Information Technology as a Factor of Economic Development: Evidence from Developed and Developing Countries

Abstract

This article explores the role of information and communication technologies (ICT) and of its individual components as factors of economic development. An augmented production function is employed to estimate the total ICT effect on labor productivity growth as well as the impact exerted by its components (hardware, software, and communications). The empirical analysis is based on a sample of 42 developed and developing countries, covering the 1993–2001 period. A positive and significant ICT growth effect is estimated in both country samples, with the highest impact observed in developed ones. This effect stems entirely from the hardware and communication components. Estimates concerning the individual components of ICT slightly differentiate between developed and developing countries, with respect to their statistical significance. The results are robust to possible endogeneity biases.

† An earlier version of this paper has been previously presented in the 9th International Conference on Marketing and Development, Thessaloniki, Greece, June 8–11, 2005.

Keywords:

• Productivity
• Growth
• Information technology
• Economic growth

Acknowledgements

I, gratefully, acknowledge the financial support of the Greek State's Scholarship Foundation.

Notes

† An earlier version of this paper has been previously presented in the 9th International Conference on Marketing and Development, Thessaloniki, Greece, June 8–11, 2005.

¹Countries are classified as developed ones, if their average GDP per capita, in 1993–2001, exceeds the average value for the entire sample of 42 countries. These include USA, EU-15 (except Greece and Portugal, whereas Luxemburg is not included due to data limitations), Japan, Canada, Australia, Switzerland, Norway, New Zealand, Hong Kong, Singapore, and Israel. As developing countries the following are considered: Greece, Portugal, Poland, Hungary, Romania, Turkey, Egypt, China, India, Korea, Malaysia, Thailand, Indonesia, Philippines, Brazil, Argentina, Chile, Colombia, Mexico, and Venezuela.

²Several other authors (e.g. Dewan and Kraemer, 2000; Brynjolfsson and Hitt, 1993) provide econometric specifications to test the ICT impact at the aggregate or firm level starting from a Cobb–Douglas specification.

³Barro and Sala-i-Martin (1992) justify this approach when studying economies in discrete time periods. Further-more, Beck and Levine (2004) have recently used this approach to investigate the growth effect of financial development indicators.

⁴The one year lagged variable was preferred for reasons of retaining as many observations as possible and fully exploit information across time. Further justification is provided by Barro and Sala-i-Martin (1992).

⁵A low value of this indicator (values range between 1 and 10) implies a high level of corruption believed to exist in the corresponding country.

⁶The Hausman statistic is distributed as a chi-square variable whose value reaches 41.07 (p-value: 0.00), in the entire sample, when the initial hypothesis is that the difference in coefficient estimates is not systematic. Similar values are obtained for the sub-samples of developed and developing countries.

⁷In countries for which no initial estimate is given, the capital stock variable is calculated as the sum of gross investments that have been realized until previous year minus their accumulated depreciation. The depreciation for each year is calculated using the Winfrey mortality function.

⁸Under Daveri (2002) assumptions, hardware investment is estimated as the 58.6% of total hardware spending, communications equipment is 33% of total communications spending, and software investment is about 205% of total software spending. Furthermore, according to Daveri (2002), the service lives of hardware, software and communications' equipment are equal to seven, four and eleven years, respectively, so it is necessary to go backwards to 1987, 1990 and 1983 and add up the investment flows to obtain the 1993 capital stock values. In doing so, we use the case of USA as a proxy for other countries by assuming that the growth rates of hardware, software and communication investments display similar patterns across countries. Concerning the deflation of nominal investment flows, hardware, software, and communication deflators are constructed, based on price indices provided by the Bureau of Economic Analysis (2004) and grounded on the assumption that the price level of ICT products against that of all other capital goods is the same for all countries (including USA). At this point, it is important to note that Daveri (2002) instead of comparing price level ratios, used ratios based on yearly price changes. Furthermore, the existing million dollar values are converted to internationally comparable figures by utilizing the PPP data provided by Heston et al. (2002). Finally, a log-normal probability distribution is employed for the purpose of ICT capital depreciation.

⁹The shares of total and ICT capital are obtained by dividing them to GDP. As it is evident, the value of total capital stock exceeds that of GDP, an empirical regularity found in many economies of the world (see the data provided by Heston and Summers, 1991).

¹⁰No multicollinearity problem exists even when ICT is replaced by its individual components (hardware, software, communications), which are not included for reasons of brevity.

¹¹By construction, the differenced error term is first-order correlated, but this does not imply that so does the original error term.

¹²Arellano and Bond (1991) have proposed a two-step estimator which results after relaxing the assumptions of independence and homoskedasticity and constructing a variance–covariance matrix obtained by the residuals of the first step. As shown by Arellano and Bond (1991) and Blundell and Bond (1998), the coefficient estimates of the two-step estimator are asymptotically more reliable but the asymptotic inference is more efficient in the case of the one-step estimator.