Technical efficiency and ICT investment in Italian manufacturing firms

Abstract

The importance of Information and Communication Technologies (ICTs) is a much debated question with extensive literature aimed at understanding the role of ICTs in increasing economic growth, firm productivity and firm efficiency. Different methods to estimate firm efficiency are used in this study. In particular, both the translog and the Cobb–Douglas production functions are used in order to estimate the impact of ICT on Technical Efficiency (TE) in Italian manufacturing firms over the period 1995 to 2003. Results show that ICT investments positively and significantly affected firms’ TE. Moreover, group, size and geographical position have a positive influence on TE. Finally, the results show that older firms are, on average, more efficient than newer ones.

Keywords:

• ICT investment
• productivity
• stochastic frontier
• Italian manufacturing firms

JEL Classification::

• D21
• L63
• O33

Acknowledgements

I gratefully acknowledge Davide Infante e Carol Newman for valuable suggestions and comments.

Notes

¹ Solow is quoted as such in Gordon (2002).

² In the Pavitt taxonomy the sectors are classified in the following way: supplier dominated (Pavitt 1), scale intensive (Pavitt 2), specialized supplier (Pavitt 3) and science based (Pavitt 4).

³ The three periods are: 1995–1997, 1998–2000 and 2001–2003.

⁴ The FRONTIER 4.1 package uses the three-step estimation method procedure. These three steps provide a maximum likelihood estimate of the parameters of the stochastic frontier production function. The first step is an Ordinary Least Squares (OLS) estimate of the function. Here all the estimators β, with the exception of the intercept β ₀, will be unbiased. In the second step, a grid search on γ is conducted. The value for the parameters β (excepting β ₀) are set to the OLS value, β ₀ and σ ² parameters are adjusted and all other parameters (μ, η and δ) are set to zero. In the final step, the values in the grid search are used as starting values in an iterative procedure to obtain the maximum likelihood estimates.

⁵ The likelihood ratio test is equal to: (2*(Unrestricted−Restricted)) and follows a chi-squared distribution.