to be engaged in multiple rather than single strategies. Such issues are for future research, and lie beyond the scope of this paper. Instead, this section concludes by briefly summarising, in Table 1, and in a highly stylised fashion, what each of the three perspectives outlined above would lead us to expect in terms of the nature and sources of innovation in services. 3-6

8 3 The CIS-2 and the Data to be Examined The analysis in this paper draws on the responses of service enterprises to the second European Community Innovation Surveys (CIS-2). These surveys, which relate to innovation activities in the period, were carried out in 1997 in 13 western European countries 4. The surveys were based on the revised version of the OECD s Oslo Manual (OECD, 1997), a document originally intended to harmonise surveys of innovation activities amongst manufacturers. Services were 4 Austria, Belgium, Denmark, Germany, France, Finland, Ireland, Luxembourg, the Netherlands, Norway, Portugal, Sweden and the UK. 3-7

9 surveyed alongside manufacturing, although using as slightly changed questionnaire (largely involving the replacement of the word product for the word service ). There is debate about the appropriateness of this type of instrument for investigating innovation in services (Djellal and Gallouj, 2000) with some commentators especially from the Lille school - arguing that significant aspects of service activities (such as client interactions and service delivery) are overlooked. As a consequence, the survey neglects important aspects of innovation behaviour in services. However, nearly 12,000 service enterprises responded to the CIS-2, which makes it a valuable if imperfect resource for examining services and their innovation activities. Not all services were included. Only enterprises with 10 or more employees were surveyed, and only a restricted range of service sectors were investigated. The surveyed sectors were 5 Wholesale trades (NACE Rev.1 51). These were not included in the French survey. Transport services, including land (NACE Rev.1 60) water (NACE Rev.1 61) and air transport (NACE Rev.1 62) services. Amongst these, land transport services tend to dominate, especially in terms of the number of enterprises, the unit of analysis for the CIS-2. Financial services, including banking and other financial intermediation (NACE Rev.1 65), insurance and pensions (excluding compulsory social security)(nace Rev.1 66), and activities auxiliary to financial intermediation (NACE Rev.1 67). Computer services (NACE Rev.1 72). Technical services - i.e., architectural and engineering activities and related technicalconsultancy (NACE Rev ). As mentioned in the introduction, the analysis which follows is based on simple statistics which highlight variety within and between these five sectors. The use of simple statistics is largely because of the nature of the data provided by Eurostat, and because of restrictions placed on the use of these data. To simplify the reporting task, European averages, derived by the method outlined in Appendix A, will be discussed. The reader should be aware, however, that significant differences arose between countries (even after controlling for firm-size and sector of activity) in the pattern of response, notably in the proportion innovators amongst the firms responding to the survey, and in the extent to which these 5 Telecommunications services (NACE Rev ) were also surveyed, but few responses were gathered from that sector, largely because it is highly concentrated, being dominated by a few very large enterprises (See Appendix Table 1 for details of the sample). Because of this, and because of the peculiar nature of that sector, telecommunications will not be analysed in this paper. Some significant service sectors were excluded from the CIS-2 survey, including retailing, business services other than technical services (such as advertising, consulting, legal services and accountancy services), as well as (generally) public services such as education and health. Some countries included a wider range of services than that agreed with Eurostat, in this paper the analysis is confined to the commonly agreed sectors 3-8

10 innovators engaged in R&D and collaborative arrangements for innovation, amongst other variables. The most likely reason for this variation is differences is interpretation of the survey questions, rather than any fundamental differences in the nature of these activities between countries. By reporting European averages, these differences have been suppressed. Readers interested in the national data can find these in a report for the European Commission (Tether et al., 2001) which is available from the author. The interpretive flexibility of the survey questions (e.g., what constitutes innovation and R&D in services, and what should be understood as information and co-operation) is certainly a significant source of variation in the data, between countries, between sectors, and within sectors. Because of this interpretive flexibility, we stress the indicative rather than the exact nature of the findings. The relatively high level of sectoral aggregation is also a problem, as many quite different activities may by combined into a sector. This is a source of intra-sectoral variety, but also raises the question of what is the appropriate level of analysis. This is an issue to which we shall return in the conclusions. Despite the high level of aggregation, and the limited number of service sectors included, those sectors selected are interesting because their patterns of innovation behaviour can be expected to vary significantly. Following Evangelista (2000), transport services, for example, can be expected to have a high proportion of technology users that is, many firms with a high dependence on acquired technologies. By contrast, science and technology based services (Evangelista, 2000) - such as computer and technical services can be expected to show greater engagement in internal knowledge generating activities, such as R&D. Meanwhile, the adoption and use of information technologies is likely to be particularly significant in interactive and IT based (Evangelista, 2000) financial services. Because of the potential for such differences between sectors, a sectoral rather than a national approach is followed in this paper. 4 The Extent and Sources of Innovation The first question concerns the extent to which services engage in (technologically) innovative activities. The CIS-2 approached this by asking the firms: Between , has your enterprise introduced onto the market any new or significantly improved services, or methods to produce or deliver services? This is the basic innovation question, which was accompanied by a rather long and convoluted definition of technological innovation (see Appendix A). There are some problems with this question, and the black and white identification of innovators and non-innovators that it effects (see Tether, 2001). Accentuating the positive, it is apparent that the proportion of innovators varied significantly between sectors, with only a quarter of transport service firms, and a third of wholesalers, introducing new services or methods to produce or deliver them, compared with 44% of technical service firms, almost half of financial service firms, and two thirds of computer service firms (Table 2). Such differences are to be expected, as some services (such as transport services) are clearly engaged in much more stable activities and provide a much more stable set of services than others (such as computer services), amongst which innovation is more continuous and central to competitiveness. Nonetheless. it is perhaps most surprising that a third of the computer service firms denied having introduced new or significantly changed services, or methods to produce or deliver 3-9

11 them, over the three year period. This may be due to differences in interpretation as to what constitutes innovation. The fact that a firm declared it did not innovate does not necessarily mean it did not invest in new technologies. Further analysis on the UK data (Tether, 2001) shows that many of the non-innovators introduced new technologies such as and Internet access. From studies of manufacturing, it has become customary to expect the propensity to innovate to increase with firm size. This has been explained by large firms tending to engage in more activities, and thus having greater scope for innovation (Sirilli and Evangelista, 1998). This pattern is found for most of the service sectors examined here, but not, interestingly, for computer services, amongst which small firms appear as likely to innovate as large firms. In general there also appears to be a weaker relationship between firm size and innovation in services than is commonly found for manufacturing, suggesting there are fewer economics of scale (which would help explain the lower concentration of most service activities except finance - as compared with manufacturing). Unfortunately, one of the weaknesses of the CIS-2 is that it did not ask the firms that did not innovate why they did not. It would be especially interesting to know how firms that are active in highly innovative sectors such as computer services survive without innovating. These questions were not investigated, and instead the remainder of the survey addressed only the innovators. Another weakness of the survey is that it did not ask about the impact of the innovations on the firm, for example in terms of the percentage of turnover due to new services. 3-10

12 4.1 The Sources of Innovation The fact that a firm claimed to have innovated need not mean it developed its own innovation(s). The questionnaire asked the innovators : who developed these services or methods? Three responses were available: mainly other enterprises or institutes ; your enterprise together with other enterprises or institutes ; or mainly your enterprise. This might be an oversimplification, for the answer can be a matter of perspective, particularly if the innovation is be based on an idea, the implementation of which required the adoption externally developed technologies. Is such an innovation found in the idea or the technology required for its implementation, or both? Another limitation is that for innovators that involved external enterprises or institutes, the type of organisation(s) (i.e., customers, suppliers, etc.) involved was not detailed. According to the findings (see Table 3), about half the firms claimed that they had mainly developed their own innovations, with a third declaring the innovations were developed jointly with others, and about a fifth admitting they were mainly developed by others 6. These proportions differed significantly between sectors, with the external development of innovations being most common in transport services and least frequent in computer services. The pattern for the internal development was roughly the reverse of this, whilst that for the joint development of innovations varied less, but was greatest amongst financial services. These findings point to different patterns in the innovation processes of different sectors. Given their characterisation as technology users (Evangelista, 2000), the relatively high dependence on externally developed innovations in transport services is unsurprising, especially if it is assumed that innovations mainly developed by others were in the main developed by suppliers. But it is perhaps surprising that the extent of dependence on. externally developed innovations was not greater. Over two fifths of the transport service firms claimed to be the authors of their own 6 These findings treat each innovator equally, but it is possible that if the significance of the innovations introduced were taken into account (Tether, 1998), then the significance of these different sources might change.in particular, the significance of jointly developed innovations might be greater, because higher levels of innovation activity tend to involve greater interaction and collaboration between firms (Tether, 2002) 3-11

13 innovations. Although the source of the innovation can be a matter of perspective, if correct these results cast doubt on the adequacy of the characterisation of transport service firms as (passive) technology users. By contrast, computer services appear to depend mainly on their own internal resources for innovation, although innovation is sometimes undertaken jointly with others (which presumably refers to customers in most instances). Indeed, the relatively high rate of joint innovation in financial services may well reflect collaborations with computer service firms, which are engaged for the development of bespoke or significantly customised IT systems. These simple findings raise more questions than they answer, but they do illustrate considerable diversity within as well as between service sectors with respect to the sourcing of innovations. Many services admit a significant involvement of others - even a dependence on others but most contend that they have undertaken (at least some of) the creative activities necessary for the introduction of their own innovations. This contrasts with the widely held perception that services are mere passive, supplier dominated adopters of externally developed technologies. These findings also suggest the strong emphasis the Lille school has placed on interactive innovation may be exaggerated, for even in technical services, which might have been expected to exhibit a high level of interactive innovation, only about a third of the firms admitted to having innovated jointly with others. 4.2 Engagement in Research and Development Activities Further evidence that services actively engage in technologically creative activities is found in the extent to which they undertake R&D. Traditionally, services other than those in the peculiar sectors of telecommunications and computer services have been assumed not to engage in R&D. Indeed, Pavitt s classification of services as supplier-dominated partially derived from the statistical evidence available at that time which showed that services undertook very little R&D 7. In recent years, R&D surveys have recorded substantially more R&D as being undertaken by services, and not just by computer services and telecommunications (Young, 1996). The extent to which this reflects real change, or a growing recognition by statistical agencies (and the responding firms) that services engage in R&D is unclear, but it does suggest that services are more technologically creative than has been widely supposed. 7 His classification was also derived from working with the SPRU innovations database, which is a list of significant technological innovations introduced in the UK between 1945 and Arguably, as services tend not to produce tangible innovations, their innovative outputs were much less likely to be recorded in this dataset. 3-12

14 The survey found that about half of the firms that engaged in innovative activities undertook R&D, with roughly half of those doing so on a continuous basis (Table 4). 8 As expected, considerable variation existed between sectors in the proportion of innovating firms engaged in R&D this being lowest (as expected) amongst transport services, and highest (again as expected) amongst computer services. But it is perhaps surprising that as many as a 30% of the innovating transport service firms claimed to be engaged in R&D, even if most only undertook R&D on an occasional basis. Firm size also had some influence here, as did the sourcing of innovations. Unsurprisingly, firms that introduced innovations (mainly) developed by others were less likely to engage in R&D, but even amongst that group some did (although very rarely on a continuous basis). This suggests that even amongst technology users significant learning can be necessary to introduce innovations; learning which is associated with investing in the firm s absorptive capacity (Cohen and Levinthal, 1989). Amongst those that developed their own innovations or which developed innovations jointly with others, more than half (claimed to have) engaged in R&D, with roughly 30% doing so continuously. Unfortunately, the survey does not provide information on other sources of innovation internal to the firm. It may well be that many firms were undertaking R&D type activities but did not recognise these as R&D. R&D is often perceived by respondents to be a formal distinct activity (Howells et al., 2001). 8 It is also worth noting that the figures presented in Table 4 again consider each innovator equally, but it may well be that those engaged in higher levels of innovation activity are more likely to engage in R&D. 3-13

15 Table 4 also shows, for each sector, the median proportion of R&D employees in total employment for firms engaged occasionally and continuously in R&D. This varies considerably between sectors, but especially for computer and technical service firms, this data indicates substantial variation in commitment to R&D amongst firms operating in the same sector. Some firms appear to be committing substantial resources to R&D, whilst others are not (or did not claim to be) engaged in R&D. How far this variation is real, or due to interpretive differences, is not clear, but it is likely that significant real variation does exist. 4.3 Expenditures on Innovation-Related Activities 9 Another indicator of the variation between and within sectors is found in the extent to which firms committed resources to innovation activities. This can be assessed through the question that asked the firms about their expenditures, in 1996, on seven innovation-related activities. These were expenditures on: internal R&D; the acquisition of externally undertaken R&D services [Acquired R&D]; the acquisition externally developed machinery and equipment [Acquired M&E]; the acquisition of other externally developed technologies [Acquired OET]; on preparations for the introduction of innovations [Preparations]; on training directly linked to innovation [Training]; and. 9 This section is based on data from France, Germany, Ireland, Norway Portugal, Sweden and the UK. For simplicity each country s the data has been treated as a simple rather than a weighted sample, with the findings for each country then weighted in accordance with the methodology outlined in Appendix A to provide the reported finding. An alternative processing of the data would result in slightly different patterns, but as emphasised throughout the paper, the findings should be interpreted as indicative, rather than as exact. 3-14

16 on the market introduction of innovations [Market Introduction.]. The sum of spending on these activities provides the total expenditure on innovation activities (in 1996) These data are interesting because they indicate the extent to which firms committed resources to innovative activities, although they also come with a health warning as the categories are subject to interpretation. For example, whether an expenditure on training is linked to innovation depends on what is understood by innovation (and indeed training ). For those that provided this data, Figure 1 shows - for each of the five sectors - the median total expenditures per employee, together with the expenditures of the firms at the 25th and 75th percentiles of each sector s distribution. This shows two things. Firstly, the amount recorded as having been spent on innovation varies significantly between sectors. On a per employee basis, the median computer service firm is recorded as having spent roughly four times the amount that the median transport service firm spent. But the figure also shows that the level of expenditure recorded varies widely between firms active in the same sector. It is common for the firm at the 75th percentile of the expenditure distribution to be recorded as spending up to 10 times (per employee) the amount of the firm at the 25th percentile. They are also skewed to the left the mean expenditure would be considerably greater than the median. To some extent, this variation may be a product of the methodology. Apart from differences in interpretation as to what constitutes innovation, the data refer to a single year, 1996, and it is quite possible that a firm s expenditure in that year was atypical of its normal pattern of expenditure on innovation. Especially in small firms, innovation expenditures can vary widely year on year, especially if these involve substantial lumpy investments in capital equipment. For these reasons, there may be less intra-sectoral variation in reality than appears from the data. It is likely, however, that if these methodological problems were overcome, substantial intra-sectoral variation in commitments to innovation would remain. The second interesting aspect of this data is the composition of innovation costs and how this varies both between and within sectors. The mean distribution of innovation costs amongst firms active in the various sectors shows the importance of the various activities differs substantially between these sectors (Figure 2) 10. In particular, internal (or intramural) R&D accounts on average for only about 10% of total innovation costs amongst transport service firms, but over 40% of these costs amongst computer services (which, as has been shown, also tend to claim much more expenditure per employee on innovation). The relative importance of investments, both in machinery and equipment and in other externally developed technologies (such as computer software), tends to follow the opposite pattern, together accounting for (on average) only about a quarter of innovation costs in computer services, but about 60% of these costs in transport services. In all five sectors the remaining quarter to 30% of innovation costs were spent on preparations for the introduction of innovations, on training directly linked to innovation, and on the market introduction of 10 Note that this treats each firm (which provided innovation costs data) equally, whether it spent a lot of a little on innovation. This is partly because of the methodological difficulties discussed above in relation to the total spending on innovation. 3-15

17 innovations 11. Preparations is especially vague, but the finding that training (directly linked to innovation) accounted for on average - less than 10% of total innovation costs, may suggest that on the whole relatively little (off the job) learning on the part of labour was required for the introduction of the innovations. Meanwhile, the finding that the market introduction of innovations tended to constitute only a small proportion of total innovation costs suggests that product innovations tended to be introduced into already established market relations, and consequently the high costs of market formation were rarely incurred. It must be emphasised that the above is just an average picture (based on sample means) and as such conceals substantial variation. One illustration of how this average varies is shown by considering two sub-samples: one of those firms that (claimed to have) incurred expenditures on internal R&D; the other of those that did not. For the firms that did not incur expenditures on 11 In each sector, a small proportion of innovation costs were spent on acquired R&D services. 3-16

18 internal R&D, acquired technologies accounted for (on average) over 60% of total innovation costs in all sectors except computer services (where they accounted for about 45% of costs). The three other activities of preparations, training and market introduction, accounted for the almost all of the remainder. But amongst firms that declared expenditures on internal R&D, this activity tended to account for a substantial share of total innovation cost (which in all sectors tended to be higher than amongst firms that did not declare expenditures on intramural R&D). Even in transport services, an average of about 40% of total innovation costs were assigned to R&D amongst those (few) firms that claimed expenditures on this activity. In computer services this proportion reached 60%. These findings suggest that when R&D is undertaken it tends to constitute a substantial proportion of total innovation related activities, even in firms active in supplier dominated / technology user sectors. But again we caution about possible mismeasurement are all of those firms recorded as not undertaking R&D genuinely not doing so, or are some not recognising some of their activities as R&D and therefore not recording this as an innovation expenditure? Such an under-recording of R&D and total innovation costs can lead analysts to infer that services are less innovative or less committed to innovation than is truly the case. At a broader level, however, these distributions highlight both the diversity that exists between sectors (in terms of average behaviour) - from a high dependence on acquired technologies in transport services, to a dominance of internal activities in computer services - as well as what appears to be significantly differentiated behaviour within sectors. 4.4 Participation in Co-operation Arrangements for Innovation This section examines the extent to which services interact with other organizations for innovation, particularly their participation in co-operative arrangements for innovation. Such arrangements were defined in the survey as active participation in joint R&D and other innovation projects with other organisations. It does not necessarily imply that both partners derive immediate commercial benefit from the venture. Pure contracting out work, where there is no active participation, is not regarded as cooperation. A supplier-dominated perspective would suggest that services rarely engage the inter-active learning associated with co-operative arrangements for innovation. Instead, they adopt standard, off-the-shelf technologies. On the other hand, the Lille school suggests services are highly interactive, and co-operation, particularly with customers, is common place. Indeed, it may even be considered the modus operandi of innovation in services. The survey found that about 30% of the innovating service firms had engaged in cooperative arrangements for innovation with external partners (Table 5), a proportion which suggests these arrangements are quite widespread (implying significant inter-organisational learning), but not necessary for innovation in services. It may be that different types of innovation tend to be achievedthrough co-operations as compared with those that are achieved by firms going it alone. In particular, low-level technological adoption or imiatitive innovation might predominate amongst firms that innovated independently, whilst co-operations may be associated with more substantial, complex or uncertain developments, particularly were these involve changes to the services provided (Tether, 2002). The unsurprising finding that co-operations were more common (at around 40%) amongst firms engaged in the more technologically dynamic activities of computer and technical 3-17

19 services lends supports to this interpretation, but unfortunately the questionnaire did not ask the firms directly about the nature or purpose of the collaborations, nor about how significant they were for the firms innovation activities. When assessed by the different types of collaboration partner, suppliers and customers (or clients) were the most widely engaged, but the proportion of firms with these arrangements was perhaps surprisingly small, especially from the Lille school s perspective. Even amongst the technical (and computer) service firms, which tend to be characterised as highly interactive and providers of co-produced innovations, only a fifth of the innovators had engaged in collaborative arrangements for innovation with their customers. Universities and research institutes were the least widely engaged partner types, but some firms did have collaborations with these. Again, such collaborations may be associated with higher-level innovation activities, and consequently the small proportion of firms with these arrangements may understate their true significance (Tether, 2002). 4.5 Sources of Information for Innovation The analysis of co-operative arrangements for innovation suggests these are neither rare nor commonplace amongst innovating service firms. The evidence on the broader importance of various sources of information for innovation is now examined. This derives from a question that asked the firms to rank the important of various sources of information between 0 not relevant and 3 very important. What constitutes information for innovation is not clear, and was not specified by the survey, but it was intended to identify the sources of ideas for innovations and how to implement these The nature of the information is likely to vary significantly between sources for example between the technical information found in patents and the market information sought from customers but unfortunately there is no further classification. Another difficultly is how the question was answered, as the importance of a source may be judged relative to other sources, or in an absolute sense. 3-18

20 Table 6 ranks these sources from highest to lowest in accordance with the proportion of firms that indicated they were very important. Also shown is the proportion of enterprises that indicated the source was relevant. Sources (of information) within the firm itself were the most widely recognised as relevant and very important source for innovation, which echoes the findings of Sundbo and Gallouj (2000) that services typically identify their own innovations. Unfortunately this was not disaggregated further to identify which sources within the firm were used, but it does suggest most firms (including many that did not claim to undertake R&D and only introduced externally developed innovations) perceive themselves to be taking an active role in sourcing innovations; they are not just passive adopters of technology. Amongst the various external sources of information for innovation, the most widely identified as both relevant and very important were customers (or clients), suppliers, and competitors. Over 80% recognised customers as a relevant source of information for innovation, and it is notable that significantly more firms recognised customers as a very important source of information than engaged in collaborative arrangements for innovation with them. This suggests that although formal collaborations with customers were not especially common, informal interactions were frequent. Meanwhile, three-quarters of the firms recognised suppliers and competitors as relevant sources of information for innovation, but it is perhaps surprising that greater proportions did not identify these and especially suppliers - as very important. This suggests the great majority of services do not see themselves as being dependent on suppliers for innovation, even though (as has been shown) 3-19

21 a substantial share of their innovation costs are accounted for by investments. Thus, while suppliers are an important source of technology, they are perhaps not a major source of ideas that lie behind the innovations (which may draw mainly on either internal business process knowledge or market knowledge). Instead, services seem to source ideas for innovation from many internal and external sources, and may invest in new technologies from suppliers to implement their innovative ideas. This wide sourcing of ideas perhaps explains the high proportions of firms that identified fairs and exhibitions, professional meetings and journals, and computer networks as relevant sources of information for innovation. As the third most widely identified external source of information for innovation, the importance attached to competitors is notable. It is sometimes argued that service innovations (as opposed to internal process innovations) are difficult to protect from imitation (Miles et al., 2000). In this context, a high level of competitor watching is to be expected, whereby firms observe the behaviour of their competitors and seek to monitor the impact of any innovations they may introduce. If their rivals innovations appear successful, the firm will seek to rapidly imitate them. This problem of rapid imitation may be less frequent when competitors have more differentiated, and less imitable capabilities, and it is notable that competitors are less likely to be seen as a very important source of information amongst technical and computer services, which are likely to have more distinctive and less imitable technological capabilities. The four least widely recognised sources of information for innovation were consultants, universities and research institutes, and patents. None of these sources was identified as very important by more than 10% of the innovating firms 13. This suggests these sources are rarely important for innovation in services, but it may be that they are used more frequently by service firms engaged in higher levels of innovative activity. Indeed, further analysis of the data using factor analysis suggests three groups of sources, other than those within the firm, suppliers and customers. These are: 1. competitors which suggests these firms may be following an imitative strategy, watching their rivals and seeking to copy their successes whilst avoiding their mistakes; 2 a combination of professional conferences, meetings and journals, computer networks and fairs and exhibitions, which suggests a search strategy of actively scanning the environment for ideas and technologies for innovations; and 3 universities, research institutes and patents which suggests an importance of disembodied technical information, perhaps associated with higher levels of innovation. Although none of these were exclusive to firms in any particular sector, the first was more common amongst financial services and wholesalers, and was less common amongst computer and technical services; the second was especially common amongst technical services but was uncommon amongst computer and transport services; whilst the third was also uncommon amongst transport services but more common amongst technical and computer services. These patterns require fuller analysis, but they do suggest variety within as well as between service sectors in the sourcing of information for innovation. 13 Although individually financial services attached slightly greater importance to consultants. 3-20

22 4.6 Why Innovate? The Aims or Objectives on Innovation The final question is why firms innovate. To address this, the survey asked the firms to rank, between 0 not relevant and 3 very important, the importance of ten aims or objectives of innovation. These were: replace (old) services being phased out; improve service quality; extend the service range; open up new markets or increase market share; fulfil regulations and/or standards; improve internal business process flexibility; reduce labour costs; reduce materials consumption; reduce energy consumption; and reduce environmental damage. Of these, improving service quality was the most widely recognised; 95% of the innovating firms declared it relevant to their innovation activities, and over 60% claimed it was very important. Opening new markets (or increasing market share) and extending the service range were also widely recognised (each was relevant to about 85% of the innovating firms, and very important to about half). In general, the high significance attached to these aims suggests a strong product orientation to the innovation activities of the service firms, but as service outputs typically lack an independent physical existence (Gallouj and Weinstein, 1997), there is commonly a close nexus between the process (and organisation) of provision, and the nature of the services provided. This means it can be difficult to apply the conventional distinction between product and process innovation 14. Put simply, changes to the services provided can be achieved through changes to the process of provision (Hipp et al., 2000). Improving the internal business process and reducing labour costs were recognised as relevant by about 80% and 70% of the innovating firms respectively, and each of these was very important to about 30%. These aims are generally associated with process innovation, but these changes may impact directly on the nature of the services provided (i.e., they may have product innovation type effects)(hipp et al., 2000). For example, the aim of increasing the flexibility of internal business processes may relate to a more fundamental objective of responding more fully (or rapidly) to differentiated or changing users needs. Such flexibility can be a source of differential pricing, and higher profit margins. The relatively high significance of reducing labour costs is unsurprising given the high labour intensity of many services, although changing the nature of the labour employed can impact significantly on the nature of the services provided. It is notable in this context that reducing 14 Perhaps it is better to distinguish between inward and outward looking innovation activities. Inward looking activities are concerned with how the firm undertakes its activities. Changes to these, which are often associated with reducing costs, may have knock-on effects on the services provided. Outward looking innovation activities are related to how the firm interacts with others, most notably its customers. The firm may seek to provide improved services to better meet its customers requirements, but it may also introduce new or alternative means of delivering services (e.g., internet banking). This contrasts with the conventional association of delivery with process innovation. 3-21

23 labour costs is less widely recognised as a significant aim of innovation in computer services. The other objectives of innovation were less widely recognised 15. Factor analysis was undertaken to analyse these aims of innovation further. This was done for France, Germany and for the UK and Ireland (pooled 16 ), and discovered very similar results (see Table 8) 17. In each case a three factor solution was found (with initial Eigenvalues > 1), which explained about 60% of the total variation. In all three cases the first factor was associated with cost reductions (especially energy and materials costs), but also with regulatory compliance (or fulfilling standards) and with reducing the environmental impact of services. Reducing labour costs also loaded on this factor, although this was also important to the second factor, which was associated with service enhancement and continuous improvement. Reducing labour costs apart, this second factor loaded heavily on improving the flexibility of the business process, and improving the quality of the services provided (and in the case of France with replacing obsolete services). This combination points to finding the appropriate balance between efficiency and effectiveness in service. provision. The third factor was clearly associated with market expansion, loading heavily on opening new markets (or increasing market share) and extending the service range. 15 Fulfilling (or complying with) regulations and/or standards, and replacing previous services were both recognised as relevant by more than half the innovating firms, and roughly half claimed each of reducing the environmental damage, reducing materials costs, and reducing energy costs were relevant objective of innovation 16 These two countries have been combined to increase the sample size and this is justified because there is no language difference so the understanding of innovation should be similar. 17 Further analysis on the Portuguese data, and that for Norway and Sweden (pooled) also found similar results (especially for Portugal). 3-22

24 It comes as no surprise that the first of these factors had a strong sectoral association, with transport service firms (and to a lesser extent wholesalers) being more likely to score highly on this cost reduction factor, whilst computer services were less likely to score highly on this factor. The second and third factors had weaker sectoral associations, although financial services tended to score highly on the improvement / service enhancement factor, and computer services tended to score more highly on the market expansion factor. But while firms in certain sectors may have been more or less likely to score highly on each of these factors, none of the factors was associated exclusively with one or two sectors. Instead, firms in all sectors had high and low scores for each of the factors, which highlights intrasectoral variety in the aims or objectives of innovation. How the pursuit of these different strategies influences innovation behaviour within a given sector is a matter that deserves fuller investigation. An initial analysis of the French data (which provides the largest sample) shows that, after controlling for firm size and sector of activity, those firms that scored highly on the market expansion factor were more likely to engage in R&D and were more likely to have co-operative arrangements for innovation. However, these finding was not repeated for the German or UK and Irish samples. A fuller analysis of these relationships is not provided in this paper, partly because of space restrictions, but also because of restrictions placed on data analysis and reporting by Eurostat. The pursuit of different objectives of innovation amongst firms in the same sectors does however illustrate the scope for strategic positioning within service sectors. Moreover, the initial findings from the French data suggest these different strategies have an influence on the nature of the activities engaged in for innovation. 3-23

25 5 Conclusions Cinderella has decidedly come to the ball - so declares Ian Miles (2000, p. 372), one of the pioneers of studies on services and their innovation activities. Miles is commenting on the recent rapid expansion in interest shown in services and their innovation processes, not only by academics but also by policy makers. It is increasingly recognised in both these communities that as services dominate economic activities in advanced nations it is important to understand their innovation activities more fully 18. Prior to the late 1990s services, other than telecommunications and computer services, received little attention from scholars of innovation. The little attention they did receive initially gave rise to highly generalised one size fits all theories particularly that services are technology adopters and supplier dominated. Research undertaken in the late 1990s, partially through the use of new data sources such as large scale datasets, has argued for the existence of multiple innovation patterns in services, and the early one size fits all approaches have now given way to taxonomies and typologies (e.g., Evangelista, 2000; Miozzo and Soete, 2001), which tend to be closely linked to sectoral analyses. Thesedevelopments are useful, not least in highlighting differences that exists between sectors. The analysis presented in this paper shows that these typologies provide useful guideposts for understanding differentiated innovation behaviours amongst services. But these typologies also effectively reify sectors as a unit of analysis, and so encourage the expectation of a high degree of homogeneity within sectors. In this paper we have consistently found diversity in the behaviour of service firms with respect to innovation not only between but also within sectors. To some extent this apparent variation in behaviour may be due to the high level of aggregation used the sectors were mainly defined at the 2 digit level. However, even technical service firms, which are defined at the four digit level (74.20), displayed considerably diversity in behaviours. An important difficultly in analysing these differences (particularly within sectors) is that there are two sources of variety: the first is genuine differences between firms in their activities; the second is differences in interpretation with respect to the meaning of the concepts (innovation, R&D, etc.) being investigated by the survey instrument. Future versions of the CIS should seek to reduce the scope for interpretive flexibility in answering the survey. One of the reasons for the variation within sectors in innovation behaviour is that sectors, even when defined at four-digit level, contain a considerable diversity of activities. Take computer services, as defined at the two-digit level in the standard industrial classification (NACE Rev.1 72). This sector is usually considered high technology, knowledge intensive, and innovative, but it contains some very routine activities, such as data entry. It is reasonable to expect significant variation in behaviour between these activities and others, such as writing computer software or 18 That services are now being taken more seriously is demonstrated by their inclusion in the second (and third) European Community Innovation Surveys. The third CIS was undertaken in 2001 and relates to innovation activities between 1998 and

26 providing consultancy services. But this is not just an argument for finer sectoral definitions, for there can also be significant differences between activities that are very similar. Consider, for example, the differences in innovation trajectories of the traditional airlines and the low cost or no frills carriers such as easyjet and Ryan Air, all of which would be classified by NACE Rev.1 as Both the traditional carriers and the low cost airlines are innovating, but in different ways. Their innovation strategies are based on positioning for different segments of demand, and different understandings or conjectures about future market development. It is this differential behaviour amongst firms within the same sector which leads to the evolution (and perhaps splitting) of the sector. Sectors are therefore not static, but dynamic. Research to date has made considerable progress in understanding differences in innovation activities between sectors (as conventionally defined), but greater knowledge is now needed about the extent and consequences of variation within sectors. This raises the issue of appropriate levels of analysis. Arguably sectors as conventionally defined are problematic; they rarely include only firms that provide genuine substitutes or that are in direct competition with each other. One alternative approach would be to investigate firms that are actually in direct competition, and to investigate the consequences of these firms taking different approaches to innovation. This approach sees innovation as closely related to competition, and making strategic choices, with innovation being about disrupting the existing situation, both in terms of seeking to gain an advantage over competitors and seeking to reduce competitive disadvantages. A second alternative is to study strategic networks (Gualti et al, 2000) or systems of innovation (e.g., Carlsson et al., 2002), that is of service and other firms that are interdependent in their innovation activities. The innovations produced are the result of their collective efforts, rather than being due to independent efforts by individual firms. Clearly the approach depends on the question being investigated, but a significant weakness of the CIS is that it fails to link innovation activities to commercial performance. This relates to the relationship between innovation and wealth and employment creation, which is a central interest, especially for policy makers. This paper has sought to review the evidence provided by the CIS-2 on innovation in European services. The paper covers many issues and certainly sacrifices depth of analysis for breadth. This is partially because of some important data restrictions and methodological problems that prevented more detailed analyses. However, although only descriptive, the analysis presented points to significant variation both between and within sectors. Miles analogy of Cinderella coming to the ball is therefore not wholly appropriate, for there appears to be not one but many innovation patterns in services. Some firms may be innovating by simply copying their rivals and/or by adopting externally developed, off-the-shelf technologies. This requires little creativity and risk taking, and it is questionable whether these firms deserve the title innovators. But others are undertaking genuine innovation, and some are committing substantial resources to innovation activities, including R&D. The point is there is considerable variety amongst innovating service firms that is not wholly explained by the nature of the particular sectors in which they operate. This variation deserves much fuller investigation and explanation in the future. 3-25

27 Appendix The Survey and the Presentation of the Results This appendix outlines the methodology used to prepare the data presented in the paper. With the exception of the factor analysis assessing the aims of innovation, only European averages are presented. This conceals significant and substantial variations between countries. The presentation of European averages has been done to simplify the reporting task, but should not be taken to imply that these differences are of no importance. As mentioned in the paper, the Community Innovation Survey was developed on the basis of the OECD s Oslo Manual (OECD, 1997). The great achievement of the Oslo Manual is to institute an extension of the study of innovation beyond R&D. The intention of the CIS is partially to understand more about innovation, but also to gather internationally comparable statistics about innovation performance. The prominence of the national systems of innovation literature (Lundvall, 1993; Nelson, 1993) means homogeneity in results from all countries should not be expected, and nor can the differences between countries be dismissed as mere statistical noise. However, the CIS-2 discovered substantial differences between countries in the extent to which firms were innovating (and in the characteristics of the innovators). This was true in services (where 15% of firms in Belgium were innovators compared with 60% in Ireland), but also in manufacturing (Tether et al., 2001). Although some of the differences in the patterns of response may reflect genuine differences between countries (for instance, the Nordic countries may genuinely have higher rates of co-operative activity for innovation), it is likely that the bulk of the inter-country differences arose due to different cultural or linguistic understandings of the concepts, such as what constitutes innovation, R&D and co-operation, etc. Such differences arose despite the use of a common questionnaire (although translated into the various national languages 19 ). One reason for the differences in the understanding is due to innovation being something that is difficult to define. Accompanying the question about the introduction of new and significantly changed services and methods to produce or deliver them, the survey provided a long and rather convoluted definition of innovation. This read: Technological innovations comprise implemented new or significantly improved services and new or significantly improved ways of producing or delivering a service. An innovation has been implemented if it has been introduced on the market or used in producing or delivering services. The service should be new (or significantly improved) to the to enterprise (it does not necessarily have to be new to the enterprise s market). A new or significantly improved service is considered to be a technological innovation when its characteristics and ways of use are either completely new or significantly improved qualitatively or in terms of performance and technologies used. The adoption of a production or delivery method which is characterised by significantly improved performance is also a technological innovation. Such adoption may involve change in equipment, organisation of production or both and may be intended to produce or 19 Minor differences in methodology may have also contributed to the different findings 3-26

28 deliver new or significantly improved services which cannot be produced or delivered using existing production methods or to improve the production or delivery efficiency of existing services. The introduction of a new of significantly improved service or production or delivery method can require the use of radically new technologies or a new combination of existing technologies or new knowledge. The technologies involved are often embodied in new or improved machinery, equipment or software. The new knowledge involved could be the result of research or utilisation of specific skills and competencies. The following changes are not technological innovations if they are not directly related to the introduction of new or significantly improved services or ways of producing or delivering them: Organisational and managerial changes such as the implementation of advanced management techniques, the introduction of significantly changed organizational structures and the implementation of new or substantially changed corporate strategic orientations. The implementation of a quality standard such as ISO Innovation activities are all those steps necessary to develop and implement new or significantly improved methods to produce or deliver services. There are several problematic aspects of this definition, but its sheer length suggests it was probably rarely read and fully considered, with respondents instead basing their answers on their instinctive (and culturally differentiated) understandings of innovation (For a fuller discussion of some of the problems with the CIS approach see Tether, 2001) The difficulty that is being faced in this paper is one of attempting to report results from an instrument that is still in the process of calibration. Differences in innovation behaviour between and within sectors have been concentrated on, partially because the results appear more consistent and hence more robust than the differences between countries. Readers interested in the detailed results pertaining to each country (and each sector within each country) can find these in the report to the European Commission (Tether et al., 2001), which is publicly available and can be provided by the author. Having decided to concentrate on sectoral rather than national patterns of innovation, a methodology was needed to summarise the findings from the national level data with sectoral breakdowns as supplied by Eurostat. Eurostat s sectoral data adjusted the original pattern of response to the survey to make it representative of all enterprises, or innovating enterprises, with 10 or more employees active in each of the various sectors 20. Eurostat also provided, for each sector, 20 Thus Eurostat s unit of analysis is the enterprise, and enterprises with few employees are given the same weight as those with thousands. There are pros and cons to this adjustment see Tether et al. (2001) for a discussion. 3-27

29 averaged findings for the whole of Europe, where this was weighted by the number of relevant firms in each country for the sector in question. This increases the importance of responses from firms in large countries, such as Germany, but also favoured countries in which a broad view of innovation tends to be taken (i.e., those that report a high rate of innovation) and in which the sector is highly disaggregated (i.e., where there are many small firms, rather than a few large firms). This can mean the results for one country dominate Eurostat s average for the whole of Europe, even though that country may have provided only a small proportion of the total response to the survey. For example, according to Eurostat, amongst the 13 European countries that undertook the CIS-2 on services, Germany accounted for 70% of the technical service firms, and over three-quarters of those that innovated (Tether et al., 2001) 21. Unsurprisingly, Eurostat s findings for the whole of Europe are very close to those for Germany. But as Germany only provided about 10% of the responses from the firms in this sector, any peculiarities within the relatively small sample of German responsees will be magnified and exaggerated in the European average This is a remarkable and surprising level of dominance. It may well be false, reflecting differences in definition, for example between enterprises and establishments, rather than genuine national differences. 22 France, despite returning more than 3 times as many responses from technical service firms only contributes about 5% of Eurostat s total weighted average, whilst the Netherlands and Norway, both of which also gathered more returns than Germany from technical service firms, contributed only 2.7% and 1.1% respectively to the weighted average. 3-28