Andrews, MichaelSmith, Alexa2024-08-072024-08-072023-10-31https://dx.doi.org/10.2139/ssrn.4591418http://hdl.handle.net/11603/35166How should research resources be allocated across space to have the most beneficial impact on society? Prior studies suggest that scientists are influenced by the local ideas they are exposed to, and hence local conditions shape the direction of science. We investigate this hypothesis in the context of agricultural research, constructing a measure that quantifies the extent to which land grant colleges are located in counties that grow different distributions of crops than the rest of their states, which we call agricultural unrepresentativeness. Consistent with the prior literature, land grant colleges in more agriculturally unrepresentative counties produce research focusing on unrepresentative crops and create more geographically limited productivity spillovers. Because college locations are not determined randomly, these results may reflect endogenous sorting by state policymakers rather than causal effects of local agricultural conditions. We isolate exogenous variation in land grant college’s agricultural unrepresentativeness using historical college site selection natural experiments. When using only this exogenous variation, we find no correlation between land grant counties’ unrepresentativeness and the unrepresentativeness of agricultural research. To understand this null result, we investigate actions land grant colleges can take to overcome the effects of local agricultural conditions and find that colleges exogenously placed in agriculturally unrepresentative counties invest more in extension services to interact with more distant constituents. We conclude that local agricultural conditions need not determine the direction of science so long as researchers can take actions to obtain non-local information.81 pagesen-USHigher EducationEconomic HistoryDirection of ScienceText