Organizational choices and performance in distribution systems

Abstract

This article studies the performance of distribution networks as the result of a range of organizational choices. The analytical part of the article surveys the vast literature devoted to franchising and dual distribution. From this framework, several testable propositions linking network performance to organizational choices are derived. Three complementary criteria of performance are taken into account: the internationalization rate, the expansion rate and the market share. This article provides evidence for the simultaneity between these performance criteria, analytically related as indicators of the network commercial performance. Thus, the econometrical model is defined as a system of simultaneous equations, free of endogeneity regarding the explanatory variables. The estimations on recent French data obtained using the three-stage least squares method provide robust results and show that the type of distribution network, the number of company-owned outlets in the network, the type of sector and the choice to manage several networks simultaneously affect the performance.

Keywords:

• franchising
• endogeneity
• simultaneous equations
• three-stage least squares

JEL Classification:

• C31
• L14

Acknowledgements

The authors thank the French National Institute of Statistics and Economic Studies for permission to use its original data sets. We are grateful to the National Secretary of Higher Education, Science, Technology and Innovation of Ecuador for its financial support. We also thank the anonymous referees for their constructive advice. None of these, however, are responsible for any remaining errors.

Notes

¹ ‘Plural forms’ and ‘dual distribution’ are used as synonyms.

² as H₁.

³ The statistical theory suggests that using the method of averages to complete the missing data introduces bias in the value of the estimator and its variance. Rubin (1996) proposes multiple imputation as a solution. This method uses Monte Carlo simulations to replace the missing data from a number (m > 1) of simulations. In each simulation, the complete data matrix is analysed using conventional statistical methods. Finally, the method combines the results to generate robust estimators, their SE and their confidence intervals. Thus, the multiple imputation method replaces missing values at random and does not generate bias in the allocation of imputed values.

⁴ Hausman (1976), Nakamura and Nakamura (1981).

⁵ All the detailed estimation results regarding this article are available upon request from the authors.

⁶ Because no good instrumental variables are available and considering the complexity of the econometrical model defined here as a system of equations, we opt to remove the endogenous variables instead of correcting the endogeneity to minimize complexity.