Essays on the economics of science and the communication of science

Abstract

In these essays I explore in theoretical models how research is produced, potentially falsified and communicated to interested third parties. The aim of a researcher is always understood to be to present himself or his research in the most favourable light. However, this imposes limitations on his credibility, as third parties anticipate potential falsification. I investigate this tension first in an individual interaction and then in the framework of competition between researchers. In the first chapter, I consider a pharmaceutical company that tries to persuade a regulator to approve a drug by presenting verifiable evidence about its quality. The company knows the quality of the drug and always wants to get it approved. The regulator only wants to approve drugs of sufficiently high quality, but does not observe the quality. The pharmaceutical company generates evidence from a costly, sequential testing process. I contrast the case where the company can suppress unfavourable evidence to the case where it has to report all evidence obtained. I show that the pharmaceutical company prefers the possibility of suppression when the regulator is already close to approving without evidence. The regulator always weakly prefers no suppression. In the second chapter,I consider the the effect of competition among researchers on falsification. I develop a simple matching contest model to study the replication crisis in the empirical sciences. Scientists can choose the risk of their projects, riskier projects are less likely to succeed, but more likely to be published if successful. Scientists can also falsely claim that projects have succeeded, at a risk of reputation damage. I show that competition among scientists leads to riskier projects, but also a more congested publication process, where the rate of publication among equally valuable projects declines. Riskier projects lead to more falsification, but congestion decreases falsification, giving rise to an ambiguous effect of competition. I also show that a policy of targeted detection of falsification in fact leads to more falsification in equilibrium.