Date of Award

May 2019

Degree Type


Degree Name

Doctor of Philosophy (PhD)




Stuart Rosenthal


Innovation, Inventor Teams, Social Networks

Subject Categories

Social and Behavioral Sciences


My dissertation comprises two papers on urban economics, social networks, and innovation. The main question that connects these two papers is what the trade-off between inventor teams and social networks in patent development is. Two papers provide empirical evidence and answer the question from different perspectives.

Chapter 1 examines the joint effect of inventor social networks and team size on patent impact. Using data from the United States Patent and Trademark Office (USPTO) from 1975 to 2010, results confirm that the marginal effect of network size on patent impact decreases as team size expands, indicating that inventor networks and team size are substitutes. The substitution effect is stronger for teams and networks within a shared technology field. The marginal effect of inventor networks on the amount of knowledge and the speed with which new knowledge in a field is adopted also diminishes as team size increases.

Chapter 2 identifies the causal effect of non-compete clause enforceability on patent inventor team size, external network size and accessibility using a quasi-natural experiment of the Michigan Antitrust Reform Act (MARA) in 1985. Compared with states with similar enforceability from 1980 to 1990, patents in Michigan are 4% less likely developed by single inventors and 1.3% more likely produced by large teams with at least four inventors. Old firms, especially those relying heavily on hiring inventors from other firms for patent development, experience a fast expansion in inventor team size in the short run. Meanwhile, patent development in Michigan is 2.6% less likely to access external knowledge and information through inventors recently hired from other firms within the same state. However, there is no clear relationship between MARA and the external network sizes. Most of these findings are in support of the results of Chapter 1.


Open Access