In this dissertation, I will conduct a critical analysis of several methods typically used for modeling patent data, and then use insights gleaned from that analysis to explore four instances of public financing of innovation in the form of renewable-energy technology.
Chapter 1 introduces my overall research objectives, frames the research in terms of its policy relevance, and provides a brief preview of the major results. Chapter 2 provides background on innovation, causes and results thereof, with a particular focus on patent data, and thereby a framework for understanding the relevance of chapters 3 and 4.
The first main essay (chapter 3) addresses statistical regression techniques frequently used with patent-count time-series data, namely negative binomial regressions and log-log regressions. It reveals high rates of spurious correlation (false positives, also known as Type I errors) when using these techniques on patent data, investigates possible ways of addressing this problem, and creates a method for detecting it when using those and other regression techniques on similar data types.
Using lessons learned from the first essay, the second essay (chapter 4) examines public research and development (R&D) funding of four renewable-energy technologies – wind turbines, solar photovoltaics, solar thermal electric, and solar water heating. It employs several novel patent analysis techniques that allow patents to more closely represent the date of inventive activity. It finds that, contrary to popular narratives of public R&D funding driving increased invention (patenting), a diverse set of relationships exist in the renewable energy sector. These relationships range from changes in funding being correlated with future changes in invention rates, to changes in invention rates preceding changes in funding, to changes in funding and invention rates having no discernible relationship.
Taken together, these essays demonstrate the interdependent relationship between appropriate analytic techniques and accurate analysis when examining the sometimes subtle effects of complex policies.