Why do countries have different broadband penetration rates?

Broadband penetration is a term I'm using to refer to the proportion of a country's citizens who have access to high-speed broadband Internet services. For my economics research seminar, I'm analyzing the factors that affect broadband penetration at the national. Using a fixed-effect regression, I model broadband penetration against a variety of factors such as GDP and population density. I use data freely available from the World Bank.

This paper is only a rough draft, but I welcome comments and constructive criticism that will help me refine my research. If you know of a paper that would be helpful, please send it along.

Contents

Introduction


Today’s markets are global, and both developed and developing countries are increasingly producing service-sector goods. Widespread access to high-speed Internet can be a competitive advantage in today’s knowledge economy (Choudrie and Lee 2004; OECD 2008).

At the same time, broadband infrastructure requires significant investment. Due to economies of scale, the market may be uncompetitive—dominated by a few large firms—and rural or low-income areas may be underserved (Gillett et al 2003; Choudrie and Lee 2004). Since broadband is excludable but largely nonrival—once the infrastructure has been installed in an area, one household’s use does not significantly diminish the service provided to other households—it might be useful to consider broadband as a public good. Governments may want to encourage broadband penetration in order to address equity concerns or modernize their countries.

Gillett et al (2003) suggest that to promote broadband penetration, a government can assume the role of 1) user, 2) rule-maker, 3) financier, and/or 4) infrastructure developer. In the first role, the government stimulates demand; in the latter three, it encourages supply.

In 1999 and 2000, South Korea experimented with several initiatives to encourage broadband penetration (Choudrie and Lee 2004). With programs such as “Cyber Korea 21” and “Ten Million People Internet Education,” the government promoted Internet literacy to stimulate demand. The government also deregulated the telecommunications industry, provided “US$77m of loans [to service providers] at preferential rates,” and prepayed for broadband service to public buildings. The government planned to commit an additional US$926m to extend broadband service to rural areas by 2005. Broadband adoption in South Korea was spurred by the prevalence of Internet cafés that introduced the population to high-speed Internet access and by a highly-competitive telecommunications sector that provided next-generation services at cutthroat rates. South Korea’s high population density also facilitated the spread of the new infrastructure.

The literature suggests that broadband penetration will be influenced by several demand- and supply-side factors.

Demand
  • Population: Aggregate demand for broadband will be greater among larger populations. One also expects that demand will be greater from more youthful populations. 
  • Wealth: Since broadband is a “premium” service, demand should be greater from wealthier countries. 
  • Education: Gillett et al (2003) argue that “white collar workers . . . are more likely to use advanced communications services.” One expects that broadband penetration will be greater in countries that have more highly-educated populations and that trade primarily in communications and services. 
Supply
  • Telecommunications sector: One expects that supply will be greater in countries with robust telecommunications industries. 
  • Population density: Broadband penetration should be greater in densely-populated countries. 

Gillett et al have verified this framework at the level of local municipalities in the United States using data provided by the American Public Power Association (2003), but no study has yet examined broadband penetration rates at the international level.

Empirical Framework


To analyze broadband penetration from time-varying panel data, a linear fixed-effects regression model is used.
where

  • Y = broadband penetration
  • X = demand-side variables
  • Z = supply-side variables

Broadband penetration is measured as the number of fixed broadband Internet subscribers per 100 citizens.

Demand-side variables include the total population as well as the percentage of the population between ages 15 and 64 and the percentage of the population over the age of 64. Since the total population varies widely among the sample data, it has been transformed logarithmically. The collinearity between these three variables is insignificant. One expects all to be positively correlated with broadband penetration.

The latter two variables are included in order to isolate which age group is driving the demand for broadband. One expects that youthful populations will have a greater demand for broadband.

GDP—transformed logarithmically—has been used to measure the wealth of a country. This variable should be positively correlated with broadband penetration. Education should also increase with income levels, and although data on adult and youth literacy rates was available, it would have presented collinearity problems and was not rich enough to be included in the regression. Hence, GDP will be used as a proxy for both the wealth and education levels of a country.

The “white-collar” demand of a country is measured by the logarithmic transformations of commercial service imports and exports (both in U.S. dollars). Countries that trade a high volume of commercial services will likely have a greater demand for high-speed Internet access. The demand of highly-educated citizens is also estimated by the number of researchers and technicians per 1 million citizens. Broadband penetration should be greater in countries that are engaged in more “knowledge work.”

The sole supply-side variable is mobile cellular subscriptions per 100 citizens. Since cellular subscription rates often increase as broadband subscription rates increase (Yang et al 2009), and since broadband providers often also provide cellular service (Yang et al 2009), this variable should indicate the robustness of a country’s telecommunications sector. However, especially in developing countries, high cellular subscription rates may also indicate a lack of fixed-line infrastructure. Even in developed countries (for example, Japan), cellular subscriptions could substitute for fixed-line broadband access and may dampen demand for broadband.

Data


The World Bank freely provides data on the preceding variables for a variety of countries. However, the data was more robust for some countries than others. The dataset excludes countries for which broadband penetration rates were not available for half of the sample period (years 2001 through 2009). Since even fixed effects regressions require variation in order to run, countries for which broadband penetration did not vary from 0 during the sample period were also excluded. Hence, as one might expect, countries that were excluded from the dataset were primarily small, developing nations for which data was not readily available. The final dataset still includes a mix both developing and developed countries. (See the Appendix for a full list. The dataset is also available from the researcher upon request.)

68 countries and 318 observations were included in the sample set.

Results


Figure 1
An F-value of 79.23 and an R2 of 0.74 indicate that this regression is highly significant and explains a large degree of the variation in international broadband penetration rates. Nearly all of the variables, with the exception of GDP and commercial service exports, are as significant.

The parameter total population carries a coefficient of 26.16 and is highly significant, indicating that total population is by far the largest determinant of broadband penetration. Surprisingly, the percentage of the population between ages 15 and 64 significantly decreases broadband penetration, while the percentage of population over the age of 64 increases it by half the amount but is insignificant. These results suggest that while total population drives the demand for broadband, the working age and retiree population of a country does not.

These results could arise from including a disproportionate number of developing countries in the dataset, since in developing countries the working age population may be more likely to be engaged in manufacturing or manual labor—occupations which do not demand high-speed broadband access—rather than knowledge work. Since the number of researchers and technicians in a country do significantly drive demand (albeit only slightly), this conclusion seems valid.

The wealth and education of a country, as measured by the logarithmic transformation of GDP, would decrease the demand for broadband were it significant. Income distribution rather than total wealth may be a more important driver of broadband demand, since broadband (as a premium communication service) is more likely to be demanded by the upper classes. This is supported by the results of the regression in Figure 2. With an F-value of 133.62 and an R2 of 0.83, this regression is actually more significant (but less telling) than the one in Figure 1. In the regression in Figure 2, GDP per capita increases broadband penetration only slightly but is highly significant.

After total population, the second-largest determinant of broadband penetration appears to be a country’s commercial service imports. While commercial service exports also positively influences broadband penetration, this parameter is insignificant. These results indicate that countries heavily trading in services will have greater broadband penetration rates. Only commercial service imports could drive broadband demand if wealthier countries are importing services from developing ones, e.g., if developed nations like the United States are outsourcing programming or accounting jobs to developing countries like China and India.

Mobile cellular subscriptions also influence broadband penetration, albeit only slightly. This suggests that a robust telecommunications sector is integral to widespread broadband penetration. Because mobile cellular subscriptions appear to increase broadband penetration, there is no significant trade-off between cellular subscriptions and broadband subscriptions. Instead, the two may be complements.

Figure 2
I plan to run a similar regression using only OECD countries.

Conclusions


At the international level, demand-side variables determine broadband penetration more than supply-side variables. The largest determinant of demand is total population, which is largely outside a government’s control. From a policy perspective, governments can best encourage broadband penetration by stimulating the telecommunications and service sectors of their economies. Service-sector industries drive the demand for broadband and provide a healthy market for premium, twenty-first century telecommunications services.

Literature Review


Sharon E. Gillett, William H. Lehr, and Carlos Osorio, “Local Government Broadband Initiatives,” 2003 [pdf].

Jyoti Choudrie and Heejin Lee, “Broadband development in South Korea: institutional and cultural factors,” 2004.

Jyoti Choudrie and Anastasia Papazafeiropoulou, “Lessons learnt from the broadband diffusion in South Korea and the UK: Implications for future government intervention in technology diffusion,” 2006 [pdf].

Organisation for Economic Co-operation and Development, “The Future of the Internet Economy,” 2008 [pdf].

Heedong Yang, Youngjin Yoo, Kalle Lyytinen, and Joon-Ho Ahn, “Diffusion of Broadband Mobile Services in Korea: The Role of Standards and Its Impact on Diffusion of Complex Technology System,” 2009 [download pdf].

Acknowledgements


My thanks to Professor Ron Cheung for generously guiding my research.

Appendix


The full list of countries in the dataset is as follows: Afghanistan, Albania, Algeria, Andorra, Angola, Antigua and Barbuda, Argentina, Armenia, Aruba, Australia, Austria, the Bahamas, Bahrain, Barbados, Belarus, Belgium, Belize, Benin, Bermuda, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil, Brunei Darussalam, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon, Canada, Cape Verde, Chad, Chile, China, Colombia, Congo, (Democratic Republic), Costa Rica, Cote d'Ivoire, Croatia, Cuba, Cyprus, Czech Republic, Denmark, Djibouti, Dominica, Dominican Republic, Ecuador, Egypt, El Salvador, Equatorial Guinea, Estonia, Ethiopia, Faeroe Islands, Fiji, Finland, France, French Polynesia, Gabon, Gambia, Georgia, Germany, Ghana, Greece, Greenland, Grenada, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, Honduras, Hong Kong SAR, Hungary, Iceland, India, Indonesia, Iran, Iraq, Ireland, Israel, Italy, Jamaica, Japan, Jordan, Kazakhstan, Kenya, Korea (Republic of), Kuwait, Kyrgyz Republic, Lao PDR, Latvia, Lebanon, Lesotho, Libya, Liechtenstein, Lithuania, Luxembourg, China, Macedonia, Madagascar, Malawi, Malaysia, Maldives, Mali, Malta, Mauritania, Mauritius, Mexico, Micronesia, Moldova, Mongolia, Montenegro, Morocco, Mozambique, Myanmar, Namibia, Nepal, Netherlands, New Caledonia, New Zealand, Nicaragua, Niger, Nigeria, Northern Mariana Islands, Norway, Oman, Pakistan, Palau, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Poland, Portugal, Puerto Rico, Qatar, Romania, Russia, Rwanda, Samoa, San Marino, Sao Tome and Principe, Saudi Arabia, Senegal, Serbia, Seychelles, Sierra Leone, Singapore, Slovak Republic, Slovenia, Solomon Islands, Somalia, South Africa, Spain, Sri Lanka, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Sudan, Suriname, Swaziland, Sweden, Switzerland, Syrian Arab Republic, Tajikistan, Tanzania, Thailand, Togo, Tonga, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, United Arab Emirates, United Kingdom, United States, Uruguay, Uzbekistan, Vanuatu, Venezuela, Vietnam, Virgin Islands (U.S.), West Bank and Gaza, Yemen, Zambia, Zimbabwe.

4 Ways the Web is Changing Communications

The web is changing how we organize information in four ways:
  1. Taxonomies (think the Dewey Decimal System or biological classification) used to be necessary to organize information, but the web recognizes that categories are often fluid. Would my hypertext novella be found in the nonfiction, new media, prose or poetry section of a bookstore? The answer is neither and all four—it depends on who is looking for the book. With hyperlinks and tags, the web's architecture is primarily relational.
  2. We rarely enter websites via their splash pages and instead access the page we want via Google, Bing, or another search engine. Even though technology allows news to break faster and memes to spread virally, old information is not so much forgotten as pushed to the fringes. For instance, Google "Ryan Gosling" and three of the top six hits will be images from The Notebook, even though the movie is seven years old and Gosling has starred in a movie every year since (including 2010's Blue Valentine, which won two awards and was nominated for seventeen more). Since old information can be accessed as easily as new, the web is inherently nonlinear.
  3. With a blockbuster movie, a prime-time television show, or even a print book, the dialogue is one-way—from artist to audience. But the web has given consumers a voice and it rewards them for using it. If web content does not facilitate audience participation, surfers will take their attention elsewhere. The web is by nature interactive.
  4. Literary writing is largely absent from the web because designers have not yet found ways to humanistically display longer works. The scroll bar on the side of a browser window actually represents a technological regression (think Egyptian papyrus scrolls). Longer documents, unless intuitively organized, lose all the advantages of print books (the abilities to instantly jump from beginning to end and to see how far you have progressed in the text). Because screen reading can be somewhat uncomfortable, web writing tends to be digestible in shorter chunks (e.g., the length of a blog post).
For my creative writing senior project, I'm writing a prose-poetry hypertext novella—a book-length work organized as a website. My novella tries to takes advantage of the above properties. It's partly an homage to my pre-digital childhood and partly a collage of memories (our present encompasses our past). Think of it as a fictional Wikipedia or a digital choose-your-own-adventure—it's an experiment in nonlinear storytelling.

I'm hoping these guys will publish it.