Institutions and Geography: A "Two Sides of the Same Coin" Story of Primary Energy Use in Sub-Saharan Africa.

• Author: Lawson, Late Ayao

1. INTRODUCTION

   The acceleration of ecosystem depletion and gas emissions has motivated systematic studies on the socio-economic drivers of natural resources and fossil energy use. This is the case in recent studies by Medlock and Soligo (2001), Wolde-Rufael (2009), Van Benthem and Romani (2009), Ozokcu and Ozdemir (2017) and Antonakakis et al. (2017), among others, on the energy use and economic development nexus. The present paper offers a new perspective into primary energy use characteristics of SSA countries by investigating an issue which has surprisingly received less attention from the literature so far: The role of political institutions and geographical characteristics in primary energy use.

   Existing works on energy use predominantly focus on the income-energy link, analysing causality, long-run dynamics and the so-called Environmental Kuznets Curve hypothesis for energy use. Further aspects question the drivers of energy demand and the channels through which energy availability enhances economic development. In this scholarship, relatively rare mentions of the role of institutional and geographical characteristics of countries (location, weather differentials and spatial spillovers) can be identified. Aiming to fill that gap, our paper analyses how institutions and geography influence energy consumption in SSA, accounting for usual socio-economic characteristics such as income and urbanization level and population dynamics, among others. Thereby, we argue that neighbouring SSA countries sharing similar geographical characteristics also show comparable profiles of energy use. This relation is also enhanced by institutional quality.

   SSA countries being pre-industrial and highly resources-dependant economies, investigating topics relative to endowments in energy resources and primary energy use as well as related environmental consequences seems pertinent. Moreover, the "Geography versus Institutions" debate mainly animated by Acemoglu et al. (2001, 2005, 2008), who stressed the role of institutions, against Sachs (2003), Gallup et al. (1999); Sachs et al. (2001) and Diamond (1999), who supported the predominance of geographical characteristics, points out the importance of both factors in economic development. Hence, institutions and geography being among factors explaining economic growth, they likely play a similar role in energy consumption. In this perspective, introducing geographical factors and political institutions in a study of the determinants of primary energy use in SSA appears promising.

   The contribution of our spatial analysis of primary energy use is twofold. Firstly, observing Figure 1, we can claim that location matters to endowments in fossil energies as well as in primary energy use, since high energy use is mostly observed in coastal located countries. Additionally, looking from the South to the North, it appears that countries with lower energy use such as Chad, the Democratic Republic of Congo, Mali, and Ethiopia are mostly surrounded by countries with higher levels of energy consumption. As coastal and geographically contiguous SSA countries show comparable energy consumption levels, there might be some geographical spillovers in primary energy use and this analysis intends to consider that. Secondly, since SSA countries are being classified among the fastest growing economies, institutional and geographical factors seemingly play a significant role in their economic performances and therefore in energy use. Thus, our analysis proposes to account for institutions and geography, arguing that both factors also affect primary energy consumption, acting as the "two sides of the same coin".

   The outline of the paper is as follows. Sections 2 and 3 present the related literature and the data. Section 4 comprehensively describes our econometric approach. Section 5 presents and discusses the estimation results. In Section 6, we provide some robustness check. Section 7 concludes the paper.

2. RELATED LITERATURE

   The empirical literature on energy demand and economic development seems large in terms of contribution and methodological approach. Our paper broadly addressing the determinants of energy use, this literature review focuses on causality and long-run dynamics analyses, the energy-income nexus as well as the socio-economic drivers of energy use.

   Being input in production activities, energy use reversely depends on income level. In addition to investigating the direction of causality, a first group of authors contributes by estimating the long-run relationship between energy and income. This is the case in Glasure and Lee (1998), Asafu-Adjaye (2000), Soytas and Sari (2003), Altinay and Karagol (2004), Lee (2005, 2006), Huang et al. (2008), Joyeux and Ripple (2011), Tang and Tan (2012) and Omri (2014), among others. Reviewing this literature, a lack of unanimity regarding the direction of the causality is noticeable (Ozturk, 2010), motivating systematic meta-analyses (Menegaki, 2014; Bruns et al., 2014; Sebri, 2015), which lead to even ambiguous conclusions.

   In the SSA context, the empirical results do not contrast with the previous ones. At country level, the results by Odhiambo (2009a) point to a stable long-run relationship with a unilateral causality running from energy to GDP in Tanzania, while in South Africa there seems to be a bidirectional causality (Odhiambo, 2009b). The conclusions by Ebohon (1996), Wolde-Rufael (2009) and Esso (2010) partly support this bidirectional causality. Akinlo (2008) provides country-level analyses to find a bidirectional causality in Gambia, Ghana and Senegal, while no causality is observed in Cameroon, Cote d'lvoire, Nigeria, Kenya and Togo, similarly to Dogan (2014) in the case of Benin, Congo and Zimbabwe. In the West African Economic Community, Ouedraogo (2013) claims a causality running from GDP to energy, while Kebede et al. (2010) point out regional disparities in energy demand and Wesseh and Lin (2016) work out the role of capital, labor, renewable and non-renewable energies in driving economic performance in Africa. It is to observe that these works on SSA countries are characterized by very limited samples in addition to providing less evidence of an EKC.

   In a slightly different perspective, researchers analysed the energy-income relationship, challenging the existence of an Environmental Kuznets Curve (EKC) for energy. (1) This is the case in empirical works by List and Gallet (1999), Nguyen-Van (2010) and Antonakakis et al. (2017), among others. While conflicting, this literature largely does not support the EKC hypothesis. (2) Thus, in thorough discussions of stylised facts on the energy-income nexus from 1971 to 2010, Csereklyei and Stern (2015); Csereklyei et al. (2016) find a stable increasing relationship over the last four decades coupled with a decreasing energy intensity of GDP (Ruhl et al., 2012). Definitely, besides global convergence in energy intensity, there seems to be no doubt that increases in energy consumption are largely driven by economic growth.

   Further relevant aspects of the literature concern drivers of energy use and the channels through which energy availability enhances GDP growth. On the latter point, Toman and Jemelkova (2003) and Birol (2007) argue that energy availability supports the improvement of health and education system and also increases productivity in industry and agriculture.

   Regarding potential drivers of energy use, Medlock and Soligo (2001) explore the role of economic sectors to conclude that industrial, transportation and residential energy demand substantially increases at early stage of economic development. Metcalf (2008); Van Benthem and Romani (2009) note end-use prices, the latter authors pointing out the role of agricultural and residential sectors in increasing energy demand in developing countries. Liddle (2013) assesses the relationship between economic growth, urbanization and energy consumption and finds results supporting a long-run relationship between these phenomena. Similar results are observed in the recent study by Dogan and Turkekul (2016), where a cointegration relationship appears between energy, urbanization and trade openness. Finally, in a systematic analysis of the determinants of energy consumption, Azam et al. (2016) mention the significant role played by income level, trade, urbanization, population growth and foreign direct investment. (3) Besides the latter socio-economic drivers, very few are research papers addressing the role of institutional and geographical characteristics in energy consumption.

   Based on this literature review, recurrent determinants such as per capita GDP, industrial and agricultural production, urbanization, population growth, trade openness and foreign direct investments (FDI) should be accounted for in our empirical analysis.

3. DATA AND DESCRIPTIVE STATISTICS

   Assessing how political institutions and geographical characteristics affect primary energy use, we intend to control for the most recurrent energy drivers identified in existing studies along with some SSA contextual elements derived from the comparative development literature. For this purpose, we have drawn socio-economic variables from the World Development Indicators, data on primary energy use from the U.S. Energy Information Administration and indicators of political institutions from the Worldwide Governance Indicators (WGI). Due to missing values, the sample is reduced to 42 SSA countries observed between 1990 and 2013.

   Dependent variable: As indicator of energy consumption, this study exploits the total primary energy use expressed in kilo-joules (Kj) per capita. It is to admit that such a synthetic measure of primary energy consumption does not provide any information concerning its composition or renewable structure. However, it serves as a good proxy for fossil and biomass energy use across SSA.

   Explanatory variables: Considering socio-economic...