Energy needs for Morocco 2030, as obtained from GDP-energy and GDP-energy intensity correlations

Energy Policy 88 (2016) 45–55

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Energy needs for Morocco 2030, as obtained from GDP-energy and GDP-energy intensity correlations Amin Bennouna n, Charaf El Hebil Physics Department, Faculty of Sciences Semlalia Marrakech, Cadi Ayyad University, Marrakech, Morocco

H I G H L I G H T S








We We We We We

present several mathematical models for country-level energy intensity. introduce the concept of the “partial energy intensity” for each energy segment. use mathematical models that historically work best for each one. extrapolate the models into the future to get their forecasts. use these lasts to calculate forecasts of needs in terms of final energy.

art ic l e i nf o

a b s t r a c t

Article history: Received 18 May 2015 Received in revised form 5 October 2015 Accepted 6 October 2015

We present forecasts of the energy consumption of Morocco towards 2030. Two models have been developed and their results compared: one based on the energy intensity (IE) and another one on a link with the country urbanization rate (URB). The IE model allowed to segment energy consumption in four posts while the URB model only in two posts. For the sensitivity analysis to economic growth, three future GDP evolution scenarios are proposed. The retrospective correlations of both models are excellent but their future extrapolations finish in slightly different results. Through their correlation to electricity consumption, peak power forecasts are also presented. A forecast of the country energy intensity is commented. As the average yearly increase of electricity should still be between 4.9% and 7.1% during 2020–2030, the electric equipment program continuation after 2020 must soon be clarified and avoid the former implementation delays. As the white combustibles needs should yearly increase between 6.3% and 7.8% in 2020–2030, electrical equipment programs should also make provisions for the case of deployment of electric cars. Butane subsidies widen the gap with other fuels and must be removed very soon possible to reduce the growth of its consumption and energy intensity. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Energy consumption Energy-income correlation Energy intensity History and forecast Developing country Morocco

1. Introduction Economical sustainability is a critical issue for countries having a very high energy dependence index like Morocco (495% in 1999–2008). More than for others, for a country so exposed to the increase and volatility of the international primary energy prices, the energy intensity must absolutely stay at the lowest level possible. 1.1. General considerations The notion of “causality”, related to variables having “caused” a given phenomenon, is, from our point of view, often wrongly n

Corresponding author. E-mail address: [email protected] (A. Bennouna).

http://dx.doi.org/10.1016/j.enpol.2015.10.003 0301-4215/& 2015 Elsevier Ltd. All rights reserved.

confused with the notion of “correlation” which is a statistically established mathematical link between an observed phenomenon and one or more variables. The existence of a correlation does not necessarily mean that the variables are the cause of the observed phenomenon, but that it adopts a mathematical behavior related to the changes of the aforementioned variables. To analyze and understand the behavior of the energy consumption of a given country, macroeconomic aggregates offer a wide choice of variables chief of which is, in general, the gross domestic product, as shown in Fig. 1. Its data are taken from the IMF, International Monetary Fund (2010) and the US Department of Energy (2006). Increasing energy costs and availability of more efficient equipments for energy conversion (power plants, vehicles, lighting, household electrical appliances, etc.) both should moderate the correlation of energy consumption and GDP. However, this correlation remains strong in DCs because the process of access to energy is incomplete (for their citizens and their production tools).

46

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

causality tests for 77 countries in 44 studies on the energy consumption-economic growth nexus, it can be stated that the topic has been intensively studied over the 35 last years, since Kraft and Kraft (1978) article, which is recognized as seminal by most authors. In addition, basing their remarks on 24 representative publications on the topic (dated between 1978 and 2012), Coers and Sanders (2013) have shown that a consensus on modeling has still not emerged and that previous work to 2013 can broadly be divided at least into:

Fig. 1. 2006 Energy and GDP adjusted for purchase power for 156 countries on left scale in blue and its related average energy intensity on right scale in red. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Electricity sold by utility (kWh/cap)

Simulated

Annual increase

Simulated

900 kWh

8%

800 kWh

7%

700 kWh

6%

600 kWh

5%

500 kWh

4%

400 kWh

3%

300 kWh

2%

200 kWh

1%

100 kWh 1975

1980

1985

1990

1995

2000

2005

2010

0% 2015

Fig. 2. Evolution of electricity per capita called by the grid in Morocco (on left scale in blue) and its related yearly increase (on right scale in red). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 2 shows how, the 1995–2002 period was a turning point in Morocco between two electricity consumption regimes: before 1995, it was increasing by around 11 kWh per year and per capita, while after 2002, it went to a faster regime, increasing around 39 kWh per year and per capita, essentially (but not only) because of the Global Rural Program of Electrification (PERG) grid extension. As this work is devoted to a segmented forecast of the Moroccan energy consumption, the presence of such turning points in the Moroccan past imposes to use historically well-established correlations capable to transcend cyclical phenomena like the one shown in Fig. 2. Its data are taken from the Moroccan utility ONEE, Office National de l’Électricité (1996–2012a, 1996– 2012b). If a statistical quality index (like the correlation coefficient) of a given retrospective correlation defines the precision of the model, the duration of its past validity is a security for the model extension towards the future. This is why this work will use the oldest available reliable data from Moroccan sources (from 1980 when available, 1985 otherwise). 1.2. Quick literature overview and the positioning of this paper 1.2.1. On energy-income correlations In a study of over more than hundred countries, Chontanawat et al. (2008) found that the causal relationship between energy consumption and economic growth is more pronounced in developed than in developing countries. But we hope that things have changed in the last decade otherwise their energy efficiency policy efforts would be simply useless to reduce CO2 emissions while continuing growing. As Bruns and Gross (2012) have counted no less than 534

1. Five categories of methodologies: simple causality tests, bivariate and multivariate Vector Error Correction (VEC) model in addition to bivariate and multivariate panel VEC model. 2. Four categories of results: no causation, causation from energy to GDP, causation from GDP to energy in addition to bidirectional causation. 1.2.2. On energy intensity Analyzing several works, Adom (2015) highlighted two contrasting arguments concerning the effect of income increase on energy intensity: 1. When dedicated to purchases of energy-using appliances, income rises may result in energy intensity increases, except if the purchase is used for energy efficient ones, where the energy intensity may decrease. According to this argument, an income increase causes a scale effect. 2. When dedicated for replacements of old equipment with energy-efficient ones, income rises may result in an energy intensity fall (income rise higher than energy use increase). Developing countries have high economic growth rates and Bernardini and Galli (1993) outlined three reasons why higher income may slow down energy intensity: 1. First, rising income changes the final demand structure, passing through different industrialization phases. 2. Second, rising incomes or higher GDP leads to technological progress that improves energy efficiency. 3. Third, technological progress also leads to the usage of substitute equipment that is less energy intensive. Overtime, the rebound effect argues that efficiency gains may lead to increase energy consumption. Mixed income effects create the possibility of a nonlinear behavior, which opens another strand of the literature which has focused on the nonlinear relationship between energy intensity and income. 1.2.3. On energy-GDP studies including Morocco Energy-income nexus of Morocco has been studied essentially for comparison purposes with: 1. Senegal and Ghana by Adom et al. (2012) who found that carbon dioxide emission acts as a limiting factor to economic growth in Morocco and Ghana. 2. South Africa, by Mans (2014) who compared the links to local green economies, 3. Other 4 North African countries by Marktannera and Salman (2011) who argued that the identification of any energy alternative as superior is hardly convincing unless certain standards of inclusive governance are met; they also highlighted the political–economic differences of energy importers like Morocco, 4. Other 13 members of the Organization of Islamic Conference by Gabbasa et al. (2013) who focused on their energy supply status

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

for sustainable development. 5. Other countries from the MENA Region:
Studying 14 countries, Omri (2013) showed that Morocco has their lowest CO2 emissions and energy consumption as well as the least energy consumption volatility and found that:
its energy consumption had an insignificant positive impact on GDP and a significant positive impact on CO2 emissions,
its GDP had a significant positive impact on energy consumption and an insignificant positive impact on CO2 emissions,
its CO2 emissions have an insignificant negative impact on both GDP and energy consumption,
Studying 11 countries, Ozturk and Acaravci (2011) found that, in Morocco, there is no cointegration between electricity consumption and the income, making impossible to estimate causal relationships within the dynamic VEC model.
Studying 24 countries, Tang and Abosedra (2014) studied the impacts of tourism, energy consumption and political instability on economic growth within the neoclassical growth framework. 6. Other African countries:
Studying 24 countries, Wolde-Rufael (2005) found that, in Morocco:
there was an opposite causality running from energy use to economic growth,
energy use negatively impacts on economic growth.
Studying 24 countries, Wolde-Rufael (2006) found that, in Morocco, there was:
a long-run relationship with GDP when electricity consumption is the dependent variable,
a positive bi-directional causality (income demands more electricity use and vice-versa).
Studying 24 countries, Wolde-Rufael (2009) confirmed that, in Morocco, causality was reversed from energy consumption to economic growth, to the opposite causality running from economic growth to energy consumption.
Studying 21 countries, Eggoh et al. (2011) found that decreasing energy consumption decreases growth and vice versa, and that increasing energy consumption increases growth, and vice versa, and that this applies for both energy exporters and importers.
Studying 21 countries, Mandelli et al. (2014) depicted the current energy situation of Africa, describing it as far as the concept of sustainable development is concerned, and tried to see if and how energy policies promoted by local players fit with this asset. They carried out overview of the energyrelated policies and action plans developed by different local players in the African continent with the goal of providing remarks by coupling such plans with the energy analysis.

47

energy-GDP test. 1.3. Positioning and limits of this paper In the relation between income and energy consumption shown in Fig. 1, if the “average behavior” is evident, the dispersal around it is substantial because of a set of economic, climatic, social and cultural factors, which describe the “national specificities”. Unlike most other articles, GDP is here expressed in local currency (Dirham, Dh) because there is no intend of benchmarking but it is hoped from this work a local thought-provoking about the Moroccan energy efficiency policy by highlighting some of the national problems arising in the 2030 forecasts of energy consumption and its intensity. Our results will show that, at the bottom line, it was possible to transcend some cyclical phenomena (like the one shown in Fig. 2) with relatively simple empirical correlative models. Of course, the risk remains that the causality itself, between GDP and energy consumption (and vice versa) will be missing but we have shown that already several authors have studied this. Doubtless, cyclical causes can describe better causality phenomena but we think that, to be precise, they require multiple hypotheses and variables as well as very big samples of timeseries data. Rather than this, a limited number of macroeconomic variables and hypotheses will be used here but still keeping the exigency of an acceptable long-lasting historical energy-GDP correlation (27–35 years back) to be used to forecast energy consumption in Morocco to 2030 (18 years forward) and to extract some comments and their consequences. Given the limits of this article scope, we avoided the dilemma of the chicken and the egg induced, sometimes, by the use of the “causality” concept and we will limit ourselves to the “correlation” one. Here, it is not pretended, neither arguing about theories of the energy-income nexus, nor having a deep enough expertize to do it.

2. Methods Except when duly mentioned, all our numbers come from Moroccan national sources: the state’s bank, Bank Almaghrib (1986–2012), who publishes a yearly report containing the energy consumption segments used here, or the national Direction de la Statistique (1983–2010) or the national Haut Commissariat au Plan (2015). One of the authors keeps updated what is the most complete private database about energy in Morocco allowing him to write a book on the topic, Bennouna (2011). 2.1. Global approach of the work In this paper we will:

7. Other countries, at the international level:
by Narayan and Popp (2012), who found that, in the 93 countries studied, Morocco is among the 44 having a significant long-run causality relationships GDP to total primary energy consumption.
by Apergis and Tang (2013) who found that, in the 85 countries studied, Morocco is among the 9 who reject the strong support to the energy-local economic development growth hypothesis.
by Bruns and Gross (2013), who found that in most of the 65 countries studied, the results of total energy-GDP causality tests frequently coincide with the results of energy type (electricity, petroleum products or Renewables)-GDP tests and, that Morocco is among the 82% of countries for which at least two energy-type GDP tests match with the total

1. present the few mathematical models used here for the country-level energy intensity analysis, 2. introduce the concept of the “partial energy intensity” for each final energy segment, 3. use the mathematical models that historically work best on this “partial energy intensity”, 4. extrapolate the models into the future to get forecasts of “partial energy intensities”, 5. use these lasts to calculate forecasts of needs in terms of final energy segments. 2.2. Correlative models based on energy intensity (IE model) Economical sustainability is a critical issue for countries, like

48

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Morocco, having an important energy dependence index. In this view, an important aggregate is the energy intensity which must stay at a sustainable level for the country economy. 2.2.1. Introduction Energy intensity (I) is defined as the ratio of energy consumption (E) to the value added (P) it contributes to create, the Gross Domestic Product in our case. Energy intensity is smaller for countries having an efficient economic activity but also for those having an important share of services in their GNP. In the specific case of Morocco, the primary sector generates around 16% of GDP, the secondary 29% and the tertiary 55%. In developed countries, competition requirements are imposed by globalization and energy intensity may tend to decrease over time, sometimes even with a growing GDP. But this does not stand for countries (mainly DC’s) that are still in the implementation phase of the access to energy for the benefit of their citizens and their economic production. 2.2.1.1. Power function behavior (IE-POW). In this model, it is considered that, during a given time interval, energy intensity varies with GDP according to a power function:

I = (E/P ) = β. (P − Po)n Where Po is the GDP generated without the use of energy E, β is a constant whose unit depends on the exponent n, which characterizes the variation mode of I with P: 1. When n is bigger or equal to 1: energy intensity I increases at least as fast as the income, with respectively a concave (J-shape) or linear shape, but economic sustainability prevents such a model to settle indefinitely, 2. When n is positive but smaller than 1: energy intensity I increases more slowly than income, with a convex shape, ensuring access to energy without destabilizing too much the country's economy, but such a behavior is not sustainable in the very long term, as the energy still increases faster than income, 3. When n ¼0, energy intensity I is constant and ensures an energy consumption E following income: even if it looks as a perennial model, it is not a competitive behavior in an environment of global race for energy efficiency, 4. When n is negative, energy intensity I decreases with income and this is the most appropriate behavior to face global economic competition and climate change challenges. In terms of mathematical method, after a graphical visualization of I against P: 1. For graphs apparently going to Po E0, the coefficient (β) and exponent (n) can be simply extracted from a log–log representation of I against P (or even E which should vary like E. Pn þ 1). 2. For graphs apparently going to Po≠0, we seek the Po value for which we obtain an optimal linear fit of ln(I) against ln(P Po). 2.2.1.2. Exponential behavior (IE-EXP). Although the modernization of the productive sector and the need for citizens’ comfort led to an energy intensity increasing with income, this increase cannot continue indefinitely because it would become economically unsustainable, especially when primary energy prices grow. The need for mathematical models that describe correlations directing energy intensity to a ceiling naturally drives towards exponential functions:

I = (E/P ) = I∞{1 − exp[ − (P − Po)/a]}

Where I1 represents the energy intensity achieved if the model could last infinitely, Po, the GDP generated without the use of energy E, and α, a coefficient which characterizes the leading edge of the intensity energy against GDP and having the same unit than the GDP. In terms of mathematical method: ln│1  (I/I1)│ is calculated and the value of I1 is changed until obtaining an optimal linear behavior. 2.2.2. Retrospective “partial” energy intensities of Morocco (IE MODEL) For Morocco, there are still no official series on energy consumption by sector. Thus, there is no question, for the time being, of going to a sector analysis without making risky assumptions in the segmentation. However, we venture in this work with an analysis of the final energy by applying the previous analytical behavior to “partial energy intensities,” due to specific segments of the final energy:

I = Ielectricity + Iwhite combustibles + Ibutane + Iothers

I = (Eelec/P ) + (Ewhi.comb./P ) + (Ebutane/P ) + (Eothers/P ) It will be voluntarily chosen to represent GDP in MDh (million Dirham) or GDh (billion Dirham) 2012 constant prices, the most telling unit for money for the more recent data used, although a different reference year would have changed nothing. Over 2002– 2012, one Dh counted for about 0.09 €. 2.2.2.1. Electricity. Hereafter we call “electricity” the net electrical energy called annually by the entire Moroccan power grid. We will just take this, even if it is true that it would have been better to consider the electricity delivered to customers; but reaching this last would have required the grid annual performance, its yearly fluctuations and a forecast for it too. Unfortunately all contributions to the Moroccan grid performance are not known in the terms needed for this study. Calculated on the basis of an average thermal energy conversion factor around 0.243 (2011), electric power, with its 7.438 Mtoe, is the first post of the final energy consumption, exceeding slightly 39.8% of the total for 2012. Fig. 3 shows the behavior of electricity in Morocco against GDP (blue diamonds related to the left scale), and its partial energy intensity (red squares related to the right scale). The absence of a “plateau” in the curve I(P) suggests avoiding the exponential function model. That is why the simulations shown in the graph are obtained after the intensity I(E) has been adjusted using I ¼a. (P  Po)n. Like shown in Fig. 4, the mathematical adjustment of electricity by the power function (n ¼0.531 and Po ¼ 23.5 GDh) is provided

Fig. 3. Simulation of electricity by the IE-POW model. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Simulated

Intensity (toe/MDh 2012)

2,0

2,5 2010

2005

3,0

2,0

2000

1,6

1,5

1995

1,2

1,0

0,4

0,5

1990

0,8

1985

Energy (Mtoe)

Simulated

2,4

0,0 0

100

200

Energy intensity (toe/MDh)

Butane gas (Mtoe)

49

0,0

300

400

500

600

700

800

GNP (constant 2012 MDh)

Fig. 4. log–log representation: electricity on left scale in blue and white combustibles on right scale in red. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Simulation of butane consumption with the IE-POW model. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

with a correlation coefficient of 99.1% despite the sharp regime change in consumption per capita at the end 1990s described earlier (Fig. 2). Indeed, it is difficult to distinguish, as in Fig. 3, a singularity between the periods before the PERG (before 1996) and that which followed it (after 2002). In our view, this can only be explained by an increment of GDP which the rural electrification program (PERG) had himself created, leaving this timeless representation in a continuous mode.

are obtained after the intensity I(E) is adjusted by the behavior I¼ β.(P  Po)n. Like shown in Fig. 4, the mathematical adjustment of the white fuels energy by the power function (n ¼0.137 and Po ¼0) is provided with a correlation coefficient of 97.1% justifying the choice of the power function with Po ¼0.

2.2.2.2. White liquid combustibles. White fuels accounted for 36.6% of total shares in 2011. With the growing shares of electricity, they stood behind electricity since 2002, in the second position of the final energy consumption in Morocco. White fuels are primarily used in transport though marginally in the agricultural pumping and generators: 1. Diesel which represents 29.2% of the total in 2011 has a calorific value of 1.01 toe per ton, 2. Premium gasoline which represents 3.7% of the total in 2011 has a calorific value of 1.07 toe per ton, 3. Aviation fuels accounted for 3.7% of the total in 2011 have a calorific value of 1.03 toe per ton, 4. Regular gasoline which supply disappeared since 2006 has a calorific value of 1.05 toe per ton. Fig. 5 shows the behavior of white fuels energy consumed in Morocco (blue diamonds related to the left scale) against GDP, as well as their partial energy intensity (red squares related to the right scale). Their masses were previously converted into toe with the above mentioned typical calorific values. The absence of a “plateau” in the curve I(P) suggests, also here, to avoid the exponential function model, so the simulations shown in the graph White combustibles (Mtoe)

Simulated

Intensity (toe/MDh 2012)

Simulated

2005

2000

4

2

1990

2

6

1985

1995

4 1980

Energy (Mtoe)

6

0

Energy intensity (toe/MDh)

8 2010

8

0 0

100

200

300

400

500

600

700

2.2.2.3. Butane. As being the third post of the final energy consumption in Morocco, butane represented 11.9% of the total in 2011. Fig. 6 shows the behavior against GDP of butane energy consumed in Morocco (blue diamonds related to the left scale) and simulation, after his energy intensity (red squares related to the right scale) has been adjusted by the exponential behavior I1{1  exp[  (P  Po)/α]}, given the shape of the curve and the appearance of a plateau. The correlation coefficient of the adjustment reached 99.7% for Po ¼234 GDh and I1 ¼2.69 toe/MDh. 2.2.2.4. Other energies. We have named “Other”, the remainder of the final energy (11.4% of total final energy in 2011) containing fuels, mainly for professional uses and heat production, which are therefore not used in electricity power plants: 1. Coal, which represents 4.35% of the total in 2011 with a calorific value of 0.66 toe per ton, 2. Fuel oil, which represents 5.64% of the total in 2011 with a calorific value of 0.99 toe per ton, 3. Propane, representing 1.16% of the total in 2011 with a calorific value of 1.10 toe per ton, 4. Natural gas, which represents 0.22% of the total in 2011 with a calorific value of 0.86 toe per m3. Fig. 7 shows the behavior, against GDP, of other final energies consumed in the country (blue diamonds related to the left scale). In view of the decreasing shape, the simulations shown in the graph (lines) are obtained after the intensity (red squares related to right scale) has been properly adjusted by the behavior I ¼ β. (P  Po)n. The correlation coefficient of the adjustment of these consumed “Other Energy” reached 92.9% for n ¼  0.504. 2.2.2.5. Comments. By observing the concavity (or convexity) of the blue curves of energies against GDP, it is easy to note that:

800

GNP (GDh 2012)

Fig. 5. Simulation of white fuels consumption with the IE-POW model. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

1. 23% of energy consumption in 2011, represented by the fuel thermal applications (sum of butane and “Other thermal”), is growing more slowly than GDP, 2. 77% of energy consumption in 2011, represented by the sum of

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Intensity (toe/MDh 2012)

2005

Simulated

1990

1,2

3,2 2000

1985

Energy (Mtoe)

1,6

4,8 4,0

1995

2,0

Simulated 2010

Other energies (Mtep)

2,4

2,4

0,8

1,6

0,4

0,8

0,0

Energy intensity (toe/MDh)

50

0,0 0

100

200

300

400

500

600

700

800

GNP (constant 2012 MDh)

Fig. 7. Simulation of “other energies” consumption with the IE-POW model. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

electricity and white fuels (mainly used for transport) continues to grow faster than GDP. Because of this, these two are the posts of energy that could endanger the energy efficiency of Morocco. 2.3. Correlative model based on a link to urbanization (URB model) This model considers that it is the combination of urbanization and the country’s revenue that lead to energy consumption. Not that the countryside is outside the scheme but that direct and indirect energy consumption drivers are in town. The model consists on considering that the energy consumption is proportional to urbanization rate and GDP in the form:

E = e(U x P m) where ε is a constant not depending on time and the exponent m are to be defined. In terms of mathematical method: the product of the urbanization rate and the constant GDP (U  Pm) is normalized to its value in the last year (2012) in order to have an abscissa stopping at 100%. Initially addressed to the energy work group “Prospective Morocco 2030”, Haut Commissariat au Plan (2007), this model was only recently published by Bennouna (2011). Here, it has been updated with later data to 2005. The scope of using such a model here is: 1. to check that it is still valid after updating, 2. to compare its extrapolations with the energy intensity model ones (chapter 2.2.2). Here, we could split the energy consumption only in 2 parts: electrical energy and the rest of the final energy consumption, referred to as “non-electric”.

Fig. 8. Correlation between (urbanization  GDPm) and both electric and not electric energies. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

related to the right scale) the linear relationship between nonelectric energy and the product of the urbanization of the country (in% of the population) by the constant GDP raised to the 0.758. The exponent required for linearization, was also adjusted so that the least squares straight line is optimal and the slope (10.121) matches the actual non-electric energy consumption at year 2012 With m ¼0.758, the correlation coefficient of the adjustment of the electricity reaches 99.0%. 2.4. Correlative model for electricity peak power Once a year, and for a short time (in summer, since several years now), the electric power demand of the Moroccan power grid reaches its annual maximum. This “peak power” is extremely important because, to avoid power cuts, local electric power capacities plus interconnections with neighboring countries must secure this power demand. Annual power peaks are directly correlated to the annual electric energy as shown in Fig. 9 which shows a linear relationship passing through the origin. The correlation coefficient of the fit of the peak power as a function of yearly electrical energy demand is therefore 99.5% with a coefficient 0.1779 MW/GWh per year will be used directly in our hereafter forecast. 2.5. Temporal pattern of GDP In order to make energy consumption forecast, a model for future GDP growth is indispensable. To do this, a retrospective model is first established as follows: 1. The variation of GDP over time t, P(t), is first smoothed, by a polynomial (6th degree in this case) to optimally adjust the

E = (Eelec ) + (Enon elec )

2.3.1. Retrospective Simulation of energy in Morocco (URB model) 2.3.1.1. Electricity. Fig. 8 shows (in blue diamonds related to the left scale) the linear relationship between electricity and the product of the urbanization of the country (in % of the population) by constant GDP raised to the power 1.284. The exponent required for linearization, was adjusted so that the least squares straight line is optimal and that the slope (29.97 TWh) matches the actual electric consumption at year 2012. With m ¼1.284, the correlation coefficient of the adjustment of the electricity reaches 99.70%. 2.3.1.2. Non-electric energy. Fig. 8 also shows (in red squares

Fig. 9. Linear correlation between yearly maximum power and electric energy consumption.

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Simulated 12%

4%

490 GDh

0%

370 GDh

-4%

250 GDh 1985

1990

1995

2000

2005

2010

-8% 2015

Fig. 10. Historical analysis of GDP. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

entire period 1980–2012 (blue diamonds related to the left scale of Fig. 10):

P (t ) =

∑i = 0 ci. t i

where ci are the seven constant coefficients of the polynomial. 1. Calculation of the average past growth Ap(t) is obtained from the logarithmic derivative, [dP(t)/dt]/P(t), of the above polynomial. The results are represented by the red curve related to the right scale of Fig. 10 which is itself given by:

A p (t ) =

∑i = 1 i. ci. t i − 1/ ∑i = 0 ci. t i

Fig. 10 also shows the actual annual growth (red squares related to the right scale) and explains why we have adjusted the GDP instead of the annual growth, with its huge fluctuations (mainly due to the dependence of the agricultural GDP to rainfall).

3. Results 3.1. The 3 scenarios of GDP forecasts A future evolution of the economic growth Af(t) is assumed to be parabolic:

Af (t ) = d0 + d1. t + d2. t ² where t is the time (year) and dj three constant coefficients which have to be determined with three different equations. Placing two constraints gives two continuity equations between Ap(t) and Af(t): 1. continuity of A(t) in year 2012: Af(t) ¼Ap(t), 2. continuity of its derivative in year 2012: [dAf(t)/dt] ¼[dAp(t)/dt]. The third equation comes from growth value hypothesis at year 2030. Three scenarios are developed for 2030:

Annual increase, HIG GNP

2 000 GDh

4,75%

1 800 GDh

4,50%

1 600 GDh

4,25%

1 400 GDh

4,00%

1 200 GDh

3,75%

1 000 GDh

3,50%

800 GDh

3,25%

600 GDh

3,00%

400 GDh

2,75%

200 GDh 1985

1990

1995

2000

2010

2005

Finally, the GDP evolution for each of the 3 future scenarios is

2020

2025

2,50% 2030

Fig. 11. Definition of three GDP scenarios used. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.) Table 1 Summary of the 3 GDP scenarios, figures in GDh. Scenario

Average of A 2012–2030

2010

2020

2030

LOW AVE HIG

3.8% 4.2% 4.6%

755 755 755

1163 1171 1178

1622 1727 1839

calculated and also shown in Fig. 11 (blue lines related to the left scale). These three scenarios have no other aim than to be in continuity with the past and to assess future energy consumption. However, as GDP data obtained represent the input for the following forecast, we summarize them in Table 1 (in GDh, in 2012 prices): 3.2. Energy forecast – IE models 3.2.1. Non-electric energy To avoid too much graphs, we only show the planned changes obtained with the medium scenario of the structure of non-electric energy obtained using models based on the adjustment of energy intensity (Fig. 12). Detailed numbers are shown in Table 2. Note that the share of white fuels in non-electric energy would increase from 59% in 2010 to nearly 66% by 2030 This, of course, unless the introduction of electric cars in mass to disturb that evolution would lead to transfer part of 9.4 Mtoe in 2020 or 14.6 Mtoe in 2030 to electricity which is already the heaviest final energy position, with all what that implies in revising to the electrical power generation plans. Because part of the rural Moroccan world is still in the process of replacing firewood by butane gas, this last is still growing of close to 1.6 kg/capita per year as shown by Bennouna (2011). Butane gas also enjoys great popularity because: White combustibles (Mtoe) [IE, AVE)

Butane gas (Mtoe) [IE, AVE)

Other energies (Mtoe) [IE, AVE)

100%

3,16

2,53

80% 60%

7,34

9,32

11,66

14,41

2015

2020

2025

2030

40% 20% 0% 2010

1. Af(t) increases up to 4.7% on 2030, the highest value of the past period (red dotted line related to the right scale of Fig. 11, scenario HIG) 2. Af(t) decreases down to 2.7%, the lowest value of the past period (solid red line related to the right scale on Fig. 11, LOW scenario), 3. Af(t) achieves 3.7% which is the average of the previous two extremes (discontinuous red line related to the right scale of Fig. 11, scenario AVG).

2015

2,90

610 GDh

GNP (GDh 2012), HIG GNP

Annual increase, AVE GNP

4,66

8%

GNP (GDh 2012), AVE GNP

Annual increase, LOW GNP

2,18

730 GDh

GNP (GDh 2012), LOW GNP

2,65

Annual increase

3,86

Simulated

2,41

GNP (GDh 2012) 850 GDh

51

Fig. 12. Evolution forecast of the structure of non-electric energy (IE model, AVG scenario).

52

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Table 2 Comparison of results obtained with the two models for the 3 scenarios.

110 TWh 100 TWh

30 Mtoe Electricity (TWh) [IE, LOW]

28 Mtoe

Electricity (TWh) [IE, AVE]

Simulation

Models GDP

2015

2020

2025

2030

90 TWh 80 TWh

White combustibles (Mtoe) Butane (Mtoe)

Other thermal (Moe)

Non-electric energy (Mtoe)

IE

IE

IE

IE

URB

Electricity (TWh)

IE

URB

Total final energy (Mtoe)

IE

URB

Energy intensity (toe/ MDh)

IE

URB

Electric peak power (GW)

IE

URB

LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG LOW AVE HIG

7.359 7.362 7.366 2.532 2.533 2.534 2.221 2.222 2.222 12.112 12.117 12.122 11.723 11.726 11.730 39.178 39.203 39.228 39.259 39.280 39.301 22.298 22.310 22.322 21.115 21.124 21.133 23.59 23.59 23.60 22.34 22.34 22.34 6.99 7.00 7.00 7.00 7.01 7.02

9.320 9.384 9.448 3.126 3.145 3.164 2.463 2.470 2.478 14.909 14.999 15.090 14.175 14.240 14.305 54.272 54.786 55.305 52.959 53.371 53.786 29.019 29.244 29.469 26.845 27.009 27.173 24.93 24.97 25.02 23.06 23.07 23.07 9.69 9.78 9.87 9.45 9.53 9.60

11.472 13.587 11.793 14.591 12.122 15.663 3.756 4.360 3.848 4.643 3.943 4.942 2.697 2.904 2.730 2.996 2.763 3.090 17.925 20.851 18.371 22.229 18.828 23.694 16.730 19.168 17.041 20.102 17.357 21.077 72.211 91.046 74.996 100.374 77.877 110.582 68.821 85.278 71.002 92.438 73.245 100.151 36.700 44.523 37.870 48.326 39.076 52.445 33.195 39.570 34.028 42.218 34.880 45.038 26.26 27.44 26.44 27.97 26.63 28.52 23.75 24.39 23.76 24.44 23.77 24.49 12.89 16.25 13.39 17.92 13.90 19.74 12.28 15.22 12.67 16.50 13.07 17.88

1. it is almost the exclusive energy source for cooking, 2. it is also used for irrigation water pumping, 3. above all, it is subsidized by the Compensation Fund when packaged in less than 12 kg per bottle. It is difficult to predict to what extent the forecast related to butane (Fig. 12 or others) would be amended without knowing the details of the expected subsidies redeployment Government plans. “Other thermal” energy needs represent about 2.5 Mtoe in 2020 and nearly 3 Mtoe in 2030. The natural gas plan, currently under design, Fédération Nationale de l’Energie (2012), has to be considered a substitute of part of “Other thermal”. 3.2.2. Electric and non-electric energies Fig. 13 shows the planned changes of both electrical and nonelectrical parts of the energy obtained by extrapolating the models based on the adjustment of energy intensity (non-electric being the sum of those represented in Fig. 12). Detailed numbers are shown in Table 2.

70 TWh

26 Mtoe

Electricity (TWh) [IE, HIG] Non-electric energy (Mtoe) [IE, LOW]

24 Mtoe

Non-electric energy (Mtoe) [IE, AVE]

22 Mtoe

Non-electric energy (Mtoe) [IE, HIG]

60 TWh

20 Mtoe

50 TWh

18 Mtoe

40 TWh

16 Mtoe

30 TWh

14 Mtoe

20 TWh

12 Mtoe

10 TWh 2010

2015

2020

2025

10 Mtoe 2030

Fig. 13. Evolution forecast of energies obtained with the IE model. 110 TWh 100 TWh 90 TWh 80 TWh 70 TWh

30 Mtoe Electricity (TWh) [URB, LOW]

28 Mtoe

Electricity (TWh) [URB, AVE] Electricity (TWh) [URB, HIG]

26 Mtoe

Non-electric energy (Mtoe) [URB, LOW]

24 Mtoe

Non-electric energy (Mtoe) [URB, AVE] Non-electric energy (Mtoe) [URB, HIG]

22 Mtoe

60 TWh

20 Mtoe

50 TWh

18 Mtoe

40 TWh

16 Mtoe

30 TWh

14 Mtoe

20 TWh

12 Mtoe

10 TWh 2010

2015

2020

2025

10 Mtoe 2030

Fig. 14. Evolution forecast of energies obtained with the URB model.

3.4. Maximum peak powers obtained by both IE and URB models Since November 2009, Morocco launched successively the Moroccan Solar Plan (PSM, 2 GW) and the Moroccan Integrated Wind Energy Plan (PMIEE, 2 GW) that will help to lower somewhat the energy dependence index of Morocco (above 95% from 1999 to 2008) using domestic renewable resources. But the intermittent nature of these sources can still not be relied upon facing the demand at peak hours (still in evening in the major part of the year). The best future we can hope for this renewable electricity is to be exported to Europe under any green electricity tariff, even importing “not-green” electricity in counterpart… cheaper, of course. Fig. 15 shows the evolution curves of the yearly electric peak power obtained using the linear correlation shown in Fig. 9 when combining the two models with the three scenarios (Fig. 13 and Fig. 14): 1. LOW: 15.2–16.2 GW in 2030, according to the adopted model, 2. AVG: 16.5–17.9 GW in 2030, according to the adopted model, 3. HIG: 17.9–19.7 GW in 2030, according to the adopted model. 20 GW Yearly maximum electric power (GW) [IE, LOW]

18 GW

Yearly maximum electric power (GW) [IE, AVE] Yearly maximum electric power (GW) [IE, HIG]

16 GW

Yearly maximum electric power (GW) [URB, LOW] Yearly maximum electric power (GW) [URB, AVE]

14 GW

Yearly maximum electric power (GW) [URB, HIG]

12 GW 10 GW

3.3. Energy forecast – URB model 3.3.1. Electric and non-electric energies Fig. 14 shows the planned changes in electrical and non-electrical parts of the energy obtained when extrapolating the urbanization-based model. Detailed numbers are shown in Table 2.

8 GW 6 GW 4 GW 2010

2015

2020

2025

2030

Fig. 15. Predictive evolution of power peaks arising from different models.

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

Actual energy intensity (toe/MDh) Simulated [IE, GNP AVE IES]

Simulated [IE, GNP LOW IES] Simulated [IE, GNP HIG IES]

Table 3 Summary of results with range for each scenario of economic growth.

28

Energy intensity (toe/MDh)

26 24 22 20 18 16 1985

1990

1995

2000

2005

2010

2015

2020

2025

53

2030

Fig. 16. Predictive evolution of overall energy intensity.

The values obtained are shown in Table 2. Extreme values shown in the document of the national energy strategy for 2030 offer a wider range within which lie all our results, namely: 1. 12 GW, for its lowest scenario, Ministère de l’Energie, des Mines, de l’Eau et l’Environnement (2009), is below our lowest model/ scenario combination, 2. 20 GW, for its disruptive scenario, Ministère de l’Energie, des Mines, de l’Eau et l’Environnement (2009), is above our higher model/scenario combination. The presentation of these results could not be avoided here. However, comparing these yearly electric peak power forecasts with the equipment plan provided by the concerned national institutions, alone accounts for another job… 3.5. Energy intensity Fig. 16 shows the planned changes in energy intensity with the IE models (Fig. 14) combined with the three GDP scenarios in (Fig. 11). Between 2010 and 2020, the 3 scenarios are similar and lead to an average increase in energy intensity of around 1% per annum. Between 2020 and 2030, the 3 scenarios lead respectively to annual average growth of 0.8%, 1% and 1.2% per annum.

4. Discussion IE models yielded higher values than those obtained by the URB model predictions. Failing to be affordable to all, most of the energy is becoming geographically available throughout Morocco and we do not see why a model based on urbanization (URB) would be perpetuated. So, even if it is clear that the URB model gives results close to the IE model, we would rather go in favor of the IE model which seems more “universal”. So, rather than sweeping the URB by this simple consideration, it is permissible to consider that both models (IE and URB) give the extremes of a range of values, which we present in Table 3 for each of our 3 scenarios. In Table 3, the total final energy is calculated by summing the non-electrical energy to electricity, previously converted in Mtoe based on the average thermal conversion factor of electricity in Morocco: 0.243 toe/ MWh (2011). In essence, the models remain consistent with the spirit of “continuity scenario”, called “S1”, of Morocco Energy Prospective 2030 of the Haut Commissariat au Plan (2007) but, five years later, the present work has helped to add a sensitivity analysis to economic growth scenarios starting from the year 2012. In the Haut Commissariat au Plan (2007) document:

Simulation

Model

Non-electric energy (Mtoe) Electricity (TWh)

IE and LOW URB AVE HIG

Total final energy (Mtoe) Energy intensity (toe/ MDh) Electric peak power (GW)

PIB

2015

2020

2025

2030

11.7–12.1 11.7–12.1 11.7–12.1

14.2–14.9 14.2–15 14.3–15.1

16.7–17.9 17–18.4 17.4–18.8

19.2–20.9 20.1–22.2 21.1–23.7

IE and LOW 39.2–39.3 53–54.3 URB AVE 39.2–39.3 53.4–54.8 HIG 39.2–39.3 53.8–55.3 IE and LOW 21.1–22.3 26.8–29 URB AVE 21.1–22.3 27–29.2 HIG 21.1–22.3 27.2–29.5 IE and LOW 22.3–23.6 23.1–24.9 URB AVE 22.3–23.6 23.1–25 HIG 22.3–23.6 23.1–25

68.8–72.2 71–75 73.2–77.9 33.2–36.7 34–37.9 34.9–39.1 23.7–26.3 23.8–26.4 23.8–26.6

85.3–91 92.4–100.4 100.2–110.6 39.6–44.5 42.2–48.3 45–52.4 24.4–27.4 24.4–28 24.5–28.5

IE and LOW 6.99–7.00 9.45–9.69 12.28–12.89 15.22–16.25 URB AVE 7.00–7.01 9.53–9.78 12.67–13.39 16.50–17.9 HIG 7.00–7.02 9.60–9.87 13.07–13.90 17.88–19.74

1. The forecasts made in 2007 of a total energy requirement to 42 Mtoe, 43 Mtoe or our previous 42.3 Mtoe, in Bennouna (2011), for 2030 are consistent with the results obtained with the medium scenario (see Table 3), 2. Former forecast of 95 TWh in electricity, by the Ministère de l’Energie, des Mines, de l’Eau et l’Environnement (2009) is consistent with our high scenario but our old forecast at 83.8 TWh, in Bennouna (2011), even based on economic growth of 4.4%, is now located near the low income growth scenario. Fig. 17 shows the evolution of the cost of weighted average of fossil fuels toe imported into Morocco (coal, and hydrocarbons). The data after 1993 have been downloaded from the Office des Changes (1993 to 2010) while earlier data are from El Ouadi (2002). The average curve of Fig. 17 started from a value oscillating around 1’240Dh/toe between 1986 and 1999 to multiply by 3 in the last decade and reach nearly 3’700 Dh/toe in 2010, an average growth rate near to 10% in current value and 8.5% in constant value! Even if GDP growth in Morocco has exceeded this value (6 times, all of them before 1997), its economy has never held such a stride. Thus, with unit fuel prices growth of 8.5% weighed by 0.8– 1.2% growth in energy intensity, it is difficult to imagine how the Moroccan economy can “absorb” such a net energy bill increase. Unfortunately, fuel costs are not depending on Morocco wishes, and without inventing the gunpowder, we have to emphasize here the urgent need for energy efficiency measures to absorb, at least, the impact of the increase in the energy intensity. The sum of white fuels and electricity accounted for 76.4% of the energy consumption of Morocco at the end of 2011. They must Net cost of imported energetic products (Dh/toe) 6 000 5 000 4 000 +9,3% per y

3 000 2 000 1 000 0 1985

1990

1995

2000

2005

2010

2015

Fig. 17. Costs, in current value, of energy products imported to Morocco.

54

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

represent the main target of any energy efficiency action because if they account for nearly three quarters of the final energy consumption, but their share is increasing respectively 1,531 and 1,137 times faster than GDP at constant prices (Fig. 4). Ideally, the energy as a function of constant GDP should become convex, or at least linear, to keep constant the energy intensity. In light of future new information, we plan, every five years, to re-adjust historical data and update the extended resulting forecasts.

5.3. Butane and the possible arrival of local natural gas


The need for butane should increase by between þ70 and

5. Conclusions and policy implications In essence, the prediction models adopted are not disruptive and based on macroeconomic correlations; they turn out finally rather simple, considering the real complexity of the problem. 5.1. Electricity


As shown in Table 3, electricity is the final energy which will have the higher increase between 2015 and 2030 (between þ130 and þ160% more). As Morocco cannot bet on the hydroelectricity load factor neither on its stability, to avoid substantial electricity imports during years of low rainfall. Even if the electrical equipment program for 2020 is known: 1. Five years before 2020, the stakeholders have still not clarified its continuation for 2020–2030, even if the forecasted average increase for the period lies between 4.9% and 7.2% yearly (calculated from extreme scenarios of Table 3). 2. Measures should be taken to avoid past delays in its implementation. In the last twenty years, each delay in the electric equipment program has forced to import more electricity (essentially from Spain through the Gibraltar Straight Euro-Moroccan grid interconnection) each year in which hydroelectricity could not cover the deficit.


As shown by Bennouna (2015), the global grid efficiency has lost approximately 2% in the twenty years, between the 1990s and the 2010s. The ideal would be to return to the situation of early 1990s, even with the larger grid. This fully justifies the allocations to improve electrical network performance included in the Program Agreement signed in 2014 between the Government and the National Utility. If the two missing points (88.5–86.5%) could be recovered it is, at current rate, almost 650 GWh of injected net electricity which could be saved, equivalent to: 1. Nearly 400 million Dh (44 MUS$) annually avoided coal electricity purchases from IPP, 2. The 2013 production of the 204 MW of public wind parks which cost nearly 3 billion Dh (330 MUS$). 5.2. White combustibles


The need for family of white combustibles should almost double between 2015 and 2030.


There is still no scenario considering a fast deployment of electric cars, if this happens significantly in the country in the next 15 years, electrical equipment programs should make provisions for this. The counterpart of this should reduce the need for white combustibles.




þ95% more from 2015 to 2030. It seems that the present combination of formal/informal distribution channels of 3 to 12 kg butane bottles is efficient enough to face such an increase. The most probable cause of the growth of butane needs is that these popular sizes of bottle are subsidized and, with the prices set for a long time (0.35 US$ per kg), the gap with other fuels widens. Not only to ease public funds, subsidies of butane must be removed as soon as possible to reduce the growth of its consumption and contribute to reduce energy intensity. Between 2012 and 2014, Morocco used 95% of its natural gas needs for electricity generation but produced less than 5.5% of it. Even if recent natural gas discoveries in the Gharb (about 40 ktoe per year) will change significantly this last figure, they are still small with respect to the national needs (21,000 ktoe). But if natural gas discoveries continue, it is not impossible to imagine, a natural gas network substituting butane gas bottles in big cities like Casablanca or others, then, some recommendations of the Symposium on Natural Gas by the Fédération Nationale de l’Energie (2012) should be very seriously considered. In this case, scenarios with partial substitution of butane by natural gas should be considered.

References Adom, P.K., Bekoe, W., Amuakwa-Mensah, F., Mensah, J.T., Botchway, E., 2012. Carbon dioxide emissions, economic growth, industrial structure, technical efficiency: empirical evidence from Ghana, Senegal, Morocco on the causal dynamics. Energy 47 (1), 314–325. Adom, P.K., 2015. Business cycle, economic-wide energy intensity: the implications for energy conservation policy in Algeria. Energy 88, 334–350. Apergis, N., Tang, C.F., 2013. Is the energy-led growth hypothesis valid? New evidence from a sample of 85 countries. Energy Econ. 38, 24–31. Bank Almaghrib, 2012. Rapports Annuels. Yearly editions. Bennouna, A., 2011. Monographie de l’énergie au Maroc. ISBN 978-9954-30-099-2. Edited by ISTICHAR. Marrakech, Morocco. BennounaA., 2015. L’efficacité du réseau électrique marocain. Unpublished. Bernardini, O., Galli, O., 1993. Dematerialisation: long term trends in the intensity of use of materials, energy. Futures 25 (4), 431–448. Bruns, S.B., Gross, C., 2012. Can declining energy intensity mitigate climate change? De-composition, meta-regression results. Max Planck Papers Econ. Evol. 1211, 1–10. Bruns, S.B., Gross, C., 2013. What if energy time series are not independent? Implications for energy-GDP causality analysis. Energy Econ. 40, 753–759. Chontanawat, J., Hunt, L.C., Pierce, R., 2008. Does energy consumption cause economic growth? Evidence from systematic study of over 100 countries. J. Policy Model. 30, 209–220. Coers, R., Sanders, M., 2013. The energy–GDP nexus; addressing an old question with new methods. Energy Econ. 36, 708–715. Direction de la Statistique, 1983–2010. Le Maroc en chiffres. Yearly editions. Eggoh, J.C., Bangake, C., Rault, C., 2011. “consumption, economic growth revisited in African countries. Energy Policy 39 (11), 7408–7421. El Ouadi, N., 2002. Ministère de l’Economie, des Finances de la Privatisation et du Tourisme DEFP, DPEG. Approche économique du secteur énergétique. Edited by Ministry of Finance, Rabat. Morocco. Fédération Nationale de l’Energie, 2012. Actes du colloque international sur le gaz naturel. Gabbasa, M., Sopian, K., Yaakob, Z., Zonooz, M.R.F., Fudholi, A., Asim, N., 2013. Review of the energy supply status for sustainable development in the organization of Islamic conference. Renew. Sustain. Energy Rev. 28, 18–28. Haut Commissariat au Plan, 2007. Prospective Maroc 2030: Quelles Options énergétiques pour le Maroc. Edited by HCP, Rabat, Morocco. Haut Commissariat au Plan, 2015. 〈www.hcp.ma〉. International Monetary Fund, 2010, World Economic Outlook Database. Kraft, J., Kraft, A., 1978. On the relationship between energy, GNP. J. Energy Dev. 3, 401–403. Mandelli, S., Barbieri, J., Mattarolo, L., Colombo, E., 2014. Sustainable energy in Africa: a comprehensive data, policies review. Renew. Sustain. Energy Rev. 37, 656–686. Mans, U., 2014. Tracking geographies of sustainability transitions: Relational, territorial aspects of urban policies in Casablanca, Cape Town. Geoforum 54, 150–161. Marktannera, M., Salman, L., 2011. Economic, geopolitical dimensions of renewable

A. Bennouna, C. El Hebil / Energy Policy 88 (2016) 45–55

vs. nuclear energy in North Africa. Energy Policy 39 (11), 4479–4489. Ministère de l’Energie, des Mines, de l’Eau et l’Environnement, 2009. Stratégie énergétique nationale-Horizon 2030. Assises Nationales de l’Énergie. Narayan, P.K., Popp, S., 2012. The energy consumption-real GDP nexus revisited: empirical evidence from 93 countries. Econ. Model. 29 (2), 303–308. Office des Changes, 1993–2010. Harmonized System of Statistics of External Exchanges of Morocco. Chapter no. 27. 〈www.oc.gov.ma〉. Office National de l’Électricité, 2012a. Brochures de chiffres clés, Yearly editions. Office National de l’Électricité, 2012b. Rapports annuels, Yearly editions. Omri, A., 2013. CO2 emissions, energy consumption, economic growth nexus in MENA countries: evidence from simultaneous equations models. Energy Econ. 40, 657–664. Ozturk, I., Acaravci, A., 2011. Electricity consumption, real GDP causality nexus:

55

evidence from ARDL bounds testing approach for 11 MENA countries. Appl. Energy 88 (8), 2885–2892. Tang, C.F., Abosedra, S., 2014. The impacts of tourism, energy consumption, political instability on economic growth in the MENA countries. Energy Policy 68, 458–464. US Department of Energy, 2006. International Energy Annual of the Energy Information Administration. Wolde-Rufael, Y., 2005. Energy demand, economic growth: the African experience. J. Policy Model. 27 (8), 891–903. Wolde-Rufael, Y., 2006. Electricity consumption, economic growth: a time series experience for 17 African countries. Energy Policy 34 (10), 1106–1114. Wolde-Rufael, Y., 2009. Energy consumption, economic growth: the experience of African countries revisited. Energy Econ. 31 (2), 217–224.