Association for Public Policy Analysis & Management

There is a growing literature documenting the increasingly prominent role of teaming in science. These trends have been largely attributed to new technology and the increasingly complex nature of frontier knowledge. Less attention has been paid to obtaining causal estimates of the effect of team size or how broader characteristics of network structure impact innovation, e.g. centrality or degree. Although recent work relying on publication and patent data has suggested that larger teams produce more output but that smaller teams are more disruptive, questions persist concerning selection and endogeneity. We explore these questions using big data on U.S. patent filings and scholarly publications from 2000 to 2015. We make progress towards casual estimation by developing a novel methodological approach and exploiting quasi-random variation in academic conferences as well as geo-spatial administrative borders. We also make use of UMETRICS data on sponsored research conducted at a subset of R1 universities. This data provides a rich picture of the broader collaboration network and helps us explore questions related to selection into authorship. From a policy perspective, the findings from our study are particularly relevant to inventors and firms engaged in science as well as to program officers making funding decisions.