The economics of innovation and intellectual property rights

Abstract

This thesis makes contributions to the economics of intellectual property rights (IPR)from different perspectives in three distinct but related empirical studies. First, patentand trademark statistics are used as innovation measures to examine the long-runrelationship between innovation and output in countries with long-established IPRsystems. The findings show that innovations may not always play a positive role indriving economic growth. Post-World War II evidence for some countries with extensivemeasured innovations (the US, and Germany) shows innovation's non-positiveeffects on economic growth, despite innovation's positive effects for the previous period.However, innovation retains a positive role in Japan, France and Australia.Despite the importance of innovation, risk often decreases the incentive to innovate,and can lead to R&D under-investment problems relative to the social optimum.Patents play an essential role in addressing this problem. This role is evaluated inthe Australian context by estimating the value of patent rights and calculating thecorresponding equivalent subsidy rate (ESR). The average value of patent rights forAustralian patents filed during 1980-1992 ranges from AU$9,000 to AU$17,000, whichis lower than the findings of the European and US studies. However, the ESR rangeof 3.2% to 8.4% is higher than that of large developed economies, indicating that thepatent system of Australia has outperformed the systems of other countries.One shortcoming of using patents as an innovation measure is the small number ofpatent users, which is less than secrecy users. Consequently, we examine determinantsof firms' choices of patenting versus secrecy using Australian data, with a focus onthe theory of Henry and Ponce (2011), predicting that firms' preference for secrecyover patents increases with knowledge tradability. In an important improvement overstandard empirical practice, a trivariate-probit model is constructed to correct for theendogeneity of the key dummy regressor in a bivariate-probit model. As a robustnesscheck, the potential sample selection bias caused by using only the innovator subsamplewas corrected. Key findings include that knowledge-trading firms and major R&Dinvestors are more likely to use secrecy than patents, providing evidence for theory andimportant insights to inform R&D and IPR policy.