Skip to main content

Advertisement

SpringerLink
Log in

Entrepreneurial finance and regional ecosystem emergence

  • Research article
  • Published:
Small Business Economics Aims and scope Submit manuscript

Abstract

We present a novel framework for studying the evolving role of entrepreneurial finance over the stages of emergence for a regional entrepreneurial ecosystem. Drawing on entrepreneurial ecosystems, entrepreneurial finance, and territorial servitization, we explore how three different finance sources impact firm survival and how they relate to each other during ecosystem emergence. We analyze entrepreneurial firms in one industry and region over 36 years. We find that firm survival is differentially affected by funder type based on the stage of ecosystem emergence. Finance sources also have different interrelations depending on the stage of emergence. Based on our results, we abductively articulate a framework for stage-dependent ecosystem emergence microfoundations. This rectifies contradictory results that examine single sources of finance and use cross-sectional data. Had we not measured the emergence process, the results would have led to markedly different theoretical implications and practical takeaways for entrepreneurial finance and ecosystem emergence.

Plain English Summary

Measuring the stage of ecosystem emergence is key to success in financing start-up firms and developing entrepreneurial ecosystems. Using detailed firm-level data, we present a novel framework for studying entrepreneurial finance within a local entrepreneurial ecosystem over time. Entrepreneurial finance sources, whether public or private, interact with firms differently depending on the stage of entrepreneurial ecosystem emergence, as measured by firm density. We find that funding sources are associated with different rates of firm survival during different stages of ecosystem emergence, with a federal source more important during the nascency stage and venture capital more effective as firm density increases. Our results demonstrate that policymakers and ecosystem champions could make better decisions if finance is considered part of an emergence process.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1

Similar content being viewed by others

Notes

  1. Though we use the term “stage,” our conceptualization of stages is distinct from life cycle studies (e.g., Auerswald & Dani, 2017; Cantner et al., 2021). Our study moves away from this deterministic model and focuses on the role of microfoundations and adaptive approaches (Kim et al., 2022; Klepper, 2007).

  2. Data was obtained from the PLACE: RTP database in February of 2019. This region includes 13 counties which encompass the Raleigh and Durham-Chapel Hill MSAs surrounding Research Triangle Park as defined by the NC Research Triangle Regional Partnership in 2014. The 13 counties are Chatham, Durham, Franklin, Granville, Harnett, Johnston, Lee, Moore, Orange, Person, Vance, Wake, and Warren.

  3. When aggregated, NC IDEA funding accounts for less than 4% of the state funding. There are 27 grants totaling approximately $50,000 to $60,000 each. While this small amount is unlikely to contribute significantly to the estimates, we decided to keep this funding in the model to provide a complete picture of state funding.

  4. For some observations, the dollar amount of the investment was missing. If the amount could not be determined from online press releases or other sources, we used $100,000 as a modest imputation.

  5. The second partial derivative with respect to time in (1) measures the rate of change in the firm density dynamic.

  6. Estimation uses the threshold command in StataSE 15.1.

  7. An important artifact from model 2, which imposes four stages, is that we still detect the same time periods for our three main stages: nascent stage (1980–1997), accelerated growth (1998–2009), and stabilization (2010–2016). By imposing a fourth stage, we detect the date when the dynamic of the acceleration stage becomes less convex. Overall, these results are robust. They confirm three industry stages from 1980 to 2016 and demonstrate the utility of our method for non-deterministically identifying industry stage changes.

  8. Details are available from the authors upon request.

  9. To determine the appropriate duration dependence, we take the hazard rate value obtained from the life table method and test which of three duration dependence specifications is able to explain the hazard rate. Efficiency criteria indicate that the polynomial form is most appropriate.

  10. There are three main ways to model these unobservable variables: a normal distribution, gamma distribution, or non-parametric approach, which fits an arbitrary distribution using a set of parameters. These parameters comprise a set of mass points and the probabilities of a firm being located at each mass point. Additional details are available upon request.

  11. It was impossible to determine values for some control variables for 31 firms. Eight firms are missing all founder data. The other 23 firms have founders with some missing data. For these firms, we used either the minimum or average value to replace missing values, whichever would cause any bias to be in the direction of a null funding. We inflated the same prior industry experience variable by founding team size for firms for which only some founders were missing this variable.

  12. The dynamic random effects probit is appropriate because our setting is dynamic where lagged values of funding from each source are needed to estimate the probability of funding from that source in the current period. We are interested in the probability of securing funding, given the prior distribution of sources. As such, we must account for state-dependent processes. Furthermore, we do not use the funding amount because we are more interested in the links between funding sources and want to avoid a value effect from funding caused by noise related to large or erratic values. Thus, we have a binary outcome variable. Finally, the relationships between state, federal, and private VC funding are nonrecursive so we cannot estimate for each funding type simultaneously.

  13. The \({c}_{i}\) term accounts for heterogeneity between firms and is written:

    $${c}_{i}={\alpha }_{0}+{{\alpha }_{1}y}_{0}+{\alpha }_{2}\overline{X }+{\alpha }_{3}{X}_{0}+{\alpha }_{4}$$

    While \({\alpha }_{0}\) and \({\alpha }_{4}\) (the firm-specific time-constant error) are purely random, \({y}_{0}\) represents the initial value of the dependent variable. \({X}_{0}\) is a row matrix of the initial value of the time-varying covariates, and \(\overline{X }\) is a row matrix of the mean value of the time-varying covariates for all firms (the within-firm average based on all time periods). \({c}_{i}\) then accounts that the probability of receiving one type of funding is likely influenced by other funding sources in prior periods (Grotti & Cutuli, 2018; Rabe-Hesketh & Skrondal, 2013; Wooldridge, 2005). The \(X\) are a subset of \(Z\).

  14. We use “Stage” when referring to the technical “regime”-switching analysis. The dynamic random effects probit model assesses state dependence in the probability of receiving funding but does not account for differences across regional industry stages.

  15. Effect sizes for coefficients in survival analysis are calculated by the formula: 100*(exp(\((\widehat{\beta })\) − 1).

  16. Meaning the likelihood of funding from one source given prior funding from the same or different sources.

  17. Full results are available in Appendix Tables 11 and 12.

  18. This could also be an artifact of our use of the 1 year. Future studies may extend examination of the temporal lag.

References

Download references

Acknowledgements

Funding for the development of the PLACE database was provided by the National Science Foundation and the Kauffman Foundation. This research was also supported by the University of North Carolina Dissertation Fellowship. We thank Christos Kolympirisa for NIH data. We would also like to thank Mercedes Delgado, Ludovic Dibiaggio, Andrew Nelson, Jose Lobo, and participants at AOM, APPAM, the Danish Research Unit for Industrial Dynamics (DRUID), the Atlanta Conference on Science and Innovation Policy, the UNC/TIM Emergence Workshop, and seminar participants at Kenan-Flagler Business School’s Entrepreneurship Working Group and MIT for comments and suggestions.

Author information

Authors and Affiliations

  1. School of City & Regional Planning, Georgia Institute of Technology, 245 Fourth St. NW, Ste 204I, Atlanta, GA, 30313, USA

    Paige Clayton

  2. Department of Public Policy, Arizona State University, 411 N Central Ave, Ste 450, Phoenix, AZ, 85004-0687, USA

    Maryann Feldman

  3. SKEMA Business School, Université Côte d’Azur (GREDEG), OFCE SciencesPo, 60 Rue Fedor Dostoïevski, 06902, Valbonne, France

    Benjamin Montmartin

Authors
  1. Paige Clayton
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Maryann Feldman
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Benjamin Montmartin
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Maryann Feldman.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix. Stage-switching analysis modeling and net effect calculation

Appendix. Stage-switching analysis modeling and net effect calculation

Here, we provide an example of the stage-switching regression. The dynamic random effects probit model with stage-switching private VC in time (t) reads as follows:

$${\mathrm{PRIV}}_{it}={\alpha }_{1}+{\beta }_{p1}{\mathrm{PRIV}}_{it-1}+ \Delta {\beta }_{p2}{\mathrm{PRIV}}_{it-1}+{\Delta \beta }_{p3}{\mathrm{PRIV}}_{it-1}+{\beta }_{s1}{\mathrm{State}}_{it-1} + \Delta {\beta }_{s2}{\mathrm{State}}_{it-1}+{\Delta \beta }_{s3}{\mathrm{State}}_{it-1}+ {\beta }_{f1}{\mathrm{Federal}}_{it-1}+\Delta {\beta }_{f2}{\mathrm{Federal}}_{it-1}+{\Delta \beta }_{f3}{\mathrm{Federal}}_{it-1}+ \Delta {\alpha }_{2}+{\Delta \alpha }_{3}+{Z}_{it}+ {c}_{i}$$

\({\beta }_{pi}\) represent the focal funding source and coefficients of key interest (in this case, private VC), with \({\beta }_{si}\) and \({\beta }_{fi}\) representing lagged values of state and federal funding. \({\beta }_{i1}\) represent coefficients for stage 1 (base effect in nascent stage), while \(\Delta {\beta }_{i2}\) represent the change of the effect from stage 1 to 2 (nascence to acceleration).\({\Delta \beta }_{i3}\) represents the changing effect in stage 3 with respect to stage 1 (nascence to stabilized growth). Here, the \({c}_{i}\) unobserved effect covariates include only state and federal funding.

We calculate the net effect by stage by adding the coefficient of the base effect for the nascency stage and the coefficient of the associated change for the subsequent two stages of acceleration and stabilization. We calculate standard errors using the covariance matrix of the coefficients and calculate a t-statistic to discern statistical significance at the usual levels. All models are heteroskedastic robust. Full stage-switching results are shown in the Tables 9, 10, 11, and 12.

Table 9 Discrete event history analysis with and without stage-switching
Full size table
Table 10 Non-parametric discrete event history model for firm failure by funding source(s)
Full size table
Table 11 Dynamic random effects probit estimation with stage-switching
Full size table
Table 12 Dynamic random effects probit with stage-switching and unobserved heterogeneity
Full size table

Results presented in Table 10 examine variations on the functional forms of the state, federal, and private VC variables. In this specification, we include all possible varieties of funding dummies that a firm might receive as a series of indicator variables, with the reference category being firms that received no support. These results confirm the previous results, demonstrating robustness. Receiving only private VC or any combination of sources is associated with a decreased probability of failure. Receiving only state funding or only federal funding has no influence, while receiving state and federal funding, even without private VC, does decrease the probability of failure. However, firms that received both private VC and federal funding, and firms that received all three sources of funding, have the lowest magnitude chances of failure.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Clayton, P., Feldman, M. & Montmartin, B. Entrepreneurial finance and regional ecosystem emergence. Small Bus Econ 62, 1493–1521 (2024). https://doi.org/10.1007/s11187-023-00827-6

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11187-023-00827-6

Keywords

JEL Classification

Search

Navigation