Gender Difference in Innovation Recognition : A Textual Analysis Approach

This paper studies gender inequality in the recognition of innovations. Using modern natural language processing on over 1.5 million patents between 1980 and 2015, we find significant gender differences in inventors’ patent citations. Changing the gender structure of an inventor team from all-female to all-male increases the probability of omitting an all-female patent by roughly 15.6 percentage points. When we restrict the sample to citations added by patent examiners only, we find no gender difference. Exploring potential mechanisms, we find that the inventor gender gap does not vanish when citing and cited inventors become closer in collaboration network, when they work in the same firm, or when the cited inventors become more senior/prestigious, which are all evidence inconsistent with a pure statistical discrimination interpretation. Additionally, we observe large heterogeneity across different time periods and technological fields, but little heterogeneity across different firm tiers. Finally, we find suggestive evidence that racial disparity is also present in patent citations