The Inventive, the Educated and the Creative: How Do They Affect Metropolitan Productivity?

Abstract

A longstanding research tradition assumes that endogenous technological development increases regional productivity. It has been assumed that measures of regional patenting activity or human capital are an adequate way to capture the endogenous creation of new ideas that result in productivity improvements. This process has been conceived as occurring in two stages. First, an invention or innovation is generated, and then it is developed and commercialized to create benefits for the individual or firm owning the idea. Typically these steps are combined into a single model of the “invention in/productivity out” variety. Using data on Gross Metropolitan Product per worker and on inventors, educational attainment, and creative workers (together with other important socioeconomic controls), we unpack the model back to the two-step process and use a SEM modeling framework to investigate the relationships among inventive activity and potential inventors, regional technology levels, and regional productivity outcomes. Our results show almost no significant direct relationship between invention and productivity, except through technology. Clearly, the simplification of the “invention in/productivity out” model does not hold, which supports other work that questions the use of patents and patenting related measures as meaningful innovation inputs to processes that generate regional productivity and productivity gains. We also find that the most effective measure of regional inventive capacity, in terms of its effect on technology, productivity, and productivity growth is the share of the workforce engaged in creative activities.

Keywords::

• Innovation
• productivity
• regional technology
• patents
• human capital
• creative class

Notes

¹ There has not been much research either on the factors and circumstances that lead some individuals to become inventors and some inventive individuals to become prolific inventors. For one of the few forays into this research area, see Shockley (1957).

² Consider, as an example, the patent granted in 1917 to Claurence Saunders (patent # 1,242,872) for a “self serving store” which became the model for all subsequent self-service stores and the modern supermarket.

³ Earlier versions of this work also used citations per worker as well as issued patents per workers as measures of inventive activity. Citations per worker generated estimation results similar to those obtained when patent applications per worker and which are presented here. The statistical strength of the results obtained when using issued patents per worker is significantly weaker. All results are available from the authors upon request.

⁴ A year-to-year correlation analysis for the same creative employment variable for the year 2006–2009 illustrates the slow change over time, with correlations of approximately 0.8–0.95 (Stolarick and Currid-Halkett, 2013).

⁵ In addition to running an analysis with the creative class based on the Richard Florida definition and with the McGranahan and Wojan definition, we also ran all regressions using two additional human capital variables: one for the age group 25–44, and one for age group 25–64, which we believe would be an even more interesting group to examine, since they account for the labor force but exclude retirees. Overall, these alternative specifications lead to only minor changes in the results (which are available from the authors upon request).

⁶www.uspto.gov/go/classification/selectnumwithtitle.htm

⁷ The correlation between the three is 0.576 (inventors and educated), 0.447 (inventors and creatives) and 0.484 (educated and creatives).