LAUSR

We study the matching of patent applications to examiners at the U.S. Patent and Trademark Office. The distribution of technology classes is more concentrated than would occur under random matching and F-tests reject the hypothesis that family size and claim scope are randomly distributed across examiners. Using the application text, we show that examiner specialization persists even after conditioning on technology sub-classes. Specialization is less pronounced in computers and software than other technology fields. More specialized examiners have a lower grant rate. These findings undermine the idea that random matching justifies instrumental variables based on examiner behaviors or characteristics.