Distribution of Internet Use: A Cross-National Analysis1, The

Journal of Private Enterprise, Fall 2003 by Fallow, Chris, Norton, Seth

The wide adoption of the Internet is one of the most dramatic economic and technological developments in recent years. The International Telecommunications Union reports that in 2002 there were 591 million Internet users worldwide. In some countries, Internet adoption has had wide impact beyond pure consumption-affecting the internal operations of firms as well as dramatic innovations in marketing (Kenney 2003).

The dispersion of the Internet across the populations of the world is also dramatic. A common measure for the incidence of the Internet is the number of "internet hosts" in a country. The measure is "... the number of computers direcdy connected to the worldwide network of interconnected computer systems" (World Bank 1999,245). The International Telecommunications Union reports that as of 1996, 96 percent of the Internet host computers were in high income countries diat included only 16 percent of the world's population. There were more Internet hosts in Finland than there were in all countries comprising Latin America and the Caribbean. New York City had more Internet hosts than all the countries of Africa.2

Table 1 contains some highlights of the variation of Internet connections across the world. The data in Table 1 show remarkable dispersion of Internet hosts. Angola has one hundred Internet host for every 10,000 inhabitants. The Netherlands has more than nineteen hundred Internet hosts per 10,000 inhabitants. Indeed, the gap between the relatively poor countries in column one and the relatively rich countries in column two is striking and certainly seems to provide a simple explanation-the dispersion of the Internet reflects the dispersion of the income across the countries of the world.

The data in Table 1 also reveal other dimension of dispersion. Consider the comparison of Angola and Botswana. Botswana has proportionately more than seven hundred times more Internet hosts than Angola. The Philippines have proportionately more than 26 times more Internet hosts than Madagascar. Thus, there is considerable dispersion within the relatively poor countries and even within the relatively rich countries, as evidenced by comparisons of Australia and Austria. Understanding the reasons for such dispersion seems central to understanding the modern information based economy. This paper examines the question why Internet use is so prevalent in some nations and nearly nonexistent in other countries.

Background

A popular explanation for economic growth is the neoclassical growth model, most commonly associated with Solow, but also widely cited in an augmented form stressing human capital as well as traditional capital (Solow, 1956; Mankiw, Romer, and Weil, 1992). The model emphasizes the role of savings or investment, population growth, and the natural diffusion of technology across counties as determinants of growth. A simple interpretation would be that if disparities exist in some application of technology, then they should be reasonably transitory. New technology should ultimately be widely distributed around the world. When new technology is developed, it is presumed to diffuse like an epidemic worldwide (Baumol, 1994). The framework implies that all countries worldwide should possess similar stocks of technology in equilibrium due to decreasing returns or imitation costs that are less than innovation costs. In short, neoclassical theory suggests there are plausible reasons to expect similar stocks of technology across countries. However, as a first approximation, the data in Table 1 clearly do not support the homogeneity of technology across countries. It is true that widespread Internet use is recent, so the market may not be in equilibrium. However, the data in no way fit closely with neoclassical growth model.

Other explanations for growth exist. These include "endogenous growth models" that permit continued relatively high growth for rich countries and that are attributable to increasing returns from various investments (Romer, 1986). Further explanations for growth focus on the role of institutions as engines of economic growth (Barro, 1992; Barro and Salia-I-I Martin, 1995; Knack and Keefer, 1995; North, 1981; Scully, 1988). Such models constitute a loose theoretical perspective that could be labeled the neoinstitutional theory of growth. This alternative explanation permits important differences across countries in terms of investment and growth, including the stock of technology. It is compatible with the neoclassical theory or the endogenous growth theory, but it also asserts that a nation's technology is driven by the efforts put forth by its citizens. The efforts citizens exert are dependent upon incentives established in the nation. These in turn are established through laws and customs in a country. Laws and customs can encourage or discourage productive behavior and the gains-from-trade by lowering the costs of contracting and the costs of less formal transactions.

A related issue is that a number of human characteristics vary across countries and contribute to the speed and extent of the diffusion of technology. Abramovitz (1994, 97) classifies these factors under the rubric "social capabilities" The level of human capital investment (e.g., years of education) of the citizenry is the most salient example.