Reputation and competence in publicly funded scientific research

This paper studies an unusually comprehensive data-set of the universe of research groups that applied to an Italian 1989 93 research programme in biotechnology and bio-instrumentation. We develop a model to test whether past publications influence the expected grants obtained by the units in this programme, and hence their publication output. We find that this effect is important, and suggest that institutional features of resource allocation in public research may have serious implications for long term divergence in research performance. We also find that to maximise the total expected publications of the programme about 85% of its budget should be allocated to the top quartile of the selected units (about 10% of applicants) ranked according to past publications. Our estimated elasticity of publications on budget is 0.6, and we find evidence of a trade-off between publications and "industrial transferability": all else held constant, projects conducted with industrial applications in mind produce about 20% fewer publications.