The Urban Renewable Energy Transition: Impact Assessment and Transmission Mechanisms of Climate Policy Uncertainty

Abstract

The transition to renewable energy is a critical pathway for achieving low-carbon development and addressing global climate change problems. Therefore, we expand the conventional province-level energy balance table to the urban level, providing a refined assessment tool for evaluating renewable energy transition (RET). This study investigates the impact of climate policy uncertainty (CPU) on urban RET and explores the underlying mechanisms. The findings reveal that CPU significantly inhibits urban RET, with this effect being particularly pronounced in non-capital and inland cities. The mechanisms through which CPU hinders urban RET include exacerbating capital and labor misallocation and suppressing industrial structure upgrading. Furthermore, the moderation model indicates that high-intensity government supervision and low public environmental awareness exacerbate the negative impact of CPU on urban RET. Our findings provide governments with adopting forward-looking climate policies to mitigate the adverse effects of urban renewable energy transition.

1. Introduction

Recently, a series of climate change issues, including global warming and extreme weather events, have had profound impacts on human society [1,2]. To address these challenges, governments worldwide have implemented various climate policies aimed at transitioning to a renewable energy structure centered on carbon neutrality, thereby addressing the negative economic effects of global warming and climate change [3]. As cities are critical units for energy consumption and transformation, exploring how to promote urban renewable energy transition (RET) is an essential measure for realizing the energy revolution and fostering harmonious coexistence between society and nature.

Existing literature has examined the factors affecting the RET from financial development [4], technological innovation [5], policy incentives [6], public environmental awareness [7], and professionally skilled workforce [8]. However, scholars have neglected the objective climate policy uncertainty (CPU), and little literature has examined its effect on the renewable energy transition. In fact, due to the multidimensional nature of climate change and the balance of economic development and emission reduction targets, climate policies formulated by governments are often accompanied by uncertainty [9], generating investment uncertainty that exacerbates financial market volatility [10], and inflow capital into the renewable energy sector. Therefore, exploring the impact of CPU on the market aspects of renewable energy is important for the urban RET.

In addition, quantifying the impact of climate policy uncertainty on the transition to renewable energy can help build a more sustainable energy system. First, in an unstable climate policy environment, market participants face a lot of uncertainty when making renewable energy investment decisions, which exacerbates the volatility of the renewable energy market and hinders the flow of capital to renewable energy development [11]. Second, an increase in the CPU as a proxy variable for measuring climate risk implies that the cost of clean energy investments could subsequently increase, and CPU may lead to a less efficient allocation of capital and labor, triggering a waste of urban resources and an overall decline in economic efficiency, thus affecting the urban renewable energy transformation [12]. Climate policy uncertainty may not only have a direct impact on the urban RET but also exacerbate capital and labor mismatches and inhibit industrial upgrading.

Therefore, this study measures the urban RET based on Chinese cities from 2005 to 2020. Then, we use a two-way fixed panel model to deeply explore the impact of climate policy uncertainty on urban renewable energy transition and its mechanism. The results show that the presence of CPU significantly inhibits urban RET, and the inhibiting effect is more significant in non-capital and inland cities. Further, our evidence shows that intensifying capital and labor misallocation and industrial structure upgrading are the important mechanisms of CPU to inhibit urban renewable energy transition. From a government–public perspective, we confirm that the intensity of government supervision can exacerbate the negative impact of CPU, while public environmental awareness could mitigate the negative impact.

The marginal contributions are mainly reflected in the following aspects: First, we extend the traditional provincial energy balance sheet to divide electricity consumption into four major energy types, namely non-renewable electricity, hydro, wind, and solar, to construct a more detailed urban energy balance sheet. Due to data availability, the quantitative assessment of RET in existing studies mainly focuses on the national and provincial levels [13]. However, as an important unit of renewable energy consumption and transformation, the internal transformation dynamics and impacts of cities need to be thoroughly explored and accurately quantified. Second, we innovatively integrate CPU and RET into the same theoretical framework and systematically explore the impacts between them. Compared to previous studies, the impacts of urban renewable energy transformation are mainly explored from the perspectives of technological innovation [5], policy incentives [6], and public environmental awareness [7]. Third, we examine the impact mechanism of CPU on urban RET from aggravating capital mismatch, labor mismatch, and inhibiting industrial structure upgrading, and adopt a moderation model to find that high-intensity government regulation and low public environmental awareness exacerbate the negative impact of CPU on urban RET.

This paper is organized as follows: Section 2 encompasses the literature review; Section 3 presents a theoretical analysis and research hypothesis; Section 4 outlines the research strategy; Section 5 shows the empirical results; and Section 6 is the conclusions.

2. Literature Review

In recent years, the intensification of extreme weather events has become a phenomenon that cannot be ignored, and it has increasingly attracted academic attention [14]. In the relevant studies on the economic impacts of climate policy uncertainty (CPU), the existing literature focuses mainly on the macro and micro effects. At the macro level, CPU is thought to encourage carbon abatement [15], widespread use of clean energy [16], increase in market volatility [17], and delay in investment decisions [18]. At the micro level, the existing studies proposed that CPU can affect firms’ interregional investment [19], operational efficiency [20], and total factor productivity [21]. For example, Sun et al. [19] found that climate policy uncertainty may lead firms to rethink their geographic business layout to reduce potential risks. In summary, existing literature examines the impact of CPU on market investment and firm behavior, while ignoring its impact on the renewable energy transition (RET).

However, the current measurement of RET remains at the provincial level and has not yet been measured at the urban level due to data availability. The main metrics focus on the share of renewable energy in the energy mix, such as the share of renewable energy consumption in the primary energy supply [22] and the share of the final consumption of renewable energy in total energy consumption [23]. The existing literature is limited by the small amount of data, and the indicator measurement of urban RET is still blank, which makes it difficult to reflect the impact of climate policies on urban renewable energy transition.

Furthermore, existing research has explored the impact mechanisms on urban RET from the perspective of capital misallocation, labor misallocation, and industrial structure upgrading. First, capital misallocation affects urban renewable energy transitions through factors such as increasing financing costs [24], exacerbating regional development imbalances [25], and hindering technological innovation [26]. Second, labor misallocation affects the urban renewable energy transition by reducing employment opportunities [27] and intensifying industrial mismatch [28]. Finally, the upgrading of industrial structure affects the urban renewable energy transition by promoting changes in the energy demand structure [29], improving energy use efficiency [30], and facilitating regional coordinated development [31].

Existing studies have examined the factors influencing RET, focusing mainly on the positive and negative factors. In terms of positive factors, technological innovation [5], policy incentives [6], and public awareness of environmental protection [7] are considered to be the main driving forces for the development of renewable energy. For example, Barnea et al. [6] found that strong regulation can significantly promote renewable energy. However, renewable energy also faces some challenges in the transition process. High initial investment costs [32], erratic renewable energy output [33], and shortage of skilled labor [8] are all important factors that hinder the transition to renewable energy. While most existing studies have explored the impact of technological progress, external policy and social support, and renewable energy characteristics on renewable energy transitions, few studies have explored the impact of urban renewable energy transitions in the context of climate policy uncertainty.

In summary, although the CPU and its impact on renewable energy transition have begun to attract attention, there are still some research gaps. First, existing studies mostly focus on internal drivers such as technological innovation and investment costs on the RET, but relatively ignore the potential impact of the external factors. The CPU may have a series of cascading effects on investors and market expectations, thereby affecting the overall process of the RET. However, existing research has not fully uncovered this complex causal relationship and its specific manifestations in different scenarios. Second, the quantitative assessment of the RET in existing studies focuses mainly on the national and provincial levels, while the quantitative assessment of the RET at the city level remains blank. It limits in-depth research on the transition to renewable energy at the city level and makes it difficult to formulate targeted policies to promote the development of urban RET. Third, although existing studies have examined the impact mechanism of RET at the city level from aspects such as resource misallocation, few studies have considered climate policy uncertainty, which may have an indirect impact on the urban RET through the comprehensive impact of various factors. However, the existing studies have not yet deeply explored these potential impact mechanisms, and there is a lack of systematic analysis and empirical research on the urban RET.

3. Theoretical Analysis and Research Hypothesis

The scarcity of fossil energy predicts its price increase as resources decrease, a trend that makes it economically viable for renewable energy to replace fossil energy [34]. However, due to the multifaceted nature of climate change and the balance between economic development and emission reduction targets, climate policies formulated by governments are often fraught with uncertainty [9], which may hinder the urban RET. To better understand the logic of this study, its research framework is shown in Figure 1. From an investment uncertainty effect perspective, CPU complicates investors’ ability to precisely forecast future policy conditions and market returns, hence elevating investment risk. Renewable energy projects often have high investment costs, long payback cycles, and high technical risks, which may cause investors to delay or reduce renewable energy projects, thus hindering the transition and growth of renewable energy in cities [12]. From a cost-effect perspective, using the CPU index as a proxy variable to measure climate risk, its increase means that firms need to invest more resources to assess the impact of policy changes on projects, increasing the cost of decision-making. This can also cause financial institutions to be more cautious in providing credit support for renewable energy projects, increasing the difficulty and cost of project financing, which can weaken the competitiveness of renewable energy, hindering the transition to urban renewable energy [35]. Therefore, we propose Hypothesis 1.

Hypothesis 1:

Climate policy uncertainty can inhibit urban RET.

First, according to institutional theory, a good institutional environment can reduce transaction costs and improve market transparency, thereby promoting the efficient allocation of capital [36]. However, in the context of climate policy uncertainty, the lack of complete information available to market participants may exacerbate distortions in factor markets, leading to inefficient capital flows in the renewable energy sector. Compared to general technologies, renewable energy technologies have double externalities. Second, according to externality theory, the traditional energy industry is often associated with significant negative externalities in the production process, such as environmental pollution. Under the influence of climate policy uncertainty, market participants may reduce capital investment in the real economy and green innovation and increase investment in financial assets after assessing policy risks [37]. While physical investment and green innovation are the main drivers to promote industrial upgrading, the phenomenon of “de-real to virtual” investment could hinder industrial structure updating [38], thus hindering the urban RET. Third, from the perspective of labor market search theory, both workers and employers search for the best match, but climate policy uncertainty increases search costs and makes the search process more difficult and inefficient [39], exacerbating the labor mismatch problem. Labor shortages and skill mismatches inhibit production scale-up and technological upgrading in the renewable energy industry, which in turn hampers urban RET. Therefore, we propose the following.

Hypothesis 2:

Climate policy uncertainty can exacerbate capital and labor misallocation and hinder industrial structure upgrading to inhibit urban RET.

The theory of bounded rationality holds that enterprises are constrained by limited information and limited cognitive ability when making decisions. The higher the intensity of government supervision, the more compliance costs and policy risks enterprises will face, and it will be difficult to accurately assess the return on investment, thus inhibiting their willingness to invest in renewable energy [40]. On the one hand, when the intensity of government supervision increases, so does policy uncertainty. In response to this uncertainty, enterprises may choose to respond to short-term regulatory pressure through greenwashing behavior [41], thus exacerbating the inhibiting effect of CPU on urban RET. As an external force, public monitoring is crucial in the transition to renewable energy in urban areas [42]. From the perspective of stakeholder theory, the low environmental concern of the public means that society does not pay enough attention to environmental issues and the renewable energy transition. On the other hand, stakeholder theory emphasizes the synergy and tradeoffs among various stakeholders. The low level of public environmental concern has led to a weakening of synergies between investors, governments, and other stakeholders in terms of environmental responsibility and the transition to renewable energy [43]. Investors may reduce investment in renewable energy projects due to a lack of public support, and governments may reduce subsidies and support for related projects due to low public attention, thus exacerbating the inhibiting effect of climate policy uncertainty on urban renewable energy transition. Therefore, we propose that.

Hypothesis 3:

The stronger the government supervision and the lower the environmental public concern can exacerbate the inhibiting effect of CPU on urban RET.

4. Research Design

4.1. Data Source

This study selects Chinese urban data from 2005 to 2020 as the research sample, and data sources include the China Energy Statistical Yearbook, China Urban Statistical Yearbook, and CSMAR database, which ensure reliability and timeliness. Following Gao et al. [11], we meticulously evaluated the data and eliminated samples with significant missing information to guarantee the precision of the study. In addition, to mitigate the impact of outliers on the study’s results, we used a 1% truncation treatment across all variables. After this processing, we ended up with 3200 valid data covering 215 cities in

Table 1. Descriptive statistics.

VarName	Obs	Mean	Median	S.D	Min	Max
CPU	3200	1.243	1.183	0.579	0.000	3.397
RET	3200	6.172	4.323	6.076	0.011	35.963
Dev	3200	10.203	10.267	0.811	7.847	13.056
Fin	3200	2.058	1.790	0.945	0.588	7.519
Urb	3200	1.236	1.340	1.073	0.132	2.114
Road	3200	3.125	3.171	0.532	0.468	5.207
FDI	3200	0.003	0.002	0.003	0.000	0.061
Caiz	3200	0.507	0.476	0.218	0.026	1.541

Notes: This table presents summary statistics for the variables in the sample. The definitions of which are detailed in Section 4.3.

.

4.2. Model Construction

4.2.1. Baseline Model

To explore the potential inhibitory impact of CPU on the urban RET, we employ the panel data model that integrates two-way fixed effects following Gao et al. [44].

$${\mathrm{R}\mathrm{E}\mathrm{T}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}+\mathsf{\lambda }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

where ${\mathrm{R}\mathrm{E}\mathrm{T}}_{\mathrm{i}\mathrm{t}}$ is a dependent variable, which represents the renewable energy transformation of the i city in the t year. ${\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}$ is the explanatory variable, representing the CPU of the i city in year t. ${\mathrm{X}}_{\mathrm{i}\mathrm{t}}$ represents a set of variables designed to control for other factors that may affect urban RET. In the model, ${\mathsf{\beta }}_0$ represents the intercept term, ${\mathsf{\beta }}_1$ represents the estimated coefficient of the corresponding explanatory variable, ${\mathsf{\mu }}_{\mathrm{i}}$ is the individual fixed effect, ${\mathsf{\gamma }}_{\mathrm{t}}$ is the time-fixed effect, and ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ is the residual term.

4.2.2. Mechanism Analysis Model

To investigate how climate policy uncertainty affects the urban renewable energy transition, this study examines the mechanism effects from three dimensions: capital mismatch index (Cap), labor mismatch index (Lap), and industrial structure upgrading (Isa). Referring to Yi et al. [45], we establish the mechanism analysis model in Equation (2).

$${\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\alpha }_0+{\alpha }_1{\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}+{\alpha }_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

In the above formula, ${\mathrm{M}}_{\mathrm{i}\mathrm{t}}$ represents the mechanism variables of city i in year t, including capital mismatch index (Cap), labor mismatch index (Lap), and industrial structure upgrading (Isa), and other variables are the same as above.

4.2.3. Moderation Effect Model

To further explore the complex relationship between CPU and city-level RET, we construct a moderation model.

$${\mathrm{R}\mathrm{E}\mathrm{T}}_{\mathrm{i}\mathrm{t}}={\mathsf{\nu }}_0+{\mathsf{\nu }}_1{\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\nu }}_2{\mathrm{N}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\nu }}_3{\mathrm{N}}_{\mathrm{i}\mathrm{t}}\ast {\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\nu }}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\gamma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(3)

where ${\mathrm{N}}_{it}$ represents the moderating variables, namely government supervision (Gs) and public environmental concern (Pc). ${\mathrm{N}}_{\mathrm{i}\mathrm{t}}\ast {\mathrm{C}\mathrm{P}\mathrm{U}}_{\mathrm{i}\mathrm{t}}$ is the interaction between the CPU and the moderation variable. ${\nu }_3$ represents the coefficient of the interaction term, reflecting the moderation effect between CPU and urban RET.

4.3. Variable Description

4.3.1. Independent Variable

The climate policy uncertainty (CPU) is the independent variable. Referring to Ma et al. [10], we use the MacBERT model and deep learning algorithm to construct the urban CPU index in China from 2005 to 2020 based on the news text mining of major Chinese newspapers. The specific steps for CPU measurement are as follows.

Step 1: Data collection and cleaning. We select news articles published by six major Chinese newspapers from 2005 to 2020, namely People’s Daily, Guangming Daily, Economic Daily, Global Times, Science and Technology Daily, and China News Service, as the data source. The title and body of the article are merged into a single text document, and the redundant information that is least relevant to the study is eliminated. Step 2: Manual verification. We develop standardized audit methods and guidelines to ensure consistency and accuracy of audit results. At the same time, this study carries out several iterative training and pre-evaluation to ensure the accuracy of the audit results. Step 3: Building the training model. The MacBERT model is used to classify messages and construct the CPU index. The initial model parameters are pre-trained with a large number of Chinese texts to learn Chinese language rules and features. The MacBERT model is then trained with a manually reviewed dataset, the classification function is implemented by the linear layer and softmax activation function, and the standard cross-entropy loss function is used to optimize the training task. Step 4: Urban CPU index construction. We calculate the proportion of CPU news for each newspaper, that is, the number of CPU news stories compared to the total number of news stories. Following Baker et al. [46], we perform a time-series variance calculation on the proportion of CPU news for each newspaper and normalize it to the unit standard deviation. Specifically, we average the standardized values of all newspapers to obtain the CPU index.

4.3.2. Dependent Variable

Renewable energy transition is the dependent variable. Referring to Yang et al. [34] and Qin and Tan [47], we subdivide electricity consumption into four key energy types in the traditional provincial energy balance sheet, namely, non-renewable energy, hydropower, wind power, and solar power, to construct more detailed provincial data. Then, regarding the research of Shan et al. [48], this study assumes sectoral energy intensity uniformity in cities and provinces. On this basis, socio-economic factors P (P = Index_city/Index_province × 100%) are adopted. Where Index_city and Index_province represent key indicators at the city and provincial level, such as GDP, population size, etc., the seven energy types in the refined provincial EBT are reduced to the city level, that is, E_city = E_province × P. Finally, we use city-level hydropower, wind, and solar power as a share of total energy consumption to build the city RET.

4.3.3. Mechanism Variables

Capital mismatch index (Cap): Following Gao et al. [49], we construct a C-D production function with constant returns to scale. The specific model is as follows:

$$Y_{it}=A{K}_{it}^{{\beta }_{Ki}}{L}_{it}^{1-{\beta }_{Ki}}$$

(4)

where $Y_{it}$ is the actual GDP of province i at year t, $A$ is the technical level constant, ${\mathrm{K}}_{\mathrm{i}\mathrm{t}}$ is the fixed capital stock, and ${\mathrm{L}}_{\mathrm{i}\mathrm{t}}$ is the annual average employment of the province. We then take the logarithms of both sides of the model and add the regional dummy variable ${\mu }_i$, the time dummy variable ${\theta }_t$, and the error term ${\epsilon }_{it}$:

$$\ln \left({\displaystyle \frac{Y_{it}}{L_{it}}}\right)=lnA+{\beta }_{Ki}\ln \left({\displaystyle \frac{K_{it}}{L_{it}}}\right)+{\mu }_i+{\theta }_t+{\epsilon }_{it}$$

(5)

We calculate the capital distortion coefficient to measure the difference between actual capital use and optimal capital use, as follows:

$${\gamma }_{Ki}={\displaystyle \frac{\left(\frac{K_i}{K}\right)}{\left(\frac{S_i{\beta }_{Ki}}{{\beta }_K}\right)}}$$

(6)

where ${\gamma }_{Ki}$ is the capital distortion coefficient. Finally, the capital mismatch index is calculated as follows:

$${\mathrm{C}\mathrm{a}\mathrm{p}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{1}{{\gamma }_{Ki}}}-1$$

(7)

Labor mismatch index (Lab): The same method is used to measure the capital mismatch index, and the final formula is:

$${\mathrm{L}\mathrm{a}\mathrm{b}}_{\mathrm{i}\mathrm{t}}={\displaystyle \frac{1}{{\gamma }_{Li}}}-1$$

(8)

Industrial structure improvement (Isu): According to Gan et al. [50], we choose the ratio of the output value of the tertiary industry to the output value of the secondary industry as the index to measure the degree of industrial structure upgrading. This index can reflect the tendency of economic structure to develop towards services, and directly and effectively reflect the dynamic change trend of industrial structure. The higher the proportion of tertiary industry in the economic structure, the higher the service orientation of the economic structure.

4.3.4. Moderation Variables

Government supervision (Gs): Following Song et al. [51], we calculate the share of heavy industry in the GDP of prefecture-level in the province, and then cross-multiplied it with the frequency of words related to “environmental protection” in the provincial government report to construct the supervision intensity of prefecture-level cities. They are divided into two groups based on the median of the government supervision intensity.

Public environmental concern (Pc): Previous studies have mostly used questionnaire surveys to measure public environmental concern. However, this method has limitations such as insufficient representation of collected data, sample bias, and low accuracy. Based on Wang and Zhao [52], this study introduces the Baidu search index as an indicator to measure the public’s environmental concerns. Specifically, the annual Baidu search index with keywords such as “smog” and “environmental pollution” is standardized and divided into two groups of high and low public environmental concern based on the median.

4.3.5. Other Control Variables

City-level variables affecting the RET are incorporated into the model. According to Li et al. [53] and Gao et al. [21], we add the following variables: (1) economic development level (Dev): the logarithm of per capita GDP is used to measure; (2) financial development level (Fin): the ratio of deposits and loans of financial institutions to GDP is used to measure; (3) urbanization rate (Urb): the ratio of permanent urban population to total population is measured by logarithm; (4) infrastructure (Road): the value of per capita kilometers is measured; (5) foreign direct investment (FDI): quantified by the ratio of FDI to GDP; (6) fiscal decentralization (Caiz) is quantified by the ratio of public revenue to public spending.

5. Empirical Results

5.1. Benchmark Regression

In

Table 2. Baseline regression.

	(1)	(2)	(3)	(4)
	RET	RET	RET	RET
CPU	−0.151 *	−0.608 ***	−2.171 ***	−0.162 **
	(0.085)	(0.086)	(0.359)	(0.078)
Dev		3.125 ***	0.083	−0.364
		(0.185)	(0.830)	(0.512)
Fin		2.032 ***	0.970 *	0.282
		(0.223)	(0.519)	(0.296)
Urb		−0.561 ***	−0.698 **	−0.163
		(0.204)	(0.294)	(0.186)
Road		−0.923 ***	−2.045 **	−0.308
		(0.232)	(0.910)	(0.460)
FDI		−28.822	−133.599 **	−16.346
		(37.014)	(63.930)	(26.493)
Caiz		−1.229	−5.744 **	0.861
		(0.820)	(2.552)	(0.766)
Constant	6.359 ***	−26.236 ***	14.890 *	9.880
	(0.106)	(1.549)	(8.481)	(6.040)
City FE	√	√	×	√
Year FE	√	×	√	√
R2	0.437	0.519	0.247	0.538
N	3200	3200	3200	3200

Notes: The table reports the benchmark regression results in Equation (1), estimating the impact of CPU on urban RET. The √ is Yes, × is No, * p < 0.1, ** p < 0.05, and *** p < 0.01, and the standard errors are clustered at the city level in ().

, column (1) displays the regression result without adding control variables, columns (2) and (3) are the regression result with control variables but only controlling individual or time fixed effects, and column (4) is the regression result with adding control variables and controlling both individual and year fixed effects. The regression results of the four columns show that the regression coefficients of CPU on urban RET are all significantly negative, indicating that it does have a restraining effect on urban renewable energy transition. Specifically, for every unit increase in climate policy uncertainty, a city’s level of urban renewable energy transition is projected to decline by 0.162, a result similar to Alharbey and Ben-Salha [12]. The possible reason is that from the perspective of the investment uncertainty effect, climate policy uncertainty increases market risk expectations, leading investors to wait and reduce investment. This uncertainty makes it difficult for investors to accurately assess the long-term impact of policies, and thus, they are more inclined to invest in areas with lower risk [54]. In addition, from the perspective of cost effect, renewable energy technology usually has a high upfront investment cost and a long return cycle, and the rise in the CPU increases the cost of renewable energy, weakens its competitiveness with traditional energy sources, and hinders urban energy transformation.

5.2. Mechanism Analysis

In this paper, Equation (2) is used to explore the impact mechanism of resource mismatch and industrial structure upgrading. The regression results are shown in

Table 3. Mechanism test.

	(1)	(2)	(3)
	Cap	Isu	Lab
CPU	0.024 **	−0.038 ***	0.022 **
	(0.010)	(0.008)	(0.010)
Constant	0.993	3.406 ***	−3.407 ***
	(0.896)	(0.735)	(0.949)
Control	√	√	√
City FE	√	√	√
Year FE	√	√	√
R2	0.838	0.801	0.835
N	3200	3200	3200

Notes: This table shows the results derived from Equation (2), which explore the channels through CPU to inhibit urban RET. Other variables are the same as above. The √ is Yes, ** p < 0.05, and *** p < 0.01.

. The regression result presented in column (1) indicates a significant positive relationship between CPU and capital mismatch, with the coefficient achieving statistical significance at the 5% level. From the perspective of institutional theory, the reason for this result may be that CPU increases the opacity of market information, which makes financial institutions cautious about financing support for renewable energy projects, which hinders the flow of capital to the renewable energy sector, which leads to the misallocation of resources, and thus hinders the urban RET.

Column (2) is listed as the regression result of climate policy uncertainty on industrial structure upgrading. The coefficient of CPU for industrial structure upgrading is notably negative at the 1% level, suggesting it can impede industrial structure advancement. The possible reason for this is that CPU leads market agents to reduce investment in the real economy and green innovation and instead increase investment in financial assets. According to the externality theory, this “de-real to virtual” trend weakens the force of industrial upgrading, increases the dependence on the traditional energy industry, and hinders the transformation to a more efficient energy structure.

Column (3) is reported as the regression result of climate policy uncertainty on labor mismatch. The result shows that the coefficient of climate policy uncertainty on the labor mismatch index is significantly positive at the 5% level, indicating that climate policy uncertainty exacerbates the level of urban labor mismatch. The possible reason is that, from the perspective of labor market search theory, CPU increases the cost for workers and employers to find a suitable match, resulting in labor mismatch, which makes it difficult for the renewable energy industry to attract the necessary talent and limits the innovation and promotion of renewable energy-related technologies. Therefore, Hypothesis 2 is verified. The visual analysis of the mechanism analysis is shown in Figure 2.

5.3. Robust Test

5.3.1. Excluding Municipalities

Considering the unique economic, political, and geographical location of the enterprises in the municipalities directly under the central government, the municipalities may have a stronger ability to withstand the negative impact of CPU, which is quite different from other cities. Therefore, the four municipalities directly under the central government, “Beijing City”, “Shanghai City”, “Tianjin City” and “Chongqing City”, are excluded from the samples, and the regression is carried out on the excluded samples. The results of the regression are shown in

Table 4. Excluding specific samples.

	(1)	(2)	(3)	(4)
	RET	RET	RET	RET
CPU	−2.324 ***	−0.608 ***	−2.171 ***	−0.162 **
	(0.490)	(0.086)	(0.359)	(0.078)
Constant	9.415 ***	−26.236 ***	14.890 *	9.880
	(0.813)	(1.549)	(8.481)	(6.040)
Control	×	√	√	√
City FE	×	√	×	√
Year FE	√	×	√	√
R2	0.177	0.919	0.247	0.938
N	3376	3200	3200	3200

Notes: This table reports the results of excluding specific samples. We test the robustness of the benchmark results after excluding samples from direct-controlled municipalities. Other variables are the same as above. The √ is Yes, × is No, * p < 0.1, ** p < 0.05, and *** p < 0.01.

. The regression results show that even if the four municipalities are excluded from the regression samples, the coefficient of CPU on urban renewable energy transition is still significantly negative, which is consistent with the baseline regression results, proving the robustness of the regression results in this paper.

5.3.2. Exclusion of Specific Event

In light of the potential exogenous influence of the COVID-19 pandemic on the regression outcomes, the sample period is reduced to 2010–2019, and the regression is re-conducted. The regression results are presented in columns (1)–(4) of

Table 5. Exclude the specific event and replace the regression model.

	(1)	(2)	(3)	(4)	(5)
	RET	RET	RET	RET	Tobit
CPU	−2.459 ***	−0.577 ***	−2.312 ***	−0.165 **	−0.274 ***
	(0.495)	(0.092)	(0.367)	(0.078)	(0.0681)
Constant	9.806 ***	27.076 ***	13.201	10.739 *	24.86 ***
	(0.839)	(1.603)	(8.569)	(6.012)	(0.799)
Control	×	√	√	√	√
City FE	×	√	×	√	√
Year FE	√	×	√	√	√
R2	0.182	0.920	0.250	0.939	0.401
N	3150	2985	2985	2985	3440

Notes: This table shows the results compressed to the year 2019 in columns (1)–(4), and column (5) shows the regression results using the Tobit model. Other variables are the same as above. The √ is Yes, × is No, * p < 0.1, ** p < 0.05, and *** p < 0.01.

. The regression results show that even when the sample year is shortened to 2019 and all city-level control variables are included in the regression, the coefficient of climate policy uncertainty on urban renewable energy is still significantly negative at the 1% or 5% significance level, and the regression results are consistent with the baseline regression results of this paper.

5.3.3. Replacing Regression Model

To further test the robustness of the regression results, the static panel model is replaced by the Tobit model for the regression. The regression results are reported in column (5) of Table 5. The regression results show that even when the regression model is replaced, the regression coefficient of climate policy uncertainty on the urban renewable energy transition is still significantly negative at the 1% significance level. This result is consistent with the benchmark regression results in this paper.

5.4. Heterogeneity Analysis

This study aims to examine the variations in the impact of CPU on the transition to renewable energy across cities with distinct characteristics. We divide the selected samples into coastal and inland cities, and provincial and non-provincial cities according to geographical location and administrative region. In the face of climate policy uncertainties, different cities can rely on their abundant resources to reduce costs and enhance the competitiveness of renewable energy to cope with risks [55]. The grouping method is used to verify the heterogeneous effects.

Compared with non-provincial capital cities, provincial capital cities often have priority in obtaining government pilot project layouts and subsidy policy support in the process of renewable energy transformation, thus fully exerting their exemplary and leading role [56]. There are differences in economic development levels between provincial capital cities and non-provincial capital cities. Columns (1) and (2) in

Table 6. Heterogeneity tests.

	(1)	(2)	(3)	(4)
	Provincial Cities	Non-Provincial Cities	Coastal Cities	Inland Cities
CPU	0.230	−0.181 **	−0.112	−0.226 *
	(0.316)	(0.081)	(0.113)	(0.120)
Constant	17.802	10.073	23.038 ***	−0.813
	(12.823)	(6.490)	(7.269)	(9.807)
Control	√	√	√	√
City FE	√	√	√	√
Year FE	√	√	√	√
R2	0.956	0.936	0.901	0.953
N	307	2893	1595	1605

Notes: This table is the heterogeneity analysis of urban administrative division and geographical location. Other variables are the same as above. The √ is Yes, * p < 0.1, ** p < 0.05, and *** p < 0.01.

show the results for provincial capitals and non-provincial capitals, respectively. The research results show that CPU only has a significant inhibiting effect on the renewable energy transition of non-provincial capitals, while the coefficient of provincial capitals is not statistically significant. The reason for this result may be that, according to the external funding sensitivity theory, provincial capital cities usually have larger economic scales and stronger innovation capabilities, which can attract a large amount of capital, technology, and talent to enter the renewable energy field [57]. Therefore, they are less affected by CPU, while non-provincial capital may face investment capacity constraints due to lower levels of economic development [58].

Table 6 presents the regression results for coastal cities in column (3) and inland cities in column (4). The regression results reveal that CPU only negatively impacts inland city renewable energy transitions, while the coefficient in coastal cities is not statistically significant. The reason for this result may be that, from the perspective of regional economic resilience theory, coastal cities, compared to inland areas, have a higher industrial economic maturity, better risk management and climate change adaptation capabilities, and can deal more effectively with the challenges posed by climate policy uncertainty [59]. Therefore, coastal cities are less affected by climate policy uncertainty. The visual diagram of heterogeneity analysis is shown in Figure 3.

5.5. Further Analysis

To further investigate the boundary conditions of climate policy uncertainty on the urban renewable energy transition, this study further examines the moderation effects from the dual perspective of public and government supervision.

The regression results of government supervision are shown in columns (1) and (2) of

Table 7. Moderation effects.

	Government Supervision		Public Environmental Attention
	High	Low	High	Low
CPU	−0.241 ***	−0.046	0.071	−0.202 **
	(0.093)	(0.124)	(0.071)	(0.084)
Constant	11.515 *	−0.581	−3.321	4.106
	(6.295)	(7.212)	(6.297)	(6.156)
Control	√	√	√	√
City FE	√	√	√	√
Year FE	√	√	√	√
R2	0.944	0.947	0.992	0.943
N	1923	1268	638	2560

Notes: The table is the result of introducing the cross-product terms of government supervision (Gs), public environmental attention (Pc), and CPU in Equation (3). Other variables are the same as above. The √ is Yes, * p < 0.1, ** p < 0.05, and *** p < 0.01.

. It shows that in areas with high government supervision intensity, the coefficient of CPU, is significantly negative at the 1% level, while in areas with low government supervision, intensity can be ignored. The reason is that, from the perspective of limited rationality theory, under the mandatory supervision of the government, excessive pollution emissions can be subject to environmental punishment. To avoid environmental punishment, enterprises make strategic responses and adopt strategic green activities such as green bleaching rather than substantive green transformation activities. Therefore, the negative effects of CPU are more pronounced in regions with high levels of government control.

In Table 7, columns (3) and (4) indicate public environmental concern regression findings. It shows that with low public environmental concern, the coefficient of CPU is significantly negative at 5%, while the coefficient of CPU in areas with high public environmental concern is not statistically significant. The reason is that, according to the stakeholder theory, in areas with low public environmental concern, residents have a lower awareness of environmental issues, which means that they rarely consider environmental factors. Investors may reduce investment in renewable energy projects due to a lack of public support, and governments may reduce subsidies and support for related projects due to low public attention, making them more vulnerable to the negative effects of climate policy uncertainty. Therefore, the regression results in Table 7 verify the validity of Hypothesis 3.

6. Conclusions

This study quantitatively analyzes the influence of CPU on urban RET over the past decade, employing a two-way fixed panel model to explore the impact between CPU and RET. The results show the following: First, climate policy uncertainty significantly inhibits urban renewable energy transition. Second, capital mismatch, labor mismatch, and industrial structure upgrading are the key ways that climate policy uncertainty affects urban renewable energy transition. Third, high-intensity government supervision and low public environmental concern can exacerbate the inhibiting effect between them. Fourth, the heterogeneity analysis shows that the inhibiting effect of climate policy uncertainty on urban renewable energy transition is particularly obvious in non-provincial and inland cities.

Although this study has quantified the impact of climate policy uncertainty on the renewable energy transition in cities, there are still some limitations. Firstly, this study has not deeply analyzed its potential spatial spillover effects. Future research can delve into the spatial spillover effects generated by climate policy uncertainty, which could help to more comprehensively understand the transmission mechanism of policy uncertainty among different regions. Additionally, this study focuses on theoretical analysis but is lacking in practical application. Future research can apply the research results to different countries and cities, propose specific case analyses, and provide scientific suggestions for cities around the world to formulate targeted renewable energy transition policies, thereby promoting the process of global green development.

The following policy implications are put forward. Firstly, governments should enhance the continuity and stability of policy-making. To advance the urban renewable energy transition, the government must prioritize consistency and steadiness of policies when devising climate-related regulations. During the policy-making process, a thorough analysis of potential future changes in climate conditions and their possible impacts on the urban energy framework is essential. This is crucial to prevent negative consequences such as reduced capital utilization efficiency, increased labor allocation costs, and hindered industrial structure upgrading, which may arise due to policy instability. By strengthening the continuity and stability of policies, investor confidence can be bolstered, encouraging them to invest in the renewable energy sector.

Secondly, it is essential to develop differentiated policies tailored to cities at distinct stages of development. By formulating such policies, non-provincial capitals and inland cities can be guided and encouraged to boost their investment in renewable energy infrastructure. Providing these cities with increased policy support and financial subsidies will help offset potential economic losses incurred during the transition to renewable energy. This support will also stimulate innovation and expand the scale of renewable energy utilization. Policymakers should promote partnerships between provincial and coastal cities and non-provincial and inland regions within the renewable energy sector. By leveraging complementary resources and advantages, these regions can derive substantial economic benefits from the construction of renewable energy projects. This approach will not only contribute to economic balance across regions but also effectively stimulate the enthusiasm and innovation potential of non-provincial cities and inland areas in their transition to renewable energy.

Thirdly, it is imperative to enhance the public supervision mechanism. To enhance the transparency and efficiency of policy implementation, the public supervision mechanism must be reinforced. By elevating citizens’ right to information, enabling them to fully understand the objectives and anticipated outcomes of policies, the public will be better positioned to assess and oversee the rationality and effectiveness of these policies. Furthermore, the enhancement of citizens’ participation rights will not only serve to augment the democratic nature of policies but also ensure that they align more closely with the actual needs and expectations of the public. Additionally, the consolidation of public supervision and participation has the potential to create a driving force, prompting enterprises to direct greater attention toward the development and utilization of renewable energy. Consequently, this will encourage the development of renewable energy technology, optimize the energy structure, and enhance the sustainability of energy use.

Fourth, the government should optimize the regulatory mechanism and reduce the enterprises’ compliance costs. It should avoid excessive interference in the normal business operations of enterprises and reduce unnecessary administrative approvals and cumbersome compliance procedures. By simplifying regulatory procedures and lowering the compliance costs for enterprises, a more relaxed business environment can be created. Flexible regulation should be promoted, such as through policy guidance, technical guidance, and consulting services, to help enterprises better adapt to environmental policy requirements, rather than relying solely on punitive measures. A dynamic monitoring and evaluation mechanism should be established, with differentiated regulatory indicators based on the industry characteristics, scale, and environmental risks of enterprises. “One-size-fits-all” regulation should be avoided to ensure the precision and effectiveness of regulatory measures. Big data and artificial intelligence technologies should be introduced to enhance the intelligence level of regulation, enabling real-time monitoring of enterprises’ environmental behaviors and timely identification and correction of potential environmental issues.