Byrdes of on kynde and color flok and flye allwayes together

— William Turner (The Rescuing of Romish Fox, 1545)

1 Introduction

Much has been written about the so-called gender gap in terms of the availability of third-party financial resources for women-owned firms (Gicheva and Link, 2013, 2015; OECD, 2019). Regarding the implications of this gender gap, the evidence in the literature on firm performance, in terms of survival rates and growth rates, is mixed (Szerb et al., 2007; Link and Strong, 2016; Brush et al., 2017). An important aspect of the gender gap literature that has rarely been studied is the relationship between gender and the innovative performance of firms. This paper addresses that topic.

Furthermore, gender congruence between individuals at different levels in a power structure has been shown to be an important factor for female advancement in many settings, including labor markets and educational organizations. Conspicuously absent from this literature is an emphasis on successful innovative activity by females in a homophilic relationship.Footnote 1

In this paper, we explore the innovative performance of firms resulting from their Phase II Small Business Innovation Research (SBIR) research-funded projects in terms of what we call the gender dynamics of the firms. In particular, we take into consideration the gender of the owner of the firm as well as the gender of the principal investigator (PI) involved in the SBIR project in our study of whether or not the Phase II project was successful as measured by whether or not it resulted in a commercialized innovation.

In Section II, we briefly describe the U.S. SBIR program and we introduce the National Institutes of Health (NIH) data that we use in our empirical analyses. In Section III, we summarize the relevant literature on economic performance in general and the homophily among females. Based on that literature, we posit a hypothesis about differences in the innovative performance of female PIs involved in Phase II research depending on the gender of the firm owner. In Section IV we present descriptive evidence based on the NIH data, and we discuss our empirical findings. Section V concludes the paper with summary remarks and our view of an agenda for future research.

2 The U.S. SBIR program

Briefly, the SBIR program was established through the U.S. Small Business Innovation Development Act of the 1982, Public Law 97–219.Footnote 2 This act was passed by Congress as an amendment to the Small Business Act of 1953 which established the Small Business Administration. Congress motivated the passage of 1982 Act by stating as a preamble that small businesses have been the principal source of significant innovations in the nation and they are among the most cost-effective performers of research and development (R&D) that results in new products.

The purposes of the SBIR program, when established in 1982, were as statedFootnote 3:

  1. (1)

    to stimulate technological innovation;

  2. (2)

    to use small business to meet Federal research and development needs;

  3. (3)

    to foster and encourage participation by minority and disadvantaged persons in technological innovation; andFootnote 4

  4. (4)

    to increase private sector commercialization of innovations derived from Federal research and development.

SBIR funding to a small firm (less than 500 employees) is first given as a Phase I award. These are small awards that are intended to assist the firm assess the feasibility of an idea’s scientific and commercial potential in response to the funding agency’s objectives.Footnote 5 Firms that complete their Phase I research can apply for a follow-on Phase II award. These awards are given on the basis of a several criteria, one of which is the potential for the funded technology to results in a commercialized product or process.Footnote 6

As part of Congress’ due diligence, it authorized as part of the SBIR program’s reauthorization in 2000 the National Research Council (NRC) within the National Academies to study the SBIR program and to offer to Congress suggestions for program improvements. These suggestions for the program would be taken into account when Congress considered the 2008 reauthorization of the program.

As part of the NRC’s efforts, in 2005 it conducted a survey of a representative sample of Phase II projects funded by the five largest agencies with SBIR programs: the Department of Defense (DoD), NIH, the National Aeronautics and Space Administration (NASA), the Department of Energy (DOE), and the National Science Foundation (NSF). After the reauthorization of the SBIR program,Footnote 7 Congress authorized additional surveys of representative samples of both SBIR-funded and STTR-funded Phase II projects.Footnote 8 In 2011, DoD, NASA, and NSF Phase II projects were surveyed; NIH and DOE Phase II projects were surveyed in 2014.

In this paper, we focus on commercialization from the Phase II research as the relevant performance variable as was stated in the 1982 Act, and we examine data on NIH funded Phase II projects.Footnote 9 The DoD SBIR program is the largest program measured in terms of number of Phase II awards and allocated research dollars. Second to the DoD program in these dimensions is the NIH program. The other three agency programs are much smaller.Footnote 10 We chose to examine the data from the NIH from both the 2005 and the 2014 survey not only because of the program’s size, but also because the survey measures of commercialization are more market-based than in the DoD data.Footnote 11 Many of the technologies developed through DoD funded Phase II awards are in fact purchased directly by the DoD.

Figure 1 shows the NIH sample of projects by the year the project was funded (n = 966). The only overlap between the 2005 and 2014 surveys is in 2001; all projects awarded prior to that were surveyed in 2005, while all subsequent awards enter the 2014 sample. The number of projects awarded in a given year ranges from 19 in 1992 (which is also the first year in the data) to 101 in 2001.

Fig. 1
figure 1

Distribution of Phase II Projects by Year of Funding (n = 966). Note: The 2005 survey includes projects funded in 1992—2001; the 2014 survey includes projects funded in 2001—2010

Full size image

3 Gender and firm performance

There is a small literature related to the non-homophilic success of females in small innovation-intensive firms. For example, Weber and Zulehner (2010) study the gender composition of the initial hires of start-up firms in Austria and show that firms that start off with more female employees have higher survival rates. Joshi, Inouye, and Robinson (2018) find that SBIR-funded projects with female PIs are more likely to succeed if the funding agency is comprised of more females; this finding complements the earlier finding of Andersen, Bray, and Link (2017). Ding, Murray, and Stuart (2006) identify a gender gap in commercialization and patenting to which they link women scientists’ low exposure to the commercial sector and the perception that commercialization activities detract from academic pursuits. Link and Ruhm (2009), who also use SBIR data, find that women-owned businesses are less successful in commercializing their technology from Phase II projects. Shane et al. (2012) find evidence of bias against women inventors among university technology transfer officers, which play a significant role in directing resources for commercialization, and a predisposition to inventions by Asian inventors with previous industry experience. Finally, Dohse, Goel, and Nelson (2019) show that female owners are more likely than female managers to introduce new innovations to the market.

There is evidence in the literature that female-led firms (e.g., female-owned firms) tend to hire and retain more females at lower levels. Matsa and Miller (2011) show that in a sample of publicly traded U.S. companies, higher female representation on the board of directors leads to more female top executives at the company. Kurtulus and Tomaskovic-Devey (2012) find a strong positive correlation between the share of female top- and mid-level managers in a large sample of U.S. firms. Hensvik (2014) also finds that among firms in Sweden, those with more female managers also tend to hire more female workers at lower levels.

Several recent studies find evidence that having women in top management is associated with better labor market outcomes for lower-ranked females at a firm. Examples include Tate and Yang (2015), who use plant- and firm-level data on non-farm establishments with paid employees from the U.S. Census Bureau’s Longitudinal Business Database, and Flabbi, Macis, and Schivardi (2019), who use data on Italian manufacturing firms. Cardoso and Winter-Ebmer (2010) obtain similar results for Portugal, showing that female employees benefit more from female firm leadership in small (100 or fewer employees) firms. See also, Fox et al. (2017) and Shane et al. (2015) for additional examples.

Educational settings provide further evidence that females perform better under female supervisors. Carrell, Page, and West (2010) show that when high-achieving female students take an introductory math or science course with a female professor, they are more likely to take more math and science and to major in a STEM field. Dee (2007) finds positive impacts for eighth-grade students of having a teacher of the same gender. Bednar and Gicheva (2014, 2018) show that in collegiate athletics, female head coaches perform better and have lower turnover rates when working under an athletic director who is generally supportive of females.

Several explanations have been proposed in the literature for why female workers may be more productive when female representation in upper levels of the firm is higher. Athey, Avery, and Zemsky (2000) offer a theoretical model underscoring the importance for firm performance of same-sex mentorship between workers in upper and lower levels of a company. Flabbi et al. (2019) present a model, consistent with Cornell and Welch’s (1996) seminal work, based on the assumption that CEOs derive more accurate signals about the productivity of workers of their own gender, so female CEOs are better able to allocate female workers to the most appropriate tasks.

Based on this literature, we hypothesize that the relationships uncovered by prior studies will carry over to the small innovative firms that comprise our sample, making Phase II research projects with a female PI more successful when the owner of the firm is also female. If females are more productive in homophilic relationships, we also hypothesize that among female-owned firms, those with a female PI on their Phase II research project will tend to be more successful. The existing theories do not address the relative success of all-male compared to all-female teams, but we further explore whether the projects conducted by female-owned firms with a female PI tend to outperform projects in male-owned firms with a male PI.

We make several contributions to the literature in this paper. First, we test existing theories positing that females tend to perform better when working under a female supervisor in the context of small innovative firms, where this relationship has not been examined previously. Second, we examine firm performance in homophilic teams using a well-defined and meaningful measure that is aligned closely with the stated goal of the SBIR program.Footnote 12 Previous studies of the effectiveness of female firm leadership have used other categories of performance measures such as stock prices and firm assets (e.g., Ahern and Dittmar 2012, Wolfers 2006). Flabbi et al. (2019) use sales per worker, value added per worker, and Total Factor Productivity (TFP), while Matsa and Miller (2013) use firm profits. A third contribution of our paper is that it is the first to analyze the 2014 SBIR survey data.

4 Empirical analysis

We estimate the following probit model of the likelihood that firm i commercializes the technology developed with the Phase II award:

$$ \Pr \left(C{ommercialization}_i=1|{FemaleOwner}_i\times {FemalePI}_i,{\boldsymbol{X}}_{\boldsymbol{i}}\right)=\Phi \left(\alpha \left({FemaleOwner}_i\times {FemalePI}_i\right)+{\boldsymbol{X}}_{\boldsymbol{i}}\beta \right) $$
(1)

where Commercialization and FemaleOwner x FemalePI are defined in Table 1 along with the controls in vector X, and Φ(∙) is the standard normal distribution.

Table 1 Definition of Variables
Full size table

Regarding the variable Commercialization, the relevant NRC survey question associated with this variable is: “Has your company and/or licensee had any actual sales of products, processes, services or other sales incorporating the technology developed during this project?” If the respondent to the survey answered YES to having actual sales of either a product, a process, or a service, we defined the variable Commercialization to equal 1, and 0 otherwise.

Our hypothesis is that the estimated coefficient on FemaleOwner x FemalePI is positive, indicating that Phase II projects led by a female PI are more likely to commercialize in female- than in male-owned firms and that female-owned firms with a female PI working on the Phase II project are more likely to commercialize than female-owned firms with a male PI.

The variables used to test empirically our hypothesis are defined in Table 1 and descriptive statistics are presented in Table 2. Within our sample, 15% of all Phase II projects were awarded to female-owned firms, and 18% of all Phase II projects in our sample were led by female PIs. Only 8.4% of all Phase II projects were awarded to female-owned firms that also had a female PI, but these statistics suggest that, consistent with findings in prior studies from different settings, females tend to be disproportionately more likely to work with other females across different levels within a company.

Table 2 Descriptive Statistics on the Variables
Full size table

Regarding our control variables, our argument for holding constant firm size is that larger firms, that is the larger firms among our sample of small firms, have more internal resources to devote to the commercialization of their Phase II project than do smaller firms. Previous Phase II awards that relate to the technology being researched in the current Phase II project are expected to have a positive impact on the efficiency of the current Phase II project, and we assert that the more efficiently the current Phase II project is conducted the more likely the developed technology will have commercial potential. In some cases, generally in the smaller firms, the PI is also either the founder of the firm or the CEO of the firm. Thus, in these firms, there is a dual homophiletic relationship, and we control for such situations. This information is available only in the 2014 survey, and so we set the variable to equal 0 for all observations from the 2005 data. We control for the cases in which the Phase II project is not funded through a SBIR award but rather through a STTR award. Finally, the specifications include indicators for year when the Phase II project was funded and for survey year (2005 or 2014).

In Table 3, we present the mean value of key variables by what we call project composition. That is, we divide the sample of Phase II projects into the four categories noted in the table. Based on mean values, the likelihood that a Phase II project’s technology is commercialized, that is the likelihood that the Phase II project is successful in terms of the main criterion of the SBIR program, is nearly 50% greater in firms with a female owner and a female PI compared to the other three categories of firms, where the probability of commercialization is approximately the same. Also noteworthy in Table 3 is that, on average, firms with both a female owner and a female PI are smaller than any of the other types but they have over 3 times more previous related Phase II awards compared to firms with other gender composition of the owner and PI.Footnote 13

Table 3 Mean Values by Gender Composition of the Project Team (standard deviation in parentheses, n = 966)
Full size table

Our econometric findings are presented in Table 4, which shows coefficient estimates from the probit model in eq. (1) and heteroskedasticity-robust standard errors. The specification in column (1) includes an indicator for whether the firm received any Phase II awards in the past that are related to the current technology, while the model in column (2) replaces this indicator with a quadratic in the number of related prior Phase II awards. The findings mirror the unconditional descriptive statistics shown in Table 3. We find that within our sample, the likelihood of commercialization is highest among female-owned firms with a female PI relative to any other category. These results support our hypothesis and are the punchline of our paper. The estimates presented in Table 4 also confirm the observation from Table 3 that male-owned firms with a male or female PI and female-owned firms with a male PI all have similar success rates.

Table 4 Estimated Results from the Probit Model in Eq. (1) (n = 966)
Full size table

Figure 2 illustrates the marginal effects of the gender variables based on the probit estimates in column (2) of Table 4. Specifically, the plot shows the predicted probability of commercialization for each of the four categories of firms, along with the 95% confidence interval. These results again illustrate that female firm owners (PIs) are more successful when working with a female PI (owner). We also document that in this sample of Phase II projects funded by the NIH, there exists a statistically and economically significant difference in performance between male-owned firms with a male PI and female-owned firms with a female PI.

Fig. 2
figure 2

Predicted Probability of Commercialization. Note: Average marginal effects based on column (2) in Table 4. The error bars show 95% confidence intervals

Full size image

Finally, regarding the coefficient estimates on control variables in Table 4, the estimated effect of firm size is small and only statistically significant at the 10% level in column (2). The average marginal effect of one standard deviation increase in firm size (45 employees) is a 2.7 percentage point increase in the probability of commercialization. And, previous related Phase II awards, as well as the number of such awards, are a driver of the likelihood that the Phase II project will be commercialized.

5 Concluding statement

We have illustrated empirically in this paper the synergetic effect of what we call gender dynamics, that is of female-owned firms that have a female PI leading their Phase II research project. While our contribution to the economics literature on homophily is unique to only one dimension of innovative behavior, it does, in our view, provide an important starting point for future research on more broadly defined innovation effects of such relationships. Future research might focus on the acquisition of technology-oriented resources, such as venture capital; technology transfer activities, such as patenting behavior, or collaborative research ventures, such as participation in R&D alliances.

To expand the literature more broadly, there might also be other dynamics associated with age, race, ethnicity, orientation, or nationality that have yet to be investigated in general, much less within the innovation literature. Of course, data availability is always a delimiting factor, or even a factor that keeps the research door closed. Creatively designed and executed firm histories over long periods of time might be a relevant vehicle through which to begin.