Professional Documents
Culture Documents
9.1 INTRODUCTION
In keeping with its title, the purpose of this book is to illuminate cluster
policies and the institutional, governance and firm dynamics that have been
unfolding in Europe. As argued in the introductory chapter, its main aim is to
address the following paradox: in spite of increasing political, business, and
research interests on cluster related policies and institutional frameworks that
can allegedly enhance the performance of cluster firms, there is still very
limited scientific understanding about the role of policies and institutional
dynamics in clusters. Redressing this shortfall of scientific understanding is
of paramount importance especially given the ever-increasing pressures on
cluster firms due to regional integration and globalisation tendencies. In
Europe this has been felt even more strongly due to the recent Eastern
enlargement. The accession of ten new members in conjunction with the
aforementioned tendencies has unleashed economic and institutional
dynamics with significant impacts to both old and new EU members.
The above aim translates into the main objectives of this study concerning
cluster learning dynamics. As introduced in chapters one and two these
objectives are articulated in terms of two key research questions: a) Does
cluster MLG exhibit learning dynamics?, and b) Does MLG support cluster
learning dynamics? In answering these two key research questions, this study
implements an original approach combining three analytical foci.
Firstly, this study focuses on the multi-level dimensions of cluster policy
and governance in the understanding that the local nature of cluster policy
and institutional dynamics are largely related to specific MLG patterns.
Secondly, it focuses on learning processes associated with policy and
institutional dynamics. This is accomplished by considering the two related
aspects of learning: i) the governance processes that induce learning in cluster
firms, and ii) the learning process of governance itself.
Thirdly, it focuses on the development of a robust analytical comparative
framework supporting the rigorous study of the above phenomena and their
multifaceted and heterogeneous characteristics. This framework has been
developed and tested throughout the stages of this study: from extensive
quantitative and qualitative data collection utilising a range of instruments by
243
244 Cluster Policies in Europe: Firms, Institutions, and Governance
In the penultimate section of this chapter the implications of the findings for
theory and practice will be presented focusing in particular on the
assumptions and treatments related to MLG and cluster learning. Finally, this
chapter and study will close by summarising the key areas for further
research that emerge from the above.
W Midlands automotive
Bohemia automotive
Slovenia automotive
Slovenia furniture
Herford furniture
Importance
ABT footwear
ABT clothing
Scottish ICT
Prague film
Rome film
Mean
Types of Institutions
Local banks & VC firms 3.17 3.00 2.59 2.76 2.00 2.50 3.05 2.77 3.69 2.38 3.31 2.10 2.63
Other 2.00 5.00 1.90 4.00 4.40 1.00 1.33 2.43 1.44 2.33 1.81 3.00 1.00
Universities and HEIs 2.36 2.36 2.95 1.79 2.50 2.74 1.94 2.05 1.72 1.88 0.88 1.60 1.87
Chambers of commerce 2.58 1.62 2.20 1.69 1.53 2.00 2.83 1.93 2.19 2.80 1.06 1.57 1.13
Mean 2.33 2.25 2.05 1.99 1.97 1.96 1.92 1.91 1.90 1.89 1.66 1.62 1.43
Sectorial associations 3.36 2.08 2.12 2.03 1.67 1.81 1.70 1.89 1.88 1.88 1.06 2.00 1.13
Certification laboratories 2.56 2.08 2.40 2.11 1.53 2.85 1.94 1.88 1.84 2.00 1.06 1.10 1.03
Service centres 2.89 1.91 1.69 1.70 1.76 1.52 1.40 1.84 1.40 1.42 2.25 1.97 2.13
Consulting firms 1.75 1.92 1.50 1.62 2.06 2.19 2.11 1.81 2.00 1.88 1.75 1.40 1.50
Technology centres 1.63 1.50 2.70 1.38 1.94 2.00 1.44 1.70 1.88 1.72 2.06 1.00 1.17
Trade unions 1.88 2.00 1.14 1.48 1.00 1.63 1.78 1.40 1.44 1.33 1.07 1.03 1.07
Non-profit foundations 1.44 1.27 1.36 1.32 1.24 1.30 1.56 1.34 1.40 1.17 1.94 1.00 1.07
Notes: White cells contain the sample means of each of the twelve clusters for each of the eleven
types of institutions on a scale of 1 (not important) to 5 (most important); whereas black cells
contain the means of these twelve clusters and eleven types of institutions. The columns and
rows are ordered in terms of descending means.
Sources: Authors’ elaboration, based on interviews with firm representatives (n=303) reported in
Tables 3.2, 3.5, 4.2, 4.5, 5.2, 5.5, 6.2, 6.5, 7.2, 7.5, 8.2, 8.5.
the 2.5 importance threshold in four of the studied clusters: the Bohemia
automotive, Slovenia furniture, Scottish ICT, and Val Vibrata clothing and
substantially below the threshold in the other eight clusters. This can also be
interpreted as suggesting that the range of listed institutional types does not
fully exhaust the variety of the cluster firms’ needs in at least about a third of
the studied clusters. That is, given their high importance of ‘other’.
Moving to the third most important type of institution, viz. Universities
and HEI a more consistent picture seems to appear (i.e. smallest standard
deviation among the top-four most important types of institutions). The
general positive perception of higher education for local growth and their
importance in the firms’ local context has been suggested in a number of
studies, e.g. Benneworth and Dawley (2005), Jones-Evans et al. (1999).
However, it has to be pointed out that its third place is a relative position and
that in absolute terms its mean is below the 2.5 importance threshold. In
actual fact this type of institutions scores above the threshold only in a
quarter of the studied clusters, viz. Saxony automotive, Slovenia automotive,
and Scottish ICT (in descending order of importance). The relative large
science and technology research base of these three clusters along with the
presence of strong and highly relevant HEIs explains to a large extent the
heightened importance of this type of institutions in these three clusters.
Nonetheless, the relative position of HEIs in the studied clusters is to be
interpreted in the context of the on going transformation of higher education
systems in Europe, the search in OECD countries for a more active role of
Universities in their local economic context, and in particular, the recent
normative emphasis on the entrepreneurial role of Universities for innovation
(Mueller 2006; Etzkowitz 1998).
Finally the fourth most important type of institution is Chambers of
commerce. Although there appears to be some variation of its importance
across the twelve clusters its standard deviation is not that excessive (i.e. only
slightly higher than HEIs). Part of this variation could be explained by
differences among countries for compulsory versus voluntary membership
and thus differences of each Chambers resources and gravitas. However, this
tells only part of the story as in the case of Germany where the Chambers’
legal framework and standing is the same, there is still some difference
between the two clusters. The conclusion thus seems inescapable that
Chambers are not equally appreciated by the various specialisations they
serve in their vicinity, which may be due to variety of reasons (including
among others historic or contemporary ties between the Chambers and
specific specialisations). Obviously Chambers are not the only type of
institution prone to such caveats. In the case of Romania for example the
firms in the footwear cluster tend to value Chambers lower and banks higher
than the firms in the clothing cluster. Thus the variations in the above and
subsequent findings concerning institutional importance need to be
interpreted within the context of the specialisations served.
248 Cluster Policies in Europe: Firms, Institutions, and Governance
and equally interrelated reasons, e.g. due to at least one third of the sample
being constituted by leading firms in each cluster that may see little value in
supporting new firms and/or relocation of other firms, and/or because most
leading firms would require little policy help for relocation, and/or because
incubation needs may be already relatively well catered in the sample of
clusters.
It thus has to be emphasised that these five key observations are based on
aggregate data, and to remind of the variation encountered in each cluster as
discussed at some length in the previous six chapters. Having alerted to these
caveats the discussion around Figure 9.1 would nonetheless be incomplete
unless something was said about any west-east, national, and/or sectoral
patterns that have been absent from it so far. The reasons that such patterns
have not figured prominently in the discussion so far is because there is very
little to report about them. That is, there are no west-east nor any national
patterns in policy relevance worthy of report; whereas the only sectoral
pattern worthy of report concerns the four automotive clusters that seem to
rank the relevance of all policy areas on average above other specialisations.
However, the latter pattern is diluted in the future with the clothing
specialisation taking the lead. This set of weak otherwise patterns forms the
sixth key observation.
2.5
2.0
1.5
1.0
0.5
0.0
Environmental
Education &
Firm networks
recruitment
development
Information
R&D
relocation
Venture/risk
incubation
attraction
Customised
New firm
Start-up &
infrastructure
diffusion
Labour
services
training
Physical
Firm
capital
Quality
policies
Notes: The reported means are derived from the means of each cluster on a scale of 1 (not
relevant) to 5 (most relevant).
Sources: Authors’ elaboration, based on interviews with firm representatives (n=285) reported in
Tables 3.3, 3.6, 4.3, 4.6, 5.3, 5.6, 6.3, 6.6, 7.3, 7.6, 8.3, 8.6.
Last but not least, a few words ought to be said about the views of cluster
institutions. As elaborated in the previous six chapters cluster institutions
Comparative Findings and Conclusions 251
tend to overvalue most policy areas. Thus a respective figure for cluster
institutions would indicate means in excess of the 2.5 relevance threshold for
all policy areas; even at present. This is also the main reason why such a
figure has been omitted from this section. Nonetheless, cluster institutions
converge with firms in expecting the relevance of all policy areas to increase
in the future. Where they differ is on the extent of sharing such relevance at
present and in the future or in other words on the so called relevance gap and
policy convergence both of which will be examined at some length in section
9.5.
W Midlands automotive
Bohemia automotive
Slovenia automotive
Herford furniture
Locality of
Type of
ABT footwear
knowledge
ABT clothing
knowledge
Scottish ICT
Max sources
Prague film
source
Rome film
Mean
Technical 4 100.0 75.0 75.0 100.0 75.0 77.1 100.0 100.0 75.0 75.0 50.0 50.0 50.0
Internal
Organisational 5 100.0 100.0 100.0 100.0 78.2 80.0 80.0 80.0 80.0 60.0 40.0 40.0
Technical 9 44.4 44.4 33.3 44.4 33.3 26.9 22.2 11.1 22.2 11.1 11.1 33.3 11.1
Local
Organisational 7 57.1 57.1 42.9 42.9 23.4 28.6 14.3 14.3 0.0 0.0 0.0 0.0
External
Technical 9 55.6 55.6 44.4 33.3 22.2 31.5 33.3 22.2 22.2 33.3 11.1 33.3 11.1
National
Organisational 7 57.1 57.1 14.3 42.9 20.8 14.3 14.3 0.0 0.0 14.3 14.3 0.0
Technical 9 55.6 44.4 22.2 22.2 11.1 17.6 0.0 0.0 11.1 0.0 22.2 22.2 0.0
International
Organisational 7 42.9 28.6 0.0 28.6 11.7 0.0 0.0 0.0 0.0 28.6 0.0 0.0
Mean 64.1 54.9 52.2 44.6 44.5 35.9 34.8 30.2 28.1 24.9 24.7 24.1 14.0
Notes: Values are reported in percentage terms of sources above the 2.5 importance threshold as a share the max possible sources in each category.
Sources: Authors’ elaboration, based on based on interviews with firm (n=303) representatives reported in Tables 3.2, 3,5, 4.2, 4.5, 5.2, 5.5, 6.2, 6.5, 7.2, 7.5,
8.2 and 8.5.
Comparative Findings and Conclusions 253
MLG
Adaptability Vertical directionality
Clusters
Herford furniture Very high
Scottish ICT
High Top-down and bottom-up
Slovenia automotive
Rome film Medium
Saxony automotive
W Midlands automotive
High
Prague film
Slovenia furniture
Top-down
Bohemia automotive
ABT footwear Medium
ABT clothing
Val Vibrata clothing Low
Notes: All rankings use the following five point scale: nil, low, medium, high, very high; except
MLG vertical directionality using a three point: top-down , bottom-up, or both.
Sources: Authors’ elaboration, based on Tables 3.9, 4.9, 5.9, 6.9, 7.9, and 8.9.
Firstly, it is noticeable in Table 9.3 that there is no cluster amidst the sample
of twelve with solely ‘bottom-up’ dynamics. As well as that the majority of
clusters have only ‘top-down’ dynamics. This appears as rather surprising
given the decentralisation or quasi-decentralisation tendencies in most
European countries during the last decade. Although most such clusters are in
CEECs, the presence of three Western clusters amidst them indicates that the
lack of ‘bottom-up’ dynamics is not exclusively a CEEC phenomenon.
Secondly, there are no clusters with ‘low’ MLG adaptability and both
‘top-down’ and ‘bottom-up’ dynamics. This can be taken to suggest that the
two criteria have only a small overlap concerning extreme cases of ‘low’
Comparative Findings and Conclusions 255
Table 9.4 All twelve clusters policy relevance gaps between firms and institutions
Policy Gaps
(absolute values)
W Midlands
Policy mean
Val Vibrata
automotive
automotive
automotive
automotive
Bohemia
footwear
Slovenia
Slovenia
furniture
furniture
clothing
clothing
Scottish
Herford
Saxony
Prague
Rome
ABT
ABT
film
film
ICT
Policy areas
l) Improving availability of venture/risk capital -2.64 -1.50 -1.75 -1.95 -0.25 -2.30 -1.96 1.41 0.14 -1.63 -1.58 -0.53 0.73
e) Research and technological development -1.75 -2.49 -1.90 -1.33 -1.13 -0.50 -2.14 1.31 -1.86 -0.80 -0.71 -0.80 -0.33
d) Education and training -1.48 -2.45 -1.05 -1.21 -2.53 0.67 -1.10 1.27 -1.87 -1.58 0.09 0.12 -1.05
a) New firm attraction to the cluster -0.81 -1.79 -0.19 -1.50 -2.18 -1.49 -0.82 1.22 -1.00 -1.33 -1.24 -1.90 0.36
k) Start-up and incubation support -2.66 -0.50 -2.31 -2.00 -1.75 -0.94 1.22 -0.68 -0.18 -1.33 -0.93 0.10
i) Establishing firms’ networks -1.06 -1.97 -0.92 -2.21 -0.92 -1.92 -1.57 1.14 -0.61 -0.85 -1.38 0.00 0.24
b) Support for cluster firm relocation -1.77 -1.62 -0.46 -2.00 -1.89 -1.13 1.10 -0.70 -1.12 0.17 0.14
Cluster mean (absolute values) 1.76 1.71 1.32 1.24 1.17 1.13 1.07 1.07 0.90 0.87 0.73 0.55 0.45
Cluster mean gap size VL VL VL L L L L L M M M S S
j) Improving firm quality development -1.66 -2.24 -0.90 -1.46 -1.09 -0.79 -0.73 1.04 -0.45 -1.02 -0.89 -0.43 0.86
h) Labour recruitment -2.50 -1.71 -1.79 -0.50 0.92 -0.04 -0.75 1.02 -1.59 -1.25 -0.73 0.35 -0.05
m) Environmental policies -1.92 -2.14 -2.05 -1.51 0.06 -1.00 -0.81 1.00 -0.61 -0.79 -0.14 -0.43 0.57
c) Physical infrastructure development -1.75 -1.39 -1.35 -1.46 -2.00 -0.40 -0.81 0.99 -0.91 -0.35 0.15 0.98 -0.36
g) Customised services to firms -1.71 -1.19 -0.32 -0.23 -0.63 -0.60 -0.94 0.64 -0.79 -0.68 -0.13 0.14 0.32
f) Information diffusion and accessibility -1.17 -1.10 -1.02 -0.25 -0.31 -1.36 -0.26 0.59 -0.53 0.00 -0.05 -0.37 0.71
Notes: Policy areas and clusters in descending order of means on a scale of 1 (not relevant) to 5 (most relevant). A minus sing refers to firms undervaluing the
respective policy area; whereas its absence the opposite. ↑ (↓) indicates a future increase (decrease) in excess of +0.5 (-0.5). Gap sizes refer to the present and
utilise the following five point scale: none (0-0.2), small (0.3-0.6), medium (0.7-0.9), large (1.0-1.2), and very large (>1.3). Blank cells indicate missing data.
Sources: Authors’ elaboration, based on interviews with firm (n=285) and institutional (n=99) representatives reported in Tables 3.7, 4.7, 5.7, 6.7, 7.7, and 8.7.
Comparative Findings and Conclusions 257
Obviously and especially for gaps that are at present already ‘large’ or ‘very
large’ such increases ought to be alarming.
Thirdly and focusing on the rows of Table 9.4 it is relatively clear that the
policy areas with the smallest relevance gaps occupy the last rows. Those in
the last three rows in particular, viz. c) physical infrastructure development,
g) customised services to firms, and f) information diffusion and accessibility
have on average ‘small’ or ‘medium’ gaps suggesting that they are the least
controversial. Moreover, cross-referencing these three policy areas with
Figure 9.1 it is clear that they are not just the least controversial but also the
three most relevant ones for the firms in the twelve clusters of this study. At
the other extreme, i.e. at the top rows of Table 9.4 lie the most controversial
policy areas or the largest relevance gaps, viz. l) improving availability of
venture/risk capital, e) research and technological development, and d)
education and training. The way to interpret and deal with relevance gaps is
to accept that they indicate true disagreements within clusters about specific
policy areas and that thus clusters with such gaps should engage in
conversations (i.e. linguistic exchanges in the course of a joint exploration)
regarding their collective views and policy needs for the present and future
(see implications in section 9.7).
Fourthly and focusing on the columns of Table 9.4 it is similarly clear
that the clusters where there is most controversy surrounding policy (on
average) are encountered on the left-hand side columns, e.g. Saxony, Prague
and West Midlands. This is alarming especially for such clusters that exhibit
in addition upward pointing arrows in several policy areas. Cross referencing
Table 9.4 with the previous criterion (e.g. Table 9.3) it could be concluded
that MLG adaptability and vertical bi-directionality do not guarantee less of a
drift between firms and institutions. As even those clusters with ‘high’ or
‘very high’ adaptability and ‘top-down’ and ‘bottom-up’ vertical
directionality, viz. Herford furniture, Scottish ICT, and Slovenia automotive
have respectively ‘large’ and ‘medium’ gaps. Moreover for Herford furniture
and the Slovenia automotive clusters such gaps are also anticipated to
increase in the future. These findings suggest in addition that the two sets of
criteria are pretty independent as well as that having both ‘highly’ adaptable
and ‘bottom-up’ MLG learning dynamics does not guarantee convergence of
policy preferences and in particular ‘small’ policy relevance gaps.
Fifthly it needs to be at least briefly, albeit explicitly stated that the
aforementioned patterns are much stronger than any west/east, national, or
sectoral ones. That is, the size and distribution of gaps and their future
tendencies seem to be uncorrelated with such variables.
Sixthly and last, some obvious caution needs to be issued concerning
clusters with (on average) similarly sized gaps in that the reasons behind such
similarities can be very different. For example the five clusters with the
smallest gaps (occupying the right-hand side of Table 9.4) turn out to be
pretty dissimilar. In Slovenia the factors underpinning the ‘small’ size gap
258 Cluster Policies in Europe: Firms, Institutions, and Governance
seem to be the industrial structure of the country and the active involvement
of cluster firms in institutional strategies; whereas in Romania the situation
differs because policy needs are apparent and obvious to both cluster firms
and institutions, albeit with cluster institutions being few and apart.
The comparative analysis can now proceed to examine the qualitative
data or the so called ‘top-four relevant policy areas share’ as part of the
second convergence criterion, summarised in Table 9.5 below.
Table 9.5 All twelve clusters firms’ and institutions’ top-four relevant policy
areas degree of convergence
Notes: All rankings use the following five point scale: none (zero top-four policy area shared),
low (one policy area), medium (two policy areas), high (three policy areas), very high (four
policy areas).
Sources: Authors’ elaboration, based on interviews with firm (n=285) and institutional (n=99)
representatives reported in Tables 3.8, 4.8, 5.8, 6.8, 7.8, and 8.8.
at least a ‘medium’ degree of convergence. Given this, the finding that 70%
of the clusters seem to have ‘medium’-to-‘high’ convergence also sound as
rather positive. Seen from the ‘half empty’ perspective it could be equally
claimed that 80% of the clusters have ‘medium’-to-‘low’ convergence which
does not sound like a positive finding. Presented with such competing
interpretations one should adopt a cautionary principle, especially given the
uncertainty surrounding clusters due to globalisation and Europeanization
tendencies.
As it may be remembered there are no less than thirteen policy areas
examined in this study. So even if cluster firms and institutions converge in
the top-three most relevant ones there are at least ten more where they could
be going astray. This is to say that policy-makers and stakeholders in general
need to be more considerate towards the priority set of policy areas for public
action.
In conclusion, the findings from this and the so far sections of this chapter
can be useful in the following sense. The policy areas that seem to enjoy the
highest degree of convergence, cross-referenced with the findings reported in
Figure 9.1 and Tables 9.1-3, should perhaps form the core of future EU27
cluster policy. In other words, these findings and identified policy areas could
serve to outline the overall lines of action in the EU regional policy.
Having completed the comparative analysis, presentation, and discussion
of findings concerning the first key research question of this study and its
respective sets of criteria the second key research question can now be
tackled. This is undertaken in the next section.
The policy focus on knowledge deficiencies, the second criterion for this
key research question, is assessed by cross-referencing the findings from all
previous criteria (including those of the first key research question) with the
findings on knowledge sources and processes at the cluster firm level (e.g. as
summarised in section 9.4). In other words, this last criterion examines the
extent to which governance is addressing the relative strengths and
weaknesses of the firms in each of the twelve clusters in terms of their
organisational and technical knowledge sources.
As such one ought to expect slightly more overlap between the two
criteria of this second key research question in comparison to the criteria of
the first key research question. This is so because the data used in connection
to the first criterion are a subset of the data used in connection to the second
criterion of this key research question. Bearing this in mind the respective
findings are summarised in Table 9.6.
Notes: All rankings use the following five point scale: nil, low, medium, high, very high.
Sources: Authors’ elaboration, on based on interviews with firm (n=303) and institutional (n=99)
representatives reported Tables 3.9, 4.9, 5.9, 6.9, 7.9, and 8.9.
Starting with the aforementioned overlap between the findings in the two
criteria of this key research question the twelve clusters in this study seem to
have a wide spread across the respective scale. However, the spread is more
even for the first rather than the second criterion suggesting that the overlap
is not as extensive as one may have surmised initially. Nonetheless, there is
more overlap than in the respective criteria for adaptability and vertical
directionality. As it may be remembered in Table 9.3 the overlap concerned
only the lower extreme. However, in this set of criteria the overlap is
detectable at both extremes as well as for some clusters in the middle (e.g.
Rome scoring ‘medium’ in both criteria).
Besides the above, there are no strong west/east, national and/or sectoral
Comparative Findings and Conclusions 261
patterns. In fact, the picture painted by this set of findings is rather mixed; i.e.
with no immediately apparent patterns.
Thus the message to be taken from this set of findings could be to focus
attention on those clusters that rank the highest in both criteria and to
highlight (as areas of policy learning for less fortunate clusters) the specific
ways through which their institutional capacity is used to tackle deficiencies
and problems in their respective clusters. Obviously according to Table 9.6
the five such clusters are: the two German and UK ones along with the
automotive one of Slovenia. They are basically, those clusters in which
public, semi-public, and private institutions have been very active at
promoting collective strategies for the cluster firms, and particularly active in
developing and/or assisting the development of knowledge-based resources
in key areas for their respective specializations.
To that extent the two CZ clusters also ought to be praised as in spite of
their ‘medium’ degree of problem-solving capacity, their focus towards
knowledge deficiencies is high. Obviously such lessons need to be paid even
more attention by clusters less fortunate in either and especially in both of the
aforementioned criteria.
Having completed the comparative analysis, presentation, and discussion
of the findings of this research in the next and penultimate section of this
study the comparative findings are summarised, their implications are raised
and respective recommendations are made.
Consistency of such views will also allow for the development of more effective
policy frameworks that help firms adapting constantly to new market
requirements. In actual fact, the findings make obvious (see Figure 9.1) that
across all clusters firms foresee an increase in the relevance of policy at the
future. This temporal consistency is telling of the need for policy to be
anticipatory as well as cognizant of current and future firms’ strengths and
weaknesses so to be able to create a supportive environment for firm adaptation.
This leads to the third implication of the findings stemming from the policy
relevance patterns. As discussed in section 9.3, the four most relevant policy
areas at present and in the future are: information diffusion, physical
infrastructure, customized services, and education and training (in this order). As
this is an aggregate pattern, the particular policy preferences of each cluster may
deviate from it (e.g. as detailed in each national chapter). Nonetheless, there is a
startling commonality in this aggregate pattern that is not just limited to the top-
four most relevant policy areas but extends to the top-six; i.e. including policies
areas concerned with R&D, and quality development. That is, all of the top-six
policy areas (with the somewhat exception of physical infrastructure) have an
important element of enhancing knowledge capacities at the firm level. Or in
negative terms policy areas concerning the environment, new firm attraction, and
venture capital that are less knowledge-oriented tend to score lower in terms of
relevance. Hence, a further implication of this aggregate pattern is that, in
general, cluster firms appear in need of more active policy initiatives that can help
them improve their knowledge base. The formulation of recommendations on
such matters should also bear in mind the particular knowledge base being
addressed by policy action. The former, as elaborated at the respective sections of
each cluster; appears to be highly dependant on the precise nature of the cluster
and the knowledge deficiencies of its firms. Such issues of knowledge-base
oriented policy action are also linked to the finding of this study concerning the
rather low percentage of knowledge sources deemed as important by the sample’s
firms; which brings to the next implication and recommendation.
Thus, the fourth implication stems from the set of findings relating to the
cluster firms’ knowledge sources. Their aggregate patterns indicate that the
sample’s firms rely more on internal than on external sources of knowledge.
Table 9.2 is rather telling in this respect. In addition the findings seem to suggest
that firms tend to consider as important far too few sources of knowledge among
all possible sources. On average, firms tend to consider as important around 36%
of all possible sources of knowledge. This undervaluation of knowledge is a
complex phenomenon that needs further examination. Although the fullness of
time may also point towards some additional implications there are still some that
should not be ignored at present, viz. the obstacles to respective increases appear
as having more to do with the availability of such knowledge sources (e.g. lack of
local publicly available information). To be more precise, the findings about the
firms’ underutilization of knowledge sources correspond well with the firms’
clear demand for more knowledge-enhancing types of policy initiatives for their
264 Cluster Policies in Europe: Firms, Institutions, and Governance
Policy focus on
directionality preferences
capacity
Directionality
policy areas
Adaptability
Top-four
MLG
share
Gap
MLG
size
Clusters
Herford furniture VH TdBu L M VH VH
Slovenia automotive H TdBu M H H H
Scottish ICT H TdBu L Lo VH VH
Saxony automotive H T VL M VH H
W Midlands automotive H T VL Lo H H
Prague film H T L M M H
Slovenia furniture H T M H Lo Lo
Rome film M TdBu L M M M
Bohemia automotive M Td L M M H
ABT clothing M Td S Lo Lo Lo
ABT footwear M Td M M Lo Lo
Val Vibrata clothing Lo Td S Lo Lo Lo/N
Notes: All rankings use the following five point scale: none/nil (N), low/small (Lo/S), medium
(M), high/large (H/L), very high/large (VH/VL); except MLG vertical directionality using a
three point: top-down (Td), bottom-up (Bu), or both (TdBu). Gap sizes refer to present while an
upward arrow indicates a future increase in excess of 0.5 on the five point scale used.
Sources: Authors’ elaboration, based on Tables 9.3, 9.4, 9.5, and 9.6.
Comparative Findings and Conclusions 265
Table 9.7 on the previous page brings forth the fifth implication of this study
relating to the importance of ‘bottom-up’ dynamics and the adaptability of MLG.
Four out of twelve clusters exhibit simultaneous ‘top-down’ and ‘bottom-up’
dynamics. The most important remark being that the combination of ‘top-down’
and ‘bottom-up’ dynamics seems rarer than ‘top-down’ dynamics alone.
Different national traditions of regionalisation and administrative decentralisation
in the countries studied are pertinent for understanding the shape of these
dynamics. However, it is worth noting that simultaneous ‘top-down’ and
‘bottom-up’ dynamics are found in clusters like the Scottish ICT, Rome film, and
Slovenia automotive, where administrative decentralisation is well-established
but not as extensively developed as in Federal systems like Germany. It is also
worth noting that the only two clusters in this sample exhibiting a combination of
‘very high’ levels of problem solving capacity and ‘very high’ levels of policy
focus on knowledge deficiencies feature both ‘top-down’ and ‘bottom-up’ MLG
dynamics.
The picture beyond this remark however is less clear. Firstly, as it can be seen
in Table 9.7, the Slovenian automotive and Rome film clusters exhibit such bi-
directional dynamics in the vertical directionality of MLG, but they only score
‘high’ and ‘medium’ levels of problem-solving capacity and policy focus on
knowledge deficiencies. Moreover, clusters like the Saxony, West Midlands, and
Bohemia automotive show ‘high’-to-‘very-high’ levels of policy focus on
knowledge deficiencies; although essentially of a ‘top-down’ nature. It is though
worth noting that these three clusters show rather adaptable patterns of MLG
dynamics. Even with this not-so-clear picture of the role of ‘bottom-up’
dynamics, there are still some important implications. They emerge on the basis
of the arguments put forward above, viz. the need for strategic policy-making in
contexts of scarce public and private resources. Admittedly, some of the cases in
this study indicate that ‘top-down’ dynamics suffice to support cluster-learning
dynamics. However, the development of cluster-specific policy strategies might
be more adequate, more user-relevant and more effective (in terms of being more
likely to provide ‘high’ and ‘very high’ policy focus on knowledge deficiencies)
if accompanied by ‘bottom-up’ dynamics. Due to these, one of the
recommendations of this study cannot but be that ‘bottom-up’ dynamics need to
be further stimulated in Europe. This is supported by similar recommendations
made almost a decade ago (Roelandt and den Hertog 1999b).
The previous five implications along with the findings summarised in Table
9.7 also lead to the sixth implication of this study, viz. the importance of taking
into consideration the role of MLG patterns for supporting cluster learning. The
findings indicate that there is considerable variation to this extent. As discussed in
section 9.6, five of the twelve clusters exhibit ‘high’-to-‘very high’ problem-
solving capacity and focus on knowledge deficiencies as two key elements for
fostering learning in cluster firms. These have been the clusters where public,
semi-public and private institutions have been very active in promoting collective
cluster strategies. This calls for a focus on procedural issues of strategy design.
266 Cluster Policies in Europe: Firms, Institutions, and Governance
As Rodrik put it: ‘The analysis of industrial policy needs to focus not on the
policy outcomes – which are inherently unknowable ex ante – but on getting the
policy process right. We need to worry about how we design a setting in which
private and public actors come together to solve problems in the productive
sphere, each side learning about the opportunities and constraints faced by the
other, and not about whether the right tool for industrial policy is, say, direct
credit or R&D subsidies’ (Rodrik 2004, p. 3).
Last but not least, this study’s findings illustrate the complexity, diversity and
dynamics of cluster MLG patterns in Europe, both in terms of learning dynamics
of MLG and in terms of stimulating cluster learning. The seventh and most
general implication of the findings is thus that learning is essential both for cluster
firms and for public, semi-public and private institutions, and that it takes
multiple and diverse forms. Hence, the final recommendation of this study refers
to the need to create and vitalize platforms stimulating such learning, in particular
those platforms that expose the cluster institutions to new insights and
perspectives. More specifically, the recommendation is to create and foster the
existing platforms for cross-cluster and cross-national learning stimulating the
exchange of experiences about the content and process of cluster-related policy-
making.
Finally, it also ought to be remarked, that although some of the studied
policies, institutions, and even clusters may have been products of explicit design
few if any of the studied MLG structures are products of such labour. As such
there is plenty of scope for their (re)design. Moreover, as the wider cluster
literature attests, the twelve clusters have no monopoly of the studied misfortunes
and successes. As such the scope of the implications and recommendations
discussed in this study may also spread to several other clusters, even beyond the
EU27 confines.
they will present an even larger puzzle for the understanding of the
knowledge-based economy, and more precisely for knowledge-based theories
of clusters (Breschi and Malerba 2005; Maskell 2005). Without challenging
the valuable insights of this normative and analytical concept, which is
largely shared by the authors of this book, future research efforts would need
to understand the current dynamics and limitations of European firms’
knowledge production and commercial exploitation. This is particularly in
relation to recent postulates about the US ‘open innovation’ paradigm
according to which most innovative firms in the US are increasingly relying
on external sources of highly advanced knowledge for keeping ahead in the
competitive edge, rather than on internal sources from their own R&D
laboratories (Chesbrough 2006). Such patterns and dynamics need careful
analysis, not least in terms of cross-national (e.g. US vs. Europe) and cross-
industrial sector comparisons.
Last, but not least, the exploratory findings of this study demonstrate that
MLG learning and support for cluster learning needs to be carefully
contrasted with future research involving larger samples and inferential
statistics. MLG, public action, and institutions provide frameworks that are
conducive albeit in varied degrees (cum positive/negative externalities) to the
performance of firms. Naturally, institutional frameworks alone cannot by
themselves make a cluster competitive, mainly because firms remain
invariably the productive engine of any cluster. However, as the findings
demonstrate, different degrees of institutional capacity seem to be associated
to specific vertical MLG dynamics. Hence, any subsequent set of research
questions ought to be directed towards the better understanding of the
mechanisms and associations between MLG and multi-level performance
(e.g. at the level of the firm, cluster, and the economy be it European,
national, or regional). This is to say that future research needs to
acknowledge from its outset the importance of institutions and the relevance
of policies, integrating them as an object of study in more ambitious research
designs. In other words, it is about time to start looking far more seriously at
institutions, policies, and MLG dynamics as part of the transition to the
knowledge economy and the never-ending search for competitiveness.
NOTES
1.
Several EU countries offer schemes for different kinds of consultancy or professional
services, e.g. in Denmark there is a start-up firm voucher scheme for basic services such as
legal or accounting advice; whereas in section 7.2 a number of SME programmes in the CZ
were reported to that effect (viz. MARKET, MARKETING, CONSULTANCY).
2.
Their order is relatively stable under alternative weights, e.g. averaging each set of criteria
on its own followed by averaging the averages of each set, and other variations thereof.