The Effect of Carbon Trading Pilot Policy on Resource Allocation Efficiency: A Multiple Mediating Effect Model of Development, Innovation, and Investment

Abstract

This study extends the existing research on carbon trading policies from the perspective of mediating effects. Based on the difference-in-differences method, this study helps to understand the relationship between China’s carbon trading policies and resource allocation efficiency. The study finds that carbon trading policy promotes the optimization of capital allocation efficiency but does not promote the optimization of labor allocation efficiency. This conclusion has passed a series of robustness tests. Moreover, our analysis shows that carbon trading policies can influence resource allocation efficiency through per capita GDP, foreign direct investment, and innovation levels using multiple mediating models. Factors such as market size, the number of emission entities, and the behavior of market participants affect the resource allocation efficiency in the carbon trading process. Finally, the spatial spillover effect of the carbon trading policy is verified. This paper provides empirical evidence and policy implications for achieving the dual carbon goal and sustainable development.

1. Introduction

Emissions trading, one of the three adaptable emission reduction options outlined in the Kyoto Protocol, is where the idea of carbon trading originated [1]. The world’s first carbon trading market, the EU Emissions Trading System (EU ETS), was formally introduced in 2005 and has provided invaluable expertise for the creation and growth of the global carbon trading market [2]. A total of 194 parties from around the world agreed to work together to address climate change when they signed the Paris Agreement [3]. International acknowledgment of carbon trading as a critical tool for meeting emission reduction targets has gained ground over time. Currently, 25 emissions trading systems (ETS) covering 17% of the world’s greenhouse gas emissions are in place [4]. The EU ETS, North America (including California and the Regional Greenhouse Gas Initiative, or RGGI), the New Zealand ETS, and other markets are major players in the carbon trading space [5]. Since its inception in 2005, the EU ETS—the largest carbon trading market globally—has experienced a number of modifications and improvements that have progressively broadened its scope and increased market efficiency [6]. Korea ETS, the first national carbon trading market in East Asia, was founded in January 2015 and presently includes more than 600 firms and 73.5% of the nation’s carbon emissions. China started carbon emission market experiments in 2013 and has since gained great experience in seven provinces and cities, including Shanghai, Beijing, and Guangdong. With the formal inauguration of China’s national spot trading market for carbon emission rights on 16 July 2021, it became the largest carbon trading market globally [7].

Carbon emissions have become an international concern due to climate change and the growing focus on environmentally sustainable development [8,9]. Numerous nations and areas have enacted carbon trading pilot policies (CTPPs) in an effort to lower carbon emissions and reach the target of lowering global greenhouse gas [10,11,12]. Through the establishment of carbon emission licenses and the ability for emission units to purchase and sell their trading carbon quotas, CTPPs are typically considered as a market-driven approach [13,14] to encouraging energy conservation of businesses [15,16]. Most people agree that the carbon market may reduce carbon emissions while simultaneously increasing carbon productivity, optimizing structural effects, fostering technological innovation, and generating synergistic effects between pollution and carbon reduction.

Resource allocation efficiency is a key indicator of high-quality development [17]. As a key economic concept, it emphasizes the importance of the effective allocation of production factors to economic development. The total factor productivity of a country depends not only on technological progress but also on resource allocation efficiency [18]. In reality, due to institutional friction and policy distortion, the entry and exit of enterprises deviate from the efficiency principle, which can harm the total factor productivity [19,20,21]. Low-productivity enterprises may exist in the market for a long time, while high-efficiency enterprises withdraw from the market, resulting in a “resource mismatch”. What impact will the introduction of CTPPs have on the resource allocation efficiency?

On the one hand, as a tool for incentives, it may significantly aid in the transition of businesses towards environmentally friendly production and encourage them to increase their investments in emission-reducing technologies, as well as the effective use and distribution of resources. On the other hand, there is a chance that the CTPPs will fail in the market. When prices in carbon markets do not accurately reflect the true societal cost of carbon emissions, market failures can be expressed as price distortions that lead to the misallocation of resources. Therefore, these possible market failures must be carefully studied and addressed during the CTPP implementation process.

It is of significant practical and theoretical value to study the effect and mechanism of CTPPs on resource allocation efficiency. (1) As the severity of the global climate change issue grows, cutting greenhouse gas emissions has emerged as a shared objective of the worldwide community. CTPPs are a crucial market mechanism that lowers carbon emissions while also improving resource allocation and fostering low-carbon development. (2) It can be obtained that clear knowledge of the mechanism by which this policy promotes the transformation of the economy into a low-carbon one by examining the effects of the CTPPs on resource allocation efficiency. This research will also help formulate more sensible and successful low-carbon development policies by offering theoretical and practical guidance. (3) Furthermore, China has put out the “3060” two-carbon objective, and one of the key strategies for achieving this objective is the CTPPs. Examining its impact on resource allocation efficiency may offer stronger policy backing and ensure China achieves the “double-carbon” target.

2. Literature Review

Numerous academics have attested to the environmental benefits of CTPPs, particularly their ability to promote the decrease of carbon emissions. Zhang et al. pointed out that CTPPs reduce emissions by reducing economic output [22]. Zhao et al. attempted to explore a balanced strategy based on carbon emission distribution under the background of CTPPs [23]. Shi et al. used a different approach to explore the emission decrease effect of carbon quota and price and discovered the emission decrease effect was most effective under the historical emission method [24]. From the perspective of research methodologies, academics primarily utilize simulation analysis [25,26,27], the multiplier method [28,29,30], the synthetic control method [31], and other techniques.

The existing literature and analysis have established a correlation between CTPPs and carbon dioxide emission reduction [32,33,34]. However, the economic benefits of CTPPs have not been extensively studied. Some researchers utilize the geographical difference Durbin approach to explore and think that the impact of the carbon trading market is mainly achieved by improving energy efficiency [35], promoting low-carbon innovation [36] and adjusting industrial structure [37]. Foreign scholars use the dynamic CGE model [38], PSM-DID [39], and other methods to study the economic growth effect of EUETS. Dong et al. found that the Porter effect was not significant and that the pilot policy had little influence on GDP [40]. Zhang et al. investigated how CTPPs affected the green technology innovation of companies and found that, at this point, CTPPs inhibit innovation in green technology and call for tight control over the carbon quota’s size [41]. Furthermore, a few researchers still explore the supply chain’s production decisions in light of greenhouse gas reduction initiatives [42,43,44].

Resource allocation efficiency is a fundamental idea in the fields of economics and management that has long been the focus of research [45,46]. Achieving sustainable development has become more dependent on how resources can be allocated optimally and how to maximize resource utilization efficiency in light of the world economy’s rapid growth and growing resource scarcity [47]. In order to maximize economic and social advantages, resource allocation efficiency refers to the optimal use of resources through prudent resource allocation and usage under particular social and economic situations. Data envelopment analysis [21], input–output ratio, and Pareto efficiency [48,49] make up the majority of the measurement criteria. The body of research indicates that a wide range of factors, including industrial structure, technical innovation, market mechanisms, and environmental policies, influence how efficiently resources are allocated [50].

In conclusion, the majority of the currently published material concentrates on how CTPPs reduce carbon emissions, with less attention paid to how effectively they allocate resources. The innovations are shown in the following: (1) Method innovation: For the purpose of doing linear fitting analysis and significance testing on the influencing elements of resource allocation efficiency, the convolutional neural network (CNN) model is creatively created in this article. (2) Perspectives innovation: the impact of reducing carbon emissions is currently the primary focus of the literature on CTPPs. This study fills a vacuum by creatively testing how CTPPs affect the effectiveness of resource allocation. (3) Process innovation: The spatial effects of CTPPs are examined in this research. The two-way fixed effect spatial Durbin model is utilized for empirical analysis, and the total effects are broken down to determine the spatial spillover effects. (4) Viewpoint innovation: This study concludes that CTPPs affect resource allocation efficiency locally in various locations, and it makes policy recommendations based on these findings for the future growth of various regions.

3. Description of the Variables, Models, and Method

3.1. Variable Description

(1)

Explained variable

The explained variable is resource allocation efficiency. Generally speaking, market distortion is the main reason for resource mismatch. The market distortion that leads to inefficient resource allocation can be roughly divided into product market dislocation and factor market dislocation, and the latter is mainly considered in capital and labor factor markets. Drawing on the research of Chen Hongwei and Hu Weimin [51], the resource mismatch rate is measured by the capital mismatch index (CAE) and the labor mismatch index (LAE). The resource mismatch index measures whether the allocation of production factors is reasonable and whether it has reached the optimal state. The greater the resource mismatch index, the lower the resource allocation efficiency. The calculation process is as follows:

$${CAE}_i={\displaystyle \frac{\left(\frac{K_i}{K}\right)}{\left(\frac{s_i{\mathsf{\beta }}_{Ki}}{{\mathsf{\beta }}_K}\right)}}$$

(1)

$${LAE}_i={\displaystyle \frac{\left(\frac{L_i}{L}\right)}{\left(\frac{s_i{\mathsf{\beta }}_{Li}}{{\mathsf{\beta }}_L}\right)}}$$

(2)

K_i and L_i represent the capital stock and labor force of region i, K and L represent the total capital stock and labor force of the country, s_i is the ratio of each region to the national GDP, ${\mathsf{\beta }}_{Ki}$ is the input–output elasticity of capital factors, and ${\mathsf{\beta }}_{Li}$ is the input–output elasticity of labor factors.

(2)

Explanatory variable

The carbon trading interaction item DID is the main explanatory factor. The calculation process is shown in Formula (3), in which treat_i is the regional dummy variable, and the values of the seven pilot areas of CTPPs take the value of 1, and 0 for the other provinces; post_it, is the time dummy variable, has values of 1 for the pilot areas following the adoption of the policy, and 0 for the rest of the provinces. The experimental group is seven pilot provinces, while the other is control groups.

$${DID}_{it}={treat}_i\times {post}_{it}$$

(3)

(3)

Control variables

The following controlled factors may have an impact on how efficiently resources are allocated in different regions: foreign direct investment (fdi) [52], specifically the proportion of foreign direct investment to GDP; industrial structure (ind) [47], specifically the advanced level of industrial structure; the level of economic development (pgdp) [53], specifically the logarithm of per capita GDP; human capital (edu) [54], specifically the years of education per capita; urbanization level (urb) [55], specifically the share of the population that does not farm; population density (pd) [47], specifically the logarithm of the total population per square kilometer; area energy-saving and emission-reduction targets (t1, t2) [56], notably the relationship between each province’s goals for these two areas during the 12th and 13th Five-Year Plan periods and the following years; the level of innovation (il) [57], specifically the logarithm of domestic invention patent applications; the degree of industrial agglomeration (iad), specifically the proportion of employed individuals to the area covered by administrative divisions. A summary of control variables is shown in

Table 1. Control variables.

Variables	Abbreviations	Reasons for Selection
Foreign direct investment	fdi	FDI brings not only capital but also advanced technology and management experience, which can directly affect resource allocation efficiency.
Industrial structure	ind	The difference in industrial structure will directly affect the consumption and emission mode of resources.
The level of economic development	pgdp	The level of economic development is closely related to the level of resource consumption, emission and resource allocation efficiency.
Human capital	edu	The improvement of human capital level can improve production efficiency and innovation ability and then promote the efficient use of resources.
Urbanization level	urb	Areas with high levels of urbanization tend to have higher resource efficiency and more stringent environmental standards.
Population density	pd	The level of population density will directly affect the supply and demand relationship and the distribution efficiency of resources.
Area energy-saving and emission-reduction targets	t1, t2	The setting and implementation of regional energy conservation and emission reduction targets will directly affect the emission behavior and resource allocation strategy of enterprises.
The level of innovation	il	Innovation is one of the important means to promote the efficiency of resource allocation and carbon reduction.
The degree of industrial agglomeration	iad	The degree of industrial agglomeration will affect the centralized utilization of resources and emission efficiency. Regions with high industrial agglomeration tend to have more complete industrial chains and more efficient resource allocation mechanisms.

.

(4)

Other variables

The intensity of environmental regulations (rg) refers to the degree of constraints imposed on the production and operation activities of enterprises by various policies, regulations and standards implemented by the government to protect the environment. It reflects the government’s attention to environmental issues and the intensity of intervention in environmental externalities through administrative means. It is measured by the ratio between the completed investment in industrial pollution control and industrial added value.

Carbon trading amount (amount) refers to the total amount of transactions reached between buyers and sellers by trading carbon emission rights (such as carbon dioxide emission quotas) in the carbon trading market. It reflects the activity of the carbon market and the willingness of participants to reduce emissions, and it is one of the important indicators used to measure the effectiveness of the carbon market mechanism.

3.2. Model Construction

3.2.1. Base Model

CTPPs can be thought of as a quasi-natural experiment, and the connection between resource allocation efficiency and carbon emissions trading can be examined using the DID model. We may examine if there is a substantial difference in resource allocation efficiency between pilot and non-pilot provinces prior to and after the opening of the China CTPPs using the multi-period DID model, holding other affecting factors constant. The corresponding model is as follows:

$${CAE}_{it}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_{1}DID+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(4)

$${LAE}_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_{1}DID+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(5)

CAE_it stands for capital allocation efficiency, LAE_it stands for labor allocation efficiency, and subscripts i and t stand for provinces and time, respectively; DID, as described above, indicates whether CTPPs were implemented in province i in year t; X_it stands for the control variables.

3.2.2. Dynamic Effect Model

It will take some time for the policy to be published and implemented, and then the effect will appear. China is putting the CTPPs into practice for the first time, and there are some functional defects in the market system. Each pilot area needs a period of time to adapt to the policy. Thus, a dynamic effect model is constructed in the following way to investigate if carbon emissions trading has a dynamic effect:

$${RAE}_{it}={\mathsf{\theta }}_0+{\mathsf{\theta }}_1\times {treat}_i+\sum _{j=2013}^{2020}{\mathsf{\theta }}_j\times {\pi }_{jt}+\sum _{j=2013}^{2020}{v}_j\times {treat}_i\times {\pi }_{jt}+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(6)

Among them, RAE_it is resource allocation efficiency, which is divided into capital allocation efficiency (CAE) and labor allocation efficiency (LAE); ${\pi }_{jt}$ is a dummy variable of year, with a value of 1 when t = j and 0 when t ≠ j; $v_j$ is the cross coefficient between the experimental group and the dummy variable of the year, which illustrates how the resource allocation efficiency of these two groups differed following the implementation of the policy. The meanings of the other variables remain unchanged.

3.2.3. Regional Heterogeneity Model

In addition to the heterogeneity analysis of the pilot area, this paper also analyzes the differences in the effects of CTPPs on resource allocation efficiency under different environmental regulation intensities and carbon trading amounts. Accordingly, the following regional heterogeneity model is constructed:

$${RAE}_{it}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_{1}DID\times rg+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(7)

$${RAE}_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_{1}DID\times amount+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(8)

“DID × rg” represents the interaction term between environmental regulation and DID, and “DID × amount” represents the interaction term between carbon trading amount and DID.

3.2.4. Intermediate Effect Model

The mediating effect model can be used to investigate the mechanism of the effect. To test whether the degree of economic development, innovation, and foreign direct investment are the transmission channels through which CTPPs influence resource allocation efficiency. The level of economic development serves as an important index for assessing a region’s comprehensive strength, influencing its investment capacity and potential in resource allocation, technological advancements, and environmental conservation. Innovation is a key factor in promoting economic development and improving the efficiency of resource allocation. The implementation of CTPPs requires technological innovation to support the realization of emission reduction targets, such as developing low-carbon technologies and improving energy efficiency. Furthermore, FDI not only brings capital but also brings advanced technology and management experience. These technologies and experiences will help improve the efficiency of resource allocation and promote green economic development. In summary, these three variables reflect the economic and social background of the implementation of CTPPs, which is of great significance in studying the impact of carbon trading on resource allocation efficiency.

The intermediate effect model is constructed by introducing the following intermediary variables: innovation (il), foreign direct investment (fdi), and degree of economic development (pgdp):

$${RAE}_{it}={\mathsf{\alpha }}_0+aDID+{\mathsf{\alpha }}_1{treat}_i+{\mathsf{\alpha }}_2{post}_{it}+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(9)

$${CV}_{it}={\mathsf{\alpha }}_0+cDID+{\mathsf{\alpha }}_1{treat}_i+{\mathsf{\alpha }}_2{post}_{it}+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(10)

$${RAE}_{it}={\mathsf{\alpha }}_0+{\mathsf{\alpha }\prime }_{1}DID+b\times {CV}_{it}+{\mathsf{\alpha }}_1{treat}_i+{\mathsf{\alpha }}_2{post}_{it}+\mathsf{\gamma }{X}_{it}+{\mathsf{\theta }}_i+{\mathsf{\mu }}_t+{\mathsf{\epsilon }}_{it}$$

(11)

Among them, CV_it represents the intermediary variables, namely GDP per capita, innovation level and FDI. The overall impact of CTPPs on either labor or capital allocation efficiency is represented by coefficient a; the impact of the policy on intermediary variables is represented by coefficient c; and coefficients ${\alpha \prime }_1$ and b are the impact of CTPPs and intermediary variables on capital allocation efficiency or labor allocation efficiency, respectively.

3.2.5. Spatial Spillover Effect Model

The LM, LR, Wald, and Hausman tests are among the applicability tests for spatial econometric models. Selecting an appropriate spatial econometric model determines how effective these empirical results are. The sample is subjected to the aforementioned tests, with

Table 2. Test results.

Test	Variables	CAE	LAE
LM test	Spatial error:
	Moran’s I	3.3920 ***	7.9080 ***
	Lagrange multiplier	201.0160 ***	57.3930 ***
	Robust Lagrange multiplier	57.6910 ***	7.3410 ***
	Spatial lag:
	Lagrange multiplier	184.3850 ***	65.7380 ***
	Robust Lagrange multiplier	41.0600 ***	15.6860 ***
LR test	chi 2	16.3400 ***	4.1100 ***
	p-value	(0.0600)	(0.0900)
	chi 2	669.0700 ***	1044.7800 ***
	p-value	(0.0000)	(0.0000)
Wald test	chi 2	6.4900	25.6700
	p-value	0.0110	0.0000
Hausman test	chi 2	4.8900	6.2100
	p-value	(0.0000)	(0.0040)

Standard errors are in parentheses, *** indicates significance at the 1% level.

displaying the outcomes. Both the spatial lag and error model pass the LM test. Furthermore, the LR test demonstrates that the time and spatial fixed effects are statistically significant, suggesting that the two-way fixed effects are more appropriate. The Wald test indicates that the spatial Durbin model does not degenerate into the spatial lag model or the spatial error model. The fixed effects model is more suitable, according to the Hausman test results. In conclusion, a more suitable instrument for examining the spatial spillover effect of CTPPs is the spatial Durbin model with a two-way fixed effect. The two-way fixed effect spatial Durbin model is built, as shown by Formulas (12) and (13). This model’s regression can accurately identify the spatial spillover effects and assess the policy’s transmission effect across various locations.

The spatial econometric model is as follows:

$${RAE}_{it}={\partial }_0+{\rho }_{0}W{Y}_{it}+{\partial }_{1}DID+{\rho }_{1}WDID+{\partial }_2{X}_{it}+{\rho }_2{WX}_{it}+{\theta }_i+{\mu }_t+{\epsilon }_{it}$$

(12)

$${\epsilon }_{it}={\rho }_{3}W{\epsilon }_{it}+{\vartheta }_{it}$$

(13)

Among them, W is the spatial 0–1 weight matrix; ρ₀ is the spatial lag term of the dependent variable; ρ₁ is the spatial lag term of the independent variable; ρ₂ is the spatial lag term of the control variables; ρ₃ is the spatial lag term of the error term; the other variables are consistent with the preceding ones.

3.3. Data Source

Accounting for data accessibility, the panel data are gathered from 30 Chinese provinces and cities between 2011 and 2020. The China Statistical Yearbook and China City Statistical Yearbook are the main sources of the data.

Table 3. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
CAE	300	0.3397	0.3922	0.0093	2.8576
LAE	300	0.0027	0.3419	0.0818	1.0916
DID	300	0.1700	0.3763	0.000	1.0000
fdi	300	0.0194	0.0183	0.0001	0.1210
ind	300	2.3742	0.1291	2.1663	2.8357
pgdp	300	2.2242	0.4310	1.1401	3.3557
edu	300	9.1673	0.9888	4.2219	12.7820
urb	300	0.5901	0.1222	0.3503	0.8960
pd	300	474.4162	709.1202	7.8638	3949.5615
t1	300	62.9500	66.1936	0.0000	187.0000
t2	300	97.9733	101.2273	0.0000	272.0000
il	300	9.5701	1.4042	5.3181	12.2852
iad	300	0.0262	0.0385	0.0004	0.2171

displays the statistical descriptions of the variables.

3.4. Method

Firstly, the link between the resource allocation efficiency level and CTPPs is confirmed from China’s point of view using the multi-stage difference-in-difference (DID) model. Secondly, the effectiveness of resource allocation and its mechanism are assessed in relation to the CTPPs using the mediation model. Thirdly, the spatial spillover effect of China’s CTPPs is studied using the two-way fixed effects spatial Durbin model. This study framework is shown in Figure 1.

4. Empirical Analysis

4.1. Benchmark Regression

Benchmark Regression Result

The impact of the CTPPs on the resource allocation efficiency is examined using the multi-period DID approach, and all regression results account for the fixed effects of year, province, and interaction.

Table 4. Regression results of CTPPs on resource allocation efficiency.

Variables	(1) CAE	(2) CAE	(3) LAE	(4) LAE
DID	0.0495 (0.1119)	−0.1883 ** (0.0586)	−0.0156 (0.0419)	0.0358 ** (0.0129)
fdi		0.7721 (0.7874)		0.9858 (0.6066)
ind		0.2625 (0.4807)		−0.0180 (0.2484)
pgdp		0.4145 ** (0.1669)		0.2952 ** (0.0898)
edu		−0.0031 (0.0058)		0.0001 (0.0026)
urb		−5.1749 ** (2.1896)		1.0767 * (0.5770)
pd		0.0016 (0.0015)		−0.0012 ** (0.0004)
t1		0.0018 * (0.0010)		−0.0008 (0.0008)
t2		0.0008 (0.0005)		−0.0001 (0.0006)
il		−0.0624 * (0.0325)		0.0075 (0.0142)
iad		14.3884 ** (3.8768)		−1.7954 (1.6867)
_cons	0.2542 *** (0.0385)	0.9561 (1.5177)	0.0161 (0.0158)	−0.5090 (0.7908)
Control	No	Yes	No	Yes
Observations	300	300	300	300
R-squared	0.2231	0.6978	0.2434	0.6857

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

presents the benchmark regression results of models (4) and (5). Columns (1) and (3) list regression results without adding control variables, and (2) and (4) list the regression results with adding control variables. DID’s estimation coefficients before and after adding control variables are inconsistent, which may be because resource allocation efficiency is influenced by many factors and is sensitive to changes. It is easy to cause errors without considering control variables. Therefore, only the regression findings following the addition of control variables are taken into account.

While the implementation of the CTPPs can significantly lower the capital mismatch index and optimize capital allocation efficiency, it can also somewhat increase the labor mismatch index, which is not beneficial to improving labor allocation efficiency. The possible reasons may be as follows:

(1) Capital investments are more likely than workforce expansion to help businesses meet their carbon emission reduction goals in the CTPPs market, for example, by purchasing more environmentally friendly equipment or adopting more efficient production processes, which is conducive to capital allocation efficiency optimization. (2) CTPPs encourage the use of cleaner and low-carbon technologies, which rely more on capital-intensive equipment and systems and are not suitable for labor-intensive production modes. (3) CTPPs pay more attention to environmental protection and carbon emission reduction targets and less attention to labor market and employment issues. The labor market balance may be broken and it will take some time to adapt, which is not conducive to labor allocation efficiency optimization. (4) A range of industries may be impacted by the introduction of CTPPs, and some may be subject to greater pressure than others to adapt. This may result in some barriers to labor resource transfer, which could lead to the creation of the labor mismatch phenomenon. (5) A segment of the workforce may be unable to afford new environmental technologies and regulations, necessitating retraining or skill upgrades. When the skills of the labor force do not meet the demands of the market, labor mismatch arises.

4.2. Robustness Test

4.2.1. Parallel Trend Test

The parallel trend test must be conducted to ensure that the data sets have the same tendency before the accomplishment of the CTPPs, which is a prerequisite for using the DID model. The base period is set in 2013 to prevent multicollinearity. Figure 2 shows the findings. The dark dash line in the graph represents the 95% confidence interval. It can be seen that the trend of the experimental group and the control group is similar before the implementation of the CTPPs, but there is a significant difference after the implementation of the CTPPs, and there is a certain lag. Overall, this result demonstrates the robustness of the benchmark regression results.

4.2.2. Placebo Test

Other factors may also affect how efficiently resources are allocated in addition to CTPPs. We can further confirm whether there is a causal relationship between CTPPs and resource allocation efficiency by the placebo test, which is an important robustness test method for DID models. The vertical dashed lines are the true regression coefficients of the DID model and the 0 values of the X-axis. Figure 3 shows the results that are carried out to confirm the change in resource allocation efficiency is primarily due to the CTPPs and not other unobservable factors. This analysis confirms the robustness of the above benchmark regression results.

4.2.3. Adjusted Sample Test

Sichuan started CTPPs, but no quota trading was carried out, so it was not able to be a pilot area. However, the resource allocation efficiency may be impacted by this. In order to avoid this influence and ensure the robustness of the above baseline results, this research proposes that it is a pilot area and treats Sichuan as a control group sample.

Table 5. Regression results of adjusted sample test.

Variables	(1) CAE	(2) CAE	(3) LAE	(4) LAE
DID	0.0352 (0.0973)	−0.1799 *** (0.0537)	−0.0130 (0.0368)	0.0300 ** (0.0133)
fdi		0.6701 (0.7817)		0.9897 (0.6170)
ind		0.4148 (0.4694)		−0.0412 (0.2546)
pgdp		0.4368 ** (0.1663)		0.2942 *** (0.0894)
edu		−0.0023 (0.0056)		0.0000 (0.0026)
urb		−5.1278 ** (2.1713)		1.0469 * (0.5862)
pd		0.0015 (0.0014)		−0.0012 *** (0.0004)
t1		0.0020 * (0.0010)		−0.0008 (0.0007)
t2		0.0010 * (0.0005)		−0.0001 (0.0006)
il		−0.0595 * (0.0318)		0.0071 (0.0143)
iad		14.5133 *** (3.8956)		−1.8136 (1.7078)
_cons	0.2342 *** (0.0517)	0.5110 (1.4475)	0.0269 (0.0183)	−0.4449 (0.8155)
Control	No	Yes	No	Yes
Observations	300	300	300	300
R-squared	0.2210	0.6985	0.2429	0.6837

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

displays the regression findings. The results are consistent with the above results and confirm the robustness of the above benchmark regression results.

4.2.4. CNN Test

AI technology is innovatively introduced to test the robustness of empirical results. The local connection and shared weight characteristics of the CNN help it to efficiently extract key features from complex data sets. Compared with other neural network models, CNN has advantages in task characteristics, computational efficiency, and model complexity.

In view of the excellent performance of CNN in processing image and text data, but there is a mismatch problem when dealing with non-image, especially binary classification features such as the DID variable, this paper adopts an alternative strategy. Specifically, the conceptual framework of DID variables was extended to continuous variables and carbon trading volume was chosen as a proxy variable to indirectly reflect the implementation effect of CTPPs. Constructing a network model based on carbon trading volume can take advantage of the powerful learning ability of AI technology to explore the correlation between CTPPs and resource allocation efficiency.

To linearly fit the effect of CTPPs on resource allocation efficiency, seven data samples from the control group are used in a CNN. Figure 4 displays the findings, in which the blue dots represent the selected samples and the red lines represent the fitting results. It can be found that the labor mismatch index is positively impacted by CTPPs, while the capital mismatch index is negatively impacted. This is consistent with the results of previous empirical analysis and supports the robustness of policy effects.

It is noted that CNN was selected to attempt the application of AI techniques to empirical analysis and to see if the model could still learn some useful information or patterns in the simplest convolutional configuration (i.e., kernel_size = 1). Although the model uses the CNN framework, the convolutional layer does not perform feature extraction in the traditional sense but as an exploratory design choice.

4.2.5. Dynamic Effect Analysis

By comparing the differences in the effects at different time points, it is possible to reveal whether the effects change over time and thus test the robustness of the baseline regression results.

Table 6. Dynamic effect results.

Variables	(1) CAE	(2) LAE
2013	0.0314 (0.0710)	−0.0429 (0.0426)
2014	0.0471 (0.0913)	−0.0649 (0.0579)
2015	0.0864 (0.1164)	−0.0910 (0.0753)
2016	0.3327 * (0.1666)	−0.2301 ** (0.0966)
2017	0.3451 * (0.1841)	−0.2751 ** (0.1106)
2018	0.3780 * (0.2059)	−0.3015 ** (0.1258)
2019	0.3815 * (0.2229)	−0.3232 ** (0.1400)
2020	0.4212 * (0.2380)	−0.3607 ** (0.1521)
Control	Yes	Yes
_cons	0.9561 (1.5177)	−0.5090 (0.7908)
Observations	300	300
R-squared	0.6978	0.6857

Standard errors are in parentheses, * and ** indicate significance at the 10%, and 5% levels, respectively.

displays the results of the regression of model (6). Since 2016, the impact of the CTPPs on the effectiveness of resource allocation has become apparent and has gotten stronger over time. This shows that CTPPs have a lag effect on the capital and labor allocation efficiency, with the policy influence increasing progressively. Meanwhile, this analysis confirms the robustness of the above benchmark regression results.

4.3. Regional Heterogeneity

4.3.1. Heterogeneity Analysis of Pilot Areas

By introducing the interaction terms between carbon markets and DID, we can investigate whether the effect of CTPPs on the resource allocation efficiency is significantly different in different pilot regions. The results of regional heterogeneity are displayed in

Table 7. The regression results of regional heterogeneity analysis.

Variables	(1)	(2)
	CAE	LAE
DID	−0.2861 *** (0.0508)	0.0545 ** (0.0220)
Beijing	0.2831 (0.1830)	−0.0659 (0.0418)
Tianjin	0.2291 ** (0.0956)	0.0366 (0.0517)
Shanghai	1.0905 *** (0.2443)	0.2565 * (0.1451)
Guangdong	0.2060 ** (0.0944)	−0.0472 (0.0322)
Hubei	0.0538 (0.0442)	−0.0541 ** (0.0219)
Chongqing	0.1404 *** (0.0472)	0.0084 (0.0289)
_cons	1.6740 (1.5503)	−0.1085 (0.7679)
Control	Yes	Yes
Observations	300	300
R-squared	0.7479	0.7210

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

. Shanghai’s carbon market is huge, and there are many market participants, which makes carbon trading more active and effectively promotes the optimization of local capital allocation efficiency. In addition, Shanghai’s carbon market contributes optimally to the efficiency of labor allocation; however, due to its immaturity and small scale, Hubei’s and other regions’ carbon markets may not be able to adequately meet market demands, which will have an adverse impact on labor allocation efficiency. In Hubei and Chongqing, the carbon market has a negligible effect on the effectiveness of resource allocation. One explanation could be that although there are a lot of carbon emitters in these places, not all of them have been completely included in the carbon market.

In conclusion, the efficiency of resource allocation is affected differently by the CTPPs in different regions, and the growth and size of carbon markets in those regions have a direct bearing on the efficiency and efficacy of those markets. Key elements influencing the effectiveness of resource allocation include the development of markets, the rise of emitters, and the participation of active market participants. We can better accomplish the goal of a low-carbon economy by directing the robust and active growth of the carbon market, deepening the reform of the carbon market, enhancing the regulatory framework, and consistently raising the market participation of carbon emitters.

4.3.2. Environmental Regulation and Carbon Amount

The purpose of this part is to analyze the difference in this effect under different environmental regulation intensities and carbon trading amounts. The intensity of environmental regulation reflects the strength and determination of each region in promoting green transformation and implementing environmental protection policies. The amount of carbon trading directly reflects the activity of the carbon market. By analyzing the difference in effects under different environmental regulation intensities or carbon trading amounts, we can more comprehensively reveal the differentiated impacts of CTPPs in different regions. Models (7) and (8) were regression, and the results are shown in

Table 8. Heterogeneity regression results of environmental regression and carbon amount.

Variables	(1) CAE	(2) LAE	(3) CAE	(4) LAE
DID × rg	4.1223 * (2.2614)	−2.0244 ** (1.3702)
DID × amount			−0.0122 ** (0.0304)	−0.0372 * (0.0301)
fdi	1.3949 (0.9004)	0.8845 (0.6235)	2.0057 (2.3971)	−0.4835 (1.0403)
ind	0.1122 (0.5112)	0.0296 (0.2577)	−3.551 * (1.5242)	0.6644 (1.5897)
pgdp	0.2833 (0.1785)	0.3231 *** (0.0893)	−0.0093 (0.1427)	0.2929 *** (0.0727)
edu	−0.0036 (0.0064)	0.0006 (0.0026)	0.1128 (0.1093)	−0.0627 (0.0855)
urb	−4.2077 * (2.1594)	0.8741 (0.5884)	−20.9833 *** (1.0027)	1.1315 * (0.5752)
pd	0.0010 (0.0015)	−0.0011 ** (0.0004)	0.0024 (0.0023)	0.0007 (0.0005)
t1	0.0006 (0.0011)	−0.0005 (0.0008)	0.0309 (0.0543)	−0.0244 (0.0135)
t2	−0.0001 (0.0007)	0.0001 (0.0006)	−0.0085 (0.0277)	−0.0098 (0.0064)
il	−0.0677 * (0.0340)	0.0088 (0.0148)	−0.1912 (0.1754)	−0.0866 (0.0589)
iad	14.5268 *** (4.4066)	−1.8745 (1.7894)	0.4304 (5.9139)	−1.3645 (2.8943)
_cons	1.4181 (1.6902)	−0.6509 (0.8101)	18.1873 * (9.185)	1.3709 (5.4277)
Observations	300	300	50	50
R-squared	0.6542	0.6819	0.9671	0.9031

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

.

The CTPP’s optimization effect on capital allocation efficiency will be weakened by the tightening of environmental regulations, but its optimization effect on labor allocation efficiency will be strengthened. The most likely explanation is that more stringent environmental regulations force businesses to spend more money to reduce their carbon emissions, which could harm capital allocation efficiency. Furthermore, businesses may exhibit a stronger inclination to swap labor for capital in response to environmental laws, which would decrease capital allocation efficiency and increase labor allocation efficiency. Concurrently, this will encourage businesses to focus more on creating environmentally friendly sectors, which calls for higher labor force involvement and enhances the effectiveness of labor allocation. The policy’s optimizing effect on the efficiency of labor and capital allocation will be strengthened by the expansion of carbon amount. The carbon market will become more dynamic and liquid as the amount of carbon increases, facilitating the effective allocation of capital. Businesses can more easily purchase carbon offsets or emission reduction initiatives, which maximizes capital use. An increase in the amount of carbon trading will force businesses to increase their R&D expenditures for green technologies, resulting in the creation of more highly skilled and valuable jobs as well as improved labor allocation efficiency.

4.4. Intermediary Mechanisms

In exploring the complex mechanism of how the carbon trading system affects the efficiency of resource allocation, a dimension that cannot be ignored is the analysis of intermediary effects. This part will focus on the analysis of intermediary effects, aiming to reveal how carbon trading can indirectly affect resource allocation efficiency by influencing these intermediary variables. This variable is not one of the control variables when examining the mediating variable’s mediating effect in order to prevent over-control.

Table 9. Intermediary effect regression results.

Variables	(1) CAE	(2) CAE	(3) CAE	(4) LAE	(5) LAE	(6) LAE
DID	0.2582 *** (0.0671)	0.3902 *** (0.0615)	0.4077 *** (0.0557)	−0.0876 * (0.0482)	0.1823 *** (0.0533)	0.2190 *** (0.0453)
pgdp	0.1881 *** (0.0586)			0.5593 *** (0.0421)
il		−0.0069 (0.0165)			0.0613 *** (0.0143)
fdi			−4.0978 *** (1.1434)			8.0916 *** (0.9307)
_cons	−0.1225 (0.1259)	0.3399 ** (0.1552)	0.3498 *** (0.0308)	−1.2318 *** (0.0905)	−0.6199 *** (0.1344)	−0.1967 *** (0.0251)
Observations	300	300	300	300	300	300
R-squared	0.1619	0.1333	0.1687	0.4297	0.1441	0.2754

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

displays the model findings of model (9)–(11).

The efficiency of capital and labor allocation is significantly mediated by GDP per capita and foreign direct investment, while innovation level has significant intermediary effects on labor allocation efficiency but not significant mediating effects on capital allocation efficiency. This situation may be because innovation often requires the introduction of new technologies and methods, which will prompt employees to receive new training and education, which will improve employee productivity and efficiency and thus promote labor allocation efficiency. The level of innovation has less of an impact on capital allocation efficiency, which may be because capital investment is more restricted by capital and market factors and has less direct correlation with innovation. Furthermore, GDP per capita serves as a comprehensive indicator of a nation’s overall economic development and standard of living. A high GDP per capita encourages greater investment in infrastructure development, human capital, and innovation, all of which improve the effective use of labor and capital. Therefore, the intermediary role of GDP per capita may be caused by the positive influence of the overall economic development level on resource allocation. The double intermediary effect of FDI on capital allocation efficiency and labor allocation efficiency may result from its influence on economic structure. FDI usually brings new production technologies, management expertise and market opportunities and promotes a more effective allocation of capital and labor. In addition, FDI may also increase the level of competition among enterprises and promote the efficiency of the whole economic system by improving efficiency and innovation.

5. Spatial Spillover Effect

Although the DID model can effectively evaluate the direct effects of CTPPs, it ignores the possible spatial spillover effects of CTPPs. Spatial spillover effects refer to the indirect effects of policies or economic activities in one region on neighboring regions. In the field of carbon trading, such spillover effects may be expressed as indirect effects of emission reduction actions in pilot areas on carbon emission behaviors and resource allocation efficiency in surrounding areas.

5.1. Spatial Features of Resource Allocation Efficiency

Firstly, K-means cluster analysis can intuitively express the distribution of resource allocation efficiency across the country, and the identification of this distribution is the basis for the subsequent spatial spillover effect analysis. Then, the Moran I index is used to analyze the spatial correlation of resource allocation efficiency on global spatial features, and the verification of global spatial correlation provides a theoretical basis for the subsequent spatial spillover effect analysis. The Moran scatter plot can be used to visualize local spatial autocorrelation, that is, to reveal the aggregation or discrete mode of resource allocation efficiency in local space.

5.1.1. Spatial Distribution Features

The resource allocation efficiency in China varies significantly between provinces and demonstrates regional correlation due to the diverse geographical positions, characteristics, and levels of economic growth of China’s cities and provinces. There are distinctions and correlations in the resource allocation efficiency among regions, and the economic relations between neighboring provinces and cities are very close. Since resource allocation efficiency is somewhat influenced by economic development level, a regular spatial distribution of resource allocation efficiency may exist. Based on this, the K-means clustering approach is used to cluster the capital and labor allocation efficiency in various regions of China to obtain the geographical distribution of resource allocation efficiency in various geographic areas. Figure 5 presents the findings. The color represents the level of resource allocation efficiency. A darker color indicates a higher resource allocation efficiency.

The provinces with high resource allocation efficiency are mainly concentrated in the coastal area, showing dispersion from coastal areas to inland areas, while the labor allocation efficiency is more complicated. The following are the causes of this phenomenon: (1) The central and southeastern areas are home to a large number of manufacturing and traditional industries, which form a stable employment market, so the labor allocation efficiency is relatively high. (2) Coastal provinces usually attract more outstanding labor talents. At the same time, there are also problems of brain drain and high mobility, resulting in an uneven distribution of labor allocation efficiency. (3) Compared with the developed coastal areas, the labor costs in the central and northeast regions may be lower, making it easier for enterprises to attract more labor in these areas and improving the labor allocation efficiency in this area. The labor allocation efficiency in Anhui province is quite different from that in the surrounding areas, which may be because the overall skill level of labor in Anhui is likely shorter than that in the surrounding advanced areas, and backward infrastructure restricts the flow of labor between different regions, and the labor force is idle or over-concentrated.

K-means clustering is used to identify regions with higher and lower resource allocation efficiency, which may play different roles in the spatial spillover effects of CTPPs. In addition, it can be found that the geographical distribution presents certain rules, which confirms the necessity of studying the spatial spillover effect.

5.1.2. Global Spatial Features

This study constructs the geographic weight matrix comprising thirty Chinese provinces and cities to test the spatial correlation of sample data.

Table 10. Global correlation test of capital allocation efficiency.

Year	I	E(I)	Sd(I)	Z	p-Value
2011	0.144	−0.035	0.074	2.401	0.016
2012	0.138	−0.035	0.071	2.408	0.016
2013	0.136	−0.035	0.069	2.477	0.013
2014	0.131	−0.035	0.067	2.474	0.013
2015	0.124	−0.035	0.065	2.448	0.014
2016	0.119	−0.035	0.063	2.427	0.015
2017	0.113	−0.035	0.061	2.407	0.016
2018	0.107	−0.035	0.060	2.347	0.019
2019	0.100	−0.035	0.060	2.252	0.024
2020	0.102	−0.035	0.060	2.269	0.023

and

Table 11. Global correlation test of labor allocation efficiency.

Year	I	E(I)	Sd(I)	Z	p-Value
2011	0.311	−0.035	0.080	4.309	0.000
2012	0.294	−0.035	0.080	4.103	0.000
2013	0.278	−0.035	0.080	3.908	0.000
2014	0.253	−0.035	0.080	3.599	0.000
2015	0.236	−0.035	0.080	3.375	0.001
2016	0.218	−0.035	0.080	3.162	0.002
2017	0.210	−0.035	0.080	3.054	0.002
2018	0.217	−0.035	0.081	3.117	0.002
2019	0.204	−0.035	0.081	2.950	0.003
2020	0.224	−0.035	0.080	3.249	0.001

present the test findings. There is a significant spatial agglomeration effect between China’s capital and labor allocation efficiency, as evidenced by Moran’s I index from 2011 to 2020 above 0 and the z-test values above 1.96. Based on this, the study of resource allocation efficiency should fully consider the spatial factors. If neglected, there may be a deviation between the measurement results and the reality. In addition, the positive value of the Moran’s I index indicates that there is a spatial positive correlation between the resource allocation efficiency, that is, high-efficiency regions are often adjacent to high-efficiency regions. This global spatial correlation provides a possibility for the spatial spillover effects of CTPPs.

5.1.3. Local Spatial Features

The local spatial correlation of the provincial resource allocation efficiency can be further evaluated using the local Moran scatter graph. Figure 6 shows the local Moran index plots of capital and labor allocation efficiency from 2011 to 2020. The abscissa z represents the observed value of the sample, the ordinate Wz is a spatial lag term, and the slope is the Moran’s index. In quadrants 1 and 3, high–high values gather together, and low–low values gather together, indicating spatial aggregation. In quadrants 2 and 4, low–high values are gathered, and high–low values are gathered, indicating spatial dispersion. The fact that most provinces are concentrated in quadrants 1 and 3 is evident, and this further supports the existence of the partial geographic agglomeration impact of resource allocation efficiency. These agglomeration areas may exhibit stronger internal interaction and mutual influence in the spatial spillover effect of CTPPs. For example, an efficient agglomeration area may have a stronger spatial spillover effect on the surrounding area through technology diffusion and market integration.

5.2. Regression Result

The spatial spillover impact includes three aspects: direct, indirect, and total effects. CTPPs have two distinct effects: one is direct and affects the resource allocation efficiency in the pilot region. The other is indirect and affects resource allocation in the areas surrounding the pilot region. The pilot region and the areas around it indicate the common spatial spillover impact, which is the total effect.

Table 12. Regression results of spatial spillover effect.

Variables	(1) CAE	(2) LAE
DID	0.1056 *** (0.0346)	−0.0491 *** (0.0163)
W × DID	−0.0382 (0.1106)	−0.0255 (0.0517)
Direct	0.1078 *** (0.0359)	−0.0503 *** (0.0172)
Indirect	−0.0537 (0.0948)	−0.0579 (0.0693)
Total	0.0541 (0.0971)	−0.1083 (0.0750)
Control	Yes	Yes
Observations	300	300
R-squared	0.1641	0.1093

Standard errors are in parentheses, *** indicates significance at the 1% level.

displays the regression results of the model (12) and (13).

The DID estimate coefficients, which are the fundamental explanatory variables for capital and labor allocation efficiency, exhibit statistical significance at the 1% level, indicating the presence of the spatial spillover effect. When the spatial impact is added, the absolute value of the DID is somewhat less than that of the benchmark regression, suggesting that neglecting the spatial factor may lead to an overestimation of the optimization effect of the CTPPs on the effectiveness of resource allocation. In order to more precisely evaluate the effect of CTPPs on the overall regional resource allocation efficiency, the results highlight the need to take spatial spillovers into account when evaluating the effects of CTPPs. They also offer strong policy suggestions for pertinent decision-making.

The aforementioned study leads us to the following conclusions: the CTPPs have a notable regional spillover impact. In particular, the indirect effect demonstrates how the pilot policy has impacted the neighboring areas, which has also enhanced resource allocation efficiency, whereas the direct effect primarily demonstrates the improvement in resource allocation efficiency inside the pilot area. The common spatial spillover effect, which raises the overall efficiency of resource allocation, is evident when we take into account the effects of the experimental area and its neighboring areas taken together.

These findings demonstrate that the CTPPs have a significant impact on the pilot region and also help the neighboring areas. By increasing the effectiveness of resource distribution and encouraging creativity, CTPPs indirectly serve as a model and source of inspiration for the neighboring communities, thereby advancing the regional economy. Through the introduction of cutting-edge technology and managerial expertise, CTPPs have directly aided in the enhancement of productivity and resource allocation efficiency within the pilot zones. They have also assisted in the economic structure’s modernization and optimization. Overall, CTPPs have given resource allocation and economic growth both inside and outside the region new life, and the shared geographical spillover effect has sped up the improvement of overall efficiency. The study’s conclusions have practical ramifications for the ongoing advocacy of comparable policies and highlight the need for regional collaboration and experience exchange in fostering effective resource allocation and economic expansion.

5.3. Robustness Test

There could be a strong degree of correlation between independent variables when interactive elements are added, which could produce erroneous regression findings. Therefore, we eliminated the interaction between each province’s conserve energy targets and emission-reduction targets in the 12th and 13th Five-Year Plan and the corresponding years in order to prevent collinearity, and we then created a regression.

Table 13. Robustness test.

Variables	(1)	(2)
	CAE	LAE
DID	−0.1871 *** (0.0251)	0.0312 ** (0.0124)
W × DID	−0.0495 (0.0820)	−0.0682 * (0.0394)
Direct	−0.1877 *** (0.0269)	0.0342 *** (0.0131)
Indirect	0.0239 (0.0608)	−0.0614 * (0.0329)
Total	−0.1638 *** (0.0582)	−0.0272 (0.0325)
Control	Yes	Yes
Observations	300	300
R-squared	0.4126	0.0492

Standard errors are in parentheses, *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

displays the results, and it is in accordance with previous experimental results.

Although the interaction term was removed, the regression results showed that the conclusion was still basically the same as the previous experimental results. This implies that the spatial spillover effect is critical to the overall effect and that CTPPs have a considerable effect on resource allocation efficiency even when the interaction is disregarded.

6. Conclusions

This study collects data from 30 Chinese provinces and cities from 2011 to 2020 to examine how CTPPs influence resource allocation efficiency. In addition, the intermediary effect of this effect and the spatial spillover effect of CTPPs are also studied. We set seven pilot regions as experimental groups and measured resource allocation efficiency from two aspects: capital allocation efficiency and labor allocation efficiency. Using DID model for empirical analysis, it is found that CTPPs achieved remarkable results in optimizing capital allocation efficiency and provided strong support for the economic green transformation in pilot areas. However, the relatively limited effect of CTPPs on improving the labor allocation efficiency suggests that policymakers need to consider more comprehensive and targeted measures to address labor market complexity. This result passed a series of robustness tests, including the parallel trend test, the placebo test, the CNN test, the adjusted sample test, and the dynamic effect analysis. However, this effect is different in the pilot areas, and the effect of resource allocation efficiency in Shanghai is the most significant, which may be closely related to the economic foundation, policy implementation and market environment of this region. In addition, it is found that the tightening of environmental regulations will weaken the optimization effect of CTPPs on the capital allocation efficiency but enhance its optimization effect on the labor allocation efficiency, and the optimization effect on the labor and capital allocation efficiency will be strengthened with the expansion of carbon amount. Based on the multiple mediating models, it is found that the effect of CTPPs on resource allocation efficiency is mediated by GDP per capita, innovation level, and FDI, providing a new perspective for understanding the transmission mechanism of policy effects. This will not only enrich the theoretical framework of policy effect evaluation, but also provide empirical basis for future policy optimization. Finally, spatial spillover effects of CTPPs are discovered using the dual fixed effects analysis method of the spatial Durbin model, and this conclusion passes the robustness test. This finding highlights the importance of interregional policy coordination and cooperation, and offers the possibility of achieving greater economic and environmental win-win outcomes.

In view of the effect of CTPPs on optimizing capital allocation efficiency, the government should continue to increase its support and investment in such policies. However, this policy has failed to improve labor allocation efficiency, so the government should re-examine the policy and institutional environment of the labor market to find the root causes that hinder the effective allocation of labor. At the same time, the government should take measures to further raise the level of per capita GDP, innovation, and foreign direct investment to make use of the mediating role of these variables and promote the optimization of resource allocation efficiency. Moreover, the spatial spillover effect of CTPPs should be fully utilized to strengthen regional cooperation in carbon emissions trading, improve the flow of factors between regions, and establish a community of interests.

There are still some limitations in this study. It ignores how resources are allocated in a particular sector or industry, such as services or agriculture. In fact, different industries may see different effects. Moreover, the above regression is constructed at the provincial level and does not take into account the Shenzhen market data. For future research, the Shenzhen carbon trading market can be classified as the Guangdong carbon trading market.