The Varying Faces of Collaboration Networks


New research by Professor Maxim Sytch shows different industries create distinct collaboration networks.

Not all collaboration is created equal.

That’s according to new research by Michigan Ross Professor Maxim Sytch, who found that different industries create vastly different collaboration networks, and they tend to stay that way over time. That’s because the way individual companies approach research and development (R&D) collaborations with other companies impacts the collaborative network of the entire industry.


​Sytch and his co-authors studied the automotive, biotech, pharmaceutical, chemical, microelectronic, new material, and telecommunication industries to see what kinds of collaborative networks firms in these industries formed over time in pursuing joint research and development efforts.

It turns out that companies in more technologically dynamic industries — such as biotech and microelectronics — have open networks that depend more on individual companies forming research and development agreements with partners that are not connected to each other. These partners are more likely to come from different pockets of the industry and different geographies.

The picture in more technologically stable industries, such as chemicals or automotive, is vastly different. In such industries, companies enter in research and development partnerships with others that are well known to one another, through past or ongoing collaborations.

“If you look at a picture of partnerships between automotive companies and one between biotech companies, they look very different,” says Sytch, professor of Management and Organizations. “It’s because in some cases, the need for new knowledge in a short period of time overrides the safety and reliability of a more closed network. In other industries it’s exactly the opposite.”

Sytch’s findings with co-authors Adam Tatarynowicz of the Lee Kong Chian School of Business, Singapore Management University and Ranjay Gulati of Harvard Business School were published in the journal Administrative Science Quarterly.

Both types of networks have advantages and disadvantages. For example, automotive companies form partnerships with known, reliable partners who can vouch for and refer newer partners. Company secrets will be safe, and the penalties for violating trust and cooperative norms in this network are high. The drawback is that it limits knowledge of diverse information and can become an echo chamber.

The more open network favored by companies in more technologically dynamic industries is more likely to provide companies with access to novel, diverse knowledge and information. The drawback here is that violations of trust in partnerships are more likely to go unpenalized and are more risky. Because the focal firm would have no shared partners with the potential violator, it has limited ability to manage the relationship using possible reputational sanctions.

Sytch and his co-authors also find these individual firm R&D approaches have a drastic impact on the collaborative network of the entire industry.

“Industries tend to create networks that promote the viability of the collective and support the grand goal of the industry,” Sytch says.

For example, collaborative research and development networks in technologically dynamic industries help spread innovation quickly that others can build upon. The resulting networks have short pathways where many companies in the industry can reach one another through just a handful of partnerships.

In contrast, technologically stable industries — where knowledge diffusion and innovations are arguably less important — form more fragmented networks. They resemble groups of densely interconnected firms that rarely collaborate across groups. Knowledge and and innovations therefore spread more slowly.

These types of industry networks appear stable over time in a given sector. But it raises a question: As some of the less dynamic industries become more dependent on new technology, would they benefit from a different collaboration network?

“Think of the fast-developing autonomous driving technologies in the traditionally slow-moving automotive industry,” Sytch says. “How long will the fragmented network serve that industry’s collective purpose of trying to put the best and the safest driverless car on the market, without repeating one another’s mistakes and duplicating R&D efforts?”

Using this network view can help inform companies and policy makers how to facilitate the ideal connections for various industries.

“When we know what kinds of network structures facilitate learning and information sharing in industries, we can use joint ventures, industry associations, and industry-wide events to help create the necessary connections among companies,” he says.


Media Contact: