[Measure] Social preferences and the organization of production

This project will focus on the study of collaborative platforms in order to elicit and analyse the social preferences within various organizations and firms.

Most organizations rely on multiple motivations, rather than monetary compensation exclusively, to encourage their members to produce. This is especially the case for groups and associations aimed at providing public goods. But cooperation can also be an essential asset for organizations and firms. The sharing of information and effort between workers with complementary skills is an essential component of the efficiency of production.

However, measures of social preferences within organizations are scant. The literature on intrinsic and extrinsic motivations within organizations is mostly theoretical. The research has mostly examined the heterogeneity and sorting of workers by missions and motivations (Besley and Ghatak, 2005; Kosfeld and Siemens, 2011). An important result from this theoretical literature is the absence of pooling of preferences within an organization, despite the fact that workers’ motivations are heterogeneous. Theoretically, if “conditional cooperators” interact with “selfish workers”, cooperation will progressively erode. On the empirical side, most evidence comes from lab experiments (Bartling et al., 2012). A few papers look at representative real-world organizations. But they tend to use productivity under various incentives as a proxy for social preferences (Bandiera et al., 2005) or to use regional trust from subjective social surveys as a proxy for cooperative management (Bloom et al., 2012).

To open the black box of the measurement of social preferences within organizations, this project begins by taking a look at peer production, particularly  among open software developers. This is an ideal natural experiment to start off with because developers’ motivations are likely to be highly heterogeneous (ranging from intrinsic to extrinsic), there are numerous peer-production projects (Linux, Appache…) available to be studied, and observational data on production and productivity are readily available for analysis. It is thus possible to match social behavior with independently collected observations on production, at the individual and organizational levels.

Research article

Research team


Yann Algan

Professor of Economics

Sciences Po


Cathy Bénard

Administrative & financial management

Sciences Po, Dep. of Economics