Renewable and Non-Renewable Energy Consumption on Economic Growth: Evidence from Asymmetric Analysis across Countries Connected to Eastern Africa Power Pool

Round 1

Reviewer 1 Report

Paper is interesting and has important contribution potential. However, many issues should be addressed. There is an especially very important problem with the empirical section. The following should be carefully addressed by the authors, very carefully. Then, the paper should be evaluated before publish. So, my decision is a major revision.

1.In the abstract, you followed abbreviations in a way that it necessitates Common to be common, Autoregressive to be autoregressive. Optional: you can also change them to Autoregressive Distributed Lag form, or the opposite, all in small letters.

2. Last part I pasted below from abstract is too general instead of highlighting contributions. Highlight nonlinear results, also, you can give examples from beta estimates. Policy recommendation could be shorter but also more direct. Make it shorter and emphasize the positive impacts of nonrenewables, focus on the returns of investments towards nonrenewables in the long-run and it's necessity for societies.

3.In the literature, emphasize other nonlinear and asymmetric literature approaches. Giving examples would be nice to state in a similar strand of literature on environment, nonlinearity is highly investigated with Markov Switching, STAR, neural networks approaches. Also I suggest to note petrol's impact in this relation.

4.An investigation of the relationship between the biomass energy consumption, economic growth and oil prices
Economic growth and CO2 emissions: an investigation with smooth transition autoregressive distributed lag models for the 1800–2014 period in the USA
Asymmetry in the environmental pollution, economic development and petrol price relationship: MRS-VAR and nonlinear causality analyses
The nonlinear relationship of environmental degradation and income for the 1870-2011 period in selected developed countries: the dynamic panel-STAR approach
Markov-switching vector autoregressive neural networks and sensitivity analysis of environment, economic growth and petrol prices

5.Empirical model:

a.There is a strong problem with the specification of models starting from Eq. 2 and following application section. I am sorry but this leads to a major revision. EC is the total of NREC and REC. These variables are "perfect" functions of eachother and leads to perfect multicollinearity suspicion. This should be corrected. Divide models into three. Make a model with EC, followed by second model with NREC and a third with REC. This is serious and don't avoid it please. Also, showing correlations and underlining that there is weak form of collinearity would not be followed in this correction. Still, it is highly suspicious. Do the revisions.

b.In Table 2, LNREN is written twice. For one variable, sample size is written as 28 for many countries. These must be typos to be corrected.

c.In CIPS, what is the lag selected? Make sure that you select them from IPS type tests in eviews, then in stata, give this lag to the test. It would lead better results if you did not do it. Because, we should make sure that those energy variables are I(0) or not. We prefer them to be I(1). Also, in data expalantion section, write their denotations.

d.You don't need to report individual test results. Remove table 6. If you don't want to do so, I suggest to put it in appendix since it is not necessary. Further, their statistic values are not given anyways. Instead, I suggest here edvancing the CIPS table. Add conventional UR tests, in addition, add Maddala Wu test which is also good for its properties under crosssectional dependence. By doing this, we confirm with robust methods the degree of integration for variables. Even though ardl is marketed as to work good with I(0) and I(1) variables, in fact, we maintain I(1) variables. With different tests, if there are results that series is I(1) and in some test I(0), under this we an use ARDL. This is the way in the original papers. But be careful, make sure that dependent variable is still I(1).

e.Dependent variable is determined a-priori as LnGDP. However, it should be made sure that there is only one cointegrating vector. This is not done in the paper. Same for the Westerlund test. Howeever, due to the correction sugggested above regarding instead of estimating all energy variables one by one, this test will be simplified. You should report ARDL F test statistic of cointegration in a form such as: Y I REN, K, L; Y I NREN, K, L. Example table is given in the literature I noted above as additional references.

f.Since all tables will be recalculated in terms of seperating EN, REN and NREN, correct interpretations in relevant places including the conclusion. Correct the IRF functions as well due to the change.

g.Asymmetric and Symmetric Tables are obtained by following a bivariate approach? If so, no need to update them.

h.Figure 4 should be changed if it will be necessary to do so after calculating new emprical results.    

For this paper, a major revision is needed. If the paper is published as is, it would be a real problem due to the econometric questionings. This problem is a well known problem at the foundation of econometrics theory. Corrections should be done carefully. Other than this, there are no problems such as Grammar and English. The paper will have important contributions after the corrections are made by the authors.  

Best Regards.



   

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This study explored the non linear relationship between economic growth and energy using NARDL and CCEGM approaches for nine East African countries and found  that nonrenewable energy has negative impacts while renewable energy has positive impacts in the region. There are some additional observations to improve the quality of the paper.

There is much need to establish the theoretical non linear relationship between energy consumption and economic growth. It is well known fact that energy consumption in developing countries increase the economic growth while deteriorating the environment so how this study is different from the already established facts. 

Presenting some numerical facts in the introduction will signify the study.

The authors are required to justify the selection of the sample and time period 0f 1980-2017.   

Why do authors need to apply the CCEMG and NARDL at the same while both tests depict some different features of data. The main focus of authors is to explore the nonlinear relationship then how CCEMG is justified?

The authors are required to put some explanation of findings of each table like table 2, 3, 4, 5.

There is need for justification to select the NARDL approach while more efficient approaches are there to estimate such type of data.

Some robustness tests are also required for significance of the empirical findings.

A separate section of discussion is necessary which will elaborate that how findings are consistent or inconsistent with earlier literature.

The authors should incorporate some latest studies on the topic. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

I appreciate the authors’ efforts in the work concerning renewable and non-renewable energy consumption on economic growth: evidence asymmetric analysis from Eastern Africa power pool countries. The paper is well structured and it provides an interesting analysis of the topic:

-        The information is relatively easy to navigate, and the structure of the paper allows readers to analyze the concepts approached.

-        The introduction and literature review sections provide a good background of the domain.

-        The authors bring relevant and interesting arguments to the investigated field.

However, to enhance the quality of the study, it would be wise to pay attention to several issues:

(1)   The paragrapgs from section 2: ”Relative studies in the countries connected at EAPP and EAPT” (lines 130 - 145) are simply copy-paste of the journal requirements. These must be eliminated and replaced with what the authors found regarding the studies in the countries connected at EAPP and EAPT.

(2)   Is the mathematical model the contribution of the authors? If not, references should be provided.

(3)   The authors should revise the manuscript for grammar and punctuation errors. For example: in the sentence: ”from nine countries(the time span is 1980-2017 )were used” in lines 507-508 there are some spaces misused.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Authors addressed critiques. Majority of them are corrected.  Most important was, concerns were raised for emprics which is not corrected by the authors. Suggestions were given to overcome the problem I mentioned in the last round. They addressed with stating that after transformations of logarithms and due to the method used, the problems are controlled. Sorry, but this is a very simple problem, that is, EC=NREC + REC. You cannot use them in a single model as explanatory variables simultaneously since all expalantory variables are perfect functions. For example, REC = EC-NREC. This is an basic econometric issue that leads to deviation of unbiasedness. Solution is not to include all in one model. Method is not robust to such situations and is incosistent under such situations. 

Though I do not understand the hesitation to estimate models seperately to solve the problem, I saw that there are examples in the literature following a similar approach. Therefore, my consideration is, my concerns still hold but I suggest another second best option. My suggestion is to discuss this problem and how it is solved by the method in the paper. How is the existing multicollinearity solved with the estimator that is not consistent under multicollinearity? If it is, then, they should present this explanation and discussion method should be explained with this respect. They can give examples from literature to strenghten their rebuttal against my suggestion to estimate seperate models with EC, NREC and REC seperately. However, this is a second-best option and still is not satisfactory. My suggestion is the first option I should suggest. If they do not want to do so, the second-best option can be considered. 

I regrest to inform that my decision is revision.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have incorporated the suggested changes.

Author Response

Thank you for your review！