2.1. Assimilation studies

Keith Pavitt (Reference Pavitt1984) can be considered the forerunner of the so-called ‘assimilation’ approach to service innovation. His classification of technological trajectoriesFootnote ² across sectors included services in the ‘passive adopters’ category. The underlying conjecture was that opportunities for innovation in services are restricted to the mere adoption of new technologies developed in the capital goods sector. This is based on the observation that many service activities are heavy users of information and communication technologies (ICTs) available ‘off the shelf’. In this approach, services are considered – paraphrasing Teece (Reference Teece1986) – more like specialized assets; that is, activities subordinated to manufacturing. A few years later Barras (Reference Barras1986, Reference Barras1990) went further down this road with a study on innovation in the financial sector. Reversing the product cycle theory (Abernathy and Utterback, Reference Abernathy and Utterback1978), he argued that adoption of new technologies is followed by quality improvements and, eventually, by new services – referred to as ‘user sector’. According to many, this study of ICTs adoption overlooks relevant aspects such as organizational and institutional changes, which were crucial complements to ICT in financial services’ innovation (Uchupalanan, Reference Uchupalanan2000; den Hertog, Reference den Hertog2000; Consoli, Reference Consoli2005a). A broader source of criticism is that Barras sought to generalize his model indiscriminately to all service activities (Tether et al., 1997; Sundbo, Reference Sundbo1998). Subsequently, Soete, and Miozzo (Reference Soete and Miozzo1989, Reference Soete and Miozzo2001) took the taxonomic approach a step further with a classification of services structured according to functional criteria. They singled out four categories of activities:

1. (i) Production intensive, scale intensive, and network services (e.g. banks and telecommunication services);

2. (ii) transport and travel services, wholesale trades, and distribution;

3. (iii) specialized technology suppliers and science-based services (e.g. software and business services);

4. (iv) public and social services, personal services, and retail trade (e.g. health-care services).

Soete and Miozzo's analysis unpacked the functional characteristics of service activities and offered an undoubtedly richer picture of the variety of innovation processes involved. In so doing, they also undermined Pavitt's conjecture that all services are supplier dominated, as this applies only to one category of the new taxonomy, namely the last. Summing up, taxonomic models deserve credit for the very effective use of the data available at the time, and for the systematic categorization of a wide range of economic activities. Even so, the view they offer of economic systems as aggregates of static sectors leaves key questions open, particularly those concerning the possibility that new activities or new agents emerge, bearing influence on the structure of the system.

2.2. Demarcation studies

The criticisms raised towards taxonomic models paved the way to a new body of work in the second half of the 1990s. According to the main tenet of the ‘demarcation’ approach, specific features such as intangibility and co-determination confer uniqueness to services. This point was made with clarity by Gallouj and Weinstein (Reference Gallouj and Weinstein1997) who revised the Lancastrian model by Saviotti and Metcalfe (Reference Saviotti and Metcalfe1984) to argue that the combination of characteristics embodied in a service is unique because its output is intangible. This implies that, unlike manufactured products, a service outcome can never be replicated exactly, no matter how many times it is reproduced. The argument goes even further when customization induced by user–producer is considered (Gadrey and Gallouj, Reference Gadrey and Gallouj1998; Sundbo and Gallouj, Reference Sundbo, Gallouj, Metcalfe and Miles2000; Gallouj, Reference Gallouj2002). As service firms often customize their output around customers’ requirements, in principle any variant in the mode of service provision can be considered an innovation.

It seems clear that striving to portray services as ‘different’ led demarcation studies to take arguably too far an analysis built upon a selective view of physical characteristics (or lack thereof) of service activities. In particular, they aroused two sources of criticism. First, the association between services and manufacturing is downplayed from the outset. As will be seen, the benefits of such a divide are debatable when considering that empirical studies rather indicate complementarity between the two (Antonelli, Reference Antonelli1998; Howells, Reference Howells2004). Second, the conjectured flexibility of services due to user–producer interactions is a generalization that would require further qualification (Tether, Reference Tether2005). Though empirical evidence underscores the role of demand in shaping innovation in the service sector (European Commission, 2004: 24), provider–user interactions are generally rather heterogeneous, and their coordination implies considerable costs of gathering highly dispersed information (von Hippel, Reference von Hippel2005). This suggests a cautious consideration of the variety of user–producer interactions and of the impact that these exert, depending on the content and the mode of service provision. Miles (Reference Miles, Fagerberg, Mowery and Nelson2005) makes this point clearly: activities such as Knowledge Intensive Business Services (KIBS), which consist in the exchange of information, are more easily influenced by users’ feedback than other types of services where such feedback is indirect or not relevant at all.

The picture portrayed by these two approaches on service innovation is a cause for discomfort for manifold reasons. First, most studies looked at static characteristics of services, insisting particularly on a number of features (Miles, Reference Miles2000, Reference Miles, Fagerberg, Mowery and Nelson2005): the outcomes of services are intangible and, thus, do not undergo processes of storage and transportation like manufactured goods (co-substantiality); as the acts of production and use (often) coincide, many services involve some form of cooperation between users and providers (co-terminality); most service activities are information ntensive, although the role of information can vary enormously across them. In sum, most of these studies focussed on what services ‘are’ within given sectoral boundaries, rather than on what they do and how. Second, and as a consequence of the former, the tendency to lump together service activities in a homogeneous clod contributed to hide the considerable heterogeneity. The latter applies to the markets they serve, their modes of organization, and the institutional framework in which they emerge and develop (Tether, Reference Tether2003; Miles, Reference Miles, Fagerberg, Mowery and Nelson2005). Service industries are characterized by greater variety of skills and greater intensity in the use of technology than had been previously perceived (OECD/Eurostat, 1997). Yet one more reason for dissatisfaction is that early studies on ‘service innovation’ assimilated concepts and tools of the analysis of technological change that had been developed in relation to manufacturing. As we have seen in the review of the literature, this gave way to extreme views of services as either subordinate or totally distinct from manufacturing.

The view we take here is that services produce value by generating transformations over time and space, the outcome of which can be either physical (e.g. a haircut, a surgery operation) or immaterial (e.g. consultancy, an online bank statement) (Miles, Reference Miles1996; Tether and Metcalfe, Reference Metcalfe, Tether and Malerba2004). Empirical works increasingly suggest that innovation in services involves re-organization of the set of activities underpinning their systems of provision. This implies that, first, such a process cannot be detached by the institutional conditions of access to and use of knowledge, and, second, that not all services are relevant to the innovation process, or at least not to the same degree. Much attention has been paid in recent years to services that are organized in network structures spanning various upstream and downstream markets, thereby defying traditional sectoral boundaries. These systems are characterized by distributed processes of innovation in which organizational, strategic, and relational factors come to play a key role, beyond the archetypal ‘product and process’ umbrella (Howells, Reference Howells, Andersen, Howells, Hull, Miles and Roberts2000; Metcalfe and Tether, Reference Metcalfe, Tether and Malerba2004; Miles, Reference Miles, Fagerberg, Mowery and Nelson2005). Growing empirical evidence points out that their mode of operation is shaped by the emergence of platforms of activities stretching across sectors and industries, both within services as well as between services and manufacturing (Antonelli, Reference Antonelli1998; Howells, Reference Howells, Andersen, Howells, Hull, Miles and Roberts2000, Reference Howells2004).

These observations provide ground for a new conceptual perspective that departs from the dichotomies that characterized the early literature on service innovation. In the next section, we present the ‘synthesis’ approach, which emerged recently in the context of innovation studies as a promising alternative to pursue this line of research.

3.1. The conceptual elements

The synthesis approach to service innovation blends the theoretical elements from innovation studies of Neo-Schumpeterian tradition with empirical analysis drawn from detailed case studies (Coombs and Miles, Reference Coombs, Miles, Metcalfe and Miles2000; Tether, Reference Tether2003; Drejer, Reference Drejer2004). Central to this approach is the concept of system of innovation, which is the logical structure that connects the purpose for which a system exists to the set of activities that are set out to achieve it. Innovation in this context is understood as an endogenous response to the limitations that make the current set of activities inadequate for such a purpose (Andersen et al., Reference Andersen, Metcalfe, Tether, Metcalfe and Miles2000; Tether and Metcalfe, Reference Metcalfe, Tether and Malerba2004). Such changes generate a path of problems and solutions through the application of new knowledge and the reorganization of the activities, which shift the system's purpose in an open-ended fashion (Metcalfe and Ramlogan, Reference Metcalfe and Ramlogan2005). Allied to this is the idea that innovation involves reconfiguration of knowledge across partially connected technical, organizational, and institutional domains (Langlois, Reference Langlois1988; Loasby, Reference Loasby1991; Metcalfe, Reference Metcalfe2002). But as knowledge is heterogeneously dispersed across individuals (Hayek, 1945), changes in its structure entail reconfiguration of the mechanisms that make communication possible and effective (Nelson, Reference Nelson2003; Metcalfe and Ramlogan, Reference Metcalfe and Ramlogan2005). Therefore, as put aptly by Loasby (Reference Loasby1991), a system thrives on the diversity of the forms of specialization within, but, at the same time, requires coherence through coordination across the activities in which knowledge is embodied.

This notion of coordination as engine of economic development can be traced back to a seminal article by Richardson in which he claims (Reference Richardson1972: 888; emphasis added):

This understanding of industry as an economic environment in which sets of organized activities are the unit of analysis shares common grounds with the system perspective elaborated in the context of synthesis studies. Both approaches emphasize that such environments are dynamic in a two-fold sense. First, they span an ‘indefinitely’ large number of activities and are, therefore, open to discovery (Klein, Reference Klein1977; Loasby, Reference Loasby1991). Second, and related to the former, such activities are organized around a purpose – ‘discovery’ according to Richardson – the achievement of which depends on the division of labour through specialization. The approach developed in the context of synthesis studies however is necessarily broader, for three reasons. The first is that services involve both physical and non-physical transformations. The second is that the development of most systems of service provision entails the creation of platforms of activities stretching across sectors and industries. In so doing, the object of analysis of synthesis studies becomes the process through which such systems develop, which is obviously beyond Richardson's scope. The third, related to the former, is that the search for and the implementation of solutions is uncertain because it depends on adaptation of the institutional architectures that determine the conditions of access to and use of knowledge (Tether and Metcalfe, Reference Metcalfe, Tether and Malerba2004; Antonelli, Reference Antonelli2005).

Summing up, synthesis studies place the dynamics of service activities within a broader conceptual frame. In particular, they emphasize the interplay among institutional conditions and the interactive processes in the creation and application of knowledge. Four complementary elements are central to this approach:

• • the context, consisting in the institutional structure and set of instituted practices as defined by a specific historical pattern;

• • the emergence of a problem, or some type of limitation in the existing system;

• • solution design, involving new knowledge and ranging from advances in basic science to engineering physical artefacts; and finally

• • The Implementation of the solution, which requires coordination of various types of knowledge with the elements of a system. This is effectively the connection of various domains like technologies, organizations, and institutions.

We will now present three case studies developed in the context of the synthesis approach.

3.2. Empirical evidence from cross-sectoral case studies on service innovation

In this section we review case studies on service innovation in three areas: air traffic control, retail banking, and health care. The criteria for their inclusion are threefold. First, we hold them representative of a wider typology of services organized around network structures which have come to play a strong role in the innovation process (Howells, Reference Howells, Andersen, Howells, Hull, Miles and Roberts2000). Second, this selection consists of activities that fall into three (out of four) service macro-sectors according to standard classifications for accounting purposes.Footnote ³ In so doing, the paths of development of the systems are coherent with the characteristics that their contexts bear. Third, the evolution of these systems of service provision is characterized by interactions cutting across different, yet progressively complementary, sectors and industries. This, in turn, draws attention towards the development of various forms of coordination across the associated technologies, organizations, and institutions.

1. (i) Air traffic control: The first work by Tether and Metcalfe (Reference Metcalfe2003) looks at the incidence of procedural innovation in air traffic management, a service consisting in directing aircrafts during landing and take-off procedures. The biggest provider in the UK is National Air Traffic Services Ltd. (NATS), which is organized as a network made of four control centres distributed across the country. The case study takes steps from data indicating the growing pressure exerted by air traffic on major European airports'Footnote ⁴ operating procedures. The obvious solution of increasing capacity to accommodate the increased traffic volumes (i.e. the sheer volume of departures and arrivals) through the construction of additional runways was ruled out in some cases, including London Heathrow, on environmental grounds. Nevertheless, the study shows that greater efficiency was achieved without expansion in the size of the physical facilities. The example of Heathrow airport is paradigmatic: despite the runway infrastructure having been in place since the 1960s, data show that the airport not only avoided saturation but also managed to achieve better capacity management. A key ingredient of this successful recipe was a mix of physical and procedural improvements. New technologies such as radar systems or recent aircraft technologies, as well as changes in the layout of the runways, fall in the former group. But a form of innovation that proved crucial to enhance efficiency was the institution of new rules for traffic management that have been implemented through cooperationFootnote ⁵ between pilots and air traffic controllers (Tether and Metcalfe, Reference Steedman2003: 449). This case study provides important insights into the development of dynamic capabilities through the organization of activities and the correlation of different forms of knowledge. In particular, it demonstrates that capacity growth does not necessarily require size expansion when efficient reconfiguration of existing practices can suffice. This is referred to as the ‘soft’ side of innovation in services (Tether, Reference Tether2005).

2. (ii) Retail banking: The second case study looks at the experience of British banking, a sector that has been radically transformed during the last three decades (Consoli, Reference Consoli2005a). A number of important steps contributed to shape this pattern of industrial development. In the beginning, limitations in the ability to manage large information datasets stimulated the automation of existing practices in banking, predominantly in the US and the UK, between the 1950s and the 1960s. The process of automation triggered a sequence of problems and solutions that proved crucial in shaping the design of banking services. The first step was to implement processors to re-organize back-office operations, soon followed by the development of new front-office channels such as automated teller machines (ATMs). Over the years, however, the process of automation of automated transactions became increasingly complex and involved cooperation among banks (Consoli, Reference Consoli2005b, Reference Clark2005d), between banks and firms with ICTs expertise (Child and Loveridge, Reference Child and Loveridge1990; Fincham et al., Reference Fincham, Fleck, Procter, Scarbrough, Tierney and Williams1994), and more recently between banks and retailers (Howells and Hine, Reference Howells and Hine1993). But the evolution of banking is a systemic process that could have not been possible without a new regulatory framework, which was a catalyst for competition,Footnote ⁶ and without the impact of consumers’ adoption of new payment instruments, which operated effectively as a selection mechanism. Demand feedback determined the success or the failure of some types of automated banking services and undermined the notion that massive investments in ICTs lead de ipso facto to lower operative costs (Chiesa and Manzini, Reference Chiesa and Manzini1998). Despite the removal of time and space constraints, adoption of banking services depends on the uneven distribution of capabilities across customers (Consoli, Reference Consoli2005b, Reference Consoli2005c; Corrocher, Reference Consoli2006).Footnote ⁷

3. (iii) Health care: Three case studies in different areas of health care provide interesting insights on the emergence of systems of medical innovation.

The first study on ophthalmology tracks the development of diagnostic techniques for glaucoma (Consoli et al., Reference Consoli, McMeekin, Ramlogan, Mina, Tampubolon and Metcalfe2005). Without delving into technical details, suffice it say that for many years diagnosis of glaucoma was based on measurements of intra-ocular pressure (IOP).Footnote ⁸ This lock-in into what would turn out to be the ‘wrong’ scientific paradigm offers a clear example of how, paraphrasing Loasby (Reference Loasby1998), the way in which the science base defines a problem forges the range of techniques, for good or for bad. An important breakthrough for research on glaucoma was the dismantling of the IOP conjecture, which is indeed insufficient for a correct diagnosis. More than this, the discovery that glaucoma is not a single disease but rather a heterogeneous group of disorders provided solid ground for the design of new techniques and the development of associated diagnostic procedures through the institution of various forms of cooperation. On the one hand, information exchanges fostered through scientific publications and informal networks of practice provided much scope for the individuation of new areas of research. The evolution of scientific understanding of glaucoma is mirrored also by qualitative changes in the channels of communication. The bibliometric analysis by Consoli et al. (Reference Consoli, McMeekin, Ramlogan, Mina, Tampubolon and Metcalfe2005) shows that other than traditional ophthalmology outlets, scientific publications on glaucoma have been increasingly featured in journals concerned with the wider field of molecular genetics. On the other hand, greater resonance within the scientific community stimulated the institution of units of cross-disciplinary research such as university-based research hospitals. The latter represent a pivotal organizational innovation in which laboratory research is blended with the provision of health-care services. In other words, research hospitals bridge the traditionally separated spheres of basic science and technology. In recent years these have become key nodes for the coordination of design improvement and testing of new diagnostic and therapeutic procedures in ophthalmology as well as in other areas of health care (Mina et al., Reference Mina, Ramlogan, Tampubolon and Metcalfe2005).

The second study on medical innovation focuses on the development of a surgical technique for the substitution of the intra-ocular-lens (IOL) in patients with cataracts, a condition in which the crystalline lens of the eye becomes opaque. Metcalfe et al. (Reference Metcalfe, James and Mina2005) analyse the process of radical innovation that shaped the technique of replacing the cataract with a functioning IOL. Initially the achievement of an individual, this breakthrough provided the basis for a system of service delivery the creation and development of which relied upon coordinated interactions across different domains. Various forms of knowledge allowed the translation of the scientific discovery into standard ambulatory procedure. First, advances in basic science allowed smaller incisions in surgical procedure. Second, market knowledge paved the way for commercialization of foldable lenses, whose manufacturing had been perfected through experimentation with new materials and designs.

The third study focuses on the evolution of the medical knowledge on Percutaneous Translumimal Coronary Angioplasty (PTCA) in cardiology (Mina et al., Reference Mina, Tampubolon, Ramlogan, Venetucci, McMeekin and Metcalfe2004, Reference Mina, Ramlogan, Tampubolon and Metcalfe2005). PTCA is a procedure which has established itself as a standard surgical treatment for patients with constricted coronary arteries. The problem and the theorized solution are rather intuitive. As the flow of blood in the artery is constricted by a plaque, it is necessary to dilate it by using a balloon-tipped catheter in order to restore normal circulation. PTCA, however, managed to achieve wide diffusion only after years of cumulative trial and error and systematic interactions with other areas of vascular surgery, as well as diagnostic services (i.e. angiography). A key role in this process was played, again, by the coordination of various types of knowledge. Technical knowledge was necessary to contrast unforeseen problems such as the discovery that arteries tend to block again after the insertion of catheters. This, in turn, needed to be complemented by market knowledge in the large-scale commercialization of stents,Footnote ⁹ which are commonly used to support the wall of the artery. Similar to the other cases, PTCA became a standard ambulatory practice only after a process of institutional adaptation, which required implementation of informal and formal types of regulation.