Forecasting Carbon Price in China: A Multimodel Comparison

Round 1

Reviewer 1 Report

Thank you for the article and your efforts.

1-The introduction section also includes a literature review. The section title should reflect that. 

2- Regarding the relationship between carbo price and energy price I have few questions:

-First, why do choose one the price of one type of crude oil and two types of coal?  We know that the crude oil plays more important role in China's energy mix.

-Second, why did you choose WTI crude oil price rather than other types of oil which China imports more? For example Arabian Light, Dubai, or Iran Light? Why not Brent?

Third, Why have you chosen Henry Hub gas price for comparison between carbon price and gas price? We know today Exxonmobil gas price in Henry Hub is about US$6 pet million BTU and in East Asia is about $33 MBTU.

3- Article says in line 210: "As a result, this paper takes carbon prices as explanatory variables and takes carbon prices and energy prices as explanatory variables to conduct empirical research". What do you mean?

4- You have written in line 415: "However, these studies only use the time series data of carbon prices and do not consider the related variables such as oil price, coal price, and natural gas price". Could you describe it for me in details?

5- Figures are not so bright, I could not absorb them very well.

6- The authors should avoid using multiple references in the manuscript. To this extent, no more than 3 references in a sentence are appropriate.

7- The Policy Recommendations part is very little. The policy recommendations should be extended and refined.

8- In line 147 why the word Parameter is capital? 

Author Response

First of all, we would like to thank you for reading our article and for your valuable comments. Below is our response to all comments made.

Point 1: The introduction section also includes a literature review. The section title should reflect that. 

Response 1: According to your suggestion, we have divided the introduction into two parts: introduction and literature review. Thank you for your suggestion to make our article better.

Point 2: Regarding the relationship between carbo price and energy price I have few questions:

First, why do choose one the price of one type of crude oil and two types of coal?  We know that the crude oil plays more important role in China's energy mix.

Second, why did you choose WTI crude oil price rather than other types of oil which China imports more? For example Arabian Light, Dubai, or Iran Light? Why not Brent?

Third, why have you chosen Henry Hub gas price for comparison between carbon price and gas price? We know today Exxonmobil gas price in Henry Hub is about US$6 pet million BTU and in East Asia is about $33 MBTU.

Response 2: Thank you for your advice, which gives me great inspiration.

Firstly, why do I choose two coal prices and one crude oil price? To be honest, at first, we wanted to select domestic and foreign energy price data, but the domestic natural gas price was not so sure, so we gave up this idea. At that time, these data were just collected, and some literature studied the relationship between these data and carbon prices, without considering whether two coal prices and one crude oil price would affect the prediction results. Therefore, according to your suggestion, I changed the energy price data into a crude oil price, a natural gas price, and a coal price.

Secondly, why do we choose WTI crude oil price, on the one hand, WTI is one of the international crude oil pricing benchmarks, and on the other hand, there are literature studies on the relationship between WTI and carbon price (Wang and Guo, 2018; Zhuang et al., 2014), so other crude oil prices are not considered. But after careful consideration of your proposal, we changed the crude oil price to Brent crude oil price during the revision process. Brent crude oil price is chosen because it and WTI are both international crude oil pricing benchmarks. In addition, according to the data released by London Intercontinental Exchange (ICE), this Brent price is used as the pricing benchmark by more than two-thirds of the world's international oil trade. And Zhao et al. (2021) chose Brent crude oil as the input variable when predicting the carbon price. Therefore, we think Brent is more representative.

Thirdly, the reason why Henry Hub gas price is chosen, like other data, is that there are literature about the relationship between Henry Hub gas price and carbon price (Duan et al.,2021). In the process of revision, we thought that a more representative natural gas price should be selected, so we changed the natural gas price to NYMEX natural gas price. Since the listing of the NYMEX natural gas futures contract in 1990, the trading volume and positions have been increasing. NYMEX natural gas futures contract price is also widely used as the benchmark price of natural gas.

Point 3: Article says in line 210: "As a result, this paper takes carbon prices as explanatory variables and takes carbon prices and energy prices as explanatory variables to conduct empirical research". What do you mean?

Response 3: As for article says in line 210: "As a result, this paper takes carbon prices as explanatory variables and takes carbon prices and energy prices as explanatory variables to conduct empirical research", which is to express that we take carbon price and energy price as input features to predict the future carbon prices. It has been revised in the article. Thank you for your valuable suggestions.

Point 4: You have written in line 415: "However, these studies only use the time series data of carbon prices and do not consider the related variables such as oil price, coal price, and natural gas price". Could you describe it for me in details?

Response 4: As the article says in line 415, we want to express that the selected existing studies all adopt deep learning to predict the carbon price. However, in the selection of input features, they only choose the carbon price and consider other data that affect the carbon price. Comparing the research results, shows that it is effective to improve the accuracy of carbon price prediction by taking both energy price and carbon price as input features. We have revised it in the article. Thank you for your valuable suggestions.

Point 5: Figures are not so bright, I could not absorb them very well.

Response 5: We have modified the graphics in the text. Thank you very much for your suggestion to make our article better.

Point 6: The authors should avoid using multiple references in the manuscript. To this extent, no more than 3 references in a sentence are appropriate.

Response 6: Thank you for your suggestion. We have revised the sentences with more than three quotes in the text according to your request.

Point 7: The Policy Recommendations part is very little. The policy recommendations should be extended and refined.

Response 7: According to your suggestion, we have rearranged the policy suggestions, and made suggestions to the carbon market regulators, enterprises, and the government respectively. Thank you for your advice, which enriches our articles.

Point 8: In line 147 why the word Parameter is capital? 

Response 8: In the review of related literature, we found only Parameter sharing (Shi et al. 2018; Jain et al. 2016) and rarely use capital sharing mechanism. So here we use Parameter. Thank you for your advice.

 

References：

Wang Y, Guo Z. The dynamic spillover between carbon and energy markets: New evidence[J]. Energy, 2018, 149: 24-33.

Zhuang X, Wei Y, Zhang B. Multifractal detrended cross-correlation analysis of carbon and crude oil markets[J]. Physica A: Statistical Mechanics and its Applications, 2014, 399: 113-125.

Zhao L T, Miao J, Qu S, et al. A multi-factor integrated model for carbon price forecasting: market interaction promoting carbon emission reduction[J]. Science of The Total Environment, 2021, 796: 149110.

Duan K, Ren X, Shi Y, et al. The marginal impacts of energy prices on carbon price variations: evidence from a quantile-on-quantile approach[J]. Energy Economics, 2021, 95: 105131.

Shi Y, Fernando B, Hartley R. Action anticipation with rbf kernelized feature mapping rnn[C]. Proceedings of the European Conference on Computer Vision (ECCV). 2018: 301-317.

Jain A, Zamir A R, Savarese S, et al. Structural-rnn: Deep learning on spatio-temporal graphs[C]. Proceedings of the ieee conference on computer vision and pattern recognition. 2016: 5308-5317.

Reviewer 2 Report

After getting familiar with the reading-matter of the reviewed article, I state the following:

1. The main focus of this article is coal price forecasting in China using Multivariate Long Short-Term Memory (LSTM), Multilayer Perceptron (MLP), Support Vector Regression (SVR), and Recurrent Neural Network (RNN).
2. This issue does not fall within the mainstream of the International Journal of Environmental Research and Public Health.
3. Incorrect entries of numerical values in the text and tables - numerical values starting with four digits should be written with the comma every three digits.
4. in the case of the followig Figures: 1, 2, 3 and 4, their titles should be followed by explanations of individual symbols (Figs. 1 and 2) and what the figures: a, b, c and d mean (Figs. 3 and 4).
5. Notes organizing the work were introduced in the track changes mode.

An article on coal price forecasting, in the opinion of the reviewer, should NOT be published in the International Journal of Environmetal Research and Public Health, but in other journals dealing with such issues. The reviewer suggests publishing this article in the Journal Policy Modeling or Energies.

Author Response

First of all, we would like to thank you for reading our article and for your valuable comments. Below is our response to all comments.

Point 1: This issue does not fall within the mainstream of the International Journal of Environmental Research and Public Health.

Response 1: Thank you for your advice. We found some articles about carbon prices in The International Journal of Environmental Research and Public Health (Luo et al., 2022; Wang et al., 2019; Yun et al. 2022). In this article, Yun et al. (2022) also studies a carbon price forecast, but its research object is European Union Allowance Future (EUAF). In addition, with the global attention to carbon dioxide, the carbon emissions trading market is becoming more and more important. As the core element of the carbon market, the accurate prediction of carbon price plays an important role in understanding the dynamics of the carbon trading market and achieving the national emission reduction target.

Point 2: Incorrect entries of numerical values in the text and tables - numerical values starting with four digits should be written with a comma every three digits.

Response 2: We have modified the table in the text according to your suggestion. Thank you for your suggestion.

Point 3: in the case of the followig Figures: 1, 2, 3 and 4, their titles should be followed by explanations of individual symbols (Figs. 1 and 2) and what the figures: a, b, c and d mean (Figs. 3 and 4).

Response 3: Thank you for your advice, which makes our article better. According to your request, we re-made the figures and explained the symbols in Figs. 1 and 2 and the sub-diagram (a, b, c and d) in Figs. 3 and 4 in their titles.

Point 4: Notes organizing the work were introduced in the track changes mode.

Response 4: According to your request, all our revisions are made in revision mode. Thank you for your advice.

References:

Luo R, Li Y, Wang Z, et al. Co-Movement between Carbon Prices and Energy Prices in Time and Frequency Domains: A Wavelet-Based Analysis for Beijing Carbon Emission Trading System[J]. International Journal of Environmental Research and Public Health, 2022, 19(9): 5217.

Wang L, Yin K, Cao Y, et al. A new grey relational analysis model based on the characteristic of inscribed core (IC-GRA) and its application on seven-pilot carbon trading markets of China[J]. International journal of environmental research and public health, 2019, 16(1): 99.

Yun P, Zhang C, Wu Y, et al. Forecasting Carbon Dioxide Price Using a Time-Varying High-Order Moment Hybrid Model of NAGARCHSK and Gated Recurrent Unit Network[J]. International Journal of Environmental Research and Public Health, 2022, 19(2): 899.

 

Reviewer 3 Report

The authors aim to forecast carbon price in Hubei and Guangdong (China)  taken  carbon price and the price of four sources of energy as explanatory variables. They use different forecasting technique including Neural Network.  They find that one model is more suitable for carbon price forecast and conclude that carbon price can be priced through energy price.

Before it becomes of publishable standards however, I would strongly encourage the authors to consider the following points:

-The relationship between the price of the different energy sources  and  the carbon price need to be explained with more references  in section 3.2. For instance, explain how the rise in international oil price increases exchange rate.

-Please reduce the use of acronyms in the text if possible, for instance, use oil or gas and not ECO or WTI.

- I believe that the authors should better mention that forecasting techniques have been developed in other fields and not only in the field of  carbon price. For instance,  page 2, line 70, instead of “In addition, in the existing literature, a large number of forecasting techniques have been applied to carbon price forecasting…”  write “Researchers have developed a large number of forecasting in different fields, among them manufacturing (Lewis, 1982); finance (Trippin and Turban, 1996)  tourism demand Álvarez-Díaz et al., 2019); energy (Wang et al, 2021) or carbon price forecasting….”.

- Please be a little bit more explicit in the conclusions when you mention the policy implications . What options do the governments have to  priced the energy sources? How would government intervention affect the possibility of enterprises  to predict carbon price?

 

References:

Álvarez-Díaz M, González, M and Otero Giralde MS (2019) Forecasting International Tourism Demand Using a Non-Linear Autoregressive Neural Network and Genetic  programming. Forecasting , 1 90-106. doi:10.3390/forecast1010007

Lewis, C.D. Industrial and Business Forecasting Methods; Butterworths: London, UK, 1982; ISBN-13: 978-0408005593.

Trippi, R.R.; Turban, E. (Eds.) Neural Networks in Finance and Investment; McGraw-Hill: New York, NY, USA, 1996; ISBN 1557384525.

Wang, J., Cao, J., Yuan, S., Cheng, M. (2021) Short-term forecasting of natural gas prices by using a novel hybrid method based on a 442 combination of the CEEMDAN-SE-and the PSO-ALS-optimized GRU network. Energy, 233, 121082.

Author Response

 

First of all, we would like to thank you for reading our article and for your valuable comments. Below is our response to all comments made.

Point 1:The relationship between the price of the different energy sources  and  the carbon price need to be explained with more references  in section 3.2. For instance, explain how the rise in international oil price increases exchange rate.

Response 1: Thank you for your advice. For the explanation of energy price data in 4.2 (3.2 of the original manuscript), we consider that different crude oil (or coal, natural gas) prices may have different impacts on carbon prices. In this part, we revise it as follows: first, we introduce how energy prices have an impact on carbon prices, and then we introduce the reasons for choosing one representative of energy prices. Therefore, there is no separate writing about how the price of crude oil (or coal and natural gas) affects the price of carbon.

Point 2: Please reduce the use of acronyms in the text if possible, for instance, use oil or gas and not ECO or WTI.

Response 2: Thank you for your advice, which makes our articles easier to read. After the revision, we only use abbreviations to distinguish the carbon prices in Hubei(HBEA) and Guangdong(GDEA), and the rest of the energy prices are not expressed by abbreviations.

Point 3: I believe that the authors should better mention that forecasting techniques have been developed in other fields and not only in the field of  carbon price. For instance,  page 2, line 70, instead of “In addition, in the existing literature, a large number of forecasting techniques have been applied to carbon price forecasting…”  write “Researchers have developed a large number of forecasting in different fields, among them manufacturing (Lewis, 1982); finance (Trippin and Turban, 1996)  tourism demand Álvarez-Díaz et al., 2019); energy (Wang et al, 2021) or carbon price forecasting….”.

Response 3: We have combed the literature review again. According to your suggestion, we showed the development of forecasting technology in various fields, including stock price(Nelson et al., 2017), wind speed (Torres et al., 2005), power load (Hafeez et al., 2020), wind power (Wang et al., 2017), carbon price and so on. Thank you for your suggestions to make our articles richer and more meaningful.

Point 4: Please be a little bit more explicit in the conclusions when you mention the policy implications . What options do the governments have to  priced the energy sources? How would government intervention affect the possibility of enterprises  to predict carbon price?

Response 4: Thank you for your advice. According to your suggestion, we have rearranged the policy suggestions and made suggestions to the carbon market regulators, enterprises, and the government respectively. Specifically,(1) Regulators of the carbon market should pay attention to the energy price closely related to the carbon price. In the development of the carbon trading market in China, regulators can use changes in energy prices to predict carbon price changes, assess the risks brought by carbon price fluctuations in advance, and put forward corresponding preventive measures. In this way, adopting appropriate risk control measures to avoid risks can improve the stability of the carbon trading market. (2) Enterprises should take full advantage of the relevant information of the energy market and control the carbon emission cost of enterprises by adjusting the energy consumption structure or improving the energy utilization rate. When the energy price changes, enterprises can predict the carbon price. If the carbon price is on the rise, enterprises should choose more clean energy or develop new technologies to improve the energy utilization rate, thus reducing the demand for carbon emissions and reducing the carbon emission cost of enterprises. (3) The government should ensure the stability of energy prices. Our research results show that the energy price can predict the carbon price well, so the stability of the energy price helps the government to set the carbon price reasonably. Specifically, the government should pay attention to improving the technical level of energy exploration and exploitation in China, increasing the domestic energy supply, and forming a more independent energy market. This can better control the source of carbon price fluctuations and fundamentally guarantee the stable and healthy development of the carbon trading market.

 

References：

Nelson D M Q, Pereira A C M, de Oliveira R A. Stock market's price movement prediction with LSTM neural networks[C]. 2017 International joint conference on neural networks (IJCNN). IEEE, 2017: 1419-1426.

Torres J L, Garcia A, De Blas M, De Francisco A. Forecast of hourly average wind speed with ARMA models in Navarre (Spain)[J]. Solar energy, 2005, 79(1): 65-77.

Hafeez G, Alimgeer K S, Khan I. Electric load forecasting based on deep learning and optimized by heuristic algorithm in smart grid[J]. Applied Energy, 2020, 269: 114915.

Wang H, Li G, Wang G, Zhou B, Peng J. Deep learning based ensemble approach for probabilistic wind power forecasting[J]. Applied energy, 2017, 188: 56-70.

 

Round 2

Reviewer 3 Report

The authors have managed to successfully address my comments, hence I now suggest that the paper should be accepted for publication.