Identifying the Age Profile of Patent Citations

A growing body of research uses patent citations to analyze economic phenomena, and many of these papers are interested in the distribution of citations over the life of a patent. However, this question leads directly to the age-year-cohort identification problem, i.e. co-linearity between the birth year, citation year, and age of a patent. Existing research has relied on functional form assumptions to separate these three effects. This paper proposes an alternative non-parametric identification strategy which uses the lag between application and grant as a source of exogenous variation. We provide statistical evidence to support our assumption that the citation clock should not start ticking until a patent actually issues, and we examine the potential bias introduced by our method if the lag between application and grant is correlated with citation levels. Finally, we use our proposed identification strategy to re-examine some prior results on the citation age profile of patents from different technological fields