Modelling entrepreneurship in small-scale enterprises

Abstract

In the small-scale enterprises, the entrepreneurs and managers are often one and the same. This paper attempts to estimate the entrepreneurial competence in small-scale enterprises decomposing into entrepreneurial ability and managerial efficiency using the frontier production function. The empirical results show that most of the sample enterprises fail to realize their full entrepreneurial potential. The important implication of these results is that small enterprises should generate a different set of rules than large enterprises in the decision-making process that best conduces them to move toward their potential frontier.

Acknowledgements

The earlier version of this paper was presented at the International conference on ‘Small and Medium Sized Enterprises in a Global Economy: Economic Resilience in East Asia – Role of SMEs and Stakeholders’, 5–7 July 2004, Kuala Lumpur, Malaysia. The author thanks seminar participants for their views and comments without implicating for any errors remain.

Notes

Cantillon (1730) was the first to place the entrepreneurial function in the field of economics. Say (1816) and Knight (1921) subsequently worked on the role of entrepreneurs. However, Schumpeter's (1934) contribution to the theory of entrepreneurship gave the momentum of the subject. Kirzner (1973) contributed significantly to the subject among the contemporary economists.

In the presence of agency problems, firm managers may choose an input structure that fails to maximize the firm's physical output.

William Baumol (1968) notes that the subject of entrepreneurship is conceptually elusive, and the term does not always have clear theoretical content.

For details, please see Kalirajan (1988).

Translog production function is considered here for its flexible nature and no a priori restrictions on the specified technology.

Calculating and then rearranging the terms, the marginal productivity conditions can be written as C_i  = β _i + Σ _j γ _ij + w_i quad i =  1, 2, 3, where w is the sum of allocative error and statistical random error and follows a multivariate normal distribution with mean ξ and variance covariance Σ and C_i  = p_i x_i /p_y y. Hence, the likelihood function of (ln y ₁, …, ln y_n , C ₁, C ₂, …,C_m ) is as follows:

This likelihood function is maximized with respect to the parameters (β, α, γ, δ, σ², ξ, and Σ) and thereby the optimal frontier output (

) can be calculated.