What to Know About Locating in a Cluster

There is ample evidence that geography matters for innovation. Innovation flourishes when workers are in close proximity to each other, where face-to-face encounters and job changes foster the flow of knowledge and the exchange of ideas. As Michael E. Porter of Harvard Business School and other scholars have noted, many companies choose to locate in industry clusters — regional concentrations involving a particular industry — on the presumption that they will gain an advantage in learning or in hiring workers with relevant skills and knowledge, and by being near suppliers and complementary businesses. Face-to-face interactions, formal and informal, frequent and repeated, become more valuable in an uncertain, complex world in which context is very important (what Eric von Hippel of the MIT Sloan School of Management calls “sticky knowledge”).

One of the most effective ways knowledge gets transferred is via people changing jobs. The knowledge might be know-how on tricky production methods or experience in conducting research. Businesses that impose noncompete provisions on their employees recognize this as “residual knowledge,” knowledge that a person cannot be prevented from taking upon leaving. That’s also why they target specific competitors when recruiting. In recruiting, local job switches are easier than long-distance ones — hence, the benefit of being in a cluster.

If the knowledge carried around in people’s heads is one of the valuable assets of a cluster, then tracking job changes should yield insights. Are there patterns that help companies be more competitive or respond to market threats? Does the makeup or the organization of a cluster make one cluster more attractive than another? How can a manager leverage the strengths of a cluster to maximum advantage? Are there any downsides to being in a cluster?

To explore these questions, we used LinkedIn’s Recruiter Lite tool to trace the career paths that many professionals display in their profiles. Professional profiles typically contain the individual’s current and past employers, education level, schools attended, industry, job function, and self-declared areas of expertise. We were able to sort people by experience level and present geographic location. We compared two Danish clusters: a vibrant life sciences cluster centered in Copenhagen, Denmark, and a faded wireless telecommunications cluster in North Jutland, near Aalborg, Denmark.

Case Study: Life Sciences

Denmark is home to hundreds of biotech, pharmaceutical, and medical technology research and development (R&D) centers that are strong in the fields of cancer, diabetes, inflammation, and neurological disorders. Nearly 1.8 million professionals in Denmark have LinkedIn profiles, more than 30% of the population. In pharmaceuticals and biotechnology, we found close to 33,000 profiles, with 6,219 listing their functions as research, 2,521 as operations, and 1,730 as program and project management. This represented about two-thirds of the Danish population working in pharmaceuticals and biotechnology.

We looked at the career paths of employees working at major global pharmaceutical and biotechnology companies, universities, and research institutes. We then studied specific subdisciplines, such as protein chemistry, immunology, and fermentation, and tabulated how many people moved from one organization to another over their careers. We input this information into a network-mapping tool to identify the community structure in the network. For major disciplines (for example, protein chemistry and immunology), subcommunities emerged that generally included Danish companies and Danish universities with a few links outside. Students at Danish universities largely did not go in great numbers to companies outside of the cluster, and Danish companies largely drew from Danish universities and other businesses in the subcommunity, with few researchers from abroad. Paralleling the situation in Denmark, Genentech, based in South San Francisco, California, drew heavily from local schools such as the University of California system, but hired few from Denmark. Researchers didn’t like to move very far. We mapped numerous specialties and saw similar patterns.

We found that it wasn’t unusual for production know-how to move between companies that were not direct competitors. For example, among 89 individuals employed at global health care company Novo Nordisk A/S who specified expertise in “fermentation,” 56 previously worked at industrial enzyme specialist Novozymes A/S, 28 at contract manufacturer CMC Biologics, 27 at food ingredient specialist Chr. Hansen A/S, 84 at medical devices company Coloplast A/S, and nine at Carlsberg A/S, the world’s fourth-largest brewing company. Review of individual profiles showed that 103 Carlsberg master brewers ended up in various pharmaceutical product development and production roles across the Danish cluster. NNE Pharmaplan A/S (formerly Novo Nordisk’s plant engineering arm) employed several Carlsberg master brewers as well.

Companies seemed to recognize the knowledge sharing, and they engaged in precompetitive collaborations with neighbors in different sectors to foster such sharing without loss of proprietary advantage. One such collaboration was Biopro, which brought together pharmaceutical, food ingredient, industrial enzyme, and energy producers to work together on advanced extraction, fermentation, and purification process technologies that all could use.

The movement of people among organizations in Denmark helped explain some of the behaviors we uncovered in our qualitative interviews. For example, private businesses in Denmark have played an unusually important role in basic research, which is normally considered a public good because private businesses have a hard time capturing a return from it and are therefore loath to invest. The Carlsberg Laboratory, established by the Danish brewer in 1875, maintains a world-class basic science research program. The lab’s charter stipulates that its activities be shared publicly, and since much of the resulting knowledge (in the form of trained people) stays within Denmark, the Danish public, and therefore Carlsberg, ultimately benefit. This was evident from the career paths of the 57 research scientists in our sample who had worked there. Most stayed within the cluster, leveraging their know-how on fermentation for employers in pharmaceutical R&D and manufacturing at Novo Nordisk, NNE Pharmaplan, Chr. Hansen, Novozymes, and so on.

People movement also helped explain how businesses in the cluster can meet the challenges posed by radical technological shifts. Novo Industri and Nordisk Gentofte, which merged in 1989 to form Novo Nordisk, got their start extracting insulin from pigs and cattle, exploiting their proximity to the Danish agricultural sector. In the late 1970s, when biotech company Genentech offered to share its method for producing human insulin using recombinant DNA (rDNA) technology via fermentation using E. coli, Novo Industri declined. (At the time, there was much wariness about the production or use of rDNA products.) When the company changed its mind, it didn’t have to look far. Mads Krogsgaard Thomsen, Novo Nordisk’s chief science officer, explained that Novo was very fortunate when it changed course, as it could recruit extensively from the University of Copenhagen and could draw heavily on other local businesses. As Thomsen recalled, some of the world’s best gene technologists and yeast physiologists were from Carlsberg. Novo also drew broadly upon skills in protein chemistry and fermentation, and it did a fill-in acquisition of a company with expertise in therapeutic and recombinant proteins. Today, Novo Nordisk is the global market leader in recombinant insulin, producing 50% of the world’s supply.

Denmark’s life sciences cluster benefits greatly from numerous shared resources. For example, the Novo Nordisk Foundation funds four major research centers at the University of Copenhagen, covering protein research, basic metabolic research, stem cells, and biosustainability. Novo Nordisk has an opportunity to hire the students trained there. When people stay within a cluster, companies have a good opportunity to reap returns from their investments. As explained by Else Marie Agger, a director at Denmark’s Statens Serum Institut, “Where else should you go? Within life sciences in Denmark, there are really not a lot of choices, and the number of relevant positions is restricted.”

Case Study: Wireless Communications

The experience involving Denmark’s wireless telecom sector has been starkly different. The telecom cluster in North Jutland pioneered the development of modern mobile telephone systems. The Nordic mobile telephone (NMT) system, designed and built from 1975 to 1982, was the world’s first multinational standards-based mobile telephone system. At the time, it had the most subscribers, biggest geographic reach, and highest adoption rates anywhere in the world.

Building NMT created demand for engineering talent from the newly established Aalborg University in Aalborg, Denmark, and from local radio engineering companies, including Dancom (later renamed Dancall Radio), a spinoff from SP Radio A/S, which specialized in maritime communications. At the time, the leading equipment suppliers included Ericsson of Sweden, Storno in Copenhagen, and Nokia of Finland. Another maritime communications company, Shipmate (a spinoff of Dancom), entered the NMT market as Cetelco in 1985. The Danish companies spawned numerous startups in and around North Jutland. By the end of the 1980s, the area was central to NMT development and deployment; several large multinationals, including Siemens and Alcatel, entered the market. In an effort to tap the specialized knowledge in the cluster, Motorola purchased Storno in 1986. Other acquisitions followed.

The success of NMT gave Danish engineers a strong voice in the development of the second generation of mobile phone technology (known as Global System for Mobile Communications or GSM), and the region became a mecca for telecom technology development. Dancall and Cetelco collaborated in creating the core platform, but they opted to manufacture phones separately. Soon, bigger companies such as Motorola, Ericsson, and Siemens came to Denmark to expand their operations there or acquire local businesses to be part of the thriving cluster.

By the end of 2000, about 40 telecom companies were in the area, employing 4,200 people. But beneath the surface of activity, there was trouble. Keeping up with the technical requirements of GSM development was difficult, and soon Dancall and Cetelco were each acquired by foreign companies. A bigger challenge came with the industry shift to third-generation (3G) mobile technology, which relied on expertise that local players didn’t have. Before long, large companies such as Ericsson and Siemens invested in 3G development centers outside of Denmark.

The former Dancall unit was subsequently sold to Robert Bosch GmbH of Germany in 1997, which later divided it and resold it in parts. After a peak in 2003, the number of employees and businesses in the mobile telecom cluster started to decline. LinkedIn profiles showed that, in contrast to people in the life sciences cluster, most wireless communications cluster employees went to other industries, many in industrial products.

Lessons From the Two Clusters

Our use of LinkedIn data highlights the direct movement of knowledge and expertise in a way that is timelier than more traditional trackers such as patent citations or publications. We can directly map the movement of skills when people declare areas of expertise, something they are motivated to do in their profiles. From this evidence we derived several insights:

Clusters with core platform strengths that span noncompeting sectors are exceptionally attractive. The Danish life sciences cluster is attractive because of generalized skills embodied in shared platform technologies like fermentation. This is shared across industrial sectors with minimal overlap: different pharmaceutical specialties, food ingredients, beverages, industrial enzymes, neutraceuticals, biofuels, and others. One contact at Biogen Inc., based in Cambridge, Massachusetts, told us that accessing fermentation skills was the major reason Biogen located a manufacturing facility in Denmark’s life sciences cluster. By contrast, the Danish telecom sector had a much narrower base. What’s more, many of the companies were in direct competition with each other. While competition in a cluster helps keep companies on their toes, in the telecom case, the foreign-based companies mainly extracted know-how and then moved on.

What makes clusters attractive also makes it harder for companies to control proprietary knowledge. As the two Danish clusters show, the very factors that can make a cluster attractive — easy people movement and knowledge spillovers — can also make it harder to control loss of proprietary know-how. Early on, when the life sciences companies were more open, there were many direct spillovers. According to Birgitte Skadhauge, vice president of research at the Carlsberg Research Center, a lot of the basic research in the Carlsberg lab dated back to the 1950s. She noted, “There were not a lot of patents, and people were not so protective as they are today.”

The unenforceability of noncompete clauses in California and other places where noncompetes are hard to implement provides additional support for the growth of the tech sector. The free movement of people fosters knowledge spillovers within Silicon Valley and encourages cluster growth. However, such movement can be a two-edged sword: It helps companies recruit for particular skills, but they run the risk of losing people they’d like to keep.

Companies can capture returns from investment in public goods if they have a geographic presence and are tied to the right subcommunities. The traditional argument against private investment in public goods (such as sponsoring R&D at universities) is the difficulty of capturing the returns. But to the extent that researchers have preferred geographies and stay within subcommunities, companies actually can have a very good opportunity to capture a significant portion of the benefits. One needs a combination of a strong geographic presence and being part of the subcommunities that regularly exchange employees.

Precompetitive collaborations foster healthy knowledge spillovers and enable building of scale. Once a company locates in a cluster, precompetitive collaborations are a good way to leverage diverse capabilities in the cluster. Choosing partners that don’t compete directly minimizes the risk, something that is easier when the cluster has diverse members that can share platforms. In the life sciences example, the members came from different sectors with little competitive overlap.

Healthy clusters have core institutions with scale and scope in relevant fields. The Danish examples underscore the importance of breadth of knowledge coverage. This scope can come from historical investments by universities or can be embodied in the diversity of local organizations. The idea is to absorb local knowledge and apply it to the problems at hand. Novo Nordisk was able to draw upon skills in recombinant methods and fermentation when it needed it. However, clusters must have the breadth to support future needs. In this regard, the Danish telecom sector came up short on technological expertise and resources during the transition to 3G.