Can African countries efficiently build their economies on renewable energy?

Can African countries efficiently build their economies on renewable energy?

Renewable and Sustainable Energy Reviews 54 (2016) 161–173 Contents lists available at ScienceDirect Renewable and Sustainable Energy Reviews journa...

599KB Sizes 3 Downloads 115 Views

Renewable and Sustainable Energy Reviews 54 (2016) 161–173

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

Can African countries efficiently build their economies on renewable energy? Presley K. Wesseh Jr, Boqiang Lin n Collaborative Innovation Center for Energy Economics and Energy Policy, China Institute for Studies in Energy Policy, Xiamen University, Fujian 361005, PR China

art ic l e i nf o

a b s t r a c t

Article history: Received 8 January 2015 Received in revised form 15 June 2015 Accepted 18 September 2015

The translog production model is estimated to provide insights on the effectiveness of renewable energy for Africa. The analysis shows that capital, labor, renewable energy and nonrenewable energy drive output in African countries; with renewable energy being a higher driver of growth than the conventional fossil fuels over the sample period. This finding is reflective of the fact that renewable sources like wind, hydro and solar account for a greater share of power generation in most African countries. Output elasticities computed for both energy types reaffirm the results and suggest that Eastern and Central African countries are more renewable energy dependent than the other three regions. In addition, technological progress is driven mainly by the efficiency with which various factors and energy inputs are used. While Africa has great potential of facing out conventional fossil energy, the discussion provided in this study suggests that such transition is limited in practice due to issues of scale, economics and sitting problems. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Renewable energy Nonrenewable energy Economic growth Africa

Contents 1. 2. 3. 4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Renewable energy consumption and economic growth in Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Review of relevant studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. The data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. The model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Production structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Estimation technique. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Renewable energy, nonrenewable energy, technological progress and economic growth in African countries . . . . . . . . . . . . . . . . . . . . 5.2. Marginal product of renewable and nonrenewable energy with respect to output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. State of technological progress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. Substitution elasticities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Renewable versus nonrenewable energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. Economics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. Sitting problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

161 162 163 165 165 166 167 168 168 168 169 169 169 169 170 171 171 172 172 172

1. Introduction n Corresponding author at: Collaborative Innovation Center for Energy Economics and Energy Policy, China Institute for Studies in Energy Policy, Xiamen University, Fujian, 361005, PR China. Tel.: +86 5922186076; fax: +86 5922186075. E-mail addresses: [email protected], [email protected] (B. Lin).

http://dx.doi.org/10.1016/j.rser.2015.09.082 1364-0321/& 2015 Elsevier Ltd. All rights reserved.

Nowhere in the world is the need for energy so critical than in Sub-Saharan Africa. At the same time, carbon dioxide emission from the use of petroleum in Africa has witnessed increasing trend

162

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

4,000 3,500

Million Metric Tons

3,000 2,500 2,000 1,500 1,000 500 0 1980

1985 Africa Middle East

1990 Asia & Oceania North America

1995

2000

Central & South America

2005 Eurasia

2010 Europe

Fig. 1. Total African CO2 emissions from the consumption of petroleum compared with rest of the world. Source: EIA.

since 1980 (Fig. 1). There are fears that Africa, with its low capital, would be more vulnerable to extreme weather conditions. For instance, unlike richer nations, it would be difficult for African countries to pay for air conditioning, import food from far away regions, and build out of the rage of rising waters. To substantiate these claims, the African Development Bank (AfDB) has highlighted the transition to green growth as the key focus of its new Ten-Year Strategy (2013–2022) in order to promote a more resource efficient and sustainable development. The point is climate change would still persist if African countries reduce emissions and other countries do not. More besides, Africa would be able to reduce the effects of climate change if accelerated wealth creation coming at the hands of cheaper and more stable production recipe (like fossil energy) is allowed. State-of-the-art discussions on how to achieve green growth pathways seem to concentrate mainly on the development of renewable energy [3]. Notwithstanding, developing renewable energy would sound much reasonable if such opportunities are able to end the energy poverty stimulate economic growth and ensure environmental sustainability in Africa. The present study therefore aims at achieving the following: first, develop reliable estimates of the economic impact of both renewable and nonrenewable energy, classical factors and technological progress in African countries using country-level panel data for 34 African countries. Second, produce estimates of the substitution potential between renewable and non-renewable energy. Third, discuss African energy challenges in the context of the empirical findings. Finally, advance relevant policy implication for African energy future. Given the low energy content, sitting problems and capacity factors inherent in most renewable energy sources (wind and solar for instance), a substantial amount of backups from conventional fuels are required to maintain supply stability, thus, offsetting the potential environmental gains in most cases.1 A study of this nature is not only useful to determine whether economic development in African countries has been affected by the inherent shortfalls in renewable energy sources, but also sheds light on the potential of shifting towards cleaner energy sources while maintaining the ability to fuel the economy and mitigating greenhouse gas emissions. Furthermore, Africa is still at the early stage of industrialization and modernization. As such, this study argues that environmental concerns should be long-run driven while ‘ending the energy poverty’ and ‘boasting real GDP’ should 1 A discussion on these issues and a more detailed cost-benefit analysis of renewable energy shall be the domain of future research.

be Africa's key priorities. In other words, environmental protection cannot come at the cost of growth and African countries will need to take a more deliberate action in promoting their competitiveness. Indeed, the originality and scientific contribution of this study adds value to the literature. First, this study is the first of its kind approach to renewable energy – economic growth potential for such a large group of African countries.2 Second, substitution elasticities between renewable energy and non-renewable energy have never been estimated for Africa before despite the significant implications these may have on Africa's economic development [37,75]. Finally, the applied methodology is novel in the energy consumption – economic growth literature; although the approach has been recently applied by very few studies to estimate energy substitution effects.3 For we use the translog production model, which in applied production analysis, is considered to be the most flexible functional form. The remainder of the paper proceeds as follows. Section 2 presents an initial analysis of renewable energy consumption and economic activities in Africa. Section 3 reviews the relevant literature for Africa. Section 4 presents the data and documents technical details of the applied methodology. Section 5 reports the estimated results. Section 6 discusses Africa's clean development possibilities in the context of the empirical findings and Section 7 draws the conclusions advancing relevant policy suggestions.

2. Renewable energy consumption and economic growth in Africa The African continent constitutes 54 countries, with each one facing different levels of economic growth and energy sector challenges. On the whole, African countries consume only 25% of the global average energy per capita [25]. Fossil fuels, biomass and hydro are the most predominant sources of energy in Africa (Fig. 2). It is now becoming a general consensus among policy makers that large-scale deployment of renewable energy technologies, due to recent technological advancement and cost 2 Work by Aissa et al. [4] is a little close. The authors employed panel error correction model to investigate the relationship between trade, renewable energy consumption and output in 11 African countries and found a positive long-run relationship. 3 To the best of the authors' knowledge, only five studies have applied the translog production model to the investigation of problems in the energy economics literature namely: [33,35,38,65,66,75].

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

163

Fig. 2. Share of total primary energy supply in 2009. Source: [3].

Table 1 Renewable energy potential across Africa. Source: [3]. Region

Wind (TW h/yr)

Solar (TW h/yr)

Biomass (EJ/yr)

Geothermal (TW h/yr)

Hydro (TW h/yr)

East Central North South West Total Africa

2000–3000

30000

1–16

3000–4000 16 0–7 5000–7000

50000–60000 25000–30000 50000 155000–170000

20–74 49–86 8–15 3–101 2–96 82–372

578 1057 78 26 105 1844

reductions, provide a cost-effective recipe for sustainable growth in African countries. Indeed, documentations have shown that Africa has a huge potential for renewable power generation of which less than one quarter of this potential is actually utilized. According to AfDB, wind power alone has a potential that is five times more that Africa's current installed power generation capacity. Nevertheless, there are severe sitting problems as different countries and regions in Africa have different mix of resource endowments. For instance, East Africa accounts for almost all the geothermal resources in Africa while Central Africa enjoys large share of the continent's hydro resources. More abundant solar and wind resources are located in the North and West and in the North and East respectively (Table 1). To set the stage for a comprehensive analysis, the total renewable electricity consumption in Africa has been plotted against economic growth. A cursory look at Fig. 3 seems to suggest no evidence of close co-movements between the two variables. For instance, any fall in renewable electricity consumption is not accompanied by a corresponding fall in real GDP. Notwithstanding, this observation is not sufficient to draw the conclusion that renewable energy consumption has no role in explaining Africa's economic growth. Certainly, several variables including nonrenewable energy, capital, labor, adjustment costs and technology for instance, contribute to the movement in real GDP. Hence, an empirical analysis is necessary to test the definite impact of renewable energy.

3. Review of relevant studies The literature on energy consumption – economic growth nexus is extensive with studies for both developed and developing

1–16

countries. In his recent review of the energy consumption – economic growth literature, Omri [51] documented various methods that have been applied in the literature including Granger causality; Sim technique; Cointegration; ECM; VECM; Multivariate VAR model; ARDL bonds test and Toda–Yamamoto causality test. The bootstrap empirical distribution has also been applied by few authors recently. As mentioned previously, these different modeling techniques coupled with different proxy variables for energy consumption have led to mixed results. In what follows, a summary of the relevant studies grouped according to different proxy variables and regions are presented. For a detailed review of these studies, interested readers are referred to Omri [51]. A number of studies have investigated the effect of aggregate energy consumption on economic growth in different countries. Omri [51] review of the literature shows that 29% of the studies in this category provide support for the growth hypothesis; 27% support the feedback hypothesis; 23% show evidence of the conservation hypothesis and 21% conform to the neutrality hypothesis. A summary of these studies including the applied methodologies can be found in Omri [51]. There are also studies that have used electricity consumption as a proxy for energy of which 40%, 33% and 27% shed light on the growth hypothesis, feedback hypothesis and conservation hypothesis respectively. No evidence of the neutrality hypothesis was found for country-specific electricity consumption – economic growth nexus studies. The documented publications in this category include Ramcharran [57] on Jamaica; Yang [83] and Hu and Lin [24] on Taiwan; Ghosh [20] on India; Shiu and Lam [63], Yuan et al. [88], Yuan et al. [89] and Zhang and Cheng [91] on China; Altinay and Karagol [9], Halicioglu [21], and Acaravci [2] on Turkey; Yoo [86] on Korea; Narayan and Smyth [46] on Australia; Yoo and Kim [87] on Indonesia; Zachariadis and Pashouortidou [90] on Cyprus; Alam et al. [7] and Mozumder and Marathe [45] on

164

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

1,000 100

Real GDP

800 80 600 60 400 40

Total Renewable Electricity Consumption

120

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

200

Real GDP (Constant Billion Dollars)

Total Renewable Electricity Consumption (Billion Kilowatthour)

Fig. 3. Renewable electricity consumption and real GDP in Sub-Saharan Africa. Source: EIA; World Bank.

Table 2 Energy consumption and growth studies for Africa Study

Country

Period

Method

Conclusions

Ebohon [16] Jumbe [26] Wolde-Rufael [78] Wolde-Rufael [79] Akinlo [5] Wolde-Rufael [80] Odhiambo [48] Odhiambo [49] Akinlo [6] Menyah and Wolde-Rufael [81] Esso [18] Odhiambo [50] Ouedraogo [52] Kebede et al. [28] Kouakou [32] Eggoh et al. [17] Al-mulali and Sab [8] Wesseh and Zoumara [73] Tamba et al. [69] Kahsai et al. [27] Richard [58] Wandji [72] Solarin and Shahbaz [67] Behmiri and Manso [11] Fuinhas and Marques [19] Ouedraogo [53] Kumar and Kumar [30] Bélaïd and Abderrahmani [12] Mensah [42] Lin and Wesseh [37] Aissa et al. [4]

Nigeria and Tanzania Malawi 19 African countries 17 African countries 11 African Countries 17 African countries South Africa Tanzania Nigeria South Africa

1960–1984 1970–1999 1971–2001 1971–2001

Granger causality Error Correction Toda and Yamamoto Toda and Yamamoto ARDL Toda and Yamamoto

Feedback hypothesis Feedback hypothesis All four hypothesis All four hypothesis All four hypothesis Growth hypothesis Feedback hypothesis Growth hypothesis Growth hypothesis Growth hypothesis

Burkina Faso 20 African countries Cote d'Ivoire 21 African countries 30 African countries Liberia Cameroon 12 African countries Cameroon Angola 23 African countries Algeria and Egypt ECOWAS countries South Africa and Kenya Algeria 6 emerging African economies South Africa 11 African countries

Bangladesh; Ho and Siu [23] on Hong Kong; Hu and Lin [24] on Fiji Islands; Aqeel and Butt [10], Shahbaz and Feridun [61] and Shahbaz and Lean [62] on Pakistan; Abosedra et al. [1] on Lebanon; Bowden and Payne [14] on the USA; Chandran et al. [15] and Tang [70] on Malaysia; Lorde et al. [39] on Barbados and Shahbaz et al. [60] on Portugal. Few publications exist which examine the nuclear energy – growth nexus. Results from these studies show 60% in favor of the neutrality hypothesis and 40% for the growth hypothesis; with no evidence of the feedback and conservation hypotheses [51]. These include Yoo and Jung [85] on Korea; Payne and Taylor [55] and Menyah and Wolde-Rufael [43] on the USA and Wolde-Rufael [81] on India. Contributions which focus exclusively on renewable energy – economic growth nexus are few and suggest 40%, 40% and 20% in

1971–2006 1971–2006 1980–2006 1965–2006

ARDL Cointegration ARDL

1968–2003

ARDL

1971–2008 1970–2006 1980–2008 1980–2008 1975–2008

Error Correction Panel Cointegration Panel Model Bootstrap tests Error Correction

1971–2008

Hidden cointegration Error Correction ARDL Panel causality

1971–2009 1985–2011 1965–2010 1980–2008 1971–2010 1971–2010 1980–2008

Panel Cointegration ARDL Error Correction ARDL Bootstrap tests Panel Error Correction

Feedback hypothesis Growth hypothesis Feedback hypothesis Feedback hypothesis Growth hypothesis Feedback hypothesis Feedback hypothesis Conser. and growth Growth and Neutrality Feedback hypothesis Feedback hypothesis Feedback hypothesis Growth hypothesis Growth hypothesis Feedback hypothesis Growth hypothesis Growth hypothesis Growth hypothesis

favor of the neutrality, conservation and growth hypotheses respectively. No evidence was found for the feedback hypothesis. These include Payne [54], Sari et al. [59], Menyah and WoldeRufael [81], Payne [56] and Yildirim et al. [84] on the USA [51]. So far, only country-specific studies focusing on regions other than Africa have been presented. To put the study into proper perspective, publications which direct attention to African countries, be it country-specific or multi-country, are documented. There are a number of country-specific studies which provide support for the feedback hypothesis in African countries. In other words, these studies suggest bidirectional Granger causality between energy consumption and economic growth. In contrast, some studies on Africa have supported the growth hypothesis. This hypothesis asserts that energy consumption complements capital and labor as important factors in the production process.

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

This means that energy is crucial for growth as the economy is energy dependent. As a result, energy conservation policies may have negative effects on real GDP. The vast majority of these studies have employed cointegration and error correction techniques. These include: Ebohon [16], Odhiambo [48] and Akinlo [6] on Nigeria and Tanzania; Jumbe [26] on Malawi; Lin and Wesseh [37], Kumar and Kumar [30], Odhiambo [49] and Menyah, and WoldeRufael [44] on South Africa and Kenya; Ouedraogo [52] on Burkina-Faso; Kouakou [32] on Cote d'Ivoire; Wesseh and Zoumara [73] on Liberia; Tamba et al. [69] and Wandji [72] on Cameroon; Solarin and Shahbaz [67] on Angola and Fuinhas and Marques [19] and Bélaïd and Abderrahmani [12] on Algeria and Egypt. There also exist multi-country studies for Africa in which all four hypotheses including feedback hypothesis, growth hypothesis, conservation hypothesis and neutrality hypothesis are supported. These studies include Wolde-Rufael [78], Wolde-Rufael [79], Akinlo [5], Wolde-Rufael [80], Esso [18], Odhiambo [50], Kebede et al. [28], Eggoh et al. [17], Al-mulali and Sab [8], Kahsai et al. [27], Richard [58], Behmiri, and Manso [11], Ouedraogo [53], Mensah [42] and Aissa et al. [4]. A review of energy consumption – economic growth nexus studies on Africa is summarized in Table 2. In light of the literature overview presented above, one may notice that the translog production model has never been applied to the investigation of problems in the energy consumption – economic growth nexus literature before despite its attractiveness and ability to provide insights on the efficiency with which various inputs are used. Moreover, the vast majority of studies have either focused on nonrenewable energy consumption or a combination of renewable and nonrenewable energy while neglecting the exclusive role of renewable energy. Giving the increased pressure on governments to limit their energy choices and make a transition towards renewables, it becomes necessary, especially when only few publications exist, to examine the effectiveness of renewable energy in alleviating energy poverty, mitigating greenhouse gas emissions and stimulating economic growth. Furthermore, given the mixed findings in the literature, it would play well to draw on deeper insights from more robust methodologies. Such an endeavor would not only provide valuable basis for reaching conclusion on the direction of causation between energy and economy, but would as well offer openings for further research and model specification. Another characteristic of this paper that sets it apart from previous studies in the literature is the fact that efforts are made to investigate the substitution effect between energy and other inputs in the model. By so doing, this study would provide estimates of the possibilities of substituting nonrenewable energy for renewable energy or vice versa. In fact, there are a number of studies dealing exclusively with inter-fuel/ inter-factor substitution possibilities with studies for European countries, Asian countries as well as North and South American countries. A review of these studies is given in Stern [68] and Smyth et al. [65,66]. Surprisingly, very little research on energy and resource substitution possibilities have been conducted for Africa countries despite the general consensus that these countries need to reduce their consumption of environmentally harmful resources and switch towards cleaner fuels like renewable energy. Moreover, given AfDB ambitious targets for green growth, research along these lines is necessary to explore the possibilities of replacing fossil energy with cleaner energy. To the authors' best knowledge; there is only one published study for Africa currently. Wesseh et al. [75] investigates the potential for inter-factor and inter-fuel substitution between capital, labor, petroleum and electricity in Liberia. The authors employed Ridge regression to estimate the translog production function and reached conclusions that all inputs considered are substitutes. Notwithstanding, the study pointed out that opportunities to substitute petroleum for

165

electricity; or labor and capital for electricity are limited in practice because of Liberia's current low scale electricity generation. Despite the contributions of the above study, one limitation is that it focuses exclusively on the substitution between fossil fuels and factors while neglecting the possibilities of substituting renewable energy for fossil fuels in Africa. The present study would therefore add value to the literature by attempting to fill the gaps that have been discussed.

4. Data and methods The applied dataset, model and methods of estimation form the domain of this section. 4.1. The data Variables used for analysis in this study include output, represented by real GDP; renewable energy, represented by total renewable electric power consumption; nonrenewable energy, represented by total nonrenewable electric power consumption (calculated as the difference between total electric power and total renewable electric power); labor; capital and the state of technological progress, represented by a time trend. Using electricity as a representation of energy is appropriate since, order than the transportation sector; it serves as the major fuel for households, industries, agriculture and service. Thus, power generation accounts for the vast majority of secondary energy. Electricity and real GDP data are available for almost all countries in Africa, but long time series for labor and capital are available for a relatively small number of African countries. Only those countries that have data available for at least 5 observations for each variable are considered (Table 3). The study collects data from 34 African countries covering the period 1980–2011, resulting in an unbalanced panel database with a total of 998 observations. This time domain permits to examine convergence issues inherent in the literature while the choice of country allows for a comprehensive geographical spread over the entire continent. Several transformations of the dataset were performed (see [34,36,73,74,76,77]). For instance, we define the point of approximation by normalizing all data around the sample mean before taking logs. This was done by dividing all observations by their sample mean. Because the translog production function is a Taylor series approximation, a good number of economic calculations are simplified at the point of approximation. In our case, such a point would correspond to a hypothetical country whose inputs, level of technology and generic outputs represent those of the sample mean. Without normalizing the data around the sample mean, it Table 3 Sample countries and geographical regionsa. Eastern

Central

Northern

Southern

Western

Burundi Ethiopia Kenya Madagascar Malawi Mauritius Mozambique Rwanda Tanzania Uganda Zambia Zimbabwe

Cameroon Congo (Braz.) Congo (DRC) Gabon Sao Tome and Principe

Algeria Egypt Morocco Tunisia

Lesotho Namibia South Africa Swaziland

Benin Burkina Faso Ghana Guinea Mali Nigeria Senegal Sierra Leone Togo

a

Categorization in table is based on United Nations country grouping.

166

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

remains questionable as to whether the production model would be able to satisfy all regularity conditions [71]. Output and capital stock were also calculated at constant prices (2005 ¼100) in order to eliminate any impact of inflation. Data on output or real GDP in our case, gross capital formation (formerly gross domestic investment) and labor are published by the World Development Indicators (WDI). Labor in this study represents employment to population ratio multiplied by the active population. Since data on capital stock are not provided by WDI, this variable is calculated at constant prices (2005¼100) for individual countries using the perpetual inventory method. This approach to inventory allows for continued adjustment by balancing inflows with outflows and does not require such strong assumptions as the proportionality method. The mathematical expression used in the calculation is given by K it ¼ K it  1 ð1  δÞ þI it

ð1Þ

where K it is the current capital stock for the ith country at time t for i ¼ 1, 2…34 and t ¼ 1.2…32; K it  1 is the capital stock of the previous year; δ represents the capital depreciation rate and is taken as 5% which is based on the World Bank total wealth estimates [75]; and I it is the capital investment in the current year. To obtain the initial capital stock, the following equation is used: K 0 ¼ I 0 =ðg þ δÞ

ð2Þ

K0 in the above expression is the initial capital stock, I 0 symbolizes the initial capital investment, and g is the average growth rate of capital investment. Electricity data are from the United States Energy Information Administration (EIA). In order to begin a comprehensive initial analysis, it is necessary to make use of tables and charts. Table 4 reports descriptive statistics of the data. The results in Table 4 show that all means are far away from 0 with sample standard deviation within the range of 1.48–2.44. Measure of skewness indicates that capital, labor and renewable energy are skewed to the left while output and nonrenewable energy are skewed to the right. The kurtosis values (3 for normal distribution) and Jacque–Bera statistics demonstrate that the distributions of all series have thicker tails than normal. The results in Table 4 also show that, on average, slightly more electricity is generated from renewable sources across African countries relative to nonrenewable sources. 4.2. The model Considerable amount of efforts have gone into specifying general forms of the cost or production function. The vast majority of these forms is applicable to econometric estimation and incorporates economic effects. Furthermore, they are consistent with properties regarding the input requirement set. While selecting an appropriate functional form would for the most part depend on the particular study in question, most studies have relied on Table 4 Summary statistics for all logarithmic transformed variables. Statistic

Mean Std. Dev. Skewness Kurtosis Jarque-Bera

Variable Output

Capital

Labor

REa

NREb

22.6 1.48 0.02 2.76 24.11***

20.3 2.44  2.40 11.8 4223***

15.5 1.49  1.08 4.26 263***

6.31 1.97  0.55 3.09 51.9***

6.26 2.35 0.26 3.12 11.89***

consistency of production technologies with the theoretical properties as the basis for selection. Whatever the choice of a functional form may be, one thing that is clear is that popular hypotheses like homotheticity, homogeneity and separability, remain to be tested. Hence, the need to specify a more flexible production function which places fewer restrictions on the underlying technology becomes enormous [35]. As we have already mentioned, previous studies which apply panel data to this problem in the energy economics literature have all employed the translog cost function and this requires data on input prices (e.g. [13,40,41]). Because these data are not available for most African countries, this study employs a log linear translog production function, thus featuring a very important innovation. The translog production function has been applied in the energy and production economics literature to examine input substitution possibilities, aggregation and separability, productivity growth and technical change, and production efficiency. The translog production function is a second-order Taylor approximation or a second-order differential approximation at a particular point. This simple locally flexible functional form places no restrictions on the production technology, i.e. no requirement is imposed on the value of the function or its first and second derivatives at the point of approximation. Even though global flexibility is preferred,4 most of the applied production analyses in economics have considered local approximation since nothing much is known about the properties of global approximation. Given that the marginal rate of technical substitution in African countries may be affected over time, it is assumed here that technical change is non-neutral and scale-augmenting. Hence, we specify the general functional form of the translog production function in the contest of panel data as: ln Y it ¼ β 0 þ

J X

βj ln xjit þ

j¼1

þ

J X

αj ln xjit t 

J J 1X X 1 β ln xjit ln xkit þ γ 1 t  þ γ 2 t 2 2 j ¼ 1 k ¼ 1 jk 2

ð3Þ

j¼1

In the expression above, i ¼ 1; 2;…,34 represents the crosssectional units; t ¼ 1; 2;…,32 indicates the time periods; j; k ¼ 1; 2;…,Jare the applied inputs; ln Y it is the logarithm of the output associated with the ith country in time t; ln xjit is the logarithm of the jth input associated with the ith country in time t; t  is the time-trend representation of technical change; β ; γ and α are parameters to be estimated. In particular, βij's are the parameters associated with inputs that give information about input substitutability; γ 1 and γ 2 are the pure (or autonomous) component of technical change which indicate a neutral-shift effect on the production function that is non-attributable to any particular input; and αij's are the biased technical change which is a scale expansion effect influencing efficiency in the use of various inputs. In order to estimate Eq. (3), a number of conditions will have to be met. First, symmetry is necessary. In other words, Young's theorem must be satisfied and this requires that βjk ¼ βkj for allj; k. This implies that the production structure in Eq. (3) has 1 neutral-scale parameter (β0 ), jþ 2 first-order parameters (βj ; γ 1 ; γ 2 ) and ðj þ 1Þðj=2Þ þ j second-order parameters ðβjk ; αj Þ: Second, the marginal products of inputs should be positive, i.e. MP jit ¼ ∂yit =∂xjit 40(monotonicity). With regards to the translog production function, we calculate the marginal product of input j by multiplying the average product of inputj by the logarithmic marginal product. Therefore, the below equation has to be positive in order for the monotonicity condition of the translog

a

Renewable electricity. Nonrenewable electricity. *** Indicate significance at the 1% level. b

4 This is so because local flexibility can impose unknown restrictions away from the unknown point of expansion.

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

specification to hold. MP jit ¼

J X

!

∂yit yit ∂ ln yit yit ¼ : ¼ : βj þ βjk ln xkit þ αj t  40 ∂xjit xjit ∂ ln xjit xjit k¼1

ð4Þ

From Eq. (4), it becomes clear that monotonocity would depend on the sign of the term in parenthesis given that both y and xj are positive numbers. Finally, the marginal products of inputs should as well satisfy the law of diminishing marginal productivities, i.e. they should be decreasing in inputs. Hence, the following expression should be satisfied: ! ! J J X X   βj þ βjk ln xkit þ αj t : βj  1 þ βjk ln xkit þ αj t o 0 ð5Þ k¼1

k¼1

It is important for researchers to check the quantitative nature of the terms in Eqs. (4) and (5). In summary, positive marginal products for each input requires βj 4 0 for all j while diminishing marginal productivities calls for β j ðβ j  1Þ o 0 for all j, if and only if 0 o βj o 1 for all j. Generally, if these conditions are met for a sufficient number of observations, then the production function is said to be well-behaved. 4.3. Production structure The translog production function has proven to be very useful in applied production analysis due to its flexibility. Its ability to approximate a number of popular models has made the translog specification to be considered the most superior functional form. For our purposes, the translog specification is especially useful because output and substitution elasticities are allowed to vary with the levels of inputs, hence homotheticity is not imposed. A homothetic production function is one in which the marginal rate of technical substitution is homogenous of degree zero in inputs. The translog production function is homogenous of degree λ in inputs if: J X j¼1

β j ¼ λ;

J X

βjk ¼ 0 and

j¼1

J X

αj ¼ 0

ð6Þ

j¼1

If λ ¼1 in the above, then the production technology is said to exhibit constant returns to scale or linear homogeneity. In this case when input doubles, output has to double. Also, the translog production function is said to be strongly separable if βjk ¼ 0 for all j; k and this is equivalent to a Cobb–Douglas function with inputbiased technical change. On the other hand, if βjk ¼ 0; αj ¼ 0 and γ 2 ¼ 0 for all j; k; then Eq. (3) would reduce to a Cobb–Douglas technology with Hicks-neutral technical change. The following therefore shows three key parameter restrictions on (3) which shall be tested for in the empirical analysis: J X

βjk ¼ 0 ðhomotheticityÞ

ð7Þ

βj ¼ 1 ðlinear homogeneityÞ

ð8Þ

k¼1 J X j¼1

βjk ¼ 0 ðseparabilityÞ

ð9Þ

As noted by Tzouvelekas [71], all of the above hypothesis can be tested using any of the available conventional statistical tests. At this junction, one must not forget that the translog specification is so flexible to the extent that no priori restrictions are placed on the value of output elasticities, returns to scale, elasticities of substitution and technical change. In the first place, the

output elasticity of the jth input from Eq. (3) is given by: ! J J J J X X X X ∂ ln yit ηjit ¼ ¼ βj þ βjk ln xkit þ αj t  ∂ ln xjit j ¼ 1 j¼1 j¼1 k¼1

167

ð10Þ

Where ηjit are the output elasticities that vary with the levels of input and state of technology. As can be seen, these are simply the logarithmic marginal product of the translog function and measure the degree of responsiveness of output to a percentage change in the jth input. Economics of scale estimates from the translog function is given as the sum of the output elasticities. To summarize the results of output elasticities and facilitate policy planning and regional allocation of resources, gaps in output elasticities between renewable and nonrenewable energy are computed and reported for different regions in Africa and the continent in general. The relationship we develop to calculate the elasticity gap is given by

ηgap ¼ ηRE  ηNRE

ð11Þ

In the above expression, ηgap is the output elasticity gap between renewable and nonrenewable energy, ηRE is the output elasticity of renewable energy and ηNRE is the output elasticity of nonrenewable energy. Positivity of ηgap implies that renewable energy contributes much greater to economic growth than nonrenewable energy and negativity means that nonrenewable energy plays a greater role in economic growth. Better still, a zero value of ηgap implies that both renewable and nonrenewable energy contributes equally to economic activities across Africa. Next, the marginal elasticities of output make it possible to derive the substitution elasticities that vary with the quantity of inputs used. Where MP is the marginal product, the relevant symmetric substitution elasticity between inputs j and k for the ith country in time t is given as: " #1  β jkðitÞ þ ðηjit =ηkit ÞβkkðitÞ σ jkðitÞ ¼ 1 þ ð12Þ  ηjit þ ηkit Inputs j and k are substitutes if σ jkðitÞ 40; independent if σ jkðitÞ ¼ 0 or compliments if σ jkðitÞ o0. Lastly, output elasticity with respect to the rate of technical change can be calculated as:   J X d ln yit  ¼ γ1 þ γ2t þ αj ln xjit ð13Þ   dt dxj ¼ 0 j¼1 Technical change in the above expression is both time and country specific and also varies with the level of input. While this measure is usually nonnegative, negativity could arise in the instance where countries face tightened or new regulations. In the sense of Hicksian, technical change is input j using if αj 4 0; neutral if αj ¼ 0 or saving if αj o 0. Tzouvelekas [71] highlights that factor-augmenting technical change with equal rates of augmentation is equivalent to Hicks-neutral technical change for constant returns to scale. Notwithstanding, if there are no constant returns to scale, then αj ¼ 0 for all j is a necessary requirement for Hicks-neutral technical change. For simplicity and clarity of model, Eq. (3) is expanded in the form below: ln Y it ¼ β0 þ βK ln K it þ βL ln Lit þ β RE ln REit þ βNRE ln NREit þ β ðREÞðKÞ ln REit ln K it þ β ðREÞðLÞ ln REit ln Lit þ β ðREÞðNREÞ ln REit ln NREit þ βðNREÞðKÞ ln NREit ln K it þ β ðNREÞðLÞ ln NREit ln Lit þ βðREÞðREÞ ðln REit Þ2 1 þ β ðNREÞðNREÞ ðln NREit Þ2 þ γ 1 t  þ γ 2 t 2 þ αK ln K it t  2 þ αL ln Lit t  þ αRE ln REit t  þ αNRE ln NREit t 

ð14Þ

In the above, K, L, RE and NRE are inputs of capital, labor, renewable energy and nonrenewable energy respectively. To

168

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

reduce the number of parameters and facilitate estimation with random effects techniques (since number of cross section has to be greater than number of coefficients), the translog components for capital and labor and the output and substitution elasticities between these factors and energy inputs are excluded from the model [33,34]. 4.4. Estimation technique Perhaps the most difficult problem in drawing inferences from nonexperimental data5 is how to statistically control for variables that cannot be observed. In the panel/cross sectional data literature, the most commonly estimated models have been fixed effects and random effects models. The ability of first differencing to remove unobserved heterogeneity also underlies the family of estimators that have been developed for dynamic panel data (DPD) models. On the one hand, if the researcher believes that there are no omitted variables in the model, or that omitted variables are uncorrelated with the explanatory variables, then a random effects model is probably best. This model will not only use all the available data, but it will also produce unbiased estimates of the coefficients and yield the smallest standard errors. However, there is still likelihood that at least some bias in the estimates will be generated since omitted variables are not controlled for. On the other hand, if there are omitted variables, and that these variables are correlated with the variables in the model, then a fixed effects model may provide a means of controlling for omitted variable bias. In fixed-effects models, each individual is used as his or her own control. The basic idea is that whatever effects the omitted variables have on any individual at one time, they will also have similar effect at a later time; hence their effects will be fixed. Some characterizations from experimental research have suggested that random effects modeling techniques are nearly always preferable (see [31]). In the framework of nonexperimental data however, nothing could be more distant from the truth. According to Wooldridge [82], unobserved heterogeneity should always be regarded as random variables. This therefore means that the distinguishing factor between fixed effects and random effects models lies in the structure of the correlations between the observed variables and the omitted variables. Unless these correlations are allowed for, one cannot control for the effects of omitted variables. This therefore makes the fixed effects methodology very attractive. At this junction, one must not forget that the interaction terms arising from our application of the translog model imply that there may be heterogeneity among countries, even in the instance of equal slope coefficients across countries. Moreover, fixed effects estimates may have substantially larger standard errors than random-effects estimates, leading to higher p-values and wider confidence intervals. This is because random effects estimates use information both within and between individuals. Fixed effects estimates, on the other hand, use only within-individual differences, essentially discarding any information about differences between individuals. If independent variables vary greatly across individuals but have little variation over time for each individual, then fixed effects estimates will be rather imprecise. Rather than imposing uncommon coefficients from the outset, Eq. (13) was first estimated by OLS estimator which is equivalent to the fixed effects model. Next, the likelihood ratio test was used to check for common intercept and slope. However, the results of this 5 For experimental researchers, the solution to that problem is simple. Random assignment to treatment groups makes those groups approximately equal on all characteristics of the subjects, whether those characteristics are observable or unobservable. But in nonexperimental research, the classic way to control for potentially confounding variables is to measure them and put them in some kind of regression model. Without measurement, there is no control.

Table 5 Parameter estimates of the translog model. Variable

Coefficient

Std. error

t-statistic

p-value

β0 βK βL βRE βNRE βðREÞðKÞ βðREÞðLÞ βðREÞðNREÞ βðREÞðREÞ βðNREÞðKÞ βðNREÞðLÞ βðNREÞðNREÞ γ1 γ2 αK αL αRE αNRE

14.89*** 0.1550*** 0.2117*** 0.1268*** 0.0532** 0.0045*  0.0201**  0.0256*** 0.0079***  0.0008 0.0005 0.0158***  0.0043 6.8200 0.0007***  0.0007* 0.0007* 0.0008** 0.99

0.9515 0.0165 0.0533 0.0350 0.1377 0.0027 0.0083 0.0041 0.0024 0.0040 0.0061 0.0024 0.0070 9.8200 0.0001 0.0004 0.0004 0.0003

15.658 9.3512 3.9716 3.6160  3.9142 1.6828  2.4066  6.1219 3.2509  0.2226 0.0818 6.3760  0.6187 0.6948 6.0650  1.7037 1.7905 2.2694

0.0000 0.0000 0.0001 0.0003 0.0454 0.0927 0.0163 0.0000 0.0012 0.8239 0.9348 0.0000 0.5362 0.4873 0.0000 0.0888 0.0737 0.0235

R2 *

Indicates significance at the 10% levels respectively. Indicates significance at the 5% levels respectively. *** Indicates significance at the 1% levels respectively. **

analysis showed no evidence of fixed effects across countries. For this reason, the random effects model was also estimated using the standard GLS estimator.6 We then applied Hausman [22] tests to check which model, whether fixed effects or random effects was more suitable. The tests results, which are not reported for the sake of conserving space, suggested no evidence of correlation between the individual country effects and the other parameters in the model, implying that the random effects model is more appropriate. Hence, this study reports estimates of the random effects model. Subsequently, all calculations are based on these estimates.

5. Empirical results This section reports coefficient estimates of the translog production model including output elasticities and the rate of technical change and energy substitutability. 5.1. Renewable energy, nonrenewable energy, technological progress and economic growth in African countries Given the high number of estimated parameters in our translog formulation, we begin with an investigation of multicollinearity in the data using the approach of Kmenta [29]. According to Kmenta, a simple measure of the degree of multicollinearity is obtained by regressing each of the independent variables on the remaining independent variables. Clearly, the translog production model specified in this study is well behaved and satisfies all regularity conditions at the point of approximation including diminishing marginal productivities since all the estimated βj coefficients are positive and fall between 0 and 1.The calculated coefficients of 2 determination ðR^ Þ from the regression can then be used as a measure of the degree of multicollinearity in the sample. In this 2 study, the values obtained for R^ corresponded to 0.17, 0.44, 0.16 and 0.20 for capital, labor, renewable energy and nonrenewable 6 Results from GLS estimator when compared with those of GMM estimator looked quite similar. These results are not presented in this paper to conserve space but are certainly available upon request from the authors.

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

energy respectively, thus suggesting that multicollinearity is not severe and consequently not a problem in the estimated model. The estimated results of the parameters of the translog production model in Eq. (13) are reported in Table 5. It can be seen that almost all major parameters have the expected sign and all but two of the coefficient estimates are statistically significant at conventional levels. The results show that economic growth in Africa is driven by capital, labor and both renewable and nonrenewable energy. In particular, Table 5 shows that a unit increase in capital, labor, renewable and nonrenewable energy would boast economic growth by 15%, 21%, 12% and 5% respectively. This suggests that, given Africa's current energy structure, renewable energy seems to play a much greater role in economic growth in general than the conventional fossil fuels. These findings are reflective of the fact that renewable sources like wind, hydro and solar account for a greater share of power generation in most African countries. Insignificance of the gamma parameters and significance of the alpha parameters suggests that technological progress is scaled biased or driven mainly by the efficiency with which various inputs are used rather than a neutral shift effect. To further strengthen these findings and given that the translog production model does not have a direct interpretation of the elasticities of output and the rate of technical change, marginal products of renewable energy and nonrenewable with respect to output are computed at each data point. 5.2. Marginal product of renewable and nonrenewable energy with respect to output Output elasticities of renewable and nonrenewable energy give a measure of how output changes due to a unit change in renewable and nonrenewable energy. The measures here are both Table 6 Estimated marginal products of renewable and nonrenewable energy. Country/Region

Renewable energy

Nonrenewable energy

Algeria Benin Burkina Faso Burundi Cameroon Congo (Brazzaville) Congo (DRC) Egypt Ethiopia Gabon Ghana Guinea Kenya Lesotho Madagascar Malawi Mali Mauritius Morocco Mozambique Namibia Nigeria Rwanda Sao Tome and Principe Senegal Sierra Leone South Africa Swaziland Tanzania Togo Tunisia Uganda Zambia Zimbabwe

0.591 0.800 0.883 0.889 0.959 0.914 0.955 0.925 1.019 0.859 0.954 0.878 0.980 0.910 0.902 0.939 0.908 0.799 0.890 0.910 0.839 0.967 0.930 0.730 0.857 0.851 0.852 0.772 0.988 0.832 0.814 0.941 0.884 0.869

0.483 0.576 0.474 0.384 0.389 0.350 0.416 0.532 0.391 0.380 0.385 0.429 0.393 0.308 0.419 0.349 0.391 0.450 0.538 0.430 0.417 0.489 0.369 0.343 0.504 0.421 0.633 0.428 0.397 0.453 0.572 0.406 0.393 0.462

169

time- and country-specific and useful for drawing inferences on resources allocation over time. To preserve space, however, only average values over time for individual countries and regions are presented in Table 6. Consistent with results reported in Table 5, the estimated output elasticities in Table 6 indicate that changes in renewable energy stimulate higher economic growth in African countries (as indicated by the higher marginal products of renewable energy) as compared to nonrenewable energy with growth elasticities ranging between 0.591–1.019 (Algeria and Ethiopia) and 0.308–0.633 (Lesotho and South Africa) for renewable energy and nonrenewable energy respectively. The lower marginal product of nonrenewable energy in South Africa despite the high dependency on coal for power generation (about 72% electricity from coal), can be attributed to the fact that output elasticities are based on parameter estimates for Africa in general (Eq. 14). Results from the analysis on computed output elasticity gaps are reported in Fig. 4. As we have indicated in Eq. (11), the red line lies far above the green line demonstrating the contribution of renewable energy over nonrenewable energy to economic growth in African countries. Gaps in output elasticities (represented by the bars) are all positive and higher for the Eastern and Central regions of Africa (0.517 and 0.507 respectively) compared to the Northern, Southern and Western regions (0.27, 0.39 and 0.42 respectively). The results here show that Eastern Africa and Central Africa have higher dependence on renewable energy sources than the other three regions. For Africa as a whole, the average estimated output elasticity gap between renewable energy and nonrenewable energy is 0.41 indicating the significant role of renewable power in Africa's economic activities. Since technical change is non-neutral, the next analysis focuses on the efficiency with which renewable and nonrenewable energy are used. 5.3. State of technological progress Technical change as defined previously measures the effects of technology accumulation. In this study, technical change is both time and country specific and varies with the level of inputs. Descriptive statistics of the estimated rates of technical change over countries and time as reported in Table 7 show that the average rate of technical progress in African countries, driven mainly by the efficiency with which various inputs are used, ranges between 6.822 and 6.837 with mean value of 6.829. The biased component of technical change has reported positive or technical progress for all countries under study. According to the results, innovations in technology observed during the sample was saving in labor (negative), but using in capital, renewable energy and nonrenewable energy (positive). 5.4. Substitution elasticities Based on the parameter estimates in Tables 5 and 6, substitution elasticities between renewable and nonrenewable energy were computed (Fig. 5). Except for Algeria which reports negative substitution elasticities for a couple of data points, all estimates are positive and greater than 1 suggesting a high degree of substitutability between renewable and nonrenewable energy in African countries.

6. Renewable versus nonrenewable energy Energy is a driving force behind all socio-economic activities and as such, it is squarely in this sector that Africa's battle for longterm development and economic prosperity will be won or lost. Sustainable solutions focusing on supply security, energy poverty

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

1.0

0.8

0.6

0.4

Output elasticity gap

fri ca To ta lA

Af ric a es te rn W

Ea st er n

C en tra lA fri ca N or th er n Af ric a So ut he rn Af ric a

0.2

Af ric a

Output elasticities of renewable and nonrenewable energy in Africa

170

RENEWABLE

NONRENEWABLE

Fig. 4. Gaps in output elasticities between renewable and nonrenewable energy in Africa. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.) Table 7 Descriptive statistics of the estimated rate of technical change across African countries. Statistica Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis Jarque-Bera

important disadvantages of renewable energy (or advantages of nonrenewable energy) are discussed within the context of Africa's transition to industrialization and urbanization. 6.1. Scale

6.829 6.828 6.837 6.822 0.003 0.141 2.691 7.278**

a All values have been rounded up to three decimal places. ** Indicates rejection of the null hypothesis of normality at the 5% level of significance.

eradication and mitigation issues are therefore being placed at the forefront of most African energy policy discussions. Results of this study have demonstrated remarkable growth records driven largely by renewable energy and the possibilities of energy conversion. Notwithstanding, one must not forget that despite Africa's growth, over 50% of the African population lives below the poverty line and the average electrification rate is only 28.5% and less than 2% in Liberia and Sierra Leone.7 Perhaps Africa's high reliance on renewable sources of electricity which currently suffer stronger limitations challenges the effectiveness of relying on renewable energy while transitioning through industrialization and urbanization. In order to guide policy planning in Africa, Table 8 summarizes major advantages and disadvantages of renewable and nonrenewable energy. Table 8 shows, among other things that, while nonrenewable power seems to enjoy scale of production, cost-effectiveness and sitting convenience, exhaustibility and environmental pollution are important disadvantages. To ensure energy security and reduce carbon dioxide emissions, there is immense pressure for African governments to limit their use of fossil fuels and make a switch towards renewable energy. In what follows, the three most 7 Presley K. Wesseh, Jr., contributing writer, FPA. Available at: http://frontpa geafricaonline.com/index.php/op-ed/commentaries-features/535-davos-world-eco nomic-forum.

The energy content in wind and solar are far less than that contained in conventionally energy or nuclear fuels. Conventional fuels have capacity factors between 85% and 90% while wind and solar vary between 21.7 and 34.4% [47]. This means that between 65.6–78.3% of the time in every given year, wind and solar will not be able to supply power. It is therefore not surprising while most African countries still suffer severe electricity gaps and deindustrialization crises. Africa is developing and requires large-scale stable power supply. On the one hand, the power density of wind or amount of power flow per unit area is about 1 watt/m2 on average compared with 550 W/m2 and 1100 W/m2 for coal and gas-fired generation technologies respectively (Table 9). Therefore, for Africa to reach its full wind potential of 300 GW with maximum capacity factor, this would require a total land area of 300,000 km2 which is about the land area of Italy or more than half the total land area of the whole of Eastern Africa. As a result, this would bring about higher capital investment especially in land and result in social acceptability issues leading to further delay in deployment. Solar power is also afflicted by the same general problem. As may be seen from Table 9, solar PV has an average power density of about 6.5 W/m2 of which only 20–30% is convertible to electricity depending on the technology. On the other hand, large-scale production of biofuels for use in the transportation sector is not also feasible for African countries in terms of technology development and land requirement for feedstock production. In the first place, growing feedstock for biofuels production would require about two to three times more land area than wind and solar would require for electricity generation. Second, replacing the transportation infrastructure like switching to electric vehicles and hydrogen cells is not only expensive but highly unlikely especially for the truck industry. Finally, most African countries are poverty stricken primarily because agriculture has not been mechanized due to poor energy services. Where growing sufficient food to feed its people is a serious challenge, there are doubts that Africa will be able to grow large-scale feedstock for biofuels production.

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

171

Substitution elasticities

5 0 -5 -10 -15 -20 Fig. 5. Estimated substitution elasticities between renewable and nonrenewable energy in African countries.

Table 8 Pros and cons of renewable and nonrenewable energy. Renewable energy

Table 9 Power densities of various generation technologies.Source: [64].

Nonrenewable energy

Pros

Cons

Pros

Cons

Non-exhaustible

Low energy content Costintensive Seasonal/ Affected by the weather Sitting issues

High energy content Less costintensive Not affected by the weather

Exhaustible

Environmentally friendly Available globally

Job creation

Low capacity factor

Not environmentally friendly

Abundantly available Easily transported High capacity factor

6.2. Economics Although electricity generated with renewable energy sources enjoy zero fuel cost, the higher capital cost on the one hand and value of that generation on the other hand, challenges the competitiveness of renewable electricity. First, one must not forget that not all electricity creates equal value. Specifically, power produced at periods of peak demand is more valuable than offpeak generation, whether during a given daily cycle or across annual seasons. In this regard, wind generation in particular is problematic because in general there is an inverse relationship between the daily hours of peak demand and wind velocities as winds tend to blow at night. Second, most electric generation capacity fueled by renewable energy sources cannot be assumed to be available upon demand because of the uncertainties caused by the unreliability of wind and sunlight suggesting that system planning and optimization cannot assume that such power will be available when it is expected to be most economic. As a result, scheduling or dispatching becomes impossible. To fill the gap, backup generation capacity is required to preserve system reliability. Consequently, the cost of that needed backup capacity becomes a crucial parameter which is usually excluded from public discussions on wind and solar power. If rough estimates of backup costs are included, the EIA estimates of onshore wind in 2016 stand at about $517 for wind and $625–$764 for solar generation per megawatt-hour compared with $80–$110 per megawatt-hour for gas- or coal-fired generation. Indeed, the projected cost of renewable power in 2016 through 2019 including the cost of backup capacity is at least five times higher than that for conventional electricity. Given Africa's thin financial resources and the severe energy poverty, a more realistic development model based on cheaper

Power source

Natural gas Coal Solar (PV) Solar (CSP) Wind Biomass

Power density (W/m2) Low

High

200 100 4 4 0.5 0.5

2000 1000 9 10 1.5 0.6

and more stable power sources like natural gas and coal would present opportunities. Removing China from the equation, one would realize that global poverty has increased far above the 1981 levels. By utilizing its cheap resources, especially energy resources like coal, China has been able to alleviate poverty and achieve close to 100% electrification rate. The Chinese are gradually moving towards renewable energy, but only after eradicating energy poverty and achieving a certain level of economic growth. Africa accounts for about 5% of global coal production but exports majority of this amount out of the continent; this trend has to be changed. Regulations on the use of coal or other fossil sources for power generation should rather be long-run driven. 6.3. Sitting problems In principle, traditional power plants can be built or placed almost anywhere. Fuels like coal, oil and natural gas can be transported to the generation utility thus optimizing land costs, reliability issues, environmental factors, transmission line losses, etc. On the other hand, most renewables like wind and solar must be placed close to where the wind blows and the sun shines sufficiently and for a long time (transmission line issues might not apply to solar PVs of smaller applications). Since the most useful and most appropriate sites are limited as those ones with the lowest costs are utilized first, the successive or marginal cost of utilizing these sites must increase. In general, the low energy content of most renewable energy sources, sitting problems and subsequent transformation into a form useable for modern applications requires massive capital investment including land. This implies that the energy that can be extracted from renewable sources, relative to that from traditional forms, by its very nature is limited and expensive. As reported in Table 9, the two major disadvantages of nonrenewable energy are exhaustibility and environmental pollution. One must not forget that the back-ups requirement and switching factor for renewable power generation brings about higher emissions when compared to direct generation from fossil fuels. Moreover, even if nonrenewable energy is exhaustible, it would still be better to fully

172

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

utilize such cheaper resources from the start and then make a gradual transition after achieving a certain energy generation benchmark.

A comprehensive cost analysis of developing renewable energy for Africa relative to nonrenewable energy or a combination of the two would form a valuable avenue for future research and model development.

7. Conclusions Acknowledgement This study presents a comprehensive discussion and general framework on the estimation properties of the translog production function to provide insights on the effectiveness of renewable energy as a model for powering Africa's development. The study utilizes country-level panel data for 34 African countries over the period 1980–2011. The framework in this study provides support for the neoclassical production theory regularity conditions of convexity and monotonicity. Several findings have been documented from the estimation of a random-effects model using generalized least squares estimator. First, the results show that capital, labor, renewable and nonrenewable energy drive output in African countries; with renewable energy being a higher driver of growth than the conventional fossil fuels. Output elasticities computed for each input mirror these results and suggest that the Eastern and Central African countries have higher dependency on renewable energy sources than the other three regions. Given that electricity has been used as a measure of energy, these findings make sense since electricity in most African countries comes from renewable sources. Second, the analysis shows that real technological progress is weak and driven mainly by the efficiency with which various inputs are used rather than a neutral shift effect. In particular, innovations in technology observed during the sample was saving in labor but using in capital, renewable energy and nonrenewable energy. Third, the study documents that African countries have the potential of substituting between renewable and nonrenewable energy while boasting output and mitigating greenhouse gas emissions at the same time. However, the low energy content of most renewable energy sources, sitting problems and subsequent transformation into a form useable for modern applications requires massive capital investment including land. This implies that the energy that can be extracted from renewable sources, relative to that from traditional forms, by its very nature is limited and expensive suggesting that African countries cannot rely on such energy sources to power their development. Instead, African countries can first focus of reducing their energy poverty and hastening their pace of growth and development by scaling up the use of conventional energy, with a gradual shift towards cleaner energy sources coming only after achieving these benchmarks. Finally, the applied model reinforces the assertion that imposing restrictions like homotheticity, homogeneity or separability on the production technology is unrealistic and should rather be a testable hypothesis within any applied production analysis. While the use of a flexible functional form may generate valuable insights, there are also certain limitations that must be pointed out. In the first place, a flexible functional form limits the range of underlying technologies that can be characterized. In other words, a technology is able to satisfy all properties of the production function if that technology is consistent with homotheticity. Drawing on insights from fundamental duality theory, this characteristic of flexible functional forms might not be very surprising since an explicit implication of the theory is that any specification of a production or cost function places some restrictions of the technology. In additional, it is being increasingly recognized that most flexible functional forms, if not all, become very inflexible especially when considering separable technologies. Because of these reasons, the translog production function cannot be treated as a panacea for solving all possible model specification problems especially in applied production analysis.

The paper is supported by the Grant for Collaborative Innovation Center for Energy Economics and Energy Policy (No: 1260Z0210011), Xiamen University Flourish Plan Special Funding (No: 1260-Y07200), and Ministry of Education (Grant No. 10JBG013).

References [1] Abosedra S, Dah A, Ghosh S. Electricity consumption and economic growth, the case of Lebanon. Appl Energy 2009;86:429–32. [2] Acaravci A. Structural breaks, electricity consumption and economic growth: evidence from Turkey. J Econ Forecast 2010;2:140–54. [3] AfDB. African development report 2012—towards green growth in Africa. [4] Aissa MSB, Jebli MB, Youssef SB. Output, renewable energy consumption and trade in Africa. Energy Policy 2014;66:11–8. [5] Akinlo AE. Energy consumption and economic growth: evidence from 11 African countries. Energy Econ 2008;30:2391–400. [6] Akinlo AE. Electricity consumption and economic growth in Nigeria: evidence from cointegration and co-feature analysis. J Policy Model 2009;31:681–93. [7] Alam MJ, Begum IA, Buysse J, VanHuylenbroeck G. Energy consumption, carbon emissions and economic growth nexus in Bangladesh: cointegration and dynamic causality analysis. Energy Policy 2012;45:217–25. [8] Al-mulali U, Sab CNBC. The impact of energy consumption and CO2 emission on the economic growth and financial development in the Sub Saharan African countries. Energy 2012;39:180–6. [9] Altinay G, Karagol E. Structural break, unit root, and the causality between energy consumption and GDP in Turkey. Energy Econ 2004;26:985–94. [10] Aqeel A, Butt MS. The relationship between energy consumption and economic growth in Pakistan. Asia Pac Dev J 2008;8:101–10. [11] Behmiri NB, Manso JRP. How crude oil consumption impacts on economic growth of Sub-Saharan Africa? Energy 2013;54:74–83. [12] Bélaïd F, Abderrahmani F. Electricity consumption and economic growth in Algeria: a multivariate causality analysis in the presence of structural change. Energy Policy 2013;55:286–95. [13] Bousquet A, Ladoux NJM. Flexible versus designated technologies and interfuel substitution. Energy Econ 2006;28:426–43. [14] Bowden N, Payne JE. The causal relationship between US energy consumption and real output: a disaggregated analysis. J Policy Model 2009;31:180–8. [15] Chandran VGR, Sharma S, Madhavan K. Electricity consumption-growth nexus: the case of Malaysia. Energy Policy 2010;38:606–12. [16] Ebohon OJ. Energy, economic growth and causality in developing countries: a case study of Tanzania and Nigeria. Energy Policy 1996;24:447–53. [17] Eggoh JC, Bangake C, Rault C. Energy consumption and economic growth revisited in African countries. Energy Policy 2011;39:7408–21. [18] Esso LJ. Threshold cointegration and causality relationship between energy use and growth in seven African countries. Energy Econ 2010;32:1383–91. [19] Fuinhas JA, Marques AC. Rentierism, energy and economic growth: the case of Algeria and Egypt (1965–2010). Energy Policy 2013;62:1165–71. [20] Ghosh S. Electricity consumption and economic growth in India. Energy Policy 2002;30:125–9. [21] Halicioglu F. Residential electricity demand dynamics in Turkey. Energy Econ 2007;29:199–210. [22] Hausman JA. Specification tests in econometrics. Econometrica 1978;46:1251– 73. [23] Ho CY, Siu KW. A dynamic equilibrium of electricity consumption and GDP in Hong Kong: an empirical investigation. Energy Policy 2007;35:2507–13. [24] Hu JL, Lin CH. Disaggregated energy consumption and GDP in Taiwan: a threshold co-integration analysis. Energy Econ 2008;30:2342–58. [25] IRENA. International Renewable Energy Association; 2013. [26] Jumbe CBL. Cointegration and causality between electricity consumption and GDP: empirical evidence from Malawi. Energy Econ 2004;26:61–8. [27] Kahsai MS, Nondo C, Schaeffer PV, Gebremedhin TG. Income level and the energy consumption – GDP nexus: Evidence from Sub-Saharan Africa. Energy Econ 2012;34:739–46. [28] Kebede E, Kagochi J, Jolly CM. Energy consumption and economic development in sub-Sahara Africa. Energy Econ 2010;32:532–7. [29] Kmenta J. Elements of econometrics. New York: Macmillan Press; 1986. [30] Kumar RR, Kumar R. Effects of energy consumption on per worker output: a study of Kenya and South Africa. Energy Policy 2013;62:1187–93. [31] LaMotte LR. Fixed-, random-, and mixed-effects models. In: Kotz S, Johnson NL, Read CB, editors. Encyclopedia of statistical sciences. New York: John Wiley & Sons; 1983.

P.K. Wesseh Jr, B. Lin / Renewable and Sustainable Energy Reviews 54 (2016) 161–173

[32] Kouakou AK. Economic growth and electricity consumption in cote d'Ivoire: evidence from time series analysis. Energy Policy 2011;39:3638–44. [33] Lin B, Wesseh Jr. PK. Estimates of inter-fuel substitution possibilities in Chinese chemical industry. Energy Econ 2013;40:560–8. [34] Lin B, Wesseh Jr. PK. What causes price volatility and regime shifts in the natural gas market. Energy 2013;55:553–63. [35] Lin B, Wesseh Jr. PK. Valuing Chinese feed-in tariffs program for solar power generation: a real options analysis. Renew Sustain Energy Rev 2013;28:474–82. [36] Lin B, Wesseh Jr PK, Owusu Appiah M. Oil price fluctuation, volatility spillover and the Ghanaian equity market: Implication for portfolio management and hedging effectiveness. Energy Econ 2014;42:172–82. [37] Lin B, Wesseh Jr. PK. Energy consumption and economic growth in South Africa reexamined: a nonparametric testing approach. Renew Sustain Energy Rev 2014;40:840–50. [38] Lin B, Xie C. Energy substitution effect on transport industry of China-based on trans-log production function. Energy 2014;67:213–22. [39] Lorde T, Waithe K, Francis B. The importance of electrical energy for economic growth in Barbados. Energy Econ 2010;32:1411–20. [40] Ma H, Oxley L, Gibson J, Kim B. China energy economy: technical change, factor demand and interfactor/interfuel substitution. Energy Econ 2008;30:2167–83. [41] Ma H, Oxley L, Gibson J. Substitution possibilities and determinants of energy intensity for China. Energy Policy 2009;37:1793–804. [42] Mensah JT. Carbon emissions, energy consumption and output: a threshold analysis on the causality dynamics in emerging African economies. Energy Policy 2014;70:172–82. [43] Menyah K, Wolde-Rufael Y. CO2 emissions, nuclear energy, renewable energy and economic growth in the US. Energy Policy 2010;38:2911–5. [44] Menyah K, Wolde-Rufael Y. Energy consumption, pollutants emissions and economic growth in South Africa. Energy Econ 2010;32:1374–82. [45] Mozumder P, Marathe A. Causality relationship between electricity consumption and GDP in Bangladesh. Energy Policy 2007;35:395–402. [46] Narayan PK, Smyth R. Electricity consumption, employment and real income in Australia evidence from multivariate Granger causality tests. Energy Policy 2005;33:1109–16. [47] NEMS. 2016 Levelized cost of new generation resources from the annual energy outlook 2010. Available at: 〈http://www.eia.doe.gov/oiaf/aeo/pdf/ 2016levelized_costs_aeo2010.pdf〉; 2010. [48] Odhiambo NM. Energy consumption and economic growth nexus in Tanzania: an ARDL bounds testing approach. Energy Policy 2009;37:617–22. [49] Odhiambo NM. Electricity consumption and economic growth in South Africa: a trivariate causality test. Energy Econ 2009;31:635–40. [50] Odhiambo NM. Energy consumption, prices and economic growth in three SSA countries: a comparative study. Energy Policy 2010;38:2463–9. [51] Omri A. An international literature survey on energy-economic growth nexus: evidence from country-specific studies. Renew. Sustain. Energy Rev 2014;38:951–9. [52] Ouedraogo IM. Electricity consumption and economic growth in Burkina Faso: a cointegration analysis. Energy Econ 2010;32:524–31. [53] Ouedraogo NS. Energy consumption and economic growth: evidence from the Economic Community of West African States (ECOWAS). Energy Econ 2013;36:637–47. [54] Payne JE. On the dynamics of energy consumption and output in the US. Appl Energy 2009;86:575–7. [55] Payne JE, Taylor J. Nuclear energy consumption and economic growth in the US: an empirical note. Energy Sources Part B: Econ Plan Policy 2010;5:301–7. [56] Payne JE. On biomass energy consumption and real Output in the U.S. Energy Sources Part B: Econ Plan Policy 2011;6:47–52. [57] Ramcharran H. Electricity consumption and economic growth in Jamaica. Energy Econ 1990;12:65–70. [58] Richard OO. Energy consumption and economic growth in Sub-Saharan. Afr: Asymmetric Cointegr Anal Int Econ 2012;129:99–118. [59] Sari R, Ewing BT, Soytas U. The relationship between disaggregate energy consumption and industrial production in the United States: an ARDL approach. Energy Econ 2008;30:2302–13. [60] Shahbaz M, Tang CF, Shabbir MS. Electricity consumption and economic growth nexus in Portugal using cointegration and causality approaches. Energy Policy 2011;39:3529–36. [61] Shahbaz M, Feridun M. Electricity consumption and economic growth empirical evidence from Pakistan. Qual Quant: Int J Methodol 2012;46:1583– 99. [62] Shahbaz M, Lean HH. The dynamics of electricity consumption and economic growth: a revisit study of their causality in Pakistan. Energy 2012;39:146–53.

173

[63] Shiu A, Lam P. Electricity consumption and economic growth in China. Energy Policy 2004;32:47–54. [64] Smil, V. Power density primer: understanding the spatial dimension of the unfolding transition to renewable electricity generation. Available at: 〈http:// www.vaclavsmil.com/wp-content/uploads/docs/smil-article-power-densityprimer.pdf〉;, 2010. [65] Smyth R, Narayan PK, Shi H. Substitution between energy and classical factor inputs in the Chinese steel sector. Appl Energy 2011;88:361–7. [66] Smyth R, Narayan PK, Shi H. Inter-fuel substitution in the Chinese iron and steel sector. Int J Prod Econ 2012;139:525–32. [67] Solarin SA, Shahbaz M. Trivariate causality between economic growth, urbanization and electricity consumption in Angola: cointegration and causality analysis. Energy Policy 2013;60:876–84. [68] Stern DI. Interfuel substitution: a meta-analysis. J Econ Surv 2012;26:307–31. [69] Tamba JG, Njomo D, Limanond T, Ntsafack B. Causality analysis of diesel consumption and economic growth in Cameroon. Energy Policy 2012;45:567– 75. [70] Tang CF. Electricity consumption, income, foreign direct investment, and population in Malaysia: new evidence from multivariate framework analysis. J Econ Stud 2009;4:371–82. [71] Tzouvelekas E. Approximation properties and estimation of the translog production function with panel data. Agric Econ Rev 2000;1:33–47. [72] Wandji YDF. Energy consumption and economic growth: evidence from Cameroon. Energy Policy 2013;61:1295–304. [73] Wesseh Jr. PK, Zoumara B. Causal independence between energy consumption and economic growth in Liberia: evidence from a non-parametric bootstrapped causality test. Energy Policy 2012;50:518–27. [74] Wesseh Jr. PK, Niu L. The impact of exchange rate volatility on trade flows: new evidence from South Africa. Int Rev Bus Res 2012;P8:140–65. [75] Wesseh Jr. PK, Lin B, Owusu-Appiah M. Delving into Liberia energy economy: technical change, inter-factor and inter-fuel substitution. Renew Sustain Energy Rev 2013;24:122–30. [76] Wesseh Jr. PK, Lin B. Renewable energy technologies as beacon of cleaner production: a real options valuation analysis for Liberia. J Clean Prod 2015;90:300–10. [77] Wesseh Jr. PK, Lin B. A real options valuation of Chinese wind energy technologies for power generation: do benefits from the feed-in tariffs outweigh costs? J Clean Prod 2015. http://dx.doi.org/10.1016/j.jclepro.2015.04.083. [78] Wolde-Rufael Y. Energy demand and economic growth: the African experience. J Policy Model 2005;27:891–903. [79] Wolde-Rufael Y. Electricity consumption and economic growth: a time series experience for 17 African countries. Energy Policy 2006;34:1106–14. [80] Wolde-Rufael Y. Energy consumption and economic growth: the experience of African countries revisited. Energy Econ. 2009;31:217–24. [81] Wolde-Rufael Y. Bounds test approach to cointegration and causality between nuclear energy consumption and economic growth in India. Energy Policy 2010;38:52–8. [82] Wooldridge JM. Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press; 2001. [83] Yang HY. A note on the causal relationship between energy and GDP in Taiwan. Energy Econ 2000;22:309–17. [84] Yildirim E, Sarac S, Aslan A. Energy consumption and economic growth in the USA: evidence from renewable energy. Renew Sustain Energy Rev 2012;16:6770–4. [85] Yoo SH, Jung KO. Nuclear energy consumption and economic growth in Korea. Prog Nucl Energy 2005;46:101–9. [86] Yoo S. Electricity consumption and economic growth: evidence from Korea. Energy Policy 2005;33:1627–32. [87] Yoo SH, Kim Y. Electricity generation and economic growth in Indonesia. Energy 2006;31:2890–9. [88] Yuan J, Zhao C, Yu S, Hu Z. Electricity consumption and economic growth in China: cointegration and co-feature analysis. Energy Econ 2007;29:1179–91. [89] Yuan J, Kang J-G, Zhao C, Hu Z. Energy consumption and economic growth: evidence from China at both aggregated and disaggregated levels. Energy Econ 2008;30:3077–94. [90] Zachariadis T, Pashouortidou N. An empirical analysis of electricity consumption in Cyprus. Energy Econ 2007;29:183–98. [91] Zhang XP, Cheng XM. Energy consumption, carbon emissions, and economic growth in China. Ecol Econ 2009;68:2706–12.