Inventive and innovative activity over time and geographical space: the case of Sweden

Two ‘stylized facts’ about innovations form the underlying motivation for this paper. First, various studies have found that innovations tend to be geographically concentrated. A number of theoretical propositions have been put forward to explain this phenomenon, most of them related to agglomeration economies. Secondly, innovation data based on patent counts indicate that most patents have very low economic value. It has also successfully been shown that ‘quality-adjusting’ patent data (by means of e.g. citations received from later patents) bring them a great deal closer to the innovation concept. Nonetheless, in the existing literature are such quality-adjustments entirely lacking geographical scope. Given that the value distribution of innovations is highly skewed, s confirmed by quality-adjustment, we expect quality-adjustment to make innovations even more unevenly geographically distributed. This paper examines whether quality-adjusted patents indeed are even more unevenly distributed across regions. This is done by comparing quality-adjusted patents using factor analysis with unadjusted patent counts. Contrary to expectations, taking quality into account seems to reduce innovation concentration. Therefore, basing measures on geographical innovation only using patent counts without taking quality into account, may therefore overstate the geographical concentration of innovation.