How Does Energy Misallocation Affect Carbon Emission Efficiency in China? An Empirical Study Based on the Spatial Econometric Model
Round 1
Reviewer 1 Report
See the comments in the attached file.
Comments for author File: 
Comments.pdf
Author Response
Response to Reviewer 1 Comments
Point 1: In line 141, 150 and in formula (1) and (2) is not was clear what is being represented by Ei and E.
Response 1: Added: Ei represents the energy consumption of region i, E represents the total energy consumption.
Point 2: In line 161, it is referred that “...the C-D production function...” will be use. It is not explained in what consists this function.
Response 2: Added: Yit in Equation (3) represents each year’s total output in each of the provinces, Kit, Lit and Eit represents each year’s capital, labor and energy input in each of the provinces, βKi, βLi and βEi represents the contribution proportion of capital, labor and energy to the output in each provinces.
Point 3: In formula (3), line 166, Yit is now representing which variable?
Response 3: Yit in Equation (3) represents each year’s total output in each of the provinces.
Point 4: Line 169: Taking the logarithm of both sides of equation (3) is not equivalent to equation (4), as indicated; In equation (4) what is represented by the terms μi, λt and εit?
Response 4: Taking the logarithm of both sides of equation (3) is equivalent to equation (4). This equation has been adopted (Bai et al., 2018 and Pu et al., 2015). μi in equation (4) is the individual effect, λt is the time effect and εit is the random error term.
Point 5: Line 171, equation (5) does not appear perceptible in the pdf and it was impossible to analyse it.
Response 5:
Equation(5):
Point 6: In lines 183-184, the authors refer that “... to regress the model...”, but this statement is unclear, what is the principal objective? It is to fit a regression model? In order to estimate which variable? It should be further develop this statement.
Response 6: This paper uses the panel data of each region from 2005 to 2016 to regress model (5) and estimate βEi , the contribution proportion of energy to each region’s output, to calculate the energy misallocation index in each province.
Point 7: In line 283, the energy misallocation index is used as “the core explanatory variable”, however this variable has been estimated according with section 3.1 using the values in table 2. What impact does this has in the error of the considered model?
Response 7: The core explanatory is the independent variable. It is obtained by regression, so there is error. Hence the error term εit in model (7).
Point 8: In line 297, the says that “ (...)μi and λt are virtual variables (...)”, What do you mean by that?
Response 8: Revised the sentence: μi represents the non-observable individual effect, λt represents the time effect.
Point 9: In line 298, it is refered that W is the spatial weight matrix, what kind of spatial weight matrix was used? It is the one based in the contiguity criterion or based in a distance approach, for example?
Response 9: W is the 0 ~ 1 spatial contiguity matrixes. When i province is adjacent to j province, W is 1, otherwise, W is 0.
Point 10: Line 312, Yi represents the observation value of the i -th province, but of which variable in this case? It is not clear.
Response 10: Yi represents carbon emission efficiency of the i-th province.
Point 11: According with line 298 it is being used a spatial weight matrix. If it is used this kind of matrix, W does not just take values 0 and 1, as is indicated in lines 313-314. In this case should has been used a spatial contiguity matrix instead.
Response 11: The reviewer’s point is taken. Revised to: spatial contiguity matrix.
Point 12: Figure1: In order to identify spatial association between provinces, I suggests to use a Moran's scatter plot instead.
Response 12: Moran's I of carbon emission efficiency reflects the global spatial autocorrelation. It is adopted to verify the existence of spatial effect. To further investigate the local features of the autocorrelation, we need a Moran's scatter plot. I think this paper mainly studies the impact of energy misallocation on carbon emission, so I replaced Figure 1 with Table 6.
Point 13: Line 330: The authors refer: “ (...) according to the criterion of Anselin(1996) (...)”this statement should be further develop, what criterion are being used? It is not explicit that the Lagrange Multiplier Tests are being used and why. Moreover, these tests have to verify several assumptions, has this been validated?
Response 13: Moran’s I only verifies the existence of the spatial effect. Whether to adopt spatial lagged model or spatial error model is determined according to the criterion of Anselin (1996), thus Lagrange Multiplier-lag(LM-lag) Test and Lagrange Multiplier-error(LM-err) Test are needed. If neither of these two passes the significance test, robustness test is required, namely, Robust LM-lag (R-LM-lag) Tests and Robust LM-error(R-LM-err) Tests.
Point 14: Line 335-336, refers to several effects in the model. These effects are refre-sented by which factors/variables in the model?
Response 14: Joining spatial lagged dependent variable term and spatial error autocorrelation term in the Ordinary Least-squares model and we have the two frequently adopted models in Spatial Econometrics: spatial lagged model and spatial error model.
Point 15: Line 344, The authors refer: “After comparison (...)”, the chosen model is selected based in what measure?
Response 15: According to the goodness of fit R2, we can see that the spatial lagged model with spatial and time period fixed effect has the highest value of R2.
Point 16: In general, there are several used acronyms throughout the text, that are not defined or whose definition appears latter in the text, which makes it a bet difficult to read. For example the acronyms LM-lag, R-LM-lag, LM-err and R-LM-err, in table 6 or LM that appears in line 388, etc.. The same for other acronyms that are used in the paper, this could be improved.
Response 16: Revised: Lagrange Multiplier-lag(LM-lag) Test and Lagrange Multiplier-error (LM-err) Test, Robust LM-lag (R-LM-lag) Tests and Robust LM-error(R-LM-err) Tests.
Author Response File: Author Response.pdf
Reviewer 2 Report
Please find the attached pdf file.
Comments for author File: Comments.pdf
Author Response
Response to Reviewer 2 Comments
Point 1: Lines 7-10 are repetitive.
Response 1: Revised.
Point 2: The manuscript uses the word "researches" as an attempt to take a plural form of the singular noun "research." However, this does not follow proper English grammar conventions. At least it is incredibly rare to use this plural version of research — for example lines 48 and 160.
Response 2: Revised.
Point 3: Line 76 is not a well-structured English sentence.
Response 3: Revised.
Point 4: I disagree with the statement presented in Line 81. Not all energy sources pollute the environment. This brings me to my first question: What type of energy is the study considering? Is it non-renewable or renewable energy sources or are the authors just considering energy in general? If it is the latter, there should be some mention of the proportion of energy sources consumption in China.
Response 4: Revised and reworded. This paper focuses on fossil fuels like coal, oil and natural gas, which are non-renewable energy sources. According to the Report on the Big Data of Energy in China 2018, judging by the data from the last decade, coal and oil accounted for 80%~90% of the total primary energy consumption in China while the proportion of clean energy such as natural gas constantly increased, showing great potential.
Point 5: The presentation of equation (2) should be improved. I had to read this paragraph several times to understand it. I suggest adding summation symbol to E so that we understand that Ei is a portion of E. This is later stated in line 149, but it is confusing.
Response 5: Added: Ei represents the energy consumption of region i, E represents the total energy consumption.
Point 6: Line 149. The formula of ?E is not correct. I arrived at this formula instead: ?E =()( E / Ei).
Response 6: Revised to: Weighted contribution proportion of energy to the output can be expressed as.
Point 7: is defined as the ratio of energy use and value. However, in lie 153, the authors define it as a cost measure. However, I could not understand how ?Ei became a cost measure. The authors should stick to one interpretation of .
Response 7: Revised: It indicates that the actual energy allocation is above the theoretical level. is the relative distortion coefficient of energy prices.
Point 8: Line 161, it should read Cobb -Douglas production function. This brings me to another point about this manuscript. The authors extensively use acronyms that are not defined in the text. For instance, L-M should be defined as Lagrange Multiplier in Table 6. See comment 19 below.
Response 8: Added.
Point 9: What does μi , λt , εit represent in equation 4?
Response 9: μi in equation (4) is the individual effect, λt is the time effect and εit is the random error term.
Point 10: After finishing in Line 170 is not a complete sentence.
Response 10: Revised.
Point 11: Equation 5 is unreadable in Line 171. Why is the output variable divided by Labor? Given that Energy is the primary focus of the paper and that one of the mechanisms identified by the authors is innovation (Line 100) why not divide by capital?
Response 11:
Equation(5): . Energy is the primary focus of the paper. This paper followed the practice of Bai et al.(2018) and Pu et al.(2015), and obtained Equation(4) and Equation(5). The output variable divided by Labor is a regular practice.
Point 12: According to the regression results presented in Table 1, βEi from equation 5 is not the same as the βEi in equation 4. This is because of the manipulation of equation 4 to arrive at equation 5.
Response 12: The βEi obtained through equation (5). This deed followed the practice of Bai et al.(2018) and Pu et al.(2015).
Point 13: Given my previous comment, I do not think that βEi represents the energy output elasticity.
Response 13: Reworded. βEi represents the contribution proportion of energy to the output in each provinces.
Point 14: In Line 197, the authors explain the interpretation of the energy misallocation index. I would suggest to explain it in terms of deviations from zero (which is efficient allocation).
Response 14: This paper focuses on the insufficient or over sufficient energy allocation and mainly studies the impact of energy misallocation on carbon emission.
Point 15: Equation 6, line 239. Note that the authors use the same Greek letter, rho, to define the carbon emission coefficient and the spatial autoregressive parameter.
Response 15: Revised.
Point 16: Equation (7), the variable of interest has been estimated by the authors, therefore it should be included in this regression as . In addition, this variable will be correlated with the other error components in equation 7. Any concerns about the empirical estimation of this equation?
Response 16: The core explanatory is the independent variable. It is obtained by regression, so there is error. Hence the error term εit in model (7).
Point 17: Figure 1 is an unconventional way to present a statistic and its corresponding p-value. I think a table or a scatter plot might be more appropriate.
Response 17: Replaced Figure1 with Table 6.
Point 18: Line 320, the fluctuation of the Moran’s I has only been from about 0.28 to 0.38. Other than having a positive spatial autocorrelation, I do not see the need to point out this fluctuation (if any).
Response 18: Replaced with Table 6.
Point 19: Table 7 What does SLM and SEM acronyms mean? Somewhere in the text, it should be noted that you ran two different regression structures: Spatial Lag Model (SLM) and Spatial Error Model (SEM).
Response 19: Revised.
Point 20: Line 347, it is the second time, I read core variables in the text. Do the author(s) mean variable of interest or main variables?
Response 20: The core explanatory is the independent variable. Revised.
Point 21: Line 392, the authors “strongly reject the original hypothesis” which is what?
Response 21: Revised: The assumption of random effect model is rejected and the fixed effect model is selected.
Point 22: Line 385, if the tests presented in Table 6 call for the use of a spatial lag model and a spatial error model, why was the spatial error model dropped from the estimations in Table 7?
Response 22: According to the goodness of fit R2, we can see that the spatial lagged model with spatial and time period fixed effect has the highest value of R2.
Point 23: Line 385, the authors are using a panel spatial regression model. The use of a panel allows the authors to estimate fixed and time effects for each province in China. Why use a regional analysis? What is the value of each element in the weight matrix for results presented in Table 7?
Response 23: There are differences between regional regression results and regression results of the whole country. Revised the spatial weight matrix to spatial contiguity matrix. When i province is adjacent to j province, W is 1, otherwise, W is 0.
Point 24: Mechanisms explained in line 100 and 118 were never identified in the regression model presented. In fact, they are presented in the model in the following way. Innovation(measured as province's proportion of R&D) affects carbon emission efficiency positively. However, we do not have any empirical evidence that this effect is as a result of energy misallocation. The same logic can be applied to the exports (measured as the openness degree variable) argument.
Response 24: Revised the analysis of Mechanisms.
Point 25: It is a standard practice to present the standard errors.
Response 25: Added in Table 8 and Table 9.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
No further comments on the revised manuscript.
