Policy Impacts and Price Fluctuation Characteristics of China’s National Emissions Trading System

1.1 Research background

The issue of climate change is becoming increasingly urgent, prompting governments worldwide to prioritize the mitigation of greenhouse gas emissions [1]. Among the various strategies employed, the Emissions Trading System (ETS) has emerged as a critical policy instrument in the fight against global warming, achieving notable success in many regions [2]. Since 2006, China has overtaken the EU ETS as the largest emitter of carbon dioxide globally [3, 4]. The implementation of a national ETS has become a crucial strategy for tackling the urgent issue of carbon emissions. This market is a vital tool for China to attain its low-carbon goals, and underscores the significance for carbon trading. In October 2011, the Chinese government approved the implementation of the Twelfth Five-Year Plan, which aims to establish a domestic ETS. Subsequently, in 2013, eight pilot ETSs programs were launched across various provinces and cities. By July 2021, a nationwide carbon trading market encompassing the entire country was officially launched for trading [5].

As of July 2024, the CN ETS encompasses 2,257 key emission entities, with annual emissions amounting to approximately 5.1 billion tons. This represents over 40% of China's total carbon dioxide emissions, positioning the country as a global leader in greenhouse gas emissions trading. However, despite this achievement, the CN ETS is still in its early stages, with the institutional framework, sectoral coverage, trading mechanisms, market functions, and data management systems all requiring further improvement. Accelerating and enhancing the development of the CN ETS is essential to fully leverage the critical role of market prices in resource allocation. This is crucial for meeting emission reduction commitments, achieving emission control targets, and minimizing the costs of emission reductions across various industries [6]. Consequently, a comprehensive analysis of the policy impacts and price fluctuation characteristics of the CN ETS can provide valuable insights for refining the trading system, optimizing the pricing mechanism, mitigating market risks, and encouraging greater corporate participation in carbon trading to achieve emission reduction goals.

Firstly, the Bai-Perron test for multiple structural breaks was utilized to identify the precise number and timing of structural breakpoints in the CETP within the CN ETS. Secondly, using the event study method, we comprehensively analyzed the impact of various policy types, such as centralized trading before the compliance deadline, improving data quality and regulation, and introducing the auction mechanism, on the price stability of CN ETS. Thirdly, to gain deeper insights into the characteristics of the CN ETS and the nature of price volatility, the GARCH family models were employed to analyze the fluctuation patterns of CETP. The contributions of this study are manifold. Primarily, the impact of policy and regulatory changes on CETP in CN ETS is examined in detail for the first time, demonstrating the nonlinear interaction between regulatory intervention and price dynamics. Secondly, it analyzes the volatility patterns of CN CETP to reveal the nonlinear dynamics of the policy-sensitive market. This analysis can help assess market risk and stability. This study provides regulatory reference for policy makers, helps market participants grasp the law of price volatility patterns, and provides important inspiration for the sustainable development of CN ETS and emerging carbon markets.

The structure of the remainder of this research is organized as follows: Section 2 provides an overview of the relevant literature; Section 3 outlines the research methodology employed; Section 4 details the data sources utilized and presents the empirical analysis results; and Section 5 concludes the paper by summarizing the key findings and offering corresponding policy suggestions.

3.1 Structural breaks test model

Various structural breaks tests are available to identify changes in an econometric series. The Bai and Perron test can effectively identify multiple structural changes, which is essential for understanding how policies or events impact time series data. The model is primarily used to identify diverse breakpoints in a linear model, i.e., a significant change in the statistical properties of the time series at one or more points in time [40, 41].

The structural breaks test model assumes that the time series could be divided into multiple sub-intervals, with the data within each sub-interval following a different linear model. The Bai and Perron test methodology identifies the optimal number and location of breakpoints by comparing the model fit under different breaks assumptions. The CETP of CN ETS is known as the Chinese Emission Allowance (CEA). Drawing on structural breaks test conducted by previous researchers in other markets [8, 42], an equation is formulated to examine structural changes in the CEA.

$\begin{gathered}C E A_t=a_j+u_t \\ \left(t=T_j, T_{j+1}, \ldots, T_{j+1}-1 ; j=0,1, \ldots m\right)\end{gathered}$                  (1)

We consider a time series containing m potential mean breakpoints, and this sequence results in m+1 different time states. For each observation $\left(T_j, T_{j+1}, \ldots, T_{j+1}-1\right), C E A_t$ represents CETP in CN ETS, where is the expected CETP under regime j, and is the error in period t; $a_j$ denotes the mean of the price series, while the error term $u_t$ denotes the heteroskedasticity or correlation series. By reducing the residual sum of squares of the model equations and applying a specified information criterion that accounts for the number of breakpoints, we computed and evaluated the associated statistics $supF_T^*(k ; q)$, Dmax statistic (UDmax and WDmax), and ${Sup}_T(i+1 \mid i)$ to achieve efficient localization of the breakpoints. Finally, combining the event study methodology, we analyze how changes in policies and regulations affect CETP, and summarize the specific impacts of these policies on CEA fluctuations. This study not only deepens our comprehension of the dynamics within the carbon market, but also offers valuable reference information for policymakers [43].

3.2 GARCH family models

To study the fluctuation characteristics of the CETP in CN ETS, this study employs the GARCH family models, including GARCH, TGARCH, and EGARCH. These approaches are constructed to characterized and forecast conditional variance. They have significant advantages in capturing the fluctuation characteristics of financial time series data. When using these models, the analysis typically utilizes the returns on financial prices, which are calculated as the first-order logarithmic differences of closing prices between consecutive trading days. This study uses the return on CEA (r) for analysis. First, the return (r) is defined as follows:

$r_t=\ln \left(\frac{C E A_t}{C E A_{t-1}}\right)$             (2)

where, $C E A_t$ represents the CETP at time t. In all GARCH family models, the basic yield model and error term model are as follows:

$r_t=\mu+\epsilon_t$              (3)

$\epsilon_t=\sigma_t z_t$            (4)

In this context, $r_t$ signifies the return of the CETP at time t. The parameter μ represents the mean of these returns, $\epsilon_t$ denotes the error term or residual, $\sigma_t$ is the conditional standard deviation, and $z_t$ follows a standard normal distribution, characterized by independence. Then, this study will construct the conditional variance equations for the three models and analyze the fit comparison, and finally adopt the model with the best fit.

3.2.1 GARCH

The Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model represents an enhancement of the ARCH model, designed to more effectively capture the persistent volatility features, or long memory characteristics, inherent in financial time series data [44-46]. The conditional variance equation for the GARCH (p, q) model is as follows:

$\sigma_t^2=\alpha_0+\sum_{i=1}^p \alpha_i \epsilon_{t-i}^2+\sum_{j=1}^q \beta_j \sigma_{t-j}^2$                (5)

In this formulation, $\sigma_{t}^2$ represents the conditional variance. The term $\alpha_0$ denotes the constant component, while $\alpha_i$ and $\beta_j$ are the coefficients corresponding to the ARCH and GARCH terms, respectively. The variables p and q signify the orders of the error lag and the conditional variance lag, respectively. The parameters $\alpha_0, \alpha_i, \beta_j$ are subject to estimation.

3.2.2 TGARCH

The “Threshold GARCH” (TGARCH) model is frequently employed to capture the leverage effect in financial markets. This is achieved through the incorporation of a threshold mechanism that differentiates between the disparate impacts of positive and negative news on volatility. For the TGARCH(1,1), the conditional volatility is formulated as:

$\sigma_t^2=\alpha_0+\alpha_1 \epsilon_{t-1}^2+\gamma_1 \epsilon_{t-1}^2 I\left(\epsilon_{t-1}<0\right)+\beta_1 \sigma_{t-1}^2$                 (6)

where, $\alpha_0, \alpha_1, \gamma_1, \beta_1$ are the parameters to be estimated. $\alpha_1$ is the coefficient of the lagged error squared term; $\gamma_1$ is the coefficient of the additional error squared term, which denotes the additional effect of the negative errors on the volatility; $\beta_1$ is the coefficient of the lagged conditional variance term. When I = 1, it signifies the influence of positive news, whereas I = 0 indicates the influence of negative news. The parameter $\gamma_1$ is utilized to capture the asymmetry in the effects of positive and negative news on the financial market's volatility. A notably non-zero coefficient for $\gamma_1$ suggests an asymmetric reaction of conditional variance to good and bad news. Specifically, a negative value for $\gamma_1$ implies that bad news has a more pronounced effect on volatility compared to good news, while a positive value for $\gamma_1$ indicates that positive news exerts a greater influence [37, 47].

3.2.3 EGARCH

The “Exponential GARCH” (EGARCH) model addresses the limitations of the GARCH model in detecting volatility asymmetries by modeling the logarithm of conditional variance [48-50]. Its equation is as follows:

$\ln \left(\sigma_t^2\right)=\omega+\beta \ln \left(\sigma_{t-1}^2\right)+\alpha z_{t-1}+\gamma\left[\left|z_{t-1}\right|-E|z|\right]$                (7)

where, ω, α, β and γ are the parameters to be estimated; ω is a constant term indicating the average of the long-run volatility level; β is the autoregressive coefficient of conditional variance reflecting the persistence of volatility; α captures the asymmetric effect (the different impacts of positive and negative shocks). When α < 0, $\ln(\sigma_t^2)$ rises (falls) in response to the price decline (rise) of the negative market shock z. $\gamma$ captures the symmetric volatility effect (the effect of the absolute value of the shock), $\gamma\left[\left|z_{t-1}\right|-E|z|\right]$ measures the magnitude effect of the innovation. If $\gamma$> 0, then $\gamma\left[\left|z_{t-1}\right|-E|z|\right]$ raises (lowers) $\ln(\sigma_t^2)$ when the extent of the market shock exceeds (is below) expectations [51].

3.3 Data description

This study analyzes the CETP using data on China Emission Allowance (CEA) from CN ETS, which launched on July 16, 2021. The daily average price of CEA was collected from July 16, 2021, to July 31, 2024, sourced from China's Wind database. Given that the vast majority of missing values in the data occur during holiday periods, we choose to delete these missing values to ensure data authenticity.

5.1 Comparative analysis with the EU ETS

The EU ETS, launched in 2005, has gradually expanded to cover multiple high-emission sectors such as power generation, industry, and aviation, making it the most mature and representative carbon market globally. In contrast, CN ETS, at its initial stage, covers only the power sector, with a coverage of approximately 5.1 billion tons of CO₂ emissions, accounting for over 40% of China's total emissions. This positions it as the largest carbon market globally.

5.1.1 Comparison of market design and operational mechanisms

As the two most prominent ETSs worldwide, the EU ETS and CN ETS share commonalities in their market design, operational mechanisms, and policy goals, while also displaying notable distinctions. Their similarities are primarily rooted in the adoption of a cap-and-trade framework, which imposes a cap on carbon emissions and allocates emission allowances to encourage enterprises to meet emission reduction targets through market-driven mechanisms. Furthermore, both markets have progressively expanded their sectoral coverage and placed significant emphasis on improving data quality and regulatory oversight. It is worth noting that the EU ETS experienced a similar pilot phase during its early stages, gradually incorporating more sectors and emission sources, mirroring the developmental trajectory of China's ETS as it transitioned from pilot programs to a nationwide market.

There are three main differences. First, the allocation of allowances. The EU ETS has gradually introduced the auction mechanism since the third phase (2013), while CN ETS is still dominated by free allocation. Second, the price mechanism. The EU ETS regulates CETP volatility by the establishment of the Market Stabilization Reserve (MSR), and has already established a future market, but the CN ETS has not yet established a similar price stabilization mechanism. Third, the data quality and compliance. While the EU ETS has strict data quality and compliance requirements with strong penalties for violations, the CN ETS still needs to be further improved in terms of data quality management and corporate compliance.

5.1.2 Comparison of the impact of policy changes on CETP

During its early development, the EU ETS underwent several major policy adjustments that significantly influenced CETP. This is comparable to the policy sensitivity observed in CN ETS, as highlighted in this study.

i. Concentrated trading before compliance deadlines

The 2024 Carbon Market Report emphasized that the surge in market demand prior to compliance deadlines often leads to an increase in CETP. By analyzing trading data from 2022 and 2023, the report found that the CETP increases observed in the first quarter of each year were directly linked to concentrated trading activity driven by compliance requirements [57]. This pattern closely aligns with the CETP growth observed in the CN ETS ahead of the deadline for its first compliance cycle, suggesting that this is a common characteristic of carbon markets rather than a phenomenon unique to the CN ETS.

ii. Delayed and Uneven Allowance Allocation

The European Parliament noted that uneven allocation of allowances could lead to market distortions, particularly when certain sectors receive more free allowances. Such policies may reduce market liquidity, thereby driving up CETP [58]. This is similar to the phenomenon observed in this study, where inconsistencies in allowance issuance timing across provinces in China led to supply-demand imbalances and CETP increases. However, due to the more mature market mechanisms and higher liquidity of the EU ETS, the impact of such issues is relatively smaller.

iii. The relationship between the auction mechanism and CETP

As highlighted in the literature review, some studies suggest that increasing the number of auctioned allowances is a key driver of rising CETP in the EU ETS [24]. However, the 2024 Carbon Market Report noted that the total volume of allowances auctioned in 2023 increased by 7%, reaching 523 million tons of CO₂ equivalent. This rise was partially influenced by the renewable energy acceleration plan of EU, which prompted the early auctioning of certain allowances to facilitate the energy transition. This temporary increase in supply placed downward pressure on CETP [57]. This phenomenon illustrates the complexity of the auction mechanism's impact on CETP, as it depends on factors such as the specific implementation method, timing, and market expectations. When auctions are viewed as a long-term strategy to enhance market efficiency and improve price discovery, they tend to drive CETP upward. In contrast, when auctions are employed to temporarily boost the supply of allowances, they can exert downward pressure on CETP.

This complexity carries significant implications for China's ETS. When implementing an auction mechanism, it is crucial to thoroughly assess the objectives, timing, and prevailing market conditions to prevent excessive price volatility resulting from improper execution. The observed CETP increase in CN ETS, driven by expectations surrounding the introduction of an auction mechanism, likely reflects the market's optimistic anticipation of auctions as a component of long-term market-oriented reforms, rather than a response to short-term supply fluctuations.

iv. The impact of strengthening data quality and regulation

After the EU ETS implemented of several regulatory enhancements, such as new anti-fraud measures, registry regulations, and the Market Stability Reserve (MSR), the CETP increased from €25 per ton of CO₂ in 2020 to nearly €100 per ton of CO₂ in 2022 [26]. A similar CETP increase was observed in the CN ETS in response to policies focused on improving data quality and regulatory oversight, as identified in this study. This underscores the importance of a robust regulatory framework in fostering market confidence and enabling effective price discovery.

v. The impact of sectoral expansion

The effect of expanding sectoral coverage on CETP within the EU ETS reveals a complex and dynamic relationship. Hu et al. [59] found that integrating high abatement cost sectors with substantial allowance demand, such as the aviation industry, into the EU ETS led to increased allowance demand, thereby driving carbon prices higher. In contrast, Hintermann et al. [60] noted that incorporating Kyoto Protocol flexible mechanisms into the EU ETS introduced additional allowance supply, which lowered the market equilibrium price. This complexity differs from the upward price impact of sectoral expansion observed in the CN ETS in this study, potentially reflecting variations in the developmental stages and market structures of the two systems.

5.1.3 Comparison of the CETP fluctuation characteristics

Jiang et al. [30] showed that CETP fluctuations in the EU ETS display typical characteristics such as price spikes, fat tails, autocorrelation, volatility clustering, and conditional variance, reflecting significant risks of extreme price movements in carbon markets. These findings align closely with the CETP fluctuation patterns observed in the CN ETS, indicating that such traits may be universal features of carbon markets.

Nevertheless, there are notable differences in the specific manifestations of volatility between the two systems. Yu et al. [31] observed that CETP returns in the EU ETS exhibit a distinct non-normal distribution, marked by left skewness and high kurtosis, which points to the presence of extreme values and fat-tail phenomena in price fluctuations. In contrast, this study finds that the returns of CN ETS exhibit slight left skewness, with a kurtosis of 9.76, significantly higher than that of the EU ETS, suggesting that extreme fluctuations are more frequent in CN ETS.

Regarding market efficiency, the EU ETS demonstrated weak-form efficiency only during Phase II (2008-2012). In contrast, inefficiencies were evident in Phase I and Phase III, driven by factors such as high transaction costs, information asymmetry, and irrational investor behavior, which contributed to heightened price volatility. Similar findings are noted in this study for the CN ETS, where high volatility persistence and significant autocorrelation indicate that both markets encounter challenges related to market efficiency at various stages of development.

Moreover, Li et al. [32] identified that the EU ETS exhibits multifractal characteristics and nonlinear complex correlations, implying that price volatility possesses long-term memory and dependencies across multiple time scales. This aligns with the long-term memory traits of price volatility observed in the CN ETS in this study. However, this feature may be more pronounced in CN ETS, as evidenced by the TGARCH model, where the α+β value reaches 1.86, far exceeding the constraint condition of 1.

It is worth noting that the two markets exhibit significant differences in terms of the leverage effect. The EU ETS demonstrates a typical leverage effect, where negative news exerts a greater influence on price volatility compared to positive news [31, 33]. In contrast, the CN ETS displays an "anti-leverage effect." This distinction highlights a fundamental disparity in the composition of market participants between the two regions. The EU ETS includes a diverse array of participants, such as industrial enterprises, financial institutions, and professional traders. It possesses strong financial characteristics, with active derivatives trading, and market prices are primarily determined by supply and demand mechanisms. Conversely, the CN ETS is predominantly composed of allowance-regulated enterprises, with limited involvement from financial institutions. Transactions in this market are more policy-driven, rendering it particularly sensitive to favorable policy signals, while its financial attributes remain relatively weaker.

5.2 Policy implications

5.2.1 Improve the market mechanism and strengthen policy coordination

The Chinese government needs to design a market stabilization and reserve mechanism to prevent extreme volatility. At the same time, the compliance mechanism and allowance allocation should be optimized to ensure that the allowance issuance time for each province is consistent, as well as the timing and quantity of auctions should reasonably designed to avoid downward pressure on CETP caused by a large the number of short-term auctions. In addition, it is necessary to improve the transparency and predictability of policies to reduce the impact of sudden policy changes on the market. Ensure the consistency of carbon market policies with energy policies and industrial policies to avoid large price fluctuations caused by policy conflicts.

5.2.2 Improve data quality and supervision efficiency

A more rigorous and transparent Monitoring, Reporting and Verification (MRV) system is fundamental to improving market confidence, and the EU ETS experience should be used to improve data quality and reliability. Meanwhile, the CN ETS regulators should strengthen market supervision, increase the cost of violations, and strengthen the binding force of compliance. In addition, they should improve the information disclosure mechanism, regularly release market operation reports and policy interpretations, improve market transparency, and help participants form reasonable expectations.

5.2.3 Enrich market structure and functions

The CN ETS authorities should steadily promote the expansion of sectors, orderly include high-emission sectors in the CN ETS, and expand the scale of the market. According to the characteristics of "anti-leverage effect" found in this study, the scope of market participants should be expanded, financial institutions and individual investors should be introduced in an orderly manner, the current single market structure dominated by performing enterprises should be improved, and market liquidity should be increased. And the Chinese regulators should develop derivatives such as carbon futures and options, provide diversified risk management tools, and enhance market depth.

By implementing these measures, the issues of price volatility and policy sensitivity in the CN ETS can be effectively mitigated. This would facilitate the market's progression toward greater maturity, stability, and efficiency, while also offering valuable lessons for other emerging ETSs.