Does size still matter? How micro firms and SMEs vary in network learning

ABSTRACT

How can geographically constrained Micro firms and SMEs in emerging markets upgrade their products? How might firm size affect the benefits from accessing diverse knowledge and from participating in different learning relationships? Although access to diverse knowledge can be vital to innovation, smaller firms have a limited understanding about which new knowledge is most relevant to their context and how they may convert their capabilities into a more innovative organization. These latter aspects, we argue, come largely from relationships embedded in interfirm networks and certain types of non-market institutions that act as knowledge bridges and provide tutelage. Such institutions appear to benefit Micro firms in processing diverse knowledge. In contrast, product upgrading for SMEs improves more from their inter-firm networks, but these networks do not aid with managing diverse knowledge. We argue that differences in complementary capacities explain these variations. We use unique survey data from the Argentine wine industry.

KEYWORDS:

• Public private institutions
• economic development
• innovation
• small firms
• networks

Acknowledgments

We would like to thank Michael Gruber, Jonathan Kramer, Valentina Marano, Kurt Norder and Maura Pape for their outstanding research assistance; and IDR Mendoza and IAE Buenos Aires for collaboration in data collection. We are especially grateful to the Reginald Jones Center at Wharton, the University of Pennsylvania Research Fund, the William Davidson Center at the University of Michigan, and CIBER at the Moore School of Business and the Smith School of Business at UMD, for their generous financial support of this project. All errors and omissions are our own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The precise cut off points between Micro firms and SMEs vary often in emerging market countries. Micro firms range from 10 to 20 employees, where SMEs are below 200–250 employees. Following the standards from development agencies and as we discuss later, we use a revenue cut off, which is more reliable due to some seasonal and informal employment common in our industry and the region. Micro firms have annual revenues under $350,000. (Ferraro and Stumpo 2010; Henriquez 2009; ECLAC 2014).

² Following Breznitz (2005), Thelen (2003) and Mair et al. (2012), GSIs would be any institutions with the government contributing directly to the budget and having a representative in its governance. PPIs are a subset that in their bi-laws require formal board membership from and resource contributions from certain government and non- government actors.

³ The quotes in this article are from interviews that were conducted by the authors between 2004 and 2009. The larger project included interviews with over 50 actors in the industry, such as managers, engineers, policy makers, and civic leaders.  For more details, see McDermott et al. 2009, McDermott 2007.

⁴ Our Sales data is categorical, with boundaries at $170,000, $350,000, $500,000, $700,000 and $1,600,000. Sample size make impossible to calculate the models with cuts below $170,000 (30 observations) or above $500,000 (34 observations). To test the sensitivity to different thresholds we calculated the models for samples above $170,000 and below $500,000. In both cases, parameter estimates for the lower and higher thresholds were in between the ones estimated with the above and below $350,000 threshold; the p-values tend to be smaller, and closer to the ones for the full sample estimation. This is consistent with the mixing of two heterogeneous populations, which confirm the appropriateness of the $350,000 threshold utilized by the Argentinean government and international agencies.

⁵ All the wineries in our sample qualify as PYMES (Spanish acronym for Small and Medium firms). From secondary data we confirmed that these wineries are below the upper threshold for Medium firms ($16,000,000 annual sales).

⁶ Box-Cox transformation yields the transformed variable most proximate to a normal distribution. Log transformation is a particular case of BoxCox (lambda = 0) Greene, W. H. 2000. Econometric analysis (4th ed.). Upper Saddle River, N.J.: Prentice Hall.

⁷ Estimation available upon request from the authors.

Additional information

Funding

This work was supported by the Center for International Business Education and Research, Moore School of Business at University of South Carolina; Center for International Business Education and Research, Smith School of Business at University of Mariyland; University of Pennsylvania Research Fund [na]; William Davidson Center at the University of Michigan; and Reginald Jones Center at The Wharton School - University of Pennsylvania.