Measuring Innovation Using Patent Data

Abstract

Firms and governments spend billions of dollars on R&D every year. To increase social welfare, the results of R&D must be commercialized so that consumers can benefit from improved products and lower prices. One measure of R&D output is patents; however, most patent databases contain no information on whether patents have been commercialized, i.e., whether innovations have been introduced in the market. This paper applies a new method to identify innovations in patent databases by relating traditional patent quality indicators (patent renewal, patent equivalents and forward citations) to patent commercialization variables. For this purpose, I use a unique database on Swedish patents that includes information on whether patents are commercialized and whether the commercialization is profitable. The estimations show that commercialization is strongly positively correlated with both patent renewal and patent equivalents but only moderately positively correlated with forward citations. Further, successful innovations are most positively related to patent renewal. Based on the traditional patent quality indicators and estimated parameters in the model, probabilities of commercialization and successful innovations can be predicted. The developed parameters may be used to identify innovations across sectors and regions in other patent databases.

Keywords: Patents, commercialization, innovations, profitability, patent renewal, patent equivalents, forward citations, predicted probabilities

JEL Classification: O31, O34