Abstract
The present paper builds a simple model of patent citations not based on the rich-get-richer aspect of preferential attachment. In our model the dynamics of citations are driven by known heterogeneities in the applicability of existing patents and aging. The model matches closely the hazard rates of citations for the vast majority of patents in a random sample of patents granted by the USPTO between 1975 and 1999. Furthermore, we show that the long run distribution of patent citations is well fitted when the distribution of applicability across patents follows a Gamma-distribution. We also discuss the possibility that popularity of patents might influence citation decisions if innovators are not perfectly informed about patents' applicability. We find that popularity matters but the size of the effect is very small. Finally, the possibility to distinguish between citations to patents within the same class and to different classes allows us to show that the magnitude of the influence of popularity is increasing in technological distance.