Close to You? Bias and Precision in Patent-Based Measures of Technological Proximity

Patent data have been widely used in research on technological innovation to characterize firms' locations as well as the proximities among firms in knowledge space. Researchers could measure proximity among firms with a variety of measures based on patent class data, including Euclidean distance, correlation, and angle between firms' patent class distributions. Alternatively, one could measure proximity using overlap in cited patents. We point out that measures of proximity based on small numbers of patents are imprecisely measured random variables. Measures computed on samples with few patents generate both biased and imprecise measures of proximity. We explore the effects of larger sample sizes and coarser patent class breakdowns in mitigating these problems. Where possible, we suggest that researchers increase their sample sizes by aggregating years or using all of the listed patent classes on a patent, rather than just the first.