استقلال علی بین مصرف انرژی و رشد اقتصادی در لیبریا: شواهد از آزمون علیتی بوت استرپ غیر پارامتری

This contribution investigates causal interdependence between energy consumption and economic growth in Liberia and proposes application of a bootstrap methodology. To better reflect causality, employment is incorporated as additional variable. The study demonstrates evidence of distinct bidirectional Granger causality between energy consumption and economic growth. Additionally, the results show that employment in Liberia Granger causes economic growth and apply irrespective of the short-run or long-run. Evidence from a Monte Carlo experiment reveals that the asymptotic Granger causality test suffers size distortion problem for Liberian data, suggesting that the bootstrap technique employed in this study is more appropriate. Given the empirical results, implications are that energy expansion policies like energy subsidy or low energy tariff for instance, would be necessary to cope with demand exerted as a result of economic growth in Liberia. Furthermore, Liberia might have the performance of its employment generation on the economy partly determined by adequate energy. Therefore, it seems fully justified that a quick shift towards energy production based on clean energy sources may significantly slow down economic growth in Liberia. Hence, the government’s target to implement a long-term strategy to make Liberia a carbon neutral country, and eventually less carbon dependent by 2050 is understandable.

Since the 1970s oil price shocks and the seminal work of Kraft and Kraft’s (1978), a number of studies have attempted to examine the causal relationship between energy consumption and economic growth in both developed and developing countries. This interest has not only been fueled by the increasing economic activities across countries which have triggered a growing demand for energy across the world, but also the notion that energy prices directly affect spending decisions of households, firms, and the overall performance of the economy. In fact, research examining the causal relationship between energy consumption and economic growth should be of interest to both economists and policy makers due to the significant implications for governmental energy policy (for example see Narayan and Smyth, 2005). However, the question one may ask is which variable causes the other? Does the use of energy lead to economic growth or does economic growth lead to energy consumption? The literature regarding the causal relationship between energy consumption and economic growth includes studies for OPEC countries, G7 countries, Asian countries, African countries, etc. In his study “A literature survey on energy-growth nexus”, Ozturk (2010) noted that the literature produced conflicting results and there is no consensus, neither on the existence nor on the direction of causality between energy consumption (electricity consumption) and economic growth. In fact, there are currently four views. The first view preaches a unidirectional causality running from economic growth to energy usage and that as the economy grows the demand for energy from different sections of the economy increases. This means that the economy is non dependent on energy and that a continuous increase in GDP may imply a permanent increase in energy consumption. The second view, however, argues a unidirectional causality running from energy usage to GDP growth which means that it is the consumption of energy that causes economic growth. Better still, the third argument emphasize a bi-directional causality relationship between energy usage and economic growth. This means that both energy consumption and economic growth cause each other. Contrary to these three views which support some form of causal relationship between energy consumption and economic growth, the fourth and less probable view contends that there is no interdependence between energy consumption and economic growth. In other words, both energy consumption and economic growth are neutral with respect to each other. Like developed and other developing countries, Africa is no exception to the above debate. In fact, recent empirical studies on the relationship between energy consumption and economic growth in African countries failed to reach a consensus as to the direction of causation. What is even more shocking is that, up to now and to the best of our knowledge there is no extensive study about links between growth and energy consumption for the oldest independent African country currently in transition – Liberia. Recognition of energy usage – economic growth links is very important for countries in transition like Liberia, which experienced recent years of civil war and a slow-down in economic growth. It therefore seems fully justified to examine in particular whether the economic growth in Liberia depends on the level of energy consumption. Answering this question is crucial not only for the Liberian energy policy but also for the energy policies of other African countries in the context of Africa’s goal of a substantial reduction of greenhouse gas emissions. In the literature, the most popular approaches for establishing causal interdependence between variables have been the traditional parametric approaches (like the traditional Granger and the Toda and Yamamoto approaches) which are based on asymptotic theory. As a matter of fact, all studies on causal independence in the context of African countries have used the traditional parametric approach. While traditional approach to testing for Granger causality is appealing, since the test reduces to determining the significance of the coefficients of the terms in the regression that depend on past and current values of the dependent variable, it can be said however that the methods describe above have several drawbacks. First, parametric tests require modeling assumptions (such as linearity of the regression structure, etc.) and hence, the application of asymptotic theory may lead to spurious results if suitable modeling assumptions do not hold (see Hacker and Hatemi, 2003). Second, even if all modeling assumptions are generally fulfilled, it is now known that the distribution of the test statistic may still be significantly different from an asymptotic pattern when dealing with extremely small samples (see Gurgul and Lach, 2012), a situation which is typical of most countries including African countries. Third, parametric tests based on prediction errors will be sensitive only to causality in the mean, while higher order structure such as heteroskedasticity, will be ignored (see Diks and DeGoede, 2001 and Diks and Panchenko, 2006). Hence, this suggests that in order to obtain more robust estimates of causality would require implementation of a more general testing method which is sensitive also to nonlinear causal relationships and overcomes the above difficulties. In order to examine the Granger causality relationship between energy consumption and economic growth in Liberia, this paper employs an alternative testing method which departs from standard traditional approaches. Efron (1979) pioneer a bootstrap testing technique which is less sensitive to possible model misspecification such as neglected nonlinearity, time series properties and size distortion. These features make application of the non-parametric bootstrap technique particularly attractive for a country like Liberia not least because of the need to adopt a different testing technique (as suggested by Ozturk, 2010) but largely because of the relatively small sample size (only 29 observations in our case) as well as evidence of heteroskedacity in the data (as revealed by White’s hetereskedacity test). Unlike parametric approaches, the method is used for estimating the distribution of a test statistic by resampling the data non-parametrically. Since the estimated distribution depends only on the available dataset, it may be reasonable to expect that the bootstrap approach does not require such strong assumptions as parametric methods. Even though this approach has been shown to reduce size distortions (as we also show in this paper) and is believed to provide more precise test inferences than the asymptotic method in many applications especially when the available sample size is small (Hacker and Hatemi, 2003, Hacker and Hatemi-J, 2006, Mantalos, 2000, Horowitz, 1994 and Mantalos and Sukur, 1998), our paper will be the first application of the bootstrap technique to any African country. Summarizing the main results, we find evidence in favor of bidirectional Granger causality between energy consumption and economic growth in Liberia. In addition, the results show that employment in Liberia Granger causes economic growth. The results apply irrespective of whether the causality is estimated in the short-run or in the long-run formulation. To explore the potential benefit of using a bootstrap method in estimating the causal links between energy consumption and economic growth in Liberia, we examine its test size and power properties1 relative to the asymptotic method. To this end, we conduct a Monte Carlo experiment to investigate whether the asymptotic test has larger size distortion than the bootstrap test in this application. Results from the experiment show that the asymptotic test has larger size distortion than the bootstrap test thus implying that the asymptotic Granger causality test suffers size distortion problem for Liberian data. Hence, this suggests that the bootstrap methodology employed in this study is more appropriate. The rest of this paper is organized as follows: Section 2 presents an overview of the energy sector in Liberia. A review of the literature on energy consumption and economic growth in African countries is given in Section 3. Section 4 describes the traditional parametric and bootstrap Granger causality approaches. Section 5 provides a summary statistics of the data and the empirical results. Section 6 concludes the paper providing some policy implications.

The purpose of this paper is to investigate causal interdependence between energy consumption and economic in Liberia. In so doing, we propose the application of a bootstrap test. Unlike traditional approaches, the method is used for estimating the distribution of a test statistic by resampling the data non-parametrically. Since the estimated distribution depends only on the available dataset, it may be reasonable to expect that the bootstrap approach does not require such strong assumptions as parametric methods. Our research was conducted for primary energy consumption. In order to reflect the causality between energy consumption and economic growth properly, we performed our investigations in a three-dimensional framework with employment chosen as an additional variable. The empirical applications to reliable Liberian data (1980 to 2008) document significant improvement in the size and power of the non-parametric bootstrap test over the asymptotic test. The main findings are: First, there is evidence of distinct bidirectional Granger causality between energy consumption and economic growth. In addition, the results show that employment in Liberia Granger causes economic growth. The results apply irrespective of whether the causality is estimated in the short-run or in the long-run formulation. Second, the long run estimates from the ARDL model reinforce the Granger causality tests by indicating that both energy consumption and economic growth are positively correlated in the long-run. Third, evidence from a Monte Carlo experiment revealed that the asymptotic Granger causality test suffers size distortion problem for Liberian data, and hence suggesting that the bootstrap methodology employed in this study is more appropriate. This finding is supportive of the hypothesis that asymptotic distribution theory performs better for longer time series. An important point to consider when analyzing the results in this paper is that in recent years, the share of the manufacturing and service sectors in Liberian GDP and employment has risen considerably. Thus, it is not surprising that increasing energy consumption in these sectors was significantly related to the economic growth of Liberia. The implication then is that energy expansion policies like energy subsidy and low energy tariff for instance, would be necessary to cope with demand exerted as a result of economic growth in Liberia. Furthermore, Liberia might have the performance of its employment generation on the economy partly determined by adequate energy. On the contrary, energy conservation measures would harm the economic growth in Liberia and should not be endorsed now. A quick shift to clean energy would be an unreasonable consideration to make. In light of the empirical results, it seems fully justified that a quick shift of the Liberian economy towards energy production based on clean energy sources may significantly slow down the economic growth of the country. In view of the above, the government’s target in the National Energy Policy to implement a long-term strategy to make Liberia a carbon neutral country, and eventually less carbon dependent by 2050 is understandable. The repetition of all calculations in this study on the basis of including capital and energy price is likely to be an interesting research avenue for the future.