Journal of Internet Business

Issue 3 - 2006

ECONOMICS OF THE INTERNET: A COMPARISON BETWEEN ASIA-PACIFIC AND OECD Ramesh Mohan* Bryant University

ABSTRACT The revolutionary trends in computer software and hardware, especially after the surge in Internet utilization in the mid-1990s, has dramatically altered the way organizations and individuals operate. The importance of internet and its interactions with the economy today have established the need for economists to address this matter. In this study, Internet users as a percentage of population was regressed on various independent variables using panel data for 22 OECD and 11 Asia-Pacific countries.

Keywords: Economics of the Internet, Asia-Pacific, OECD.

* Assistant Professor, Department of Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI02917 USA. Phone: +1-401- 232-6379. Email:[email protected].

Journal of Internet Business

Issue 3 - 2006

INTRODUCTION In recent years, advances in Information and Communication Technology (ICT) have largely driven economic changes such as restructuring of economies, globalization of markets, and expansion in trade and capital flows as well as increases in information activities. As the economy heads into liberalization, Internet plays an important role in spearheading the development of each sector of the economy. Today’s corporate leaders are facing a number of challenges in managing ICT: the level of understanding required for using ICT, the facilitating task of ICT for knowledge processes, the status of the ICT infrastructure, and other user contemplations (Hendriks, 2001). These directives have become particularly pertinent in all sectors that face new competitive pressures at the global level. An operation with outdated ICT systems is fast becoming a recipe for failure. The importance of Internet technology and its interactions with the economy today have established the need for economists to address this matter.

The R&D

activities induced by new technologies are powerful tools that can be used to allocate resources efficiently and economically. Since technical change is difficult to separate from economic change, knowledge of technology could stimulate economic progress. Technology simplifies and reduces tasks that would otherwise require manual skills and effort. The use of appropriate technology in well-planned systems can have dramatic effects on operations. Properly applied, it can increase productivity. Technology can also influence the way organizations interact with customers, suppliers, and competitors.

Journal of Internet Business

Issue 3 - 2006

Internet Schumpeter (1939) foresaw that innovation (Internet in this study) not only changed production and produced a new product but also created a gateway to a new market. Powerful and cheaper hardware development greatly complemented the development of software. The revolutionary trends in computer software and hardware, especially after the surge in Internet utilization in the mid-1990s, has dramatically altered the way organizations and individuals operate. According to the International Telecommunication Union (1999) report, it took 74 years for the telephone to reach 50 million users, but it took only four years for the Internet users to reach 50 million. The report further examines the usage and benefits of the

Internet

in

various

sectors

including

business,

health,

education,

and

telecommunication. The report concluded that the price of Internet usage, shortage of telephone lines, and English language proficiency were major barriers for Internet diffusion in developing countries. Four major studies examined the determinants of Internet usage. Hargittai (1999), using the number of Internet hosts for Western Europe as dependent variable, found that GDP per capita, number of telephone lines, monopoly provider, and number of personal computers to be the main determinants. Kiiski and Pohjala (2001), using the Gompertz model of technology diffusion, investigated the factors that determined technology diffusion for OECD and a larger sample. Using Internet hosts per capita as the dependent variable, they found that only two variables, GDP per capita and Internet access cost are significant for the OECD sample.

Dasgupta et al. (2001) found that urban population

and nature of regulation or competition in the sector were the important variables that influenced Internet usage for a group of developing countries in Africa, Asia, and Latin

Journal of Internet Business

Issue 3 - 2006

America. The empirical analysis of Muller (2002) in Central and Eastern Europe showed that per capita income, openness to foreign influences, educational level, state of liberalization of the telecommunications sector, and the degree of political and civil freedoms were the notable determinants of Internet usage.

DATA AND VARIABLES In this study, the dependent variable is Internet user as a percentage of population for a period of 1994 to 2002. The choice of independent variables for the Internet usage regression was based on Hargittai (1999), Kiiski and Pohjala (2001), Dasgupta et al. (2001), and Muller (2002). The independent variables are number of internet hosts (HOST), telephone lines per 1000 people (TELLINE), rural population as a percentage of total population (RURAL), ICT expenditure per capita (ICTPC), personal computers installed in education facilities (PCEDU), mobile phones per 1000 people (MOBILE), household final consumption expenditure per capita in constant 1995 US$ (HFCE), average telephone cost of local calls in US$ per 3 minutes (TELCOST), cable TV subscribers per 1000 people (CABTV), communications, computers, etc. expenditure as a percentage of secondary

service exports (ICTEXP), and percentage of the labor force with

education

(LFSE).

The

data

were

obtained

from

International

Telecommunication Union and World Development Indicators. Since the data for the variables are available for the period of 1994-2002, a panel data study for 11 Asia-Pacific countries and 22 OECD countries was conducted. See Table 1 for acronym, descriptions, expected signs, and justifications for using the variable in the Internet usage regression.

Journal of Internet Business

Issue 3 - 2006

Table 1: Variables, Descriptions and Their Relationships with Internet Usage Description Internet host

HOST

Expected sign +

TELLINE

Telephone lines (per 1000 people)

+

RURAL

Rural population (% of total population) ICT per capita

_

ICTPC

+

PCEDU

Personal Computer Installed in Education

+

MOBILE

Mobile phones (per 1000 people) Household Final Consumption expenditure per capita (constant 1995 US$) Average telephone cost of local calls (US$ per 3 mins) Cable TV subscribers (per 1000 people)

+

ICTEXP

Communications, computers, etc. (% of service exports).

+

LFSE

Labor force with Secondary Education (% of total)

+

HFCE

TELCOST

CABTV

+

Rationale The more hosts the more Internet usage since competition drives Internet access prices down. Availability of basic infrastructure for dial-up access induces internet usage. Higher rural population, lesser internet usage. Higher ICT expenditure per capita indicates importance given to ICT in the country, which enhances Internet access. To capture how far the number of PCs in education encourages internet usage. Proxy for telecommunication industry liberalization, and thus ICT. Higher purchasing power indicates Internet access affordability.

_

Proxy for the cost of Internet access. Higher cost, lower usage.

+

Proxy for high-speed internet availability since most providers offer them as a package. Captures whether availability of ICT related industry helps Internet accessibility. Higher literacy among labor force induces Internet usage.

Muller (2002) found that the correlation coefficient between the Internet host (HOST) and Internet usage is around 0.703 in the European Union and Central and Eastern Europe. The author further indicated a large variation of the number of Internet hosts (per 10,000 inhabitants) in 2001, ranging from 0.08 in Uzbekistan to 3,714 in the U.S.

Journal of Internet Business

Issue 3 - 2006

Telephone mainlines (TELLINE) is the basic infrastructural component for Internet dial-up access. The ITU (1999) report revealed that in developing nations infrastructure shortage, especially telephone lines, is a big barrier to Internet access. However, although Internet access through telephone dial-up lines is increasingly facing competition and giving way to broadband technology, which works 100 times faster through fiber optic cables, dial-up still plays an important role in many developing countries. Internet access among the rural population (RURAL) is very limited even in developed nations.1 Investigating the level of internet usage in rural areas is important in determining the level of market failure in terms of Internet awareness in rural communities. It is critical that investments are made to expand the extension of and familiarity with the Internet in rural areas. Comparatively, lower levels of Internet usage in rural areas can possibly be attributed to the fact that the majority of the rural population is older, less wealthy, have less education, as well as being faced with poorer infrastructure. The level of ICT per capita (ICTPC) indicates the importance given to ICT by the private and public sectors. One would expect a higher numbers of Internet users in a country that promotes ICT usage. The number of personal computers installed in education (PCEDU) is used to capture how much computer availability in education encourages Internet usage. One study of college and high-school students showed that those who studied algebra with computers instead of conventional paper and pencil,

1

See Bell, Reddy, and Raine (2004), Rural Areas and the Internet, PEW Internet and American Life Project, available at http://www.pewinternet.org/reports/pdfs/PIP_Rural_Report.pdf.

Journal of Internet Business

Issue 3 - 2006

performed well in a series of exams. Furthermore, computers have been found to improve students' communication skills and the quality of their presentations.2 Mobile phone usage (MOBILE) is used as a proxy for telecommunication industry liberalization.3 Flexibility is important in improving levels of communication through diversifying accessibility to different types of telecommunications systems. Liberalization in the telecommunications and communications industry encourages investment and competition in the sector. This potentially indicates the importance given to ICT in the country. Household final consumption expenditure (HFCE) is the market value of all goods and services purchased by households. It is an indication of purchasing power of consumers, and hence, indicates Internet access affordability. Higher average telephone cost of local calls (TELCOST), as a proxy for the cost of Internet access, is expected to have an adverse impact on Internet usage. Cable TV subscribers (CABTV) is used as a proxy for high-speed Internet access. A study using capital cost models based on case studies of the PSICable use in Cambridge, MA, and the Internet over ISDN service offered by Internex, Inc., in the San Francisco, CA, area found that cable's shared-bandwidth method has greater economic distinctiveness: per-subscriber cost per bit of peak bandwidth for 500 Kbps cable service is $0.60 versus close to $16 for ISDN. In terms of service, cable-based access can support 2

Study quoted in New York Times editorial, Rethinking Computers: More Than A Toy? November 30,1997.

3

Research work at Keio University, Shonan Fujisawa, Japan involved conducting an ethnography study of new uses of and services for cellular phones found mobile e-mail increasingly replacing voice telephony as the leading means of telecommunications among youngsters. http://www.ojr.org/japan/wireless/1043770650.php

Journal of Internet Business

Issue 3 - 2006

both full-time Internet connections and higher peak bandwidths, such as a four Mbps cable service that offers 32 times the peak bandwidth of ISDN.4 ICTEXP is used to capture whether availability of ICT related industry helps Internet accessibility. One would expect that higher export of ICT related goods and services positively influence Internet usage. Literacy among the labor force, as measured by the labor force that has a secondary education as a percentage of the total labor force (LFSE), is likely to be an important factor determining Internet access. The ITU (1999) report found that education is one of the key aspects for Internet diffusion. Development of human resources in ICT and other skill areas is one of the key requirements for the information economy to prosper.

EMPIRICAL METHODOLOGY Economists generally use the logit model in technology diffusion models. Equation below illustrates the reduced form of a logit model: P [ Y it = 1 ] =

e β ' X it 1 + e β 'X

it

(1)

where Y is the internet users as a percentage of population. X are the independent variables. i = 1,……….N

indexes country.

t = 1,……….T(i) indexes period. X = k x n matrix, β = 1 x k matrix, and Y = 1 x n matrix. Transforming equation (1) by taking logarithms on both sides, and rewriting it produces

4

See http://rpcp.mit.edu/

Journal of Internet Business

⎛ Y it ln ⎜⎜ ⎝ 1 − Y it

Issue 3 - 2006

⎞ ⎟⎟ = ln e β ' X it = β ' X ⎠

(

)

it

(7)

The transformation generates a linear regression of the form Yit = β ' X it + ε it

(8)

Following Greene (2002), for the proportion data, when one carries out the transformation, the weighted least squares with dependent variable is given by5

⎡ Y ⎤ logit ( P (Y it )) = ln ⎢ it ⎥ = β ' X it ⎣ 1 − Y it ⎦

(9)

In general, FE logit transformation of proportion data model was estimated in the form of:

⎡ Y ⎤ logit( P(Yit )) = ln ⎢ it ⎥ =η1δ 1it + η 2 δ 2it + ......... + β ' X it + µ it ⎣1 − Yit ⎦ where

(10)

πit is the Internet users as a percentage of population in country i =1,……, N, year t = 1,….,T(i). Xit is the vector of independent variables. δjit is the group specific year dummy variables. ηi is the individual specific constant or the country effect. µit is a classical disturbance with E[µit] = 0, var[µit] = σ2µ . White’s robust, heteroscedasticity corrected covariance matrix was used.

EMPIRICAL RESULTS In Table 2, regression results of the FE logit transformation of Internet usage for 22 OECD countries and 11 Asia-Pacific countries are reported. Internet users as a

5

Two limitations apply: First, the transformation is not possible when the dependent variable is zero or one. In this case, the result will be a missing value. The second limitation is that when the fraction is from large population; the variance will be low, leading to low standard error and high t-ratios in the minimum chi-square regression.

Journal of Internet Business

Issue 3 - 2006

percentage of population was regressed on various independent variables using panel data. For OECD, eight of the 11 variables were statistically significant and nine of the 11 variables were significant in the case of Asia-Pacific at the 10% level or better.

Table 2: Regression Estimates of Internet Usage Variables

HOST

OECD Coefficient 0.913

t-ratio 11.034***

ASIA-PACIFIC Coefficient 0.806

t-ratio 10.788***

TELLINE

-1.098

-3.071**

1.133

2.612**

RURAL

-3.868

-2.534**

-3.522

-2.532**

ICTPC

0.845

1.300

0.699

2.034**

PCEDU

0.365

1.370

0.298

1.209

MOBILE

0.336

2.789***

0.320

2.699**

HFCE

3.334

2.478**

3.900

3.996***

TELCOST

0.654

1.987

-0.980

-1.709**

CABTV

0.256

1.402*

-0.209

-0.335

ICTEXP

0.433

1.677**

0.804

2.236**

LFSE

0.633

2.666**

0.114

1.697*

R2

0.652

0.678

Adjusted R2

0.638

0.671

F-Value

80.16***

125.33***

No. of obs.

198

99

Note: *** , **, and * denotes significance at the 1%, 5%, and 10% respectively.

Journal of Internet Business

Issue 3 - 2006

In both the Asia- Pacific and OECD sub-samples, the number of hosts is positive and highly significantly related to Internet usage. Contrary to the expected sign, number of telephone lines (TELLINE) has a negative sign and is significant for OECD. One possible explanation is that investment in telecommunication was made perhaps 50 to 100 years ago. Current additions of telephone lines in the OECD are for upgrades, thus producing a negative relationship. However, TELLINE is positive and significant in the case of Asia-Pacific, as expected. The telephone line is the basic infrastructure for Internet dial-up access in many developing nations. This is supported by the results of Hargittai (1999) and Kiiski and Pohjola (2001). Parameter estimates of rural population (RURAL) were strongly negative and significant in the Asia-Pacific and OECD samples. This confirms the expectation that Internet access among rural populations is very limited not only in the Asia-Pacific but also in the OECD. This is not surprising since most of the rural population are employed in agriculture, are less literate, and many are senior citizens. In addition, rural areas have less well-developed infrastructure. ICT per capita (ICTPC) was positively and significantly (at the 5% level) associated with Internet usage in Asia Pacific but not significant in the OECD. The estimates of personal computers installed in education (PCEDU) were not significant in either the OECD or Asia-Pacific sample. Mobile phone subscribers (MOBILE), as a proxy for a deregulated telecommunication market, were significant in both OECD (at 1% level) and Asia-Pacific (at 5% level). Muller (2002) also found this variable to be highly significant. As Muller concluded, perhaps mobile phone use not only captures the level of liberalization but the

Journal of Internet Business

Issue 3 - 2006

overall telecommunication sector’s role in enhancing Internet usage. Mobile phone usage is increasingly changing communication patterns, especially short messaging system (SMS) and mobile-email. Household final consumption expenditure (HFCE) is highly positive and significant in both the OECD and Asia-Pacific.

The variable indicates that high

purchasing power of consumers in a country indicates Internet access affordability and thus, higher Internet usage. The ITU (1999) report revealed that wealth is an important factor that encourages Internet Usage. In many low-income countries, lower Internet usage is partly associated with high access cost. As expected, higher average telephone cost of local calls (TELCOST), used as the proxy for the cost of Internet access, was negatively related to Internet usage and significant at the 5% level in Asia-Pacific. However, TELCOST has the wrong sign and is not statistically significant in the case of OECD. This is perhaps partly due to the increasing popularity of high-speed Internet access through fiber-optic cable in many high-income countries. Cable TV subscribers (CABTV) was used to capture the high-speed Internet access. The broadband technology works through fiber optic cables used to deliver cable television. Parameter estimate of CABTV was positive and statistically significant in the OECD at 10% level but negative and not significant in the Asia Pacific sample. This new technology, which allows connection speeds 100 times faster than those currently in use, is more popular in the OECD, and is slowly increasing in Asia-Pacific. Even in the case

Journal of Internet Business

Issue 3 - 2006

of high-income countries, Internet providers are merging with cable companies to use existing fiber optic cables, primarily to avoid high infrastructure cost.6 The empirical estimates of ICTEXP were positive and significant (5% level) in both the OECD and Asia-Pacific, which indicates that export of ICT related goods and services has a positive impact on Internet usage. This clearly shows that availability of ICT related industry in the country helps Internet accessibility. As found in the ITU (1999) study, literacy among the labor force, as measured by the fraction of labor force with secondary education (LFSE), was positive and significant at 5% and 10% levels in the OECD and the Asia-Pacific, respectively. However, the coefficient of LFSE in AsiaPacific is comparatively very small. Literacy is an important factor for Internet access, and most of the countries under investigation have comparatively high literacy rates. There are several limitations to the current study. First, Internet access costs data were not available for most of the countries under investigation. Using telephone local access cost as a proxy is not a good choice for explaining accessibility charges. Using telephone access cost was misleading in many countries where high-speed Internet access is slowly becoming more prominent. Second, aggregating OECD and Asia-Pacific data into a larger sample and estimating a logit transformation yielded very poor estimation. This is primarily due to the second limitation of the logit transformation method, discussed in the empirical methodology (footnote 5).

CONCLUSIONS

6

For example, the AOL-Time Warner and AT&T-TCI-Media One mergers in the U.S.

Journal of Internet Business

Issue 3 - 2006

In terms of Internet usage, numbers of hosts, mobile phone subscribers, household final consumption expenditure, ICT exports, and labor force with secondary education were positively correlated with Internet usage in both samples.

The rural population

estimates were significant and negative in both the Asia-Pacific and the OECD samples. ICT per capita and telephone line show a relatively strong and significantly positive effect on Internet usage in Asia-Pacific. CABTV was positively associated with Internet usage in the OECD, and telephone local access cost was negatively related to Internet usage in Asia-Pacific. Looking at the Internet usage empirical results, a few important policy issues should be considered.

Internet usage was inversely related to rural population in both

samples. This suggests that Internet usage is primarily concentrated in the cities. Internet usage may be lower in rural areas because many people are older and poorer, many are illiterate and economically poorer, and infrastructure is not as developed as in the urban areas. It is critical to devise policies for investments in communication infrastructure to expand the knowledge and accessibility of Internet in the rural areas. Probably government intervention is needed to correct the market failure by providing relevant ‘info-structure’. Mobile phone technology is becoming increasingly popular.7 Development of Wireless Application Protocol (WAP)-enabled mobile phone in developing countries is really amazing. An important breakthrough is mobile messaging through SMS. The introduction of MMS (Multi Media Messaging) in 2002 takes SMS beyond simple text 7

See IDC research report on various uses of mobile technology by users in Exploring Usage Models In Mobility: A Cluster Analysis of Mobile Users. Available at http://www.idcresearch.com/getdoc.jsp?containerId=pr2003_11_20_091841.

Journal of Internet Business

Issue 3 - 2006

messages. The advancement from SMS to MMS for mobile phones is comparable to the transition from DOS to Windows for personal computers.8

In many developing

countries, SMS through mobile phones has become affordable and a popular mode of communication, even among low-income households. Dasgupta et al. (2001) pointed out that e-mail or SMS through mobile phones is less expensive for the average household in developing countries, and easier and cheaper to learn than computers. The International Data Corporation (IDC) projected that mobile phones will dominate as the Internet access platform compared to telephone lines and cable TV (high-speed Internet) in developing countries.

Thus, two important

implications should be considered. First, including mobile technology as an important component of ICT and/or Internet diffusion is an important area for future research. Second, liberalization of the telecommunication industry in any given country would allow ICT to contribute more to the development of that nation. The number of cable TV subscribers (CABTV) was used to capture high-speed Internet access. One disturbing trend that should be addressed in future research is that Internet providers are merging with cable companies to use existing fiber optic cables, primarily to avoid high infrastructure costs. The dominance of cable TV providers in the high-speed Internet market creates a monopoly in the media and communication markets. This implies an increase in market power for a selected number of firms and implies a lack of choices for consumers, which might lead to an increase in prices and poor service. Cooper (2003) suggests the need for policymakers to create new sets of rules and

8

See http://www.mobilesms.com.

Journal of Internet Business

Issue 3 - 2006

regulations to ensure competition in the telecommunication and Internet industry, or broadly speaking, the ICT sector as a whole, given the current trend.

Journal of Internet Business

Issue 3 - 2006

REFERENCES Dasgupta, S., Hall, S., and Wheeler, D., 2000, Policy Reform, Economic Growth and the Digital Divide:An Econometric Analysis, World Bank Working Paper No. 2567, Washington D.C. Greene, W. H., 2002, Econometric Analysis, Prentice Hall. Hargittai, E., 1999, Weaving the Western Web: Explaining Differences in Internet Connectivity among OECD countries, Telecommunication Policy, Vol.23:701-18. Hendriks, Paul H.J., 2001, Many Rivers To Cross: From ICT To Knowledge Management Systems, Journal of Information Technology, Vol. 16, No. 2. Hsiao, Cheng, 2002, Analysis of Panel data, Cambridge University Press. ITU (1999), Challenges To The Network: Internet For Development, Article No 16212 Oct. 1999, International Telecommunication Union, Geneva, Switzerland. ________ (2002), World Telecommunication Development Report: Reinventing Telecoms, International Telecommunication Union, Geneva, Switzerland. Kiiski, S., and Pohjola, M., 2001, Cross Country Diffusion of the Internet, UNU/WIDER Discussion Paper No. 2001/11. Muller, P., 2002, Internet Use in Transition Economies: Economic and Institutional Determinants, UNU/Wider Discussion Paper No.2002/95. World Bank, 2003, World Development Indicators, The World Bank.