Diversity is considered key to research, innovation and growth. However, throughout the literature, the exact meaning of the notion of diversity is often left in the midst. What is more, whilst the relation between diversity and productivity growth by now has been extensively addressed, still few empirical studies exist that address the relationship between diversity and innovation directly. To fill in this gap in the literature, this paper offers an empirical analysis of the relationship between technological diversity and the impact of innovation at the EU regional level whilst including different measures of diversity oriented at different levels of technological detail. Using EPO patent data, first, a set of different measures related to diversity is created exploiting the hierarchical structure of patent classes of the International Patent Classification (IPC). Second, the impact of innovation is captured by two citation-based indicators. Whilst the count of (field-normalized) citations is used to proxy the quality of innovation in terms of average impact, the number of highly cited patents is used as a proxy for top-quality innovation in terms of research excellence. Concerning the average impact of innovation, we clearly observe an advantage of specializing in innovation activities in few and related technological sectors. Localization economies can operate not only at low levels of aggregation, but also at a very high level, depending on the classification scheme. Results for research excellence are similar, with two main differences though. First, the impact of evenness on research excellence is non-monotonic in the level of aggregation and can be significant also for a very high level of aggregation. Second, while localization has positive impact on research excellence at the lowest level of aggregation similarly to average quality, a negative impact is found for all the other levels. The main conclusion of this paper holds that related specialization (conceived in terms of either similarity or related variety) and localization have a positive effect on the impact of EU regional innovation the more technological detail is taken into account. Our results then first and foremost stress the importance of taking into account (i) the relations among different technologies and (ii) the appropriate level of technological detail along which relations among technologies play out. As related specialization has a positive effect on technological impact, policymakers could consider either one of two policy options in order to boost the impact of innovation in Europe. One option concerns steering related specialization and localization at a coarse-grained technological level further as to increase the impact of innovation. If related diversification and not related specialization is considered to be a viable policy option in Europe, then the reward system of innovation should be drastically revised.
