Using the Variation in Transaction Costs to Explain the Propensity to Patent: Some Firm-Level Evidence

Abstract

Do lower transaction costs induce innovation? We exploit the variation in transaction costs across firms and countries to explain the firm-level propensity to innovate. Drawing on a large dataset from European Patent Office (EPO) covering over 15,000 firms from 100 countries over the period 1991-2015, we compute the conditional probabilities of individual firms to acquire a valid based on the validation outcomes and examine whether lower transaction costs encourage firm-level innovation. Our empirical strategy addresses firm-level idiosyncrasies, country-level unobserved effects and common technology shocks that potentially invoke omitted variable bias in the effects of transaction costs on the propensity to patent. Numerous indicators of transaction costs are clustered into four specific categories: (i) Shleifer-La Porta (administrative), (ii) Northian (cross-border), (iii) Coasean (procedural & contractual) and (iv) Williamsonian (intra-firm). Our investigation reveals that higher transaction costs systematically account for the failure of firms to acquire a valid patent and to innovate. The results plausibly indicate that higher costs across firms and countries are associated with a marked drop in firm-level propensity to patent. Controlling for unobserved heterogeneity across firms, we employ the negative binomial model to examine the contribution of transaction costs to firm-level patenting propensities. The evidence suggests that not all transaction costs are created equal. Higher administrative transaction costs are associated with persistently higher number of valid patents as such costs tend to address moral hazard, opportunistic behavior and other sources of asymmetric information and potentially deter copy-cat imitating firms from entry. On the other hand, decreasing cross-border, procedural and intra-firm transaction costs leads to consistently higher number of valid patents being granted. The estimated propensities to patent are further tested using extreme bounds analysis by running more than 13 million combinations of the core regression model which confirms the importance of transaction costs in explaining differential propensities to patent across firms and countries.