1

Innovation for Growth – i4g

Policy Brief N° 8

Smart specialisation and the New Industrial Policy agenda

Dominique Foray

2

1 Introduction

The policy concept of smart specialisation has enjoyed a short but very exciting life! Elaborated by a group

of innovation scholars in 20091, it very quickly made a significant impact on the policy audience, particularly

in Europe. The concept is now a key element of the EU 2020 innovation plan2 - the Commission has decided

to build a platform of services (S3) to support regions in their efforts to devise and implement a smart

specialisation strategy3. Moreover, in Annex IV of the general SF draft regulation, smart specialisation is set

as an ex ante conditionality for two thematic objectives of the future Cohesion Policy (R&I target and ICT

target)4; and other international institutions (OECD, the World Bank) are launching activities for promoting

and measuring smart specialisation5.

The recent rapid success of the term “smart specialisation” is a very pleasing result for the academics at the

origin of the concept. But it is also a perfect example of “policy running ahead of theory”6. While smart

specialisation seems to be already a policy hit and policy makers have become actively engaged in

formulating smart specialisation strategies, the concept is not yet tight – it lacks transparency, verifiability

and broad consensus. Many statements and arguments about smart specialisation are not yet based on

sound empirical foundations, advocacy in favour of smart specialisation as a policy and development of the

tools and instruments to implement a smart specialisation strategy may reflect wishes and hopes or worse,

opportunistic special pleading, rather than a robust and defensible strategic case for action. There is, in

short, a worrisomely growing gap between the policy practice and the theory.

The proposed paper aims at articulate a coherent vision of the policy approach that is evoked by that term

and then explore and elaborate the requirements and implications that are consistent with giving

operational content to that conceptualisation.

1 - D. Foray, P.A. David and B. Hall, “Smart specialisation: the concept”, in Knowledge for Growth: Prospects for

science, technology and innovation, Report, EUR 24047, European Union, 2009; D. Foray, P. A. David and B. Hall,

“Smart specialisation: from academic idea to political instruments, the surprising career of a concept and the

difficulties involved in its implementation”, Working Paper series, 2011-01, Management of Technology and

Entrepreneurship Institute, EPFL, 2011

2 - see Europe 2020 Flagship Initiative Innovation Union: Transforming Europe for a post-crisis world, Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, European Commission, COM(2010) 3 - see Regional Policy contributing to smart growth in Europe 2020, Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, SEC, 2010, 1183. 4 - See Fact Sheet: Regional Innovation Strategies for Smart Specialisation, DG Regional Policy, European Commission, 2011. 5 - see Comparative advantage through “smart” knowledge-based specialization: implications for science, technology and industry policies, Working Party on Innovation and Technology Policy, DSTI, OECD, 2011; Research and innovation for smart specialization strategy, The World Bank, draft, 16 June 2012 6 - See W.E. Steinmueller, “Economics of Technology Policy”, in Hall and Rosenberg (eds.), Handbook in Economics of innovation, vol.2, North-Holland, 2010.

3

2 – The economic fundamentals of smart specialisation

2.1 – Towards definitions

Smart specialisation is not a new word for regional innovation policy as a whole. In particular it does not

involve horizontal measures and neutral policy aiming at improving general framework conditions and

capabilities (good universities, human capital, intellectual property rights, research and ICT infrastructure,

competition and openness, and so on)7. Smart specialisation deals with a more vertical and non neutral

logic of intervention. The process of identification and selection of desirable areas for intervention is about

some technologies, fields, sub-systems that could be favoured.

And the difficult policy challenge here is to emphasize such a vertical logic of prioritization while avoiding

the government failures usually associated with top-down and centralized bureaucratic process of

technology choices and selection. Vertical prioritization is difficult; this is why smart specialisation is about

defining a method to help policy makers to identify desirable areas for innovation policy intervention. In

short, it is a policy approach that involves letting and helping the regional economy to discover new

activities with strong potential; making a sound analysis of potential and defining a process which will

empower those actors most capable of realising the potential.

How to prioritize and favour some R&D and technological activities, some sub-systems or some fields while

not dissipating the extraordinary power of market-driven resource allocation in boosting decentralized

entrepreneurial experiments? This question is at the top of the agenda of the so-called “New Industrial

Policy” and the answer lies in inventing intelligently designed policies8.

This distinction between horizontal and vertical or neutral and non neutral is needed for two reasons. The

first one is that horizontal policies might be difficult to do but the identification of what to do – the

domains of intervention – is not so difficult (everybody knows about the direct and indirect conditions to

foster innovation). In contrast, the identification of desirable areas of intervention in a vertical logic – what

technology, what group of firms – is extremely difficult and so a vertical policy such as smart specialisation

needs to put a strong emphasize on the process and procedures of identification and choices and on the

design of new policy instruments (observation and evaluation systems). The other reason to make the

distinction is simply that vertical prioritization; i.e. concentrating resources and getting focused on some

specific technological fields and group of firms that are present and activities in such field – is important. It

is important because even in the information age, the logic of specialisation is intact. Scale, scope and

spillovers are important determinant of R&D and other innovation-related activities and the ability to

realize economies of scale and scope and to capture spillovers is strongly conditional to size. Significant

returns to size in R&D are empirically identified in numerous academic papers (e.g. NBER empirical works

on firms’size diversity, Anchor Tenant, scale, scope and spillovers)9. All this empirical evidence based on

7 - A neutral policy is a policy that does not select projects according to preferred fields or any such criteria, but

responds to demand that arises spontaneously from industry (definition taken from M.Trajtenberg, “Government

support for commercial R&D: lessons from the Israeli experience”, Innovation Policy and the Economy, vol.2, 2002).

8 - D. Rodrik, Industrial policy for the twenty-first century, CEPR, Discussion paper Series, n°4767, November 2004

9 - See for example: R.Henderson and I.Cockburn, “Scale, scope and spillovers: the determinant of research

productivity in drug discovery”, The RAND Journal of Economics, vol.27, n°1, 1996; A.Agrawal, I.Cockburn and

A.Oettl, “Innovation and the firm size diversity hypothesis”, Draft, 2010; A.Agrawal and I.Cockburn, University

research, industrial R&D and the anchor tenant hypothesis”, Draft, 2002, M.Trajtenberg, 2002, op.cit.

4

different methods and illustrating various dimensions of inventive and innovative activities, says the same

thing: there are substantial indivisibilities in knowledge production at both micro and macro-levels. Gains

from specialisation are central in R&D; even the ability to capture knowledge spillovers generated by others

depends as well on the existence of a sufficiently large nearby R&D sector.

Small is not necessarily more beautiful in the information age. If you are small, you are not in a good

position to benefit from these returns to size and so you have to be smarter: concentration of resources in

a few domains, focus of efforts in order to generate these size effects (scale, scope and spillovers) that you

will not get if you do a little of everything10.

It is also clear that focusing and concentrating ressources in a limited number of activities (scale, scope and

spillovers rationale) is not enough; regions should do it by developing distinctive and original areas of

specialisation. “They need to particularize themselves”!11

2.2 – On the process and procedures of smart specialisation

The central insight of smart specialisation is that beyond the horizontal programs that are essential to

improve framework conditions and general capabilities, it is crucial to prioritise, concentrating resources in

specially-selected domains dealing with a particular kind of technology, field, disciplines and sub-systems

within a sector or at the interstices of different sectors.

Activities - that

i) show potentials - they are new, aim at experimenting and discovering technological and

market opportunities and have the potentials to provide learning spillovers to others in the

economy – and;

ii) have scale and agglomeration economies or produce the characteristics of coordination failures

(profitable activities can fail to develop unless upstream and downstream investments are

made simultaneously)

-are natural candidates for prioritization. However principles i) and ii) are very general principles and

identifying new activities as priorities in the real life is not trivial.

Prioritizing some technologies or domains always entail a risk because this implies guessing future

development of technologies and markets. Business as usual strategies to minimize these risks are of two

sorts:

- “cafe para todos” (!)

- Imitating the other region

Smart specialisation suggests a new approach to minimize these risks. This new approach is based on five

principles.

Entrepreneurial discovery and entry

10 Same line of argument by Rodriguez Pose (2001) quoted in “Structural funds for innovation and growth” (I4G internal paper)

11 - Oral communication, Paul David, Knowledge for Growth meeting

5

Smart specialisation involves a self-discovery or entrepreneurial discovery process12 that reveals what a

country or region (will) does best in terms of R&D and innovation. There is always an element of bet and

risk in any policy aiming at identifying and prioritizing firms, technologies or sectors to be supported; and

the best bet is entrepreneurial trial and error. Priorities will be identified where and when opportunities

are discovered by entrepreneurs. Prioritization is no longer the role of the omniscient planner but involves

an interactive process, in which the private sector is discovering and producing information about new

activities, and the government assesses potential and then empowers those actors more capable of

realizing the potential.

This principle allows introducing a clear-cut distinction between the smart specialisation approach and

older policy style that involved centralised or indicative planning methods for identifying industrial

development priorities. These old approaches to the problem of prioritisation and resource concentration

involved formal exercises based on rational and robust theories (inter-sectoral matrixes, technological

interdependencies and hierarchical structures, technological complexities). They were, however, by their

very nature, driven by pre-conceptions regarding industrial priorities and technological opportunities. Such

approaches, which claimed to be very scientific and rational in their ways of identifying priorities, targets

and objectives, were actually often very naïve because they excluded an essential knowledge for success -

entrepreneurial knowledge13.

Entrepreneurs in the broadest sense (innovative firms, research leaders in higher education institutions,

independent inventors and innovators) are in the best position to discover the domains of R&D and

innovation in which a region is likely to excel given its existing capabilities and productive assets. This

principle is so important that any model that did not include this provision would have an entirely different

character.

While entrepreneurial discovery dates the opening of exploitation opportunities, entry constitutes the

confirmation that others see this discovery as meaningful. When the initial experiment and discovery are

successful and diffused, other agents are induced to shift investments away from older domains with less

potential for growth than the new one.

Imitative entry is a key ingredient of smart specialisation so that agglomeration externalities can be

realized: the discovery of a potential domain in which a region could become a leader should very quickly

result in multiple entrants to the new activity. This is the onset of the clustering phase of a smart

specialization process.

Prioritization concerns ‘activities’ not sectors

The response to the big question “what are my priorities?” is not given at the sector level (i.e. agriculture

versus mechanical engineering) nor at the individual level of companies. Sectoral level prioritization is what

12 - The notion of “entrepreneurial discovery” used in the smart specialisation framework draws on works in development economics; in particular Hausman and Rodrik’s view of development as “a self-discovery process”; see R. Haussmann and D. Rodrik, “Economic development as self-discovery”, Journal of Development Economics, vol.72, December 2003. 13 Entrepreneurial knowledge involves much more than knowledge about science and techniques. Rather, it combines and relates such knowledge about science, technology and engineering with knowledge of market growth potential, potential competitors as well as the whole set of inputs and services required for launching a new activity.

6

the old fashioned industrial policy did, based on a very weak and controversial rationale. In case of smart

specialisation, priorities are identified at the level of “new activities”.

For example, think of the case of companies exploring the potentials of nano tech to improve the

operational efficiency of the pulp & paper industry (Finland). In such case, the priority is not the pulp and

paper sector as a whole but the activity involving the development of nanotech applications for the pulp

and paper industry. In the case of plastics firms exploring diversification from the car industry to biomedical

innovations (Basque Country), what should be prioritized is not the plastic industry as such but the activity

of exploring diversification opportunities towards biomedical applications. In the case of automotive

subcontractors exploring diversification towards new sectors (British Midland), again what should be

prioritized is not the whole sub-contracting sector but the activity of exploring a transition path from the

car industry towards new markets.14

These cases above describe entrepreneurial explorations, experiments and discoveries (not simple

innovation) which are about innovational complementarities between a general purpose technology15 (or a

key enabling technology) application and a traditional sector (case of the pulp and paper) or about a

transition path from an existing set of collective capabilities to the foundations of a new business or about

potential economies of scope between two different activities. Such discoveries have the potentials to

generate learning spillovers to the rest of the economy about the value of the new activity. Governments

should support initiatives like these to help the considered activities to grow, through measures such as

solving coordination problems, securing key suppliers, attracting service providers and other firms.

In doing so, what the government is supporting is neither the whole sector nor one single firm but the

growth of a new activity. This achieves two things: it (indirectly) improves the general performance of the

sector; while at the same time building capabilities and expanding the knowledge base towards some new

fields (i.e. the development of applications of nano/bio etc.)

In most cases the discovery process bridges present with future strengths of the economy

In general what is discovered as future priorities are those activities where innovative projects complement

existing productive assets. The pulp & paper/nanotechnology case exemplifies a process of modernization

of a traditional industry. The plastics/medtech case exemplifies a process of diversification or transition

from an existing set of capabilities to a new business. All these cases are involving the generation of related

variety16

The outcome of the process is thus much more than a “simple” technological innovation but rather a

structural evolution of the whole regional economy. Indeed, the entrepreneurial discovery that drives the

14 - These examples come from the following case studies: T.Nikulainen, “Open innovation and nanotechnology: an

opportunity for traditional industries”, draft, April 2008; M.Navarro, M.J.Aranguren and E.Magro Montero, “Smart

specialisation strategies: the case of Basque Country”, Orkestra WPseries, 2011-R07; D.Bailey and S. Mac Neill, “The

Rover task force: a case study in proactive and reactive policy intervention?”, draft -

15 - T. Bresnahan, “General Purpose Technologies”, in Hall and Rosenberg (eds.), Handbook in Economics of Innovation, vol.2, North-Holland, 2010

16 - See K. Frenken, F. Van Oort & T. Verburg (2007), “Related Variety, Unrelated Variety and Regional Economic

Growth”, Regional Studies, 41:5, 685-697

7

process of smart specialisation is not simply the advent of an innovation but the deployment and variation

of innovative ideas in a specialized area that generates knowledge about the future economic value of a

possible structural change. Structural changes as the main outcome of a smart specialisation process

invariably involve some kind of related diversification, a process that builds upon existing capabilities and

industrial knowledge and that is animated by the development of R&D and innovation activities. In other

words, structural evolution is an accumulative process that bridges present with future strengths of a

regional economy in a particular domain of activity and knowledge. Different logics of related

diversification may be identified:

- Transition is characterized by a new domain emerging from an existing industrial commons (a

collection of R&D, engineering, and manufacturing capabilities that sustain innovation).

- Modernisation is manifest when the development of specific applications of a general purpose

technology produces a significant impact on the efficiency and quality of an existing (often

traditional) sector.

- Diversification in a narrow sense is a third pattern. In such cases the discovery concerns potential

synergies (economies of scope, spill overs) which are likely to materialise between an existing

activity and a new one. Such synergies make the move towards the new activity attractive and

profitable.

Another pattern involves the radical foundation of a domain. In this case, the discovery is that R&D and

innovation in a certain field has the potential to make some activities progressive and attractive that had

not been previously.

Priorities emerging today will not be supported for ever

While at t0 some priorities emerge and subsequent activities will be supported, it is expected that 3 or 4

years later other discoveries will be made in other parts of the regional system and the subsequent

emerging activities will be supported as well. Smart specialisation entails strategic and specialized

diversification. This principle is important to help policy makers to make choices and decide priorities.

These choices are not so tough since those activities not selected now have a chance to be supported in the

future.

Before going to the 5th criteria, we can see that at a first glance the main goals of smart specialisation

involve:

i) facilitating the emergence and early growth of new activities which are potentially rich in

innovation and spillovers;

ii) diversifying the regional systems;

iii) generating critical mass, critical networks, critical clusters.

The experimental nature of the policy and the need for evaluation

Clear benchmarks and criteria for success and failures are needed. Because of its nature this policy is

experimental: this is the nature of entrepreneurial discovery that not all investments in new activities will

pay off. Evaluation is, therefore, a central policy task so that the support of a particular line of capabilities

formation will not be discontinued too early nor continued so long that subsidies are wasted on non-viable

projects.

8

To summarize these five principles, we can highlight the following key words - non neutral policy, keeping

market forces working (self-discovery), interactive process between policies and the private sector, activity

as the right level of intervention, experimental nature of policy, what is important here is the process which

helps reveal areas of desirable interventions. These keywords sound very familiar to those who are

interested in the New Industrial Policy agenda.17. In the second part, I will discuss the design of programs

that might help to achieve the goals just identified and to assess these programs.

3 - Smart specialisation programs

In line with the framework previously elaborated, one can think of the following programs. Designing and

implementing a smart specialisation strategy involve efforts and actions directed to:

i) maximize “public-private entrepreneurial discoveries”

ii) provide operational facilities for continuous observation, detection and evaluation

iii) support early growth of the prioritized activities

iv) align incentives

3.1 - Tools and mechanisms to support entrepreneurial discovery

Information externalities

In the recent literature addressing the problems of entrepreneurial discovery, the simple and only rationale

for policy is given by the case of informational externalities18: “good” discoveries are expected to result in a

multiplication of “entries” in the new activity; which is a good thing for the regional evolution towards

smart specialisation. But this will raise an appropriation issue. The entrepreneur who has made a discovery

will not be able (and actually should not be able) to capture a significant fraction of the social value of her

initial investment. Consequently, there is a risk that not enough agents and organisations will invest in this

particular type of discovery. So – according to these authors - the correction of imperfect appropriation is

the main policy problem and it is a difficult one since imitative entry is desirable to a certain extent (and so

the problem should not be corrected by mechanisms such as patents of wide scope which have the effect

of blocking entry).

Capabilities

This information externality raises an important issue and requires the design of some mechanisms to

subsidize the costs of discoveries. But the objective of building an economy with an intensive level of

entrepreneurial experimentation and discovery requires other types of actions than “simply” correcting this

market failure; and this is particularly true for regions which are relatively poor in entrepreneurial

capabilities. This goal also requires creating conditions for multiple micro-systems of experiments and

discoveries to emerge. The performance of entrepreneurs and firms in experimenting with and discovering

potential domains for future specialisation may depend upon the way in which they build an external

organisation of connections with universities, laboratories, suppliers and users. The main policy problem

therefore appears to be one of helping to design such inter-organisational connections and coordination of

efforts in the sphere of experimentation and discovery19.

17 - D.Roddrik, 2004, op.cit.

18 See Haussman and Rodrik, 2003, op.cit. as well as D. Rodrik, 2004, op.cit. 19 P.A. David and S. Metcalfe, Universities and public research organisations in the ERA, Report prepared for the Knowledge for Growth expert group, EC (DG-Research); and P. Aghion, P.A. David and D. Foray, “Science, technology

9

In regions which are poor in entrepreneurial capabilities, the main question is, therefore, not insufficient

incentives (informational externalities) that might impede the private effort of the existing entrepreneurs

but, instead is the insufficient supply of local entrepreneurs. Policymakers concerned with this kind of

regions will face different options for launching a smart specialisation strategy, including the mobilization of

extra-regional resources20.

Guiding discoveries?

An important research question deals with the role of policy not only in supporting entrepreneurial

discovery but also in influencing the “direction” in which experiments and discoveries should be oriented.

Under what conditions such policy action can be undertaken without causing the usual failures of wrong

choices and market distortions? In the section 2 (above), a typology of structural changes has been

suggested (modernization, diversification, transition, radical foundation). This typology outlines central

elements in the policy process. It provides policy makers with the possibility to think ahead and to identify

the most desirable structural evolution of the regional economy given its strengths and weaknesses. The

policy maker can search for the needed entrepreneurial knowledge and discoveries which will materialize

and validate the policy vision. There is therefore a feedback mechanism from a policy vision –as structured

by the identification of what structural change would be particularly desirable for the regional economy –

to the search for the entrepreneurial knowledge in the sectors and institutions corresponding to such a

vision. However subsequent decisions and choices – whether to help and support a particular trend as a

potential domain for future specialisation - is conditioned by the quality of entrepreneurial discoveries that

will be (or not) made.

Funding experiments and discoveries

Determining the appropriate method of financing experiments and discoveries as well as the initial

development of a new activity is no trivial matter. The uncertainty associated with starting a new activity is

coupled here with the uncertainty and risk related to the fact that, very often, this activity will be carried

out in a region that is little developed. The uncertainty, informational asymmetries and moral hazard21 are

considerable and are likely to permit opportunistic behaviour on the part of entrepreneurs. It will therefore

be difficult to attract private investors or even win a share of development funds set up by banks as part of

their corporate responsibility.

The combination of high uncertainty, asymmetric information and moral hazard, and the fact that R&D

typically does not yield results instantaneously, imply a particular funding mechanism: venture capital

organisations (VCs). While R&D carried out by small entities and entrepreneurs are often characterised by

considerable uncertainty and informational asymmetries, permitting opportunistic behaviour by

entrepreneurs, VC organisations employ a variety of mechanisms to address these information problems. In

short, the environment in which VCs operate is extremely difficult. It is the mechanisms associated with the

and innovation for economic growth: linking policy research and practice”, Research Policy, 38(4), 2009, for a more general treatment of the importance of such “linkages” in growth policy design at the national level. 20 - -See Rodrik, 2004, op.cit. for a careful consideration of the role of diaspora in regional innovation policy.

21 - Moral hazard refers to inefficient behavior by one actor in a transaction brought on by differences in information available to parties in the transaction – on application about finance and innovation, see B. Hall and J. Lerner, “Financing R&D and Innovation”, in Hall and Rosenberg (eds.), Handbook in Economics of Innovation, vol.1, North-Holland, 2010.

10

VC funds that are critical in ensuring that they receive a satisfactory return. These circumstances have led

to VC organisations emerging as the dominant form of equity financing for privately held technology

intensive businesses. At the same time, there are reasons to believe that despite the presence of private VC

funds, there still might be a role for public VC programmes in the kind of difficult contexts described above.

There are several arguments for public investments:

- The structure of venture investments may make them inappropriate for many projects (venture

funds tend to make quite substantial investments, even in young firms, and so VC organisations are

unwilling to invest in projects that require only small capital infusions).

- The VC industry is limited: VCs back only a tiny fraction of technology-oriented businesses and VC

funds are highly geographically concentrated.

- If public VC awards could certify that projects are of high quality, some of the information problems

could be overcome and investors could confidently invest in these firms.

- Finally, public finance theory emphasizes that subsidies are an appropriate response in the case of

activities that generate positive externalities.

All these reasons for which public VC might be a complement and extension to private VC are valid in the

case of projects aiming at discovering new areas for future specialisation. Such efforts often have financial

requirements that are too small in relation to the average financing scale. Project may be located in not

very well advanced regions, which increases the informational problems to such an extent that the

customary monitoring mechanisms set up by the VC may seem insufficient or increase the costs too sharply

compared to the anticipated profitability. Finally, the essence of entrepreneurial discoveries is the

generation of informational spillovers (effects of demonstration and emulation) that in themselves

represent a rationale for public financing.

An important policy tool to study and specify is therefore a public VC fund; that is to say a public financing

mechanism addressing the problems of entrepreneurship and entrepreneurs’ projects given the difficult

contexts and circumstances of many regional economies.

3.2 - Observation, detection and evaluation

Fine grained observation is needed to see what’s happening at sub-system level. This is the right level to

observe ‘what are the pieces of the knowledge economy’ that a region can take as a basis for smart

specialisation. Observation of micro-dynamics is obviously a difficult task and theredore new incentives are

needed for encouraging firms to elicit information. The idea is to transform the approach to detect

entrepreneurial discoveries from one of ‘what does the policy maker knows and how can she find out what

she does not know’ to one of ‘how those who knows, the entrepreneurs, can be induced to come forward

with that knowledge.

Beyond observation and detection, monitoring and assessing are central policy tasks so that the support of

a particular line of capabilities formation will not be discontinued too early nor continued so long that

subsidies are wasted on non-viable projects. The ability to detect, monitor and assess is closely connected

with ex ante assessment capabilities of policy makers. They need to observe, to be able to screen between

“simple” innovation and discoveries that have the potential to spawn new areas of specialization and which

might be the corner stone of a smart specialisation strategy.

11

It is essential to operationalize the process of assessing potentials to reduce risks in policy implementation

and practice of smart specialisation. Precise ex ante estimation of the future value of an R&D specialisation

that would be required for a cost-benefit analysis is a nearly impossible task and one better left to

investment markets. The well-known “blind giant” metaphor tells us that it is always very difficult at an

early stage to assess the stability and sustainability of a specialisation22. This concern is why the smart

specialisation approach is positioned at a particular point in the development cycle, one in which a degree

of local commitment and development has already occurred and achieved a measure of stability.

Evaluation and subsequent decisions (support) should happen at a certain point in the development cycle

where degrees of local commitment and development have already occurred (to avoid the lottery of very

early stages)

3.3 - Support of early stage and growth of new activities

Fixing coordination failures Most projects with the potential to give rise to a new activity require simultaneous large-scale investments

to be made in order to become profitable. All the necessary services and complementary activities have

fixed costs and are unlikely to start unless the potential provider have enough positive expectations

regarding the future of the smart specialisation strategy. Profitable new activities can fail to develop unless

upstream and downstream investments are made simultaneously. Such coordination problems have

several solutions, not necessarily based on subsidization23. Fixing coordination failures involves also

supporting the provision of adequate supply-responses (in human capital formation) to the new

“knowledge needs” of traditional industries that are starting to adapt and apply the GPT, by subsidizing the

follower region’s access to problem-solving expertise from researchers in the leader region, and by

attending to the development of a local personnel that can sustain the incremental improvement, as well

as the maintenance of specialised application technologies in the region.

Public good provision

The most obvious coordination measures to consider are those that would provide specific

complementary public goods – such as programs of further education, pre-competitive R&D and tax

credits or other subsidies for on-the-job training in relevant skill domains where these are not present

elsewhere in the regional economy. Also included under this heading is public support for the

provision of adequate supply-responses (in human capital formation and research) to the new

“knowledge needs” of traditional industries that are starting to adapt and apply a GPT, which might

take the form of subsidising expenditures to gain access to problem-solving expertise from

researchers in an external, technologically leading region.

3.4 – Aligning incentives through intelligent policy design

Building intelligent policy design involves essentially solving the potential conflict between two kind of

incentives that are needed along the process i) incentives to reward those who discover new domains and

activities and ii) incentives to attract other agents and firms and facilitate entries so that agglomeration and

scale effects materialize at the next stage. The two series of incentives are not perfectly aligned.

22 - P.A.David, “Path-dependence in economic processes: implications for policy analysis in dynamical system contexts”, in Dopfer (ed.), The Evolutionary Foundations of Economics, Cambridge University Press, 2005 23 D.Rodrik, 2004, op.cit.

12

The entry phase – once initial discoveries have been made and have led to first entrepreneurial success- is

when a single discovery begins to be translated into a collective phenomenon, so that agglomeration

externalities can be realized. As noted earlier, there is a tension between the need for entrepreneurs who

make discoveries to capture private returns and the need to prevent this appropriation from foreclosing all

of the social value of the discovery. For example, the first firms that have been deemed eligible for smart

specialisation subsidies on the basis of their business plans to pursue opportunities in the new line of

regional specialisation should be under some obligation to disclose (for publication by the government

authority administering the subsidy programs) the principle generic features of their respective

business plans.

The “social value” failure occurs when there are not enough agents and organisations willing or able to

invest in the direction and domain indicated by the initial discovery. Since imitative entry is desirable to a

certain extent, the necessary correction of imperfect appropriation raises a difficult problem that needs to

be addressed. What are the mechanisms that will allow the initial discoverer to capture adequate private

returns while not foreclosing the additional social returns stemming from entry?

(see diagram 1 in appendix for an example of a policy design addressing this kind of problem).

3.5 The space of smart specialisation

Neither pre-defined “regions” nor specific sectors can be used ex ante to determine the boundaries of a

smart specialisation dynamics, as articulated above. Whatever we call it – the knowledge ecology or the

industrial commons – that is to say the collective R&D, engineering and manufacturing capabilities that

sustain innovation – are not necessarily deployed and contained within strict regional boundaries and their

development and evolution is likely to defy administrative frontiers. In other words, resources in the

knowledge economy are not immobile and specific to each region. Extra-regional entrepreneurship, like

extra-regional finance, and skilled “business service” can initiate and carry on new enterprises in regions

where those factors of production are scarce. By the same token, such extra-regional resources (including

research services) can develop and expand the capacity of small regional enterprises that have been

launched by local entrepreneurs. This opens the question of the larger ecology of innovation to which the

particular regional system belongs.

4 – The policy potential (political salience)

4.1 - Not only for the best.

This concept provides strategies and roles for any region. Indeed, the concept is built around the fact that

there is not only one game in town in terms of R&D and innovation i.e. innovation is multi-dimensional and

there are many other kinds of productive and potentially beneficial activities apart from the invention of

fundamental knowledge needed for the development of general purpose technologies and tools (GPTs)

such as information and communication technology (ICT) or biotechnology. There are in fact different logics

or orders of innovation24. In other words, innovation often involves the development of applications of a

GPT which has been invented elsewhere. Some regions can indeed specialise in the invention of the GPT

while others will invest in the ‘co-invention’ of applications to address particular problems of quality and

productivity in one or a few important sectors of their economies. ‘Co-invention’ is an important notion

24 T. Bresnahan, 2010, op.cit.

13

here because it means that the very act of adopting some ICTs (or any other GPTs) to improve operational

efficiency or product quality in a given sector of industry or service is by no means a simple task. ICT

applications are not ready and waiting on the shelf for new users. The co-invention of applications involves

a great deal of R&D, design and redesign, i.e. a collection of knowledge-driven activities. Smart

specialisation therefore implies rejecting the principle of a sharp division of labour between knowledge

producers and knowledge users. All regions face challenges in terms of improving the operational efficiency

and product quality in their business and industries and making these improvements is often a matter of

R&D, capabilities development and innovation which generates a certain kind of structural change (e.g.

“modernisation” or “capabilities upgrading”).

The smart specialisation strategy seeks to avoid petrifying relative positions between followers and leaders

with the less advanced regions being locked in to the development of applications and incremental

innovations. Of course smart specialisation has no magical properties to transform laggards into global

leaders. However, at minimum, a smart specialisation strategy transforms less advanced regions into good

followers: a region in transition which is building capabilities and is agglomerating knowledge resources in

a certain domain of application, so that it will be able to capture knowledge spillovers from the leaders

(those who are inventing the basic technology), to attract further knowledge assets and to develop an

ecosystem of innovation with the prospect and the realistic hope of becoming a leader! A leader? Yes but a

leader not in inventing the generic technology but in co-inventing specific applications (for example ICTs

applied to logistics or biotechnology applications for monitoring agricultural production).

This means that the follower regions and the firms within them, by designing and implementing a smart

specialisation strategy become part of a more realistic and practicable competitive environment -- defining

an arena of competition in which the players (other regions with similar strategies) are more symmetrically

endowed, and a viable market niche can be created that will not be quickly eroded away by the entry of

larger external competitors.

4.2 - ..but also for the best

Perhaps the best regions or countries have super-efficient systems in which discoveries are made

continuously and good framework conditions make new activities growing well so that strategic

diversification is happening at any time. The Silicon Valley for example is well equipped to catch the new

waves of opportunities because of its “innovation habitat”. It is a habitat that is good at incubating not just

IT start ups. May be! But in most cases of successful regions, the success of today is not a guarantee of

success for tomorrow. Successful clusters are not protected against the disease of innovation routinization,

creative myopia and collective inertia. Many historical cases tell the same story of very successful clusters

or regions not capable of re-inventing themselves when new waves of technologies and market needs

come. Moreover, when innovation is particularly concentrated in a single large firm, it is proved that such

firm and the people employed suffer from creative myopia. They are not inclined to look outside, to learn

from others25

25 - see A.Agrawal, I.Cockburn and C.Rosell, “Not invented here : creative myopia and company towns », draft, 2009

14

Are there enough experiments and discoveries beyond the current innovative routines? In leading regions

too, entrepreneurs exploring new domains beyond and outside the innovation routines need to be

detected and supported.

4.3 - System’s level good properties of smart specialisation (vs. cluster policies)

The political salience of smart specialisation is also due to its potential contributions to greater efficiencies

in resource allocation (human capital, research infrastructures, specialised services for innovation) at the

system level (for instance the level of an integrated regional system such as the EU.). Smart specialisation

offers a means for local strategies based on identifying and developing original, distinctive and fertile areas

of specialisation for the future that are likely to promote greater diversity in the areas of knowledge and

expertise within the system, thereby rendering the entire regional economy more able to enjoy the

benefits of distinct local agglomeration economies.

Other regional technology policies – such as cluster policies do not exhibit such efficiency properties at

system level. Of course generating a vibrant innovative cluster is a classic outcome of a smart specialisation

strategy (see below section 22); one might say an “emergent property” of a smart specialisation policy

applied to a particular region for purely local economic interests. But, cluster policies at regional level are

likely even to accentuate strongly mimetic programs of local and national industrial development –

resulting in fostering knowledge base standardization, wasteful duplication and the dissipation of the

potential agglomeration economies at system level – as a multiplicity of imitative local government

authorities compete to attract the small finite pool of mobile capital, management and knowledge

resources. The resulting duplication, unproductive uniformity and lack of imagination and vision in setting

R&D and cluster priorities can be expected to produce poor results at the EU level; with most regions

remaining unattractive and unable to compete with other territories to attract high value resources and to

retain their best resources. Smart specialisation, on the other hand, involves discovery of what makes a

local knowledge base original and distinctive and, thereby, exhibits “efficiency properties” at the system

level – i.e. for an integrated regional system as a whole, such as the EU.

5 – Conclusion

To quote Paul David, one of the co-author of the very first paper written on smart specialization: «Yet such

strategies have a chance to yield results that will be superior to the past tendencies produced by

undifferentiated EC recommendations of undifferentiated ‘best policy practices’ – encouraging policy

makers to set their sights on doing the same ‘good things’ to foster the same forms of innovation» 26

26 - P.A.David, « Comments on ‘Enhancing Bulgaria’s competitiveness and export performance through technology

absorption and innovation », 2011

15

Appendix A – Aligning incentives to support discoveries and to help subsequent growth of activities

Step A

Mechanims designed as a

contest whereby private sector entrepreneurs would

bid for public resources by

bringing forth pre-investment proposals

The reward needs to be structured in such a

way that maximizes the

spillovers to subsequent entrants and rivals (Step B)

Criteria for financing at step

A:

*Activities opening a new

(related) domain

*They have the potential to

provide information

spillovers to others

*Private entities are willing

to submit themselves to

oversight and performance audits