A question of considerable interest in a world of tightened resources is the relationship between research outputs to research inputs. At the national level, policy makers want to know the degree to which more funding leads to more research. At the micro level, funding agencies want to know the degree to which research can be attributed to the funds invested in researchers. These types of questions are sometimes answered by relating the amount of direct funding investigators receive from a foundation or agency to the number of articles published in the next two or three years.
The approach of relating publications to agency funding (PAF) highlights two issues encountered in examining the relationship between research inputs and outputs. The first is one of attribution: exactly which articles should be attributed to what funding stream? In the PAF approach all articles published in the next few years are attributed to total agency funding received in a given year. Yet some articles undoubtedly result from funding received prior to the year being studied while others relate to funding received in future years. The PAF approach also assumes that all articles can be attributed to funding from one agency. Yet many researchers have funding from more than one agency.
The PAF approach also assumes that the manner in which funding is bundled has no effect on the productivity of the lab. There are two dimensions to this neutrality assumption. First, it assumes that it makes no difference whether a principal investigator (PI) has four grants a year that sum to $1million or one grant for $1 million. Second, the PAF approach implicitly assumes that output is unrelated to the composition of the funding portfolio in terms of the relative size of each grant.
The presented paper by Michele Pezzoni (Ecole Polytechnique Federale de Lausanne, Switzerland), Jacques Mairesse (CREST-ENSAE, France, UNU-Merit, the Netherlands, and NBER), Paula Stephan (Georgia State University and NBER), and Julia Lane (American Institutes of Research) investigates the attribution and neutrality issues, using data from the California Institute of Technology for the period 2000-2010. Our data are fine-grained and allow us to observe the size of the award, the source of the funding, and the length of funding at the PI level, thus creating a panel data base with three dimensions: PI, grant and time. We measure research productivity in terms of publication counts (QUANTITY) and the average Impact Factor of the journals in which the publications are published (QUALITY).
We find that the relationship between inputs and outputs persists after addressing attribution issues and estimating the relationship at the grant level. The simulations we run provide evidence that the neutrality assumption does not hold and that productivity increases the more highly concentrated is the PI’s grant portfolio. This is consistent with the presence of economies of scale in grant administration. It is also consistent with the fact that grants of small size are used to support more risky research agendas.