Research gap and contribution of the study: Not many studies have been done on the relationship between electricity and economicgrowth in Zimbabwe. Most studies on the electricity and economicgrowth nexus in Zimbabwe have been carried out as part of regional studies, focusing on Zimbabwe as part of African countries (Wolde, 2006), Sub-Sahara Africa (Akinlo, 2008) and Common Market for Eastern and Southern Africa (COMESA), Nondo, et al (2010). The results from these studies have been conflicting. However, most studies reveal that the relationship runs from economicgrowth to electricity in Zimbabwe. These studies lack an in depth analysis of the electricity situation in Zimbabwe. In addition, using per capita electricityconsumption is not quite representative of the electricity that goes into production in Zimbabwe as a larger proportion of the country’s populace lives mainly in the rural areas with no access to electricity. The country’s electricityconsumption is mainly prioritised towards production activities in manufacturing, mining and agriculture. As argued by Wolde-Rufael, (2009) a detailed study on the relationship between electricity and economicgrowth is required in Zimbabwe.
For the case of multi-countries studies, Yoo (2006) conducted a study to examine the causal relationship between electricityconsumption and economicgrowth for four ASEAN countries namely Indonesia, Malaysia, Singapore, and Thailand. Empirical evidence indicated bi-directional causality between electricityconsumption and economicgrowth in Malaysia and Singapore. On the other hand, one-way causal relation was found from economicgrowth to electricityconsumption in Indonesia and Thailand. Similarly, Wolde- Rufael (2006) investigated the relationship between electricityconsumption and real GDP per capita (economicgrowth) for 17 African economies. Bounds testing approach to cointegration developed by Pesaran et al. (2001) was applied to examine the presence of long-run equilibrium relationship and the causality test suggested by Toda and Yamamoto (1995) was used to determine the direction of causality between the variables of interest. The results show that cointegration is only found in nine out of seventeen countries. However, causality analysis implies that electricityconsumption Granger-causes economicgrowth in Benin, the Democratic Republic of Congo, and Tunisia, while economicgrowth Granger- causes electricityconsumption in Cameroon, Ghana, Nigeria, Senegal, and Zimbabwe. Moreover, there exists bi-directional causal relationship between the variables in case of Egypt, Gabon, and Morocco. 1
In recent times, the world has experienced energy shortage. This phenomenon is due to the abrupt increase in global energy demand (Sekantsi & Okot, 2016; Tamba et al., 2017). This is because of the pivotal role energy (electricity) consumption plays in the stimulation of socioeconomic and economic activities of both developed and developing economies. The debate is still heated in the energy economics literature as to whether economicgrowth precedes energy consumption or vice versa. However, much has been documented in the energy economic literature for decades, mostly in developed economies. Little is known about this very interesting dynamic interaction in developing economies, more precisely in Sub-Saharan Africa (SSA). Thus, this current study focuses on Nigeria, which is faced with a huge and alarming electricity deficit. Recent statistics for the case of Nigeria reveal that an overwhelming 95,500,000 inhabitants of the population are without electrification, with 55 per cent of the total population without access to electricity while 45 per cent reside in urban centres and 63 per cent in rural areas (CIA, 2018). Given this backdrop, the country relies on load shedding to meet its electricity demand. Further, statistics shows that electricityconsumption rose from 13.72billion Kwh in 2000 to 24.57 billion KWh in 2018 (CIA, 2018).
The current study examines the existence of causation between electricityconsumption and economicgrowth of Togo for the period 1971-2009. Distinct from the study of Alinsato (2009), who utilised a bivariate analysis, the current paper approach the study in a multivariateframework by including capital formation, export and urbanisation rate as control variables. This is done to elude biased results caused by omission of relevant variables in causality test (Asafu-Adjaye, 2000, Lutkepohl, 1982; Stern and Cleveland 2004). Beyond the statistical advantage of multivariate approach over bivariate framework, the selected control variables are actually pertinent in the case of the relationship between electricityconsumption and economicgrowth in the country. For example, substantial portion of the available electricity is consumed in the urban centres. Roughly 80% of the peak demand for electricity, and the majority of total consumption occur in Lome (the capital city) and its environs. Moreover, the industrial and institutional sectors (chiefly concentrated in the urban centres) account for 31% and 15% of the country’s total electricityconsumption. The share of urbanisation population to the total population continues to grow as it was 5.302% in 1998 and 3.794% in 2009. Justifying the inclusion of export, the country is the fourth largest exporter of phosphate in the globe and a major exporter of cement. Two major companies in these sectors (International Fertilizer Group or IFG-Togo- and West African Cement Company or
4 Central Asia as it went through major power sector reform since it’s independence in 1991. In order to avoid bias caused by omission of relevant variables, trade openness is added to turn the study into a trivariate investigation. In practice, trade openness and electricityconsumption may individually have direct influence on economicgrowth. They may also serve as intermediate variables to each other, when impacting the economy. Economicgrowth may in turn also affect either electricityconsumption or trade openness. In case of Kazakhstan, inclusion of trade openness as a control variable is plausible as it enhances aggregate demand, which in effect causes electricityconsumption to grow.
 presented an adaptive linear, forward selecting time- series modeling technique to forecast load for space heating in buildings. It utilized ambient temperature, global radiation and wind speed as inputs to its model. The presented heat load forecasts in the study were used as input for the optimization of heat supply to buildings in smart grid applications. The recursive identification method for predicting parameters in electrically stimulated muscles was introduced by . The study improved output prediction at future times; hence, its application to predictive adaptive controllers. The adoption of multiple regression technique to develop simple energy estimation models for office buildings in five cities of China was presented by . The study analyzed weather conditions as they relate to energy use. The coefficient of determination R^2 was used to explain variations in energy use. The research estimated the likely energy savings to be obtained from analyzing data for different building schemes. The use of regression models using economic and demographic variables to develop a long-term consumption forecasting model was proposed by . The variables considered in the research were historical electricityconsumption, gross domestic product (GDP), gross domestic product per capita (GDP per capita) and population.  described the energy consumption of a supermarket in Northern England by means of a multiple regression analysis based on its gas and electricity data. As part of the study, the research utilized prevalent weather conditions such as temperature and level of building occupancy.
7 Our empirical investigation has two dimensions. The first is to examine the long-run relationship between carbon electricityconsumption and real GDP, while the second is to examine the short-run dynamic causal relationship between the variables. The basic testing procedure requires three steps. The first step is to test whether the variables contain a panel unit root to confirm the stationarity of each variable (Engle and Granger, 1987). This is done by using the Levin and Chu test, (LLC, 2002), the Im et al. test (Im, Pesaran and Shin (IPS, 2003)), the Augmented Dickey – Fuller test (F-ADF) (Maddala and Wu, 1999; Choi, 2001) and finally Breitung (2000) test. The second step is to test whether there is a long-run cointegrating relationship between the variables. This is done by the use of the Johansen-Fisher (Maddala and Wu, 1999; Kao, 1999; Pedroni, 1999, 2004) methods. Finally, the last step, if all variables are I(1) (integrated of order one) and cointegrated (Masih and Masih, 1996), short-run elasticities can be computed using the vector error correction model (VECM) method suggested by Engle and Granger (1987). In this case, an error correction mechanism exists by which changes in the dependent variables are modeled as a function of the level of the disequilibrium in the cointegrating relationship, captured by the error-correction term (ECT), as well as changes in the other explanatory variables to capture all short-term relations among variables (Pao and Tsai, 2010).
8 Where u it are residual terms and assumed to be identically, independently and normally distributed. The statistical significance of lagged error term i.e. ECT t 1 further validates the established long run relationship between the variables. The estimates of ECT t 1 also show the speed of convergence from short run towards long run equilibrium path in all models. The VECM is superior to test the causal relation once series are cointegrated and causality must be found at least from one direction. Further, VECM helps to distinguish between short-and-long run causal relationships. The VECM is also used to detect causality in long run, short run and joint i.e. short-and-long runs respectively in the following three possible ways: The statistical significance of estimate of lagged error term i.e. ECT t 1 with negative sign confirms the existence of long run causal relation using the t-statistic. Short run causality is indicated by the joint 2 statistical significance of the estimates of first difference lagged independent variables. For example, the significance of 22 , i 0 i implies that electricityconsumption Granger-causes economicgrowth and causality runs from economicgrowth to electricityconsumption can be inferred by the significance of 22 , i 0 i . The same inference can be drawn for rest of causality hypotheses. Finally, we use Wald or F-test to test the joint significance of estimates of lagged terms of independent variables and error correction term. This further confirms the existence of short-and-long run causality relations (Shahbaz et al. 2011) and known as measure of strong Granger-causality (Oh and Lee, 2004).
The BRICS countries, Brazil, Russia, India, China and South Africa are recognized as the most developed economies from the emerging countries. They have become an important force of the world economic stage. In recent years, the world energy consumption has risen rapidly driven by the slowing but still dynamic increase of energy consumption in Brics (Enerdata 2014). For instance China represented approximately 22 percent of the global energy consumption with Brazil lifting its world ranking to eight largest energy consumer. It was discovered that in 2013 Brics validated their increasing dominant role in the new global energy landscape with a share of approximately 40 percent compared to the 20 percent it experienced in 2000 (Enerdata 2014). Despite being the largest energy consumers, the Brics economies face challenges of greenhouse gas emissions. Most of the Brics countries rely on coal to meet their rising demand for electricity. Coal is regarded as the main contributor to the world energy demand growth and it is one of the main sources of carbon dioxide emissions. In the Brics group, Russia constitutes the largest contribution of fossil fuel on the energy matrix with 89.9 percent, followed by China which constitutes 86.8 percent. Brazil is the only one which stands out because it constitutes only 57.5% of fossil fuels in its energy mix. In terms of the carbon dioxide emitters, South Africa ranks 13 th
Several studies have been done on the linkage between the above key variables but up to now the area stills not well explored depending to countries’characteristics. Our work fills the void by extending the issue in three directions: (i) To assess whether the electricityconsumption per capita and economicgrowth per capita are cointegrated while trying to check if there is a long run relationship between these variables; (ii) To investigate the causal relationship between electricityconsumption and economicgrowth within a Vector Error Correction Model in a three panels of MENA countries 3 and also country-by-country.
Aims of the study were to critically examine the extent to which electricityconsumption influences economicgrowth in Ghana and also determine, if it is electricityconsumption that causes economicgrowth in Ghana or otherwise. The study employed Augmented Dickey-Fuller test, Cointegration test, Vector Error Correction Model and Granger Causality test. The study revealed that, in the long term, a hundred percent increase in electricity power consumption will cause real gross domestic product per capita to increase by approximately fifty-two percent. However, in the short run, electricityconsumption negatively affects real gross domestic product per capita. The study again revealed that unidirectional causality run from electricityconsumption to economicgrowth meaning that any policy actions taken to affect the smooth consumption of electricity in Ghana will definitely affect her gross domestic product per capita. Therefore, the current load shedding policy due to low supply of electricity will definitely affect the Ghanaian economy negatively, that is lower production levels, high inflation, and high rates of unemployment and lower standard of living. Therefore, the government of Ghana should invest massively into electricity infrastructure and conservation measures to meet the needs of the various sectors of the Ghanaian economy.
Third category includes the studies that have found bidirectional causality between energy consumption and economicgrowth. Soytas and Sari (2002) for Argentina, Paul and Bhattacharya (2004) for India, Yoo (2004) for Korea, Yoo (2005) for Malaysia and Singapore, Wolde-Rufael (2006) for Egypt, Gabon and Morocco and Lise and Montfort (2006) for Turkey have found bidirectional causality in their targeted countries. This category holds that higher energy consumption may lead to higher economicgrowth and then following a simultaneous bias effect, the higher economicgrowth may lead to even higher energy consumption and vice versa. Fourth category consists of studies that have found no causality between economicgrowth and energy consumption. Yu and Hwang (1984), Cheng (1996) for Venezuela and Mexico and Wolde-Rufael (2006) for Algeria, Congo Rep., Kenya, South Africa and Sudan have found no causality between the two variables. For this category, conservation or expansion of energy usage may not affect the economicgrowth. (The review of literature is summarized in Table 1)
Electricity is one of the basic elements of the daily routine of human’s life (i.e., personal usage to industrial production). Most often, it is claimed that the amount of electricityconsumption is directly attributed to the economicgrowth of a particular country. Due to the expansion of the global economy as well as the increase of income per capita, more demands are created for electrical-based equipment. Even in the rural area, people are eagerly connecting to the electric grid, gaining access to road transportation, and purchasing energy- used assets like electrical appliances and vehicles. These activities are also common for small, medium and large manufacturing industries, which are heavily relying on energy i.e., electricity, and thus contributing to the GDP. Given the possibility of the contribution of electricityconsumption on economicgrowth, many scholars have extensively conducted the study on electricityconsumption and economicgrowth nexus over the past two decades. Many studies have investigated the relationship between electricityconsumption and economicgrowth due to the complex links between these two variables (Ciarreta & Zarraga, 2010).
Furthermore, the total of electricity generation in MENA countries grew by an average of 6.3% per year. We depict in Figure 4 that hydro power grew slightly comparable to renewable electricity. The contribution from hydro is dominated in Egypt and Morocco (12%, 9.2%, respectively) and to lesser extent in Tunisia by 0.1%. It is also worth notable that non-hydro renewable electricity was concentrated in Egypt, Jordan, Morocco, Tunisia and Turkey (i.e. 0.8%, 0.5%, 2.0%, 0.3%, 0.3%, respectively). Algeria, Saudi Arabia and UAE don’t report any non-hydro generation. It also worth notable that the final energy used in MENA region differs per country due to the combination of a Mediterranean climate among North Africa (Algeria, Morocco and Tunsia) where space heating demand is common, i.e. the demand consists to a large extent for food production, especially during the winter season. However, in Middle East countries which are distinguished during a desert climate (especially, Oman, Saudi arabia and the UAE), the demand is absent, although a small share of domestic hot water.
3 Oil prices are a key component of energy, and their importance in economic development has been recognized by economists, policy makers, businessmen, households, and researchers. After the 1973 oil crisis, several studies (Timilsina 2015, Kilian and Vigfusson 2011, Kilian 2008, Hamilton 1983, 1985, Gisser and Goodwin 1986, Mork 1989) affirmed an inverse relationship between oil prices and economicgrowth. Economists and researchers have reached a consensus that oil price volatility simultaneously reduces economicgrowth. However, the recent literature shows the negative relationship decreasing over time because of oil alternatives and preemptive governmental measures against sudden oil price shocks (Doroodian and Boyd 2003, Jbir and Zouari-Ghorbel 2009). Oil-importing developing economies are severely affected by oil price hikes because of a lower tax share on oil prices. Moreover, developed economies have a higher tax share on oil. Therefore, such oil price shocks may be mitigated to an extent by suspending the tax share as oil prices rise. Developing countries with less of a tax share on oil have less ability to absorb oil price shocks. Consequently, oil price hikes appear to have a more adverse impact on developing economies.
Cointegration implies the existence of Granger causality, however, it does not point out the direction of the causality relationship. Granger (1988) emphasizes that a vector error correction (hereafter VEC) modeling should be estimated rather than a VAR as in a standard Granger causality test, if variables in model are cointegrated. Following Granger (1988), to test for Granger causality in the long-run relationship, we employ a two step process: The first step is the estimation of the long-run model for Equation (1) in order to obtain the long-run relationship as error-correction term (ECT) in the system. The second step is to estimate the Granger causality model with the variables in first differences and including the ECT in the systems. In our case, the VEC multivariate systems take the following forms:
2. Whether or not electricityconsumption positively affects and causes GDP, the relationship is crucial for electricity conservation policies (Narayan and Smyth, 2005b; Ghosh, 2002). If a positive unidirectional causality running from electricityconsumption to GDP does not exist then this provides a basis for elec- tricity conservation policies, such as electricity rationing. In the absence of this causal relationship, the implication is that a country does not depend on electricity for growth and development. If a unidirec- tional causality runs from electricityconsumption to GDP then reducing electricityconsumption could lead to a decrease in economicgrowth. This implies that a negative shock to electricityconsumption leads to higher electricity prices or electricity conservation policies and have a negative impact on GDP (see Narayan and Singh, 2007). Payne (2010) empha- sized bivariate causality tests results. However, a com- mon problem associated with bivariate analysis is the possibility of omitted variable bias, which draws into question the validity of the inferences of a causal rela- tionship. Furthermore, with the exception of the stud- ies by Wolde-Rufael (2006), Squalli (2007), and Tang (2008), the majority of the studies do not examine the coefficients with respect to both the sign (positive or negative) and the magnitude of the relationship between electricityconsumption and economicgrowth In Table 3, results of the studies in the litera- ture are presented. According to the results, 28.10% of the studies supported the neutrality hypothesis; 20.26% of the studies supported conservation hypothesis; 33.01% of them supported the growth hypothesis; and 18.62% of them supported the feed- back hypothesis. When these rates are examined in subcategories of developing and developed countries, which involves 65 and 90 countries respective- ly,10.7% of the studies for developed countries and 30% of the studies for developing countries support the conservation hypothesis, 24.6% of the studies for developed countries and 27.7% of the studies for developing countries support the growth hypothesis, 10% of the studies for developed countries and 20% of the studies for developing countries support the feedback hypothesis and 53.8% of the studies for developed countries and 22.2% of the studies for developing countries support the neutrality hypothe- sis. According to Payne (2010); the results for the 74 specific countries surveyed show that 31.15% sup- ported the neutrality hypothesis; 27.87% the conser- vation hypothesis; 22.95% supported the growth hypothesis; and 18.03% supported the feedback hypothesis.
The intent of the study is to re-examine the relationship between per capita electricityconsumption and per capita Gross Domestic Product (GDP) from 1971 to 2014. By employing the Granger causality test, the study found that there was absence of a long- term equilibrium relationship between per capita electricityconsumption and per capita GDP in India, but the existence of unidirectional causality running from per capita GDP to per capita electricityconsumption was reported in a Vector Autoregression (VAR) framework. The results indicate that the policymakers should encourage energy conservation measures on both the supply and demand-side which will lead to sustainable energy supply in the country. This will lead to a sustainable energy supply in the country. Moreover, if the government in collaboration with the power utility industries frames the appropriate national policy on energy conservation for entering the practical action, it will enhance economic development on a sustainable basis.
The ARDL model involves two steps in order to estimate the long- run relationships between variables. The first step involves investigating the existence of long-run relationships among the variables in the equations under estimation. The literature on ARDL model indicates that the calculated ARDL F-statistic is sensitive to the selection of lag length in the model (Shahbaz et al., 2012; Bahmani et al., 2000). After regression of Equations (1-2), the Wald test (F-statistic) was computed to differentiate the long-run relationship between the concerned variables. The Wald test can be carried out by imposing restrictions on the estimated long-run coefficients of economicgrowth and electricityconsumption. The computed F-statistic value will be evaluated with the critical values tabulated in Table CI (iii) of Pesaran et al. (2001). According to Pesaran et al. (2001), the lower bound critical values assumed that the explanatory variables are integrated at order zero, or I(0), while the upper bound critical values assumed that variables are integrated at order one, I(1). If the computed F-statistic is smaller than the lower bound value, then the null hypothesis is not rejected and we conclude that there is no long-run relationship between electricityconsumption and economicgrowth variables. Conversely, if the computed F-statistic is greater than the upper bound value, then growth-electricity nexus has a long-run relationship. On the other hand, if the computed F-statistic falls between the lower and upper bound values, then the results are inconclusive (Davoud et al., 2013). The Null hypothesis of no cointegration (H 0 ) and the alternative
Interestingly, our results show also that Iran in energy exporting countries and Turkey in energy importers are leaders in terms of the intensity of interaction between energy usage and economicgrowth (see Table 7). This may be mainly due to a good structuring of the electricity sector that leads necessarily to a positive and significant effect on economicgrowth (EIA, 2009). The favorable position of Iran and Turkey in comparisonto other countries of our set sample leads to an essential recommendation which is the reorganization of the electricity sector. This latter policy can be a useful and valuable tool of our considered economies yields slowly in each country, especially under the current energy crisis. It is also crucial to identify clearly the determinants of electrical energy demand in order to better understand changes in electricityconsumption in recent years while it hardly reflects the economicgrowth of MENA countries either in energy importers or in energy exporters (see Figure 2).