Working paper (with Alex Bell, Raj Chetty, Neviana Petkova and John Van Reenen) available upon request.
We match administrative tax data on income to the population of patent grants and applications
in order to analyze the lifecycle of 1.2 million US inventors from 1996 to 2012. Children of parents
in the top 1% of the income distribution are ten times more likely to become inventors than
those in the bottom 50%. Large differences in invention rates also exist by gender and race.
Test score data from third grade and above suggests that the majority of these differences are
not due to to innate ability but primarily result from the environmental conditions in which the
potential inventors grow up. Decompositions indicate that income-innovation gaps can largely
be accounted for by differences in human capital acquisition and early exposure to innovation via
families, parental colleagues and neighborhoods. These associations are evident for overall
patenting and persist at the detailed technology class level. We also document that the financial
returns to innovation are extremely skewed and uncertain at the start of a career. While our
analysis does not directly identify the causal mechanisms that drive innovation, our descriptive
findings shed light on which types of policy tools are likely to be most effective in sparking
innovation. In particular, "supply side" policies targeted at drawing more talented individuals
into innovation have great potential.