Science knowledge localizes

Science research and socio-economic advantage appear to co-locate geographically. As with technology knowledge spillovers, these benefits of science research are often ascribed to the localization of interpersonal science knowledge spillovers. Well-crafted causal designs have failed, however, to confirm knowledge spillovers in science or estimated the differences between technology and science spillovers. If published science did not localize – and was freely available to all – then firms would have little reason to locate near its source and nations little reason to fund science research. We isolate the mechanism of interpersonal spillovers and provide arguably causal evidence and quantitative visualization of the localization of science knowledge; following the unexpected departure of a scientist, and leveraging death as a quasi-natural experiment, papers without local co-authors receive 28 % fewer science paper, 45 % fewer medical paper, and 67 % fewer patent citations, from within a radius of 20 miles around the deceased, relative to still-living co-authors in other geographical locations. The effects first deepen and then attenuate with time, hold for within scientist estimations, and decrease monotonically up to 180 miles.