Impact of the Marketization of Industrial Land Transfer on Regional Carbon Emission Intensity: Evidence from China

Abstract

With the implementation of deep-market-oriented reforms in China, an accurate interpretation of the effect and transmission mechanism of the marketization of industrial land transfer on carbon emission reduction can provide references for further elevating the role of land policy in China’s low-carbon economic transition. Based on the balanced panel data of China’s provincial level from 2009 to 2020, this paper uses carbon emission intensity to quantify emission reduction targets, and studies the effects of the marketization of industrial land transfer on regional carbon emission intensity, as well as its internal mechanism, by describing the typical characteristics. Moreover, this paper analyzes the moderating role of environmental governance and conducts area heterogeneity analysis. The main results show that (1) the marketization of industrial land transfer can significantly reduce regional carbon emission intensity, and the negative effect is stronger in the mid-west compared with the effect in the east. (2) Environmental governance can play a positive moderating role in the relationship between the marketization of industrial land transfer and the carbon emission intensity. (3) Industry selection is the internal mechanism by which the effect of the marketization of industrial land transfer on regional carbon emission intensity operates, and the mechanism is represented more prominently in the mid-west. The main conclusions provide inspiration for land policy regulation in relation to carbon emission reduction. China’s local governments should implement specific strategies to improve the market-based operation mechanism of land, attracting industries with high energy efficiency and low carbon emissions, and strengthening the intensity of environmental governance.

1. Introduction

At present, global warming and its impacts are amongst the most severe challenges facing humans. Since the industrial revolution, a large amount of the carbon dioxide (CO₂) emitted by human activities, such as burning fossil fuels, industrial processes and land use changes, has remained in the atmosphere, and this is the main cause of climate warming [1]. To mitigate this global warming, countries around the world must work together to limit their cumulative carbon emissions and ensure no more than 1.5 °C of warming [2]. China is now the world’s largest emitter of CO₂. According to the environmental database of the World Bank [3], China’s CO₂ emissions accounted for 31.18% of the world’s total in 2019. Against this background, China’s government has set the target of reaching peak CO₂ emissions by 2030, and achieving carbon neutrality by 2060, to ensure national sustainable development while promoting global climate cooperation, as laid out at the 75th session of the United Nations General Assembly on 22 September 2020. Since the target of “carbon peaking by 2030, carbon neutrality by 2060” was put forward, carbon emission reduction, as one of the most important aspects of China’s future development, has become an issue that governments at all levels need to focus on.

With the increase in the attention being paid by various countries to carbon emission reduction, studies on this issue have gradually become more common in the field of environmental economics. Existing studies on carbon emission reduction mainly approach the issue from four perspectives: firstly, they measure carbon emissions using different methods [4,5,6]; secondly, most studies focus on the relations between regional economic development and carbon emissions, and the conclusions vary depending on the subject [7,8,9,10,11]; thirdly, recent studies have analyzed the factors affecting regional carbon emissions, finding that population size [12,13], industrial structure [14,15], environmental regulation [16,17] and urbanization [18,19] all have a significant impact; fourthly, the distributional characteristics in geography and space of carbon emissions have been analyzed, with most studies finding that the regional carbon emissions show spatial heterogeneity [20,21,22].

Land transfer policies are the important means by which China’s government participates in macro-control [23], and they can be also used to promote the development of a low-carbon economy. Governmental land transfer refers to governments, as the land owner, transferring land use rights to the land user within a certain number of years, after which the land user pays a land use right transfer fee to the government. Land transfer has typically been the main form of land use and development in China, and can be used to make up for gaps in the local governments’ fiscal expenditure, which is referred to as “land finance” [24]. Pursuing “land finance” imparts economic benefits, but reduces environmental benefits [25]. In the context of severe carbon emissions and the raising of local responsibility for emission reduction targets, China’s local governments, as land suppliers and the parties responsible for carbon emission reduction, have begun intervening in carbon emissions with the help of multi-dimensional land transfer policies. In the context of the evolution of policy, land transfer marketization has been defined as the behavior of transferring the right to use state-owned construction land through market allocation via bidding, auctioning or listing [24,26], and this approach can help limit carbon emissions through the effective allocation of resources [27]. According to the carbon neutrality research database of the CSMAR [28], the main source of China’s CO₂ emissions is the burning of fossil fuels such as coal, oil and natural gas in the processes of industrial production. Obviously, industrial land transfer policies directly affect the scale and structure of regional industrial development, and thus affect regional carbon emissions. With the implementation of deep market-oriented reforms in China [29], the marketization of industrial land transfer can be further expanded. Thus, an accurate interpretation of the effects and transmission mechanisms of the marketization of industrial land transfer on carbon emission reduction can provide references for further elevating the role played by land policy in China’s low-carbon economic transition and ecological construction.

To sum up, based on the balanced panel data of China’s provincial level from 2009 to 2020, this paper uses carbon emission intensity to quantify emission reduction targets, and empirically analyzes the impact of the marketization of industrial land transfer on regional carbon emission intensity, as well as its internal mechanism, which provides empirical references for the land policy regulation of carbon emission reduction. This paper makes three contributions to the existing literature on the relationship between land transfer marketization and carbon emissions. First, the most recent studies use total carbon emissions to quantify emission reduction targets [30,31], while this paper uses carbon emission intensity. Second, considering that industrial production is the main source of carbon emissions in China, the research object of this paper is industrial land transfer. Third, this paper examines the internal mechanism of the impact of the marketization of industrial land transfer on carbon emission intensity. The structure of the rest of the paper is as follows: Section 2 contains a comprehensive review of the relevant literature and puts forward research hypotheses; Section 3 introduces the research data and design of this paper; Section 4 concerns empirical tests and a discussion of the results; Section 5 gives the main conclusions and implications.

2. Literature Review and Hypotheses

2.1. Concerning the Governments’ Land Transfer

At the economic and social levels, scholars have tried to explore the impact of land use on carbon emissions from the perspectives of land use intensification [32,33], land expansion [34,35], land layout [36,37], the structure and intensity of land use [38,39], and land urbanization [40,41]. Governmental land transfer is the main form of land use and development in China, and can be used to produce “land finance”. The existing literature on land transfer mainly approaches this issue from the point of view of land supply and land demand. In terms of land supply, many studies have considered the distribution and effects of the land supply structure. Using Chinese land transfer data from 2006 and 2014, Xiong and Tan [42] found that the greater the amount of residential/commercial land supplied, the smaller the amount of industrial land supplied, and the more clustered the urban expansion will be, but the differences in urban form were not very obvious. Using the panel data of 270 prefecture-level cities in China from 2003 to 2015, Zhang [43] argued that there was an obvious “internal imbalance” and “spatial mismatch” in China’s urban land supply. That is, the proportions of housing and new construction land supplied were obviously low within the city, and the per capita land supply in first-tier and second-tier cities was significantly lower than that in small and medium-sized cities. Using China’s land supply data from 2001 to 2016, Fan et al. [44] found that the land supply structure affects housing prices, whereby the lower the proportion of commercial and residential land that is supplied, the higher the housing prices. In addition to the above studies on land supply structure, some of the literature has also focused on factors affecting land supply and its strategies. Using Chinese provincial-level data from 1995 to 2010, Du and Peiser [45] found that China’s land pricing system has led to land hoarding, and stronger administrative intervention in the land market has led to more land holding. Using data for the period 1973–1997, Lai and Wang [46] showed that the increase carried out by the Hong Kong government in land supply would not necessarily solve the housing shortage. Using industrial land transfer data from 2007 to 2016, Zhou et al. [47] argued that industrial land supply was a powerful tool by which China can cope with the transitions of national strategy, and all areas except the north–east showed positive responses to the national strategic transitions.

In terms of land demand, literature has mainly considered the means of prediction in order to achieve a better land supply. For example, Monkkonen [48] developed a theoretical model of the demand for land regularization in urban areas based on models of demand for the registration of agricultural land, and used a unique combination of census and administrative data from informally developed neighborhoods in Tijuana to test the model empirically. The results were mostly consistent with the predictions. Using MATLAB R2016a software modeling tools to establish a GM (1, 1) model and an RBF neural network model, Li et al. [49] predicted the demand for urban–industrial land in the Beijing–Tianjin–Hebei Urban Agglomeration, finding that the RBF neural network model is the most effective model for predicting urban–industrial land.

2.2. Concerning the Relation between Land Transfer and Carbon Emissions

As China’s economy shifts from its stage of rapid growth to a stage of high-quality development, the effects of carbon emissions caused by land use have become the main focus of academic circles. As mentioned above, scholars have explored the impact of land use on carbon emissions from a multi-dimensional perspective. From the perspective of land transfer, some consider the effects of land transfer scale [50] and structure [51] on carbon emissions. The land transfer mode is also an important factor affecting carbon emissions in China [23]. In China, land transfer includes four modes: bid transfer, auction transfer, listing transfer and negotiated transfer. Local governments can attract investment through loose land transfer policies that enable them to introduce highly polluting enterprises and enhance their production capacity, resulting in high energy consumption and pollution emissions [52], which increases the economic benefits while damaging environmental benefits [25]. In order to limit the corruption of the local governments in relation to rent-seeking behaviors, and preclude the inefficient and extensive use of land resulting from the mismatch between land resources and industrial demand, China’s land transfer mode should be shifted from the allocation of administrative plans to market-oriented transfer based on supply and demand, competition and price mechanisms [53]. In the context of evolving policy, the existing studies have defined land transfer marketization as the behavior of transferring the right to use state-owned construction land through the market allocation of bidding, auctioning or listing [24,26]. Scholars have also explored the effects of carbon emissions caused by land transfer marketization. Because land transfer marketization can help encourage more efficient enterprises featuring advanced technology through the higher demands of information disclosure [54], most studies have stated that land transfer marketization has a significantly negative effect on regional carbon emission. For example, using a panel data econometric model, Xu et al. [30] found that China’s land marketization had a significant slowing effect on carbon emissions. Based on the GMM dynamic panel metering model, Yang et al. [31] concluded that there was a significantly negative relationship between the marketization level of land transfer and carbon emissions. Jiang et al. [55] attributed the effect of land transfer marketization on industrial structure to the threshold and extrusion effects, and further studied the influence of this on green total factor productivity. Based on the double-fixed effect model, Liu et al. [27] found that land transfer marketization had a significantly inhibiting effect on carbon emissions. In terms of industrial land transfer, based on the systemic GMM and threshold regression models, Chen et al. [56] found that the proportion of the acreage of industrial land transferred by the “negotiated” mode affected industrial energy carbon emissions positively, and the impact on industrial energy carbon emissions showed a significant singular threshold effect at the economic level.

The above studies provide references for the exploration of the relationship between the marketization of industrial land transfer and carbon emissions, but there is still room for improvement. On the one hand, most studies use total carbon emissions to quantify their emission reduction targets. Jotzo and Pezzey [57] stated that carbon emission intensity can be used to measure the relationship between the total carbon emissions and economic development level of a country or region. On the other hand, the main source of China’s CO₂ emissions is industrial production. Therefore, for China to establish a balance between economic development and environmental protection, it is of more practical significance to quantify emission reduction targets using carbon emission intensity, and to empirically analyze the impact of the marketization of industrial land transfer on carbon emission intensity. Moreover, although studies have generally stated that land transfer marketization has a negative impact on carbon emissions, its internal mechanism has not been closely discussed. Most of the existing studies have stated that a negotiated transfer means that the quality of investment introduced will be worse [58]. At the present stage, the manufacturing industry in most areas of China, and especially in the mid-west, is still around the middle or bottom of the global value chain, given the low technical content and added value of products, and the weak core competitiveness [54]. With the continuous standardization of the domestic land transfer market, the phenomenon of low-price industrial land transfer has been alleviated, and so the cause of environmental pollution is more likely to be industry selection rather than the quality of the individual projects introduced. Therefore, industry selection may be the internal mechanism by which the effect of the marketization of industrial land transfer on regional carbon emission intensity operates. That is, the marketization of industrial land transfer can reduce the proportion of industrial land that is transferred out of the total high-carbon emission industries, thereby reducing the regional carbon emission intensity.

Environmental governance refers to the process of implementing a series of measures, including planning, management, monitoring, evaluation and restoration, so as to protect and improve environmental quality. Because environmental governance is an important way for local governments to reduce carbon emissions, we believe that this form of governance not only has a direct impact on the carbon emission intensity, but also plays a positive moderating role in the negative effect of the marketization of industrial land transfer on the carbon emission intensity.

In view of the above, to better serve China’s goal of “carbon peaking and carbon neutrality”, this paper uses carbon emission intensity to quantify emission reduction targets, and empirically analyzes the impact of the marketization of industrial land transfer on regional carbon emission intensity and its internal mechanism. The following three hypotheses are proposed:

Hypothesis H1.

The marketization of industrial land transfer can significantly reduce regional carbon emission intensity;

Hypothesis H2.

Environmental governance can play a positive moderating role in the relationship between the marketization of industrial land transfer and carbon emission intensity. That is, as the intensity of environmental governance becomes stronger, the negative effect of the marketization of industrial land transfer on the carbon emission intensity also becomes stronger;

Hypothesis H3.

Industry selection is the internal mechanism of the effect of the marketization of industrial land transfer on regional carbon emission intensity. That is, the marketization of industrial land transfer can reduce the proportion of the industrial land that is transferred out of the total in high-carbon emission industries, thereby reducing the regional carbon emission intensity.

Based on the above, we have constructed a framework depicting the mechanism of the impact of the marketization of industrial land transfer on carbon emission intensity, as shown in Figure 1.

3. Materials and Methods

3.1. Model Specification

To verify Hypothesis H1, we have constructed a benchmark regression model based on two-way fixed effects, as shown below:

$$C{I}_{it}={\beta }_0+{\beta }_{1}Land\_Mar{k}_{it}+{\beta }_2^{\prime }{X}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(1)

where CI_it is the dependent variable, representing the carbon emission intensity of China’s province i in year t. Land_Mark_it is the independent variable we mainly focus on, representing the marketization of industrial land transfer. X_it is the vector of observable control variables affecting regional carbon emission intensity. μ_i is the provincial dummy variable used to control the provincial fixed effects. γ_t is the year dummy variable used to control the year fixed effects. β is the vector of parameters to be estimated in model (1). ε_it is a random error term. In model (1), if β₁ is greater than 0, Hypothesis H1 is confirmed.

To verify Hypothesis H2, we have constructed a test model based on model (1), as follows:

$$C{I}_{it}={\lambda }_0+{\lambda }_{1}Land\_Mar{k}_{it}+{\lambda }_{2}Land\_Mar{k}_{it}\times E{g}_{it}+{\lambda }_3^{\prime }{X}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(2)

where Eg_it is the moderating variable, representing the environmental governance of the Chinese province i in year t. λ is the vector of the parameters to be estimated in model (2). The meanings and definitions of other variables are consistent with those in model (1). In model (2), if λ₂ is less than 0, Hypothesis H2 is confirmed.

To verify Hypothesis H3, we have constructed a mechanism test model based on two-way fixed effects, as follows:

$$Ind\_se{l}_{it}={\alpha }_0+{\alpha }_{1}Land\_Mar{k}_{it}+{\alpha }_2^{\prime }{X}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(3)

where Ind_sel_it is the mechanism variable, representing the industry selection of the Chinese province i in year t. α is the vector of the parameters to be estimated in model (3). The meanings and definitions of other variables are consistent with those in model (1). In model (3), if α₁ is less than 0, Hypothesis H3 is confirmed.

3.2. Variable Selection

Dependent variable. Carbon emission intensity (CI_it) is the dependent variable in this paper. Jotzo and Pezzey [57] argued that the carbon emission intensity can be used to measure the relationship between the carbon emissions and economic development levels of a country or region. Therefore, for developing countries seeking a balanced path between economic development and environmental protection, it is more realistic to quantify emission reduction targets using carbon emission intensity, the formula of which is as follows:

$$C{I}_{it}=CO{2}_{it}/GD{P}_{it}$$

(4)

where CO2_it is the total CO₂ emissions of China’s province i in year t. GDP_it is the regional gross domestic product (GDP). The carbon emission intensity is the CO₂ emissions per unit GDP (ton/CNY 10 thousand). In this paper, nine major energy sources, namely, coal, coke, crude oil, gasoline, kerosene, diesel oil, fuel oil, liquefied gas and natural gas, are selected to measure the total CO₂ emissions, the calculating formula of which is as follows:

$$CO{2}_{it}={\displaystyle \sum _{j=1}^9\left(C_{ijt}\times k_j\right)}$$

(5)

where C_ijt is the consumption of the energy source j by Chinese province i in year t. k_j is the CO₂ emission coefficient of the energy source j.

Core independent variable. The marketization of industrial land transfer (Land_Mark_it) is an important factor affecting the regional CO₂ emission intensity that we are focusing on. Li and Wu [59] argued that in China’s primary market of land transactions, the marketization degree of negotiated transfers is the lowest, while that of auction transfers is the highest, and those of bid transfer and listing transfer are in the middle. Therefore, we have selected the proportion of industrial land transferred by ”bid, auction and listing” out of the total transferred industrial land in the land transaction market to measure the marketization of industrial land transfer in China, the formula for which is as follows:

$$Land\_Mar{k}_{it}=\left(Bi{d}_{it}+Au{c}_{it}+Li{s}_{it}\right)/\left(Bi{d}_{it}+Au{c}_{it}+Li{s}_{it}+Ne{g}_{it}\right)=\left(Bi{d}_{it}+Au{c}_{it}+Li{s}_{it}\right)/T_{it}$$

(6)

where T_it is the total transferred industrial land in Chinese province i in year t. Bid_it is the quantity of the industrial land transferred by bid. Auc_it is the quantity of industrial land transferred by auction. Lis_it is the quantity of industrial land transferred by listing. Neg_it is the quantity of industrial land transferred by negotiation. Thus, the greater the value of Land_Mark_it, the higher the marketization degree of industrial land transfer. Because land transfer marketization can help introduce more efficient enterprises with more advanced technology through a higher degree of information disclosure, with the increase in the degree of marketization of industrial land transfer, the regional carbon emission intensity will be expected to decrease.

Mechanism variable. Industry selection (Ind_sel_it) is the mechanism variable that we are focused on. According to previous analyses, industry selection is the internal mechanism of the effect of the marketization of industrial land transfer on regional carbon emission intensity. The lower the marketization degree of industrial land transfer, the more likely it is that there will be a transfer of land to high-carbon emission industries. According to the carbon neutrality research database of the CSMAR [28], the production and supply of electricity and heat, the smelting and pressing of ferrous metals, petroleum processing, coking and nuclear fuel processing, the manufacture of raw chemical materials and chemical products, and the mining and washing of coal are the top five carbon-emitting industries in China. The total CO₂ emissions of these industries account for more than 85% of the industrial total. Thus, we have constructed the following indicator to represent the selection of high-carbon emission industries:

$$Ind\_se{l}_{it}=\left(E{h}_{it}+F{m}_{it}+P{n}_{it}+C{m}_{it}+C{o}_{it}\right)/T_{it}$$

(7)

where T_it is the total quantity of all transferred industrial land in Chinese province i in year t, as in model (6). Eh_it is the quantity of the industrial land transferred for the production and supply of electricity and heat. Fm_it is the quantity of industrial land transferred for the smelting and pressing of ferrous metals. Pn_it is the quantity of industrial land transferred for petroleum processing, coking and nuclear fuel processing. Cm_it is the quantity of industrial land transferred for the manufacturing of raw chemical materials and chemical products. Co_it is the quantity of the industrial land transferred for the mining and washing of coal. Thus, the greater the value of Ind_sel_it, the higher the proportion of industrial land transferred to high-carbon emission industries.

Because environmental governance (Eg_it) is not only the moderating variable, but also the control variable introduced into the model due to its direct impact on the carbon intensity, we describe it in the section regarding control variables later.

Control variables. By referring to the existed studies on the factors influencing carbon emissions and considering the completeness of data selection, this paper introduces some control variables at the provincial level that may affect regional carbon emission intensity into the regression models, as follows: ① Economic development level (PGDP_it), represented by the natural logarithm of per capita GDP. The greater the value of PGDP_it, the higher the economic development level. With the improvement of the regional economic development level, people’s awareness of environmental protection will be enhanced, and the regional production technology level will also be improved. Thus, the level of regional economic development is expected to have a negative effect on the carbon emission intensity. ② Industrial structure (Ind_stru_it), represented by the proportion of regional secondary industry added value out of the GDP. The greater the value of Ind_stru_it, the more mature the industrial development. In the early stage of industrial development, local governments tend to pursue greater scales of industrial production and invest less in environmental protection, meaning the carbon emission intensity increases with the increase in industrial added value. After industrial development reaches a certain level, due to the scale and agglomeration effects, industrial production technology and efficiency in specific regions are improved with the increase in industrial added value, and the carbon emission intensity will also decrease. It is expected that the impact of Ind_stru_it on CI_it will manifest an inverted U-shaped curve. Thus, the square term of the proportion (Ind_stru2_it) is also introduced into the regression model. ③ Population size (Pop_it), represented by the natural logarithm of regional resident population at the end of the year. Human activities are closely related to carbon emissions; thus, population size is expected to have a positive effect on the carbon emission intensity. ④ Urbanization (Urb_it), represented by the ratio of the regional urban population to the resident population at the end of the year. Human activities in cities are more likely to produce carbon emissions than those in rural areas.; thus, regional urbanization is expected to have a positive effect on the carbon emission intensity. ⑤ Greening level (Gr_it), represented by the per capita park green area (square meter/person). The greater the value of Gr_it, the higher the greening level. Greening is an important way to reduce carbon emissions, and it is expected that the greening level will have a negative effect on the carbon emission intensity. ⑥ Environmental governance (Eg_it), represented by the proportion of regional fiscal expenditure on environmental protection out of the general budget expenditure. The greater the value of Eg_it, the stronger the intensity of environmental governance. Environmental governance is also an important way to reduce carbon emissions, and it is expected that environmental governance will have a negative effect on the carbon emission intensity. ⑦ Tax burden (Tax_it), represented by the proportion of the regional tax revenue out of the GDP. With the increase in the regional tax burden, on the one hand, the after-tax profits of existing enterprises in the region will be reduced, and the high-polluting and low-value-added industries will be transferred out of the region, thus reducing the carbon emission intensity. On the other hand, if the most relevant enterprises emitting CO₂ are highly dependent on regional resources or markets, they will be more inclined to externalize the internal costs caused by the increased tax burden than to transfer out of the region, and will further increase production, which increases the carbon emission intensity in the region. Therefore, the impact of the regional tax burden on the carbon emission intensity is uncertain.

3.3. Data Description and Descriptive Statistics

Given the availability of data, the balanced panel data of 30 provincial administrative regions in China from 2009 to 2020 are selected as the research materials. These 30 provincial administrative regions include Heilongjiang, Jilin, Liaoning, Hebei, Henan, Shandong, Shanxi, Anhui, Jiangxi, Jiangsu, Zhejiang, Fujian, Guangdong, Hunan, Hubei, Hainan, Yunnan, Guizhou, Sichuan, Qinghai, Gansu, Shaanxi, Inner Mongolia, Xinjiang, Guangxi, Ningxia, Beijing, Tianjin, Shanghai and Chongqing. The total number of research samples is 360. Data on the annual consumption of each energy source in each province come from the annual China Energy Statistical Yearbook for 2010 to 2021. Data on the CO₂ emission coefficient of each energy source come from the carbon neutrality research database of the CSMAR [28]. Data on each type of industrial land transfer come from the China Land Market Network (https://www.landchina.com/ (accessed on 5 January 2023)). Data on the annual GDP, per capita GDP, secondary industry added value, resident population at year-end, urban population, per capita park green area, fiscal expenditure on environmental protection, general budget expenditure and tax revenue of each province come from the regional economic statistics database of the Development Research Center of the State Council [60]. The descriptive statistics of all variable data are shown in

Table 1. Variable definitions and descriptive statistics.

Types	Variable	Meaning	Measurement	Obs.	Mean	Std. Dev.	Min.	Max.	Expected Effect
Independent variable	CIit	Carbon emission intensity	CO2 emissions per unit GDP	360	2.6344	1.9738	0.2722	8.9780
Core dependent variable	Land_Markit	Marketization of industrial land transfer	The proportion of the number of industrial land transferred by ”bid, auction and listing” out of the total transferred industrial land	360	0.9011	0.1068	0.4741	1.0000	−
Mechanism variable	Ind_selit	Industry selection	The proportion of industrial land transferred to the top five most carbon-emitting industries out of the total of transferred industrial land	360	0.0782	0.0527	0.0000	0.4462
Control Variables	PGDPit	Economic development level	The natural logarithm of per capita GDP	360	10.6881	0.4988	9.2886	12.0086	−
	Ind_struit	Industrial structure	The proportion of regional secondary industry added value out of GDP	360	0.4174	0.0819	0.1597	0.6196	∩
	Ind_stru2it		The square term of the proportion	360	0.1809	0.0632	0.0255	0.3839
	Popit	Population size	The natural logarithm of the regional resident population at year-end	360	8.1990	0.7427	6.3231	9.4433	+
	Urbit	Urbanization	The proportion of the regional urban population out of the resident population at year-end	360	0.5768	0.1285	0.2989	0.8960	+
	Grit	Greening level	Per capita park green area	360	12.6576	2.8600	6.1300	21.0500	−
	Egit	Environmental governance	The proportion of regional fiscal expenditure on environmental protection out of the general budget expenditure	360	0.0305	0.0098	0.0114	0.0681	−
	Taxit	Tax burden	The proportion of regional tax revenue to GDP	360	0.0851	0.0281	0.0428	0.1882	/

Notes: ”+” indicates that the expected impact is positive; “−” indicates that the expected impact is negative; “∩” indicates that the expected impact will show an inverted U-shaped curve; “/” indicates that the expected impact is uncertain.

. “Obs.” refers to the number of research samples. The mean is the average value of the variable. Std. Dev. Is the standard deviation of the variable. Min. is the minimum value of the variable. Max. is the maximum value of the variable. Expected effect is the expected effect of each variable on the carbon emission intensity.

4. Results and Discussion

4.1. Typical Characteristics

4.1.1. Concerning the Carbon Emissions

Before conducting an econometric analysis, we need to understand the trend change in China’s carbon emissions from 2009 to 2020. As shown in Figure 2, China’s total CO₂ emissions increased gradually from 99.021 × 100 million tons in 2009, to 141.707 × 100 million tons in 2020, which is an increase of 43.11%. However, the growth rate of China’s total CO₂ emissions has been gradually declining. From 2009 to 2014, the growth rate of China’s total CO₂ emissions was 27.13%, but was 12.57% from 2014 to 2020. This was helped by the continued reduction in China’s carbon emission intensity. From 2009 to 2020, China’s carbon emission intensity dropped from 2.834 ton/CNY 10 thousand to 1.406 ton/CNY 10 thousand, a decrease of 50.39%. The continuous decline of China’s carbon emission intensity is the key to achieving carbon emission reduction.

According to their geographical distribution and economic situation, the 30 provinces in China can be grouped into east and mid-west areas. The east area includes 11 provinces, namely, Shanghai, Beijing, Tianjing, Shandong, Guangdong, Jiangsu, Hebei, Zhejiang, Hainan, Fujian and Liaoning. The mid-west area includes the others. Figure 3 displays the spatial distribution of the mean values of per capita GDP from 2009 to 2020 in the study areas. As shown, compared with the mid-west, the east features a significantly higher level of economic development, which explains the higher level of their regional production technology. Thus, their carbon emission intensity should be heterogeneous within the area.

As shown in Figure 4a, the total CO₂ emissions in the east increased from 46.652 × 100 million tons in 2009, to 63.399 × 100 million tons in 2020, an increase of 35.90%, while in the mid-west, the total CO₂ emissions increased from 52.369 × 100 million tons to 78.308 × 100 million tons, an increase of 49.53% (approximately 14 percentage points higher than that in the east). As shown in Figure 4b, the carbon emission intensity in the east dropped from 2.300 ton/CNY 10 thousand in 2009 to 1.152 ton/CNY 10 thousand in 2020, a decrease of 49.94%, while in the mid-west, the carbon emission intensity dropped from 3.571 ton/CNY 10 thousand to 1.712 ton/CNY 10 thousand, a decrease of 52.07%. This indicates that although the carbon emission intensity values in the east and the mid-west both decreased to a certain extent, in absolute terms, the carbon emission intensity in the east has been lower than that in the mid-west for a long time, but the difference has been gradually narrowing, which is closely related to the level of economic and industrial development.

4.1.2. Concerning the Industrial Land Transfer

The changing trend of China’s marketization of industrial land transfer from 2009 to 2020 must also be understood. As shown in Figure 5, the proportion of industrial land transferred by “bid, auction and listing” out of the total in China first showed a sharp rise, from 0.836 in 2009, to 0.952 in 2015, an annual increase of 1.94 percentage points. Then, it underwent a slight decline, from 0.952 in 2015 to 0.936 in 2020, an annual decrease of 0.32 percentage points. However, with the implementation of deep, market-oriented reforms in China, the marketization of industrial land transfer will be further expanded.

Figure 6 shows China’s industry selection of industrial land transfer from 2009 to 2020. As displayed, the proportion of industrial land transferred to the five top carbon-emitting industries out of the total in China fluctuated, but was generally stable at around 0.070. This indicates that these high-carbon emission industries still hold some importance in China. Figure 6 also shows China’s industry selection of industrial land transfer by area from 2009 to 2020. As displayed, the proportion in the east also fluctuated, but was generally stable around 0.065. The proportion in the mid-west underwent a decline preceded by a fluctuation from 0.087 in 2009 to 0.066 in 2014, a decrease of 2.09 percentage points. Then, it showed an increase with fluctuation from 0.066 in 2014 to 0.084 in 2020, an increase of 1.85 percentage points. However, compared with the mid-west, the east showed a lower proportion from 2009 to 2020. This indicates that the mid-west was more inclined to introduce high-carbon-emitting industries due to differences in resource endowments compared with the east.

4.2. Econometric Results

4.2.1. Benchmark Regression

Table 2. Results of the benchmark regression of the marketization of industrial land transfer and carbon emission intensity.

Variable	CIit
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Land_Markit	−2.0699 **	−1.8697 ***	−1.5381 ***	−1.5295 ***	−1.2482 ***	−1.1776 ***	−0.9151 **	−0.9943 **	−0.9900 **
	(0.9311)	(0.5808)	(0.4560)	(0.4658)	(0.4810)	(0.4365)	(0.4135)	(0.4284)	(0.4314)
PGDPit			−1.6398 ***	−1.6233 ***	−1.5503 ***	−0.8994 **	−0.8039 **	−0.7673 **	−0.7592 **
			(0.3329)	(0.3778)	(0.3892)	(0.3763)	(0.3671)	(0.3583)	(0.3583)
Ind_struit				5.0091 *	3.3168	9.4620 ***	9.0696 ***	9.1818 ***	9.1936 ***
				(2.6924)	(2.6502)	(2.9292)	(2.9178)	(2.9383)	(2.9404)
Ind_stru2it				−5.9243 *	−5.3274 *	−12.7427 ***	−12.2871 ***	−12.6458 ***	−12.5617 ***
				(3.2034)	(3.0993)	(3.3525)	(3.3709)	(3.4007)	(3.4183)
Popit					2.6697 ***	2.5550 ***	2.8114 ***	2.9255 ***	2.8975 ***
					(0.5084)	(0.5368)	(0.5345)	(0.5384)	(0.5455)
Urbit						−4.9153 ***	−4.4516 ***	−4.9739 ***	−4.9928 ***
						(1.2421)	(1.2403)	(1.2254)	(1.2363)
Grit							−0.0482 ***	−0.0492 ***	−0.0494 ***
							(0.0163)	(0.0161)	(0.0161)
Egit								−4.2249 *	−4.4656 *
								(2.5489)	(2.5158)
Taxit									1.4245
									(2.1158)
CONSTANT	4.4995 ***	2.6807 ***	20.5500 ***	19.6389 ***	−1.3371	−4.6742	−7.5762	−8.1919	−8.2583
	(0.8362)	(0.3781)	(3.7402)	(4.0830)	(6.5067)	(6.2794)	(5.8626)	(5.8062)	(5.8019)
Provincial fixed effects	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	360	360	360	360	360	360	360	360	360
R2	0.0125	0.9800	0.9830	0.9830	0.9844	0.9853	0.9858	0.9860	0.9860

Notes: The values of all variables are available in Table 1. The robust standard errors are in “( )”; “*” is significant at the 0.10 level; “**” is significant at the 0.05 level; “***” is significant at the 0.01 level.

shows the results of the benchmark regression performed using the ordinary least squares (OLS) method to estimate model (1). Column (1) shows the regression results with only the variable Land_Mark_it introduced. Column (2) shows the regression results after controlling for the provincial effects and the year fixed effects. Columns (3) to (9) show the regression results with the stepwise addition of the control variables. All the regression results show that regardless of whether the fixed effects and control variables are introduced into the model or not, the marketization of industrial land transfer (Land_Mark_it) has a significantly negative effect on the regional carbon emission intensity (CI_it); that is, with the increase in the marketization degree, the regional carbon emission intensity will decrease, which verifies Hypothesis H1 (the marketization of industrial land transfer can significantly reduce the regional carbon emission intensity). Marketization can remove institutional barriers that hinder the independent and orderly flow of factors of production, and comprehensively improve the efficiency of the coordinated allocation of factors of production [29]. Huang et al. [54] argued that through transparent transfer routes, the marketization of industrial land transfer encouraged the introduction of low-energy and high-efficiency industries, thereby reducing the regional carbon emission intensity. Moreover, we have also conducted an econometric regression of the effect of Land_Mark_it on the regional total CO₂ emissions (CO2_it) and found that its influence coefficient is not significant. This indicates that the marketization of industrial land transfer can not only reduce the regional carbon emission intensity, but also does not increase the total CO₂ emissions, which provides a model for the land policy regulation of carbon emission reduction.

In terms of control variables, as expected, the economic development level (PGDP_it), urbanization (Urb_it), greening level (Gr_it) and environmental governance (Eg_it) all have a significantly negative effect on the carbon emission intensity. This indicates that with the improvement of the regional economic development level, urbanization, greening level and environmental governance intensity, the regional carbon emission intensity decreases. Population size (Pop_it) has a significant (expected) positive effect on the carbon emission intensity, which indicates that with the increase in population size, the regional carbon emission intensity increases. Industrial structure (Ind_stru_it) has a significant (expected) effect, manifesting an inverted U-shaped curve, on the carbon emission intensity, which indicates that with industrial development, the regional carbon emission intensity firstly increases and then decreases. Moreover, the R² in column (9) is 0.986, which indicates that the regression model fits well. The regression results of these control variables can also provide guidance for the national governance of carbon emission reduction.

4.2.2. Robustness Test

The results of the benchmark regression performed in this paper may have been disturbed by the presence of abnormal samples with extreme values. To mitigate the interference of extreme values, we have processed the carbon emission intensity (CI_it) and the marketization of industrial land transfer (Land_Mark_it) via the bilateral Winsorization of 1% and 5%, and the bilateral censoring of 1% and 5%, and then performed regression on model (1) again. As displayed in

Table 3. Results of the robustness test: solving the abnormal samples.

Variable	CIit
	Bilateral Winsorization		Bilateral Censoring
	1%	5%	1%	5%
Land_Markit	−0.9795 **	−1.1788 **	−1.4544 ***	−1.6149 ***
	(0.4425)	(0.4669)	(0.4908)	(0.4368)
CONSTANT	−8.0778	−4.8389	−6.3475	−6.0348
	(5.8093)	(5.8888)	(5.7911)	(5.7962)
Control variables	Yes	Yes	Yes	Yes
Provincial fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Obs.	360	360	351	299
R2	0.9859	0.9846	0.9859	0.9857

Notes: The robust standard errors are in “( )”; “**” is significant at the 0.05 level; “***” is significant at the 0.01 level.

, whether the regression samples are processed via the bilateral Winsorization of 1% and 5% or bilateral censoring of 1% and 5%, the regression coefficients of Land_Mark_it are all significantly negative, which verifies the robustness of the results of the benchmark regression, stating that the marketization of industrial land transfer has a significantly negative effect on the regional carbon emission intensity.

We conducted the robustness test again with different variables. Firstly, we replaced Land_Mark_it with Land_Mark_Acr_it as the core independent variable. Land_Mark_Acr_it refers to the marketization of industrial land transfer, represented by the proportion of the acreage of industrial land transferred by “bid, auction and listing” out of the total acreage of all transferred industrial land. As displayed in column (1) of

Table 4. Results of the robustness test: replacing variables.

Variable	CIit	CI_Reit
Land_Markit		−1.5057 ***
		(0.5745)
Land_Mark_Acrit	−0.4886 ***
	(0.1863)
CONSTANT	−9.8297 *	−28.3698 ***
	(5.7186)	(9.3163)
Control variables	Yes	Yes
Provincial fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Obs.	360	360
R2	0.9858	0.9861

Notes: The robust standard errors are in “( )”; “*” is significant at the 0.10 level; “***” is significant at the 0.01 level.

, Land_Mark_Acr_it retains its significantly negative effect on the regional carbon emission intensity (CI_it), similar to Land_Mark_it in Table 2. Secondly, we replaced CI_it with CI_Re_it as the dependent variable. CI_Re_it refers to the carbon emission intensity, represented by the CO₂ emissions per unit real GDP, calculated using 2009 as the base period. As displayed in column (2), the marketization of industrial land transfer (Land_Mark_it) retains its significantly negative effect on the regional carbon emission intensity (CI_Re_it). These regression results also verify the robustness of the benchmark regression results.

On the one hand, the marketization of industrial land transfer and the carbon emission intensity may be causally related to each other in the same period. On the other hand, the construction period of most industrial projects is two years in China; that is, the effect of Land_Mark_it on CI_it may be lagged. Thus, we have replaced Land_Mark_it with its lagged 1 period Land_Marki_t−1 and its lagged 2 period Land_Marki_t−2 as the core independent variable, so as to consider the lagged effect. As displayed in

Table 5. Results of the robustness test: considering the lagged effect of Land_Mark_it.

Variable	CIit
	(1)	(2)
Land_Markit−1	−1.2440 ***
	(0.3564)
Land_Markit−2		−1.1131 ***
		(0.2838)
CONSTANT	−8.7583	−9.1583
	(5.6399)	(5.0786)
Control variables	Yes	Yes
Provincial fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Obs.	330	300
R2	0.9891	0.9913

Notes: The robust standard errors are in “( )”; “***” is significant at the 0.01 level.

, both Land_Marki_t−1 and Land_Marki_t−2 have a significantly negative effect on CI_it, which further confirms that the marketization of industrial land transfer can reduce the regional carbon emission intensity.

4.2.3. The Moderating Role of Environmental Governance

Because environmental governance is an important way for local governments to reduce carbon emissions, we believe that this form of governance not only has a direct impact on the carbon emission intensity, but it also plays a positive moderating role in the negative effect of the marketization of industrial land transfer on the carbon emission intensity. That is, as the intensity of environmental governance becomes stronger, the negative effect of the marketization of industrial land transfer on the carbon emission intensity also becomes stronger. Thus, we introduced an interaction term (Land_Mark_it × Eg_it) into model (1) to test the moderating role.

Table 6. Regression results of the moderating role of environmental governance.

Variable	CIit
Land_Markit	−0.6806 *
	(0.4103)
Land_Markit × Egit	−38.8309 **
	(16.0222)
Egit	−6.6010 ***
	(2.5516)
CONSTANT	−7.8827
	(5.7220)
Control variables	Yes
Provincial fixed effects	Yes
Year fixed effects	Yes
Obs.	360
R2	0.9863

Notes: In order to avoid serious multicollinearity, we performed mean centralization treatments on Land_Mark_it and Eg_it. The robust standard errors are in “( )”; “*” is significant at the 0.10 level; “**” is significant at the 0.05 level; “***” is significant at the 0.01 level.

reports the regression results regarding the moderating role of environmental governance. As displayed, the coefficient of the interaction term (Land_Mark_it × Eg_it) is significantly negative at the 0.05 level, which verifies Hypothesis H2 (environmental governance plays a positive moderating role in the negative effect of the marketization of industrial land transfer on the carbon emission intensity). Thus, in order to make good use of the carbon emission reduction effect of land policy regulation, it is necessary for local governments to strengthen the intensity of environmental governance.

4.2.4. Area Heterogeneity Analysis

Compared with the mid-west area, China’s east has a higher level of economic development and regional production technology. Moreover, the eastern area in China has a lower carbon emission intensity and exhibits a lower proportion of industrial land transferred to the top five most carbon-emitting industries out of the total. Therefore, the effect of the marketization of industrial land transfer on the regional carbon emission intensity may be heterogeneous by area. As displayed in

Table 7. Results of the area heterogeneity analysis.

Variable	CIit
	Eastern Area in China	Mid-Western Area in China	China
Land_Markit	−0.4148 **	−1.5564 **	−1.4968 **
	(0.1908)	(0.7234)	(0.6707)
Land_Markit × Ri			0.9102 *
			(0.5016)
CONSTANT	−24.7607 ***	−11.5701 *	−7.8262
	(8.4934)	(6.6077)	(5.8834)
Control variables	Yes	Yes	Yes
Provincial fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Obs.	132	228	360
R2	0.9908	0.9882	0.9861

Notes: The robust standard errors are in “( )”; “*” is significant at the 0.10 level; “**” is significant at the 0.05 level; “***” is significant at the 0.01 level.

, the marketization of industrial land transfer (Land_Mark_it) has a significantly negative effect on the regional carbon emission intensity (CI_it) in both columns (1) and (2), which indicates that the marketization of industrial land transfer can reduce the regional carbon emission intensity, whether it is in the east or the mid-west. However, the coefficient of Land_Mark_it in column (1) is −0.4148, which is much lower than the coefficient (−1.5564) of Land_Mark_it in column (2), and this indicates that the negative effect of the marketization of industrial land transfer on the regional carbon emission intensity is stronger in the mid-west area. The reasons may be that, first, the mid-west area in China has a lower level of economic development and a higher carbon emission intensity. The higher the intensity, the greater the space for emission reduction (to some extent). Therefore, compared with in the east, land policy regulation in the mid-west can play its role as an emission reduction dividend more effectively. Secondly, the mid-west area in China has a higher proportion of industrial land transferred to the top five most highly carbon-emitting industries out of the total. An important mechanism by which Land_Mark_it negatively affects CI_it is industry selection. Therefore, the marginal effect of Land_Mark_it on industry selection is stronger in the mid-west. That is, in the mid-west area, it is easier for the marketization of industrial land transfer to reduce the proportion of industrial land transferred to high-carbon-emitting industries out of the total, thereby reducing the regional carbon emission intensity.

We have also introduced an interaction term (Land_Mark_it × R_i) to test the area heterogeneity. If region i is in the east, R_i equals 1, and it equals 0 otherwise. As displayed in column (3), the coefficient of the interaction term (Land_Mark_it × R_i) is significantly positive at the 0.10 level, which further confirms the finding that in the mid-west area, it is easier for the marketization of industrial land transfer to reduce the regional carbon emission intensity.

4.2.5. Mechanism Test

As discussed earlier, an important mechanism by which the marketization of industrial land transfer negatively affects the regional carbon emission intensity is industry selection, and we verify this using the OLS method to estimate model (3). As displayed in column (1) of

Table 8. Results of the area heterogeneity analysis.

Variable	Ind_selit
	China	Eastern Area in China	Mid-Western Area in China
Land_Markit	−0.0769 **	0.0141	−0.1380 **
	(0.0389)	(0.0363)	(0.0651)
CONSTANT	2.4830 ***	1.0427	2.4358 ***
	(0.4804)	(0.6804)	(0.8380)
Control variables	Yes	Yes	Yes
Provincial fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Obs.	360	132	228
R2	0.7139	0.7196	0.7188

Notes: The robust standard errors are in “( )”; “**” is significant at the 0.05 level; “***” is significant at the 0.01 level.

, the marketization of industrial land transfer (Land_Mark_it) has a significantly negative effect on the industry selection (Ind_sel_it), which verifies Hypothesis H3 (industry selection is the internal mechanism of the effect of the marketization of industrial land transfer on regional carbon emission intensity). That is, the marketization of industrial land transfer can reduce the proportion of industrial land transferred to high-carbon-emitting industries out of the total, thereby reducing the regional carbon emission intensity. The coefficient of Land_Mark_it is not significant in column (2), but it is significantly negative in column (3), which indicates that the industry selection mechanism of the marketization of industrial land transfer is represented more prominently in the mid-west area. This result can further explain the stronger negative effect of the marketization of industrial land transfer on the regional carbon emission intensity in the mid-west area.

5. Conclusions and Policy Implications

Based on the balanced panel data of China’s provincial level from 2009 to 2020, this paper studies the effect of the marketization of industrial land transfer on the regional carbon emission intensity and its internal mechanism, thus providing guidance for land policy regulation that will enable carbon emission reductions. The main results are as follows: from 2009 to 2020, China’s carbon emission intensity decreased year by year, and compared with the east, the mid-west maintained a lower carbon emission intensity and a lower proportion of industrial land transferred to the top five most highly carbon-emitting industries out of the total for a long time. The results of the benchmark regression show that the marketization of industrial land transfer can significantly reduce the regional carbon emission intensity, which passes a series of robustness tests. Variables such as greening level and environmental governance also have a significantly negative effect on the carbon emission intensity, and environmental governance plays a positive moderating role in the negative effect of the marketization of industrial land transfer on the carbon emission intensity. The results of thearea heterogeneity analysis show that, compared with the effect in the east, the negative effect of the marketization of industrial land transfer on the carbon emission intensity is stronger in the mid-west. The results of the mechanism test show that industry selection is the internal mechanism determining the effect of the marketization of industrial land transfer on regional carbon emission intensity. That is, the marketization of industrial land transfer can reduce the proportion of industrial land transferred to high-carbon-emitting industries out of the total, thereby reducing the regional carbon emission intensity. Moreover, the industry selection mechanism of the marketization of industrial land transfer is represented more prominently in the mid-west.

The conclusions obtained in this paper have some policy implications that will enable China to further reduce its regional carbon emissions through land policy regulation, and achieve the goal of “carbon peaking and carbon neutrality”. First, the marketization of industrial land transfer is an important means to realize carbon emission reduction. Thus, China’s local governments should improve the market-based operation mechanism of land, and make the land transfer system more open and transparent, so as to improve the allocation and utilization efficiency of land resources through market-based competition. More specifically, local governments should standardize the bidding, auctioning and listing of transfer process, limit the scales and proportions of negotiated transfers, increase the transparency of the primary land market, and reduce rent-seeking motivations, and thus corruption in the process of land transfer. Second, the internal mechanism of the marketization of industrial land transfer affecting regional carbon emission intensity is industry selection. Thus, local governments should strengthen the quality review process applied to the projects introduced, avoid allocating limited land resources to industries with a low energy efficiency and high carbon emissions, and strive to create a positive business environment to attract industries with a high energy efficiency and low carbon emissions. More specifically, for industries with a high energy efficiency, the government should increase the support and preferential policies that are in place, and reduce the industries’ capital outflow. For industries with a low energy efficiency, local governments should undertake reasonable planning based on the local economic development level and industrial characteristics, and actively guide industrial transformations and upgrades. Third, environmental governance has a positive moderating role on the negative effect of the marketization of industrial land transfer on the carbon emission intensity. Thus, local governments should strengthen the intensity of their environmental governance, and establish a policy evaluation system to improve their efficiency and reduce the carbon emissions of industrial land. More specifically, local governments should raise the fiscal expenditure on environmental protection. This last policy is not only used to limit or restrain local high-polluting industries and build up the environmental infrastructure, but also to conduct environmental protection testing and publicity, so as to promote the development of local energy conservation and environmental protection.

Although this paper provides a more detailed empirical analysis of the relationship between the marketization of industrial land transfer and carbon emission intensity, whether the allocation of industrial land would have a more significant impact on carbon emissions is unclear, and this can be further discussed in the future. Moreover, the impact of the transfer of industrial output between regions on carbon emission intensity will also be considered.