Innovation in the U.S. 1920-2006 - Quality Trends and Evolutionary Path

Long run economic growth potential depends on the increase in the efficiency of resource utilization in the economy through improvements in the underlying technological capabilities. Recent economic growth slow-down, and in particularly the substantial decline in long-run productivity growth rates in the US, raises a question regarding the possible slowdown in the underlying technological growth. I set out to evaluate this question by examining the change in the potential social benefits from technology over the period of 1944-2000 in the US. I use knowledge spillovers generated by cohorts of patented innovations as proxy to the social benefits generated by those cohorts. These can be estimated using the distribution parameters for the number of citations received by patents in a cohort. The data for the work is a newly constructed patent citations database of all US patent data from 1920-2006 with full citations from 1947 on. The first chapter introduces the new data and provides an in-depth analysis of the time-trend of its various statistical characteristics. The exposed non-stationary distribution parameter for the citation data impedes its use in the time-series analysis for extracting the innovative quality trend. A de-trending treatment to correct for this non-stationary behavior is proposed and applied. The second chapter pursues the innovative quality analysis over the period. My attempts to use the Jaffe-Trajtenberg model for knowledge spillovers with the expanded data period strained some of the model's underlying assumptions to a breaking point. Instead, I introduce a new model for estimating the intensity of such spillovers (the innovative quality) based on the Log-Normal distribution of patent values as measured from their spillover effects (received citations). I compute the innovative output quality for annual cohorts of patents in narrowly defined technological fields over the period of 1937-1994. The results show a decline in the traditional mechanical and chemical technologies quality starting in the early 1960s. The modern technologies associated with electronics, ICT and medicine flourished until the early to mid 1980s,after which their quality declined as well. The last chapter examines the evolutionary path of a transformative technology using the ICT over the period of 1944 to 1994. The analysis uses the full citation network for US patents over the period of 1947-2006 and applies network analysis techniques to identify main technological trajectories for the key technological fields comprising the ICT. The pattern of technological inflows and outflows for each cohort on the trajectory provides the evolutionary timeline and technology inter-dependencies. These demonstrate the underlying process of building the essential skills and complementary devices and techniques which took place in the first 25 years of the ICT evolutionary path. The fast pace of that evolutionary path and lack of significant gaps in the time-line suggest that even under optimal conditions of existing market demand for new technologies (e.g. defense and space), it takes a new, transformative technology close to half a century to deliver its productivity gains