An Empirical Analysis of Energy Demand in Sub-Saharan Africa (1980-2014)

Ms Aisha Kolawole, Oxford Brookes University

This paper presents a comprehensive analysis of energy demand in Sub-Saharan Africa (SSA) by analysing demand functions for aggregated and disaggregated (by energy types) energy demand, in order to facilitate efficient demand management in the region. About 60% of the total population of 936.1 million in SSA lack access to electricity, with 30 countries in the region bedevilled by power shortages and rationing. The electrification rate and the average annual per capita consumption of power in the region is the lowest when compared to the other regions in the world. For instance, the electrification rate in other developing regions like South Asia and Latin America are 70% and 94%, respectively, whereas that of the Sub-Saharan Africa is 32% (IEA, 2012). Apparently, the demand for energy in most of the countries exceeds the available energy supply. The energy poverty is in sharp contrast to the abundant energy resources available in SSA (IEA Africa Energy Outlook, 2014; Onyeyi, 2014).

In the aggregate demand model which is unbalanced, the panel cointegration technique is used. The technique includes the panel unit root test, panel co-integration test and the Pooled Mean Group Estimator (PMG) is used to establish the short and long run relationships among the variables in the specified model. Evidence of mixed order of integration was found and also the variables move together (cointegrated) in the long run. Our result for the aggregated model shows that in the long run, income, urbanisation and energy prices are significant drivers of energy demand in the countries analysed. The coefficients of the variables have the expected signs predicted by economic theory and confirm that the demand for energy is inelastic.

For the disaggregated models, we have a balanced panel and the linear static models are used for the estimation. The energy types analysed are electricity, petrol, diesel, liquefied petroleum gas (LPG), kerosene and solid biomass. The Hausman selection test was used to choose between the random and fixed effects model, and the fixed effects model was chosen in most of the energy types. We carried out heteroscedasticity tests, auto correlation tests and the multicollinearity tests in the models. No evidence of multicollinarity and auto correlation was found, but there was heteroscedasticity. The Prais-Winsten heteroscedastic panel-corrected standard errors model was used to correct for this issue. Our results show that economic structure, urbanisation, population and income are the main significant drivers of the demand for the nergy types analysed. These factors have different signs and figures of estimated elasticities, but all is as we expected.

The estimated coefficients will guide academics, policymakers and other stakeholders on how to meet the demand, and also how to develop the energy markets in the countries analysed. We recommend a review of energy price subsidization policy in SSA, so as to free up capital for improving the refineries in the oil-rich countries, building infrastructure and amenities. Energy conservation policies and increased competition through the use of independent power companies to improve energy service delivery and markets in SSA are highlighted.