Abstract
The spillover effect of the energy markets and the CET plays an important role in guiding the realization of two-carbon target; using the network spillover methodology of Diebold, Yilmaz, Jozef Barunı´k, and Toma´sKrehlı´k, we examine both the static and dynamic connectedness of CO2 emissions trading (CET) market, steam coal market, new energy, and traditional energy market in China from early Dec 2013 to the end of July 2021. At last, we verified the stability of the model and obtained the following findings: (1) the total spillover effect index is 13.91% between those markets, and it is mainly focused on short term. Moreover, the dynamic spillover effect is time-varying, and it is greatly influenced by the domestic and international environment; (2) the connectedness of the CET market with other energy markets is neutral, the development of new energy market is strong, it is the main transmitter to other markets, especially to the traditional energy market except for the steam coal market, and the coal market is an effect transmitter. These results provide a theoretical reference for investors and policy makers who are concerned with the return connectedness among the CET market, new energy market, and steam coal market in China.
1. Introduction
At present, global climate change has become one of the greatest challenges and threats to human development. In the process of economic development, the carbon dioxide emission caused by energy consumption, especially fossil energy consumption, is the main cause of global climate change [1, 2]. Thus, over the past decade, global initiatives are being taken to reduce the use of traditional energy to clean energy for emission reduction [3]. Nevertheless, China is facing great pressure in the energy structure transformation. Statistically, steam coal accounts for 56% of China’s total energy consumption, oil accounts for 18.7%, and clean energy sources such as natural gas, hydropower, wind power, nuclear power, and solar power generation account for 25.3% in 2021. Although Chinese government regards new energy development as an important strategic development direction, coal will still be the main source of energy in China in the future [4]. Furthermore, the international community is increasingly attaching great importance to the sustainable development of energy [5], climate, and the environment which urge carbon emissions to become the most urgent environmental problem in China. So, in the next five years, China is further deepening energy price reform, especially the price of coal and other fossil energy, which has become an important measure to control the total energy consumption and improve energy efficiency [4].
On the other hand, in order to reduce carbon emissions, countries internalize the externalities of carbon emissions by establishing a carbon emission trading (CET) system. Since the Kyoto Protocol and the Paris Agreement were signed, Europe has set up the EU Emissions Trading System (EU ETS), which has effectively reduced the intensity of carbon emissions [6]. Since 2013, China has established eight pilot carbon emission trading markets in Beijing, Shanghai, Wuhan, Guangzhou, and so on. By October 2022, the cumulative trading volume of the 8 pilot projects was close to 196 million tons and the cumulative turnover of the 8 provinces and cities exceeded 8.58 billion yuan. However, the grandfather rules focused on the historical emissions lead to overallocation and low carbon prices. The carbon emission price is much lower than 100 CNY/ton. Therefore, it is of great practical significance to study the correlation mechanism and spillover dynamics between the CET market and a specific energy market to establish a perfect and sustainable CET market [4].
Over the past few decades, because of its cleanliness, new energy has received the favor of various countries. China has taken the development of new energy industry as a national strategy to be vigorously implemented. The new energy industry is of great significance for China to realize the transformation of the energy utilization mode and the development of green economy for the dual-carbon goal [7].
The vigorous development of the new energy industry will also attract the attention of investors in the capital market. Investors are generally optimistic about the development direction of new energy companies which reflected in the investment in the stocks of new energy companies. Therefore, it has great significance for financial investors and policymakers to acquaintance the new energy company stock price influence factors.
Above all, because the carbon emission rights have both commodity and financial attributes in the financial markets, the CET market has both resource allocation and financial functions. Due to the link of economic fundamentals, there are some connections between “carbon-finance-energy” markets by means of information transfer [8]. The financialization of the energy markets can reflect the fundamental links between energy markets through financialization means so as to better deepen the energy price reform and realize the transformation of green energy utilization. This may be why the Chinese central bank has repeatedly proposed to financialize the energy market and develop carbon finance to realize carbon control and emission reduction in a market-oriented way. With increased globalization and carbon financialization, the correlation between carbon emission market and energy market is also strengthening. This paper studies the connectivity spillover relationship among the CET market, coal market, new energy market, and the traditional energy market in China; the research results can show the total spillover effect between the CET market and the given energy markets which can further show the important role of the CET market in the task of emission reduction. On the other hand, the spillover effect of each financial market can become an important investment channel for the diversification of profits and risks, and it can provide certain reference information for investors to make investment decisions and hedge financial risks.
2. Literature Reviews
The connectivity and the spillover effect between the CET market and the energy market have attracted the attention of many scholars. Numerous scholars have confirmed the relationship between the CET market and the energy market [4, 7]. The existing literature about the connectivity between CET and energy markets has conducted studies on different aspects [9, 10]. Different energy markets have been selected for the study, such as fossil fuel [11] and crude oil [12, 13]; scholar Yang Lu also studied the spillover effects between the CET market and the cryptocurrency market [14]. A variety of research methods have been involved, such as wavelet coherency [7], multiscale entropy [15], structural equation modelling [16], quantile-on-quantile approach [17], multiscale analysis [6], and DCC-MVGARCH model [10]; it is worth mentioning that electricity as an important part of energy is of great significance to the energy transformation and utilization and carbon emission reduction. In the existing literature [18], we studied the role of peak-valley electricity price and trait factors in the information spillover mechanism between the European electricity market and the carbon market, and the result proved the dominant role of the electricity market [19]. We studied the value of renewable energy generation for emission reduction and power supply [8]. We also studied the interaction among Guangdong power, fossil fuel, and carbon market price and confirmed the long-term cointegration relationship among them.
On the other hand, the methods of DY index and BK index have been widely used to measure the connectivity among specific objects [20]. After studying the related literature, these two methods have been applied in various fields up to now, such as these methods can not only be used to measure the total connectivity among all objects [21], but it can also survey the pairwise connectivity [22] between each two objects of the system which may contain multiple objects and the net spillover effect of each object [23]; they can not only be used to measure the objects’ connectivity based on the time dimension [24, 25] but also on different frequencies [26]; just because of the unique capabilities of the DY and BK methods, this method system has been widely used by scholars [27, 28].
In summary, there were massive studies about the connectivity between CET and energy markets, while there are few about the research among CET, steam coal, and new energy markets [4]. In the few existing literature studies on CET, coal market, and new energy markets, we have the following research gaps. Firstly, most of the research is about the bidirectional causality between the CET market and the single energy market, but there is lack of simultaneous studies on the interaction between multiple markets. Secondly, the research studies on the correlation among the CET market, coal market, new energy market, and traditional energy market lack directional spillover and net spillover of impact identification and mutual influence of the complex networks of all markets. Thirdly, the relationship among the CET market, coal market, and the new energy market should be measured from both static and dynamic aspects, and whether the relationship between the CET market and the given energy markets has time-varying nature is worth exploring.
In order to fill the gap in the existing literature, this paper studies the connectivity and spillover effect among CET, steam coal, new energy, and traditional energy markets in China based on the method of Diebold and Yilmaz [29]. The reason we elect this method is that it is independent of element sorting [12]. The aim of research is to explore the connectivity, especially the directional spillover effects and net spillover effects between the various markets of the research framework. Firstly, the paper analyses the characteristics and the regular development of each market from the perspective of time sequence, and then we constructed the DY and BK indices to reveal the connectedness among the CET market, coal market, new energy market, and traditional energy market in the time and frequency domain. Finally, we study the directional spillover index and the net spillover index from a time-varying perspective.
Thus, the contributions to the existing literature are from the following aspects: first, the study from the perspective of connectivity to research the total spillover effect, net spillover effect, and the pairwise spillover effect among the CET market, coal market, new energy market, and traditional energy market, this holistic research approach simplifies the process of understanding the role of direct and indirect effects between multiple markets. Second, the study surveys the spillover effect between CET and energy markets from the perspective of static and dynamic spillover effects, as well as from the perspective of time dimension and the frequency dimension, all of this further broaden the scope of research on the given markets. Thirdly, we identified the net information recipients and net information contributors in the CET market, coal market, new energy market, and traditional energy market in the current Chinese context.
The rest of this article is organized as follows: Section 3 illustrates various descriptions and data collection as well as constructs the methodology of the DY and BK indexes. The preliminary analysis and the empirical results of the series of CET prices, new energy market prices, traditional energy prices, and coal prices are demonstrated in Section 4. In Section 5, we analyzed the empirical findings of the static and dynamic spillover effects. Finally, the conclusion, the policy implication, and the further work are summarized in Section 6.
3. Data and Methodology
3.1. Data
This paper investigates the connectedness of the CET market, new energy market, steam coal market, and traditional energy market in China.
Considering that Beijing Carbon Exchange has been running smoothly and efficiently compared to other CET markets since its inception, this study selects the CET price in the Beijing Carbon Exchange as the proxy of the CET market price. Moreover, the Chinese stock market has become quite efficient through a series of institutional and regulatory reforms after China’s accession to the WTO; thus, the market data (e.g. prices) of the listed companies can effectively reflect the relevant information of the company and the market expectation for future performance. The development of the new energy industry can be effectively measured by its corresponding stock price movement; this paper selects the CSI New Energy Index (CSINE), which is composed of 80 companies listed on the Shanghai Stock Exchange Composite and Shenzhen Stock Exchange Composite. The paper selects the CSI All Share Energy Index and the steam coal price as the proxies of the traditional energy market. Furthermore, according to the national data, traditional energy consumption accounts for 75% of total energy consumption in China, so this paper uses the CSI All Share Energy Index (ASEI) as the proxy of the traditional energy market. In particular, the paper investigates the connectedness and spillover between the steam coal market and three other markets. Since the futures price is a better representation than the spot price, this paper selects the stem coal futures price named COAL instead of the steam coal spot price as the representation of the steam coal market. Finally, the stability of the connection model is tested using CSI energy, namely, CSIEN.
The CET data were obtained from the China Beijing Green Exchange (https://www.cbeex.com), and the steam coal future contract price was obtained from Zhengzhou Commodity Exchange (ZCE). The CSI All Share Energy Index (the ticker symbol is 000986) and CSI New Energy Index (the ticker symbol is 399808) were obtained from the database provided by Wind Information Co., Ltd. (WIND). Because CET market trading is not continuous every workday, instead of utilizing daily data, we use weekly average data. The ASEI and CSINE indices use the weekly closing prices, and the CET price and the coal futures price use the weekly average of trades. The data sampling period ranges from early December 2013 to the end of June 2021, and a total of 353 observations are available. The beginning point and data size depend on the availability of data on the Beijing CET market.
3.2. Methodology
In this research, we explore the overall connectivity, the directional spillover index, and the net spillover index among CET, steam coal, new energy, and the traditional energy. Firstly, we established VAR (vector auto regression model) with indices of the markets we considered, and then we apply the measurement approaches for the connectedness among different markets set up by Diebold and Yilmaz [29], namely, the DY index, and Barunik and Křehlík [30], namely, the BK index. Those indexes were calculated on the basis of the generalized variance decomposition (GVD) of the covariance-stationary VAR () model, which is expressed by the following equation:
In equation (1), parameter is an vector that represents the endogenous variables, is the autoregressive coefficient matrices with dimension , is the lags of the model, and is the model’s random error that is independent and identically distributed.
Moreover, the basic idea of the DY approach is using the generalized variance decomposition technique. The following DY approach can obtain the contribution of the change of each variable to the other variables. Here, we describe this contribution as the spillover index, and the spillover index from market to market is denoted by . This is the proportion of the H-step prediction error variance of variable explained by variable . Therefore, the value of is from 0 to 1. Furthermore, as H increases, gradually tends to decrease until it stabilizes. The significance of generalized variance decomposition is that the analysis result will not be influenced by the sequence of variables in the VAR model, so we can obtain robust analysis results. The formula of is defined as the following equation:where is the variance matrix of the errors in the VAR (p) model, is the standard deviation of the error shown in the jth diagonal element of , and is an selection vector with one as its element and zero otherwise.
Because the sum of the composition of the own and cross-variable variance is not unity, we normalized each variance composition using the row sum, and the normalized formula is given as follows in the equation:
In equation (3), and are involuntary. shows the pairwise directional connectedness from to visually at horizon H.
In order to better analyze the connectedness relationship between variables, Debold and Yilmaz constructed a series of network spillover indices on the basis of a generalized variance decomposition matrix, and the details are as follows:
We name the total spillover index of the system, and it represents the total connectedness between each market we consider. In this study, it can measure the spillover effect of the CET market, new energy market, traditional market, and coal market.
We denoted the pairwise directional spillover index from market to market in the system as , so , and is generally not equal to . Thus, we further define the net-pairwise directional index from market to market as the deviation value between and denoted as , and .
Therefore, it is natural that the total directional connectedness from all other markets to market is denoted as , and the calculation formula is given as follows in the equation:
The total directional connectedness to all other markets from market is denoted by , and the calculation formula is given as follows in the equation:
Here, we focus on the net spillover effect of market , which is denoted as . This index measures the net spillover from market to all other markets.
4. Preliminary Statistical Analysis
The change details of the coal futures price, the CET price, the CSI new energy index, and the ASEI during the period of early December 2013 to the end of July 2021 are presented in Figure 1. The figure shows that there are different trend details in the four markets starting in early December 2013.
(a)
(b)
(c)
(d)
It can be clearly noticed that the coal price continued to fall from the end of 2013 to the end of 2015, and then the coal price rapidly returned to normal levels in half a year. Furthermore, the coal price rapidly increased after June 2020, which can be attributed to increased demand. According to our investigation, the demand for electricity is rising across the country as the economic recovery accelerates and heat persists in the postpandemic era period, and 70% of China’s power plants are coal-fired, which has pushed coal prices soaring in turn. The CET price fluctuated after June 2018 and fluctuated more in September 2019. The results can be attributed to President Xi setting the goal of peak carbon use and carbon neutrality in the Seventy-fifth Session of the United Nations General Assembly. For the last three years, the CET price has fluctuated considerably. In the energy market, China has paid great attention to develop the new energy industry in recent years, which has led to a flood of money into the new energy sector. Moreover, it can be seen that the CSINE index has been rapidly increasing since early 2020, and the traditional energy market has been declining with shocks since the end of 2015, which may be some of the results of the transition from traditional energy to new energy.
Because the return price has better statistical characteristics, the study treats the original data with formula to serve as the return index of the variable before the preliminary statistical analysis.
represents the weekly data of the CET price, coal futures price, new energy index, and traditional energy firms stock price. Therefore, in the following descriptive statistical analysis and empirical analysis, the paper will adopt the return series of the four markets for analysis.
Figure 2 shows the dynamic evolution of the CET market returns, new energy market returns, coal market returns, and traditional energy market returns. We clearly obtain the fluctuation of each return series. The coal price and the CET price have more similar volatilities. However, in the early days, the CET market fluctuated earlier than the coal market. For example, the CET market is middle in 2017, while the coal market is in the middle early in 2018. However, in the subsequent volatility, the two series’ movements are almost synchronous. This suggests that markets are sufficiently flexible and efficient to reflect market information. We can also find that the volatility of the new energy market is similar to the volatility of the traditional energy market. As shown in Figure 2, the CSINE market and the traditional energy market fluctuate more in the period of July 2015 to April 2016, and the traditional energy market tends to be stable. Furthermore, new energy has higher volatility than the traditional energy market because increasingly more fields have paid attention to the new energy field in recent years. Moreover, new energy is an inevitable choice to realize green economy development.
(a)
(b)
(c)
(d)
Table 1 shows the descriptive statistics of the variables’ weekly returns. It is evident that the mean returns of the four markets are all near zero. Furthermore, the standard deviation of the CET return price is the largest, while the standard deviation of the coal market is the smallest, which shows that the CET market has the largest volatility and the coal market has the lowest volatility. The skewness shows that the skewness of the coal futures return price is similar to a normal distribution, while the other three markets’ return prices are negatively skewed. Furthermore, the kurtosis coefficients of the four markets are greater than zero, which means that they are all leptokurtic. Moreover, the J-B test is a normality test based on the skewness and kurtosis, and the results show that the test results are all significant at 1% significance level, which indicates that the four return series do not all obey the normal distribution.
The stationarity of the four return series can be checked by the augmented Dickey–Fuller (ADF) test. It is clear that the T statistics of the above four variables are all less than the corresponding critical values from Table 1. Therefore, the null hypothesis is rejected at the 1% level, indicating that there is no unit root in the return series of the CET price, coal futures price, new energy index, and traditional energy firm stock, which are stationary series. The KPSS test also obtained the same conclusion, and this further confirms the suitability of using the VAR model for analysis.
Figure 3 depicts a visual Pearson’s correlation matrix for the four markets’ various return series. We note that the color which changes from blue to red indicates the strength of the correlation which changes from negative to positive. First, it is found that there is a significantly strong and positive correlation (0.63) between the traditional energy market and the new energy market. As expected, the correlation ship between the coal market and the traditional energy market is positive (0.22). Because coal is a major part of the traditional energy market, there is an inherent connection between them. Judging from the current data, there is a weak negative correlation between the new energy market and the traditional energy market, and there is a weak correlation between the CET market and the new energy market.
5. Empirical Results and Discussion
Our initial result of the significant correlation ship among the CET market, traditional energy market, coal market, and new energy market offers some preliminary indication of the network connectedness and spillover effects among the markets we consider. In this section, we will utilize the decomposition of the prediction error variance based on the VAR model to construct the DY and BK indices. The network connectivity and spillover effects between each energy market and the CET market are analyzed from both the static and dynamic directions, filling the gap in the existing literature in the related fields. This method can not only measure the direct effect between the variables but can also measure the directional parameters, which enables the more detailed description of the interaction relationship between the market pairs in the system.
5.1. The Full-Sample Volatility of Spillover Analysis of Return Series
We use the return series connectedness network to Chinese environmental and energy to study their spillover connectedness in a static environment. Following Jiang et al. [7] and Lin and Chen [4] who examined the systemic spillover of China’s CET market, coal market, and new energy market using the multivariate wavelet method, VAR(1)-BEKK-AGARCH(1, 1) and VAR(1)-DCC-GARCH(1, 1) models were used. Firstly, the VAR model was constructed based on the weekly data of China’s coal futures price, CO2 emissions trading price, traditional energy market, and new energy index for which the model lag order was selected after comparing the model results with the model’s Bayes–Schwarz Information Criterion (BIC). Then, we use the methods proposed by Diebold and Yılmaz in the time domain and the method of BK in the frequency domain to reflect the mutual influence and spillover effects within the four markets and construct the return spillover network based on the estimation of 100-step-ahead error variance prediction, and the results are shown in Table 2.
Table 2 is the net connectedness table for each market during the entire period from early December 2013 to the end of July 2021. The predictive horizon H is 100 weeks, which is sufficiently high so that it will not change with the additional period, and the VAR lag order is 2 weeks.
In Panel A, the th element in the 4 4 (from to ) matrix shows the estimated contribution to the forecast error variance of variable coming from market , which represents the 100-week-ahead forecast error variance of market due to the shock from market . “FROM ” and “TO ” represent the from-connectedness of market and the to-connectedness of market , respectively, e.g., in the line direction, steam coal return series’ forecast error variance was explained 98.65% by itself, while there was 1.29% explained by ASEI which presented traditional energy, and there was 1.29% forecast error variance which was explained by other markets. There was 55.42% forecast error variance of the shock of ASEI market which was explained by itself, and the shock of CSINE market explained 37.74% forecast error variance of the ASEI market. There is a total of 44.57% of the forecast error variance which was caused by other market shocks. This shows that the fluctuation of the energy market has a great impact on the traditional energy market. This can be interpreted as follows: new energy is an alternative product of traditional energy sources. With the intensification of the global warming, the use of new energy materials and products is becoming more and more popular, so the use of traditional energy-related products decreases accordingly. This impact effect will also be reflected in the corresponding stock market data. From the column direction, we can see the shock of every market contribute rate to the variance error decomposition in other markets, such as there is 5.42% forecast error variance of the ASEI explained by the steam coal market. There are similar interpretations to other data in the Panel A. As can be seen from Panel A, the degree of mutual influence between every two markets is inconsistent. In general, the more stable the market, the less it is affected by other markets and the smaller the value of “From” in Panel A [26, 31].
Table 2 shows that the total spillover index is 13.91%, which means that 13.91% of the variation in the system is due to the interaction between variables. It is obvious that for the CET market, the spillover effect from the coal market is much greater than that from the CSINE and ASEI markets. As for the traditional energy market, the spillover effect from CSINE, which reaches 37.74%, is much greater than those for other market indices. Overall, the CSINE, which represents the new energy market, has the largest spillover effect (9.54%), and this is mainly because the new energy market has a high spillover effect on the traditional energy market. Panel B shows the pairwise directional connectedness among the four markets, including the net-pairwise connectedness and the conclusion. We find that the traditional energy market is a recipient market, and the largest transmitter is the new energy market in the system. As expected, this is because these two markets have considerable substitution effects on each other. Furthermore, the CET market in China is neutral, and the spillover effect between the CET market and other markets is nonsignificant.
5.2. Analysis of the Static Return Spillover Effect Based on the BK Index
The abovementioned analysis in a static environment used the method of the DY index, and this method can examine the connectedness at a specific time. In order to study the time and frequency dynamics between the CET market and the energy markets in China, we next focus on the method proposed by Baruník and Křehlík [30] to measure the spillover effects of the return series of the CET market and the energy markets in China.
Table 3 shows the empirical findings of the return series spillover between the CET market and the energy market based on the BK index at different time frequencies. As shown in Table 3, there are three different time-frequency ranges, namely, Panel A, Panel B, and Panel C. Panel A is the table of the overflow index in the short-term (1–5 weeks) frequency band, Panel B is the table of the overflow index in the medium-term (5–20 weeks) frequency band, and Panel C is the table of the long-term (longer than 20 weeks) frequency band.
Regarding the results in Table 3, we focus on the “FROM_ABS” statistics. The total spillover index is 10.57% in the short term, and following the time-frequency band growth, the total spillover index dropped rapidly to 2.44% in the medium term and 0.90% in the long term. Therefore, the spillover effect has time-varying characteristics in the system, and the spillover effect between those markets mainly focuses on the short-term horizon.
Specifically, Panels A, B, and C show that the steam coal market is the largest spillover communicator for the CET market, while the spillover effect for the CET market to the energy market is nonsignificant. Furthermore, the new energy market is the largest spillover communicator for the traditional energy market in the short term, medium term, and long term.
5.3. Time-Varying Spillover Indices Analysis with Rolling-Window Analysis
Full sample analysis is insufficient to reveal the time variability of return series overflow; therefore, we measure the time variability of return series overflow using the dynamic spillover index which is named the DY index, and we use a rolling-sample estimation method to estimate the VAR model with the rolling window width W = 100 weeks which is approximately two years, the predictive horizon H is 100 weeks, and the VAR lag order p is 2 weeks.
5.3.1. Dynamic Total Spillover Effect Analysis among the Four Major Markets of China
The total spillover index of the return series in China’s four energy markets is shown in Figure 4. On average, the total spillover index of the four energy markets in China is 19.16%. From the analysis data, we know that this level is mainly determined by the connectedness between the traditional energy market and the new energy market. The total connectedness curve shows that the total volatility fluctuates greatly in the case of a rolling window of 100 weeks and varies from 12.68% to 24.43% because the sample periods span seven years and great changes occurred in China’s energy sector during this period. Starting in early December 2013, tremendous changes occurred in China’s economy, and China’s GDP increased from 9169.77 billion dollars in 2013 to 15711.53 billion dollars in 2020. China’s GDP growth rate remained at approximately 6%, except for being 1% in 2020, which was due to the impact of COVID-19 in that year. However, China’s economic development has shifted from a stage of rapid economic growth to a stage of high-quality development. With the growing awareness of high-quality development in China, the Chinese government has paid more attention and invested more in the energy sector and made increasingly more efforts to improve the energy environment, all of which inevitably had greater impacts on the energy environment market. Figure 4 shows that there are three obvious circles in this period.
The first circle, which maintained relatively high connectivity, started at the beginning of 2016 and ended in mid-2017. In this period, the Chinese government submitted to the United Nations “Strengthening Action on Climate Change—China’s Nationally Determined Contributions,” which proposed that China’s C02 emissions would peak in approximately 2030 and strive to reach the peak as soon as possible. The structural adjustment of China’s coal industry began in early 2016. At this point, the price of products in the energy industry will fluctuate sharply, and the risk will spill over to other energy markets so that the volatility between the markets will rise rapidly. Furthermore, various new energy cars have gradually entered the public eye, and major car companies have entered the new energy field.
The second circle began in mid-2017 and ended in Q4 2019, and China’s energy consumption structure changed rapidly in this period. The total returns connectedness shape decreased from the highest value of 24.43% in mid-2017 to the lowest value of 15.61% in mid-2018, jumped up to a temporary high point and then quickly fell back to the normal level. This small bump lasted only 2 months, and then the curve slowly rose to the end of the cycle at the end of 2019. The reason for the fluctuation at the end of 2018 can be ascribed to the influence of crude oil price fluctuations. As the impact of factors such as the resumption of U.S. sanctions on Iranian oil exports continued to decline into early 2019, the significant uncertainty in the energy market during this period resulted in a sharp increase of the spillover index among the domestic energy markets.
The third circle began in Q1 2020 and ended in mid-2021. In this period, the total spillover index fluctuated greatly, while the average level was relatively low. This phenomenon is due to the influence of COVID-19 on energy consumption sectors and the economy at the beginning of 2020. Due to COVID-19, the national economy almost halted and all energy consumption suddenly dropped until the end of June 2020 when the COVID-19 epidemic improved. China’s economic development gradually recovered, and the spillover effect of major energy markets increased sharply. This suggests that the impact of extreme events can make markets more interconnected, and the four energy markets have higher connectedness with each other over all of 2020.
5.3.2. Total Directional Connectedness over Time
In this section, we assess the dynamic total directional connectedness including the from-connectedness, to-connectedness, and net connectedness for the four energy markets in China. Figures 5–7 show each energy market’s dynamic to-connectedness, from-connectedness, and net connectedness, respectively.
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
Figure 5 shows the dynamic from-connectedness of the CET market and energy markets. Generally, the from-connectedness varies substantially across time and markets. It is obviously that the from-connectedness strength of the CET market is lower than those of the other three markets. From 2015 to 2017, the CET market had a higher from-connectedness and a lower to-connectedness. Figure 6 shows that from 2018 to 2020, the from-connectedness is lower, while the to-connectedness is higher. Notably, at the beginning of 2020, the from-connectedness hit rock bottom. This can be explained by the impacts of COVID-19 on economy, and in turn, it affects the four major energy markets. Then, the from-connectedness increased. In this system, the traditional energy market has the largest form-connectedness and the from-connectedness curve of the coal markets shows a large change during the entire period.
Figure 6 shows the dynamics of the to-connectedness for the CET market and three energy markets. We know that the connectedness of the CET market is also lower. Furthermore, the to-connectedness of the coal market and the new energy market is higher compared to their from-connectedness, and the traditional energy market is lower. Overall, the risk spillover between the CET market and the energy market fluctuates greatly, which indicates that the overall spillover of the energy-carbon system presents significant time-varying characteristics during the period of investigation.
Figure 7 shows the dynamics of the net connectedness for the CET market and the energy market in China. In the three circles, the CET market is a net receiver, net transmitter, and net receiver, respectively, and the steam coal market is always a net receiver, which may be because steam coal is the major energy source in the energy industry. The traditional energy market is always a net receiver, which indicates that traditional energy except for steam coal, such as oil and gas, is a net receiver of spillovers. The new energy market is mainly a net transmitter of volatility connectedness or shocks. The new energy industry is one of China’s emerging strategic industries, and it has a very strong momentum. In order to cope with the global warming trend, new energy must be future energy.
In order to demonstrate the role of the CET market in the financial market, this paper conducted a network analysis of the connectivity between CET markets in Beijing, Shanghai, Shenzhen, Guangdong, and Hubei exchanges. The data are selected for the weekly closing price from June 2014 to July 2021. The analysis results are shown in Table 4.
As we can see, there are 38.38 percent total variance influences from other trading markets in the system, which can demonstrate that the CET market in China can influence each other. The Beijing CET market has been most affected by other markets, while the Shanghai CET market has the biggest impact on other markets. In the given five CET markets, the Guangdong CET market price is most affected by the CET market price in Shanghai. The connectivity between the major CET markets in China shows that the carbon rights as a financial asset can effectively affect the carbon emission price.
Our findings are important for investors who have bought equities such as energy company. For example, when an investor owns a portfolio containing traditional energy shares and new energy shares, the close relationship between traditional shares and new energy shares reduces the diversified return strategy, and the investor needs to make appropriate adjustments to the trading strategy based on this time-varying information. Moreover, our findings about the relationship among the CET market, coal market, new energy market, and the traditional energy market play an important role in China’s scientific and technological development and environmental improvement. For example, through the fluctuation of the CET price, enterprises can be enforced to increase technological innovation to reduce carbon emissions.
5.4. Robustness Tests
In order to test the robustness of the aforementioned results, we use a variety of methods to test the return series spillovers among the CET market, coal market, new energy market, and the traditional energy market in China as the robustness test about eliminating the model assumptions’ condition was mentioned by Raquel M. Gaspar. In the application of the connectedness network proposed by Diebold and Yilmaz, there are three main parameters, such as the predictive horizon H for variance decomposition, the rolling window width W for the dynamic analysis, and the lag order p of the VAR model. In this section, we will test the robustness of the abovementioned results with different values of the three parameters and variable substitution method to test the model.
The results of the robustness test by the transformation parameters are shown in Figures 8–10, respectively. It is obviously reflected that the trend of the curves is consistent with the original results in different steps, different rolling windows, and different VAR lags’ order.
In addition, we use the index of the CSI Energy Index to replace the ASEI as the representative of the traditional energy market and then recalculate the spillover index and net-pairwise spillover index. The results show that the total spillover is 12.57 which is not far away from 13.91, and the net-pairwise of each market is in line with the original results. All of the abovementioned methods demonstrate the reliability of the original results.
6. Conclusion and the Policy Implication
With the intensification of the marketization process, the relationship among the CET, coal market, new energy market, and traditional market has been confirmed by many scholars. In this paper, we describe the static and dynamic influence relationship between the CET market and the new energy market, steam coal market, and the traditional market by constructing the VAR model. We first use the method of DY indices to study the network connectivity of the four markets in the temporal dimension, and then we use the method of BK to study the connectivity of CET, new energy, steam coal, and traditional in the frequency dimension; then, we study the dynamic connectivity among the four markets through a rolling window approach. At last, we test the result robust. The conclusions and enlightenment are as follows:
(1) From the static perspective, the results confirm the spillover effect among the CET, new energy, steam coal, and traditional energy with the total spillover effect index being 11.39% and the effect is mainly in the short term; (2) in all the markets, it is neutral that the spillover relationship among the CET, coal market, new energy market, and traditional energy market, while the steam coal spillover to CET is the highest with the spillover effect index being 1.33, and obviously, with the development of the CET market, the effect between them will be increased. New energy and the steam coal energy are net transmitters, while the traditional energy is a net receiver. (3) From the dynamic perspective, the spillover effect among the given markets has a time-varying characteristic, and the spillover index shows periodic changes, and it is affected by the international and domestic environment. Additionally, the results of the pairwise net directional spillover effects show that the new energy price returns play a dominant role in the total connectedness, followed by coal futures price returns. Furthermore, the traditional energy market plays the main net receiver role. Because traditional energy includes oil and gas without steam coal, we infer that the main net receiver is the oil and gas market in China.
The result indicates that the steam coal as the major energy source of the Chinese industry has a strong spillover effect, while new energy sources have strong development momentum under the background of the Chinese goal of carbon peak and carbon neutrality, and the new energy industry has been accepted by all sectors as an important way to achieve the dual-carbon goal in China. At the present stage, the aforementioned results also provide theoretical basis and support further research studies on cross-market and cross-regional information transmission and risk transmission mechanisms in the future, and it also provides a perspective to understand the connectivity and the spillover effect between the CET and the relevant energy market. The results can provide certain reference significance for marketing managers and formulate corresponding policy guidance for the policy markets, as well as for the investors. They can develop appropriate portfolios and hedge funds based on the connectivity results. In the future research, other commodity markets and a more broad range of data can be added to the research framework or we can use other methods to analyse the connectivity for the larger object.
Data Availability
The data that support the findings of this study are available on request from the corresponding author.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Authors’ Contributions
Fengwei Gao contributed to the conceptualization, writing reviews, editing, and reading the manuscript. Yimin Wu was responsible for the methodology, writing the review, and editing. Ding Chen was responsible for software. Mengyao Hu was responsible for validation. All the authors have read and agreed to the published version of the manuscript.
Acknowledgments
This research was funded by the key research project of the Suzhou University (grant: 2020yzd15 and 2022yzd14); Anhui Quality Engineering Project (grant: szxy2020xxkc04); Key Research Projects of Humanities and Social Sciences in Anhui Province (grant#: SK2021A0695).