Questioning the implementation of smart specialisation: Regional innovation policy and semi-autonomous regions

Introduction

Smart specialisation has come to dominate the regional economic development and innovation policy agendas across Europe, becoming the key approach to which regions must now conform. The idea of smart specialisation is that regions should choose specific sectors to specialise in based on an intelligent ‘entrepreneurial discovery process’ to identify their strengths, rather than simply copying trends from elsewhere. From its roots as an academic concept, smart specialisation has very rapidly transitioned into a practical policy framework, and research is still catching up (Foray et al., 2011). In essence, smart specialisation is about regions building on their strengths and reducing duplication and competition within Europe, building on past ideas and trends in economic geography and regional policy. Elements of earlier regional innovation system (RIS) thinking are key to smart specialisation, with conceptual additions such as general enabling technologies and the entrepreneurial discovery process (European Commission, 2011; Foray et al., 2011; McCann and Ortega-Argilés, 2014). Previous thinking around key sectors and building an RIS continue to dominate, following a recognised pattern of policy path-dependence (Morgan, 2013).

Through studying the implementation of smart specialisation in Wales, several conceptual shortcomings with the approach are identified. These are traced back to issues with earlier RIS approaches, which have been incorporated into contemporary smart specialisation policy, and if unresolved could lead to the replication of past mistakes. Specifically, the economic, social, historical, political and cultural definitions of the region remain persistently fuzzy, whilst drawing regional boundaries remains particularly difficult. Regional governments lack guidance on balancing an introverted systemic perspective focused on their regional needs with cross-region or cross-border governance and operation necessary to exploit multiple types of proximities and inherently intertwined systems at a European, and indeed global level. The language and discourse of regions is found to be problematic in territories with an active nation-building agenda. It is argued that each of these shortcomings has the potential to undermine the efficiency and appropriateness of the current policy approach in different regional settings across Europe, with particular pertinence for weaker regions disproportionately reliant on European funds and thus obliged to closely follow the smart specialisation agenda. Furthermore, if our policy approaches are not sufficiently interactive and evolutionary in nature to account for rapidly shifting political and economic situations, they risk becoming irrelevant or untenable.

This paper aims to increase the links between innovation theory and policy practice (cf. Martin, 2012; Mytelka and Smith, 2002), and to ensure that theory is fit for use by regional innovation policymakers and remains relevant to their work. This is achieved through an in-depth study of innovation and economic policy in Wales over the last 20 years compromising 58 interviews with key stakeholders in the innovation system from business, government, academia and intermediaries, and analysis of 20 years of policy relating to innovation and economic development. It was found that the guiding principle of RIS has failed to bring about the economic transformation hoped for and four shortcomings are identified with the theoretical foundations of smart specialisation: conceptual ambiguity, problems in defining regional boundaries, issues with the use of the term ‘region’ in a normative fashion, and potential mis-translation and application arising from conceptual ambiguities. Based on these shortcomings identified from the Wales case, concerns are raised over the appropriateness and efficacy of smart specialisation in this setting, and some suggestions are made as to how the policy approach and the theoretical underpinnings therein can be adapted to better suit the diverse range of regions within which it is being applied.

The paper is structured as follows. A theoretical background to the paper is presented outlining the smart specialisation approach and how it has evolved from earlier RIS approaches to become the predominant regional innovation policy framework at the current time. Then, the Wales case and the research conducted is introduced and four identified issues with smart specialisation presented from this case study. The paper concludes with suggestions as to how weaknesses in RIS theory can be addressed, and how in turn the smart specialisation agenda can be rendered more applicable to the varied regions of Europe. The premise of the paper is that if we do not interrogate smart specialisation (and indeed our predominant policy frameworks more generally) in real world regions, we are in danger of prescribing inappropriate policy fixes which do not suit local circumstances, or replicating past mistakes resulting from weaknesses in existing theory.

Theoretical foundations

According to Morgan (2013: 104), contemporary smart specialisation policy is designed in the context of path-dependent processes, notably policy path-dependence, conditioning future regional development prospects; as such, to plan future policy we must understand historical outcomes. Here, recent smart specialisation trends are considered as the latest iteration of concepts key to European policy over the last 20 years. Arguably, the most prominent of these has been the RIS approach, elements of which have been subsumed into smart specialisation such as the key concepts of ‘embeddedness’ and ‘connectedness’ (Camagni and Capello, 2013). Recent work has examined smart specialisation in the context of systemic failures of RIS (Grillitsch, 2016). Indeed, the strategy is articulated as ‘Regional Innovation Strategies for Smart Specialisation’ (RIS3) illustrating how smart specialisation is intended as a new and improved manifestation of early RIS approaches. Smart specialisation in its theoretical form also incorporates elements of new industrial and entrepreneurship policy, but the threads of RIS are certainly central in its current manifestation. Before delving into smart specialisation’s historical evolution, the concept as it currently stands is briefly outlined.

At its core, ‘Smart Specialisation is about R&D and innovation’ and is the key policy approach to innovation and regional economic development in Europe (Foray et al., 2011: 5). Member regions are required to produce RIS3 to bring about ‘integrated and place-based economic transformation’ (European Commission, 2013). The agenda has gained currency across Europe following Innovation Union’s publication (European Commission, 2010), which employs smart specialisation to achieve its goals of ‘smart, sustainable and inclusive growth’. The European Commission (EC) has called on its regions to design RIS3 strategies to maximize the impact of European Union (EU) structural funding in the next round through ‘thematic concentration’ (European Commission, 2011, 2012). Smart specialisation envisages agglomeration at a European scale, allowing different centres to emerge drawing knowledge into the region (European Commission, 2011). This is an apparently ‘simple idea’, based on the understanding that regions need to promote what makes them ‘unique’ and ‘superior’ (Foray et al., 2011: 4). Smart specialisation is the dominant contemporary approach to regional innovation policy in Europe.

Smart specialisation distinguishes itself from the earlier regional innovation approaches in calling for an ‘entrepreneurial discovery process’ to identify the characteristics and assets of the region; an attempt to avoid the trend of regions trying to replicate a limited set of trendy sectors where they do not already exist. It calls for coordination and mobilisation of regional stakeholders around a shared vision based on these existent strengths (European Commission, 2011). However, researchers have traced a conceptual line from RIS to smart specialisation, with common theoretical foundations being recycled and recombined (Foray et al., 2011; McCann and Ortega-Argilés, 2014). McCann and Ortega-Argilés (2013: 3) find that smart specialisation (SS) ‘largely reflects a regional innovation systems logic’. A strength of this is that tried-and-tested best practice and successful approaches can be revived; the danger is that past weaknesses or mistakes do not get resolved or corrected, and become re-packaged into a new approach. Before exploring these dynamics, the evolution of regional innovation policy in Europe from RIS to smart specialisation is discussed, which is the theoretical backdrop against which the analysis below is painted.

The RIS concept evolved as a branch of the wider systems of innovation perspective viewing innovation as an interactive and evolutionary phenomenon (Edquist, 2005; Lundvall, 1992). It gained a high degree of academic momentum and policy relevance during the 1990s and early 2000s through the seminal work of regional scholars (e.g. Cooke et al., 1997). Indeed, RIS became the theoretical underpinning of European innovation and regional policy at that time, with the Regional Technology Plans/Regional Innovation Strategies (RTP/RIS) rolled out across European regions (Asheim, 2012; Landabaso, 1997; Morgan et al., 1999). Wales, a semi-autonomous ‘home nation’ of the UK, was one of the crucibles of the development of the RIS concept through the work of prominent Welsh economic geographers (Cooke, 2004; Cooke and Morgan, 1992; Morgan et al., 1999) and as a pilot region of the original RTP (Landabaso, 1997; Welsh Development Agency, 1998, undated a, undated b). The wider systems of innovation concept has subsequently influenced policy at the regional, national and international level, utilised by the OECD, United Nations and national innovation agencies (Edquist, 2005; Lundvall, 2007). There is a strong precedent for the RIS as a framework for studying and evaluating regional innovation policy efforts and drawing out lessons for policymakers (Asheim, 2012; Cooke et al., 2004; Morgan, 1997; Nauwelaers and Wintjes, 2003).

The essence of systems theory is that interactive processes between varied and diverse actors, networks, continuous leaning processes and innovation-conductive institutions such as policy incentives and trust will give rise to economic growth, technological dynamism and competitiveness (Caniëls and van den Bosch, 2011). This underpinning notion of innovation as interactive and evolutionary (cf. Lundvall, 1992; Nelson, 1993) has been incorporated into smart specialisation, and hence transformed into a policy blueprint for regional governments across Europe.

The regional, local and national variants of the systems approach (Iammarino, 2005: 502; Morgan, 2004: 4) are built on the understanding that ‘innovation is fundamentally a geographical process’ (Rutten and Boekema, 2007: 3). Iammarino (2005: 499) describes an RIS as ‘the localised network of actors and institutions in the public and private sectors whose activities and interactions generate, import, modify and diffuse new technologies within and outside the region’. The theory’s development has historically been marked by a division between regional and national scopes. Yet, researchers are increasingly viewing territorial systems less rigidly, finding ways of understanding multi-level innovation systems and policy approaches that combine these perspectives (Magro and Wilson, 2013). This is feeding through into smart specialisation research as other complex regional arrangements such as cross-border regions (Oliveira, 2015; Van den Broek and Smulders, 2015) are examined.

RIS was elevated to the forefront of regional innovation policy consistent with a trend towards regionalism and the increasing focus on innovation and R&D in Europe over the last 20 years – the right theory, in the right place, at the right time. Regions were being strengthened, re-defined and promoted as important actors through new governance structures and a re-scaling of spaces and meanings (Paasi, 2009). The RIS literature focuses on the innovation-specific aspects of this debate and argues that a regional framework is most appropriate, with learning and knowledge having a specifically regional context and culture (Asheim and Gertler, 2005; Cooke et al., 1997, 2007; Morgan, 1997; Nauwelaers and Wintjes, 2003). Paasi (2009) links up the increasing political influence of regions with regional identity and economic development, explaining how EU cohesion policy, concerned with a redistribution of opportunities in space, was developed to motivate regions to exploit their cultural characteristics, skills and social capital for ultimately economic goals (Faludi, 2007).

Here, it is important to make the distinction between the smart specialisation concept, which is a territorially based idea, and its application at the European level, which has been specifically regional. This subtle yet important shift has meant that a concept that could theoretically be applied at a range of geographical scales (e.g. city, region, nation, etc.) has developed into a strictly regional approach. This paper argues that the reason this co-option of the concept into a regional frame has occurred is because of the strong historical precedent within European policy of taking a regional approach to innovation and economic development and the apparent ease of ‘bolting-on’ the smart specialisation concept to past RIS approaches that were highly familiar and tried and tested across Europe.

Indeed, RIS has demonstrated remarkable staying power as a guiding rationale for European policy. It has been applied in several regional settings; the seminal book on RIS (Braczyk et al., 1998) features contributions from Wales, Catalonia, Baden-Würrtemburg, Ontario and other diverse regions (Bacaria et al., 2004; Braczyk et al., 1998; Gertler and Wolfe, 2004; Heidenreich and Krauss, 2004). RIS could be argued to be a more realistic representation of the complexities and number of different players involved in innovation policy rather than reducing the situation to a tick-list of organisational players (as per the triple helix – Etzkowitz and Leydesdorff, 1997) or a narrow collection of firms in a particular area (as per cluster theory – Porter, 1998).

Despite its prominence as a policy blueprint, conceptual weaknesses with the theory have been highlighted, and this paper is concerned with investigating whether these weaknesses have in turn become absorbed into the smart specialisation approach, and what the consequences of this might be. Many question the centrality of the region and instead argue that other geographical scales of analysis are equally or more important, or that knowledge and innovation are not regionally bounded (Hadjimichalis, 2006; Harrison, 2006, 2013; Lovering, 1999; MacLeod, 2001). Lovering (1999: 380) is especially critical of the ‘new-regionalism’ and uses insights from his study of Wales to contest the two key ideas underpinning the new-regionalist school of thought: that the region has become the ‘crucible of economic development’; and the normative basis that ‘the region’ should be the prime focus of economic policy. Also, it is still unclear how well our theoretical approaches can be translated or transplanted into less favoured circumstances (Cooke, 2004; Gunasekara, 2004; Hospers, 2006; Martin and Sunley, 2003).

There is confusion around the composition, boundaries, causal relationship and measurement aspects of systems (Asheim, 2012; Carlsson et al., 2002). This ‘fuzziness’ (cf. Markusen, 2003) can be problematic as these issues remain largely unresolved as the concept continues to be used primarily as a normative/prescriptive policy tool (Uyarra and Flanagan, 2010). There remains a need for active and ‘broad based’ innovation policies (Flanagan et al., 2011), but a balance must be struck between flexibility and theoretical rigour (Lundvall, 2002, 2003). The fact that RIS is less normative and prescriptive than some other theoretical approaches could be problematic if we conflate its role as a descriptor with that of a best-practice framework. However, its inherent fuzziness lends it well to an application in different circumstances and capabilities. Iammarino (2005) highlighted some time ago that the shortcomings of the RIS theory have prevented a complete integration between concepts and measures hampering further enhancement of the theoretical model, whilst failing to attain a better normative justification for public policy. Nevertheless, the influence of these ideas persist through the smart specialisation agenda.

By examining the issues that smart specialisation has inherited from previous RIS approaches via documented issues of policy path-dependence (cf. Morgan, 2013), this paper highlights potentially problematic unresolved theoretical issues which could weaken its potential and efficacy as a policy blueprint. To investigate, a study of innovation policy in a ‘real world’ regional setting was undertaken, focusing on the design and implementation of regional innovation policy in a weaker and semi-autonomous setting.

Case context and enquiry

Wales was one of the key testing grounds for the RIS in its early formative years (Cooke, 1992; Morgan, 1997), with several important contributors continuing to explore the Wales case study and contribute to theoretical debates on regional innovation (e.g. Bristow, 2005; Cooke and Clifton, 2005; Huggins and Kitagawa, 2012; Lovering, 1999; Morgan, 2012; Pickernell, 2011). Wales’ history of closely following European innovation policy trends, and its complex semi-autonomous status render it an interesting case study for considering the practical strengths and weaknesses of smart specialisation. Wales is classified as a weaker region, in the European and British context, suffering from many of the problems associated with post-industrial regions such as: low competitiveness, high levels of economic inactivity, lack of private sector dynamism, high public sector employment, low skills levels, and few R&D performing firms (Ball, 2008; Cooke and Clifton, 2005; Cooke et al., 2003; Huggins and Thompson, 2010). Wales has the lowest GVA per head, currently at 71.4% of the UK average across all regions (Welsh Government, 2015). Half of Wales – west Wales and the valleys – qualifies for the highest level of support for economic development activities from Europe. The population is around three million.

Wales is distinct politically and culturally from the rest of the UK, and its devolved status provides an interesting case study of innovation systems from a regional policy perspective (Huggins and Kitagawa, 2012: 818). Since devolution, where power was transferred from the central British state, innovation and economic policy has been designed and implemented at the Welsh level, allowing us to adopt a regional lens. However, there are policy areas which have an important bearing on innovation and economic development that are still controlled by the UK government such as foreign policy, defence, tax and regulation. Alongside and intersecting with this complex administrative situation is the cultural context; Wales is part of the UK. However, it is often referred to both within and outside Wales as a ‘country’, has its own language and heritage, and a proud and distinct identity.

Cooke (2004: 4) describes Wales as an ‘institutional regional innovation system’ due to the pro-active policy towards innovation and regional development starting with the RTP/RIS plans during the 1990s (Asheim, 2012; Cooke et al., 2003; Morgan, 1997) and conforming closely to European trends, producing ‘standard EU fare’ designed to absorb European funding (Cooke and Clifton, 2005). In recent years, this has materialised as a smart specialisation premised approach, focussing on four ‘grand challenge’ areas (Welsh Government, 2013) in direct response to the EC’s calls for RIS3 strategies in all regions (European Commission, 2013). In spite of this close adherence to European policy directives, Welsh efforts to develop an innovation economy have not proved particularly successful, and the economy continues to lag, sparking questions about the efficacy and suitability of the historic RIS and contemporary smart specialisation premised approach. The insights into regional innovation policy that can be obtained through studying the Wales case could hold relevance for other regions struggling with problems of peripherality, weak economies, and complex multi-level or semi-autonomous governance situations.

The empirical research consisted of an in-depth review of innovation and related policy and programmes (from 1999 to the present day). Policies of the Welsh Government relating to innovation and economic development (including related areas such as science, higher education and entrepreneurship) were read and analysed for key themes in an inductive fashion. Face-to-face interviews were conducted with 58 key stakeholders in the Welsh innovation system from government, business, university, intermediary and third-sector spheres, with a roughly equal balance of respondents from each. The interviews were semi-structured lasting around 1 hour. Participants were asked to provide their reviews of innovation policies and interventions, with a particular focus on contemporary and future directions for innovation policy in light of smart specialisation. The interviews were analysed using software, using a semi-inductive approach based on looking across the different categories of respondents according to the themes discussed as per the interview pro-formas. Analysis of the policy documents and the interviews was then brought together to examine rhetoric and reality of innovation policy, and to build up a qualitative evaluation of the policies being implemented. There is precedence in the policy studies literature of combining methods thus (Burnham et al., 2004; Garnett and Lynch, 2012; Harrison, 2001). Insights gained from the case study are employed in order to fundamentally question the suitability of regional innovation theories both as a lens for examining weaker regions and as a tool for creating a best-practice plan for innovation policymakers.

Analysis

In this section, the themes that emerged when examining RIS and smart specialisation in the context of Welsh regional innovation policy are discussed. Some of these are already discussed in the literature, in which case the experiences from Wales are highlighting the persistence of thorny theoretical issues and supporting calls to resolve these shortcomings. Other insights are perhaps more specific to the Wales experience but could shed light on issues being faced by policymakers implementing smart specialisation approaches in other similar peripheral and semi-autonomous settings, often overlooked. Analysis is structured around themes, which cover the main barriers or issues that emerged inductively from the empirical research. Rather than being a definitive list of the unresolved issues or weaknesses in theory, these are pertinent issues in Wales currently.

Discussion

Through an empirical study of regional innovation policy in Wales, key issues have been identified in the implementation of regional innovation policy, in particular with reference to the current smart specialisation approach. This discussion sections links these findings from the Wales case back to extant literature on the topic, and also broadens the analysis beyond the Wales case to consider wider implications for theory and practice, and also potential insights for other regional settings beyond this specific case.

It is clear from considering the practicalities of implementing RIS in Wales that boundary definition and knowing what to include in the innovation system is highly problematic. This is especially the case in territories with complex governance situations, such as semi-autonomous or cross-border regions, such as Wales, of which there are several across Europe. The smart specialisation approach and RIS theory underpinning it do not account for places that do not fit neatly into ‘regional’ boundaries and definitions, such as semi-autonomous regions or cross-border regions. Systems of innovation theory is unhelpful in this regard, professing that the boundaries of a system of innovation cannot be sharply determined, or might be impossible to define (Edquist, 2005; Lundvall, 2007). Besides, innovation does not necessarily conform to regional or national boundaries, and proximities other than geographical, such as organisational, cognitive, institutional or social, are important drivers of economic growth and advantage (Boschma, 2005). Smart specialisation provides little guidance as to how regional policymakers build and exploit non-geographic proximities working across regional boundaries.

Little guidance exists in the literature on RIS or smart specialisation to help policymakers to know what to include in their strategies and how. Extant literature has limited insight into this issue: fundamental disagreements exist around how theoretically rigorous and prescriptive about the geographical frame innovation theory should be (e.g. Asheim and Gertler, 2005; Cooke et al., 1997, 2007; Edquist, 2005; Lundvall, 1992, 2002, 2003; Morgan, 1997; Nauwelaers and Wintjes, 2003). Part of the problem lies in the conceptual ambiguity surrounding the term region, well established in the literature (Harrison, 2006; Lovering, 1999; Macleod and Jones, 2007). The boundaries of a region can range from administrative boundaries to social practices and norms, and the various forms regions take – administrative, cultural, economic, governmental and historical – could all have different implications for the applicability of models, depending on which we include in our analysis and allocate the most weight (cf. Keating, 2004; Paasi, 2009). Our policy approaches assume that administrative and system boundaries will be the same, and that policymakers would have agency with regard to both. Clear problems arise in cross-border regions (cf. García-álvarez and Trillo-Santamaría, 2013), and in regions without a clear regional level of government to administer and actualise smart specialisation strategies. In short, innovation systems do not always delineate according to this constructed system of state, national or regional boundaries.

Foray et al. (2012) and Querejeta and Wilson (2013: 13) emphasise the importance of analysis that looks beyond regional boundaries, taking into account specialisation and capacities relative to other regions in Europe, but how this should be done is less clear. According to Morgan (2013: 124), the demands of the smart specialisation process will force regional states to recognise innovation as ‘a collective endeavour in which the capacity to work together may be the most crucial factor separating success from failure’. Learning and evaluation is encouraged at a pan European scale but issues of neighbouring regions, overlapping administrations and cross-border regions are thus far devoted little attention. The lack of dialogue and joint working between policymakers in neighbouring regions is concerning given that innovation systems are most often inherently cross-border. It is a significant oversight of smart specialisation that is does not devote sufficient attention to providing guidance and mechanisms to assist policymakers working with overlapping and complex system boundaries. If this results in a lack of interaction between policymakers and strategies in neighbouring regions, important elements of the innovation system and synergies could be missed by the strategies.

Consideration of smart specialisation implementation in Wales has found a mis-translation of the approach as it is applied in a manner akin to old cluster policy approaches (cf. Pugh, 2014). It is hard to know exactly how ‘standardised’ regional innovation policy approaches should be. The desired flexibility and rigour of the underlying systems of innovation concept remains unresolved; two of the concepts key figures disagree over the extent to which the concept should be rigorously theorised or generic and adaptable enough to suit any regional context (Edquist, 2005; Lundvall, 2007). Nevertheless, an important question that the originators of the approach at the European level need to ask is whether it is problematic that smart specialisation is being mis-interpreted and co-opted at the regional level, or whether this is desirable if it means policymakers with detailed knowledge of their local specificities can adapt and shape central policy edicts to best suit their specific contexts.

Issues of regional culture, identity and institutions all have profound influence on economic development and innovation, and should be seriously considered in any discussion of the innovation system; the economic, political, social and cultural aspects of regions change over time (Paasi, 2009) and correspondingly, our theorisation of them needs to account for this evolution. Despite its roots in evolutionary and interactive approaches, smart specialisation has little guidance as to how transitioning regions should manage their strategies. Wales is a good example of a territory with a strong identity manifested in the political sphere as a nation-building agenda, and is not alone:

A recent survey of Euro-regions found that strong identities are associated with peripheral location, economically weak areas, low level education, and high relative levels of primary employment (Paasi, 2009: 140).

Looking forwards

These criticisms of smart specialisation do not discount the fact that the concept may contain a number of useful ideas and suggestions to help regional policymakers, especially those in less favoured regions; but some re-thinking and re-packaging of the concept is required, and more attention devoted to its unresolved issues to ensure the theoretical frameworks provided for regional innovation policy practice are as sound as they can be. Theory must be able to adapt alongside political, economic and social change taking place; if not it risks becoming irrelevant, or promoting inappropriate actions.

Whilst the fundamental theoretical issues are fundamentally tricky to resolve, the rhetorical elements around our discourses of regional innovation policy are simpler. This paper suggests that re-packaging smart specialisation and being less dogmatic about the regional frame of analysis and implementation could render the concept more palatable and politically tenable for policymakers in semi-autonomous regions. Through creating a framework that is appropriate for the range of regions in which it is being applied, mis-translation and co-option of the framework could be reduced. Some suggestions have been made of alternative terminologies building on RIS, which could be more appropriate for understanding the situation in semi-autonomous regions. One suggestion is the ‘Region-State Innovation System’, which draws upon the term introduced by Ohmae (1993) and utilised by Paasi (2009), which elevates the region alongside the nation state as a central and important economic and political actor. Another is the ‘Autonomous-Community Innovation System’, borrowing the Spanish term for such regions (García-álvarez and Trillo-Santamaría, 2013), which perhaps more respectful and tactful regarding the identity politics surrounding these debates. There are some options to shift our vocabulary to one that is more appreciative of the culture and importance of sub-state territories and escape some of the negative connotations and deference to the nation state associated with the term region.

Our theoretical models fail to account for complex ‘in between’ territories that can be considered neither a nation nor a region, displaying elements of both. Systems of innovation theory has traditionally presented a binary choice between these two options, and little guidance for policymakers needing to incorporate elements of both a national and regional approach and work across different governance levels. Innovation systems theory has failed to find a middle ground that is less dogmatic in taking either a regional or national perspective.

Another suggestion is to incorporate a multi-level approach that does not try to select one geographical level to analyse, but rather examines multiple perspectives. However, there are few attempts in the SI literature to do this. Combining the systems theory with the established concept of multi-level governance (Bache and Flinders, 2004; Hooghe and Marks, 2001) could provide a framework which better accounts for the complexity of the governance of innovation in regions such as Wales. This would allow for flows and interactions between the different levels to be taken into account, and opens up the potential to study a territory such as Wales as an innovation system in its own right, but also interacting with neighbouring systems in the UK. This suggestion has been made by Cooke (2002) but practical attempts to integrate the concepts are thus far lacking, signalling an important area for further development. Magro and Wilson (2013) have made important steps in this regard, and their case for a multi-level approach to regional innovation policy via innovation system development is supported. This could hold promise for semi-autonomous regions (such as the Basque region as investigated by Magro and Wilson, or Wales) but also holds risks of greater complexity and requires testing in a range of settings.

Conclusion

By highlighting pertinent issues encountered when applying smart specialisation in a semi-autonomous region such as Wales, we gain a new perspective on theory by testing its applicability in real world settings not always considered. Identifying theoretical shortcomings in smart specialisation helps us to recognise where our policy blueprints may not be performing ideally, and where some re-thinking of our predominant approaches may be required. It is the premise of this paper that there exist some underlying theoretical issues that smart specialisation has inherited from earlier RIS approaches, and that failure to address these could lead to old mistakes being replicated and sub-optimal policy outcomes. Theoretical tightening of the concept, which has very rapidly evolved as a policy framework, could help to mitigate problems arising from misunderstanding or co-option of the concept. There are questions to be asked about whether continuing to pursue an RIS-premised policy approach at the European level is in danger of alienating policymakers and publics in such places. It is difficult to say at this stage how successful smart specialisation has proved because the strategies are currently being implemented; as such, we can reflect on observations as the agenda progresses and react to issues as they arise.

The European application of smart specialisation has drawn too heavily on past RIS elements and sidelined some of the other influences (for example from industrial and entrepreneurship policy) that originally existed. It is suggested that a return to the theory and a re-consideration is required, which softens some of the RIS elements. In its original conceptualisation, SS was merely territorial (not dogmatically regional) and more consideration of how it could work in alternative territorial settings (such as cities and city regions, cross-border regions) is needed. Due to the fuzziness of the concept, and also its similarity to older RIS- and cluster-based approaches, it has been found to be easily co-opted or misapplied on the ground. More guidance and support could be provided for policymakers to work with neighbouring regions, and also with different levels of administration, in order that strategies can be designed in a more synergistic and complementary manner. Failure to resolve issues with smart specialisation as they emerge and to adjust the concept as it is applied in different regions could lead to the concept becoming quickly outdated and no longer fit for purpose.

Questions are raised about the political tenability and the wider appropriateness of continuing to pursue a strictly regional agenda in innovation policymaking in the Europe of today. Whilst this paper speaks to the situation in Wales and the UK, there is a wider relevance to other semi-autonomous regions in Europe which have also been pursuing nation-building agendas. It is important to consider how smart specialisation interacts with these cultural and political processes taking place. There are several regions, such as Wales, where this discourse is no longer appropriate, and it is suggested that the European Commission reconsider its keenness to promote regional frameworks in the current political context. Considering the precarity of European nation states, and the wider European project at present, it seems unwise to alienate or inflame semi-autonomous territories through something as easily remedied as outdated and inappropriate language.