Has land resource misallocation increased air pollution in Chinese cities?

Abstract

As a fundamental production factor of economic development, rational land allocation is relevant to economic development and a significant factor affecting urban air pollution. This study analyzes 284 cities in China to investigate the impact of land resource misallocation on air pollution and the underlying mechanisms. The relevant findings are fourfold. First, land resource misallocation increases urban air pollution. Second, land resource misallocation inhibits technological innovation, government technology investment, and foreign direct investment, increasing local air pollution. Third, the impact of land resource misallocation on air pollution is affected by heterogeneous conditions, including geographic region, city type, and city size. Finally, based on the air pollution caused by land resource misallocation, local governments should optimize the land supply structure and improve technological innovation and environmental investment.

Introduction

As the largest developing country in the world, China has experienced rapid urbanization since implementing its reform and opening-up policy in the 1980s. China’s urbanization rate, which was just 17.8% in 1978, increased to 52.6% in 2012 and will have reached 64.72% in 2021, according to the Chinese government’s website.1 However, this rapid economic development has generated severe problems of land resource misallocation and environmental pollution (Huang et al. 2015; Gan et al. 2021). The Chinese government began implementing tax-sharing reforms in 1994, resulting in unequal financial and administrative powers for local governments. In the context of fiscal expenditure exceeding revenue, some local governments made use of their monopoly positions in land supply to transfer land to obtain financial support for infrastructure development (Wu et al. 2015). At the same time, under the influence of local governments’ pursuit of increased gross domestic product (GDP) and official promotion, urban land supply policies have tended to favor short-term economic goals, leading to structural differences in the allocation of industrial and commercial land due to the lack of scientific planning (Li and Zhou 2005; Li et al. 2019). In addition, according to the investigative report of the Asian Development Bank regarding China’s environmental problems in the 12th Five-Year Plan period, 99% of the 500 cities in China did not meet the air quality standard defined by the World Health Organization, and seven of the 10 cities with the most severe air pollution in the world were in China (Zhang and Crooks 2012). While China’s air quality has improved in recent years, this study contends that land policies in the midst of rapid urbanization are one of the key factors affecting air quality.

Considerable research has examined the influencing factors of air pollution, but most studies only investigate the issue from environmental regulation, industrial structure, and foreign direct investment (FDI) perspectives (Hille et al. 2019; Ouyang et al. 2019; Zhang et al. 2022a, b;Tian and Feng 2022). Some literature has studied the causes of air pollution from the urbanization perspective, including population density (Borck and Schrauth 2021) and urban expansion (Huang and Du 2018). However, minimal research has analyzed the causes and mechanisms that influence air pollution from the perspective of land as a production factor. Compared with some countries, local governments in China have a monopoly position in the market for land supply. Under the influence of the existing promotion mechanism and GDP target assessment of government officials, land market prices and resource allocation may be distorted, which could have an impact on the environment. Therefore, this study analyzes 284 cities in China to investigate the impact of land resource misallocation on air pollution, examining the impact mechanism of land resource misallocation on air pollution from technological innovation, government technology input, and foreign investment perspectives, and determines the different effects of land resource misallocation on air pollution under various heterogeneous conditions.

The marginal contributions of this study are as fourfold. First, compared with previous research on the influencing factors of air pollution (He and Du 2022), this study examines the impact of land resource misallocation on air pollution using production factors. Second, limited previous studies have investigated the mechanisms and paths of land resource misallocation and air pollution (Huang and Du 2018; Li et al. 2021). This study analyzes the impact mechanism of land resource misallocation on air pollution from the three aspects noted above, enriching the research in this regard. Third, in contrast to previous single regional heterogeneity analysis (Xu et al. 2022), this study also examines the impact of land resource misallocation on air pollution under different city types and sizes, reflecting the heterogeneity of land policies and air governance in different cities. Fourth, previous provincial figures have neglected to include differences in urban land resource supply and air pollution, whereas land- and city-level particulate matter (PM2.5) data are used in this study, which are more likely to reflect real air pollution and lead to more reliable conclusions.

The remaining structure of this paper is as follows. The “Literature review and theoretical hypotheses” section is devoted to a literature review and theoretical hypotheses. The “Methodology and variables” section presents the methodology and variables of this study. The “Empirical results and discussion” section details the empirical conclusions and discussions. The “Conclusions, implications, and future research” section offers conclusions, implications, and directions for future research.

Literature review and theoretical hypotheses

Literature review

With the ongoing deterioration of the global environment and other challenges, particularly the COVID-19 pandemic, the issue of air pollution has attracted increasing attention from domestic and international researchers. Air pollution is primarily caused by the large amount of pollutants emitted by intensive heavy industry, resource-based enterprises, vehicles, and residents (Shendell 2021; Song et al. 2021). The classic environmental Kuznets curve (EKC) holds that there is a nonlinear relationship between environmental pollution and economic growth (Grossman and Krueger 1991), contending that haze pollution will intensify after an economy grows to a certain threshold in the development trajectory (Ma et al. 2016; Dong et al. 2018). In-depth research has been conducted regarding potential factors that lead to increased air pollution in the process of national economic development. For example, industrial emissions are generally considered to be the predominant cause of haze and air pollution (Duzgoren-Aydin 2007). Parikh and Shukla (1995) analyzed data from 83 developing countries, finding that increased urbanization leads to increased air pollution. Wang et al. (2019) found that about 75% of global greenhouse gas emissions and most PM2.5 emissions are generated by the energy sector. Xu et al. (2019) used panel data from Chinese provinces to examine the impact of exports and FDI on air pollutants.

Land is both a spatial carrier of economic activities and a factor resource that production activities rely on and use (Restuccia and Rogerson 2017). Misallocation of land resources can have a serious negative impact on economic and social development and the ecological environment. For example, Li et al. (2016) determined that misallocation of construction land resources in Chinese cities reduces industrial enterprises’ productivity. Adamopoulos et al. (2022) found that land resource misallocation constrains more productive farmers, which reduces overall agricultural productivity in China. Moreover, Li and Luo (2017) used Chinese city- and industry-level data and demonstrated that land resource misallocation is an important restricting factor of China’s industrial structure upgrade. Han et al. (2022) used panel data from 266 prefecture-level cities in China from 2004 to 2017, determining that land resource misallocation has an inhibitory effect on urban innovation. In addition, some scholars have explored the relationship between government intervention and land resource misallocation (Huang and Du 2017). Fan et al. (2021) studied the impact of China’s land quota system and land supply structure on housing prices, concluding that the lower the supply ratio of commodity and residential land, the higher the housing price. However, minimal research has investigated the impact of land resource misallocation on environmental pollution, particularly air pollution. Although some studies have examined the impact of land resource misallocation on urban green development and carbon emissions (Zhang and Xu 2017; He and Du 2022), few studies include the specific aspect of haze pollution in cities. Existing research also lacks the explorations of the specific mechanism of land mismatch on air pollution. Therefore, to clarify the impact and mechanism of land resource misallocation on air pollution, we construct a fixed effects model using panel data from 284 cities above the prefecture level from 2009 to 2019 for systematic analyses.

Theoretical hypotheses

Land resource misallocation and subsequent air pollution in prefecture-level cities are closely related to China’s unique system of fiscal decentralization and official assessment (Xu 2011). Tax-sharing system reform in 1994 and income tax-sharing reform in 2002 greatly enhanced the central government’s fiscal transfer payment capacity; however, these reforms also placed local governments under enormous financial pressure, forcing them to pursue off-budget revenue, such as land finance. Meanwhile, the political promotion incentives provided by the personnel appraisal system with Chinese characteristics force local governments to prioritize economic performance as the sole reference for job promotion, with promotions tied to fiscal measures of economic construction (Li et al. 2019). Consequently, to relieve financial pressure and obtain more political achievements, local government officials invest land resources in capital-intensive industries or heavy industry that contribute more to the GDP, resulting in increased pollutant emissions such as soot, sulfur dioxide (SO₂), and nitrogen oxides.

Under the current land supply system in China, as a scarce resource held by local governments, monopoly and intervention in land supply have become the primary policy tools for local governments to dominate the local economy and promote expanded investment. Specifically, local governments control the total amount of land supply in the urban primary market as well as overseeing the autonomous allocation of structure of land supply for different uses—industrial land and commercial and residential land. To rapidly improve the regional economy, local governments will adopt various supply strategies for different land uses, often selling industrial land at low prices to attract investment and generate tax revenue (Han and Kung 2015) while selling commercial and residential land at high prices to supplement fiscal revenue to make money from the land (Wu et al. 2015). The mismatch scale and price distortion of industrial land and commercial and residential land have reduced the efficiency of land use, causing distortions in the allocation of land resources (Restuccia and Rogerson 2017). The improper allocation of land resources reduces land-use efficiency, inhibits corporate innovation, and results in a large number of low-quality industrial investment projects. Furthermore, local governments’ allocation of a large amount of construction land to the industrial field at low prices also squeezes the development space for the modern service industry, which inhibits the optimization and upgrading of the industrial structure. This context advances the formation of industrial clusters with severe pollution and considerable emissions, aggravating ecological environment pollution and inhibiting the improvement of air quality in cities. Based on the above analysis, the following research hypothesis 1 is proposed:

• Hypothesis 1: Land resource misallocation aggravates air pollution in cities.

Technology innovation is an important means of advancing resource and energy utilization efficiency and improving the ecological environment (Danish and Ulucak 2020), in addition to being an essential driver for improving air pollution in cities. Land resource misallocation can seriously hinder technological innovation in prefecture-level cities. First, in the context of local government competition, the misallocation of land resources provides opportunities for rent-seeking and corruption (Li and Luo 2017; Huang and Du 2017). Selling industrial land at a low price indirectly reduces industrial enterprises’ production cost, expands enterprises’ profit space, and increases the probability of unproductive rent-seeking by entrepreneurs. Meanwhile, high-quality talent and entrepreneurs with innovation potential will abandon technology research and development (R&D) activities to engage in rent-seeking based on available opportunities to obtain excess profits, impeding technological innovation (Murphy et al. 1993). Second, land resource misallocation can drive up housing prices and make economic development dependent on the real estate economy, also inhibiting technological innovation (Yan et al. 2014). Rising residential prices will also increase the cost of living and reduce residents’ utility level, leading to the “outflow” of human capital, which further impedes cities’ overall innovation (Yu and Cai 2021). Thus, we propose the following hypothesis:

• Hypothesis 2: Land resource misallocation aggravates air pollution in cities by inhibiting technological innovation.

China endeavors to optimize the environment for innovation and development through government technology investment, particularly through long-term grants and project resources. To match green and innovative development strategies, financial departments often promote industrial innovation and development by adjusting project funding methods (Li 2017), wherein long-term financial support is provided for basic green projects, and priority is given to supporting key major projects in the field of green industries. This approach will promote the advancement of green productivity, air pollution improvement, and sustainable economic development (Pang et al. 2020). Land resource misallocation establishes conditions for local governments to implement a fiscal expenditure bias of focusing on production and ignoring innovation, resulting in insufficient investment in innovative activities by local governments. Specifically, local governments’ fiscal spending decisions are influenced by the existing political incentives. Under the decentralization system of China’s “official promotion tournament” (Li and Zhou 2005), local government officials hold a fiscal expenditure bias that emphasizes productive expenditure and neglects innovative expenditure to maximize the immediate economic and political interests during their term of office. Administrative land resource allocation is a significant tool for local governments to implement such biased expenditures. By intervening in land resource allocation, a large amount of low-cost industrial land can promote rapid economic growth, while a small amount of high-priced residential land can maximize fiscal revenue (Han and Kung 2015). Therefore, to leverage greater land demand, promote regional economic growth, and increase fiscal revenue, local governments tend to strategically allocate land through the excessive supply of industrial land and restricted residential land for suppliers, using land transfer income for productive expenditure such as infrastructure construction in industrial parks, rather than innovative expenditure that has a long cycle, slow results, and generally does not generate short-term benefits. Due to the potential positive externalities of innovation results, the promotion of technological innovation in cities cannot be separated from the financial support of the government (Hall 2002; Acemoglu et al. 2016); therefore, fiscal expenditure bias that influences land resource misallocation squeezes out local governments’ innovation expenditure, weakening the development, promotion, and application of cleaner production technology that could advance the effective control of haze pollution. Based on the above, our third hypothesis is proposed as follows:

• Hypothesis 3: Land resource misallocation aggravates air pollution in cities by inhibiting government technology input.

Minimization of production costs is the main factor affecting the location layout of FDI (Buckley 1989). With the rise of new geographic economics, scholars began to focus on the significant role of land prices in cities’ industrial development (Fujita and Thisse 2009; Pflüger and Tabuchi 2010). Land resource misallocation will have an impact on the price of industrial and commercial land, which will then negatively impact FDI. To invite capital investment, local governments endeavor to attract investment in key industries and enterprises using low industrial land prices. However, local governments will also increase the price of land sales to raise fiscal revenue, which will inhibit FDI, particularly environmentally friendly FDI (Cai et al. 2016). Land resource misallocation will lead to rising commercial land prices, and the rise in housing prices will also increase the cost of foreign-funded enterprises, with a restraining effect on FDI. In addition, FDI can facilitate local resource utilization efficiency and pollution emissions through direct capital financing and technology diffusion effects (Alfaro et al. 2004; Xie and Sun 2020). In addition, environmentally friendly foreign investment in green and clean production industries and equipment and the advanced green management approach of environment-friendly foreign investment promote local pollution control (Earnhart et al. 2014; Wang et al. 2017). Therefore, land resource misallocation has crowded out local FDI, which subsequently worsens air pollution in cities, leading to our fourth and final hypothesis.

• Hypothesis 4: Land resource misallocation aggravates air pollution in cities by inhibiting foreign direct investment.

Methodology and variables

Methodology

To investigate the impact of land resource misallocation (LRM) on air pollution (AP), the reduced form econometric model constructed in our study is as follows:

$${\mathrm{AP}}_{\mathit{it}}={\alpha }_0+{\alpha }_1{\mathrm{LRM}}_{\mathit{it}}+{\alpha }_2{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 1

where $i$ and t represent city and year, respectively. ${\mathrm{AP}}_{\mathit{it}}$ is the dependent variable denoting air pollution in prefecture-level cities. ${\mathrm{LRM}}_{\mathit{it}}$ is the independent variable representing land resource misallocation. ${\mathrm{Control}}_{\mathit{it}}$ is the control variable matrix, including five aspects of economic development, population density, education level, financial development level, and government intervention. Because air pollution between different cities greatly varies, we also control for the individual effects of the city and time (${\mathrm{City}}_i$ and ${\mathrm{Year}}_t$), and ${\epsilon }_{\mathit{it}}$ represents a disturbance term.

According to the previous analysis, land resource misallocation can indirectly affect air pollution through the transmission channels of technological innovation level, government technology input, and FDI. Our study uses a recursive model referencing the mediating effect method (Baron and Kenny 1986) to test how land resource misallocation affects the transmission mechanism of air pollution.

$${\mathrm{TI}}_{\mathit{it}}={\beta }_0+{\beta }_1{\mathrm{LRM}}_{\mathit{it}}+{\beta }_2{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 2

$${\mathrm{AP}}_{\mathit{it}}={\delta }_0+{\delta }_1{\mathrm{LRM}}_{\mathit{it}}+{\delta }_2{\mathrm{TI}}_{\mathit{it}}+{\delta }_3{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 3

where ${\mathrm{TI}}_{\mathit{it}}$ represents the regional technological innovation level that is the mediating variable, and other settings mirror Eq. (1). This study uses the stepwise regression method to conduct the intermediary test as follows: the first step is to regress Eq. (1). If ${\alpha }_1$ is significant, indicating the overall effect of land resource misallocation on air pollution, the next test is conducted; otherwise, the mediating effect is not evident and the test is terminated. The second step is to regress Eq. (2) to test the impact of land resource misallocation on technological innovation level. The third step is to regress Eq. (3) to test the direct effect of land resource misallocation on air pollution and the mediating effect of technological innovation level. If ${\beta }_1$ and ${\delta }_2$ are both significant, indicating an indirect effect, the fourth step test is performed to compare the signs of ${\beta }_1\times {\delta }_2$ and ${\delta }_1$. If the signs are the same, this indicates that there is a mediating effect. If the signs differ, there is no evident mediating effect.

Similarly, we also use the mediating effect model to test the role of government technology input level and FDI.

$${\mathrm{GOT}}_{\mathit{it}}={\gamma }_0+{\gamma }_1{\mathrm{LRM}}_{\mathit{it}}+{\gamma }_2{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 4

$${\mathrm{AP}}_{\mathit{it}}={\phi }_0+{\phi }_1{\mathrm{LRM}}_{\mathit{it}}+{\phi }_2{\mathrm{GOT}}_{\mathit{it}}+{\phi }_3{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 5

$${\mathrm{FDI}}_{\mathit{it}}={\partial }_0+{\partial }_1{\mathrm{LRM}}_{\mathit{it}}+{\partial }_2{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 6

$${\mathrm{AP}}_{\mathit{it}}={\theta }_0+{\theta }_1{\mathrm{LRM}}_{\mathit{it}}+{\theta }_2{\mathrm{FDI}}_{\mathit{it}}+{\theta }_3{\mathrm{Control}}_{\mathit{it}}+{\mathrm{City}}_i+{\mathrm{Year}}_t+{\epsilon }_{\mathit{it}}$$ 7

where ${\mathrm{GOT}}_{\mathit{it}}$ and ${\mathrm{FDI}}_{\mathit{it}}$ represent the government technology input level and the level of FDI, respectively, and the other settings are the same as Eq. (1), following the inspection steps and methods above.

Variable and definitions

Explained variable

Air pollution (AP)

At present, the primary form of urban air pollution is haze pollution, and PM2.5 is the main substance that produces haze pollution; however, China did not conduct PM2.5 monitoring in all prefecture-level cities until 2015. To solve the challenge of missing PM2.5 concentration data for prefecture-level cities in China, referencing Li et al. (2017) and Zhang et al. (2020), we use the satellite-based monitoring of annual global PM2.5 averages published by the Center for Socioeconomic Data and Application at Columbia University. Using ArcGIS software to analyze the grid data of PM2.5, the specific values of annual average PM2.5 concentration in 284 prefecture-level cities in China from 2009 to 2017 are obtained to capture the degree of urban air pollution.

Explanatory variable

Land resource misallocation (LRM)

The main problem investigated in this study is the impact of land resource misallocation on urban air pollution, arguing that the unreasonable development of industry is the main cause of cities’ air pollution. However, as noted above, the low-price assignment of industrial land represented by development zones leads to unreasonable industrial development. As the China Land and Resources Statistical Yearbook does not provide various cities’ industrial land transfer data, only providing the land transfer area and price of each city according to the transfer method, most researchers regard “agreement transfer” as a synonym for “industrial land” and “low-price transfer” (Guo et al. 2013). When the land area transferred by agreement in a city accounts for a large proportion of the total land transfer area, the city is more likely to have a large amount of land occupied by development zones and low barriers to entry for enterprises, which is more likely to lead to the misallocation of land elements, resulting in increased air pollution. Therefore, we use the proportion of the land area transferred by agreement to the total land area transferred as the key explanatory variable.

Mediating variables

Technology innovation (TI)

Technology innovation facilitates the improvement of air quality in cities by advancing the production technology of traditional industries and improving the efficiency of resource utilization (Li and Lin 2018). Existing literature mainly measures TI using R&D expenditure (Kontolaimou et al. 2016), number of patents, product innovation, and construction of comprehensive evaluation indicators for technological innovation (Guo et al. 2017). We comprehensively consider the availability and representativeness of data, selecting the number of city patent grants to represent TI.

Government technology input (GOT)

Enhancing the city’s innovation capacity is inseparable from local governments’ financial support (Acemoglu et al. 2016); however, land resource misallocation crowds out local government innovation spending (Chen and Kung 2016), resulting in insufficient support for innovation activities, hindering urban air pollution. In this study, the local governments’ subsidies for promoting enterprises’ technological innovation and R&D through direct and indirect means are collectively referred to as government technology input. As there are no specific statistics on government technology subsidies at the city level, science and technology expenditures are primarily represented as line items in public budget expenditures, which can reflect local governments’ R&D subsidies to enterprises. Therefore, this study uses the ratio of science and technology expenditure to GDP to represent government technology input.

Foreign direct investment (FDI)

The previous analysis suggested that land resource misallocation has a crowding-out effect on FDI. Meanwhile, FDI can generate changes in the host country’s technology, industrial structure, and market size, thereby facilitating air quality improvement in cities (Zhou and Li 2021). Therefore, to explore whether FDI is a transmission channel for land resource misallocation to air pollution in cities, FDI is represented by the amount of foreign capital used and logarithmically.

Control variables

In addition to the core explanatory variables, this study includes other control variables that affect air pollution, such as economic development (ED), represented by the logarithm of per capita GDP. As the EKC hypothesis holds that there is an inverted u-shaped relationship between environmental pollution and economic development level (Grossman and Krueger 1991), we include the squared term of economic development in the control variables. This study selects the ratio of the output value of the secondary industry to the regional GDP to represent the industrial structure (IS) level of cities. For population density (PD), this study uses the ratio of permanent urban population to city area. To represent the regional education level (RED), we use the logarithm of the annual number of students in regular higher education institutions in cities. Financial development level (FDL) is represented by the ratio of financial institutions’ loan balance to regional GDP. Variable definitions are summarized in Table 1.

Table 1.

Variable definitions

Classification	Symbol	Definitions
Explained variable	AP	Annual average estimated surface PM2.5 concentration
Explanatory variable	LRM	The ratio of the land area transferred by agreement to the total land area transferred
Mediating variables	TI	The number of city patent grants
	GOT	The ratio of science and technology investment to public finance expenditure
	FDI	The amount of foreign capital used and logarithmically
Control variables	ED	Region per capita GDP, taking the natural logarithm
	ED2	ED taking the square term
	IS	The ratio of the output value of secondary industry to regional GDP
	PD	The ratio of permanent urban population to city area
	RED	The annual number of students in regular higher education institutions in cities, taking the natural logarithm
	FDL	The ratio of financial institutions’ loan balance to regional GDP

Data and descriptive statistics

Considering the pertinence of the research, the continuity of data availability, and the comparability between cities, this study selected 2556 points of observational data from 284 cities in China from 2009 to 2017. We obtain the related variables from the China City Statistical Yearbook, the China Environment Statistical Yearbook, and datasets from the Economy Prediction System (EPS) database.2 This study also uses the interpolation method to define some missing values in the variables. Table 2 presents a statistical description of the main variables.

Table 2.

Descriptive statistics

Variable	Number	Mean	St. dev	Min	Max
AP	2556	44.192	19.553	3.596	110.121
LRM	2556	0.106	0.131	0.000	1.218
TI	2556	6.844	1.667	1.792	11.576
GOT	2556	0.002	0.003	0.000	0.063
FDI	2556	11.274	3.223	0	16.878
ED	2556	10.545	0.653	4.605	15.675
ED2	2556	111.627	13.819	21.207	245.712
IS	2556	0.488	0.104	0.149	0.898
PD	2556	0.043	0.034	0.0005	0.265
RED	2556	10.488	1.334	5.442	13.881
FDL	2556	0.882	0.540	0.118	7.450

Empirical results and discussion

Basic model regression results

Columns (1)–(7) of Table 3 present the regression estimation results of Eq. (1). This study applies the method of adding control variables step by step to regression Eq. (1) to capture the impact on our estimation results. Table 3 indicates that the coefficient of LRM is positive and has passed the significance level test in different control variable groups. According to column (7) of Table 3, after adding all control variables and controlling for time and city fixed effects, the coefficient of LRM is 2.751, with a 5% significance level, indicating that land resource misallocation increased AP in cities during the sample period, which supports the validity of hypothesis 1. This conclusion is similar to the conclusion that LRM increases industrial pollution emissions (Du and Li 2021). When local government land resources are misallocated and unbalanced, that is, urban land expansion and supply are unreasonable, urban air pollution is increased (Huang and Du 2018).

Table 3.

Baseline regression

Variables	AP	AP	AP	AP	AP	AP	AP
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
LRM	3.733*** (3.46)	2.602** (2.41)	2.647** (2.45)	2.695** (2.52)	2.672** (2.49)	2.674** (2.49)	2.658** (2.49)
ED		− 2.581*** (− 6.22)	− 0.384 (− 0.14)	1.649 (0.88)	1.689 (0.87)	1.586 (0.80)	1.206 (0.57)
ED2			− 0.106 (− 0.81)	− 0.172* (− 1.82)	− 0.175* (− 1.79)	− 0.171* (− 1.73)	− 0.158 (− 1.52)
IS				− 16.830*** (− 3.32)	− 16.946*** (− 3.33)	− 16.936*** (− 3.34)	− 14.046*** (− 2.71)
PD					40.346 (0.82)	39.658 (0.81)	35.838 (0.70)
RED						0.157 (0.35)	0.113 (0.25)
FDL							1.647*** (2.60)
Cons	45.164*** (220.39)	72.007*** (16.55)	60.682*** (4.35)	54.989*** (5.70)	53.245*** (5.33)	52.289*** (5.12)	52.574*** (4.92)
City FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.526	0.535	0.535	0.540	0.540	0.540	0.543
N	2556	2556	2556	2556	2556	2556	2556

The values of t are given in brackets

***, **, and * represent the different significance levels (1%, 5%, and 10%, respectively)

From the control variables in column (7), the coefficients of ED and ED² are 1.206 and $-$ 0.158, respectively, both of which do not pass the significance level test. This indicates that ED has no significant EKC effect on AP. The coefficient of IS affecting AP is − 14.046, with a 5% significance level, indicating that the upgrading of industrial structure can effectively reduce cities’ AP. The coefficient of FDL affecting AP is 1.647, with a 5% significance level test, indicating that FDL significantly increased cities’ AP. The rationale for this could be due to the fact that urban financial development is not oriented to green investment and loans in the process of economic development, thus increasing urban air pollution (Canh et al. 2021). The effect of RED on AP was negative but did not pass the significance level test, and PD had a positive effect on AP but also failed the significance level test.

Robustness test

To make the estimation results of the benchmark regression equations more plausible, we perform a series of robustness tests, presenting the results in Table 4. First, we replace the explanatory variable using SO₂ as a proxy indicator for AP, as SO₂ is one of the six standard sources of air pollutants that can well represent the degree of local air pollution (Zeng et al. 2019). Column (8) presents the estimation results of replacing the explained variables in Eq. (1), revealing a 0.217 coefficient of LRM, which passes the 5% significance level test, indicating that LRM increases SO₂ emissions and aggravates urban air pollution.

Table 4.

Robustness test

Variables	SO2	AP	AP	AP	AP
	(8)	(9)	(10)	(11)	(12)
LRM	0.207* (1.95)	2.630** (2.33)	2.617** (2.40)	2.914*** (2.62)	3.410** (2.44)
ED	− 0.188 (− 0.50)	6.879 (0.75)	0.768 (0.34)	0.751 (0.32)	0.982 (0.56)
ED2	0.007 (0.34)	− 0.456 (− 1.05)	− 0.143 (− 1.31)	− 0.146 (− 1.31)	− 0.131 (− 1.58)
IS	1.129** (2.55)	− 12.159** (− 2.25)	− 13.625** (− 2.58)	− 14.104*** (− 2.67)	− 24.059*** (− 4.06)
PD	− 14.149*** (− 3.71)	90.755 (1.10)	8.908 (0.20)	3.948 (0.09)	166.582* (1.86)
RED	0.060 (1.48)	0.063 (0.12)	0.096 (0.21)	0.108 (0.24)	0.138 (0.31)
FDL	− 0.022 (− 0.26)	2.599*** (3.30)	1.995** (2.38)	1.766*** (2.63)	1.913*** (2.68)
Cons	11.214*** (5.90)	22.520 (0.48)	56.366*** (4.98)	54.449*** (4.61)	50.001*** (5.42)
City FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.426	0.545	0.551	0.551	0.542
N	2556	2556	2286	2304	1800

The values of t are given in brackets

***, **, and * represent the different significance levels (1%, 5%, and 10%, respectively)

Second, as outliers can lead to biased estimation results, we conduct outlier processing. To do so, all continuous variables are constrained by 1% above and below. In column (9), the coefficient of LRM is positive at a 10% significance level, indicating that LRM increases cities’ AP.

Third, special samples are removed. Provincial capitals, municipalities directly under the central government, and planned cities in China are clearly superior to other cities in terms of political and economic status; thus, these cities differ from other cities in terms of industrial development and environmental governance. Therefore, these cities are excluded, Eq. (1) is re-estimated, and the results are shown in column (10). The coefficient of LRM is positive at a 5% significance level, and the results are consistent with the benchmark regression equation.

Fourth, we remove heavy industrial cities. First, economic activity is more prosperous in polluted cities. These cities have a large number of manufacturing industries, such as steel, chemicals, and other pollution-intensive industries. Second, since some of the cities in the sample are heavily polluted, these cities are excluded, considering that outliers may have an impact on the estimation results. Column (11) reflects this estimation result, where the coefficient of LRM remains positive at the 5% significance level, confirming that LRM increases cities’ AP.

Fifth, we remove heavily polluted cities. Pollutant emissions are the primary cause of worsening air quality, and heavily polluted cities have more positive economic activity. This includes heavy industrial cities and those with active traditional transportation such as cars, which are major sources of air pollution. In addition, air flow can be restricted due to the geographical environment of some cities, which aggravates air pollution (Xu et al. 2019). Column (12) reflects the results of this estimation, where the coefficient of LRM is positive at the 5% level of significance, indicating that LRM increases AP. Based on the above robustness tests, it can be considered that the regression results of the benchmark equation in this study are robust, further confirming the validity of hypothesis 1.

Endogeneity test

There may be a bidirectional causal relationship between land resource misallocation and air pollution. This is because China’s current air pollution controls are mainly based on polluting companies. In particular, the government uses the ratio of different industrial and commercial land to regulate the size and number of polluting enterprises, thus achieving the goal of reducing carbon emissions and air pollution. According to the findings of Chen and Kung (2016), the steepness of terrain will have an impact on the mismatch of land resources. The higher the urban slope, the higher the cost of construction and production for industrial enterprises. As a result, the steepness of the urban terrain will limit the size of land available for industrial enterprises. At the same time, slope, as a geographical condition, will not have a significant impact on haze mainly generated by industrial emissions. Therefore, this paper chooses the slope as the instrumental variable of land resource misallocation and adopts the two-stage least squares (2SLS) method for estimation. The first and second columns of Table 5 report the estimation results of the two-stage least squares method, respectively. According to column (13), the effect of TR on LRM is significantly negative, indicating that the steeper the slope is, the less the industrial agreement land area is. At the same time, the F values of 12.49 and 13.07 for the first and second stages are both larger than 10 and consistent with the rule of thumb, which also suggests that there is no problem with weak instrumental variables. It can be seen from column (14) that the coefficient of LRM in the second stage is significantly positive. This shows that the results are consistent with the benchmark regression after controlling for the endogeneity of land resource misallocation.

Table 5.

Endogeneity test: IV (2SLS) estimation

Variables	First stage	Second stage
	LRM	AP
	(13)	(14)
LRM		211.799*** (3.56)
SLOPE	− 0.00009*** (− 3.62)
ED	− 0.126*** (− 3.14)	30.225** (2.31)
ED2	0.004* (2.30)	− 1.189** (− 2.40)
IS	− 0.177*** (− 2.22)	62.497*** (4.35)
PD	1.125 (1.78)	196.759 (− 1.30)
RED	0.003 (0.39)	− 0.715 (− 0.44)
FDL	0.007 (0.90)	− 0.203 (− 0.11)
Cons	12.829*** (3.94)	− 178.483* (− 1.74)
N	2556	2556
Cragg-Donald Wald F statistic	12.49
Kleibergen-Paap Wald rk F statistic	13.07

The first column in parentheses is the t value, and the second column in parentheses is the z value

***, **, and * represent the different significance levels (1%, 5%, and 10%, respectively)

Heterogeneity test

This study analyzes heterogeneity in the impact by LRM on air pollution (AP) from three aspects of region, city type, and city size. Table 6 presents the estimation results of Eq. (1) applying heterogeneity analysis. First, Chinese cities are divided into the eastern, central, and western regions according to geographical and economic factors to examine regional differences in the impact of LRM on AP. According to columns (15)–(17) of Table 5, the coefficient of LRM in eastern and central regions is positive, whereas the coefficient of LRM in western regions is negative; however, only the impact of LRM on air pollution (AP) in cities in the central region passed the significance level test. This indicates that the LRM has a significantly positive effect on AP in the central region, increasing air pollution in cities in the central region while having no significant effect in other regions. The rationale for this is as follows: as the government allocates resources, land is an important means for local governments to attract investment. The central region is favored by a national policy featuring “the rise of the central region,” and its economic activity is more active than that in the western region (Wang and Feng 2021). Therefore, a large amount of industrial land is needed to attract enterprise investment and support industrial park development, and the resulting land resource misallocation accelerates industrial development, increasing air pollution (Du and Li 2021). In contrast, the western region’s economic development is relatively behind, and the carrying capacity of the ecological environment is limited. Industrial land in the western region tends toward accelerated local industrial transformation and reducing the damage to the ecological environment (Hu et al. 2022), which causes land resource misallocation and reduces air pollution, but this is not significant in the sample period of this study. Furthermore, there is competition among local governments. In the context of the convergence of industrial land prices, cities in the central region may lower environmental standards to attract investment and compete with industries in the developed eastern regions, which also aggravate air pollution in the central region (Hu et al. 2019). In particular, the effect of the LRM on the AP in the eastern region is not significant, which may not be realistic. We suggest that the reason for the insignificance of the eastern region may lie in the sample time bias, as the sample used in this study is from 2009 to 2017. During this period, the cities and industrialization in the eastern region were relatively mature, making it difficult for local governments to address air pollution reduction by regulating the land supply.

Table 6.

Heterogeneity test

Variables	Area			City type		City size
	East	Center	West	Resource-based	Non-resource	Large	medium	Small
	(15)	(16)	(17)	(18)	(19)	(20)	(21)	(22)
LRM	1.717 (1.47)	5.874** (2.43)	0.516 (0.26)	5.206*** (3.11)	1.465 (1.07)	3.202 (0.65)	3.456** (2.13)	1.927 (1.38)
ED	34.039*** (3.49)	17.539 (1.05)	− 0.896 (− 0.47)	2.204 (1.60)	− 6.759 (− 0.71)	− 24.016 (− 0.88)	− 14.69*** (− 3.14)	2.634 (1.18)
ED2	− 1.659*** (− 3.62)	− 0.903 (− 1.13)	− 0.053 (− 0.57)	− 0.153** (− 2.58)	0.144 (0.33)	0.856 (0.73)	0.503*** (2.72)	− 0.217 (− 1.64)
IS	− 4.124 (− 0.69)	− 30.32*** (− 2.98)	− 18.90** (− 2.28)	− 18.26** (− 2.17)	− 5.278 (− 0.73)	19.188 (0.50)	5.955 (0.59)	− 23.47*** (− 3.86)
PD	− 78.612* (− 1.83)	173.59* (1.85)	232.39 (0.53)	49.56 (0.47)	47.46 (0.79)	87.47 (1.28)	50.46 (0.65)	71.76 (0.45)
RED	− 0.767*** (− 3.04)	− 1.214 (− 1.56)	1.039 (1.51)	− 0.813 (− 1.06)	0.682 (1.09)	− 1.934 (− 0.67)	1.703 (1.31)	0.123 (0.26)
FDL	1.520** (2.21)	2.070** (2.04)	1.670 (1.31)	1.729* (1.95)	2.042** (2.46)	− 0.439 (− 0.13)	1.392 (1.53)	1.786** (2.12)
Cons	− 114.23** (− 2.30)	− 17.42 (− 0.20)	44.11*** (3.15)	49.71*** (4.88)	93.18* (1.94)	223.717 (1.48)	120.91*** (4.78)	45.237*** (3.74)
City FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.598	0.546	0.653	0.498	0.583	0.672	0.595	0.526
N	1026	981	549	1026	1530	171	909	1476

The values of t are given in brackets

***, **, and * represent the different significance levels (1%, 5%, and 10% respectively)

Second, we examine the type of city. Compared with non-resource-based cities, resource-based cities have more serious environmental pollution due to a high dependence on resource mining and processing industries (Xie et al. 2020; Zhang et al. 2022a, b). According to the Chinese government’s classification of resource-based cities, this study divides the sample into resource- and non-resource-based industries. According to columns (18)–(19) of Table 5, the LRM coefficient of the resource-based city group is significantly positive at the 1% level, while the LRM of the non-resource-based city group does not pass the significance level test. This suggests that LRM has a more significant positive effect on AP in resource-based cities compared with non-resource-based cities. The reason is that the land supply of resource-based cities tends to be used for resource extraction and industrial land for resource-based and supporting industries, resulting in the land allocation of resource-based cities supporting the development of local resource-based industries, causing more severe local air pollution.

Third, we examine city size. There are differences in land allocation and air pollution between cities of different sizes. In this study, Chinese cities are classified into large, medium, and small cities, and according to columns (20)–(22) of Table 5, the LRM coefficients of cities of three sizes are all positive, but only medium-sized cities pass the significance level test. This suggests that LRM has a more significant positive effect on AP in medium-sized cities compared with large and small cities. The reason is related to China’s long-term implementation of an urbanization strategy of controlling the size of big cities and actively developing medium and small cities. The process of urbanization in large cities is relatively high. Within the limited land supply, industrial land should be controlled to advance industrial upgrading, and commercial land should be increased to meet the needs of economic activity. In particular, large cities have higher levels of environmental regulation, so LRM cannot significantly increase local air pollution. In comparison to large and medium cities, small cities have poor transportation infrastructure and market environments, and local governments’ resources are not attractive enough to draw industrial investment through land transfer. Therefore, land misallocation in favor of industrial land does not significantly increase local air pollution (Tang et al. 2020). Amid the transformation and upgrading of the industrial structure, medium-sized cities have become the main regions undertaking industrial transfer from larger cities, causing medium-sized cities to have internal economic development needs for industrial support facilities and infrastructure to accommodate industrial transfer. Moreover, the governments of most medium-sized cities have competitive relationships in the process of undertaking industrial transfer, making lower industrial land price the main strategy for attracting industrial transfer and investment (Han and Kung 2015; Liu and Geng 2019), leading to a significant increase in local AP in the LRM of medium-sized cities.

Mechanism analysis

We next examine the mediating mechanism of LRM on air pollution (AP), and the estimation results of columns (23)–(28) are shown in Table 7. Columns (23)–(24) test the mediating effect of technological innovation (TI), revealing that the coefficient of LRM in column (23) is − 0.260 at a 5% significance level, and the coefficient of TI in column (24) is − 0.526 at a 5% significance level, indicating a significant indirect effect. The coefficient of LRM in column (24) is 2.521 at a 5% significance level, also indicating a significant direct effect. In addition, the sign of the product between the coefficient of LRM and the coefficient of TI is consistent with the sign of the coefficient of LRM in column (24). The proportion of the intermediate effect to the total effect is 0.054 ((0.260 × 0.526)/2.521)). This result suggests that TI has a mediating effect between LRM and AP, and LRM increases cities’ AP by inhibiting TI, validating hypothesis 2. As for rationale, although enterprises’ production cost is reduced in the short term, in the long run, enterprises may increase opportunities for rent-seeking and ignore innovation (Gao et al. 2021). In addition, a small amount of high commercial land prices will inevitably lead to high housing prices, not only causing investment to flood into the real estate industry, but also impeding the ability to attract innovative talent (Li and Luo 2017; Yu and Cai 2021). Furthermore, pressure to promote officials leads local governments to favor investment in infrastructure projects with short-term outcomes over innovation projects with uncertain outcomes and long cycles. This is because infrastructure projects meet the needs of local economic development while also yielding impressive economic growth figures for the year (Li and Zhou 2005). Thus, land misallocation limits technological innovation, which subsequently hinders local investment in environmental technology innovation and increases urban air pollution.

Table 7.

Mechanism test

Variables	TI	AP	GOT	AP	FDI	AP
	(23)	(24)	(25)	(26)	(27)	(28)
LRM	− 0.260**	2.521**	− 0.001*	2.588**	− 0.955**	2.459**
	(− 2.27)	(2.37)	(− 1.84)	(2.48)	(− 2.39)	(2.36)
TI		− 0.526**
		(− 1.98)
GOT				− 133.784**
				(− 2.21)
FDI						− 0.208***
						(− 2.61)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Constant	− 8.664	48.019***	− 0.002	52.314***	− 10.36***	50.420***
	(− 0.92)	(5.26)	(− 0.27)	(3.93)	(− 2.61)	(4.67)
City FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.682	0.544	0.063	0.544	0.065	0.545
N	2556	2556	2556	2556	2556	2556

Columns (25)–(26) test the mediating effect of government technology input (GOT). The results indicate that the coefficient of LRM in column (25) is − 0.001 at a 10% significance level, and the coefficient of GOT in column (26) is − 133.784 at a 5% significance level, which indicates a significant indirect effect. The coefficient of LRM in column (26) is 2.588 at a 5% significance level, which indicates that the direct effect is significant. Furthermore, the sign of the product between the coefficient of LRM and the coefficient of GOT is consistent with the sign of the coefficient of LRM in column (26), and the proportion of the intermediate effect to the total effect is 0.052 ((0.001 × 133.784)/2.588)). This result suggests that GOT has a mediating effect between LRM and AP, and LRM increases urban AP by inhibiting GOT, validating hypothesis 3. The reason is that land transfers have become a major source of revenue for local governments; however, due to negative environmental externalities, governments not only require institutional arrangements but also long-term capital investments for environmental governance. Under the fiscal decentralization mode in China, local governments’ technical and non-technical investments in environmental governance are insufficient, and a large number of low-cost industrial land transfers under the pressure of official promotion weaken fiscal revenue capacity (Han and Kung 2015; Liu et al. 2021), aggravating the challenges of environmental governance.

Columns (27)–(28) test the mediating effect of FDI. The results show that the coefficient of LRM in column (5) is − 0.955 at a 5% significance level, and the coefficient of FDI in column (27) is − 0.208 at a 10% significance level, indicating a significant indirect effect. The coefficient of LRM in column (27) is 2.459 at a 5% significance level, also indicating a significant direct effect. Furthermore, the sign of the product between the coefficient of LRM and the coefficient of TI is consistent with the sign of the coefficient of LRM in column (28), and the proportion of the intermediate effect to the total effect is 0.081 ((0.955 × 0.208)/2.459)). This result suggests that FDI has a mediating effect between LRM and AP, and LRM increases cities’ AP by inhibiting FDI, validating hypothesis 4. The reason is that land resource misallocation raises the production and operation costs of foreign investment, which has a negative impact on FDI. Local governments under financial pressure will raise land prices, which could lead to divestment or transfer of FDI to other low-cost cities. Particularly in the context of the green development strategy implemented by the Chinese government, polluting FDI prefers cities with lower comprehensive costs (Wang et al. 2017). In addition, high housing prices inhibit the flow of human capital, which weakens business development in technology-intensive FDI enterprises (Cai et al. 2016). As a result, land resource misallocation under the high cost of industrial land constrains local attractiveness for FDI and prevents clean FDI from introducing advanced green technologies and management equipment to advance local environmental governance, which further aggravates local air pollution.

Conclusions, implications, and future research

Conclusions and implications

Based on data regarding land resource misallocation and cities’ matching data, this study investigates the impact of land resource misallocation on air pollution in 284 cities in China. The relevant results are threefold. First, land resource misallocation has a significantly positive relationship with air pollution and can lead to increased urban air pollution. Second, technological innovation, government technology input, and FDI have partial mediating roles in the relationship between land resource misallocation and urban air pollution, namely, land resource misallocation worsens urban air pollution by inhibiting technological innovation, limiting government technology input and constraining FDI. Third, heterogeneity tests demonstrate that the impact of land resource misallocation on urban air pollution is moderated by geographic region, city type, and city size, revealing that land resource misallocation has a significant worsening effect on land air pollution in central, resource-based, and medium-sized cities, with no significant effect in other cities.

Based on the above conclusions, the implications of this study are as follows. First, local governments, which play a decisive role in land resource allocation, should reasonably develop land supply policies that simultaneously align with local economic endowments, development plans, and sustainable development. Local governments must strengthen market-oriented land supply reform, optimize the ratio of industrial land to commercial land, and continuously reduce the impact on the environment. Second, local governments should clearly recognize the negative externalities of the environment, promote land supply policies that favor innovative and green enterprises, persistently stimulate innovation and business environment improvement, and eliminate the negative impact of land resource misallocation on technological innovation, government technology input, and FDI. Third, a one-size-fits-all policy for land governance must be avoided, and the principle of strategically adapting measures targeting local conditions should be adhered to. In particular, in the context of carbon neutrality, it is essential to avoid vicious competition in less-developed regions to attract investment in high-emission industries to advance the achievement of carbon peak targets and promote win–win cooperation among regions.

Future research

Although the conclusions of this study concerning the impact of land resource misallocation on urban air pollution are robust, there are at least two aspects that could be further investigated in future research. First, due to the lack of public data, this study was unable to include the impact of the three red-line policies on real estate implemented since 2021 in its examinations. Second, the global economic downturn caused by the COVID-19 pandemic and the Russia–Ukraine conflict may also affect China’s urban land supply and environmental governance policies, further increasing uncertainty regarding the impact and mechanisms of land resource misallocation on air pollution. Third, in future investigations, we will collect the bid, auction, and listing data of urban land allocation in China to further test the impact of land resource misallocation on air pollution.

Author contribution

This manuscript was contributed by all authors. Xie Wancheng proposed the title and completed the preface, literature review, research hypothesis, and conclusion, and Zhang Ming wrote the preface, empirical analysis, and conclusion. Xie Wancheng and Zhang Ming procure funds. Xie Wancheng, Zhang Ming, and Gao Wen collected and sorted out the data together. Xie Wancheng, Zhang Ming, and Gao Wen reviewed the manuscript, and Xie Wancheng made the final revision. All authors have participated in the work of this manuscript.

Funding

This research was supported by the Chongqing Social Science Planning Project (2021NDYB050; 2022NDQN28) and National Social Science Foundation of China (21FJYB039).

Data availability

Not applicable.

Declarations

Ethical approval

Not applicable.

Consent to participate

The authors have read and approved the final manuscript.

Consent for publication

The authors agree to the publication of the journal.

Conflict of interest

The authors declare no competing interests.