Digital divide: an econometric study of the determinants in information-poor countries.

Zafar, Tasneem ; Aftab, Khalid

There can not be two opinions on the importance of Information and Communication Technology OCT) for economic development. However, real disparities exist in access to and use of ICT across countries. The digital divide is a complicated matter of varying levels of access, basic usage, and applications of ICT among countries and peoples. Using the Gompertz Technology Diffusion model, this paper attempts to measure the contribution of factors such as affordability, knowledge, infrastructure, human capital, trade openness, and oconomic and social environment in the technology diffusion process, specially in the case of information-poor countries.

JEL classification: 033, L96

Keywords: Digital Divide, Information and Communication Technologies, ICT, Gompertz Model, ICT Diffusion, Economic Development, ICT Infrastructure

I. INTRODUCTION

The need for access to Information and Communication Technology (ICT) for accelerated economic development has increased manifold in the information age. Not only are the new technologies considered a key to unlocking economic growth, they impinge on and can impact virtually all aspects of development. In this regard, a number of well-known declarations concerning developmental applications of Information and Communication Technology (ICT) rest on the experiences of high- or middle-income countries, and are simply assumed to be valid in other settings as well.

With its power to influence profoundly every sector of the economy, improved access to information and communications is central to improving the lives of people in the third world. And institutions in these countries, ranging from public bureaucracies and large enterprises, to small businesses and NGOs have the obvious need to improve their efficiency and effectiveness through access to modern means of communication i.e. computers, basic software and internet. All of this and much more, would be done if there were no constraints (or relatively malleable constraints) on governments, communities and individuals attempting to improve the quality of life in the developing world--just as it has been done in the advanced industrial world. However, there are extremely serious constraints on using ICT to improve the lot of most people in the Third World. These constraints are only partially technical and to a greater extent, they are economic, social and political. They flow not only from unresolved problems of poverty and economic inequality in particular countries and regions, but also from the structure and dynamics of the global economic system. Furthermore, whatever efforts are made to improve access to ICT in these countries, these take place within extremely varied cultures and social structures which shape the outcome of technological change in particular ways. Both the need for certain ICT products and their use may, thus, differ markedly from what might be expected in advanced industrial societies.

Thus far the gains of the digital revolution have been confined to a comparatively small group of countries, mainly in the industrialised world. The unequal distribution of the new and old the ICT across countries and associated efficiency gains go by the name of Digital Divide. It is the logical consequence of the social and economic imbalances that already exist within and across the countries. Although broadening of physical access to information and communication technologies is often a necessary step in reducing the digital divide, it is almost never sufficient to do so because the problem goes beyond the physical access and is related to real access. Physical access is determined by the availability of ICT related to infrastructure and its quality. But 'Real Access' is determined by: affordability; knowledge; IT training; its usage; human capital; sociopolitical conditions; and economic infrastructure available in a country.

Section I introduces the problem and gives the analytical framework used in the papers. Section II details the methodology. Section III describes the variables included in the study in the light of the literature review. This section also analyses the data sources. Section IV reports empirical findings of the study, and Section V summarises the findings of the study.

[FIGURE 1 OMITTED]

A glance at Figure 1 shows the presence of two sets of physical and real factors which influence countries access to ICT. We use this framework to place different information poor countries along Digital Access Index (DAI) ranking scale across countries. This is shown in Table 1.

Because of considerable differences in physical and real access across countries, it would be important to estimate the relative significance of various determinants of digital divide. This should help in identifying the factors that shape the environment in which modern ICT get diffused into the economies and what makes particular applications and services useful, especially in the case of information poor countries.

II. METHODOLOGY

Using Gompertz Technology Diffusion model, this study estimates factors that are responsible for the slow technology diffusion process in the information poor countries. This kind of model was used by [Stoneman (1983)] for modeling spread of computers. The specifications of the model are as follows:

[T.sub.it] is an indicator of Information and Communication Technology (ICT) in a country [sub.i] in year 't' and [T.sub.i.sup.*] be its post diffusion or equilibrium level or value ([T.sub.i.sup.*] or equilibrium level of ICT in country 'i' will be a function of exogenous demand side variables).

Most of the models of technology of technology adoption assume that over time [T.sub.it] tends to [T.sub.i.sup.*] along an S-shaped path i.e. this model assumes that spread between the value of the ICT indicator in year 't' and its value in year 't-1 'is a function of the spread between a target value (or post diffusion value) [T.sup.*] and value in year t-1.

ln [T.sub.it] - ln [T.sub.it-1] = [[alpha].sub.i](ln[T.sub.i.sup.*] - ln[T.sub.it-1]) ... (1)

Where [alpha.sub.i] is the speed of adjustment taken to be constant in our analysis.

Moreover we assume that most of the explanatory variables change over time. We may say that [T.sub.i.sup.*] is time dependent and express it as:

ln[T.sub.i.sup.*] = [[beta].sub.io] + [[beta].sub.il] ln [Y.sub.it] + [gamma]'[Z.sub.it] ... (2)

where post diffusion level of technology is a function of [Y.sub.it], i.e. the national income of the country 'i' in year 't' and [Z.sub.it], which is the vector of other possible variables describing the demand or supply conditions e.g. infrastructure, openness to international trade, economic freedom, knowledge or educational base of country ' i' in year 't'.

The estimable equation is obtained by inserting (2) in to (1).

ln[T.sub.it] - ln[T.sub.it-1] = [[alpha].sub.i][[beta].sub.o] + [[alpha].sub.i][[beta].sub.il] ln[Y.sub.it] + [[alpha].sub.i][gamma]'[Z.sub.it] - [[alpha].sub.i]ln[T.sub.it-1] + [epsilon] ... (3)

where [epsilon] is a white noise i.e. where the error terms are uncorrelated with zero mean and [[sigma].sup.2] variance.

III DESCRIPTION OF THE VARIABLES AND DATA SOURCES

The estimates of Gompertz Technology Diffusion model are reported in Section IV of this research paper for four ICT indicators i.e. cellular mobile subscribers per 100 inhabitants, personal computers per 100 inhabitants, internet hosts per 10,000 inhabitants, and internet users per 10,000 inhabitants. The data on these variables are for the period 1998-2003. The first three variables are taken as indicators of the state of the ICT infrastructure, so they will help to study the diffusion process of ICT infrastructure, while the fourth indicator, internet users, measures access to the internet. It is worth noting that the difference between communication technology and information technology has become blurred. For example, mobile phones are primarily tools of communication, but with the advent of wireless applications, consumers can access data and information via cellular phone. The internet is mainly an indicator of information technology, yet, many internet users communicate with other users from their personal computers. Thus, all three information indicators: internet hosts, internet users and personal computers have also become tools of communication.

The first explanatory variable in estimable equation is Gross National Income (GNI) per capita measured in international dollars. This variable is included to capture affordability. GNI is converted to international dollars using purchasing power parity rates. An international dollar has the same purchasing power over GNI as a U.S. dollar has in United States. Purchasing power parity (PPP) rates provide a standard measure allowing comparison of real price levels between countries, just as conventional price indices allow comparison of real values over time. Data for GNI is taken from world development indicators database, 2003. Historical data from developed nations indicate that adoption and diffusion of ICT is highly correlated with income. Countries with higher per capita income invest more in research and development and, hence, are more able to discover and use advanced information technologies. Prior to the spread of the internet, fixed telephones Hardy (1980) and telephone infrastructure Norton (1992) were used to model communication effects on economic growth. Since mid-1990s however other indicators of ICT began to be emphasised and more robust econometric tests are being employed. In general, the association between ICT and income is expected to be strong and positive.

The first variable in vector 'Z' is education. Low levels of education and literacy are expected to hinder both real accessibility and dissemination of ICT. Since the use of knowledge-based products requires a basic level of literacy, we would expect to see higher education causing higher ICT use and its consumption. Diffusion of ICT may require higher or tertiary education, and scientific research. Kiiski and Pohjola (2002) showed that, in a sample including developing and OECD countries, tertiary education had a positive and statistically significant influence on ICT diffusion. In contrast, Hargittai (1999), and Kiiski and Pohjola (2002) have found that in the case of industrial countries, education did not seem to influence ICT diffusion. These conflicting results suggest that this can be an important explanatory variable and need to be empirically tested. Also, in a sample that included both developed and developing countries, Norris (2000) shows that education did not have a significant influence on ICT diffusion. Consistent data on tertiary education are not available for all the countries in the sample. This study uses adult literacy and the education index instead. This index is also used by UNDP in generating the human development index (HDI).

This study uses three freedom indicators, which in fact represent the economic, social and political infrastructure in the economy that create an environment conducive for the spread of modern technologies i.e. ICT. The first indicator is the index of economic freedom published by the Heritage Foundation. This index is an average score of 10 indexes measured on a one-to-five scale, with 5 indicating the highest level of economic freedom. The 10 indexes assess trade policy, monetary policy, capital flows and foreign investment, wage and price control, banking and financial regulations, intellectual property and black markets, property rights, regulation, transparency and bureaucracy, government intervention in the economy, and the fiscal burden of the government (taxes and government expenditure). At least in cross-sectional analyses, greater (higher index) economic freedom is expected to be associated with higher GDP, higher levels of education or literacy rates, and stronger ICT indicators.

The other freedom indicators are the index of political rights and the index of civil liberties. By including these indices, we follow the work of Norris (2000) and try to explore whether countries with higher levels of civil and political freedom could also have greater ICT diffusion. These two indexes are measured on a one-to-seven scale, with 7 indicating the highest degree of freedom. The correlation between these indices and income is expected to be positive.

The other variables included in vector 'Z' are trade policy indicators. Openness to international trade is one of two trade policy indicators used in this study. It is measured as the ratio of the sum of exports and imports to GDP in world prices. The role of trade policy is important. For example, Jussawalla (1999) claims that East Asian nations fostered ICT production through openness and export-oriented investments. Both exports and imports may offer a channel for increased adoption and diffusion of ICT. Some imported goods and services require the existence of specific ICT to be operational. In some cases, ICT may be embodied in the imported products. Similarly, to enhance their exports, firms find it increasingly necessary to make use of ICT. Mobile phones, internet use, computerized operations are all tools used to improve the efficiency of conducting business in the global market. These tools tend to reduce the level of imperfect information and incomplete markets. As argued by Stiglitz (1989), imperfect information results in less trade. Thus, we would expect a positive and significant correlation between ICT and openness to international trade. The second international trade variable is foreign direct investment (FDI). Inward FDI usually allows recipient economies' access to advanced technologies, managerial skills and higher level of know-how. Transnational corporations tend to standardise their operations around the world and train workers in host countries according to their skill standards, including the use of ICT. Moreover, FDI may replace ICT as a medium for information and knowledge diffusion in cases where information and knowledge associated with ICT have a proprietary feature. As emphasised by Bedi (1999), '... in such cases, the role of ICT in enabling access is limited, and other measures such as trade and foreign direct investment may be appropriate conduits for disseminating information and knowledge'. Thus, it is reasonable to expect higher inward FDI to contribute to ICT diffusion.

Other variables which have been emphasised in the literature as potential determinants of ICT diffusion include knowledge of English language Kiiski and Pohjola (2002), income distribution [Bedi (1999); Hargittai (1999) and Pohjola (2000)], and competition in the telecommunication industry [Hargittai (1999); Jayakar (1999); and Kiiski and Pohjola (2002)]. The empirical evidence on the impact of these variables, particularly in developing countries, is ambiguous or more in support of their insignificance. So, they are not included as explanatory variables.

IV. EMPIRICAL RESULTS

The results from the linear estimation of Gompertz Technology Diffusion model exploring the factors that influence ICT diffusion are reported in Tables 2-5. To test the robustness of the model, four equations were estimated. As mentioned earlier, the use of ICT in an economy can be seen through four indicators i.e. internet users, internet hosts, number of personal computers and mobile phone subscribers. Table 2 displays the statistical results from estimating the model with internet use as the relevant ICT variable. Table 3 reports the findings when personal computers were the relevant ICT indicator. Tables 4 and 5 report the results associated with internet hosts, and mobile phones, respectively.

Equations /columns (1) and (2) in each table differ in terms of right hand side variables because in each table (except internet hosts) the first equation reports the findings about those right hand side variables selected as a result of stepwise model selection procedure from all entered variables. Here one thing is worth mentioning that the selected model in above mentioned three cases as a result of stepwise selection procedure is also consistent with the model selected through forward selection procedure i.e. the both methods select the same explanatory variables. Equation (2) provides the estimates of the model including those explanatory variables selected as a result of backward model selection procedure. Moreover all the four estimated equations satisfy the basic assumptions of linear models as all have been duly checked (checking includes co-linearity diagnostic through VIF(variance inflate factor), and autocorrelation through Durbin-Watson test.

The underlying assumption here is that the diffusion process is the same in all countries i.e. the parameter values of Gompertz Technology Diffusion model take the same value for all 'i' or countries, moreover the speed of diffusion is assumed to be constant over time. However it would be more appropriate to make it time dependent as suggested by Kiiski and Pohjola (2002), but in order to make the analysis simple it is assumed so.

Table 2 displays the statistical results of internet use as the relevant ICT indicator. The empirical results indicate that in case of internet users both of the equations show that model adequately captures the diffusion process since the speed of diffusion or adjustment (coefficient on the lagged value of this variable) is highly significant in both cases. Speed of diffusion is 0.718 in case of Equation (1) and 0.780 in case of Equation (2) and in both cases significant at 99 percent confidence level. Moreover income, education, openness to international trade, stock of personal computers and internet access cost turn out to be highly significant i.e. at 99 percent confidence level in both equations. However civil liberties and economic freedom come up with correct signs, but are not as significant explanatory variables. If we make a comparison of two selected models, the model selected through backward selection procedure (Equation 2) is a little better than one selected through stepwise procedure (Equation 1) as it slightly improves the value of adjusted R-squared i.e. from 0.75 in Equation (1) to 0.77 in Equation (2).

Table 3 displays the results of the model where ICT is represented by the number of personal computers per 100 inhabitants. Again in this case the speed of diffusion ([alpha]) is highly significant in both the selected models as it is 0.180 in case of Equation (1), and 0.804 in the case of Equation (2) and is significant at 99 percent confidence level. In the case of internet users income, adult literacy, economic freedom, internet users turn out to be highly significant at 99 percent confidence level. Moreover out of the two models selected through two different selection procedures, Equation (1) is more appropriate i.e. selected through stepwise and forward selection procedures as it gives slightly improved value of R-square (i.e. 0.612) as compared to 0.6 in the case of Equation (1).

Table 4 displays the results when internet hosts was the ICT indicator in an economy or dependent variable. Although the speed of diffusion is significant at 99 percent confidence level in the selected model but the value of R-square is very low. However the results suggest that income and economic freedom are other important significant explanatory variables that too are significant at 99 percent confidence level. Moreover model fails to provide support for the influence of education or literacy on internet host diffusion.

Finally, Table 5 reports the findings when mobile phone is an indicator of ICT. Again in this case speed of diffusion adjustment is significant at 99 percent level of confidence, but model captures weakly the diffusion process because here again the value of adjusted R-square is low. Moreover, this is the only ICT indicator where income does not come out to be significant. The only significant variable is openness to international trade which is significant at 95 percent level of confidence.

In summary, the empirical results provide support for the role of income as a major determinant of ICT diffusion because it comes out to be significant at 99 percent confidence level in the case of internet use, internet hosts and personal computers. This is consistent with the conclusions in Niininen (2001), Hargittai (1999), Quah (2001), Norris (2000), and Kiiski and Pohjola (2002). Thus showing that adoption and diffusion of modern information and communication technology is highly correlated with income level. Countries with higher per capita income invest more in research and development therefore are able to acquire and use advanced information technologies.

In addition, education and literacy, especially adult literacy, appears to have direct impact on dissemination and personal computers, thus showing that education influences technology adoption. However we do not find evidence that education is a significant explanatory variable of mobile phone use. This may be due to the reason that use of mobile phones does not need as much educational or training skills as it is required in case of computer or internet use. Undoubtedly education must have a role in diffusion of information and communication technologies for at least two reasons. Firstly, education directly contributes to basic literacy and reading and writing skills which are essential in use of modern ICT as knowledge-based products. More educated people are likely to be quicker to adopt new innovations than people with less education. Secondly, based on the facts that the early users of the internet were people working in higher education and research academic institutions may play an important role in spreading of ICT. However, our findings that education is important in technology dissemination is consistent with the earlier findings of Barrow and Lee (2000) and Duncombe (2000), Caselli and Coleman (2001) and Wong (2001) and is in sharp contrast with the findings of Hargittai (1999) and Norris(2000), as they concluded that education is not important in technology dissemination.

It is surprising to find that there is no support for the influence of FDI on ICT diffusion. As mentioned earlier, FDI is an important channel through which technology enters a country and gets disseminated. Perhaps, there is a threshold that most developing countries in the sample have not yet reached or that FDI in the countries under study targets labour-intensive sectors that require negligible levels of ICT. In fact, since FDI is accounted for in the index of economic freedom, the findings do not necessarily imply that this variable has no impact on ICT diffusion.

Moreover, the estimation yields values for the speed of diffusion adjustment ([alpha]) that are consistent with the increased adoption of ICT. In a cross-sectional model including 75 developed and developing countries, Kiiski and Pohjola (2002) report values for the speed of diffusion that range from 0.186 to 0.527. However the empirical results in this research paper find that the speed of diffusion can vary from 0.400 to 0.455 for mobile phones, from 0.804 to 0.810 for personal computers and from 0.718 to 0.780 for internet use. However, given that the RHS variables are not the same, it is difficult to make a more meaningful comparison of the results derived in the two studies.

V. SUMMARY

This study has six important findings. First, income is a major determinant of ICT diffusion. Income influences both ICT infrastructure as it is shown to cause higher internet use, use of personal computers and internet hosts and access to ICT since it has an effect on internet use. Second, there is a positive impact of government trade policies on ICT. Openness fosters the adopting and adapting of technology. Third, at least in the case of two ICT indicators (mobile phones and internet hosts) political rights and civil liberties have a strong influence. Fourth, there is evidence supporting that education (literacy) has a positive impact on ICT diffusion. Moreover, the above conclusions highlight the role of demand in the market for knowledge-based products, and are consistent with the propositions in Quah (2001). It is important to note that for mobile phones openness to international trade and for internet hosts, economic freedom are important factors, while GNI does not seem to have an effect on mobile phones.

In addition, the speed of diffusion in the case of Internet users and Personal computers is shown to be much higher than in the case of internet hosts and mobile phones. This finding may reflect the recent trend in large cities where cyber cafes are mushrooming. However, it is feared that a faster diffusion of internet users (relative to internet hosts) may lead to saturation and poor access to information. The present findings seem to provide elements for hope and concern at the same time. On the one hand, there is evidence through earlier researches that ICT enhances income, and hence, it can provide an additional source of economic growth. Due to its pervasive nature, ICT diffusion may allow a leapfrogging process to occur. On the other hand, the finding that trade policies and social development variables are important determinants of ICT diffusion, as well as economic development, implies that countries with poor performance in these variables may sink even further in the information-poor and non-communicating side of the digital divide.

Notes: Information-poor countries considered in this research paper for the purpose of analysis are low-access economies, i.e., the countries have a score value of less than 0.37 according to the Digital Access Index (DAI) of ITU, 2002-03. A complete list of these countries along with their scores can be seen in Table I. Countries in this category are the poorest in the world and most are LDCs. They have a minimal level of access to the information society. The Digital Access Index (DAI) measures the overall ability of individuals in a country to access and use information and communication technologies. The DAI combines eight variables, covering five areas, to provide an overall country score. The results of the Index point to potential stumbling blocks in ICT adoption.

APPENDIX
Mathematical Derivation of Results

Details of Model Selection Procedure When Internet Users Are the
Relevant n ICT Indicator

 Variables Entered/Removed (a)

 Variables Variables
Model Entered Removed Method

1 LN-USERS-98 * Stepwise (Criteria:
 Probability-of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

2 LN-Avg GNI * Stepwise (Criteria:
 Probability-of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

3 ln-OPEN * Stepwise (Criteria:
 Probability-of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

4 ln_Internet * Stepwise (Criteria:
 tariff Probability-of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

(a) Dependent Variable: USERs (LN02-98).

Model Summary

 Adjusted Std. Error of
Model R R Square R Square the Estimate

1 .525 (a) .276 .258 .89848688218940
2 .818 (b) .669 .652 .61554392824644
3 .853 (c) .728 .707 .56438514885354
4 .879 (d) .772 .748 .52380532999675

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-Avg GNf.

(c) Predictors: (Constant), LN-USERS-98, LN-Avg GNf, In-OPEN.

(d) Predictors: (Constant), LN-USERS-98, LN-Avg GN1, In-OPEN,
In Internet tariff.

 ANOVA (e)

 Sum of Mean
 Model Squares df Square F Sig.

1 Regression 12.626 1 12.626 15.640 .000 (a)
 Residual 33.098 41 .807
 Total 45.724 42

2 Regression 30.569 2 15.284 40.339 .000 (b)
 Residual 15.156 40 .379
 Total 45.724 42

3 Regression 33.302 3 11.101 34.849 .000 (c)
 Residual 12.423 39 .319
 Total 45.724 42

4 Regression 35.298 4 8.825 32.163 .000 (d)
 Residual 10.426 38 .274
 Total 45.724 42

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-Avg GNI.

(c) Predictors: (Constant), LN-USERS-98, LN-Avg GNI, ln-OPEN.

(d) Predictors: (Constant), LN-USERS-98, LN-Avg GNI, ln-OPEN,
ln_Internet tariff.

(e) Dependent Variable: USERS (LN02-98).

Coefficients (a)

 Unstandardised
 Coefficients
 Standardised
 Std. Coefficients
Model B Error Beta t Sig.

1 (Constant) 1.708 .351 4.869 .000
 LN-USERS-98 -.411 .104 -.525 -3.955 .000

2 (Constant) -6.780 1.257 -5.396 .000
 LN-USERS-98 -.671 .081 -.858 -8.323 .000
 LN-Avg GNI 1.022 .149 .709 6.882 .000

3 (Constant) -8.138 1.242 -6.553 .000
 LN-USERS-98 -.679 .074 -.868 -9.186 .000
 LN-Avg GNI .963 .138 .668 6.997 .000
 ln-OPEN .541 .185 .249 2.929 .006

4 (Constant) -3.918 1.943 -2.016 .051
 LN-USERS-98 -.718 .070 -.918 -10.241 .000
 LN-Avg GNI .549 .200 .381 2.745 .009
 ln-OPEN .567 .172 .261 3.307 .002
 ln_internet tariff .295 .109 -.376 -2.698 .010

(a) Dependent Variable: USERs (LN02-98).

Excluded Variables (e)

Model Beta ln t Sig.

1 LN-PCs-98 .533 (a) 4.388 .000
 LN-Avg GNI .709 (a) 6.882 .000
 CL-c -.196 (a) -1.494 .143
 ECF-c -.079 (a) -.589 .559
 ln-fdi .378 (a) 2.996 .005
 ln-OPEN .336 (a) 2.693 .100
 ln_internet tariff -.694 (a) 6.423 .000
 Education index .520 (a) 4.212 .000

2 LN-PCs-98 .274 (b) 2.592 .013
 CL-c .048 (b) .487 .629
 ECF-c .011 (b) .117 .908
 ln-fdi .120 (b) 1.147 .259
 ln-OPEN .249 (b) 2.929 .006
 ln_internet tariff -.349 (b) -2.240 .031
 Education index .265 (b) 2.520 .016

3 LN-PCs-98 .125 (c) 2.242 .031
 CL-c .037 (c) .410 .684
 ECF-c .039 (c) .451 .655
 ln-fdi .124 (c) 1.299 .202
 ln_internet tariff -.376 (c) -2.698 .010
 Education index .177 (c) 1.632 .111

4 LN-PCs-98 .160 (d) 1.588 .121
 CL-c .030 (d) .357 .723
 ECF-c .075 (d) .931 .358
 ln-fdi .065 (d) .693 .493
 Education index .148 (d) 1.448 .156

 Collinearity
 Partial Statistics
Model Correlation Tolerance

1 LN-PCs-98 .570 .827
 LN-Avg GNI .736 .781
 CL-c -.230 1.000
 ECF-c -.093 .991
 ln-fdi .428 .928
 ln-OPEN .392 .985
 ln_internet tariff -.713 .762
 Education index .554 .823

2 LN-PCs-98 .383 .651
 CL-c .078 .857
 ECF-c .019 .970
 ln-fdi .181 .755
 ln-OPEN .425 .964
 ln_internet tariff -.338 .311
 Education index .374 .660

3 LN-PCs-98 .342 .628
 CL-c .066 .855
 ECF-c .073 .959
 ln-fdi .206 .755
 ln_internet tariff -.401 .310
 Education index .256 .567

4 LN-PCs-98 .253 .571
 CL-c .059 .854
 ECF-c .151 .934
 ln-fdi .113 .702
 Education index .232 .560

(a) Predictors in the Model: (Constant), LN-USERS-98.

(b) Predictors in the Model: (Constant), LN-USERS-98, LN-Avg GNI.

(c) Predictors in the Model: (Constant), LN-USERS-98, LN-Avg GNI,
ln-OPEN.

(d) Predictors in the Model: (Constant), LN-USERS-98, LN-Avg GNI,
ln-OPEN, ln_internet tariff.

(e) Dependent Variable: USERS (LN02-98).

Variables Entered/Removed (b)

 Variables
Model Variables Entered Removed Method

1 Education index, ECF-c, Enter
 CL-c, In-fdi, ln-OPEN,
 LN-PCs-98, LN-USERS-98,
 LN-Avg GNI, In Internet
 tariff (a)

2 ln-fdi Backward (criterion:
 Probability of
 F-to-remove >= .100).

3 CL-c Backward (criterion:
 Probability of
 F-to-remove >= .100).

4 ECF-c Backward (criterion:
 Probability of
 F-to-remove >= .100).

5 Education Backward (criterion:
 Index Probability of
 F-to-remove >= .100).

6 LN-PCs-98 Backward (criterion:
 Probability of
 F-to-remove >= .100).

(a) All requested variables entered.

(b) Dependent Variable: USERS (LN02-98).

Model Summary

 Adjusted Std. Error of
Model R R-Square R Square the Estimate

1 .904 (a) .816 .766 .50435849610735
2 .902 (b) .813 .769 .50178824801358
3 .898 (c) .806 .767 .50357515882357
4 .894 (d) .798 .765 .50591094508067
5 .887 (e) .787 .758 .51361817779732
6 .879 (f) .772 .748 .52380532999675

(a) Predictors: (Constant), Education index, ECF-c, CL-c, ln-fdi,
ln-OPEN, LN-PCs-98, LN-USERS-98, LN-Avg GNI, ln_internet tariff.

(b) Predictors: (Constant), Education index, ECF-c, CL-c, ln-OPEN,
LN-PCs-98, LN-USERS-98, LN-Avg GNI, In_internet tariff.

(c) Predictors: (Constant), Education index, ECF-c, ln-OPEN,
LN-PCs-98, LN-USERS-98, LN-Avg GNI, In_internet tariff.

(d) Predictors: (Constant), Education index, ln-OPEN, LN-PCs-98,
LN-USERS-98, LN-Avg GNI, ln_internet tariff.

(e) Predictors: (Constant), ln-OPEN, LN-PCs-98, LN-USERS-98,
LN-Avg GNI, ln_internet tariff.

(f) Predictors: (Constant), ln-OPEN, LN-USERS-98, LN-Avg GNI,
ln_internet tariff.

ANOVA (g)

 Sum of Mean
Model Squares df Square F Sig.

1 Regression 37.330 9 4.148 16.306 .000 (a)
 Residual 8.394 33 .254
 Total 45.724 42

2 Regression 37.163 8 4.645 18.450 .000 (b)
 Residual 8.561 34 .252
 Total 45.724 42

3 Regression 36.849 7 5.264 20.759 .000 (c)
 Residual 8.876 35 .254
 Total 45.724 42

4 Regression 36.510 6 6.085 23.775 .000 (d)
 Residual 9.214 36 .256
 Total 45.724 42

5 Regression 35.964 5 7.193 27.265 .000 (e)
 Residual 9.761 37 .264
 Total 45.724 42

6 Regression 35.298 4 8.825 32.163 .000 (f)
 Residual 10.426 38 .274
 Total 45.724 42

(a) Predictors: (Constant), Education index, ECF-c, CL-c, In-fdi,
ln-OPEN, LN-PCs-98, LN-USERS-98, LN-Avg GNI, In intemet tariff.

(b) Predictors: (Constant), Education index, ECF-c, CL-c, In-OPEN,
LN-PCs-98, LN-USERS-98, LN-Avg GNI, In internet tariff.

(c) Predictors: (Constant), Education index, ECF-c, In-OPEN,
LN-PCs-98, LN-USERS-9R, LN-Avg GNI, ln_internet tariff.

(d) Predictors: (Constant), Education index, In-OPEN, LN-PCs-98,
LN-USERS-98. LN-Avg GNI, ln_internet tariff.

(e) Predictors: (Constant), In-OPEN, LN-PCs-98, LN-USERS-98,
LN-Avg GNI, ln_internet tariff.

(f) Predictors: (Constant), In-OPEN, LN-USERS-98, LN-Avg GNI,
ln_internet tariff.

(g) Dependent Variable: USERS (LN02-98).

Coefficients (a)

 Unstandardised
 Coefficients

 Std.
 Model B Error

1 (Constant) -4.366 1.961
 LN-PCs-98 .183 .101
 LN-Avg GNI .472 .204
 CL-c 5.910E-02 .057
 ECF-c 7.594E-02 .071
 ln-fdi 4.281E-02 .053
 ln-OPEN .383 .186
 ln_internet tariff -.199 .119
 LN-USERS-98 -.779 .072
 Education index 1.034 .581

2 (Constant) -4.267 1.947
 LN-PCs-98 .163 .097
 LN-Avg GNI .491 .202
 CL-c 6.347E-02 .057
 ECF-c 8.149E-02 .071
 In-OPEN .383 .185
 ln_intemet tariff -.230 .113
 LN-USERS-98 -.780 .072
 Education index 1.097 .573

3 (Constant) -3.680 1.881
 LN-PCs-98 .144 .096
 LN-Avg GNI .443 .198
 ECF-c 8.180E-02 .071
 ln-OPEN .421 .182
 ln internet tariff -.244 .112
 LN-USERS-98 -.758 .070
 Education index .925 .553

4 (Constant) -3.531 1.886
 LN-PCs-98 .154 .096
 LN-Avg GNI .460 .198
 ln-OPEN .410 .183
 ln_intemet tariff -.224 .111
 LN-USERS-98 -.754 .070
 Education index .796 .544

5 (Constant) -3.670 1.912
 LN-PCs-98 .155 .098
 LN-Avg GNI .521 .197
 ln-OPEN .509 .172
 ln intemet tariff -.241 .112
 LN-USERS-98 -.735 .070

6 (Constant) -3.918 1.943
 LN-Avg GNI .549 .200
 ln-OPEN .567 .172
 ln Internet tariff -.295 .109
 LN-USERS-98 -.718 .070

 Standardised
 Coefficients
 Model Beta t Sig.

1 (Constant) -2.227 .033
 LN-PCs-98 .188 1.811 .079
 LN-Avg GNI .327 2.309 .027
 CL-c .088 1.031 .310
 ECF-c .085 1.066 .294
 ln-fdi .075 .809 .424
 ln-OPEN .176 2.063 .047
 ln_internet tariff -.254 -1.668 .105
 LN-USERS-98 -.995 -10.755 .000
 Education index .190 1.781 .084

2 (Constant) -2.192 .035
 LN-PCs-98 .168 1.676 .103
 LN-Avg GNI .341 2.433 .020
 CL-c .095 1.118 .271
 ECF-c .091 1.155 .256
 In-OPEN .177 2.076 .046
 ln_intemet tariff -.293 -2.042 .049
 LN-USERS-98 -.997 -10.828 .000
 Education index .202 1.917 .064

3 (Constant) -1.956 .058
 LN-PCs-98 .148 1.495 .144
 LN-Avg GNI .307 2.238 .032
 ECF-c .091 1.155 .256
 ln-OPEN .194 2.311 .027
 ln internet tariff -.312 -2.178 .036
 LN-USERS-98 -.969 -10.898 .000
 Education index .170 1.671 .104

4 (Constant) -1.873 .069
 LN-PCs-98 .158 1.597 .119
 LN-Avg GNI .319 2.323 .026
 ln-OPEN .189 2.242 .031
 ln_intemet tariff -.285 -2.010 .052
 LN-USERS-98 -.964 -10.806 .000
 Education index .146 1.461 .153

5 (Constant) -1.919 .063
 LN-PCs-98 .160 1.588 .121
 LN-Avg GNI .361 2.648 .012
 ln-OPEN .235 2.958 .005
 ln intemet tariff -.307 -2.142 .039
 LN-USERS-98 -.939 -10.564 .000

6 (Constant) -2.016 .051
 LN-Avg GNI .381 2.745 .009
 ln-OPEN .261 3.307 .002
 ln Internet tariff -.376 -2.698 .010
 LN-USERS-98 -.918 -10.241 .000

(a) Dependent Variable: USERS (LN02-98).

Excluded Variables (f)

 Collinearity
 Partial Statistics
Model Beta In t Sig. Correlation Tolerance

2 In-fdi .075 (a) .809 .424 .139 .643

3 In-fdi .084 (b) .909 .370 .154 .648
 CL-c .095 (b) 1.118 .271 .188 .765

4 In-fdi .094 (c) 1.014 .317 .169 .655
 CL-c .095 (c) 1.117 .272 .186 .765
 ECF-c .091 (c) 1.155 .256 .192 .890

5 In-fdi .106 (d) 1.136 .264 .186 .661
 CL-c .055 (d) .652 .518 .108 .828
 ECF-c .065 (d) .815 .420 .135 .927
 Education .146 (d) 1.461 .153 .237 .560
 index

6 In-fdi .065 (e) .693 .493 .113 .702
 CL-c .030 (e) .357 .723 .059 .854
 ECF-c .075 (e) .931 .358 .151 .934
 Education .148 (e) 1.448 .156 .232 .560
 index
 LN-PCs-98 .160 (e) 1.588 .121 .253 .571

(a) Predictors in the Model: (Constant), Education index, ECF-c, CL-
c, ln-OPEN, LN-PCs-98, LN-USERS-98, LN-Avg GNI, ln_Internet tariff,

(b) Predictors in the Model: (Constant), Education index, ECF-c, ln-
OPEN, LN-PCs-98, LN-USERS-98, LN-Avg GNI, ln_Internet tariff.

(c) Predictors in the Model: (Constant), Education index, ln-OPEN, LN-
PCs-98, LN-USERS-98, LN-Avg GNI, ln_internet tariff.

(d) Predictors in the Model: (Constant), ln-OPEN, LN-PCs-98, LN-
USERS-98, LN-Avg GN1, ln_Internet tariff.

(e) Predictors in the Model: (Constant), In-OPEN, LN-USERS-98, LN-Avg
GN1, In Internet tariff.

(f) Dependent Variable: USERS (LN02-98).

* Details of Model Selection Procedure When Personal Computers Are the
Relevant n ICT Indicator

Variables Entered / Removed (b)

 Variables
Model Variables Entered Removed

1 LN-USERS-98, CL-c, In-OPEN, ECF-c,
 to-fdi, LN-PCs-98, Adult literacy, LN-
 Avg GNI, In intemet tariff(a)

2 * ln-OPEN

3 * CL-c

4 * ln intemet
 tariff

 * ln-fdi

Model Method

1 * Enter

2 Backward (criterion: Probability of
 F-to-remove >= .100).

3 Backward (criterion: Probability of
 F-to-remove >= .100).

4 Backward (criterion: Probability of
 F-to-remove >= .100).

 Backward (criterion: Probability of
 F-to-remove >= .100).

(a) All requested variables entered.

(b) Dependent Variable: PCs (LN-02-LN98).

Model Summary

 R Adjusted Std. Error of
Model R Square R Square the Estimate

1 .816 (a) .666 .573 .6663679314691

2 .816 (b) .666 .584 .6571244352135

3 .8i5 (c) .664 .595 .6488837496513

4 .814 (d) .662 .604 .6412910798225

5 .812 (e) .659 .612 .6349569338436

(a) Predictors: (Constant), LN-USERS-98, CL-c, ln-OPEN, ECF-c, In-
fdi, LN-PCs-98, Adult literacy, LN-Avg GN1, ln_internet tariff.

(b) Predictors: (Constant), LN-USERS-98, CL-c, ECF-c, ln-fdi, LN-PCs-
98, Adult literacy, LN-Avg GNI, ln_intemet tariff.

(c) Predictors: (Constant), LN-USERS-98, ECF-c, ln-fdi, LN-PCs-98,
Adult literacy, LN-Avg GN1, In Internet tariff.

(d) Predictors: (Constant), LN-USERS-98, ECF-c, ln-fdi, LN-PCs-98,
Adult literacy, LN-Avg GNI .

(e) Predictors: (Constant), LN-USERS-98, ECF-c, LN-PCs-98, Adult
literacy, LN-Avg GN1.

ANOVA (f)

 Sum of Mean
Model Squares df Square F Sig.

1 Regression 28.396 9 3.155 7.105 .000 (a)
 Residual 14.209 32 .444
 Total 42.605 41

2 Regression 28.355 8 3.544 8.208 .000 (b)
 Residual 14.250 33 .432
 Total 42.605 41

3 Regression 28.290 7 4.041 9.598 .000 (c)
 Residual 14.316 34 .421
 Total 42.605 41

4 Regression 28.211 6 4.702 11.433 .000 (d)
 Residual 14.394 35 .411
 Total 42.605 41

5 Regression 28.091 5 5.618 13.935 .000 (e)
 Residual 14.514 36 .403
 Total 42.605 41

(a) Predictors: (Constant), LN-USERS-98, CL-c, ln-OPEN, ECF-c, ln-
fdi, LN-PCs-98, Adult literacy, LN-Avg GNI, ln internet tariff.

(b) Predictors: (Constant), LN-USERS-98, CL-c, ECF-c, ln-fdi, LN-PCs-
98, Adult literacy, LN-Avg GN1, ln_internet tariff.

(c) Predictors: (Constant), LN-USERS-98, ECF-c, ln-fdi, LN-PCs-98,
Adult literacy, LN-Avg GNI, ln_Internet tariff.

(d) Predictors: (Constant), LN-USERS-98, ECF-c, ln-fdi, LN-PCs-98,
Adult literacy, LN-Avg GNI.

(e) Predictors: (Constant), LN-USERS-98, ECF-c, LN-PCs-98, Adult
literacy, LN-Avg GN1.

(f) Dependent Variable: PCs (LN-02-LN98).

Coefficients (a)

 Unstandardised
 Coefficients
 Std.
 Model B Error

1 (Constant) -4.567 2.621
 LN-PCs-98 -.842 .135
 LN-Avg GNI .554 .273
 CL-c 2.519E-02 .076
 ECF-c .166 .094
 ln-fdi -4.733E-02 .071
 ln-OPEN 7.357E-02 .244
 Adult Literacy 1.528E-02 .006
 ln_Internet tariff -7.059E-02 .160
 LN-USERS-98 .324 .096

2 (Constant) -4.442 2.553
 LN-PCs-98 -.834 .130
 LN-Avg GNI .560 .269
 CL-c 2887E-02 .074
 ECF-c .164 .093
 ln-fdi -4.734E-02 .070
 Adult Literacy 1.592E-02 .006
 ln_Internet tariff -6.409E-02 .156
 LN-USERS-98 .322 .094

3 (Constant) -4.161 2.418
 LN-PCs-98 -.840 .128
 LN-Avg GNI .537 .259
 ECF-c .164 .092
 ln-fdi -4.466E-02 .069
 Adult Literacy 1.546E-02 .006
 ln_Internet tariff -6.630E-02 .154
 LN-USERS-98 .331 .090

4 (Constant) -4.939 1.587
 LN-PCs-98 -.820 .118
 LN-Avg GNI .605 .203
 ECF-c .157 .089
 ln-fdi -3.462E-02 .064
 Adult literacy -1.522E-02 .005
 LN-USERS-98 .337 .088

5 (Constant) -4.734 1.526
 LN-PCs-98 -.810 .115
 LN-Avg GNI .561 .184
 ECF-c .155 .088
 Adult literacy 1.483E-02 .005
 LN-USERS-98 .334 .087

 Standardised
 Coefficients
 Model Beta t Sig.

1 (Constant) -1.742 .091
 LN-PCs-98 -.887 -6.244 .000
 LN-Avg GNI .393 2.027 .051
 CL-c .039 .332 .742
 ECF-c .190 1.762 .088
 ln-fdi -.085 -.666 .510
 ln-OPEN .035 .301 .765
 Adult Literacy .327 2.451 .020
 ln_Internet tariff -.092 -.442 .661
 LN-USERS-98 .424 3.394 .002

2 (Constant) -1.740 .091
 LN-PCs-98 -.878 -6.411 .000
 LN-Avg GNI .398 2.084 .045
 CL-c .044 .391 .699
 ECF-c .187 1.768 .086
 ln-fdi -.085 -.676 .504
 Adult Literacy .340 2.754 .009
 ln_Internet tariff -.084 -.411 .684
 LN-USERS-98 .422 3.429 .002

3 (Constant) -1.721 .094
 LN-PCs-98 -.884 -6.581 .000
 LN-Avg GNI .381 2.074 .046
 ECF-c .187 1.791 .082
 ln-fdi -.080 -.649 .521
 Adult Literacy .331 2.767 .009
 ln_Internet tariff -.086 -.431 .669
 LN-USERS-98 .434 3.685 .001

4 (Constant) -3.112 .004
 LN-PCs-98 -.863 -6.958 .000
 LN-Avg GNI .429 2.978 .005
 ECF-c .179 1.761 .087
 ln-fdi -.062 -.541 .592
 Adult literacy .325 2.770 .009
 LN-USERS-98 .441 3.830 .001

5 (Constant) -3.103 .004
 LN-PCs-98 -.853 -7.029 .000
 LN-Avg GNI .398 3.045 .004
 ECF-c .177 1.762 .087
 Adult literacy .317 2.750 .009
 LN-USERS-98 .437 3.843 .000

(a) Dependent Variable: PCs (LN-02-LN98).

Excluded Variables (e)
 Collinearity
 Partial Statistics
Model Beta ln T Sig. Correlation Tolerance

2 ln-OPEN .035 (a) .301 .765 .053 .787

3 ln-OPEN .041 (b) .364 .718 .063 .808
 CL-c .044 (b) .391 .699 .068 .793

4 ln-OPEN .034 (c) .311 .758 .053 .821
 CL-c .046 (c) .411 .684 .070 .794
 ln_internet -.086 (c) -.431 .669 -.074 .245
 tariff

5 ln-OPEN .036 (d) .330 .744 .056 .822
 CL-c .038 (d) .349 .729 .059 .805
 ln Internet -.042 (d) -.226 .822 -.038 .277
 tariff
 ln-fdi -.062 (d) -.541 .592 -.091 .727

(a) Predictors in the Model: (Constant), LN-USERS-98, CL-c, ECF-c,
ln-fdi, LN-PCs-98, Adult literacy, LN-Avg GNI, ln_Internet tariff.

(b) Predictors in the Model: (Constant), LN-USERS-98, ECF-c, ln-fdi,
LN-PCs-98, Adult literacy, LN-Avg GNI, ln_internet tariff.

(c) Predictors in the Model: (Constant), LN-USERS-98, ECF-c, ln-fdi,
LN-PCs-98, Adult literacy, LN-Avg GNI .

(d) Predictors in the Model: (Constant), LN-USERS-98, ECF-c, LN-PCs-
98, Adult literacy, LN-Avg GNI .

(e) Dependent Variable: PCs (LN-02-LN98).

Regression

Variables Entered / Removed (a)

 Variables Variables
Model Entered Removed Method

l LN-USERS-98 * Stepwise (Criteria: Probability-
 of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

2 LN-PCs-98 * Stepwise (Criteria: Probability-
 of-F-to-enter <= ,050,
 Probability-of-F-to-remove >=
 .100).

3 LN-Avg GNI * Stepwise (Criteria: Probability-
 of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

4 Adult Literacy * Stepwise (Criteria: Probability-
 of-F-to-enter <= .050,
 Probability-of-F-to-remove >=
 .100).

(a) Dependent Variable: PCs(LN-02-LN98).

Model Summary

 R Adjusted Std. Error of
Model R Square R Square the Estimate

1 .365 (a) 0.133 0.112 .9607157410174
2 .656 (b) 0.431 0.401 .7886787234721
3 .756 (c) 0.572 0.538 .6926100879668
4 .794 (d) 0.630 0.590 .6527580184499

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GNI.

(d) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GNI, Adult
literacy.

ANOVA (e)

Model Sum of Squares df Mean Square F Sig.

1 Regression 5.686 1 5.686 6.161 .017(a)
 Residual 36.919 40 0.923
 Total 42.605 41

2 Regression 18.347 2 9.173 14.748 .000(b)
 Residual 24.259 39 0.622
 Total 42.605 41

3 Regression 24.376 3 8.125 16.938 .000(c)
 Residual 18.229 38 0.480
 Total 42.605 41

4 Regression 26.840 4 6.710 15.748 .000(d)
 Residual 15.765 37 0.426
 Total 42.605 41

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GNI.

(d) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GNI,
Adult literacy.

(e) Dependent Variable: PCs (LN-02-LN98).

Coefficients (a)

 Unstandardised
 Coefficients Standardised
 Coefficients
Model B Std. Error Beta

1 (Constant) 1.489 .379
 LN-USERS-98 .279 .112 .365

2 (Constant) 1.565 .312
 LN-USERS-98 .470 .102 .615
 LN-PCs-98 -.569 .126 -.599

3 (Constant) -4.021 1.599
 LN-USERS-98 .366 .094 .479
 LN-PCs-9R -.774 .125 -.815
 LN-Avg GNI .676 .191 .480

4 (Constant) -3.991 1.507
 LN-USERS-98 .335 .089 .438
 LN-PCs-98 -.804 .118 -.846
 LN-Avg GNI .537 .189 .381
 Adult literacy 1.311E-02 .005 .280

Model t Sig.

1 (Constant) 3.929 .000
 LN-USERS-98 2.482 .017

2 (Constant) 5.023 .000
 LN-USERS-98 4.626 .000
 LN-PCs-98 -4.512 .000

3 (Constant) -2.514 .016
 LN-USERS-98 3.906 .000
 LN-PCs-9R -6.194 .000
 LN-Avg GNI 3.545 .001

4 (Constant) -2.647 .012
 LN-USERS-98 3.748 .001
 LN-PCs-98 -6.789 .000
 LN-Avg GNI 2.846 .007
 Adult literacy 2.404 .021

(a) Dependent Variable: PCs (LN-02-LN98).

Excluded Variables (e)

Model Beta In t Sig.

l LN-PCs-98 -.599(a) -4.512 .000
 LN-Avg GNI .093(a) .552 .584
 CL-c .045(a) .304 .763
 ECF-c .105(a) .708 .483
 In-fdi .109(a) .711 .481
 In-OPEN -.010(a) -.067 .947
 Adult literacy .202(a) 1.290 .205
 In Internet tariff .073(a) .430 .670

2 LN-Avg GNI .480(b) 3.545 .001
 CL-c -.127(b) -1.003 .322
 ECF-c .075(b) .609 .546
 In-fdi .158(b) 1.267 .213
 In-OPEN .124(b) .987 .330
 Adult literacy .382(b) 3.158 .003
 In Internet tariff -.341(b) -2.211 .033

3 CL-c -.016(c) -.133 .895
 ECF-c .127(c) 1.184 .244
 In-fdi -.020(c) .158 .876
 In-OPEN .106(c) .964 .342
 Adult literacy .280(c) 2.404 .021
 In Internet tariff .022(c) .109 .914

4 CL-c .038(d) .337 .738
 ECF-c .177(d) 1.762 .087
 In-fdi -.056(d) -.469 .642
 In-OPEN .025(d) .222 .825
 In Internet tariff .013(d) .067 .947

 Collinearity
 Partial Statistics
Model Correlation Tolerance

l LN-PCs-98 -.586 .827
 LN-Avg GNI .088 .781
 CL-c .049 1.000
 ECF-c .113 .991
 In-fdi .113 .928
 In-OPEN -.011 .985
 Adult literacy .202 .872
 In Internet tariff .069 .762

2 LN-Avg GNI .499 .614
 CL-c -.161 .912
 ECF-c .098 .988
 In-fdi .201 .921
 In-OPEN .158 .933
 Adult literacy .456 .811
 In Internet tariff -.338 .558

3 CL-c -.022 .838
 ECF-c .191 .970
 In-fdi -.026 .740
 In-OPEN .156 .931
 Adult literacy .368 .735
 In Internet tariff .018 .285

4 CL-c .056 .805
 ECF-c .282 .938
 In-fdi -.078 .728
 In-OPEN .037 .825
 In Internet tariff .011 .285

(a) Predictors in the Model: (Constant), LN-USERS-98.

(b) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98,
LN-Avg GNI.

(d) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98,
LN-Avg GNI, Adult literacy.

(e) Dependent Variable: PCs (LN-02-LN98).

Regression

Variables Entered / Removed (a)

 Variables Variables
Model Entered Removed Method

 l LN-USERS-98 . Forward (Criterion: Probability-
 of-F-to-enter <= .050)
 2 LN-PCs-98 . Forward (Criterion: Probability-
 of-F-to-enter <= .050)
 3 LN-Avg GNI . Forward (Criterion: Probability-
 of-F-to-enter <= .050)
 4 Adult literacy . Forward (Criterion: Probability-
 of-F-to-enter <= .050)

(a) Dependent Variable: PCs (LN-02-LN98).

Model Summary

 Adjusted
Model R R-Square R-Square Std. Error of the Estimate

 1 .365(a) 0.133 0.112 .96071574101738
 2 .656(b) 0.431 0.401 .78867872347214
 3 .756(c) 0.572 0.538 .69261008796680
 4 .794(d) 0.630 0.590 .65275801844985

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors: (Constant), LN-USERS-9R, LN-PCs-98, LN-Avg GNI.

(d) Predictors: (Constant), LN-USERS-9R, LN-PCs-98, LN-Avg GNI,
Adult literacy.

ANOVA (e)

 Sum of
 Model Squares df Mean Square F Sig.

1 Regression 5.686 1 5.686 6.161 .017(a)
 Residual 36.919 40 .923
 Total 42.605 41

2 Regression 18.347 2 9.173 14.748 .000(b)
 Residual 24.259 39 .622
 Total 42.605 41

3 Regression 24.376 3 8.125 16.938 .000(c)
 Residual 18.229 38 .480
 Total 42.605 4I

4 Regression 26.840 4 6.710 15.748 .000(d)
 Residual 15.765 37 .426
 Total 42.605 41

(a) Predictors: (Constant), LN-USERS-98.

(b) Predictors: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GN1.

(d) Predictors: (Constant), LN-USERS-98, LN-PCs-98, LN-Avg GNI,
Adult literacy.

(e) Dependent Variable: PCs (LN-02-LN98).

Coefficients (a)

 Unstandardised
 Coefficients Standardised
 Coefficients
 Model B Std. Error Beta

1 (Constant) 1.489 .379
 LN-USERS-98 .279 .112 .365

2 (Constant) 1.565 .312
 LN-USERS-98 .470 .102 .615
 LN-PCs-98 -.569 .126 -.599

3 (Constant) -4.021 1.599
 LN-USERS-98 .366 .094 .479
 LN-PCs-98 -.774 .125 -.815
 LN-Avg GNI .676 .191 .480

4 (Constant) -3.991 1.507
 LN-USERS-98 .335 .089 .438
 LN-PCs-98 -.804 .118 -.846
 LN-Avg GNI .537 .189 .381
 Adult literacy 1.311E-02 .005 .280

 Model t Sig.

1 (Constant) 3.929 .000
 LN-USERS-98 2.482 .017

2 (Constant) 5.023 .000
 LN-USERS-98 4.626 .000
 LN-PCs-98 4.512 .000

3 (Constant) -2.514 .016
 LN-USERS-98 3.906 .000
 LN-PCs-98 -6.194 .000
 LN-Avg GNI 3.545 .001

4 (Constant) -2.647 .012
 LN-USERS-98 3.748 .001
 LN-PCs-98 -6.789 .000
 LN-Avg GNI 2.846 .007
 Adult literacy 2.404 .021

(a) Dependent Variable: PCs (LN-02-LN98).

Excluded Variables (e)

 Model Beta In t Sig.

1 LN-PCs-98 -.599(a) -4.512 .000
 LN-Avg GNI .093(a) .552 .584
 CL-c .045(a) .304 .763
 ECF-c .105(a) .708 .483
 In-f4i .109(a) .711 .481
 In-OPEN -.010(a) -.067 .947
 Adult literacy .202(a) 1.290 .205
 In_internet tariff .073(a) .430 .670

2 LN-Avg GNI .480(b) 3.545 .001
 CL-c -.127(b) -1.003 .322
 ECF-c .075(b) .609 .546
 In-fdi .158(b) 1.267 .213
 In-OPEN .124(b) .987 .330
 Adult literacy .382(b) 3.158 .003
 In_internet tariff -.341(b) -2.211 .033

3 CL-c -.016(c) -.133 .895
 ECF-c .127(c) 1.184 .244
 In-fdi .127(c) -.158 .876
 In-OPEN .106(c) .964 .342
 Adult literacy .280(c) 2.404 .021
 In_internet tariff .022(c) .109 .914

4 CL-c .038(d) .337 .738
 ECF-c .177(d) 1.762 .087
 In-fdi -.056(d) -.469 .642
 In-OPEN .025(d) .222 .825
 In_internet tariff .013(d) .067 .947

 Collinearity
 Partial Statistics
 Model Correlation Tolerance

1 LN-PCs-98 -.586 .827
 LN-Avg GNI .088 .781
 CL-c .049 1.000
 ECF-c .113 .991
 In-f4i .113 .928
 In-OPEN -.011 .985
 Adult literacy .202 .872
 In_internet tariff .069 .762

2 LN-Avg GNI .499 .614
 CL-c -.161 .912
 ECF-c .098 .988
 In-fdi .201 .921
 In-OPEN .158 .933
 Adult literacy .456 .811
 In_internet tariff -.338 .558

3 CL-c -.022 .838
 ECF-c .191 .970
 In-fdi -.026 .740
 In-OPEN .156 .931
 Adult literacy .368 .735
 In_internet tariff .018 .285

4 CL-c .056 .805
 ECF-c .282 .938
 In-fdi -.078 .728
 In-OPEN .037 .825
 In_internet tariff .011 .285

(a) Predictors in the Model: (Constant), LN-USERS-98.

(b) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98.

(c) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98,
LN-Avg GNI.

(d) Predictors in the Model: (Constant), LN-USERS-98, LN-PCs-98,
LN-Avg GNI, Adult literacy.

(e) Dependent Variable: PCs (LN-02-LN98).

Details of Model Selection Procedure When Internet Hosts
Are the Relevant n ICT Indicator

Variables Entered / Removed (b)

Model Variables Entered Variables Method
 Removed

1 LN-HOSTS-98, FDI-Avg, Enter
 OPEN-Avg, CL-c, ECF-c,
 LN-PCs-98, LN-Avg GNI,
 Education index(a) --

2 -- Education Backward (criterion:
 index Probability of
 F-to-remove >=.100).

3 -- FDI-Avg Backward (criterion:
 Probability of
 F-to-remove >=.100).

4 -- CL-c Backward (criterion:
 Probability of
 F-to-remove >=.100).

5 -- LN-PCs-98 Backward (criterion:
 Probability of
 F-to-remove >=.100).

6 -- OPEN-Avg Backward (criterion:
 Probability of
 F-to-remove >=.100).

(a) All requested variables entered.

(b) Dependent Variable: HOSTA (LN02-98).

Model Summary

 Adjusted Std. Error
Model R R Square R Square Std. Error of the Estimate

 1 .543(a) 0.294 0.106 1.19706115760003
 2 .541(b) 0.293 0.133 1.17892490645525
 3 .538(c) 0.290 0.157 1.16271328411677
 4 .532(d) 0.283 0.174 1.15089924013000
 5 .522(e) 0.273 0.187 1.14159443945720
 6 .508(t) 0.259 0.195 1.13605956508465

(a) Predictors: (Constant), LN-HOSTS-98, FDI-Avg, OPEN-Avg,
CL-c, ECF-c, LN-PCs-98, LN-Avg GNI, Education index.

(b) Predictors: (Constant), LN-HOSTS-98, FDI-Avg, OPEN-Avg,
CL-c, ECF-c, LN-PCs-98, LN-Avg GNI.

(c) Predictors: (Constant), LN-HOSTS-98, OPEN-Avg, CL-c, ECF-c,
LN-PCs-98, LN-Avg GNI .

(d) Predictors: (Constant), LN-HOSTS-98, OPEN-Avg, ECF-c,
LN-PCs-98, LN-Avg GNI.

(e) Predictors: (Constant), LN-HOSTS-9R, OPEN-Avg, ECF-c,
N-Avg GNI.

(f) Predictors: (Constant), LN-HOSTS-98, ECF-c, LN-Avg GNI.

ANOVA (g)

 Sum of Mean
 Model Squares df Square F Sig.

1 Regression 17.932 8 2.242 1.564 .178(a)
 Residual 42.989 30 1.433
 Total 60.921 38

2 Regression 17.835 7 2.548 1.833 .116(b)
 Residual 43.086 31 1.390
 Total 60.921 38

3 Regression 17.660 6 2.943 2.177 .071(c)
 Residual 43.261 32 1.352
 Total 60.921 38

4 Regression 17.210 5 3.442 2.599 .043(d)
 Residual 43.711 33 l.325
 Total 60.921 38

5 Regression 16.611 4 4.153 3.186 .025(e)
 Residual 44.310 34 1.303
 Total 60.921 38

6 Regression 15.749 3 5.250 4.068 .014(f)
 Residual 45.172 35 1.291
 Total 60.921 38

(a) Predictors: (Constant), LN-HOSTS-98, FDI-Avg, OPEN-Avg, CL-c,
ECF-c, LN-PCs-98, LN-Avg GNI, Education index .

(b) Predictors: (Constant), LN-HOSTS-98, FDI-Avg, OPEN-Avg, CL-c,
ECF-c, LN-PCs-98, LN-Avg GNI.

(c) Predictors: (Constant), LN-HOSTS-98, OPEN-Avg, CL-c, ECF-c,
LN-PCs-98, LN-Avg GNI.

(d) Predictors: (Constant), LN-HOSTS-98, OPEN-Avg, ECF-c, LN-PCs-98,
LN-Avg GNI.

(e) Predictors: (Constant), LN-HOSTS-9R, OPEN-Avg, ECF-c, LN-Avg GNI.

(f) Predictors: (Constant), LN-HOSTS-9R, ECF-c, LN-Avg GNI .

(g) Dependent Variable: HOSTA (LN02-98).

Coefficients (a)

 Unstandardised
 Coefficients Standardised
 Coefficients
Model B Std. Error Beta

1 (Constant) -5.542 3.355
 LN-PCs-98 .129 .225 .109
 LN-Avg GNI .699 .383 .399
 ECF-c .326 .188 .300
 CL-c -7.102E-02 .141 -.087
 FDI-Avg -1.385E-05 .000 -.058
 OPEN-Avg 6.690E-03 .014 .085
 Education index .401 1.540 .061
 LN-HOSTS-98 -.398 .166 -.476

2 (Constant) -5.390 3.253
 LN-PCs-98 .134 .220 .114
 LN-Avg GNI .721 .368 .412
 ECF-c .310 .175 .285
 CL-c -8.097E-02 .133 -.100
 FDI-Avg -1.314E-05 .000 -.055
 OPEN-Avg 8.004E-03 .012 .102
 LN-HOSTS-98 -.380 .147 -.454

3 (Constant) -5.329 3.204
 LN-PCs-98 .131 .217 .111
 LN-Avg GNI .708 .361 .405
 ECF-c .310 .172 .285
 CL-c -7.534E-02 .131 -.093
 OPEN-Avg 8.479E-03 .012 .108
 LN-HOSTS-98 -.378 .145 -.452

4 (Constant) -5.902 3.015
 LN-PCs-98 .144 .214 .122
 LN-Avg GNI .763 .344 .436
 ECF-c .312 .170 .286
 OPEN-Avg 8.330E-03 .012 .106
 LN-HOSTS-98 -.392 .142 -.468

5 (Constant) -6.986 2.529
 LN-Avg GNI .883 .292 .505
 ECF-c .315 .169 .289
 OPEN-Avg 9.582E-03 .012 .122
 LN-HOSTS-98 -.394 .141 -.471

6 (Constant) -6.659 2.484
 LN-Avg GNI .889 .291 .508
 ECF-c .295 .166 .271
 LN-HOSTS-98 -.379 .139 -.453

Model t Sig.

1 (Constant) -1.652 .109
 LN-PCs-98 .573 .571
 LN-Avg GNI 1.825 .078
 ECF-c 1.733 .093
 CL-c -.505 .617
 FDI-Avg -.368 .716
 OPEN-Avg .494 .625
 Education index .260 .796
 LN-HOSTS-98 -2.407 .022

2 (Constant) -1.657 .108
 LN-PCs-98 .610 .546
 LN-Avg GNI 1.962 .059
 ECF-c 1.774 .086
 CL-c -.607 .548
 FDI-Avg -.355 .725
 OPEN-Avg .646 .523
 LN-HOSTS-98 -2.578 .015

3 (Constant) -1.663 .106
 LN-PCs-98 .603 .551
 LN-Avg GNI 1.963 .058
 ECF-c 1.803 .081
 CL-c -.577 .568
 OPEN-Avg .698 .490
 LN-HOSTS-98 -2.601 .014

4 (Constant) -1.957 .059
 LN-PCs-98 .673 .506
 LN-Avg GNI 2.217 .034
 ECF-c 1.830 .076
 OPEN-Avg .693 .493
 LN-HOSTS-98 -2.765 .009

5 (Constant) -2.763 .009
 LN-Avg GNI 3.023 .005
 ECF-c 1.865 .071
 OPEN-Avg .813 .422
 LN-HOSTS-98 -2.801 .008

6 (Constant) -2.680 .011
 LN-Avg GNI 3.058 .004
 ECF-c 1.775 .085
 LN-HOSTS-98 -2.732 .010

(a) Dependent Variable: HOSTA (LN02-98).

Excluded Variables (f)

 Model Beta In t Sig.

2 Education index .061(a) .260 .796

3 Education index .054(b) .238 .814
 FDI-Avg -.055(b) -.355 .725

4 Education index .085(c) .392 .697
 FDI-Avg -.044(c) -.288 .776
 CL-c -.093(c) -.577 .568

5 Education index .102(d) .476 .638
 FDI-Avg -.038(d) -.252 .802
 CL-c -.103(d) -.648 .521
 LN-PCs-98 .122(d) .673 .506

6 Education index .147(e) .744 .462
 FDI-Avg -.050(e) -.335 .740
 CL-c -.102(e) -.647 .522
 LN-PCs-98 .141(e) .796 .432
 OPEN-Avg .122(e) .813 .422

 Partial Statistics
 Model Correlation Tolerance

2 Education index .047 .433

3 Education index .043 .435
 FDI-Avg -.064 .937

4 Education index .069 .472
 FDI-Avg -.051 .950
 CL-c -.101 .858

5 Education index .083 .480
 FDI-Avg -.044 .953
 CL-c -.112 .867
 LN-PCs-98 .116 .664

6 Education index .127 .554
 FDI-Avg -.057 .963
 CL-c .110 .867
 LN-PCs-98 .135 .680
 OPEN-Avg .138 .958

(a) Predictors in the Model: (Constant), LN-HOSTS-98, FDI-Avg,
OPEN-Avg, CL-c, ECF-c, LN-PCs-98, LN- Avg GNI.

(b) Predictors in the Model: (Constant), LN-HOSTS-98, OPEN-Avg,
CL-c, ECF-c, LN-PCs-98, LN-Avg GN1.

(c) Predictors in the Model: (Constant), LN-HOSTS-98, OPEN-Avg,
ECF-c, LN-PCs-98, LN-Avg GNI.

(d) Predictors in the Model: (Constant), LN-HOSTS-98, OPEN-Avg,
ECF-c, LN-Avg GNI.

(e) Predictors in the Model: (Constant), LN-HOSTS-98, ECF-c,
LN-Avg GNI.

(f) Dependent Variable: HOSTA (LN02-98).

Details of Model Selection Procedure When Mobile
Phones Are the Relevant n ICT Indicator

Variables Entered / Removed (b)

 Variables
Model Variables Entered Removed Method

1 LN-M-98, CL-c, ECF-c, -- Enter
 In-OPEN, In-fdi, Adult
 literacy, LN-Avg GNI,
 Education index(a)

2 -- CL-c Backward (criterion:
 Probability of F-to-
 remove >= .100).

3 -- ECF-c Backward (criterion:
 Probability of F-to-
 remove >= .100).

4 -- In-fdi Backward (criterion:
 Probability of F-to-
 remove >= .100).

5 -- Adult Backward (criterion:
 literacy Probability of F-to-
 remove >= .100).

6 -- Education Backward (criterion:
 index Probability of F-to-
 remove >= .100).

7 -- LN-Avg Backward (criterion:
 GNI Probability of F-to-
 remove >= .100).

(a) All requested variables entered.

(b) Dependent Variable: M(ln02-ln98).

Model Summary

 Adjusted
Model R R-Square R-Square Std. Error of the Estimate

 1 .602(a) .362 .248 .92768648860601
 2 .601(b) .361 .264 .91825546325226
 3 .598(c) .358 .276 .91034613791259
 4 .594(d) .352 .285 .90485855365053
 5 .577(e) .333 .279 .90866499916275
 6 .574(f) .330 .289 .90210549580296
 7 .560(g) .314 .287 .90343596063431

(a) Predictors: (Constant), LN-M-98, CL-c, ECF-c, In-OPEN, In-fdi,
Adult literacy, LN-Avg GNI, Education index.

(b) Predictors: (Constant), LN-M-98, ECF-c, In-OPEN, In-fdi, Adult
literacy, LN-Avg GNI, Education index.

(c) Predictors: (Constant), LN-M-98, In-OPEN, In-fdi, Adult
literacy, LN-Avg GNI, Education index.

(d) Predictors: (Constant), LN-M-98, In-OPEN, Adult literacy,
LN-Avg GNI, Education index.

(e) Predictors: (Constant), LN-M-98, In-OPEN, LN-Avg GNI, Education
index.

(f) Predictors: (Constant), LN-M-98, In-OPEN, LN-Avg GN1.

(g) Predictors: (Constant), LN-M-98, In-OPEN.

ANOVA (h)

 Sum of Mean
Model Squares df Square F Sig.

1 Regression 21.960 8 2.745 3.190 .006(a)
 Residual 38.727 45 .861
 Total 60.687 53

2 Regression 21.900 7 3.129 3.710 .003(b)
 Residual 38.787 46 .843
 Total 60.687 53

3 Regression 21.737 6 3.623 4.371 .001(c)
 Residual 38.950 47 .829
 Total 60.687 53

4 Regression 21.386 5 4.277 5.224 .001(d)
 Residual 39.301 48 .819
 Total 60.687 53

5 Regression 20.229 4 5.057 6.125 .000(e)
 Residual 40.458 49 .826
 Total 60.687 53

6 Regression 19.997 3 6.666 8.191 .000(f)
 Residual 40.690 50 .814
 Total 60.687 53

7 Regression 19.061 2 9.530 11.677 .000(g)
 Residual 41.626 51 .816
 Total 60.687 53

(a) Predictors: (Constant), LN-M-98, CL-c, ECF-c, In-OPEN, In-fdi,
Adult literacy, LN-Avg GNI, Education index.

(b) Predictors: (Constant), LN-M-98, ECF-c, In-OPEN, In-fdi, Adult
literacy, LN-Avg GNI, Education index.

(c) Predictors: (Constant), LN-M-98, In-OPEN, In-fdi, Adult literacy,
LN-Avg GNI, Education index.

(d) Predictors: (Constant), LN-M-98, In-OPEN, Adult literacy, LN-Avg
GNI, Education index.

(e) Predictors: (Constant), LN-M-98, In-OPEN, LN-Avg GNI, Education
index.

(f) Predictors: (Constant), LN-M-98, In-OPEN, LN-Avg GNI.

(g) Predictors: (Constant), LN-M-98, In-OPEN.

(h) Dependent Variable: M(ln02-ln98).

Coefficients (a)

 Unstandardised
 Coefficients Standardised
 Coefficients
Model B Std. Error Beta

1 (Constant) -2.680 2.099
 LN-Avg GNI .288 .259 .195
 CL-c -2.370E-02 .090 -.035
 ECF-c 5.048E-02 .118 .055
 In-fdi 5.578E-02 .085 .096
 In-OPEN .652 .297 .293
 Education index -3.989 3.670 -.714
 Adult literacy 2.992E-02 .030 .610
 LN-M-98 -.462 .108 -.640

2 (Constant) -2.840 1.988
 LN-Avg GNI .303 .250 .205
 ECF-c 5.129E.02 .116 .056
 In-fdi 5.321E.02 .083 .091
 In-OPEN .642 .292 .288
 Education index -3.902 3.618 -.699
 Adult literacy 2.964E.02 .030 .604
 LN-M-98 -.464 .106 -.644

3 (Constant) -2.637 1.917
 LN-Avg GNI .307 .248 .207
 In-fdi 5.367E.02 .083 .092
 In-OPEN .640 .289 .288
 Education index -4.339 3.450 -.777
 Adult literacy 3.265E-02 .029 .665
 LN-M-98 -.462 .105 -.641

4 (Constant) -2.647 1.906
 LN-Avg GNI .347 .238 .235
 In-OPEN .623 .286 .280
 Education index -4.398 3.428 -.787
 Adult literacy 3.375E.02 .028 .688
 LN-M-98 -.442 .100 -.613

5 (Constant) -2.293 1.890
 LN-Avg GNI .275 .232 .186
 In-OPEN .605 .287 .272
 Education index -.447 .844 -.080
 LN-M-98 -.454 .100 -.630

6 (Constant) -1.958 1.768
 LN-Avg GNI .217 .203 .147
 In-OPEN .549 .265 .246
 LN-M-98 -.455 .099 -.631

7 (Constant) -.330 .908
 In-OPEN .573 .264 .257
 LN-M-98 -.400 .086 -.555

Model t Sig.

1 (Constant) -1.277 .208
 LN-Avg GNI 1.112 .272
 CL-c -.264 .793
 ECF-c .429 .670
 In-fdi .659 .513
 In-OPEN 2.194 .033
 Education index -1.087 .283
 Adult literacy .998 .324
 LN-M-98 -4.279 .000

2 (Constant) -1.429 .160
 LN-Avg GNI 1.214 .231
 ECF-c .440 .662
 In-fdi .639 .526
 In-OPEN 2.200 .033
 Education index -1.078 .286
 Adult literacy 1.000 .323
 LN-M-98 -4.365 .000

3 (Constant) -1.375 .175
 LN-Avg GNI 1.238 .222
 In-fdi .650 .519
 In-OPEN 2.213 .032
 Education index -1.258 .215
 Adult literacy 1.141 .260
 LN-M-98 -4.387 .000

4 (Constant) -1.389 .171
 LN-Avg GNI 1.456 .152
 In-OPEN 2.175 .035
 Education index -1.283 .206
 Adult literacy 1.189 .240
 LN-M-98 -4.413 .000

5 (Constant) -1.213 .231
 LN-Avg GNI 1.189 .240
 In-OPEN 2.106 .000
 Education index -.530 .599
 LN-M-98 -4.535 .000

6 (Constant) -1.107 .273
 LN-Avg GNI 1.073 .289
 In-OPEN 2.071 .044
 LN-M-98 -4.575 .000

7 (Constant) -.363 .718
 In-OPEN 2.170 .035
 LN-M-98 -4.679 .000

(a) Dependent Variable: M(ln02-ln98).

Excluded Variables (g)

Model Beta In t Sig.

2 CL-c -.035(a) -.264 .793

3 CL-c -.036(b) -.277 .783
 ECF-c .056(b) .440 .662

4 CL-c -.026(c) -.203 .840
 ECF-c .057(c) .451 .654
 In-fdi .092(c) .650 .519

5 CL-c -.021(d) -.160 .874
 ECF-c .087(d) .711 .481
 In-fdi .102(d) .717 .477
 Adult literacy .688(d) 1.189 .240

6 CL-c -.005(e) -.044 .965
 ECF-c .097(e) .814 .420
 In-fdi .091(e) .649 .519
 Adult literacy -.032(e) -.223 .824
 Education index -.080(e) -.530 .599

7 CL-c -.047(f) -.403 .688
 ECF-c .077(f) .652 .518
 In-fdi .127(f) .968 .337
 Adult literacy .030(f) .229 .820
 Education index .005(f) .035 .972
 LN-Avg GNI .147(f) 1.073 .289

 Collinearity
 Partial Statistics
Model Correlation Tolerance

2 CL-c -.039 .826

3 CL-c -.041 .827
 ECF-c .065 .866

4 CL-c -.030 .838
 ECF-c .066 .866
 In-fdi .094 .682

5 CL-c -.023 .839
 ECF-c .102 .915
 In-fdi .103 .685
 Adult literacy .169 4.033E-02

6 CL-c -.006 .879
 ECF-c .115 .958
 In-fdi .092 .696
 Adult literacy -.032 .672
 Education index -.075 .596

7 CL-c -.057 .995
 ECF-c .092 .978
 In-fdi .136 .781
 Adult literacy .032 .805
 Education index .005 .767
 LN-Avg GNI .150 .713

(a) Predictors in the Model: (Constant), LN-M-98, ECF-c, In-OPEN,
In-fdi, Adult literacy, LN-Avg GNI, Education index.

(b) Predictors in the Model: (Constant), LN-M-98, In-OPEN, In-fdi,
Adult literacy, LN-Avg GNI, Education index.

(c) Predictors in the Model: (Constant), LN-M-98, In-OPEN, Adult
literacy, LN-Avg GNI, Education index.

(d) Predictors in the Model: (Constant), LN-M-98, In-OPEN, LN-Avg
GN1, Education index.

(e) Predictors in the Model: (Constant), LN-M-98, In-OPEN, LN-Avg
GNI.

(f) Predictors in the Model: (Constant), LN-M-98, In-OPEN.

(g) Dependent Variable: M(ln02-ln98).

Resources Memory Required 5636 bytes
 Additional Memory Required for Residual Plots 0 bytes
 Elapsed Time 0:00:00.00

Variables Entered / Removed (a)

 Variables Variables
Model Entered Removed Method

1 LN-M-98 -- Stepwise (Criteria: Probability-of-F-
 to-enter <=.050, Probability-of-F-to-
 remove >= .100).

2 In-OPEN -- Stepwise (Criteria: Probability-of-F-
 to-enter <=.050, Probability-of-F-to-
 remove >= .100).

(a) Dependent Variable: M(ln02-ln98).

Model Summary

 Adjusted Std. Error of
Model R R Square R Square the Estimate

1 .501 (a) .251 .236 .93508686061264
2 .560 (b) .314 .287 .90343596063431

(a) Predictors: (Constant), LN-M-98.

(b) Predictors: (Constant), LN-M-98, ln-OPEN.

ANOVA (c)

 Sum of
Model Squares df Mean Square F Sig.

1 Regression 15.219 1 15.219 17.405 .000 (a)
 Residual 45.468 52 .874
 Total 60.687 53

2 Regression 19.061 2 9.530 11.677 .000 (b)
 Residual 41.626 51 .816
 Total 60.687 53

(a) Predictors: (Constant), LN-M-98.

(b) Predictors: (Constant), LN-M-98, ln-OPEN.

(c) Dependent Variable: M(ln02-ln98).

Coefficients (a)

 Unstandardised
 Coefficients
 Standardized
 Std. Coefficients
Model B Error Beta t Sig.

1 (Constant) 1.608 .170 9.444 .000
 LN-M-98 -.361 .086 -.501 -4.172 .000

2 (Constant) -.330 .908 -0.363 .718
 LN-M-98 -.400 .086 -.555 -4.679 .000
 ln-OPEN .573 .264 .257 2.170 .035

(a) Dependent Variable: M(ln02-ln98).

Excluded Variables (c)

Model Beta Ln t Sig.

1 LN-Avg GNI .172 (a) 1.219 .228
 CL-c -.047 (a) -.388 .700
 ECF-c .044 (a) .363 .718
 ln-fdi .121 (a) .889 .378
 ln-OPEN .257 (a) 2.170 .035
 Education index .102 (a) .798 .429
 Adult literacy .119 (a) .948 .348

2 LN-Avg GNI .147 (b) 1.073 .289
 CL-c -.047 (b) -.403 .688
 ECF-c .077 (b) .652 .518
 ln-fdi .127 (b) .968 .337
 Education index .005 (b) .035 .972
 Adult literacy .030 (b) .229 .820

 Collinearity
 Partial Statistics
Model Correlation Tolerance

1 LN-Avg GNI .168 .719
 CL-c -.054 .995
 ECF-c .051 .993
 ln-fdi .124 .781
 ln-OPEN .291 .955
 Education index .111 .886
 Adult literacy .132 .920

2 LN-Avg GNI .150 .713
 CL-c -.057 .995
 ECF-c .092 .978
 ln-fdi .136 .781
 Education index .005 .767
 Adult literacy .032 .805

(a) Predictors in the Model: (Constant), LN-M-98.

(b) Predictors in the Model: (Constant), LN-M-98, ln-OPEN.

(c) Dependent Variable: M(ln02-ln98).

REFERENCES

Barro, R. J. and J. W. Lee (2001) International Data on Educational Attainment: Updates and Implications. Oxford Economic Papers 53:3, 541-63.

Bedi, Arjun (1999) The Role of Information and Communication Technologies in Economic Development: A Partial Survey. Centre for Development Reasearch, Universitat Bonn. (Discussion Papers on Development Policy 7.)

Caselli, Francesco and W. John II Coleman (2001) Cross-country Technology Diffusion: The Case of Computers. American Economic Review 81, 328-35.

Duncombe, R. (2000) Information and Communication Technology, Poverty and Development in Sub-Saharan Africa. Institute for Development Policy and Management, University of Manchester, UK.

Hardy, Andrew P. (1980) The Role of the Telephone in Economic Development. Telecommunications Policy 4:4, 278-286.

Hargittai, Eszter (1999) Weaving the Western Web: Explaining Differences in Internet Connectivity among OECD Countries. Telecommunications Policy.

ITU (1999) Telecommunications Indicator Handbook [Internet]. International Telecommunication Union. Retrieved from the World.

ITU (2000) World Telecommunications Indicators: Chronological Time Series 1960-1999 [STARS Software Package]. International Telecommunication Union.

ITU (2001) Country Data by Region [Internet]. International Telecommunication. Retrieved from the World Wide Web: www7.itu.inffbdt_cds/IDC/Countries.idc.

ITU (2002) International Telecommunication Indicators. Available at:

<www.itu.int/ITU-D/ict/statistics>.

ITU (2003) International Telecommunication Indicators. Available at: <http://www.itu.int/ITU-D/ict/statisties>.

ITU Secretariat. (1998) Methodological Note on the Choice of Teledensity Bands Internet International Telecommunication Union Secretariat Study Group 3 Tariff and Accounting Principal. Retrieved from the World Wide Web: <http://www.itu.int/sg3focus/choiceof.htm>.

Jayakar, Krishna P. (1999) Local Exchange Competition in Early US Network Development: Considerations for Developing Countries. Telecommuni-cations Policy 23:5, 375-87.

Jussawalla, M. (1999) The Impact of ICT Convergence on Development in the Asian Region. Telecommunications Policy 23:3-4, 217-234.

Kiiski, Sampsa, and Matti Pohjola (2002) Cross Country Diffusion of the Internet. Information Economics and Policy.

Niininen, Petri (2001) Computers and Economic Growth in Finland. In M. Pohjola (ed.)

Norris, Pippa (2000) The Global Divide: Information Poverty and Internet Access Worldwide. Available at: <www.ksg.harvard.edu/people/>

Norton, S. (1992) Information Costs, Telecommunications, and the Microeconomics of Macroeconomic Growth. Economic Development and Cultural Change 41, 175-96.

Pohjola, M. (2001) Information Technology and Economic Growth: A Cross Country Analysis. In M. Pohjola (ed.) Information Technology, Productivity, and Economic Growth. New York: Oxford University Press.

Pohjola, Matti (2000) Information Technology and Economic Growth: A Cross-Country Analysis. In M. Pohjola (ed.) Information Technology, Productivity, and Economic Growth: International Evidence and Implications .for Economic Development. New York: Oxford University Press. pp. 242-56.

Quah, Danny (2001) Technology and Economic Growth: Some Lessons for the New Economy. Available at: <http://www.econ.lse.ac.uk/staff/dquah/p/dp0522.pdf>

Stiglitz, Joseph (1989) Economic Organisation, Information and Development. In J. Behrman and T. N. Srinivasan (eds) Handbook of Development Economic Vol. 1. Amsterdam: North Holland.

Stoneman, P. (1983) The Economic Analysis of Technological Change. Oxford: Oxford University Press.

Wong, P. K. (2001) ICT Industry Development and Diffusion in South East Asia. Helsinki: UNU/WIDER. (WIDER Discussion Paper WDP 2001/08.)

Tasnocm Zafar <tzafar@wiwi.uni-frankfurt.de> is Lecturer in Economics and Khalid Aftab <vicechancellor@gcu.edu.pk> is Professor of Economics at GC University, Lahore.

******** insert table jpg 67, 71-73 ************

Table 1
'Information-poor Countries' included in Analysis with
Digital Access Index (DAI) Score Less than or Equal to
0.37 out of 1

African Countries

Country Score

Algeria 0.37
Benin 0.12
Burkina Faso 0.08
Burundi 0.10
Cameroon 0.16
Central African Rep. 0.10
Chad 0.10
Comoros 0.13
Congo 0.17
Cote d'Ivoire 0.13
Djibouti 0.15
Egypt 0.40
Equatorial Guinea 0.20
Ethiopia 0.10
Gambia 0.13
Ghana 0.16
Guinea O.10
Guinea-Bissau O.10
Kenya 0.19
Lesotho 0.19
Madagascar 0.15
Malawi 0.15
Mali 0.09
Mauritania 0.14
Mozambique 0.12
Nepal 0.19
Niger 0.04
Nigeria 0.15
Rwanda 0.15
Senegal 0.14
Sudan 0.13
Tanzania 0.15
Uganda 0.17
Zambia 0.17
Zimbabwe 0.29

Asian Countries

Armenia 0.30
Azerbaijan 0.24
Bangladesh 0.18
Bhutan 0.13
Cambodia 0.17
Georgia 0.37
India 0.32
Indonesia 0.34
Kyrgyzstan 0.32
Lao P.D.R. 0.15
Pakistan 0.24
Syria 0.28
Tajikistan 0.21
Turkmenistan 0.37
Uzbekistan 0.31
Viet Nam 0.31

American Countries

Haiti 0.15
Honduras 0.29
Nicaragua 0.19

European Country

Moldova 0.37

Oceania

Papua New Guinea 0.26

Source: International Telecommunication Indicators 1998-2003.

Note: Scores arc on a scale of 0 to 1 where 1 = highest access
and 0 =lowest score. DAI values arc shown to hundreds of a
decimal point. Countries with the same DAI value arc ranked
by thousands of a decimal point by ITI.

Table 2
Estimates of the Gompertz Technology Diffusion Model

Cross-section Results for Countries with DRI Score Less than or
Equal to 0.4

Dependent Variable (Internet Users) ln [U.sub.2] - ln [U.sub.98]

 (1) (2)

Speed of Diffusion 0.718 *** 0.780 ***
 (0.070) (0.072)

Constant [alpha][[beta].sub.0] -3.918 ** 11.267 **
 (1.943) (1.947)

GNI per Capita [alpha][[beta].sub.1] 0.549 *** 0.491 ***
 (0.200) (0.202)

Adult Literacy

Secondary and Tertiary Education

Education Index 1.097 **
 (0.573)

Civil Liberties 0.06347
 (0.057)

Economic Freedom 0.08149
 (0.071)

Foreign Direct Investment

Openness to Intcmational Trade 0.567 *** 0.383 **
 (0.172) (0.185)

Personal Computers -98 0.163 *
 (0.097)

Internet Access Cost -0.295 *** -0.230 **
 (0.109) (0.113)

F-Test 32.163 *** 18.450 ***

R 0.879 0.902

[R.sup.2] 0.772 0.813

Adjusted [R.sup.2] 0.748 0.769

Number of Observations 42 42

Note: Standard errors are in parentheses. * = Significant at
(0.10), i.e., at 10 percent, ** = Significant at (0.05), i.e.,
at 5 percent and *** = Significant at (0.01), i.e., at 1 percent.

Table 3
Estimates of the Gompertz Technology Diffusion Model

Cross-section Results for Countries with DRI Score Less than or
Equal to 0.4

Dependent Variable (Personal Computers) ln [PCs.sub.02] - ln
[PCs.sub.98]

 (1) (2)

Speed of Diffusion [alpha] 0.810 *** 0.804 ***
 (0.115) (0.118)

Constant [alpha][[beta].sub.0] -4.734 *** -3.991 ***
 (1.526) (1.507)

GNI per Capita [alpha][[beta].sub.1] 0.561 *** 0.537 ***
 (0.184) (0.189

Adult Literacy 0.01483 *** 0.01311 ***
 (0.005) (0.005)

Secondary and Tertiary Education

Education Index

Political Rights

Civil Liberties

Economic Freedom 0.155 **
 (0.088)

Foreign Direct Investment

Openness to International Trade

Personal Computers -98 0.334 ***
 (0.087)

Internet Access Cost

F-Test 13.935 *** 15.748 ***

R 0.812 0.794

[R.sup.2] 0.659 0.630

Adjusted [R.sup.2] 612 0.600

Number of Observations 41 41

Note: Standard errors are in parentheses. * = Significant at
(0.10), i.e., at 10 percent, ** = Significant at (0.05), i.e.,
at 5 percent and *** = Significant at (O.ol, i.e., at 1 percent.

Table 4
Estimates of the Gompertz Technology Diffitsion Model

Cross-section Results for Countries with DRI Scores Less than or
Equal to 0.4

Dependent Variable Internet Hosts (ln [H.sub.02] - ln [H.sub.98])

Speed of Diffusion [alpha] 0.379 ***
 (0.139)

Constant [alpha][[beta].sub.0] -6.659 ***
 (2.484)

GNI per Capita [alpha][[beta].sub.0] 0.889 ***
 (0.291)

Economic Freedom 0.295 ***
 (0.166)

F-Test 4.068 ***

R 0.508

[R.sup.2] 0.259

Adjusted [R.sup.2] 0.200

Number of Observations 38

Note: Standard errors are in parentheses. * = Signiticant at
(0.10), i.e., at 10 percent, ** = Signiticant at (0.05), i.e.,
at 5 percent and *** = Signiticant at (0.01), i.e., at 1 percent.

Table 5
Estimates of the Gompertz Technology Diffusion Model

Cross-section Results for Countries with Low DRI Scores Less than
or Equal to 0.4

Dependent Variable Mobile Phones (ln [M.sub.02] - ln [M.sub.98])

 (1) (2)

Speed of Diffusion [alpha] -0.330 -1.958
 (0.908) (1.768)

Constant [alpha][[beta].sub.0] 0.400 *** 0.455 ***
 (0.086) (0.099)

GNI per Capita [alpha][[beta].sub.1] 0.217
 (0.203)

Openness to International Trade 0.573 ** 0.549 **
 (0.264) (0.265

F-Test 11.677 *** 8.191 ***

R 0.560 0.574

[R.sup.2] 0.314 0.330

Adjusted [R.sup.2] 0.287 0.289

Number of Observations 53 53

Note: Standard errors are in parentheses. * = Significant at (0.10),
i.e., at 10 percent, ** = Significant at (0.05), i.e., at 5 percent
and *** = Significant at (0.01), i.e., at 1 percent. (This should
be cleared.)