Comparing Historical Patent Datasets

Michael Andrews, National Bureau of Economic Research (NBER)

I compare the strengths and weaknesses of several historical patent datasets and compare the suitability of each for use in social science research. I show first that a number of historical sources exist that are nearly as complete as data on contemporary patenting. Second, I describe in detail differences across the datasets in terms of patent and inventor information included, reliability of provided information, and potential sample selection issues. Third, I show that while the datasets paint a remarkably consistent picture of aggregate invention through U.S. history, they contain important differences. In particular, differences in how patents are matched to location can produce statistically significant differences in estimates of the importance of agglomeration for invention. I discuss ways in which the different datasets can be used in conjunction with one another to improve inference.

No extended abstract or paper available

 Presented in Session 236. Patent Data and Institutions