Modelling R&D expenditure data with zero observations: two-equation model

Abstract

This article attempts to analyse the determinants of firms’ Research and Development (R&D) expenditures in Korea by considering the business environment after the economic crisis in 1997. In addition, to take into account zero R&D expenditure, this article employed a two-equation model unlike models used in other studies. This method incorporates a two-level decision structure: the participation decision and the decision on the amount to spend once the issue of participation has been decided. According to the estimation results, while the proposition that larger firms are more active in R&D is true, the proposition that firms that possess market power are more active in R&D is not true for Korea. Technical cooperation among Korean firms seems to be less active than in other countries. In addition, the results indicate that foreign investment stimulates the firms’ R&D expenditure. Furthermore, a number of factors were found to play a role in promoting firms’ R&D activities: the external conditions of the firms’ R&D activities, including the location, other firms’ R&D activities in the same industry, support from the government and technical support from research institutes.

Acknowledgements

We would like to thank the Science and Technology Policy Institute for providing us with the dataset of KIS 2002.

Notes

¹ To estimate the two-equation model, the starting values are obtained from estimations of univariate models (1) and (2). ρ starts at zero. During the iterations, the estimated coefficients of ρ and σ are achieved by estimating γ = ρσ and . The specifications impose the desirable property that the SD σ should be strictly positive and the correlation coefficient ρ should range from –1 to 1. After convergence on these transformed parameters, the final output and derivatives are written in terms of ρ and σ.