MIT Sloan Management Review

materialize if they act alone. But such a relationship is unlikely to occur unless there is a sufficient level of trust to counter fears of abuses of confidential information and know-how. Contracts alone won’t help; in fact, they could inhibit innovation because they imply control of information flow and a range of legal dispositions that typically slow the project down.

Consider the case of Poclain (formerly Europe’s hydraulic shovel leader) and Potain (the leading French tower crane maker), which decided in the 1960s to enter the mobile crane industry by using their complementary technologies. The idea was that Poclain’s mastery of hydraulic circuits would combine with Potain’s competence in crane design and development to create new and improved machines. But the joint venture, under the name PPM, lagged behind because of scant innovation. The two partners’ management teams developed a system of mutual control and double signatures, even on minor expenses, necessitated by unrelieved mutual suspicion. In this atmosphere, the joint engineering projects dragged along for years with little success, delivering nothing more than “me-too” products.

Can Trust Become Detrimental? Given this example, along with many similar ones in the literature, it is tempting to suggest that higher levels of trust would have produced higher levels of innovation success. After all, trusting partners are more open, more supportive, less hostile, less competitive and therefore show more creativity — that is, generate more new ideas. 9 Of course, creativity is not innovation, which is the implementation of new ideas. But one might expect that more creativity would lead to more innovation in a higher trust partnership, which would lead to greater partner commitment and consequently, better implementation.

After close observation of numerous joint innovations, however, it is obvious that very high trust partnerships sometimes fail to be innovative. Fiat and Peugeot’s long and lucrative alliance in the commercial and passenger van markets serves as a useful example.

In the late 1970s, the two companies launched Sevel, a partnership to design and manufacture a commercial van that would be marketed separately as the Fiat Ducato and Peugeot J5. It met with immediate success. Sevel was able to exceed all

COURTESY OF MANITOWOC

expectations, exhibiting a production capacity twice that of the original plan. As a result, Fiat and Peugeot announced in 1988 the Sevelnord project, which would try to build on Sevel’s achievements and aim to compete in the passenger minivan market as well. The partners adopted the same governance structure, the same general economic principles of cost and investment sharing and even the same key executives as in the previous successful venture. After all, why fix what isn’t broken?

Both partners publicly hailed the excellent atmosphere of camaraderie they had established. “We find solutions together,” they reported. “Whether it is Fiat or Peugeot that has the idea — both groups come out ahead. There is an exchange of experience that is extremely rich.” 10 Unfortunately, after Sevelnord introduced the resulting passenger minivan — Peugeot’s 806 and Fiat’s Ulysse — in 1994, the product did not attain even half of its predecessor’s success. It was simply not a very innovative minivan and thus failed to gain much of a share in a market already dominated by competitors.

How could this happen, given the “extremely rich” possibilities? Although partners who trust each other may commit more of their resources to a joint venture, this does not ensure a high level of creativity, which requires a certain level of tension that may not exist in a high-trust environment. On the contrary, a high level of mutual trust between partners may result in soft, unchallenging and accommodating teamwork behaviors — the opposite of what is needed to develop creative solutions. In fact, recent studies do not find a positive link between R&D teams’ mutual trust and resulting creativity. 11

The Trust Experiments Case studies such as the above are not adequate for evaluating whether trust level is associated with higher or lower levels of innovativeness, as it is impossible to disentangle this factor from the many other contributors.

For this reason, we decided to set up a series of experiments. We enrolled groups with up to 30 players each, assigned them to as many as 15 pairs, and instructed each pair to design and build an object in the most creative way possible. Because we needed to ensure that the pairs represented a spectrum from very high to very low levels of trust, we chose individuals who already knew each other and who had sufficient