testing current theory on value drivers of innovations within a structural two-stage discrete choice simultaneous equation model

Patent indicators are widely used to assess innovative output. Despite the large variety of empirical studies in the field, however, the precise meaning of these indicators and their obvious relation to patent value is still based on assumptions and intuitions. This paper provides the first empirical test of patent indicators as value measures in the structural form. It disentangles the different effects reflected in patent indicators and enhances our understanding why inventions are valuable at all. Using a newly assembled data set on European polymer patents, current assumptions on the innovation incentives set by patentability requirements (novelty, inventive activity) are tested. The estimations are carried out using a custom-tailored two stage discrete choice probit model yet unknown in the literature. The results support the assumptions that novelty and inventive activity enhance a patent’s value. They confirm the importance of backward citations, family size, and forward citations as va lue indicators. However, they expand on and partly break with the respective explanations why patent indicators correlate with profitability