ICT demand behaviour: an international comparison

Abstract

This study aims to provide some empirical explanations for the gaps in information and communication technologies (ICT) diffusion between industrialized countries and especially between European countries and the USA. National macro-economic panel data are mobilized for 11 OECD countries over the 1981–2005 period. The analysis is based on factor demand estimates. It provides some original results: (i) the impact on ICT diffusion is positive for the level of education and negative for market rigidities, and both increased over time (in absolute terms) until mid-1990s; (ii) in each country, the price-elasticity of demand for ICT decreased (in absolute terms) over time, from 2 at the beginning of the 1980s to 1 in the middle of the 2000s.

Keywords:

• ICT
• investment
• factor demand
• productivity
• education
• regulation

JEL Classification :

• E22
• O47
• O57
• R24

Acknowledgements

The authors thank M. Juillard, G. Nicoletti, V. Spiezia and anonymous reviewers for their useful remarks. The views expressed in this article are those of the authors and do not necessary reflect those of the institutions for which they work.

Notes

Antonelli, Krafft, and Quatraro (2010) underline the impact of research in ICT sectors not only on ICT innovation but also on a country's ability to imitate ICT technologies. Education and market rigidities are important determinants of a country's research ability, hence our model could be presented as a reduced form including this mechanism.

This empirical specification, very intuitive, can also be deduced from a producer maximization program under weak assumptions, mainly the local approximation of the production function using a constant elasticity of substitution (CES) and constant returns to scale (see Appendix 2).

For more information on EPL and ETCR, see OECD (1999), and Conway, Janod, and Nicoletti (2005). In order to facilitate the interpretation of the estimates, these indicators are re-scaled between 0 and 1, with 1 for the strongest rigidity.

When introduced simultaneously the null hypothesis of multicolinearity could not be rejected.

As there are strong positive correlations between the market rigidity indicators, introducing both of them alternatively amounts to over-estimating each specific impact, as part of the explanatory power of the omitted indicator is carried forward to the present indicator.

For a discussion on these specific measurements errors see, Cette, Lopez, and Noual (2005).

A discussion on how to take account of cross-country methodological differences in price deflators for computers is available in Schreyer (2000).

The sets of instruments tested are combinations of the following variables lagged one or more periods: the explanatory variables or their first difference, long-term or short-term interest rates on treasury bonds, alternative capital user costs, the unemployment rate, the participation rate and the proportion of civil servants in global employment.

As the ICT diffusion threshold is chosen from a statistic selection method, the t-statistics of the parameters a5_ICT′ and a7_ICT′ is shown in Table 1 are irrelevant to conclude on the evolution of the impact of education and rigidities on ICT demand. Nevertheless, as the null hypothesis of equality could be rejected whatever the threshold tested, the statistical significativity of the evolution is not reassessed by the choice of a particular threshold.

In addition, this threshold is reached in 1995 for Denmark, in 1996 for the USA, in 1997 for the UK and Japan, in 1998 for the Netherlands, in 1999 for Finland and Germany and in 2000 for Austria, Spain, France, and Italy.

When ICT price elasticity changes are specified via an ICT diffusion threshold, the evolution is confirmed by the Bai and Perron (1998) break test.