Allocating innovation resources to their most productive uses is a challenge for innovators because they have incomplete information about which projects will be most productive. I empirically study how a group of large scientific labs traded off the exploitation of existing opportunities versus the exploration of new ones, that is whether they pursued safe projects to maximize short-term productivity or undertook high-variance projects to acquire information and improve long-term productivity. To recover how these labs made the exploitation-exploration tradeoff, I estimate a dynamic model of decision-making, assuming the labs approximated the value of exploration with a simple index. The type of index is well-studied in theory and well-used in practice but has not been applied to estimation of empirical decision models. The index model captures the labs’ decision-making well. Estimates of its free parameters suggest that the labs explored extensively. Counterfactual simulations show that, had the labs not explored, their output quantity would have decreased by 51%, and their citations would have decreased by 57%.
Ansprechpartner: Michael Rose