By Elizabeth Lyons | March 12, 2018
The rate of new startup foundings has been in steady decline over the past four decades, potentially contributing to important declines in job creation and economic growth losses. In 1979 startups accounted for 14% of all firms but only 8% in 2015, and according to a recent Brookings report, the decline in startup rates during the 2000s was associated with a decline in high-tech entrepreneurship. Moreover, while non-startups in the United States continue to introduce new products, they are increasingly focusing on improvements on existing products rather than new products or varieties.
Much of the research that has been done to better understand how innovation rates can be increased has implicitly assumed that the population of innovators is fixed, and that efforts to increase innovation should be directed at these existing innovators. This assumption has important implications for how innovators should be paid, and how innovation policy should be designed. If innovative talent is entirely innate, organizations will have to provide large amounts of compensation to attract scarce innovators. Similarly, policy should be designed to enhance the performance of these existing innovators. If, however, innovators can be created, compensation to innovators will be disciplined by potential new entrants and the cost of creating them. Moreover, policy could aim both to improve the performance of existing innovators and to increase participation among non-innovators.
More recently, particularly in research on entrepreneurship, an interest in understanding how to encourage new entrants into these careers has developed. For example, Hans K. Hvide and Paul Oyer recently provided convincing evidence that dinner table human capital contributes to the higher rate of entrepreneurship among males relative to females. This is particularly useful evidence because it demonstrates that exposure, rather than an innate skill or personality trait, seems to contribute to the gender gap in entrepreneurship. This further suggests that by exposing females to individuals with entrepreneurship experience, the rates of entrepreneurship can increase. However, it is unclear how we can reverse this differential exposure.
In a recent paper, Josh Graff Zivin and I examine whether individuals who self-select into innovative and entrepreneurial pursuits have observably different characteristics than those who do not, and second, whether small incentives can “induce” people who do not identify as innovators to take on an innovative task and perform as well as their self-selected peers. We studied this by running randomized control trials within an innovation contest for undergraduate STEM majors. Our experiment design allows us to compare self-selected innovators to a population of potential innovators with identical training, and to test whether the non-self-selected, or induced, individuals can successfully innovate. It also allows us to assess the differential effects of confidence-boosting messages across these populations to determine whether this type of managerial intervention impacts innovative outcomes, and whether the induced innovators' performance benefits more from these efforts. By observing actual innovative output through the innovation contest, we are able to determine whether innate and induced innovators have different innovative skills.
We find that those participants that were induced to participate are different than those that volunteered based on observable characteristics. Induced participants were less likely to be drawn from majors that provide the most relevant skills for the competition and had lower cumulative GPAs. Yet, despite appearing to be less well equipped to compete, the success of induced participants was statistically indistinguishable from those that were “innately” drawn to the competition. Survey-based measures of effort invested in the contest are also statistically the same across the induced and innate samples. Thus, at least on average, it looks like innovators can be created. However, these average effects mask important heterogeneity in the impacts of each treatment arm. In particular, the impacts of inducement are significantly less promising for lower-ability students, as proxied by cumulative GPA.
Similarly, our confidence-boosting intervention has no average effect on participant performance on either the induced or innate sample of participants but it has large and opposite effects on high- and low-GPA students respectively. Not surprisingly, the emails significantly improved the performance of low-GPA students. The impacts on high-ability students are much more surprising. Not only do they not benefit from this encouragement, but they appear to be harmed by it.
Combined, our results demonstrate that innovators can be created through inducement subsidies, but that targeting inducement is likely to be more cost effective especially since targeting can be based on information like GPA that is relatively easy to collect. Moreover, we demonstrate that managerial interventions aimed at increasing innovator confidence may also need to be targeted to focus on those who benefit from it while avoiding those it may harm.
Elizabeth Lyons is an assistant professor of management at the School of Global Policy and Strategy. Her research focuses on the intersection between technology and innovation strategy, international management, and organizational economics. Her work on the organization of online work has been featured in the New York Times and other major media outlets.
Her current projects include using field observational data to analyze firm hiring and organization in international labor markets, using experimental lab evidence to analyze the relationship between organizational authority and production decisions, and using data collected over four years to study the effects of entrepreneurship training on career decisions.
During her graduate studies, she was the associate director of the Creative Destruction Lab, a venture lab at the University of Toronto.