Exploring social preferences in private–collective innovation

Abstract

Firms, research institutions and individuals have long realised the advantages of innovating in an open manner. Companies such as IBM, Philips and Procter & Gamble, increasingly seek to cooperate with outside individuals and organisations to tap into their ideas for new products and services. This approach makes good business sense, but it is also difficult to achieve. In particular, mechanisms that motivate innovators to ‘open up’ are critical in achieving the benefits of open innovation. Private–collective innovation (PCI) has become an increasingly important model for explaining innovation at the boundaries of traditional, closed and open innovation regimes. Previous work has examined PCI both conceptually and empirically, and recent scholarship has focused specifically on the initiation of PCI as it relates to problems of collective action. This work shows that a project will not ‘take off’ unless the right incentives are in place for innovators to contribute their knowledge to open innovation from the beginning. Drawing on behavioural game theory, this paper expands on prior work by exploring social preferences in the initiation of PCI. The authors conducted a simulation study that shows how inequality aversion, reciprocity and fairness affect the underlying conditions that lead to the initiation of PCI. The results indicate that reciprocity and the potential gains for other participants explain changes in individual knowledge sharing in PCI.

Keywords:

• knowledge-sharing
• private–collective innovation
• game theory

Acknowledgements

This research was partly funded by the Swiss National Foundation Grant 100014_125513.

Notes

In our notation, player i denotes the leader and player j the follower.

According to Fehr and Schmidt (1999), α can be discretely distributed: 30% of the population has α=0; an additional 30% has α=0.5; another 30% has α=1, and the residuary 10% have α=4. These estimations are partly verified in the PCI model of Gächter, von Krogh, and Haefliger (2010).

The social parameter is drawn from a mixture of power-law distribution (70%) and zeros (30%). Long-tail distribution has previously been used to capture differences in social preferences (Maillart et al. 2008). We also introduce zeros to the participants’ distribution in a PCI environment to account for the free riders.