When governments support emerging technologies, they face the same problem as investors: evidence is incomplete, the future is uncertain, and no one knows which interventions will succeed. Most research has asked why governments intervene in innovation. We asked a more practical question: how do states experiment, and what distinguishes one strategy from another?
Working with Daniel Armanios and Lauren Lanahan, I helped build a national dataset of state-led technology and economic-development policy.
We web-scraped approximately 6,800 policy reports, used natural-language processing to identify relevant material, and hand-coded 1,659 unique state policy actions from 2000 to 2015. Each action was classified by its technology focus, policy instrument, and institutional setting, achieving 0.92 inter-rater reliability.
We validated the dataset against legislative records, alternative policy sources, and interviews with state officials across four U.S. regions. The final dataset and scripts were published openly through the University of Oxford.
Wealth does not explain experimentation. Economically vibrant states did not consistently experiment more. States made distinct choices about where to search, how widely to spread their bets, and which institutions to involve. How states experiment matters more than how much!
We identified four strategies based on how broadly states experimented across technology topics and institutional partners:
Hub Specialists concentrated on a narrow set of technologies and institutions.
Public Entrepreneurs explored broadly through cross-sector partnerships.
Industry Architects supported multiple technologies through a more concentrated institutional model.
Ecosystem Designers spread experiments broadly across both technologies and partners.
This framework describes not only how active a state was, but what kind of experimenter it was. Broad experimentation created more opportunity, but also more exposure. After the 2008–09 recession, most states narrowed their portfolios. Public Entrepreneurs were the exception: cross-sector partnerships helped them continue experimenting across a wider range of technologies.
Spread of State Policy Portfolio by Dimension, 2000–2015
There is no single best innovation-policy model. States are choosing among focused search, broad exploration, and resilience.
For economic-development leaders, the useful question is not simply how much a state spends on innovation. It is: How is the state structuring its search, and can that strategy survive when conditions change?
The lesson also applies beyond government. Any organization facing technological uncertainty must decide how broadly to experiment, which partners to involve, and how to preserve learning when resources tighten.
With Daniel Armanios and Lauren Lanahan. Published in Academy of Management Discoveries, 6(2), 266–299. Supported by the Kauffman Foundation.