From Kyle Myers:
This paper estimates the degree to which scientists are willing to change the direction of their work in exchange for resources. Novel data from the National Institutes of Health is used to estimate an entry model that accounts for strategic interactions. Inducing a scientist to change their direction by 1 standard deviation, a qualitatively small difference, requires a four-fold increase in funds, an extra $1 million per year. But at current levels, the costs and benefits of directed versus undirected research appear to be quite similar.