Should Regional Policies Try To Promote Clustering?

The state can’t make clusters

In economic policy it is currently very fashionable to support networks in order to increase growth. But it is doubtful whether the state will support collaboration between companies, as it is the rivalry and competition between them that creates dynamic clusters.

By Joakim Stymne
Chief Economist, Kaupthing Bank and chair of the think-tank Timbro

Translated by Phil Holmes

http://www.axess.se/english/currentissue/theme_stymne.php

On his journeys did he espy
how well Providence oft’ dictated
that rivers everywhere located
where great cities might pass by.

From "The Life of a Fool"
by Johan Henric Kellgren

AT THE STOCKHOLM SCHOOL OF ECONOMICS there was for many years a professor of economic geography called William William-Olsson. He was a professor of the old school who knew everything and frightened the wits out of his students. William-Olsson was, among other things, a pioneer in explaining how cities develop.

Illustrating the world by means of maps was one of his strong suits. He worked both on new map projections and on a graphic visualisation of the economy. One map of the structure of the Swedish economy, designed in the 1950s or 1960s, was especially beautiful. On the map colourful circles were drawn so that it appeared as if pins of different sizes had been stuck into it.

Each town had a pin whose size was proportional to the value added created there. The colour of the circle indicated the branch of industry which dominated the place. A quick glance at the map revealed where in Sweden different kinds of economic activity were being carried on. Red circles along the coast of Norrland represented the forest industry. Many black circles in the area around Lake Mälaren stood for mixed engineering. Blue circles along the coasts of southern Sweden for shipbuilding. One large green circle round the town of Borås for textiles with a lot of small green pins in the surrounding province of Västergötland. The interior of Norrland was mostly empty.

In his later years William-Olsson felt he had been misunderstood. He thought that the researchers into economic geography were rediscovering his old achievements, but were ignoring him by creating a whole new terminology. If he had been able to experience today’s cluster mode he would assuredly have pointed to his map and said, "There are your clusters—and by the way you’re lazy and have forgotten how to make good maps."

To cap it all, economic geography was for a long time a discipline in decline. Geography as a discipline did not seem able to deal with difficult issues about the nature of the world in an interesting way. It was one of the first subjects taught at the School of Economics, but now there is no department of geography there. At Harvard it is said that the only department ever to be closed was geography.

Thus it is a little ironic that the discipline of geography has now been given a new lease of life, not least with the aid of economists, business economists and politicians. Michael Porter, who popularised the concept of the cluster with his book The Competitive Advantage of Nations from 1990, is actually a professor at Harvard, even if he is in the university’s business school. At the department of social and economic geography at Uppsala university there has for a year now been an institute called CIND (Centre for Research on Innovation and Industrial Dynamics), which from the perspective of clusters is studying developments in industry, and which has attracted researchers from several different disciplines.

At CIND researchers Göran Lindqvist, Anders Malmberg and Örjan Sölvell under the heading "Swedish cluster maps" have produced maps of the geographical distribution of branches of industry across the country. The maps are admittedly less aesthetically pleasing than William-Olsson’s, but they reflect the fact that the economic geography of Sweden has changed quite considerably since he produced his map.

WHAT GEOGRAPHERS have always known, and wanted to investigate the reasons for, is that economic activities tend to clump together. Why are there towns? Markets? Why are factories sited where they are? And why do different activities sometimes cluster together geographically?

Traditionally we think of localisation factors in terms of closeness to markets or to raw materials or production factors. Johan Henric Kellgren’s fool was quite right when he observed that towns are often found on water courses, and quite wrong when he drew the conclusion that the rivers ended up where the towns were, rather than vice versa.

But there are examples of enterprises of a certain character lumping together without there being any obvious advantage in the form of raw materials or production factors. A popular example when clusters are being discussed is the carpet industry. An astonishingly large share of the world’s carpets is produced by companies located around the town of Dalton in the state of Georgia, USA. Another example is the concentration of manufacturers of semiconductors in Silicon Valley in California. In Belluno in northern Italy a cluster of 600 companies produces more than half of the spectacle frames sold in Europe and the USA. There are many more examples.

When economists have to explain why enterprises lump together they like to refer to externalities, different forces that exist outside the individual company and which affect its profits. They can be physical externalities such as the opportunities for transport (the fool’s rivers) and access to infrastructure, or institutional externalities created by various organisations, public activity and access to education. Social and cultural externalities exist in the form of attitudes and values, information externalities in the form of knowledge that can be exploited by the individual company.

At the same time as the individual company derives benefit from the externalities in the surroundings, it can itself create externalities, such that others can benefit without the company in question being paid for it. Successful companies are often characterised by the fact that they improve their products and processes. The aim of the company is of course to gain competitive advantage and earn more money. But a lot of knowledge cannot be patented and kept secret within the company where the innovation was originally made. When a company has made some innovation the knowledge of this floats about, as it were, and influences both suppliers and employees as well as customers. The knowledge becomes an externality which other people can derive benefit from without the inventor billing them.

A competing company in the same branch can get a free ride from the innovator’s knowledge and rapidly copy new products and processes. Empirical studies have shown that as only a small part of the value of an innovation accrues to the innovator, generally speaking too few innovations occur. If the individual company could gain sole control of an improved process and exploit all of the benefit from it, the company would invest in more innovations.

BUT WITH THE RIGHT KIND of interplay there can instead be an exchange between the companies, in which the transfer of knowledge goes in the opposite direction and the overall number of innovations increases more rapidly. The individual company can in its turn appropriate knowledge gained by someone else. The end result can be rapid local growth and a dynamic climate of innovation; competitors, suppliers and employees contribute to the growth of a cluster of companies competing and interacting so intensively that they are strengthened collectively. An important point is that this happens even if the companies involved are bitter rivals, and are not interested in cooperating, but are driven by a rivalry in which every company at any given moment is striving to gain advantages over the others.

Michael Porter’s contribution is to have concretised and systematised the economic thinking on what it is that drives the dynamics in an area. The Competitive Advantage of Nations maintains that traditional production factors have had their day. In his famous diamond model he focuses on four groups of factors influencing a cluster’s dynamics and ability to innovate.

1. Access to production factors such as labour and capital.

2. Demand conditions.

3. The presence of related or supporting companies in the local area.

4. Strategy, structure and rivalry.

What cluster research is saying is that it is not sufficient that some companies are located close to each other and are doing similar work in order to create a dynamic environment. The interplay between the factors is much more complex than that. One of the points in Porter’s model is that the concept of the cluster transcends traditional divisions of industries and treats entire environments with many types of interaction. For this reason William William-Olsson’s branch map is at bottom not a real cluster map: it shows the distribution in space of different industries but does not represent the dynamics that transcend traditional industry boundaries.

Pure chance can bring about the development of a dynamic cluster. In 1895 a nimble teenage girl called Catherine Evans in Dalton, in the American state of Georgia made a tufted bedspread to give as a wedding present. Many people thought the bedspread was fine and she was given orders for more than she could produce herself. Women in the area began to make textiles using the same technology and gradually inventions led to the mechanisation of the process. It soon proved that tufted carpets were considerably cheaper than woven ones. Production grew rapidly and brought in many peripheral industries. Today Georgia is responsible for almost half of the world’s carpet production. What region would not wish to repeat successes like these?

But it is necessary for Porter’s factors to interact well in order for a cluster to take off. Professor Örjan Sölwell, who works with Michael Porter and who has applied his analysis model to Sweden, points out how deficiencies in the environment can impede the development of a cluster.

"Both the Audi and the Trabant came from the same company, Autounion, after the Second World War," he said in a lecture to a cluster conference last autumn. "But the Audi developed into a wide and dynamic cluster, while the Trabant originated in a thin and static cluster."

Even though one could talk of a car cluster around Audi production in West Germany and another around Trabant production in East Germany, it is obvious that the East German environment suffered from many shortcomings which meant that the cluster was left hopelessly far behind.

A VERY WELL-KNOWN cluster is Silicon Valley in California, which for several decades has been a dynamic centre for the production of semi-conductors and other computer-related activities. Silicon Valley, even more than the carpet cluster in Georgia, has been the dream of other regions: high tech, rapid growth and beautiful weather (even if only the fool himself would believe that a successful cluster improves the weather). All the different "valleys" existing around the world today are trying with varying success to emulate the development of Silicon Valley. This is despite what AnnaLee Saxenian, who has studied the Silicon Valley cluster, says with her co-authors in the essay "’Old economy’ inputs for ‘new economy’ outcomes" (2001): "Where Silicon Valley is entrepreneurial, decentralised and only loosely and flexibly connected to broader national institutions, many efforts at imitation have been government sponsored, top down or tightly linked to established firms, perhaps ‘national champions.’"

The commentary takes us from the cluster as a more or less spontaneously arising dynamic environment to the cluster as a desirable creation which will be shaped by political actions. And a great many people want to create clusters, or at the least to gear up existing accumulations of activities into something much better. A cluster is transformed from being an object of analysis into an initiative once an organisation is formed around it. Cluster initiatives are found in every continent except the Antarctic (although Christchurch New Zealand actually lists "Antarctica" as one of its clusters).

Last autumn the Competitiveness Institute, the international organisation for "cluster practitioners" (those people who in one way or another are involved in cluster initiatives) held a major conference in Göteborg. At this conference Örjan Sölvell, Göran Lindqvist and Christian Ketels presented their Cluster Greenbook, a study of 250 cluster initiatives worldwide. They state that a cluster initiative is typically initiated by one individual "clusterentrepeneur" and is in large part publicly financed. It also appears that cluster initiatives are most popular among relatively technically qualified areas: IT, medical technology, engineering, communication technology, biotechnology and automotive-related manufacturing.

Many countries are far ahead of Sweden as regards clusters. In Denmark, cluster policy has been a priority area from as early as 1993. In the Italian region of Emilia Romagna what are now called cluster initiatives were begun as early as the 1970s. New Zealand has an enormous number of cluster initiatives. The state of Catalonia has supported twenty or so clusters over the years.

But now Sweden is also getting into clusters. In Sweden there are primarily three authorities who invest in clusters: Vinnova (Verket för innovationssystem, Swedish Agency for Innovation Systems), Nutek (Verket för näringslivsutveckling, Swedish Business Development Agency) and ISA (Invest in Sweden Agency). The government has provided SEK 70 millions to these authorities in order jointly to set up programmes with the name Visanu (National Programme for Development of Innovation Systems and Clusters), which have three years to support the development of clusters across Sweden.

In a survey carried out by Visanu last year upwards of 250 cluster initiatives were found in Sweden alone, many with witty names. There are for example TelecomCity in Blekinge, Microwave Road in Västra Götaland, Trätårtan (The Wooden Tart) in Kronoberg, Biotech Valley in Södermanland, Mobile Valley in Stockholm, Triple Steelix in Dalarna, Fiber Optic Valley in Västernorrland and Internet Bay in Norrbotten. Maria Lindqvist, project leader för knowledge development at Visanu, tells me when I spoke to her, that, however, the number of active cluster initiatives in practice numbers between 50 and 60.

Vinnova has also initiated a project under the name Vinnväxt which will pay out SEK 400 millions to cluster initiatives. In the first phase three programmes have been selected to receive 10 million each for 10 years: Robotdalen (Robot Valley) in the Mälaren Valley, Innovation in Gränsland and Uppsala Bio.

In addition to this Swedish regional policy is in a state of change; the correct nomenclature is nowadays "regional development policy." The government has instructed each county to produce "a programme for growth." These so-called regional growth programmes began to take effect from 1st January this year and will be the central tools in the long-term work towards sustainable regional development. In December the government decided to provide just over a billion kronor to the counties to support this policy.

THE CLUSTER APPROACH has a more or less prominent role in all the regional growth programmes (with one exception, see below). To take some examples from the county programmes: In Kronoberg County there is great enthusiasm and innovation systems are prioritised among their investment areas. The regional growth programme is crawling with cluster arguments and there are a number of ambitious cluster initiatives. Aluminium, heavy vehicles, Tränätverk Syd ("Timber Network South"), Internet City, media/entertainment and Glasriket (the Kingdom of Glass). Unlike the majority of other growth programmes Kronoberg has put figures to its plans for investments: of SEK 110 million during 2004, 37 million will go to cluster investments.

Västerbotten is another enthusiastic county, and reports clusters in areas of rock and minerals, timber and testing. But the county also considers that they, as they put it, "show competence" in further strong cluster formations and innovation systems, mentioning optimistically biotechnology, hydraulics, forest machinery, foodstuff development, sheet metal treatment, engineering, GIS, IT and "a host of other technologies."

Norrbotten emphasises the car and component testing cluster at Arjeplog (where car manufacturers test their cars under difficult winter conditions) as a successful cluster, but say modestly that whether "those activities relating to the ice hotel in Jukkasjärvi qualify for inclusion in the concept of a cluster is more doubtful." Norrbotten also writes, however, that these presumed clusters are "two striking examples of innovative and successful exploitation of natural resources that abound in Norrbotten… snow, ice, sparse population and vast expanses." The formulation indicates a certain gallows humour on the part of the county council up there.

Jämtland talks modestly of the fact that in "Jämtland County there are cluster-like phenomena" around tourism, sport and leisure activities, and wonders whether it is possible to develop clusters round all centres. In a clear-sighted formulation on how one might acquire co-funding of existing national programmes and their regional growth programme, they say that "from Nutek and Vinnova it is primarily the cluster development programme, the entrepreneurship programme and development of learning systems which will be in demand."

Against the background that it is easier to get money for the county there are good reasons for looking for clusters.

Gävleborg expresses a faith in politics creating clusters. They want to "survey ideas for new clusters that are assessed as creating regional growth," "develop focus areas and new clusters" and ensure that "existing clusters are developed such that they gain international competitiveness." Södermanland does not really believe in this however, and says explicitly that "we cannot create clusters."

So to the great exception: Halland. Of all the regional growth programmes Halland stands out in that the word "cluster" is never mentioned. Reading Halland’s growth programmes after having been fed with more or less convincing cluster terminology from North to South is refreshing.

For this reason I ring up Sten Liljedahl, trade and industry director for Region Halland. "It is deliberate," he replies, when I ask why the cluster concept is not employed. "You shouldn’t use concepts people don’t understand. This is fine if you are applying for money from Stockholm. But it is the businesses that will be creating growth. Those companies who are involved in the investment will earn money from it. The authorities can facilitate the birth of clusters but never actually create them. We have to look to the end user in what we do. It is the needs of the companies that matter. Our role in the system is to make it easier for them to do business. All the time we have to complement the market where necessary, not compete."

If we look more closely at Region Halland’s growth programme we can say that instead of clusters it contains three "growth areas": health technology, fostering experiences and foodstuffs. For these they plan to apply SEK 23 millions per year out of a total 174 millions in the growth programme. And despite (or as a result of?) the absence of clusters already mentioned the programme was assessed positively by the government. Halland was one of six counties that received an extra allocation for 2004 with the explicit motivation that they had succeeded well in their strategies.

THE THREE AUTHORITIES Nutek, Vinnova and ISA want further top reinforce cluster ambitions. Very recently the authorities three director-generals published a booklet entitled Vår syn på tillväxt ("our view of growth"). Here they argue for basing a modern trade and industry policy on the view that growth arises through the interplay within clusters, and that for this reason investment is needed to stimulate clusters and innovation systems.

There is, therefore, no lack of ambition and involvement as regards the cluster approach. It gained a more systematic foothold in practical politics in connection with last year’s work on the regional growth programmes. It is being pushed hard by these authorities, and the idea is said to be gaining a much stronger foothold among individual members of the Swedish parliament. But can a successful cluster really be created? Is it a good idea to do this? It is possible to distinguish three attitudes to this question:

Market liberal. The cluster focus is a way of shifting attention away from what is really important in politics, namely creating the best possible conditions for enterprise.
Market improving. Cluster policy can be a good way of correcting market failures and providing further impetus to an already dynamic environment.

Market denying. Cluster ventures are good because cooperation and planning is better than the market at achieving economic dynamics and other prioritised political goals.

Porter’s diamond model can be used to illustrate the first two points of view. He identifies those mechanisms that create development impetus. One example is the capital market. This forms a component of the corner of his diamond named production factor conditions. In an ideal world there should be access to sufficient start-up financing; new businesses must have a chance to get started. A common notion is that the market is too mean in providing start-up financing. This is therefore a failure on the part of the market, leading to too little investment being made in the proposed cluster.

THE DIRECTORS-GENERAL of the three cluster authorities think that the state has the job of putting right this market shortcoming. In their recent publication mentioned earlier they regret that the tendency of the private venture capital firms to provide start-up financing has diminished. "Let the state resume its role within start-up financing" is their suggestion for improving the market.

But they are ignoring the fact that there is very wide-ranging research with massive micro-econometric evidence that the creation of new business in a community is firmly linked to the existence of private fortunes. That is to say that the entrepreneur either earns his start-up capital himself or that someone wishing to start a business approaches a rich relative or friend, rather than turning to formal, anonymous capital markets. At the same time there are large question marks hanging over how successful state attempts at start-up financing have been.

The market liberal conclusion is that poor provision of start-up financing is a political failure rather than a market failure. The task of politics would then not be to provide financing, but to carry out changes in taxation and regulations to facilitate the formation of capital.

Another example that might be seen as a market failure is lack of information, for example about markets, production processes and so on. Cluster initiatives often involve different forms of collaboration which mean that the businesses affected gain new meeting points to exchange information.

But nor is this unproblematic if it is the cluster dynamic that you want to promote. Remember that the strength in the clusters is very much the result of rivalry between the companies; they are competing with each other, steal each other’s staff etc. Mechanisms increasing collaboration and reducing rivalry can weaken the pressure to compete and thereby the dynamics of the cluster.

It is impossible to say to what extent organised collaboration in cluster ventures in practice reduces the pressure to compete in the clusters. If rivalry is the mechanism of the spontaneous cluster in order to deal with a lack of the will to innovate within the private firm, then the county councils will risk weakening the cluster dynamic when they increase the formal collaboration for exchange of information and the like. But one should note that the county councils in their regional growth programmes do not discuss how rivalry within the cluster should be encouraged to increase rather than decrease (allowing for our not having missed a phrase somewhere). Against the background of the central role this assumes in Porter’s model, it would not have been out of place to have at least reflected on this problem.

A third example is how cluster ventures deal with the government’s demands that they shall also favour equality of opportunity and diversity. It is a plausible hypothesis that a lack of equality of opportunity is a market failure, and that more female business enterprise and greater diversity in enterprise would contribute to increasing the dynamic within a cluster. But why are there weaknesses in equality of opportunity?

In Västerbotten there is an organisation called Emma Resurscentrum (Emma Resource Centre) which has carried out several ambitious studies of women’s place in cluster ventures.

Emma Resurscentrum says that its work starts from a feminist premise. "The perspective is that there is a general hierarchy of power in society where women are subordinated to men and that this hierarchy should be broken." This perspective leads to Emma Resurscentrum concentrating to a great extent on the power of language over thought. It claims for example that entrepreneurs are ascribed typically male characteristics and that Nutek in this way creates an implied justification and idealisation of male homogeneity, which would make women less disposed to engage in cluster ventures.

It is also said women are overrepresented in branches such as child care, and care of the elderly and those with handicaps. But nowhere does Emma Resurscentrum discuss the problem that women are in this way relegated to work in which private enterprise for a long time was, in practice, prohibited and is still to a great extent impossible other than on a very small scale. To facilitate the private production of welfare services, expressly allow the profit motive for this and thereby open up women’s area of competence to enterprise and competition seems to be more fruitful than to try to get Nutek to describe entrepreneurs in less masculine terms.

THERE IS however an even greater problem with cluster ventures than the risk of a faulty analysis of which market shortcomings that might motivate them. When I speak to Michael Enright, an American professor of business, working Hong Kong and one of the founders of the Competitiveness Institute, he develops this as follows:

"It is possible to create successful clusters. The initiatives work best in areas where there are good conditions for enterprise. But you have to be clear about precisely what it is the market cannot provide before you start. Unfortunately the cluster branch is among many things a last place of refuge for old Marxists and people who see local initiatives as an alternative to a globalised world."

In this way we got into the third category of cluster motives: market denial (referring to those who do not believe in the market economy, but see publicly driven collaboration and planning as central to a desirable development of society). In the context of clusters market denial has the perfect ideological justification in something that is called the "triple helix."

In practice the triple helix is nothing more remarkable than a plea that the state, trade and industry and universities should cooperate, preferably regionally. Through this collaboration a cross-pollination takes place which leads to better growth. This trivial (if not uncontroversial) conclusion builds on a grandiose theoretical construct by sociologists Harry Etzkowitz and Loet Leydesdorff.

Anyone who does not like motivating public contributions by careful analyses of market failures can safely read the triple helix literature. Both Etzkowitz and Leydesdorff are deeply rooted in traditions far from mainstream economics. Nor do they have a model like Porter’s, which complements accepted economic theory with careful empirical work. Etzkowitz has his roots in the Marxist radical sociology movement of the late 1960s. Leydesdorff has written a book in which he develops a programme for a postmodern, deconstructionist research method for sociology.

In the following extract Professors Etzkowitz and Leydesdorff summarise what the triple helix is about.
The Triple Helix model extends the basis of structural-functionalism by introducing the notion of "meaning" from symbolic interactionism: social functions are discursively constructed, and they can be deconstructed and reconstructed reflexively. Thus, one can no longer accept a dialectics between ahistorical functions and historical institutions. The institutions are needed to carry out the functions, but they can be expected to be changed while doing so. The functions are continuously under reconstruction and the institutional elements of the systems have been generated by these reflexive operations. This is the work of charlatans. Such postmodern sociological communication theory should not be let outside the seminar rooms of academic sociologists and be allowed to form the basis for Swedish economic policy. It is this kind of theorising which without blushing would claim that geography is a social construct and that the fool’s assertion about rivers and cities is just as true as any other localisation theory, as it is language that creates reality.

The triple helix idea has for some reason become very popular in Sweden. In practice it seems, however, not to deal with much more than the fact that representatives for both the authorities, companies and universities should meet and talk and maybe agree about things. The problem is that the wooliness of the triple helix model makes it impossible to ask the question which problems a cluster venture will solve that the market itself cannot deal with.

One example is the role of the universities. Why is it universities rather than education at lower levels that should take part in the cluster collaboration? Sten Liljedahl in Halland specifically mentions that academic research is not central to success in Halland’s "non-cluster." But Etzkowitz and Leydesdorff’s torrents of words do not help the clusterpreneur to make such assessments: the collaboration is presented as desirable a priori.

Despite all objections there is one important factor supporting cluster focus in regional policy: it could have been worse. Cluster ventures are nevertheless more market-oriented than a lot of earlier trade and industry policy. By concentrating on dynamics and growth the cluster ventures imply a welcome shift in focus in regional politics, which used mostly to be about defensively facing stagnation and decline or grandiose industrial investments in the wilderness.

In Porter’s model the existence of a cluster venture is only one of 25 different factors creating a dynamic environment. Örjan Sölvell likes to talk of the cluster venture as an oilcan used on dynamic development. Maria Lindqvist, mentioned above, project leader for knowledge development at Visanu, underlines the flexibility of cluster ventures.

"You have to distinguish between organising and organisation. Better to build on what exists than to build up organisations with offices, MDs etc. We construct networks, not buildings."

OF COURSE regional grown programmes are characterised by cluster ventures sometimes more out of pious hope of a better future than from any hard analysis of realistic competition. But they undoubtedly start to a greater extent from an insight that the economy requires real entrepreneurs than many earlier trends in regional policy have done. And that is positive.

At the same time Maria Lindqvist provides a disturbing answer to the question of whether the Swedish cluster ventures are usually carried out according to Porter’s model.

"No, it is primarily triple helix that is applied," she says.

To return to the example with the Audi and the Trabant, you can see there the difference in the basic conditions. The Trabant was produced in an economy with factor markets working inefficiently, without the profit motive, without competition, with poor local demand, the lack of sophisticated distributors, and so on. There is no cluster initiative in the world that would have been able to contribute to a dynamic cluster round Trabant production without first changing the conditions for enterprise. That is to say abolishing the "actually existing socialism" and setting loose the forces of chaotic capitalism.

End

JOAKIM STYMNE
Chief Economist, Kaupthing Bank and chair of the think-tank Timbro

Axess Magazine Blomma Copyright 2004 © Axel and Margaret Ax:son Johnson Foundation for Public Benefit.
All Rights Reserved.