Research Paper

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

journal homepage: www.elsevier.com/locate/techfore

Enhancing ICT for environmental sustainability in sub-Saharan Africa

Simplice A. Asongua,b,⁎, Sara Le Rouxa, Nicholas Biekpeb

a Oxford Brookes University, Faculty of Business, Department of Accounting, Finance and Economics, United Kingdom b Development Finance Centre, Graduate School of Business, University of Cape Town, Cape Town, South Africa

A R T I C L E I N F O

JEL classification: C52 O38 O40 O55 P37

Keywords: CO2 emissions ICT Economic development Sub-Saharan Africa

A B S T R A C T

This study examines how increasing ICT penetration in sub-Saharan Africa (SSA) can contribute towards en- vironmental sustainability by decreasing CO2 emissions. The empirical evidence is based the Generalised Method of Moments and forty-four countries for the period 2000–2012. ICT is measured with internet penetration and mobile phone penetration while CO2 emissions per capita and CO2 emissions from liquid fuel consumption are used as proxies for environmental degradation. The following findings are established: First, from the non- interactive regressions, ICT (i.e. mobile phones and the internet) does not significantly affect CO2 emissions. Second, with interactive regressions, increasing ICT has a positive net effect on CO2 emissions per capita while increasing mobile phone penetration alone has a net negative effect on CO2 emissions from liquid fuel con- sumption. Policy thresholds at which ICT can change the net effects from positive to negative are computed and discussed. These policy thresholds are the minimum levels of ICT required, for the effect of ICT on CO2 emissions to be negative. Other practical implications for policy and theory are discussed.

1. Introduction

It is difficult to achieve sustainable development and environmental protection without three main elements, namely: accountability, transparency and the wider participation of the public through in- formation flow (Chemutai, 2009). The third element requires the availability of a communication tool that enables diffusion of knowl- edge, in order to reduce the informational/knowledge deficiencies that are associated with environmental degradation. CO2 emissions con- tribute to environmental degradation or pollution. Hence in this study, the amount of pollution caused by CO2 emissions will be used as a proxy for environmental degradation. This study is concerned with assessing how enhancing information and communication technology (ICT) can affect the emission of a greenhouse gas such as CO2.

1

The positioning of the inquiry is motivated by four main trends in policy and scholarly circles, namely: the need for effective commu- nication strategies in environmental governance for multilateral policy coordination; the great potential for the penetration of ICT in Sub- Saharan Africa (SSA); issues pertaining to global warming and en- vironmental sustainability; and gaps in the existing literature. These trends are considered in further detail below.

First, communication as a tool for environmental management has often been overlooked as part and parcel of environmental governance.

With regard to the United Nations Task Force on Environment and Human Settlements credible governance of the environment can ex- clusively be achieved via effective coordination at national levels. The narrative further notes that only through effective communication can departments which provide consistent guidance to special agencies on environmental management respond to specific requirements within the government. While a comprehensive implementation of Multilateral Environmental Agreements (MEAs) and Agenda 212 is achievable via advanced communications among secretariats of MEA, it is important to share environmental databases among institutions, in order to ad- dress challenges like false reporting and other information asymmetries (Chemutai, 2009).

Second, recent ICT literature is consistent with the view that com- pared to other more advanced economies in the world (Asia, Europe & North America) where ICT is reaching levels of saturation, there is more room for ICT penetration in SSA (see Penard et al., 2012; Asongu, 2013; Afutu-Kotey et al., 2017; Asongu and Nwachukwu, 2016a). Such a promising potential for penetration can be leveraged by policy makers in order to address sobering policy challenges to sus- tainable development, such as environmental pollution and global warming.

Third, environmental sustainability by means of mitigating green- house gas (such as CO2) emissions is a pressing agenda for the

http://dx.doi.org/10.1016/j.techfore.2017.09.022 Received 10 April 2017; Received in revised form 17 September 2017; Accepted 23 September 2017

⁎ Corresponding author at: Development Finance Centre, Graduate School of Business, University of Cape Town, Cape Town, South Africa. E-mail addresses: [email protected] (S.A. Asongu), [email protected] (S. Le Roux), [email protected] (N. Biekpe).

1 For the purpose of simplicity, in some narratives, the terms ICT, the internet and mobile phones are used interchangeably. 2 Agenda 21 is one of the United Nation's voluntarily implemented plans of action that is non-binding with respect to sustainable development.

Technological Forecasting & Social Change 127 (2018) 209–216

Available online 07 October 2017 0040-1625/ © 2017 Elsevier Inc. All rights reserved.

T

achievement of the Sustainable Development Goals (SDGs) (see Asongu et al., 2016a). This is particularly relevant for SSA because of the recent growth resurgence of the sub-region; the persistent energy crisis in the sub-region; poor energy management and consequences of climate change.

Over the last two decades, Africa has been experiencing a recent phase of growth resurgence after several decades of lost growth, due partly to failed Structural Adjustment Programmes (SAP) (see Fosu, 2015). Some accounts maintain that SSA has recently hosted about seven of the World's ten fastest growing economies (Asongu and Rangan, 2015).

However, the management of energy in the sub-region has not been effective, in spite of the apparent energy crisis which represents one of the most critical policy syndromes in the post-2015 development era.3

Shurig (2015) found that access to electrical energy in the sub-region is limited to 5% of the population. Moreover, the consumption of energy in SSA is below 17% of the global average: the equivalent of energy consumed by a single state such as New York, USA.

Inefficiency has recently characterised the management of energy in most African nations (Anyangwe, 2014). To put this point into per- spective, Nigeria — the most populous country in Africa, devotes a large proportion of government resources to subsidising the use of fossil fuels, instead of investing in alternative and renewable energy sources. This has led to the use of electricity generators (that burn subsidised petroleum fuel) to compensate for outages of and shortages in elec- tricity supply (see Akpan and Akpan, 2012).

It is now abundantly clear that global warming which is one of the most dominant challenges in the post-2015 development era, is a direct consequence of the unsustainable consumption of fossil fuels (see Huxster et al., 2015). Unfortunately, compared to other regions of the world, Africa may be most severely affected by the negative effects associated with global warming (Kifle, 2008), partly because compared to other continents of the world Africa lacks adequate financial re- sources with which to deal with the consequences of global warming. This sobering prospect is broadly shared by Akpan and Akpan (2012) who have posited that CO2 emissions constitute about three-quarter of the world's emissions in greenhouse gas, which cause global warming.

Fourth, this inquiry unites the three strands above by examining how enhancing ICT affects environmental sustainability through the reduction of CO2 emissions. In addition to insights supporting the in- quiry discussed in the previous paragraphs, the intuition motivating the inquiry is that by sharing information and potentially decreasing cor- responding information asymmetry, ICT can reduce transaction and travelling costs that are associated with the CO2 emissions in house- holds and corporations. Such intuition can be framed in theory- building, as opposed to the testing of theory. Accordingly, given that practical and policy implications will be provided by the study, we are consistent with the literature (see Costantini and Lupi, 2005; Narayan et al., 2011) in arguing that applied econometrics should not ex- clusively be based on the rejection and acceptance of existing theories. This is essentially because an empirical exercise that is founded on solid intuition could pave the way towards theory-building, especially in the light of growing SDGs challenges.

The positioning of this study deviates from existing literature which has been fundamentally based on investigating nexuses between energy consumption, CO2 emissions and economic growth. While the first strand of this extant literature focuses on the relationship between environmental pollution and economic growth with particular articu- lation on the Environmental Kuznets Curve (EKC) hypothesis (see Diao et al., 2009; Akbostanci et al., 2009; He and Richard, 2010),4 the

second strand is concerned with linkages between energy consumption, economic growth and environmental pollution (Jumbe, 2004; Ang, 2007; Odhiambo, 2009a, 2009b; Apergis and Payne, 2009; Menyah and Wolde-Rufael, 2010; Ozturk and Acaravci, 2010; Begum et al., 2015; Bölük and Mehmet, 2015) and nexuses between the consumption of energy and economic growth (Mehrara, 2007; Esso, 2010).

A common drawback to the highlighted studies is the absence of a policy variable with which associated policy syndromes underlying the inquiries can be addressed. We argue that inquiries motivated by nexuses between policy syndromes (CO2 emissions) and macro-eco- nomic variables (e.g. energy consumption and economic growth) pro- vide results with limited policy relevance because policy makers are not provided with policy instruments with which to address associated policy syndromes. The highlighted gap is addressed using ICT as a policy variable. In essence, we investigate how increasing ICT pene- tration can reduce CO2 emissions by establishing relevant mobile phone penetration and internet penetration thresholds above which ICT pe- netration reduces CO2 emissions.

5

The current paper deviates from the recent ICT literature which has fundamentally focused on, inter alia: Africa's information revolution from the perspective of production networks and technical regimes (Murphy and Carmody, 2015); economic prosperity (Levendis and Lee, 2013; Qureshi, 2013a); banking sector progress (Kamel, 2005); living standards (Chavula, 2013); externalities in welfare (Carmody, 2013; Qureshi, 2013b); life for all (Kivuneki et al., 2011) and sustainable development (Byrne et al., 2011) in developing nations. Hence, whereas human development and socioeconomic benefits associated with ICT have been substantially documented in the existing literature, we contribute to this stream of engaged literature by assessing how ICT can be consolidated for environmental sustainability. Accordingly, we have built on a comprehensive survey of green ICT literature (see Krishnadas and Radhakrishna, 2014) and more contemporary studies to position the study.

The above positioning is also an extension of recent technological foresight literature on the role of technology and innovation in sus- taining development and corporate outcomes, notably: the exploration of battery technology for grid-related energy storage (Versteeg et al., 2017); techno-organisational decarbonisation (Mazzanti and Rizzo, 2017); the importance of technology in consolidating the petroleum and petrochemical industry (Hossani et al., 2017); the relevance of environmental innovation in marketing capacity (Yu et al., 2017); lin- kages between unburnable fossil fuel, cumulative emissions and op- timal carbon tax (van de Ploeg and Rezai, 2017) and the connection of innovation ecosystems (Pombo-Juárez et al., 2017).

The inquiry we are engaging is relevant for policy because com- munication, which is essential in the coordination of good environ- mental governance, has often been neglected in policy circles (see Chemutai, 2009). According to the narrative, the United Nations Task Force on Environment and Human Settlements and environmental protection can be achieved fundamentally through effective coordina- tion with good communication tools.

The rest of the study is structured as follows. Section 2 engages the data and methodology. The empirical results are presented in Section 3 and Section 4 concludes with implications and future research direc- tions.

3 Fosu (2013) defines policy syndromes as situations that are detrimental to growth: ‘administered redistribution’, ‘state breakdown’, ‘state controls’, and ‘suboptimal inter temporal resource allocation’. Within the framework of this study, policy syndromes are considered as issues that merit policy action in order to achieve sustainable development.

4 According to the EKC hypothesis, in the long term, there is an inverted U-shaped

(footnote continued) relationship between per capita income and environmental degradation.

5 Accordingly, thresholds articulate levels of ICT penetration that are essential in re- ducing CO2 emissions. This objective does not suggest the existence of ICT thresholds that may be trivial if changes in some factors (e.g. desertification due to land clearing and overgrazing) contribute to decreasing CO2 emissions. Hence, the attainment of this ob- jective does not negate the relevance of other factors in decreasing CO2 emissions.

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

210

2. Data and methodology

This study investigates a panel of forty-four countries in SSA with data from African Development Indicators of the World Bank for the period 2000–2012.6 While we have provided the motivation for the scope of the inquiry in the introduction, the corresponding periodicity is contingent on constraints in data availability. The dependent vari- ables are: CO2 emissions per capita and CO2 emissions from liquid fuel consumption. In the corresponding assessment, a negative effect on the outcome variables indicates environmental protection or sustainability.

Consistent with recent ICT literature, the internet and the mobile phone are used as proxies that measure ICT (Penard et al., 2012; Asongu, 2014; Tony and Kwan, 2015; Amavilah et al., 2017; Tchamyou, 2016). Therefore, the internet penetration rate (per 100 people) and the mobile phone penetration rate (per 100 people) are used as ICT policy variables. In order to reduce concerns about variable omission bias, five control variables are used, namely: trade openness, Gross Domestic Product (GDP) growth, population growth, quality of education and regulation quality. Hence, the variables account for: globalization (trade openness), the market (GDP growth and population growth), institutional quality (regulation quality) and other human factors (quality of education).

Intuitively, we expect the quality of education and regulation quality to negatively affect the outcome variables, while trade openness and population are expected to have the opposite effect. The effect of GDP growth is ambiguous because it is contingent on a plethora of factors, notably, whether, GDP growth is broad-based or limited to a select number of industries and whether the fruits from economic prosperity are used to finance conditions for sustainable development. Conversely it may be seen that growing GDP may also contribute to environmental damages. It is also worthwhile to balance the expecta- tions in signs with the fact that country-specific features that are eliminated (by definition) from the Generalised Method of Moments (GMM) specifications may weight on the expected signs. Moreover, population growth and GDP growth dynamics could be high in Africa, which could generate some biases in the expected signs. This study assumes that primary education is a basic and necessary condition for ICT literacy. The full definitions of variables, corresponding summary statistics and correlation matrix can be found in Appendices 1, 2 and 3 respectively. While the correlation matrix is reported, issues of multi- collinearity are not relevant in interactive regressions (see Brambor et al., 2006). The Shapiro-Wilk normality test is also provided in Appendix 2 to show that the variables are not normally distributed. Hence, estimation techniques like Ordinary Least Squares are in- appropriate.

In the estimation process, we adopt a two-step GMM empirical strategy for five principal reasons: (i) the number of countries (44) is higher than the number of years in each country (13); (ii) the outcome variables are persistent because their correlation coefficients with their respective first lags are higher than the rule thumb threshold of 0.800; (iii) given that the GMM approach (by definition) builds on a panel data structure, cross-country differences are considered in the regressions; (iv) the estimation approach further controls for endogeneity by ac- counting for simultaneity in the exploratory variables through the process of instrumentation and use of time-invariant omitted variables and (v) inherent biases in the difference estimator are corrected with the system estimator.

In this study, the Arellano and Bover (1995) extension by Roodman

(2009a, 2009b) is employed because, relative to traditional GMM techniques (difference and system GMM approaches), it mitigates the proliferation of instruments (or restricts over-identification) and ac- counts for cross-sectional dependence (Love and Zicchino, 2006; Baltagi, 2008; Boateng et al., 2016).

The following equations in level (1) and first difference (2) sum- marise the standard system GMM estimation procedure in which the error term take a two-way error component form.

∑= + + + + +− =

−CO σ σ CO σ IC σ ICIC δ W υi t i t τ i t i t h

h h i t τ i t, 0 1 , 2 , 3 , 1

5

, , , (1)

∑

− = − + −

+ −

+ − + −

− − − −

−

=

− − −

CO CO σ CO CO σ IC IC

σ ICIC ICIC

δ W W υ υ

( ) ( )

( )

( ) ( )

i t i t τ i t τ i t τ i t i t τ

i t i t τ

h h h i t τ h i t τ i t i t τ

, , 1 , , 2 2 , ,

3 , ,

1

5

, , , , 2 , , (2)

= + +υ η ξ εi t i t i t, , (3)

+ = − + +− − −υ υ ξ ξ ε ε( ) ( )i t i t τ t t τ i t i t τ, , , , (4)

where, COi, t is a CO2 emissions indicator of country i at period t; σ0 is a constant, IC represents information and communication technology (mobile phone penetration and internet penetration); ICIC is the in- teraction between identical ICT variables (e.g. “mobile phones × mo- bile phones” or “internet × internet”); W is the vector of control vari- ables (trade, GDP growth, population growth, education and regulation quality); τ represents the coefficient of auto-regression, which is one for the specification, because of limited degrees of freedom; ξt is the time- specific constant; ηi is the country-specific effects (or factors that are particular to each country in the sample), υi, t is the two-way the dis- turbance term and εi, t is the error term. The definitions and sources of the variables can be found in Appendix 1.

We now devote space to discussing the identification and exclusion restriction strategy which is important for a sound GMM specification. In accordance with recent studies, we consider all the explanatory variables as predetermined or suspected endogenous and exclusively acknowledge the time invariant variables to exhibit strict exogeneity (Boateng et al., 2016; Asongu and Nwachukwu, 2016b). Within this framework, Roodman (2009b) has argued that it is not very feasible for time-invariant variables to be endogenous after a first difference.7

As concerns exclusion restrictions, consistent with the process of identification discussed above, the time invariant variables affect the outcome CO2 emissions indicators exclusively through the suspected endogenous channels. Furthermore, the criterion employed to assess the statistical relevance of this identification strategy is the Difference in Hansen Test (DHT) for instrument exogeneity. Hence, in order for the hypothesis of exclusion restriction to hold, the null hypothesis of the DHT should not be rejected. Failure to reject this hypothesis is an in- dication that CO2 emissions are influenced exclusively by the strictly exogenous variables, through the suspected endogenous indicators.

In the light of the above emphasis, the assumption of exclusion re- striction is confirmed in the reported results, when the null hypothesis of the DHT related to instrumental variables (IV) (year, eq(diff)) is not rejected. This process in GMM exclusion restriction is broadly con- sistent with the standard instrumental variable (IV) procedure, in which failure to reject the null hypothesis of the Sargan Overidentifying Restrictions (OIR) test is implies that strictly exogenous variables affect the outcome variable exclusively via the suspected endogenous variable channels (see Beck et al., 2003; Asongu and Nwachukwu, 2016c).

6 The 44 countries are: Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde, Central African Republic, Chad, Comoros, Congo Democratic Republic, Congo Republic, Cote d'Ivoire, Djibouti, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia, Niger, Nigeria, Rwanda, Sao Tome & Principe, Senegal, Seychelles, Sierra Leone, South Africa, Sudan, Swaziland, Tanzania, Togo, Uganda and Zambia.

7 Hence, the procedure for treating ivstyle (years) is ‘iv (years, eq(diff))’ whereas the gmmstyle is employed for predetermined variables (Asongu and De Moor, 2017).

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

211

3. Empirical results

Table 1 below presents the empirical results. There are two main sets of specifications pertaining to the two dependent variables. For either dependent variable, there are two subsets of specifications. Each sub-specification is characterised by ‘mobile phone’- and ‘internet’-re- lated regressions.

Four main information criteria are employed to assess whether the GMM models are valid.8 It is important to articulate two more points in the light of these information criteria. On the one hand, the second- order Arellano and Bond autocorrelation test (AR(2)) is more relevant as an information criterion than the first-order because some studies exclusively report the higher order with no disclosure of the first order (e.g. see Narayan et al., 2011; Asongu and Nwachukwu, 2016d). On the

other hand, the Hansen test is robust but weakened by instruments whereas the Sargan test is not robust but not weakened by instruments. “Weakened by instruments” implies that employing too many instru- ments increases the probability of a Type 2 error (i.e. failure to reject a false null hypothesis) or not rejecting the null hypothesis for the va- lidity of instruments. A logical way of addressing the conflict is to adopt the Hansen test and avoid the proliferation of instruments. Instrument proliferation is subsequently avoided by ensuring that the number of instruments in each specification is lower than the corresponding number of cross sections.

Net effects are computed to examine the overall effect of increasing ICT on environmental sustainability. For instance, in the third column of Table 1, the net effect from enhancing mobile phone penetration is 0.0025 ([−0.00002 × 24.428] + [0.003]). In the computation, the mean value of mobile phone penetration is 24.428, the unconditional effect of mobile phone penetration is 0.003, while the marginal effect from increasing mobile phones is −0.00002. This labelling of effects is consistent with recent literature on increasing ICT for development outcomes (Asongu and Le Roux, 2017).

The following findings can be established from Table 1. First, from the non-interactive regressions, ICT (i.e. mobile phones and the in- ternet) does not significantly affect CO2 emissions. Second, with regard to the interactive regressions, increasing ICT has a positive net effect on per capita CO2 emissions, while increasing mobile phones alone would have a negative net negative effect on CO2 emissions from liquid fuel

Appendix 1 Variable definitions.

Variables Signs Variable definitions (measurement) Sources

CO2 per capita CO2mtpc CO2 emissions (metric tons per capita) World Bank (WDI) CO2 from liquid fuel CO2lfcon CO2 emissions from liquid fuel consumption (% of total) World Bank (WDI) Mobile phones Mobile Mobile phone subscriptions (per 100 people) World Bank (WDI) Internet Internet Internet penetration (per 100 people) World Bank (WDI) Trade Openness Trade Imports plus Exports of goods and services (% of GDP) World Bank (WDI) GDP growth GDPg Gross Domestic Product (GDP) growth (annual %) World Bank (WDI) Population growth Popg Population growth rate (annual %) World Bank (WDI) Educational quality Educ Pupil teacher ratio in Primary Education World Bank (WDI) Regulation quality RQ “Regulation quality (estimate): measured as the ability of the government to formulate and implement sound policies and

regulations that permit and promote private sector development” World Bank (WDI)

Appendix 2 Summary statistics (2000–2012).

Mean SD Minimum Maximum Shapiro-Wilk Observations

CO2 per capita 0.901 1.820 0.016 10.093 0.709*** 567 CO2 from liquid fuel 78.880 23.092 0.000 100 0.842*** 567 Mobile phone penetration 24.428 28.535 0.000 147.202 0.809*** 525 Internet penetration 4.222 6.618 0.005 43.605 0.634*** 521 Trade openness 76.881 35.326 20.964 209.874 0.918*** 555 GDP growth 4.851 5.000 −32.832 33.735 0.853*** 567 Population growth 2.334 0.866 −1.081 6.576 0.958*** 529 Educational quality 43.784 14.731 12.466 100.236 0.975*** 425 Regulation quality −0.607 0.544 −2.238 0.983 0.983*** 530

Appendix 3 Correlation matrix (uniform sample size: 155).

CO2mtpc CO2lfcon Educ Internet GDPg Popg Trade RQ Mobile

1.000 −0.721 −0.369 0.411 −0.057 −0.611 −0.0004 0.593 0.558 CO2mtpc 1.000 0.246 −0.232 0.020 0.364 0.113 −0.366 −0.349 CO2lfcon

1.000 −0.444 0.104 0.515 −0.147 −0.515 −0.403 Educ 1.000 0.021 −0.580 0.256 0.536 0.718 Internet

1.000 0.074 −0.136 −0.140 −0.128 GDPg 1.000 −0.406 −0.624 −0.580 Popg

1.000 0.128 0.270 Trade 1.000 0.505 RQ

1.000 Mobile

8 “First, the null hypothesis of the second-order Arellano and Bond autocorrelation test (AR (2)) in difference for the absence of autocorrelation in the residuals should not be rejected. Second, the Sargan and Hansen over-identification restrictions (OIR) tests should not be sig- nificant because their null hypotheses are the positions that instruments are valid or not cor- related with the error terms. In essence, while the Sargan OIR test is not robust but not weakened by instruments, the Hansen OIR is robust but weakened by instruments. In order to restrict identification or limit the proliferation of instruments, we have ensured that instruments are lower than the number of cross-sections in most specifications. Third, the Difference in Hansen Test (DHT) for exogeneity of instruments is also employed to assess the validity of results from the Hansen OIR test. Fourth, a Fischer test for the joint validity of estimated coefficients is also provided” (Asongu and De Moor, 2017, p. 200).

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

212

consumption. Third, the “hysteresis” hypothesis is apparent because the absolute values of estimated lagged dependent variables are con- sistently between the intervals of zero and one, implying that past va- lues of CO2 emissions have a positive influence on future CO2 emissions. The information criterion for the “hysteresis” hypothesis is consistent with recent literature (Fung, 2009, p. 58; Prochniak and Witkowski, 2012a, p. 20; Prochniak and Witkowski, 2012b, p. 23). Most of the significant control variables have signs as expected and detailed before.

We have established that increasing ICT would have positive net impacts on CO2 emissions per capita. Fortunately, the corresponding marginal effects are negative. The implication of these negative mar- ginal effects is that at certain thresholds of ICT, the net impact can be changed from positive to negative. Accordingly, in order for these thresholds to make economic sense and have policy relevance, they must be within the range of corresponding data. Such a range denotes the minimum to maximum values provided by the summary statistics. In what follows, we first clarify the concept of threshold before

computing the thresholds corresponding to the net positive effects. The conception of threshold (also referred to as critical mass) re-

presents a point at which, further ICT penetration yields a net negative effect on CO2 per capita emissions. Hence, if the computed thresholds are within statistical range, policy makers can aim to increase ICT pe- netration beyond the established thresholds, in order to achieve the desired effect on environmental sustainability. This notion of threshold is consistent with the literature, notably: critical masses for appealing effects (Roller and Waverman, 2001; Batuo, 2015); requirements for inverted U-shaped and U-shaped patterns (see Ashraf and Galor, 2013), minimum conditions for desired impacts (Cummins, 2000) and essen- tial information sharing thresholds at which market power can be curbed in order to enhance financial intermediation efficiency (Asongu et al., 2016b).

The following are the corresponding critical thresholds of ICT pe- netration in reducing CO2 emissions: (i) 150(0.003/0.00002) for mobile phone penetration per 100 people and (ii) 42.5(0.017/0.0004) for

Table 1 Enhancing ICT and CO2 emissions.

Dependent variables: CO2 emissions

CO2 emissions (metric tons per capita) CO2 emissions from liquid fuel consumption (% of total)

Mobile Internet Mobile Internet

Constant 0.446*** 0.227*** 0.621*** 0.340*** −4.088 1.038 −6.464** −5.532** (0.000) (0.003) (0.000) (0.000) (0.107) (0.618) (0.036) (0.044)

CO2 per capita (−1) 0.897*** 0.934*** 0.924*** 0.929*** – – – – (0.000) (0.000) (0.000) (0.000)

CO2 from fuel (−1) – – – – 0.980*** 0.950*** 0.974*** 0.990*** (0.000) (0.000) (0.000) (0.000)

Mobile 0.00002 0.003*** – – 0.010 −0.047** – – (0.962) (0.000) (0.372) (0.038)

Mobile × Mobile – −0.00002*** – – – 0.0005*** – – (0.000) (0.000)

Internet – – −0.001 0.017*** – – 0.056* −0.120 (0.147) (0.000) (0.050) (0.105)

Internet × Internet – – – −0.0004*** – – – 0.004*** (0.000) (0.007)

Trade −0.0003 −0.0006** −0.001* −0.001*** 0.013 0.0006 0.030** 0.024* (0.222) (0.042) (0.051) (0.001) (0.166) (0.949) (0.029) (0.065)

GDP growth 0.0001 −0.002** −0.002 −0.002** −0.020 −0.053** −0.020 −0.013 (0.930) (0.049) (0.143) (0.044) (0.454) (0.024) (0.539) (0.641)

Population growth −0.072*** −0.075*** −0.116*** −0.072*** 1.189** 1.460*** 0.954** 0.803*** (0.000) (0.000) (0.000) (0.000) (0.023) (0.002) (0.019) (0.000)

Education −0.004*** −0.0008 −0.004*** −0.001 0.041* −0.005 0.087*** 0.069** (0.000) (0.536) (0.004) (0.228) (0.087) (0.793) (0.002) (0.012)

Regulation quality 0.005 −0.090* −0.050 −0.025 0.857 0.632 1.421 1.853** (0.861) (0.050) (0.254) (0.393) (0.426) (0.574) (0.132) (0.036)

Time effects (2000−2012) Yes Yes Yes Yes Yes Yes Yes Yes Net effects nsa 0.0025 nsa 0.0153 nsa −0.0347 nsa na AR(1) (0.109) (0.095) (0.105) (0.099) (0.006) (0.007) (0.004) (0.004) AR(2) (0.207) (0.199) (0.149) (0.153) (0.046) (0.044) (0.033) (0.032) Sargan OIR (0.000) (0.000) (0.000) (0.000) (0.826) (0.952) (0.954) (0.980) Hansen OIR (0.861) (0.390) (0.783) (0.361) (0.372) (0.610) (0.682) (0.829) DHT for instruments (a) Instruments in levels H excluding group (0.754) (0.726) (0.871) (0.828) (0.865) (0.971) (0.528) (0.636) Dif(null, H = exogenous) (0.763) (0.225) (0.566) (0.165) (0.163) (0.274) (0.652) (0.784) (b) IV (years, eq(diff)) H excluding group (0.743) (0.594) (0.978) (0.573) (0.140) (0.271) (0.287) (0.617) Dif(null, H = exogenous) (0.765) (0.213) (0.285) (0.197) (0.779) (0.914) (0.932) (0.850) Fisher 20560*** 84884*** 6308*** 287100*** 1143*** 2558*** 462.85*** 5779*** Instruments 36 40 36 40 36 40 36 40 Countries 44 44 43 43 44 44 43 43 Observations 339 339 334 334 339 339 334 334

*, **, ***: significance levels of 10%, 5% and 1% respectively. The different significant levels reported in the table are standard levels of significance used in the literature, notably: 10%, 5% and 1% significance levels. DHT: Difference in Hansen Test for Exogeneity of Instruments' Subsets. Dif: Difference. OIR: Over-identifying Restrictions Test. The significance of bold values is twofold. 1) The significance of estimated coefficients and the Fisher statistics. 2) The failure to reject the null hypotheses of: a) no autocorrelation in the AR(1) and AR(2) tests and b) the validity of the instruments in the Sargan and Hansen OIR tests. Whereas Eqs. (1) and (2) represent the standard system GMM specification, the results reported in Table 1 are based on the system GMM in the perspective of Roodman (2009a, 2009b). nsa: not specifically applicable because the regressions is non-interactive. na: not applicable because at least one estimated coefficient needed for the computation of net effects is not significant. The mean of mobile phone penetration is 24.428 while the mean of internet penetration is 4.222.

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

213

internet penetration per 100 people. Given that the mobile penetration range is between 0.000 and 147.202 per 100 people, only the computed threshold of internet penetration makes economic sense and has policy relevance. It follows that increasing the internet penetration rate to above 42.5 per 100 people, would mitigate CO2 per capita emissions and enhance environmental sustainability. Thus, the calculated policy threshold is the minimum level of ICT required, for the effect of ICT on CO2 emissions to be negative.

4. Concluding implications, caveats and future research directions

This study has examined how increasing ICT penetration in Sub- Saharan Africa (SSA) can contribute towards environmental sustain- ability by decreasing CO2 emissions. The empirical evidence is based on Generalised Method of Moments and forty-four countries for the period 2000–2012. ICT is measured with internet penetration and mobile phone penetration, while CO2 emissions per capita and CO2 emissions from liquid fuel consumption are used as proxies for CO2 emissions. The following findings have been established. First, from the non-inter- active regressions we conclude that increasing ICT (i.e. mobile phones and the internet) would not significantly affect CO2 emissions. Second, using the interactive regressions we conclude that increasing ICT has a positive net effect on CO2 emissions per capita, while increasing mobile phone penetration has a net negative effect on CO2 emissions from li- quid fuel consumption.

The negative net effects imply that ICT needs to be further con- solidated beyond certain thresholds, in order to achieve the desired negative net effect on CO2 emissions per capita. We have computed these corresponding thresholds and found them to be: 150 per 100 people for mobile phone penetration and 42.5 per 100 people for in- ternet penetration. While the former is not within practical range, and thus has no practical implication for policy, the latter is within achievable range and is of policy relevance. It follows that an internet penetration range of above 42.5 for every 100 people ensures en- vironmental sustainability by reducing CO2 emissions per capita.

The main policy implication of the findings is that ICT can be practically consolidated in order to mitigate CO2 emissions per capita. The fact that the threshold at which this is possible exclusively make economic sense for internet penetration is appealing, because the in- ternet can be used on many devices, such as the mobile phone. Hence, the internet is intuitively more relevant than the mobile phone, though both are also complementary.

Policy makers can leverage on the established findings in the post- 2015 development era, by addressing issues related to internet pene- tration, like affordability and lack of infrastructure. Furthermore, uni- versal access that encourages low pricing and broad coverage should also be considered. In order to sustainably mitigate CO2 per capita emissions, such internet penetration policy should be tailored to en- courage the adoption, access, interactions, effectiveness and reach of ICT. This policy agenda can go a long way towards making progress on some stalling environmental objectives, namely: Agenda 21 and Multilateral Environmental Agreements (MEAs). National schemes like the Rwanda Information Technology Authority (RITA) should be en- couraged in other African countries. RITA consolidates and coordinates the nation's resources in information technology (Chemutai, 2009).

The principal theoretical contribution of the study is that ICT re- duces information asymmetry associated with environmental pollution. Therefore, by decreasing informational rents that are associated with CO2 emissions, the theoretical relevance of ICT is consistent with the theoretical underpinnings of information sharing offices (e.g. private credit bureaus and public credit registries) in decreasing information asymmetry for banking intermediation efficiency (see Tchamyou and Asongu, 2017; Asongu et al., 2016c). In the light of this analogy, the theoretical motivation for consolidating financial intermediation effi- ciency through information sharing offices is consistent with the use of

ICT to tailor efficiency in CO2 emissions for environmental sustain- ability. Such efficiency is both intuitive and practical because ICT can be instrumental in sharing information to reduce unnecessary travelling and corresponding CO2 emissions and boost household and public management efficiency through the use of energy-saving ICT applica- tions.

The importance of ICT in dampening CO2 emissions is broadly consistent with the literature maintaining that ICT through network possibilities decrease cost/traffic per minute associated with economic operations (Gille et al., 2002; Esselaar et al., 2007; Gutierrez et al., 2009; Gilwald and Stork, 2008). Moreover, further effects of ICT on CO2 emissions might unveil in the future. It is essentially because these ef- fects can be further increased in the future, that the positioning of the inquiry is based on how increasing ICT can affect CO2 emission re- duction.

Three caveats to the study are worth highlighting: First, the growing share of non-ICT sectors (such as technology involved in refrigeration and air conditioning) in GDP can also exert as much influence on CO2 emission, as ICT-related factors. However, we restrict our study to ICT factors, owing to reasons already discussed in the Introduction.

Second, a methodological handicap is the omission of the CO2 emissions growth caused by anthropogenic deforestation and deserti- fication in SSA. Within this framework some variables may need ad- justment. For instance, the decrease of CO2 absorbing capacity (caused by destroying the rainforests) could be calculated as a CO2 equivalent and added to the CO2 production from the burning of fossil fuels. Unfortunately, the variables used in the study were directly obtained from World Bank Development Indicators and no further transforma- tions were done because we do not have the corresponding data with which to engage such transformations.

Third, the impact of ICT should intuitively be highly contingent on state policy-orientation. Unfortunately country-specific effects are not relevant in the modelling exercise because they are eliminated to limit the endogeneity which is associated with the correlation between the lagged endogenous variable and the error term. Moreover, African countries are treated as a homogenous group and the results are ap- plicable to all the sampled countries for the same reason that country- specific effects are not engaged in the modelling framework. Hence, controlling for fundamentals like income levels, political stability and legal origins is also not feasible. While one might think that using sub- samples or these fundamentals (i.e. income levels, political stability and legal origins) could account for more idiosyncratic and group effects, the sub-sample estimations do not pass the post-estimation diagnostic tests because of instrument proliferation. It is important to note that the N > T condition is required for the application of the GMM technique. Sub-sampling decreases N and increases instrument proliferation. Moreover, even if these fixed effects or homogenous groups were to be considered in the modelling exercise, they are by definition eliminated in the GMM approach because they are correlated with the lagged de- pendent variable.

Future studies can build on these caveats in order to advance the extant knowledge, by employing alternative estimation techniques to assess whether the established findings withstand empirical scrutiny within country-specific settings and homogenous groups (e.g. income levels and legal origins). This would be necessary for more targeted policy implications.

Acknowledgement

The authors are indebted to the reviewers and editor for their constructive comments.

WDI: World Bank Development Indicators. CO2: Carbon dioxide. S.D: Standard Deviation. CO2: Carbon dioxide. ***: 1% significance

level. CO2mtpc: CO2 emissions (metric tons per capita). CO2lfcon: CO2

emissions from liquid fuel consumption (% of total). Educ: Quality of

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

214

primary education. Internet: Internet penetration. GDPg: GDP growth. Popg: Population growth. RQ: Regulation Quality. Mobile: Mobile Phone penetration.

References

Afutu-Kotey, R.L., Gough, K.W., Owusu, G., 2017. Young entrepreneurs in the mobile telephony sector in Ghana: from necessities to aspirations. J. Afr. Bus. http://dx.doi. org/10.1080/15228916.2017.1339252.

Akbostanci, E., Turut-Asi, S., Tunc, G.I., 2009. The relationship between income and environment in Turkey: is there an environmental Kuznets curve? Energ Policy 37 (3), 861–867.

Akpan, G.E., Akpan, U.F., 2012. Electricity consumption, carbon emissions and economic growth in Nigeria. Int. J. Energy Econ. Policy 2 (4), 292–306.

Amavilah, A., Asongu, S.A., Andrés, A.R., 2017. Effects of globalization on peace and stability: implications for governance and the knowledge economy of African coun- tries. Technol. Forecast. Soc. Chang. 122 (C), 91–103.

Ang, J.B., 2007. CO2 emissions, energy consumption, and output in France. Energ Policy 35 (10), 4772–4778.

Anyangwe, E. (2014). Without energy could Africa's growth run out of steam?, The Guardian, http://www.theguardian.com/global-development-professionals- network/2014/nov/24/energy-infrastructure-clean-cookstoves-africa (Accessed: 08/ 09/2015).

Apergis, N. and J. Payne, J. E., (2009). “CO2 emissions, energy usage, and output in Central America”, Energ Policy, 37(8), pp. 3282–3286.

Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of error components models. J. Econ. 68 (1), 29–52.

Ashraf, Q., Galor, O., 2013. The Out of Africa hypothesis, human genetic diversity, and comparative economic development. Am. Econ. Rev. 103 (1), 1–46.

Asongu, S.A., 2013. How has mobile phone penetration stimulated financial development in Africa? J. Afr. Bus. 14 (1), 7–18.

Asongu, S.A., 2014. Globalization, (fighting) corruption and development: how are these phenomena linearly and nonlinearly related in wealth effect? J. Econ. Stud. 41 (3), 346–369.

Asongu, S.A., De Moor, L., 2017. Financial globalisation dynamic thresholds for financial development: evidence from Africa. Eur. J. Dev. Res. 29 (1), 192–212.

Asongu, S.A., El Montasser, G., Toumi, H., 2016a. Testing the relationships between energy consumption, CO2 emissions, and economic growth in 24 African countries: a panel ARDL approach. Environ. Sci. Pollut. Res. 23 (7), 6563–6573.

Asongu, S.A., Le Roux, S., 2017. Enhancing ICT for inclusive human development in Sub- Saharan Africa. Technol. Forecast. Soc. Chang. 118 (May), 44–54.

Asongu, S.A., Nwachukwu, J.C., 2016a. The mobile phone in the diffusion of knowledge for institutional quality in Sub Saharan Africa. World Dev. 86 (October), 133–147.

Asongu, S. A., and Nwachukwu, J. C., (2016b). The role of governance in mobile phones for inclusive human development in Sub-Saharan Africa, Technovation, 55-56 (September–October), pp. 1-13.

Asongu, S.A., Nwachukwu, J.C., 2016c. Foreign aid and governance in Africa. Int. Rev. Appl. Econ. 30 (1), 69–88.

Asongu, S.A., Nwachukwu, J.C., 2016d. Revolution empirics: predicting the Arab Spring. Empir. Econ. 51 (2), 439–482.

Asongu, S.A., Nwachukwu, J., Tchamyou, S.V., 2016c. Information asymmetry and fi- nancial development dynamics in Africa. Review of Development Finance 6 (2), 126–138.

Asongu, S.A., Le Roux, S., Tchamyou, V.S., 2016b. Essential information sharing thresholds for reducing market power in financial access: a study of the African banking industry. In: African Governance and Development Institute Working Paper No. 16/036, (Yaoundé).

Asongu, S.A., Rangan, G., 2015. Trust and quality of growth. Econ. Bull. 36 (3), 1854–1867.

Baltagi, B.H., 2008. Forecasting with panel data. J. Forecast. 27 (2), 153–173. Batuo, M.E., 2015. The role of telecommunications infrastructure in the regional eco-

nomic growth of Africa. Journal of Development Areas 49 (1), 313–330. Beck, T., Demirgüç-Kunt, A., Levine, R., 2003. Law and finance: why does legal origin

matter? J. Comp. Econ. 31 (4), 653–675. Begum, R.A., Sohag, K., Abdullah, S.M.S., Jaafar, M., 2015. CO2 emissions, energy con-

sumption, economic and population growth in Malaysia. Renew. Sust. Energ. Rev. 41 (January), 594–601.

Boateng, A., Asongu, S. A., Akamavi, R., and Tchamyou, V. S., (2016). “Information Asymmetry and Market Power in the African Banking Industry”, African Governance and Development Institute Working Paper No. 16/032, Yaoundé.

Bölük, G., Mehmet, M., 2015. The renewable energy, growth and environmental Kuznets curve in Turkey: an ARDL approach. Renew. Sust. Energ. Rev. 52 (December), 587–595.

Brambor, T., Clark, W.M., Golder, M., 2006. Understanding interaction models: im- proving empirical analyses. Polit. Anal. 14 (1), 63–82.

Byrne, E., Nicholson, B., Salem, F., 2011. Information communication technologies and the millennium development goals. Inf. Technol. Dev. 17 (1), 1–3.

Carmody, P., 2013. A knowledge economy or an information society in Africa? The in- tegration and the mobile phone revolution. Inf. Technol. Dev. 19 (1), 24–39.

Chavula, H.K., 2013. Telecommunications development and economic growth in Africa. Inf. Technol. Dev. 19 (1), 5–23.

Chemutai, B., (2009). “Achieving Effective National Environmental Governance in Africa”, I SS Today, https://issafrica.org/iss-today/achieving-effective-national- environmental-governance-in-africa (Accessed: 07/04/2017).

Costantini, M., Lupi, C., 2005. Stochastic convergence among European economies. Econ. Bull. 3 (38), 1–17.

Cummins, J., 2000. Language, Power and Pedagogy: Bilingual Children in the Crossfire. Multilingual Matters, Clevedon, England.

Diao, X.D., Zeng, S.X., Tam, C.M., Tam, V.W.Y., 2009. EKC analysis for studying economic growth and environmental quality: a case study in China. J. Clean. Prod. 17 (5), 541–548.

Esselaar, B., Gilwald, A., Stork, C., 2007. Towards an Africa E-index: Telecommunications Sector Performance in 16 African Countries. Research ICT Africa. www. researchICTafrica.net.

Esso, L.J., 2010. Threshold cointegration and causality relationship between energy use and growth in seven African countries. Energy Econ. 32 (6), 1383–1391.

Fosu, A.K., 2015. Growth, inequality and poverty in Sub-Saharan Africa: recent progress in a global context. Oxf. Dev. Stud. 43 (1), 44–59.

Fosu, A., 2013. Growth of African economies: productivity, policy syndromes and the importance of institutions. J. Afr. Econ. 22 (4), 523–551.

Fung, M.K., 2009. Financial development and economic growth: convergence or diver- gence? J. Int. Money Financ. 28 (1), 56–67.

Gille, L., Noumba Um, P, Rudel, C., and Simon, L., (eds) (2002) “A Model for Calculating Interconnection Costs in Telecommunications”. The World Bank. http://www.ppiaf. org/sites/ppiaf.org/files/publication/WB%20-%20Model%20Calculating %20Interconnection%20Costs%20Telecoms%202004.pdf (Accessed: 27/11/2014).

Gilwald, A., Stork, C., 2008. “Towards Evidence-based ICT Policy and Regulation: ICT Access and Usage in Africa”, Volume 1 Policy Paper Two. Research ICT Africa. www. researchICTafrica.net.

Gutierrez, L.H., Lee, S., Virto, L.R., 2009. Market concentration and performance in mobile markets in Africa and Latin America. OECD Development Center (mimeo).

He, J., Richard, P., 2010. Environmental Kuznets curve for CO2 in Canada. Ecol. Econ. 69 (5), 1083–1093.

Hossani, H., Silva, E.S., Al Kaabi, A.M., 2017. The role of innovation and technology in sustaining the petroleum and petrochemical industry. Technol. Forecast. Soc. Chang. 119 (C), 1–17.

Huxster, J.K., Uribe-Zarain, X., Kempton, W., 2015. Undergraduate understanding of climate change: the influences of college major and environmental group member- ship on survey knowledge scores. J. Environ. Educ. 46 (3), 149–165.

Jumbe, C.B., 2004. Cointegration and causality between electricity consumption and GDP: empirical evidence from Malawi. Energy Econ. 26 (1), 61–68.

Kifle, T., 2008. Africa hit hardest by Global Warming despite its low Greenhouse Gas Emissions, Institute for World Economics and. Int. Manag Working Paper No. 108. http://www.iwim.uni-bremen.de/publikationen/pdf/b108.pdf (Accessed: 08/09/ 2015).

Kamel, S., 2005. The use of information technology to transform the banking sector in developing nations. Inf. Technol. Dev. 11 (4), 305–312.

Kivuneki, F.N., Ekenberg, L., Danielson, M., Tusubira, F.F., 2011. Perceptions of the role of ICT on quality of life in rural communities in Uganda. Inf. Technol. Dev. 21 (1), 61–80.

Krishnadas, N., Radhakrishna, R.P., 2014. Green information technology: literature re- view and research domains. Journal of Management Systems 24 (1), 57–79.

Levendis, J., Lee, S.H., 2013. On the endogeneity of telecommunications and economic growth: evidence from Asia. Inf. Technol. Dev. 19 (1), 62–85.

Love, I., Zicchino, L., 2006. Financial development and dynamic investment behaviour: evidence from panel VAR. The Quarterly Review of Economics and Finance 46 (2), 190–210.

Mazzanti, M., Rizzo, U., 2017. Diversely moving towards a green economy: techno-or- ganisational decarbonisation trajectories and environmental policy in EU sectors. Technol. Forecast. Soc. Chang. 115 (C), 111–116.

Mehrara, M., 2007. Energy consumption and economic growth: the case of oil exporting countries. Energ Policy 35 (5), 2939–2945.

Menyah, K., Wolde-Rufael, Y., 2010. Energy consumption, pollutant emissions and eco- nomic growth in South Africa. Energy Econ. 32 (6), 1374–1382.

Murphy, J.T., Carmody, P., 2015. Africa's Information Revolution: Technical Regimes and Production Networks in South Africa and Tanzania, RGS-IBG Book Series. Wiley, Chichester, UK.

Narayan, P.K., Mishra, S., Narayan, S., 2011. Do market capitalization and stocks traded converge? New global evidence. J. Bank. Financ. 35 (10), 2771–2781.

Odhiambo, N.M., 2009a. Electricity consumption and economic growth in South Africa: a trivariate causality test. Energy Econ. 31 (5), 635–640.

Odhiambo, N.M., 2009b. Energy consumption and economic growth nexus in Tanzania: an ARDL bounds testing approach. Energ Policy 37 (2), 617–622.

Ozturk, I., Acaravci, A., 2010. CO2 emissions, energy consumption and economic growth in Turkey. Renew. Sust. Energ. Rev. 14 (9), 3220–3225.

Penard, T., Poussing, N., Yebe, G.Z., Ella, P.N., 2012. Comparing the determinants of internet and cell phone use in Africa: evidence from Gabon. Commun. Strateg. 86 (2), 65–83.

Pombo-Juárez, L., Könnölä, T., Miles, I., Saritas, O., Schartinger, D., Amanatidou, E., Giesecke, S., 2017. Wiring up multiple layers of innovation ecosystems: contempla- tions from personal Health Systems Foresight. Technol. Forecast. Soc. Chang. 115 (C), 278–288.

Prochniak, M., Witkowski, B., 2012a. Beta convergence Stability among “Old” and “New” EU Countries: The Bayesian Model Averaging Perspective. Warsaw School of Economics.

Prochniak, M., Witkowski, B., 2012b. Real Economic Convergence and the Impact of Monetary Policy on Economic Growth of the EU Countries: The Analysis of Time Stability and the Identification of Major Turning Points Based on the Bayesian Methods. Warsaw School of Economics.

Qureshi, S., 2013a. What is the role of mobile phones in bringing about growth? Inf.

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

215

Technol. Dev. 19 (1), 1–4. Qureshi, S., 2013b. Networks of change, shifting power from institutions to people: how

are innovations in the use of information and communication technology trans- forming development? Inf. Technol. Dev. 19 (2), 97–99.

Roller, L.-H., Waverman, L., 2001. Telecommunications infrastructure and economic development: a simultaneous approach. Am. Econ. Rev. 91 (4), 909–923.

Roodman, D., 2009a. A note on the theme of too many instruments. Oxf. Bull. Econ. Stat. 71 (1), 135–158.

Roodman, D., 2009b. How to do xtabond2: an introduction to difference and system GMM in Stata. Stata J. 9 (1), 86–136.

Shurig, S. (2015). Who will fund the renewable solution to the energy crisis?, The Guardian, http://www.theguardian.com/global-development-professionals- network/2014/jun/05/renewable-energy-electricty-africa-policy (Accessed: 08/09/ 2015).

Tchamyou, V.S., 2016. The role of knowledge economy in African business. J. Knowl. Econ. http://dx.doi.org/10.1007/s13132-016-0417-1.

Tchamyou, V.S., Asongu, S.A., 2017. Information sharing and financial sector develop- ment in Africa. J. Afr. Bus. 18 (1), 24–49.

Tony, F.L., Kwan, D.S., 2015. African entrepreneurs and international coordination in petty businesses: the case of low-end mobile phones sourcing in Hong Kong. J. Afr. Bus. 15 (1–2), 66–83.

Yu, W., Ramanathan, R., Nath, P., 2017. Environmental pressures and performance: an analysis of the roles of environmental innovation strategy and marketing capability. Technol. Forecast. Soc. Chang. 117 (C), 160–169.

van de Ploeg, F., Rezai, A., 2017. Cumulative emissions, unburnable fossil fuel, and the optimal carbon tax. Technol. Forecast. Soc. Chang. 116 (C), 216–222.

Versteeg, T., Baumann, M.J., Weil, M., Moniz, A.B., 2017. Exploring emerging battery

technology for grid-connected energy storage with Constructive Technology Assessment. Technol. Forecast. Soc. Chang. 115 (C), 99–110.

Simplice A. Asongu holds a PhD from Oxford Brookes University and is currently the Lead Economist and Director of the African Governance and Development Institute, Cameroon. He is a Research Associate at the Africa Growth Institute (Cape Town, South Africa), a researcher at Oxford Brookes University (Oxford, UK), PhD Supervisor at Covenant University (Ota, Nigeria), Research Associate at the University of Buea (Buea, Cameroon), MBA Supervisor at Management College of Southern Africa (Durban, South Africa) and Research Associate at the Development Finance Centre of the Graduate School of Business at the University of Cape Town.

Sara Le Roux has a PhD in Economics from the University of Exeter. She is interested in the theoretical and experimental analysis of individuals' perception of ambiguity and decision choices made by individuals in the presence of ambiguity. In addition, her sec- ondary research considers economic development in countries in Asia and Africa; in particular, financial stability, and how this impacts on poverty mitigation, gender equality, education and welfare.

Nicholas Biekpe is Professor of Development Finance and Econometrics at the Graduate School of Business at University of Cape Town, South Africa. He is also the Director of the Development Finance Centre at the University. Nicholas is also the President of Africagrowth Institute and Executive Chairman of the Chartered Institute of Development Finance. Professor Biekpe is the Executive Editor of African Finance Journal, Editor-in- Chief of Review of Development Finance and Executive Editor of Africagrowth Agenda.

S.A. Asongu et al. Technological Forecasting & Social Change 127 (2018) 209–216

216