Does climate policy uncertainty affect carbon emissions in China? A novel dynamic ARDL simulation perspective

Abstract

This study provides new insights into the impact of climate policy uncertainty, energy consumption, and economic development on China’s carbon emissions. In doing so, we develop a novel index of China’s climate policy uncertainty (CCPU). We then use the newly constructed dynamic autoregressive distributed lag (ARDL) simulation model, the frequency-domain causality (FDC) test, and the fully modified OLS (FMOLS) estimation to investigate these potential relationships from 2005 to 2021. The empirical results suggest that increasing CCPU reduces carbon emissions in most parts of China, which improves environmental degradation. Furthermore, the effects of energy consumption and economic growth on carbon emissions are confirmed to be positive in each location. Finally, the results of the FDC and FMOLS confirm the robustness of the model. Our findings suggest that information from the CCPU can be used to forecast CO₂ emissions in China. Furthermore, the government should strike a balance between economic growth and environmental regulation and promote the use of renewable energy to reduce carbon emissions. Proactively developing climate policy is important to achieve the goal of carbon neutrality.

Introduction

Climate risk has created an unparalleled obstacle to long-term economic growth in a large number of countries (Diffenbaugh and Burke, 2019; Pástor et al., 2021). The Paris Agreement, which went into force in November 2016, is a significant milestone that has strengthened the global response to climate change and provided a new framework for climate policy. Consequently, policies aimed at addressing climate change have been of high priority to governments for decades and all are working to reduce greenhouse gas (GHG) emissions (Lee and Chen, 2020; Sun et al., 2021). Nevertheless, there still exist significant uncertainties in the course of such climate policy implementation. The most recent example is the US withdrawal from the Paris Agreement in 2017, which has created significant uncertainty regarding the execution of climate policies.

Climate policy uncertainty refers to the degree of ambiguity and unpredictability surrounding government policies and regulations related to climate change mitigation and adaptation. It arises from potential changes in policy or regulation, such as the implementation or elimination of carbon taxes, subsidies for renewable energy, or the adoption of new emissions targets. These uncertainties might have far-reaching consequences for the macroeconomy and the carbon neutrality target (Lee et al., 2021; Su et al., 2021; Wang et al., 2019; Wen et al., 2022).

Over the past two decades, China has experienced the highest rate of economic expansion and energy consumption among emerging economies (Khan et al., 2022). The past economic development model has led to a multiplication of greenhouse gas (GHG) emissions such as CO₂, accounting for 27% of the total global GHG emissions in 2019. However, the Paris Agreement aims to cut CO₂ emissions and mitigate climate deterioration, with the objective of keeping the global temperature rise under 2°C within this century. As the largest carbon emitter, China needs to take action. Indeed, the Chinese government has already committed to peaking its CO₂ emissions by 2030 and achieving the carbon neutrality target by 2060 (Wang et al., 2019). To this end, it is of great policy importance to measure how China’s climate policy has changed and whether the policy changes have an impact on carbon emissions. Understanding the influence of climate policy uncertainty on carbon emissions and investigating the underlying mechanism is vital for authorities and policymakers to achieve these goals in China.

Among existing studies, several potential channels could explain the theoretical link between policy uncertainty and CO₂ emission (Jiang et al., 2019; Ullah et al., 2021; Yu et al., 2021; Wan et al., 2022; Wen et al., 2022). These include the direct policy adjustment channel (Jiang et al., 2019), the energy intensity effect (Yu et al., 2021), and the innovation investment channel (Wan et al., 2022; Wen et al., 2022). For example, high levels of uncertainty about climate risk or climate policy could hinder energy consumption and nonessential transportation for individuals and firms. At the same time, the increase in CPU might encourage renewable energy consumption and increase research and development (R&D) innovation on climate-friendly innovation, thus leading to lower CO₂ emissions (Gavriilidis, 2021).

Considering the abovementioned reality, the current study aims to analyze the influence of CPU on carbon emissions in China. The following are the main contributions of the current paper to the literature. First, we made a novel China’s climate policy uncertainty index. We take newspaper content from nine major Chinese newspapers and then search for relevant keywords to construct the index. Our novel index is in line with key domestic climate policy changes. The index construction provides a new perspective for quantifying climate policy change. Second, although some studies have examined the impact of basic determinants on CO₂ emissions such as economic growth and energy consumption, they have not elaborated on the impacts of climate policy uncertainty on CO₂ emissions. In view of this, this study considers the important role of climate policy uncertainty and analyzes whether and how uncertainty regarding climate policy affects CO₂ emissions, enriching the study of the macro effects of climate policy. Moreover, to uncover the heterogeneity of the influence of climate policy uncertainty on carbon emissions, this paper explores this relationship from a regional perspective. Finally, we employ the dynamic autoregressive distributed lag (DARDL) framework to reveal the long- and short-run effects of the concerned variables on carbon emissions and the frequency-domain causality (FDC) test for the robustness check. By doing so, our papers contribute to a better understanding of how climate policy uncertainty affects carbon emissions, which helps to achieve the goal of carbon neutrality.

Literature review

The impact of policy uncertainty, energy consumption, and economic growth on carbon emissions has been the subject of extensive research in recent years, particularly in the context of the energy transition and low-carbon development initiatives. Understanding the relationship between these factors is crucial for effective policy-making and sustainable development.

Several studies have examined the influence of energy consumption on carbon emissions. For example, Adedoyin and Zakari (2020) conducted a study on the UK and found that energy use has a significant and beneficial effect on carbon emissions in the long term. Adams et al. (2020) investigated 10 resource-rich economies and identified a significant relationship between energy use and carbon emissions. Similarly, Abbasi and Adedoyin (2021) focused on China and found that energy consumption has a substantial positive impact on carbon emissions in both the short and long term.

Moreover, for the sake of sustainable development, scholars have studied the relationship between renewable energy consumption and carbon emissions. Khan et al. (2022) examined four East Asian economies and found that renewable energy consumption mitigates CO₂ emissions. Atsu and Adams (2021) analyzed BRICS countries and found a significantly negative impact of renewable energy use on carbon emissions. However, Xue et al. (2022) found no significant long-term effect of clean energy usage on carbon emissions in France.

Besides that, the existing research has also extensively explored the relationship between economic growth and carbon emissions, often using the environmental Kuznets curve (EKC) hypothesis. Adedoyin and Zakari (2020) found a significant positive effect of economic growth on carbon emissions in the UK. Shahbaz et al. (2020) supported the EKC hypothesis by finding an inverted U-shaped relationship between carbon emissions and economic development in the UK. Adams et al. (2020) identified a positive relationship between economic growth and carbon emissions in 10 resource-rich countries. Abbasi and Adedoyin (2021) found significant positive effects of economic growth on China’s carbon emissions. Syed et al. (2022) documented that economic development increases carbon emissions across different quantiles in BRICS economies.

In addition, the impact of policy uncertainty on carbon emissions has gained attention in recent years, but the findings in the literature are inconsistent. Some studies have found positive relationships, indicating that increasing policy uncertainty can lead to environmental aggravation. Jiang et al. (2019), Adams et al. (2020), and Pirgaip and Dinçergök (2020) found positive correlations between policy uncertainty and carbon emissions. Atsu and Adams (2021) also found a significant positive impact of policy uncertainty on carbon emissions in BRICS economies. Anser et al. (2021) and Yu et al. (2021) provided further evidence of the positive effects of policy uncertainty on carbon emissions.

Contrary to these findings, some studies have shown that increased policy uncertainty could slow carbon emissions. Adedoyin and Zakari (2020), Ahmed et al. (2021), and Syed et al. (2022) found evidence of the negative impact of policy uncertainty on carbon emissions. Gavriilidis (2021) and Liu and Zhang (2022) documented the negative influences of policy uncertainty on carbon emissions in their respective studies. However, contradictory findings have also been reported, with Abbasi and Adedoyin (2021) and Nakhli et al. (2022) finding no significant influence of policy uncertainty on carbon emissions in specific contexts. The disparities in findings may be attributed to variations in sampling periods and methodological frameworks employed in these studies.

Several shortcomings and weaknesses should be addressed and improved upon based on the preceding literature research. First, although several studies in the literature have focused on the factors driving China’s carbon emissions and considered the role of energy use and economic development, the majority of them have primarily used national carbon emissions and do not consider the impact of policy uncertainty. Besides that, several types of research have uncovered that carbon emissions are influenced by policy uncertainty. However, few studies detect the effect of climate policy uncertainty on CO₂ emissions. Climate challenges have been increasingly prominent in recent years, as have the corresponding climate governance policy uncertainties. Therefore, exploring the impact of climate policy uncertainty on carbon emissions has policy implications. Based on the above discussion, the current paper aims to add to the body of evidence that climate policy uncertainty can be used to explain changes in China’s CO₂ emissions.

Theoretical model, methodology, and data

Theoretical model

In this section, we adopt the approach of Fried et al. (2021) and employ a general dynamic environment Cobb-Douglas model. This model is a well-established and widely used theoretical approach for analyzing the relationships between environmental issues and underlying determinants. Our goal is to uncover the influence of CCPU on carbon emissions. We assume that the economy comprises entrepreneurs and workers with infinite lives, which has two sectors: the “fossil” sector that emits carbon and the “clean” sector that does not. y is the final good in the economy, which is produced from labor input l, a clean intermediate input x^c, and a carbon-intensive fossil intermediate input x^f. Moreover, we assume that interest rates and labor supply are exogenous variables. Therefore, the aggregate output y can be written as:

$$y = (x^c)^\gamma (x^c)^\theta l^{1 - \gamma - \theta }$$

(1)

where γ and θ indicate the factor shares of the clean and fossil inputs, x^c is clean capital (k^c), and x^f is fossil intermediate that satisfies the minimum between fossil fuel and fossil capital (min[k^f, f]). Both production functions have constant returns to scale.

Taking prices as given, the representative firms choose the optimal share of fossil and clean inputs and labor to maximize profits. If the government implements a specific climate policy, all choices are made at the start of the term. The first-order conditions (F.O.C.) suggest the following expressions for the prices of the fossil and clean inputs, p^f and p^c:

$${{p}^{f}} = \theta (x^c)^\gamma (x^f)^{\theta - 1}\,{{{\mathrm{and}}}}\,p^c = \gamma (x^c)^{\gamma - 1}(x^f)^\theta$$

(2)

Let V^c(k^c) represent the clean firm’s value function in the steady-state before the policy change, and T^c(k^c) imply the value function in period t of the transition after the government implements the new climate policy. Therefore, the clean firm’s value function in the steady state before the policy change equals:

$$\begin{array}{l}V^c(k^c) = \mathop {{\max }}\limits_{k^{c^{\prime}}} \left\{ {p^ck^c - i^c\left. { + \left( {\frac{1}{{1 + r}}} \right)[\rho T_1^c(k^{c^\prime }) + (1 - \rho )V^c(k^{c^\prime })]} \right\}} \right.\\ s.t.\,k^{c^\prime } = (1 - \delta )k^c + i^c\end{array}$$

(3)

where p^ck^c denotes the total revenue from production, i^c represents the investment ρ is the probability of uncertainty due to the government introduction of climate policy, r indicates the exogenous interest rate, and δ denotes the depreciation rate.

Moreover, the clean firm’s value function in period t of the transition equals:

$$\begin{array}{l}T_t^c(k^c) = \mathop {{\max }}\limits_{k^{c^{\prime}}} \left\{ {p^ck^c - i^c\left. { + \left( {\frac{1}{{1 + r}}} \right)T_{t + 1}^c(k^{c^\prime })} \right\}} \right.\\ s.t.\,k^{c^\prime } = (1 - \delta )k^c + i^c\end{array}$$

(4)

Since all uncertainty is eliminated after the implementation of the established climate policy by the government, the firm’s value in the period t of the transition equals the value function in the period t + 1 of the transition.

In a similar vein, the fossil firm’s value function in the pre-policy steady-state equals:

$$\begin{array}{l}{{V}^{f}}(k^f) = \mathop {{\max }}\limits_{k^{{f}^{\prime} },f} \left\{ {p^fk^f - \xi f - {{i}^{f}}\left. { + \left( {\frac{1}{{1 + r}}} \right)[\rho {{T}_{1}^{f}}(k^{{f}^{\prime} }) + (1 - \rho )V^f(k^{{f}^{\prime} })]} \right\}} \right.\\ s.t.\,k^{{f}^{\prime} } = (1 - \delta ){{k}^{f}} + {{i}^{f}}\end{array}$$

(5)

where p^fk^f denotes the total revenue from production, ξ^f implies the expenses on fossil fuel, i^f represents the investment ρ is the probability of uncertainty due to government introduction of climate policy, r indicates the exogenous interest rate, and δ denotes the depreciation rate.

Likewise, the fossil firm’s value function in period t of the transition equals:

$$\begin{array}{l}{{T}_{t}}^{f({k}^{f})} = \mathop {{\max }}\limits_{k^{{{f}^{\prime}}},f} \left\{ {p^fk^f - \xi f - i^f\left. { + \left( {\frac{1}{{1 + r}}} \right)T_{t + 1}^f(k^{{{f}^{\prime}} })} \right\}} \right.\\ s.t.\,k^{{f}^{\prime} } = (1 - \delta )k^f + i^f\end{array}$$

(6)

Therefore, by constructing the Lagrange equation, we solve that:

$$K^c = \left( {\frac{r}{{r + \delta }}} \right)^{\frac{1}{{1 - \gamma - \theta }}}\left( {\frac{{r + \delta }}{{r + \delta + \xi + \rho }}} \right)^{\frac{\theta }{{1 - \gamma - \theta }}}\left( {\frac{\theta }{r}} \right)^{\frac{\theta }{{1 - \gamma - \theta }}}$$

(7)

$$K^f = \left( {\frac{r}{{r + \delta }}} \right)^{\frac{1}{{1 - \gamma - \theta }}}\left( {\frac{{r + \delta }}{{r + \delta + \xi + \rho }}} \right)^{\frac{{1 - r}}{{1 - \gamma - \theta }}}\left( {\frac{\theta }{r}} \right)^{\frac{{1 - r}}{{1 - \gamma - \theta }}}$$

(8)

Dividing Eq. (7) by Eq. (8), we finally obtain the following relationship between ρ and the ratio of fossil capital to clean capital:

$$\frac{{{K}^{f}}}{{K}^{c}} = \left( {\frac{{r + \delta }}{{r + \delta + \xi + \rho }}} \right)\left( {\frac{\theta }{r}} \right)$$

(9)

Specifically, the ratio of fossil capital to clean capital decreases when climate policy uncertainty increases (ρ increase). Moreover, since the equilibrium use of fossil fuels is equal to the level of fossil capital, a lower ratio of fossil to clean capital further reduces carbon emissions from firm production.

Methodology

To investigate the influence of CCPU on carbon emissions, we employ the novel DARDL simulation method to uncover this effect. Unlike the conventional ARDL model, the novel DARDL approach proposed by Jordan and Philips (2019) can forecast and automatically plot one predictive shift on the dependent variable. Moreover, the DARDL model maintains the stability of the other independent variables when analyzing the effects of short- and long-term estimations. Besides that, the DARDL approach also has higher robustness in small sample cases, which is in line with the current dataset. Another advantage of DARDL estimation is that the co-integration test can be performed as long as they are not I (2), which is much different from the traditional method.

Jordan and Philips (2019) state that the DARDL error correction form in the current study is as follows:

$$\begin{array}{c}{\it{\Delta }}({\it{ln}}CO_2)_t = \alpha _0 + \theta _0({\it{ln}}CO_2)_{t - 1} + \beta _1{\it{\Delta ln}}CCPU_t + \theta _1{\it{ln}}CCPU_{t - 1}\\ + \beta _2{\it{\Delta ln{\rm E}}}C_t + \theta _2{\it{ln}}EC_{t - 1} + \beta _3{\it{\Delta ln}}GDP_t + \theta _3{\it{ln}}GDP_{t - 1} + \varepsilon _t\end{array}$$

(10)

where CO₂ denotes the carbon dioxide emission, the explained variable in the current study. CCPU is the climate policy uncertainty in China, the main explanatory variable. EC implies energy consumption and GDP shows the gross domestic product, which are the control variables in this paper. ε_t is the error term.

Furthermore, the FDC introduced by Breitung and Candelon (2006) is used in this article to support the robustness of the DARDL model. Compared with the traditional Granger causality test, the FDC analysis method used in this study helps to predict the response between variables within a specific time frequency. The FDC equation is expressed as follows:

$$x_t = \alpha _1x_{t - 1} + \alpha _2x_{t - 2} + \cdots + \alpha _px_{t - p} + \beta _1y_{t - 1} + \beta _2y_{t - 2} + \cdots + \beta _py_{t - p} + \varepsilon _t$$

(11)

where ε_t is the error term and α and β are the estimated parameters in time (t), and lag (p), respectively. Figure 1 reveals the framework of the methodology in the study.

Fig. 1

Methodological framework in the current study.

Data collection

The primary goal of this article is to investigate the impact of CCPU on carbon emissions in six distinct areas of China. To this end, we have followed the methods of Baker et al. (2016) and Huang and Luk (2020) to construct a novel CCPU index¹. The datasets of CO₂ emissions in the current paper have been collected from the China Energy Statistics Yearbook (Table 1).

Table 1 Regional divisions.

In addition, to rule out the influence of other underlying factors on CO₂ emissions, we used the principles given by Shahbaz et al. (2018), Adams et al. (2020), Adedoyin and Zakari (2020), Shahbaz et al. (2020), Abbasi et al. (2021), and Syed et al. (2022), controlling the effect of economic growth and energy use on carbon emissions. Specifically, we have selected the total energy consumption (EC) and the gross domestic product (GDP) of each region as proxy variables. The corresponding data were collected from the Wind Economic Database and the National Bureau of Statistics of China. Considering the availability of the data, the datasets in the current study span from 2005 to 2021. In addition, we use a logarithmic form for each data point. Table 2 reports the statistical characteristics of the CO₂ emissions, CCPU, EC, and GDP in different regions of China.

Table 2 Descriptive statistics.

The descriptive statistics show that these series are clearly heterogeneous. To begin with, there are significant differences in the logarithmic pattern of the CO₂ emission series for each region. Second, we have also noticed that the averages of CO₂ emissions are much larger than those of the other variables for each region. Third, the standard deviation indicates that the GDP value is the largest, followed by the CCPU, suggesting that there have been more changes in climate policies during the sample period, and each region also witnessed a great fluctuation in economic growth. In addition, the normal distribution trend of each series has been statistically verified by the Jarque-Bera test, as each series cannot reject the null hypothesis.

Empirical result

Unit root test

Table 3 shows the unit root test results for each concerned variable. The Augmented-Dickey-Fuller (ADF) test, Phillips-Perron (PP) test, and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test are jointly utilized to examine the stationarity of these series. Based on the outcomes of the ADF test, PP test, and KPSS test, we find that CCPU and CO₂ emissions for North China and Northwest China series are I(1) processes, whereas the remaining variables are I(0) processes. Overall, the evidence of the unit root test reveals that none of the series is stationary at I(2), which meets the requirements of the dynamic ARDL model simulation.

Table 3 Unit root examinations.

ARDL bounds test analysis

Since not all series are I(1) processes, it becomes critical to test the validity of the co-integration relationship. This paper uses the ARDL bound technique to detect whether there is a long-term link between the variables. Table 4 displays the outcomes of the ARDL bound test for different regions in China. The t-statistic value is discovered to be larger than the upper limit value at the 10% level of significance, except for Northeast China. Moreover, the F-statistic is also utilized for the evaluation of co-integration and we note that the F-statistic value is higher than the upper bound value at the 1% level of significance for each region. Combining the above two statistical results, we conclude that there is co-integration between the series.

Table 4 Bounds test results.

Dynamic ARDL estimation results

The outcomes of the dynamic ARDL estimation for six regions of China are shown in Table 5. First, the adjustment speed from short-term disequilibrium to a new long-term equilibrium is assessed by the error correction term (ECT). We notice that the coefficient of ECT for each scenario is significantly negative, which suggests that 35.8% to 88.8% of the disequilibrium is corrected in the long term. The results are consistent with Abbasi and Adedoyin (2021) and Abbasi et al. (2022), who report that the value of ECT is negative indicating that the adjustment speed is significant.

Table 5 Dynamic ARDL simulation results.

Specifically, from the short-term perspective, the coefficient of CCPU is negative and significant in Eastern (–0.020) and Southcentral China (–0.051). These outcomes show that a 1% increase in CCPU would decrease carbon emissions by 0.020% and 0.051% in both regions (Liu and Zhang, 2022). In other words, the CCPU has a beneficial effect on environmental improvement. These results are supported by Pirgaip and Dinçergök (2020), Gavriilidis (2021), and Liu and Zhang (2022). However, the coefficient of CCPU is significantly positive in North China (0.093), suggesting that the positive shock of CCPU has a positive influence on environmental deterioration. North China is rich in coal resources, and the coal resources in this region are most concentrated in Shanxi Province. Therefore, industries in this region rely heavily on carbon-intensive production methods. These industries might increase production ahead of future climate policy implementation, leading to increased carbon emissions. This finding backs with previous research (Adams et al., 2020; Anser et al., 2021).

In the long-term perspective, we observe a significant and negative coefficient of CCPU for most regions, indicating that higher levels of policy uncertainty have a dampening effect on CO₂ emissions. This finding suggests that when faced with uncertain climate policies, economic agents tend to err on the side of caution and adopt measures that lead to a reduction in energy consumption and thus lower carbon emissions (Gavriilidis, 2021). This economic behavior can be attributed to the risk-averse nature of firms and individuals, who are uncertain about the future regulatory environment and therefore choose more conservative energy use strategies to make long-term investment decisions aimed at reducing their carbon footprint.

However, the estimated result for Southwest China presents a contrasting picture. We find that a 1% positive shock in CCPU leads to an increase in CO₂ emissions by 0.121%. This implies that in Southwest China, higher levels of policy uncertainty can have a stimulating effect on carbon emissions. The outcomes are in line with Atsu and Adams (2021) and Xue et al. (2022). One economic explanation for this phenomenon is that the region is characterized by industries that heavily rely on carbon-intensive production methods. In the face of uncertain climate policies, these industries may perceive a temporary window of opportunity to increase production before potential future policies are implemented, which increases emissions.

In addition, the estimated short-term and long-term coefficients of EC are significant and show positive signs in each scenario, indicating that EC has a positive influence on CO₂ emissions. This outcome is similar to Lin and Xu (2020) and Abbasi and Adedoyin (2021). This finding can be attributed to the fundamental relationship between energy consumption and economic activity. As economic growth and industrial production increase, the energy demand also increases, leading to higher levels of energy consumption and subsequent carbon emissions. Our findings suggest that the growing energy consumption without a proper replenishment mechanism will further accelerate environmental degradation and hinder its achievement of carbon neutrality goals.

In addition, the empirical outcomes show that 1% GDP growth will lead to a short-term increase in CO₂ emissions ranging from 0.116% to 0.935%, and a long-term increase ranging from 0.103% to 0.564%, respectively. The significant and positive coefficients of GDP in the short- and long-term analyses indicate a positive relationship between GDP and CO₂ emissions. This finding is related to the fact that economic growth drives an increase in consumer demand and the expansion of businesses, which leads to a large amount of energy use and a subsequent increase in carbon emissions. These results are supported by Chen et al. (2019), Wang et al. (2019), Adams et al. (2020), Abbasi and Adedoyin (2021), and Syed et al. (2022).

In order to validate the appropriate model, several statistics and diagnostics are utilized at the bottom of Table 5. The outcomes reveal that the null hypothesis of serial correlation LM tests and heteroscedasticity tests are not refuted in each scenario. Therefore, serial correlation and heteroskedasticity do not exist in the estimation model. Moreover, the null hypothesis is also not rejected by the Ramsey RESET tests and normal tests, implying that the model is correctly measured and the residuals are normally distributed. In addition, this study also uses the CUSUM to measure the stability of the coefficients in both the long- and short-term specifications, and the results indicate that the short- and long-term parameters are stable.

Dynamic ARDL simulation forecasts

While keeping the values of other variables constant, the dynamic ARDL simulation forecasts illustrate predictions of the actual regressor shift and its effect on the explained variable. In particular, a 10% positive or negative shock in the main explanatory variable (CCPU) is predicted in the current study, which could be used to measure its impact on CO₂ emissions in different areas of China. Specifically, the impulse response results are shown in Fig. 2a–f. The yellow broken line indicates predicted mean values, and the light blue area to dark blue area denotes 70%, 90%, and 95% confidence intervals.

Fig. 2: The impulse response for a 10% positive or negative shock in the CCPU and its effect on carbon emissions in six regions of China.

a The effect of a 10% CCPU shock on carbon emissions in North China. b The effect of a 10% CCPU shock on carbon emissions in Northeast China. c The effect of a 10% CCPU shock on carbon emissions in Eastern China. d The effect of a 10% CCPU shock on carbon emissions in Southcentral China. e The effect of a 10% CCPU shock on carbon emissions in Southwest China. f The effect of a 10% CCPU shock on carbon emissions in Northwest China. The yellow broken line indicates predicted mean values and the light blue area to dark blue area denotes 70%, 90%, and 95% confidence intervals.

Figure 2a represents the influence of a 10% positive and negative shock in the CCPU on CO₂ emissions in North China. The impulse response graph reveals that a 10% increase in CCPU reduces carbon emissions, which improves environmental deterioration. In comparison, a 10% decline in CCPU increases CO₂ emissions; that is, it has a positive effect on environmental deterioration. Figure 2b shows the impulse response of the influence of CCPU on carbon emissions in Northeast China. The graph reveals that a 10% increase in CCPU has a significantly negative impact on environmental degradation. However, a 10% reduction in CCPU has a beneficial effect on environmental degradation. As discussed above, CCPU plays a key role in whether there is a detrimental or favorable impact on carbon emissions.

In the same vein, Fig. 2c shows the impulse response graph in Eastern China, which indicates the impact of CCPU on carbon emissions. A 10% positive change in CCPU hurts CO₂ emissions, which improves environmental deterioration. In contrast, a 10% negative change in CCPU has a positive influence on environmental aggravation. In addition, Fig. 2d depicts the forecast for examining the influence of CCPU on CO₂ emissions in Southcentral China. A 10% increase in CCPU has a beneficial effect on environmental improvement. However, a 10% negative change in CCPU has a positive influence on CO₂ emissions, suggesting that CCPU could play a key role in the environmental deterioration in Southcentral China.

Different from the above findings, the impulse response plot in Fig. 2e illustrates that a 10% positive change in CCPU positively affects carbon emissions in Southwest China. In comparison, a 10% negative change in CCPU harms CO₂ emissions. This outcome is consistent with the dynamic ARDL estimation result. Besides that, Fig. 2f shows the connection between CCPU and carbon emissions in Northwest China. The impulse response reflects that a 10% positive change in CCPU has a beneficial influence on environmental improvement, while a 10% decline in CCPU suggests that it has a positive influence on environmental deterioration.

Sensitivity analysis results

To further validate the robustness of the DARDL estimation results, the FDC model and the fully modified least squares (FMOLS) are employed as a sensitivity check. The corresponding outcomes are shown in Tables 6 and 7, which suggest that CCPU has the expected influence on CO₂ emissions in different areas of China. Therefore, the robustness test indicates that the empirical outcomes of the current study are valid and dependable. Our outcomes are supported by Gavriilidis (2021), suggesting that high levels of uncertainty about climate policy could hinder energy consumption and, thus lead to lower CO₂ emissions.

Table 6 Frequency-domain causality test results.

Table 7 Fully modified OLS simulation results.

Moreover, the statistical shreds of evidence of EC show that EC is the key determinant of carbon emissions in China. The outcomes are similar to those (Hussain et al., 2020; Kwakwa et al., 2020; Abbasi et al., 2021). Besides that, with the exception of Southwest China, GDP has the same influence on CO₂ emissions in the long- and short-term. This outcome is reinforced by Sunday et al. (2017) and Adedoyin and Zakari (2020), indicating that economic expansion has a major influence on CO₂ emissions and the environment.

Conclusion and policy implications

The purpose of the paper is to detect whether CCPU affects the achievement of the carbon neutrality target of China and to investigate the influence of CCPU on CO₂ emissions. To this end, we constructed a novel CCPU index that is based on the methods of Baker et al. (2016). Then, we applied the newly developed DARDL model to detect the potential relationships with Chinese data from 2005 to 2021. Considering the heterogeneity of China’s CO₂ emissions across regions, this study investigates the influence of CCPU on carbon emissions from a regional perspective.

After controlling for the effect of economic growth and energy use on carbon emissions, our outcomes suggest that a rise in CCPU reduces carbon emissions in North, Northeast, Southcentral, and Northwest China, which improves environmental deterioration. In comparison, the positive change in CCPU increases CO₂ emissions in Southwest China; that is, it has a positive effect on environmental aggravation. This finding implies that the information originating from the CCPU can provide useful information to predict CO₂ emissions in China. In addition, the impact of energy consumption on carbon emissions is also affirmed to be positive in each scenario. In this case, authorities should consider other alternative renewable energy uses to deal with the negative impact of these emissions. Moreover, economic development has shown a significant positive influence on carbon emissions. Therefore, while promoting economic growth, the negative impact on the environment cannot be ignored.

These results provide the following important policy recommendations. First, the CCPU effectively affects the movement of carbon emissions in different regions of China. As a result, taking this into consideration, policymakers should pay close attention to the changes experienced in climate policies, to avoid excessive CO₂ emissions. Moreover, since regional carbon emissions show heterogeneous reactions to the CCPU, local governments should note the characteristics of the CCPU and treat the impact of the CCPU differently to control regional carbon emissions more effectively. Second, since non-renewable energy use increases CO₂ emissions, the government should spend more on renewable energy projects and encourage innovation in environmentally friendly energy use to increase green energy production. In addition, the government should support the development of renewable energy technologies and provide appropriate subsidy policies, which will aid in the reduction of carbon emissions. Third, the outcomes of this study show that economic expansion has a positive influence on carbon emissions. Therefore, the government should balance economic development with environmental governance. Economic development should not come at the expense of the environment. In addition, further improving the quality of economic development is critical to achieving carbon neutrality.

Although we have explored the impact of climate policy uncertainty on CO₂ emissions from a dynamic perspective, there are still some limitations. Firstly, although we have constructed a novel CCPU index in the current study, it is not a large enough sample size. Therefore, additional research should be conducted with richer data that can predict the influences of CCPU more accurately. Secondly, future research should explore the potential relationships between CCPU and CO₂ emissions across countries, which could shed light on the impact of CCPU on CO₂ from a worldwide perspective.