"The Cluster Initiative Greenbook: New Findings on the Process of Cluster-Based Economic Development", Christian H. M. Ketels
Rotman Competitiveness Seminar, November 21, 2003, 8 a.m.
Talk co-sponsored by the Ontario Institute for Competitiveness and Prosperity, and the Rotman School, University of Toronto
Christian H. M. Ketels, Institute for Strategy and Competitiveness, Harvard Business School
These participant's notes were created in real-time during the meeting, based on the speaker's presentation(s) and comments from the audience. These should not be viewed as official transcripts of the meeting, but only as an interpretation by a single individual. Lapses, grammatical errors, and typing mistakes may not have been corrected. Questions about content should be directed to the originator. These notes have been contributed by David Ing (firstname.lastname@example.org) at the IBM Advanced Business Institute ( http://www.ibm.com/abi).
Introduction by Roger Martin
Works with Michael Porter
When Roger Martin became the chair for the Ontario Institute for
Competitive and Prosperity, got help from Christian.
Has done research on cluster initiatives around the world.
Christian is originally from Germany, degree from London School of
Now lives partly in Boston and partly in Sweden
Institute in Toronto is a role model for the type of institution that Harvard
Institute wants to work with
Harvard Institute wants to use the network
How to use clusters in regional and national
Moving to the next level
Greenbook makes a contribution towards this.
Some confusion: people confuse clusters and
Will dissect this.
Over the last 10-15 years, since renewed interest in
clusters, how economic activity is located across geographies.
Now a sound body of research, foundations.
Understanding of how clusters work, and related to competitiveness, how
they evolve, how they create benefits.
So far, more of theoretical foundations, now moving to
As people became interested in clusters, policy-makers jumped on the
Larger context of how economic policy discussions have
Ten years ago, focused on macroeconomics, creating the legal
infrastructure of a market economy.
This has mostly been done
A good start, but not enough.
Policy-makers looking for something else to be done.
Old interventionist economics policies didn't work.
Thus, interest in cluster-based economic development
Use of clusters now moving more to the mainstream
Some have tried on a few projects, to see if it works.
Now the bar is rising: in policy advice, need to be much more
structured, use higher principles
What can we prove?
Will discuss greenbook, then what will emerge.
Clusters are intertwined with the concept of
May be thinking too narrowly about clusters: it doesn't fix
Need to think about clusters about overall business
Empirical work: what we are trying to prove?
Not employment, entrepreneurship
Prosperity as productivity: new things that customers will pay
This explains what we can afford as a standard of
Innovative capacity is important, particularly for countries like
On the forefront of creating new products and services
Need to come up with differentiated goods that others
This is why people talk about knowledge-based
Looked to U.S. regions for empirical proof
Chart: average wage (2001) and patents per 100,000 inhabitants --
What is driving productivity? What drives
companies to operate at high levels of productivity?
Macroeconomic, political, legal and social context for
Important, particularly in 1960s-1990s.
Need these: ownership rights, etc.
Also found, that this isn't enough
e.g. Argentina has stabilized economies, opened markets, but companies
Macroeconomic forms not sustained, because microeconomic foundations
In microeconomic foundations of development, two parts:
(a) Sophistication of company operations and strategy
(b) quality of the microeconomic business environment
e.g. supply chain initiatives:
(a) managers don't want to change, not adopting, even if streets and
suppliers are there to get them to a higher level of efficiency, then
productivity remains low.
(b) managers want to change, but streets are bad, or suppliers block
In some countries, a real imbalance
In Australia, a lot of improvement in the business environment, but
companies won't seize opportunity
In Germany, the other way around, where managers are sharp, but the
infrastructure is declining
In the Global Competitive Report, try to generate more
data to see if it's really true that the microeconomic factors matter.
Through a survey of business leaders, collect an aggregate measure:
what is the quality of the business environment, and the sophistication
of companies in the economy?
Most macro economic indicators are not related to level (although they're
related to growth rates)
No proof of causality, but suggested.
Canada still looks relatively strong, hasn't dropped as much as the macro
Canada is also about the regression line: GDP per capita is
higher than would be expected, based on the quality of the business
May be because of natural resource endowments.
Look at the business environment: the diamond
introduced by Porter
Lots of things matter: roads, taxes, etc. ... and it doesn't
The diamond tries to organize these, to cover the main areas that are
important for competitiveness
Factor (input) conditions
Context for firm strategy and rivalry: government (taxes, trade
barriers) and competitors (e.g. price wars)
Demand conditions: the ability to learn from customers, adjust to
regulations on demand, to pre-empt
Related and supporting industries: the presence of
Clusters are geographically colocated companies and
institutions, that have externalities from being colocated
e.g. Boston life sciences clusters.
They provide economic benefits
e.g. Novartis: Boston was the most expensive
place they could go to, but since they wanted to be a research leader,
they couldn't afford not to be there.
Other benefits overweigh cost of real estate and high wages.
Helps to save costs, if suppliers are next door --
particularly if they're specialized equipment
Can see what colleagues do.
Quicker, and learn more intensely if people are nearby.
allow for higher levels of innovation:
Heterogeneity of U.S. regions, in growth, productivity -- seen in
quicker to get feedback
cheaper: experimentaton is cheaper, if people
allows new business formation, creation of new
companies, than where there are no clusters available.
e.g. two companies have already failed, but I try again and fail,
there are other companies that I can go to -- preserving human
Boston life sciences, even though people thought about
Silicon Valley, or New York financial, or German automotive
People thought couldn't make a new cluster, but
there are many levels: in the U.S., 7 or 8 strong life science
clusters that all have different roles.
New Jersey, where large scale manufacturers
Boston, strong in research
Puerto Rico, where well-understood generics are done
Minneapolis/St. Paul where there are medical instrument
Can focus on function and/or geography
Good news: room for a lot of clusters in the
Don't need to be in Silicon Valley
For policy makers, Silicon Valley policies may not
Should decide on ambitions
Then, policy steps
e.g. a more efficient production site, rather than one that does
Declustering / outsourcing:
Lower trade barriers, costs, do drive a lot of economic
Data suggests that this doesn't mean fewer clusters.
Not just spreading everywhere, but instead moving to other clusters,
e.g. Italian shoe making companies, creating a new cluster in Romania,
bring their own suppliers and banks
Close to markets, roads are reasonably good
Clusters evolve, but that doesn't mean cluster effects aren't becoming
Clusters in different locations.
Another change in thinking over 10 to 15 years, is the
change in geographic levels
Porter was published on competitive advantage of nations
Actually, clusters are a regional phenomenon, not a national
The states and provinces are influenced and convene
One indicator: huge heterogeneity across countries, huge
disparity in economic performance, despite the whole region being subject to
the same economic policies
There are influences from all levels, and different geographic levels
play different roles
e.g. EU or NAFTA level
Have made a lot of progress in understanding these levels.
Move to what does this mean for economic policy?
Consistent with thinking about policy in a new way.
Government was in charge in infrastructure.
Companies are supposed to compete within this
But a lot of infrastructure is not in control of the
How companies and universities train.
How trade associations help (or don't)
As move from macro context to microeconomic, need a larger coalition of
A different collaborative setting
Government is more than one entity: multitude of
Regional government, state government, regulatory
In a lot of regions, there are new institutions that
play a new role.
Collective action problems: lots of companies, but who speaks for
e.g. Massachusetts, some very organized, some not so organized, some
traditional, some new
Work together to improve the quality of the business
Have found that regions that don't have these associations don't do as
well, because then it falls back to government, but government doesn't have
everything they need to make things work.
One more theoretical point, before moving into
data: clusters versus cluster policy.
Two different animals.
Implicit assumptions that led us to believe that clusters are a good
thing to do, four things we need to believe:
(1) Clusters exist, and externalities are strong
Good evidence of this
(2) Clusters provide economic benefits
Pretty solid evidence
Then move to policy
(3) Cluster development can be influenced
Data suggests this, but don't have enough supporting evidence to
(4) Not only important to influence, but need to be convinced that
cluster development has net benefits, creates more benefit than it
Almost as interesting as asking the question: can jobs be
What types of policy development do we have?
Think so, but no large scale empirical evidence.
Also need more theory around policy, as well as around just
Newer data on cluster benefits.
Then will discuss empirical reality: this is the core of the
At the end, can talk about policy
Empirical evidence, from the cluster mapping project
started by Michael Porter 3 to 4 years ago.
Using wage data by employment by region
Most previous studies tried to understand how clusters worked, and tried
to cluster industries.
This approach: if regions exist, then clusters would emerge, and
Look at colocation patterns
More information on the web site about methodology
First observation: a lot of industry is not
Call these local: local retail, local banks.
2/3 of employment is like this
Second, prevalent traded cluster areas
Industries that can choose where they want to be,
serving national or international
No reason that they have to be somewhere, e.g. natural resources
Third, natural-resource-driven companies, less than 1%
Chart: average local wage (2001) vs. average
If have good wages in the cluster, it improves wages locally
Demand-side portion that helps companies
Also reliance on strong local industries, e.g. particularly in
Found 41 cluster types:
Business services the largest
People say need to be in high-wage clusters
Looked at wage gap between region and U.S. average
Found for the median, almost 75% explained by the cluster
Only 25% explained by the mix
Thus, the problem is not being in the wrong cluster, it's that their
cluster isn't doing very well.
Based on this, shouldn't think about shifting clusters, but how to raise
wages in the local cluster.
Strong relationship between relative wage level (e.g. at the national
average) and the relative employment share (i.e. 1 means that same as U.S.
average, and higher numbers mean that you're specialized).
There are economies of scale, and get extra benefits from people getting
hired in a cluster, regardless of what cluster it is.
What share of employment is in our strong clusters?
The more people you have employed in strong clusters, the higher the
average regional wage.
Thus, a strong role for clusters.
What about change? Chart: state data
Right side: states that improved positions where they are strong,
and reduced where they are weak.
Strong correlation of increased wages where states improved where they
are already strong, than those that try to "flatten out"
Explaining average regional wages:
L.A., San Fran, New York
Seem to grow, even if no focus on clusters.
If only have one university, and they create all of the innovation, this
isn't a good sign.
More institutions are better
e.g. for patents: Boise city, have one
university that cranks out patents
e.g. Rochester, NY: lots of patents from one
Cluster breadth: each cluster has a lot of different
In how many industries, how many clusters have a strong
If positions in more types of clusters, then extra benefits, and it
helps the wage level.
The Cluster Initiative Greenbook
Almost no systematic information on cluster initiatives
Even case studies have a unit of analysis one cluster, not on the
initiative to make it better.
A lot of cluster initiatives are run by a few committed individuals, but
as researchers, have given them few tools.
First small step in Greenbook: try to take stock, with cluster
initiatives filling out a survey
Then, see how this relates to performance
The initiative for the book comes out from an association, and Swedish
Used the Harvard cluster network and Swedish network to start
This is a biased sample: more active, and people that we
Asked people from the cluster initiative, who have a strong view of what
they do, and their performance
Still, there's some heterogeneity in the data.
Also a bias in who fill out surveys, e.g. those who think that they've
Still, it's the first large-scale data collection
Flavour: web-based survey
Response rates 47%, highest in Northern Europe, Australia/New Zealand and
Two authors in Sweden.
New Zealand has a long history of cluster development
Will try to get more people throughout the world
Implicit understanding of cluster initiative
Not just a government agency, but also companies and university
Didn't test responses for these
Approach: conceptual model: there are
different things that can influence what a cluster can do
Setting: if the cluster is weak and the business environment is
terrible, could be a correlation
Objectives: what are they trying to do?
Process and structure: how is thing organized, governance
structure, who is financing
Performance: growth, did it create employment?
First, talk about descriptive statistics, then
Setting, context: strong environment for
Also tend to work with strong clusters, e.g. leading in the regional, or
leading in the world.
More heterogeneity in trust of government.
Some issues about the level of competition in the structure.
Wide distribution in age, when cluster was started
Objectives: What are they trying to achieve?
Found most cluster initiatives are trying to do a lot of things, not just
Also all cluster initiatives have a lot of activities in a lot of
Different from other types of policy development
List of 15 objectives, check off.
Networks were important, but lots of other
Almost 50% checked off 15 to 20 objectives
Active in at least five of six areas: cluster expansion, innovation
and technology, education and training ...
a process towards a goal
Process / structure: interaction between
Who started it all?
Distribution of started jointly, by government, by
In financing, government usually steps to the plate, more than companies
Not large budgets
But if you ask who are most influential, 70% respond
When did cluster initiative start? Timeline
If very young, the government pays a lot.
Over time, the funding becomes broader.
Objectives don't change much over time, thus young initiatives do as much
Cluster initiative facilitator
(May be an issue of the survey respondent, who puts self into a good
Still, these people are the few that have good knowledge.
Also, this may not be sustainable in the long term, the organization
relying on an individual or two.
Encouraging to find these people, but it's also a sign of
Most use a task force structure, identify most important initiatives,
then divide into subgroups
Lots of networking, but this may be driven by the sample
Physical infrastructure, budget: less than 70% say yes.
Not talking a huge amount of money:
$1M to $2M USD may be enough for a person for an office for a
Self-assessment in improving the success of the
Can't rely on cluster initiative alone, it's like an island
Trust in government is associated with more success.
Strong regional government
Cluster strength: more success if can work with a cluster that
already has potential.
Need a broad number of objectives
Still, can't find specific objectives that lead to
Objectives based on specific needs
Better infrastructure helps, e.g. money and an office.
If you don't have a clear strategy and goals, can't move
Didn't find a negative effect
Did find negative effect from limiting participation, e.g. not allowing
foreign countries, or if focused on small, not including medium and
Leading drivers of failiure:
Lack of consensus on cluster initiative activities
Then can't get people to work together
How can we build on this evidence to think about
cluster-based economic development more generally?
First, implications for cluster initiatives
Then, how do we think about microeconomic strategies?
Life cycle of a cluster initiative:
The dynamics of how the cluster initiative works, changes over
Moving from one stage to another, danger of things falling
Antecedent: period before a cluster initiative starts, often some
intelligence, but people haven't acted together, need to get people on
Formation: possibly from economic shock, or because a person
comes to the centre
Sometimes people analyze, want to understand where they are strong
Data brings people around, because they haven't thought about it
Less successful initiatives use this only as PR: how can we
show we're #2, as opposed to recognizing being #25, and want to
Hand-off from knowing about clusters, to doing something about
Will have a document, where all of data is collected,
Then need to move about from 5 to 6 person consulting team, to a
working group stage.
e.g. tech trends and training in Boston: then need people
responsible for licensing
Consultants would never have enough knowledge to ge this done,
can't give detailed advice about how to turn it
Starting as a project structure, need to always have new working
groups and putting other to rest
Thus, need a cluster-based initiative ongoing
Life cycle is important for practitioners to avoid falling into
Generalized: need to become more professional
How to conduct cluster initiatives, sound strategy and structure, can't
rely on people don't it in spare time, with CEO on a board when he has 20
other things to do.
Need to be careful about benchmarking:
Need to look at unique characteristics
Limited value to understanding how Nokia got started in Finland:
policies were specific to their circumstances, no way to understand how
they're relevant in another case.
Need more data:
A lot of cluster initiatives just organize and "know what we need to
do", would be better to have some data to work on priorities.
Need to be more systematic about looking at impact:
More and more requests from policy-makers, who want to convince
parliament, who want to see if money was well-spent.
As moving from clusters to cluster-based economic
development, clusters aren't a new policy tool
Evidence from Greenbook suggests the better way to think about this is
prioritization of what we need to do.
Broader issue: when looked at macroeconomics and legal environment,
there was a right direction and a wrong direction.
This is partially true in microeconomics, but there are lot of things
that could be done that would be positive (e.g. training, regulation of
consumer market, attraction of foreign investment), but then which are the
ones that should be picked for our region and cluster that are most
Prioritization important in microeconomics in a way that it isn't in
Cluster creation and cluster activatoin:
Close to economic policy.
Cluster creation: says that some clusters in the economy are more
valuable than others, if you want to be a rich and prosperous.
Thus, then want the cluster to locate there.
Government may then need incentives or protection, hope that will be
able to take it away in the long term.
Found track record of taking away is poor.
e.g. 41 states in the U.S. had a dedicated life sciences policy, and
Massachusetts wasn't one of them!
Cluster activitation: look at what you're already doing, get
together companies, and have them work together to bring the level for the
Clusters are important, but they're not everything.
Three key roles:
Engines for economic development
Can help understand what's wrong with overall business
If you ask companies, they might say tax rates, or something
Can be a new platform for people to work together in a
Should be seen as a part of an overall strategy
Politically problematic: helping one cluster, when you're not
helping 95% of the population.
This is a strategy for the region and national economy as a
Successful strategies move forward on at least these three
Still a young field
Trying to do more of mapping clusters, nationally
Have done in Canada, working on Sweden.
Want to understand economically disadvantaged regions: rural and
Focus on mapping clusters, little data on what makes them successful,
which means the business environment:
Trying to begin a cluster initiative, would you try to
define the boundaries of a cluster or measure them? Drive team?
A hard question that is less important than would think in the
When does a hill become a mountain?
It's more a map of the region.
Need to be open to anyone that things that they can add to the success of
a cluster initiative.
No systematic good answer.
Will vary by cluster, technology, region.
Rule of thumb: if you can get back there in one business day,
driving, then thinking about a cluster makes sense.
Where this matters is patent or concentration.
Could define those boundaries, e.g. one primary patenter, but would have a dozen
within a drive.
Cuts are defined by U.S. statistical boundaries, but not clear that
they're the right cuts.
Working with EDA.
If you're next to a urban area, how much does it impact? Ongoing
In Ontario, there are a lot of companies near Detroit
involved in automotive. Is this one big cluster, or clusters on their
Need to look at data.
All focused on Detroit region.
Probably makes sense to look at them as clusters in their own right, but
strong correlation to Detroit's success.
Within region, can't influence what will happen in Detroit, so makes
sense to focus on local regoin.
Where is the line between policy and cluster
Some policies are focused specifically on clusters: how to help
These may be too narrow.
Other issues, e.g. cross-cutting business environment things that
influence more clusters
Successful regions and clusters have a feeling for what we stand for in
More important at the regional economy level.
Too much focus on clusters are everything, right now, and
Has data shown a correlation between the peformance of
clusters, and engagement of universities with companies? Particularly with
Too little data for this.
Believe that in advanced economies, work on innovation, the model isn't
basic research in universities, and then companies bring to
Now more convoluted: universities and companies play at all
Seems natural to play a strong role.
In Canada right now, one of policy issues is
commericalization, particularly in university research. Clusters and
commercialization? Right now trying to understand knowledge transfer,
rather than commercialization. Currently a positive correlation. Do
you have some sense of needing to move beyond correlation with patent activity,
towards looking towards knowledge transfer towards commercialization?
Small tech-transfer patent path is a small part of the whole.
There are a small number of patents in the Boston area that cite other
patents in the Boston area, thus they're working together.
Need to have a broader picture in mind.
But there may be too small a focus on tech transfer, need to have to
overall picture in mind
How do we change the overall environment so that the knowledge in
universities is used more broadly across the region?
Define an institution for collaboration? Boards of
trade? e.g. Italy? Expansions beyond one or two person
offices. e.g. MARS project in Toronto.
Don't have larger-scale evidence to tackle this better.
There's other literature on networks, and cultures of cooperation.
Are looking at institutions, and they come from different place.
e.g. trade institutions, plastics industry in Colombia: can't just
lobby for trade liberalization and taxes, needed to work with other trade
Joined the supplier association, changing the agenda there.
Others develop from government projects
e.g. genetics in the U.S., still more in government.
In San Diego, it was more the university that was the platform for
Lots of different patterns for how they came about
All broader, trying to create a coalition
Education and knowledge creation as a cluster.
Does anyone study these clusters? Does this suggest that clusters can be
influenced? In this province, they're all public institutions, more like a
planned economy? Can you influence this cluster to develop?
Still a lot of room for research to understand.
Also, business services are huge, and need to be understood.
May be cross-economy clusters that influence a lot.
Had a broad cluster definition: some companies are related to more
than one clusters.
How clusters interlink -- needs to be understood better.
Can cluster development be influenced? Believe this.
e.g. Research Triangle in North Carolina is one of best success cases,
but it's taken a long time and a long commitment.
It sometimes creates benefits, but we can't be certain it
What interventions work?
Believe the role of universities is important, as we move to
Do you suspect that there are certain clusters or
industry, in terms of industrial organization (e.g. multiple actors or
consolidated actors) that impact on collaboration, or the success of the cluster
initiative? (No dependent variable, yet) Suspicions?
Yes, a strong hypothesis, but think that other factors are more
e.g. suppose have an industry with a lot of contracting, like footwear,
compared to life sciences: is one more likely to
In expanding clusters like sciences, may work together.
In contracting clusters, there's a need to work together.
No strong evidence, yet.
Other issues: global market or local market, so
that local companies would see themselves as collaborators. Relationship
to internal / external factors. e.g. software company that rely on
research institutions for flows in and out of people.
Importance of traded clusters, but doesn't really matter whether global
If you're a global company, you have global value chains.
If you have a national market, then might think people are more concerned
May be a secondary effect to culture or institutions.
If everyone believes that will only compete on price, then will be hard
to get companies to work together.
When differentiated, then easier to get people to the same
At http://competeprosper.ca , have 41 clusters
for Canada. First working paper has more information on clusters.
Second annual report will be released on Tuesday.
Some content on this website may be subject to prior copyrights.
Please contact the author(s) prior to reproduction or further distribution of the materials.