Measuring Innovation and the Propensity to Patent

Abstract

Patents are often taken as an indicator to measure innovativeness, because they are a lot easier to obtain than most others. There is a complication though, which is that patents are not a direct measure of innovativeness: the fit between patents and innovativeness is biased by differing propensities to patent. The propensity to patent was analyzed in the 1980’s and 1990’s, but results vary and more recent research focuses mostly at specific case studies. With the availability of the CIS (Community Innovation Survey) databases of 2000 and 2004 there is a good opportunity to test again, in a structured way, which factors are important for the propensity to patent. For this research CIS data from three North-Western European countries were analyzed: Belgium, Norway and Germany. Descriptive results show that the propensity to patent varies greatly among different types of innovation, and especially when looking at the difference between goods, services and processes. Process and service innovations turn out to be rarely patented, with average patent propensities of respectively 7.6% and 9.6%. Moreover, descriptive results confirm that the propensity to patent differs across industries. Next to that a logistic regression analysis was performed. This analysis tested existing hypotheses and explored new factors that came available in the CIS questionnaires. By only including those enterprises in the analysis that actually had innovative output, factors could be determined that increase the propensity to patent these innovations. Factors that were found significantly relevant include EU funding (+), having a new to market innovation (+), cooperation arrangements with universities (+), having a local/regional market (-), having a market outside Europe (+) and using private R&D institutions as information sources for innovation (-). On top of those it was confirmed that patent propensity is higher in specific industries and countries, as well as for different types of innovation, even when correcting for the aforementioned factors. Results provide a comprehensive overview of the importance of, and some correlations between, firm level, industry level and country level factors of the propensity to patent. These results can be used to improve patent based innovation measures, as well as to provide additional insights into appropriability conditions between sectors.