Guannan Li

A small number of star inventors account for a large share of technological output, yet whether they raise the output of the other inventors in their technology class remains less well understood. This paper estimates the effect of star inventors on the innovation output of the other inventors in the same class across 124 technology classes in the United States between 1975 and 2014, using immigrant star inventors as the source of identifying variation. To address the endogenous allocation of stars across classes, I construct an instrument that interacts the top income tax rates of 17 origin countries, relative to the United States, with each country’s revealed technological advantage in a class. I find that immigrant star inventors have a positive causal effect on the innovation output of native non-star inventors. The average class-period has 1.3 immigrant stars, and 72% have none. Weak-instrument-robust confidence sets exclude zero for both outcomes and imply that a 10% increase in the number of immigrant star inventors in a class raises the citation-weighted output of native non-star inventors by at least 0.5% and their patent counts by at least 0.8%. The OLS estimates are positive and significant, and lie close to the lower bounds of the instrumental variable confidence sets. Spillovers onto native star inventors are six to seven times larger than spillovers onto native non-stars. Using an accounting decomposition, I show that immigrant stars directly produce 0.6% of class citation output, and that including their spillovers onto native inventors raises their contribution to 15.4%. By connecting the international supply of top inventors to the output of native inventors, the paper is relevant to policies designed to attract highly productive inventors.

Transportation infrastructure facilitates the agglomeration of economic activity and the diffusion of knowledge, yet its impact on average regional innovation quality remains less well understood. This paper estimates the causal effect of interstate highways on average innovation quality in metropolitan statistical areas (MSAs). I find that a 10% increase in interstate highways leads to roughly a 1.47% increase in average innovation quality, measured by citations per inventor, over a five-year period. To address the endogeneity problem, I use the 1947 interstate highway plan, the historical railroad network and old exploration expeditions as sources of exogenous variation to estimate the causal effect. Using a simple empirical model, I demonstrate that the impact of transportation on total innovation quality is primarily driven by average innovation quality rather than by the number of inventors. In terms of the mechanism, transportation can be treated as an amenity that attracts inventors to move to the region where the transportation stock is higher. New inventors in a region can be viewed as a form of research resource, bringing new ideas to the region and increasing its average innovation output. By connecting transportation infrastructure to the spatial distribution of innovation, the paper is relevant to regional and urban development policy.