5.1 Descriptive analysis
This subsection documents the descriptive review of the data set employed in the study and a correlation analysis of the independent variables to ensure that no perfect degree of association exists among the independent variables which could affect the reliability of the result. The descriptive tools utilised by this study are the mean, median, maximum, minimum, skewness, kurtosis and standard deviation.
The mean shows the average value which measures the extent of central tendency which is derived by the sum of the values in the variable data set divided by the total number of the values; the median shows the middle value in each variable data set; the maximum shows the highest value among each variable data set and its country is highlighted; the minimum shows the lowest value of each variable in the data set; the standard deviation shows the square root of the absolute value of the variance, where the variance is the difference between each of the values in each variable data set; the skewness shows the degree of asymmetry of the distribution which could either be positively or negatively skewed; and kurtosis measures the degree to which the frequency distribution is focused on the mean or the degree of peakedness of the distribution whereby it could be mesokurtic (when the kurtosis coefficient = 0), platykurtic (when the kurtosis coefficient > 0) and leptokurtic (when the kurtosis coefficient is < 0).
Table
2 shows the descriptive analysis information, while Table
3 shows the correlation analysis carried out.
Table 2
Descriptive analysis.
Source: computed by the authors
Mean | 5.67 | 4966.93 | 270.33 | − 0.73 | 30.09 | 2.59 | 46.72 |
Median | 2.35 | 84.67 | 47.50 | − 0.80 | 20.78 | 2.06 | 46.60 |
Maximum | 42.68 | 170,493.8 | 3994.70 | 0.59 | 227.78 | 30.34 | 51.57 |
Minimum | 0.19 | 1.085 | 0.40 | − 1.71 | 4.27 | − 9.06 | 40.34 |
SD | 8.47 | 22,617.02 | 738.58 | 0.53 | 32.9 | 4.18 | 2.40 |
Skewness | 2.66 | 5.5148 | 3.85 | 0.52 | 3.56 | 2.28 | − 0.14 |
Kurtosis | 9.76 | 34.02 | 17.14 | 2.70 | 17.60 | 15.14 | 2.61 |
Observations | 165 | 165 | 165 | 165 | 162 | 165 | 130 |
Table 3
Multicollinearity test.
Source: computed by the authors
lintusph | 1.00 | | | | | | |
lintphsci | 0.41 | 1.00 | | | | | |
Lscijl | 0.07 | 0.49 | 1.00 | | | | |
Lpse | 0.35 | 0.07 | − 0.05 | 1.00 | | | |
Rule | 0.34 | − 0.14 | − 0.11 | 0.18 | 1.00 | | |
lcredit | 0.29 | 0.02 | − 0.28 | 0.56 | 0.19 | 1.00 | |
lgdpcgr | 0.11 | 0.04 | 0.09 | 0.08 | 0.33 | 0.07 | 1.00 |
From Table
2, Internet users per 100 people (INTUSPH) had an average value of 5.67 and a median value of 2.35; the highest number of Internet users per hundred people was found to be 42.48 in Nigeria as at 2014; and the lowest number of Internet users per hundred people recorded was 0.19 in Niger as at 2014. Furthermore, the number of Internet users per hundred people had a standard deviation of 8.47; positive skewness of 2.66; and kurtosis of 9.76 which shows that the distribution is platykurtic (greater than zero). Scientific and technical journal (SCIJL) capturing innovation is characterised by an average value of 270 journals; a median value of 47 journals; a maximum value of 3994 which was attained by Nigeria in 2014; and a minimum value of 0, recorded for Cape Verde in 2006.
Rule of law had an average value of − 0.73; a median value of − 0.80; and a maximum value of 0.59 for the selected countries. In terms of the standard deviation, the value is 0.53, which shows less deviation from its mean value; the descriptive result also shows that the data for institutional variable (rule) are positively skewed and are platykurtic. In terms of the primary school enrolment rate in ECOWAS for the selected time period, the average rate of primary school enrolment was 46.72%; the median value was 46.6%; the highest primary school enrolment rate was 51.57%, attained by Senegal in 2014 and the lowest primary school enrolment rate for the countries was 40.34% attained by Niger in 2004.
Gross domestic product per capita growth rate (GDPPCGR) for ECOWAS in the time period had an average of 2.59%; and a median of 2.06%. A maximum of 30.34% was attained by Nigeria in 2004 while a minimum of − 9% attained by Mali in 2004. Gross domestic product per capita growth rate had a standard deviation of 4.18 and a positive skewness of 2.28 and was platykurtic with a kurtosis value of 15.14. In terms of the credit provided by the domestic financial sector, an average value of 30%; a median value of 20.78% were observed, while a maximum value of 227.78% was attained by Liberia in 2004 and a minimum value of 4.27% attained by Guinea-Bissau, in 2004. Furthermore, a standard deviation of 32.9; a positive skewness of 3.56 and a platykurtic kurtosis of 17.60 were observed.
Data on the number of Internet users show the total number of people in a country with access to Internet facilities to surf the web for various reasons. The data set shows that Nigeria is the highest in terms of Internet usage, followed by Cote d’Ivoire, while Guinea-Bissau has the lowest Internet usage. Nigeria in 2004 had approximately 1.8 million Internet users and in 2015 had approximately 86 million Internet users. Cote d’Ivoire had 151,193 Internet users in 2004 and approximately 4.8 million Internet users by 2015. Guinea-Bissau, on the other hand, had 25,889 Internet users by 2004 and in 2015, 65,302 Internet users. It is important to note that the population of the country of interest also accounts for the number of Internet users (World Bank
2017a,
b).
Table
3 presents the correlation matrix for the variables in the model; an incidence of strong correlation among the independent variables may violate the working assumptions of the estimation technique and hereby produce unreliable results. The result indicates that the strongest correlation is seen between primary school enrolment (PSE) and domestic credit (credit), followed by the relationship between technological progress (scij) and Internet users (intusph). The overall assessment of the pairwise correlation shows that multicollinearity (a perfect relationship) is absent in the model which guarantees the reliability of the model.
5.2 Results from econometric estimation
The chosen econometric technique of analysis is the fixed effects (FEM) and random effects (REM) panel data analysis (which caters for the possible individual fixed effects that could occur from the nature of the data panel) for which the Hausman test (probability value of the Chi-square test) is usually performed after the FEM and REM to determine the most appropriate between the two. The estimated coefficients could be used to determine the degree of relationship and impact existing between the variables of interest.
The rule of thumb for deciding the most appropriate model (between the REM and FEM) states that given that the FEM was run first before the REM if the Chi-square probability value of the Hausman test is less than 0.05, the FEM is most appropriate and if the Chi-square probability value is greater than 0.05, the REM is the most appropriate for interpretation and policy recommendation. Furthermore, the use of the FEM signifies the presence of individual-specific fixed effects which could affect the result if not taken care of during the estimation process, while the choice of the REM indicates the absence of the individual-specific effects. Table
4 shows the Hausman test result.
Table 4
Hausman test result.
Source: computed by the authors
Fixed effects | Random effects | Fixed effects | Random effects | Fixed effects | Random effects |
Reject | Accept | Accept | Reject | Reject | Accept |
While juxtaposing the relationship between the number of Internet users, innovation and human development, the results in terms of individual statistical significance show that in ECOWAS, Internet usage, innovation and a combination of the former and the latter (technical knowledge) statistically impact human development (columns 1–3 at 5% level of significance). Furthermore, the value of the statistically significant coefficient result is positive. This shows that the Internet usage, innovation and technical knowledge in ECOWAS have a positive impact on human development. This implies that an increase in Internet usage and innovation individually (columns 1 and 3) and wholly (column 2), significantly, positively affects human development in ECOWAS which further implies that the a priori expectation is attained in ECOWAS. This indicates that technology improvement as a result of knowledge is essential to human development. The positive significant result of this study is consistent with Ejemeyovwi and Osabuohien (
2018). The results are displayed in Table
5.
Table 5
Panel estimation results (dependent variable: LHDI).
Source: the authors
Internet usage (lintusph) | 0.017* (4.17) | | |
Technical knowledge (Internet usage * innovation) | | 0.016* (5.75) | |
Innovation (lscijl) | | | 0.02* (4.37) |
Primary school enrolment | 1.279* (8.46) | 1.179* (8.43) | 1.430* (11.11) |
Rule of law (institution) | − 0.008 (1.05) | − 0.02* (− 2.51) | − 0.010 (− 1.27) |
Domestic credit provided by financial sector | − .006 (− 0.65) | − .010 (− 1.24) | − .004 (0.50) |
GDP per capita growth rate | − .0007 (− 0.23) | 0.0004 (− 0.15) | − 0.002 (− 0.78) |
Constant | − 2.50* (10.15) | − .2.36* (10.28) | 2.79* (13.62) |
F-statistics | | 68.71 | |
Prob > F | | 0.000 | |
Wald χ2 (5) | 269.34 | | 289.27 |
Prob > χ2 | 0.000 | 0.000 | 0.000 |
Corr (u_i, Xb) | 0 | 0.10 | 0 |
Number of observationsa | 97 | 97 | 97 |
Number of groups | 15 | 15 | 15 |
ICT usage is an effective medium for development as applied by many developed and developing countries (Zimbabwe—
e-
Hurudza phones; India—
Reuters Market Light; Zambia—
prepaid voucher,
MRIAgro; Kenya—
M-
Pesa, iCow). Human development perspective is not left out with distance learning programs, digitisation of health care products (e-health products) and means of communication. The interaction between innovation (research and development) and Internet technology utilisation refers to: (1) the constant discovery and production of new products
1 that could solve the problems/issues (such as distance between two parties) associated with human development. (2) The use of Internet facilities to enable research through literature (empirical, theoretical, methodological and practical) search and dissemination of these discoveries as journal articles online to the world.
ICT adoption could lead to efficiency through instant information dissemination which reduces the information variance between the different users at the various ends (supply and demand) of each value chain that exists across all sectors and markets. Some of the sectors through which this Internet utilisation could be helpful is the: (1) financial market—to access funds and know current financial policies in order to make available funds to drive human development; (2) labour market—to create a platform for the supply and demand in the labour market which enables firms and the government to plan effectively to improve the human development levels; foreign exchange market—to be updated with the exchange rate information and other international information in real time (spontaneously) in an economy.
The impact of the positive interaction between Internet usage and innovation on human development as found by this study has socio-economic implications for the firms, household, government. The firms experience increase in productivity (more output over the same timeframe), which contributes to the profit-maximising objective of a firm. The household also constitutes the firms, but however, work–life balance (through the usage of the social arm of the Internet usage and innovation such as the social media for leisure) would contribute to the maximisation of satisfaction (utility). The government’s objective is to maximise the welfare of the governed people, and hence, Internet usage, innovation as well as improved human development contribute to the achievement of that goal through spontaneous information dissemination coupled with the households and the firms’ achievement of their own objectives as the welfare maximisation of the government.
It is important to mention that there are crucial concerns which can be the result of the presence of heterogeneity among the different ECOWAS member countries; with respect to the variables used, it is necessary to disaggregate and hence, show the differential impact of Internet usage on each of the member countries. To achieve this objective, the two-way error component within modelling approach is utilised. This unmasks the expected heterogeneity. The two-way error component within modelling is a version of the least square dummy variable (LSDV) that assumes that the explanatory variables are independent of the error term and have a constant slope coefficient, but the intercept varies over countries and time as it has been noted by Vijayamohanan (
2016). The results are shown in Table
6.
Table 6
Two-way error component within modelling
Source: the authors
Country 2: Burkina Faso | − 0.08* (15.98) | − 0.08* (− 17.43) | − 0.08* (− 17.87) |
Country 3: Cape Verde | 0.11* (9.87) | 0.16* (9.19) | 0.17* (9.54) |
Country 4: Cote d’Ivoire | − 0.01 (− 1.53) | − 0.01* (− 2.56) | − 0.01 (− 1.91) |
Country 5: Ghana | 0.05* (7.21) | 0.04* (6.87) | 0.04* (5.86) |
Country 6: Guinea | − 0.08 (0.01) | − 0.06* (− 6.44) | − 0.05* (5.48) |
Country 7: Guinea-Bissau | − 0.08* (− 6.36) | − 0.05* (− 3.50) | − 0.02* (− 2.88) |
Country 8: Liberia | − 0.06* (− 6.65) | − 0.03* (− 2.48) | − 0.02 (− 1.88) |
Country 9: Mali | − 0.04* (− 8.54) | − 0.04* (− 8.00) | − 0.04* (− 7.61) |
Country 10: Mauritania | − 0.02* (− 3.07) | 0.006 (0.55) | 0.003 (0.33) |
Country 11: Niger | − 0.11* (19.25) | − 0.10* (− 18.07) | − 0.10* (− 17.06) |
Country 12: Nigeria | 0.001 (0.25) | − 0.03* (− 2.78) | − 0.03* (− 2.71) |
Country 13: Sierra Leone | − 0.05* (− 5.99) | 0.06* (− 7.26) | − 0.08* (− 11.82) |
Country 14: Senegal | − 0.10* (− 13.05) | − 0.07* (− 6.64) | − 0.06* (− 5.82) |
Country 15: Togo | − 0.01 (− 2.34) | − 0.005 (0.40) | − 0.004 (− 0.61) |
Constant (country 1: Benin) | − 2.49* (− 10.42) | − 2.37* (− 10.59) | − 2.74* (13.62) |
Unravelling the within-estimation results of the impact of Internet usage on human development, 11 countries had statistically significant values out of the 15 ECOWAS countries. While Cape Verde, Ghana and Nigeria reported positive coefficients, the other countries reported negative individual regression coefficients. More interesting is that the positive coefficients of the three countries outweighed the negative coefficients so significantly that the general panel estimation coefficient for Internet usage, as indicated in Table
6, had positive values of the coefficient. The significant influence of Internet usage on human development implies that Internet usage has a significant role to play to influence the outcome of human development both at the individual (country) level and in the ECOWAS sub-region.
In terms of technical knowledge (Internet usage and innovation interaction) and human development nexus, it is observed that for most of the countries, a statistically significant relationship is observed. Also observed are Cape Verde, Ghana, Guinea-Bissau and Sierra Leone which individually report positive impact of technical knowledge on human development among the ECOWAS countries, while the other countries have a negative impact. This implies that the magnitude of impact of the four variables are strong relative to the other countries within the ECOWAS countries to positively affect human development in ECOWAS sub-region as seen in Table
5.
With regard to the individual country coefficients of innovation’s impact on human development four countries (Cote d’Ivoire, Liberia, Mauritania and Togo) had statistically insignificant relationships, while for the other countries, a significant relationship was recorded (see Table
6). Cape Verde reported a positive coefficient, which can be attributed to the level of research and development ongoing in that economy as seen in the descriptive statistics explanation (see Table
2).
In terms of the elasticity of the coefficients, all of the individual coefficients for Internet usage, innovation and technical knowledge in all the countries are inelastic given that they are below 1.0 (< 1.0). This implies that the degree of impact of a change in any of the independent variables will lead to a less than proportionate change in human development. This further implies that for the human development values to be significantly increased, there would need to be a strong or high level of change in Internet usage, technical knowledge and innovation within ECOWAS. The significant impact of Internet usage, innovation and technical knowledge (interaction of Internet usage and innovation) on human development implies that Internet usage has a significant role to play to influence the outcome of human development at both the individual country level and the ECOWAS community as a whole. Also, this result proves that some countries are more involved in developing Internet usage alone, innovation alone, or technical knowledge alone and some countries, none of the three cases above.