Nonlinear Effects of Environmental Data Disclosure on Urban Pollution Emissions: Evidence from China

Abstract

Urban pollution emissions have become an unavoidable problem for China in its goal to achieve sustainable development, and environmental data disclosure is a key initiative for China to control urban pollution emissions. Based on the panel data of 120 cities in China from 2013 to 2018, this paper investigates the specific impact of environmental data disclosure level on urban pollution emissions. It was found that there is an inverted U-shaped curve relationship between the level of environmental data disclosure and urban pollutant emissions. A higher level of environmental data disclosure can alleviate information asymmetry and force polluters to take measures to ultimately reduce urban pollution emissions. However, a low level of environmental data disclosure cannot produce strong environmental constraints on polluters, although it may stimulate their speculative psychology, expand production before the advent of more efficient environmental supervision, and increase the total amount of urban pollution emissions. Therefore, the level of environmental data disclosure should be improved as much as possible, and the positive value of environmental data disclosure should be exploited to reduce urban pollution emissions. Heterogeneity shows that the impact of environmental data disclosure on urban pollution emissions is more significant in cities with higher entrepreneurial vitality, higher public environmental awareness, and stronger resource dependence. Further mechanism tests found that environmental data disclosure mainly affects urban pollutant emissions by increasing urban research investment, influencing the level of urban green technology innovation and the talent agglomeration effect, and improving urban green total factor productivity. These findings enrich the content of the literature regarding the relationship between environmental data disclosure and urban pollution emissions and present a feasible path for China to achieve emissions control goals through environmental data disclosure.

1. Introduction

China’s rapid economic growth has not only improved the material level of the people, but also brought about serious environmental pollution that seriously restricts the future sustainable development of China’s economy and society [1,2]. Environmental pollution has strong externalities that affect the health and quality of life of all people in the polluted area [3]. In order to solve the problem of environmental debt and improve the ecological environment, the Chinese government has repeatedly emphasized the need to strengthen environmental management, and since the promulgation of the Environmental Protection Law of the People’s Republic of China, dozens of laws and regulations related to resources and the environment have been introduced. The Report to the 20th National Congress of the Communist Party of China also pointed out that “Tightening resource and environmental constraints and environmental pollution were pronounced”, and it is necessary to “Intensifying pollution prevention and control”, aiming to reduce environmental pollution and strengthen environmental governance capabilities.

In order to curb the increasing environmental pollution and improve the level of environmental governance, scholars have carried out studies on the subject from the perspectives of the digital economy [4], green innovation [5], financial investment [6], government regulations, and many other aspects [7], achieving a series of research results. However, severe environmental problems have a strong relationship with the asymmetry of environmental information among the government, enterprises, and public [8]. Only by eliminating the obstruction of environmental information and having a comprehensive understanding of the current pollution situation and causes can effective wording be adopted to fundamentally reduce pollutant emissions and improve the ability of environmental governance. Therefore, this paper conducts relevant research from the perspective of environmental data disclosure. In order to ease the blockage of environmental information and promote the disclosure of environmental data, the Chinese government issued the Environmental Information Disclosure Measures (Trial) in 2007 and the National Key Monitoring Enterprise Pollution Source Supervision Monitoring and Information Disclosure Measures in 2013. Environmental data disclosure provides information support for accurate law enforcement by the government, provides a basis for the public to participate in environmental supervision, increases the cost of fines for enterprises that violate environmental laws [9], enhances the ability of higher governments to quantitatively assess the level of environmental governance by lower governments [10], and provides incentives for local governments to strengthen environmental control and for enterprises to take measures to reduce pollution emissions [11].

A high level of environmental data disclosure is certainly conducive to better control of urban pollution emissions and the promotion of the construction of green cities [12]. Environmental data disclosure is a process that occurs from a low level to a high level, so it is necessary to discuss the relationship between the level of environmental data disclosure and urban pollution emissions in detail. At present, most scholars regard environmental data disclosure as an event [8,13], and there are few studies on environmental data openness’s impact on urban pollution emissions under different levels of environmental data disclosure. This study has enriched the research on the relationship between environmental data disclosure and urban pollution emissions, has important social significance, and provides theoretical support for better promoting environmental data disclosure and pollution emissions control.

The main contributions of this study are reflected in the following aspects. (1) This paper does not regard environmental data disclosure as an event, but innovatively discusses the nonlinear impact of environmental data disclosure on urban pollution emissions according to the quantitative level of environmental data disclosure, enriching the research on the role of environmental data disclosure in urban pollution emissions. (2) By introducing environmental data into the model of the relationship between production function and pollution emissions, this paper demonstrates the inverted U-shaped influence of environmental data disclosure on pollution emissions through mathematical derivation and theoretically proves the possibility of the inverted U-shaped curve, similar to Kuznets curve, in urban pollution emissions and environmental data disclosure. (3) Further research on the mechanism of action is carried out, and it has been shown that environmental data disclosure forces cities to increase investment in science and technology, changes the level of green technology innovation and the talent agglomeration effect, and improves the effect of urban green total factor productivity on urban pollution emissions, all of which provide a rich explanation for how environmental data disclosure affects urban pollution emissions and provides important theoretical support for urban pollution emissions control.

The remainder of our paper unfolds as follows. Section 2 presents the results of a literature review, conducts a theoretical derivation of the impact of environmental data disclosure on urban pollution emissions, and presents the relevant research hypotheses. Section 3 presents the econometric model, variables, and data sources. Section 4 provides an analytical interpretation of the empirical results, conducts a heterogeneous analysis, and carries out a discussion of the mechanism involved. Section 5 presents our conclusions.

2. Literature Review and Research Hypothesis

2.1. Literature Review

Abundant environmental data are a prerequisite for improving the efficiency of pollution prevention and control, and the disclosure of environmental information is an important way to safeguard the environmental rights of the public [14]. In order to win the blue sky battle and achieve green development, China encourages various forms of environmental information disclosure. The disclosure of environmental data is mainly driven by official efforts, and data collection and disclosure are carried out through official government organizations. Currently, research related to environmental data disclosure mainly includes the evaluation of official environmental information disclosure policies and the exploration of the impact of environmental information disclosure on pollution emissions, green innovation, economic development, and other aspects.

Environmental information disclosure policies were implemented earlier in European and American countries and achieved good results in pollution control; studies have found that environmental information disclosure policies improve environmental performance by disclosing corporate environmental violations and exposing companies to direct penalties and indirect economic losses [9,15]. The low cost of environmental crime in China leads to the failure of environmental information disclosure policy in the financial market [16], but it can enhance residents’ pro-environment behavior [17]. Strong public opinion can make up for policy deficiencies, to a certain extent, and put pressure on local governments to improve environmental quality [18].

Environmental information disclosure brings limited punishment to heavily polluting enterprises, but after a period of time, it can promote the transformation of urban industrial structure and the progress of environmental protection technology, and significantly reduce urban pollutant emissions. The emissions reduction effect is sustainable and will be enhanced with the enhancement of regional pollution laws and environmental regulation [14]. Environmental information disclosure can improve public awareness of environmental conditions and scientificize intuition. For example, the public can easily obtain accurate air quality information through applications such as Mobile Weather. Open environmental information can improve the public’s environmental participation and the government’s environmental law enforcement efforts to ultimately reduce urban pollution emissions, and for non-resource-dependent cities, this emissions reduction effect is more significant [19]. The opening of environmental information not only reduces the emissions of local pollutants, but also has a spatial radiation effect, which helps to improve the ability of surrounding areas to clean up pollutants [11]. Pan et al., 2022 also confirmed that environmental information openness promotes the formation of environmental governance by the government, enterprises, and the public, increases the government’s environmental concern, stimulates the public’s motivation to monitor environmental protection, promotes enterprise restructuring, and reduces regional pollution emissions. Moreover, the effect of emissions reduction is more obvious in regions where officials have stronger awareness of avoiding environmental punishment and less financial pressure [20].

Environmental information disclosure can effectively alleviate environmental information asymmetry between central and local governments [21], change officials’ motivation for environmental protection, and urge local officials to carry out substantive environmental governance; local governments’ motivation for environmental governance will eventually increase corporate pressure for environmental protection, prompting firms to significantly increase their investment in environmental protection. This effect is stronger for firms whose executives have experience in public office [22]. Environmental information disclosure motivates local governments to enhance enterprises’ green innovation motivation through political pressure channels and law enforcement channels, and the effect is long term [23]. Enterprises’ enthusiasm for environmental innovation will eventually improve the level of urban green innovation, and the higher the level of information disclosure, the more obvious the effect of urban green innovation [24]. Enterprises with a stronger ecological concept are also more willing to publish their own environmental information, and the voluntary disclosure of environmental information has a positive impact on enterprise innovation investment and can enhance the technological innovation ability of enterprises [25]. After the disclosure of environmental information, the enthusiasm of green innovation in industries with high environmental risks is significantly improved, especially in non-patent-intensive enterprises and state-owned enterprises, and the synergistic effect of public opinion and environmental law enforcement can further strengthen the green innovation effect of environmental information disclosure. The disclosure of environmental information also provides more environmental protection materials for new media, a free media environment encourages heavily polluting enterprises to actively innovate green technology to meet the requirements of stakeholders, and pollution charges and penalties brought by regulation will also force enterprises to improve green technology innovation [26].

Environmental information disclosure can also significantly improve the urban total factor energy efficiency by improving the regional research and development levels and innovation ability, promoting regional energy conservation and emissions reduction technology levels, and replacing polluting industries with clean industries [27]. Improved energy efficiency can lead to increased green economic efficiency in cities, with each unit increase in government environmental information disclosure increasing the green total factor productivity by 20%, and the effect is nonlinear, with the increase being greater as the level of environmental information disclosure increases [28]. The disclosure of China’s environmental information will also significantly affect foreign direct investment, promote the exit of polluting foreign-invested enterprises that have entered the market, and increase the investment of clean enterprises, leading to an ecological transformation of the foreign investment structure [29]. Environmental information disclosure can also play the role of resource allocation of green finance, stimulate green orientation in the capital market, and improve green investment [30]. Most of the above literature treats environmental information disclosure as an event and analyzes the social impact of environmental information disclosure by conducting quasi-natural experiments. This method is able to measure the impact of environmental data disclosure on pollutant emissions reduction and green innovation in a broad direction but is prone to some biases in accurately estimating the relationship between the level of environmental data disclosure and air pollution, green innovation, etc. [20].

The essence of environmental information disclosure is the openness of environmental data, which is one of the results of digitalization construction. Current research on the impact of digitization and intelligence on environmental information openness and environmental pollution is also relatively abundant. Shi et al., 2018 empirically tested whether smart cities reduce urban environmental pollution by changing urban development patterns through modern information technology [31]. Technology optimization and industrial upgrading brought about by digital transformation significantly reduced the power consumption intensity, thus reducing energy consumption and pollution [32]. Xiao et al., 2023 believe that digitalization can affect enterprise carbon performance through green technology innovation, and this relationship is a U-shaped curve [33]. It can be seen that digitalization has an important impact on environmental pollution, although the impact of digitalization and intelligence on the environment is broader and deeper than the impact of data disclosure on the environment. However, environmental data disclosure and digitalization development have a close relationship. Digitalization-related research can still be used as an indirect reference for the study of the relationship between environmental data disclosure and pollution emissions.

After the implementation of the Environmental Information Disclosure Measures (Trial) in 2008, the Institute of Public and Environmental Affairs (IPE) and the Natural Resources Defense Council (NRDC) jointly published the Report on the Disclosure of Regulatory Information on Pollution Sources in Key Cities, which is a scientific evaluation of the status of the disclosure of regulatory information on pollution sources in key cities, starting in 2008 and continuing annually until 2018 with good continuity. After the Ministry of Environment and Ecology of the People’s Republic of China issued a new method for environmental information disclosure in 2013, the IPE and NRDC, after repeated discussions and argumentation, decided to revise the calculation standard of the Pollution Information Transparency Index (PITI), which has been in use for 4 years, and formed the PITI under the new standard. The new standard is more complete, reasonable, feasible, and scientific, so PITI can effectively measure the level of environmental data disclosure at the city level. This paper carries out a quantitative study of the level of environmental data disclosure on pollution emissions using six indexes from 2013 to 2018, and carries out an in-depth discussion of the impact mechanism to improve the existing research on environmental information disclosure on urban pollution emissions.

2.2. Research Hypothesis

In the era of digital economy, data and information play an important role in the strategic adjustment, production, and operation of enterprises. The disclosure of environmental data increases the cost of environmental violations for enterprises and will have an impact on enterprise productivity and pollution weight, thus affecting regional pollutant emissions.

Referring to Kovak et al. [34], the urban production function is set as the Cobb–Douglas production function:

$$Q=A{L}^{\alpha }{K}^{\beta }$$

(1)

Q represents the output; L and K, respectively, represent the input of labor and capital; A is the comprehensive technical level; α is the elastic coefficient of labor output; and β is the elastic coefficient of capital output. Referring to Bai et al. [35], the production function driven by digitalization is $Q=A\left(E\right)L^{\alpha }{K}^{\beta }$, $A\left(d\right)=A_0+{\theta }_{A}E$, A₀ is the initial technical level, θ_A denotes the propensity for technological progress due to data openness, θ_A > 0, and E is the level of environmental data disclosure. The resulting production function incorporating data openness is shown in Equation (2):

$$Q=(A_0+{\theta }_{A}E)L^{\alpha }{K}^{\beta }$$

(2)

By-products will inevitably be produced along with the desired products and, referring to Stokey and Sheng [36,37], the pollution function is written as $W=zQ$; z is the pollution factor per unit output. Digital infrastructure, information level, etc., are the basis for open data application, and these prerequisite foundations will likewise improve enterprise productivity. Drawing on the approach of Bai et al. [35], pollution weight is linked to the level of data openness, $z=z_0+{\theta }_{z}E$, z₀ is the initial pollution value before the opening of environmental data, and θ_z is the change tendency of urban pollution weight caused by the opening of data. The application of enterprise data can lead to the improvement in enterprise production efficiency, to the adoption of the existing literature hypothesis [35], and to θ_z > 0 when the enterprise production efficiency increases.

In a perfectly competitive market, the product price at competitive equilibrium is p, the fixed cost is C, the product cost is the unit variable cost b, the initial pollution cost is C_w0, and the propensity to change the marginal abatement cost is θ_w. Theoretically, when the level of environmental data disclosure exceeds a certain threshold, the total cost of urban pollution increases sharply due to the increased regulatory capacity and the limitation of pollution treatment level; the cost of pollution treatment is C_w = C_w0 + θ_w × W. Through the above setting, the total target profit function and constraint conditions of enterprises are obtained as follows:

$$\begin{gathered} \max \pi =Q\left(p-b\right)-C-C_W \\ st\left\{\begin{array}{l} Q=(A_0+{\theta }_{A}E)L^{\alpha }{K}^{\beta }\\ C_W=C_{W0}+{\theta }_{W}W, {\theta }_W>0\\ W=\left(z_0+{\theta }_{z}E\right)Q, {\theta }_z>0 \\ p>b>0, \alpha >0, \beta <1 \end{array}\right. \end{gathered}$$

(3)

Pollution emissions under optimal selection are obtained:

$$W=\frac{{\theta }_A{L}^{\alpha }{K}^{\beta }}{ {\theta }_W{\theta }_z}\left(p-b-z{\theta }_W\right)z$$

(4)

The first-order and second-order derivatives of Equation (4) can be obtained as shown in (5) and (6), respectively:

$$\frac{dW}{dE}=\frac{\left(p-b-2z{\theta }_W\right){\theta }_A{L}^{\alpha }{K}^{\beta }}{ {\theta }_W}$$

(5)

$$\frac{d^{2}W}{d{E}^2}=-2{\theta }_z{\theta }_A{L}^{\alpha }{K}^{\beta }<0$$

(6)

Before the opening level of environmental data reaches a critical value, the overall quality of data is not high, which renders it difficult to substantially improve the government and public’s ability to regulate enterprises’ environmental violations. On the contrary, it may induce enterprises’ speculative psychology. Before the opening level of environmental data can substantively affect enterprises’ fines for violating the law, enterprises will try their best to use the time window to expand production, resulting in an increase in pollution emissions. The mathematical representation is that Equation (6) is less than 0. Pollution emissions increase with the level of environmental data disclosure when E is small. When the level of environmental data disclosure reaches the threshold, environmental data can greatly enhance the ability of the public and the government to supervise the emissions of enterprises, and the cost of illegal emissions of enterprises increases sharply. At this time, pollutant emissions will decrease with the continuous improvement in the level of environmental data disclosure.

When the level of environmental data disclosure is low, the public cannot accurately grasp the current environmental situation, and the awareness of environmental protection participation will be weak. The ambiguity of environmental information also makes it more difficult for the government to enforce environmental laws and weaken the environmental supervision by higher government to lower government, resulting in a weaker overall environmental regulation in society. With the development of digitalization and intelligence, the level of environmental data disclosure will continue to improve, and the ability of environmental supervision will be strengthened. Therefore, when the level of environmental data disclosure is low, enterprises will expand their production in order to obtain more profits, which makes urban pollutant emissions increase. When the level of environmental openness is high and the cost of environmental violations increases, enterprises’ emissions behavior will gradually be restrained, and at this time, the early environmental protection investment also begins to have an effect, and urban pollutant emissions will show a downward trend. Thus, Hypothesis 1 is proposed.

Hypothesis 1.

There is an inverted U-shaped curve relationship between environmental data disclosure and urban pollutant emissions.

Pollutants are unavoidable by-products of the production process, but the use of green technologies to improve resource utilization efficiency can reduce pollutant emissions per unit of product [38], mitigate the environmental hazards caused by economic activities, and enhance the environmental and economic benefits of enterprises [39]. Green technology innovation has become a key tool to reduce pollutant emissions. A high level of environmental data disclosure alleviates the environmental information gap between the central and local governments, enterprises, and the public [24]; facilitates the environmental supervision of the higher-level governments and the public; and increases the environmental pollution costs of the lower-level governments and enterprises. In order to cope with increasingly severe environmental constraints, enterprises have been motivated to enhance green technology innovation. Porter’s hypothesis also supports this conclusion from the perspective of environmental regulation. In order to encourage green development, the government will increase investment in scientific research [40] and provide incentives, such as subsidies, to enterprises that pay attention to ecological protection, actively carry out green technology innovation, and reduce pollutant emissions. Subsidies can have a significant positive impact on enterprises’ green innovation by easing financial constraints [41]. Under the “carrot and stick” incentive policy, high-quality environmental data disclosure will inevitably force high-polluting enterprises to actively carry out green technology innovation to reduce energy consumption and pollution to cope with more stringent and precise environmental regulations in the future. However, when the level of environmental data disclosure is low, it is difficult for environmental regulation to drive enterprises to carry out innovation activities. Enterprises often choose to bear the cost of environmental violations and the pressure of environmental violations “crowding out” the ability and willingness of enterprises to carry out innovation [42]. At this time, external stakeholders cannot supervise enterprises’ green innovation activities through public low-quality environmental data, and the agency costs faced by green innovation are aggravated, which may hinder enterprises’ green innovation [43].

Nowadays, society is more concerned about environmental issues, and social public opinion and social financial support are more inclined to favor green and environmental protection enterprises. As a medium, environmental data disclosure can change the financing constraints of enterprises by transmitting the urban environment information, environmental management, and environmental awareness of local enterprises to lenders and creditors. With a low level of environmental data disclosure, it is difficult to convey effective information, which enhances the probability of green financial mismatch, distorts the financing cost constraint of green innovation, and reduces the level of green innovation of enterprises; a high level of environmental information disclosure reduces the risk concerns of lending institutions and the financing cost of environmental protection enterprises. As a result, environmental protection enterprises will be more likely to receive green financial support and more resource rationing, and will possess increased funds for green innovation [44]. Green innovation will eventually become an important driver of clean economic growth and reduced pollution emissions [45]; hence, Hypothesis 2 was produced.

Hypothesis 2.

The level of environmental data disclosure influences urban pollution emissions through green innovation.

Environmental pollution can inhibit the inflow of regional talent and promote the outflow of population and labor [46]. Lower levels of environmental data disclosure cannot effectively improve the government’s environmental monitoring capacity, and low levels of environmental regulation and poor environmental quality can increase public depression and reduce the public’s life satisfaction [47], contributing to the outflow of urban migrants; this effect is more pronounced for well-educated people [48]. An increased level of environmental data disclosure will enhance the sense of urgency of green development in cities, and cities will gradually retire high-energy-consuming and high-polluting enterprises and introduce clean industries that are mostly knowledge- and technology-intensive. Thus, urban industrial restructuring brings changes in the composition of urban human capital. More employment opportunities and a better livable environment will attract more senior talents, which will continuously improve the overall cultural literacy of the city and enhance the overall social responsibility and environmental awareness of the city [49]. The improvement in social participation awareness and environmental protection concepts will enhance public expectations of urban environmental quality, making urban pollution emissions face greater constraints. Local governments will increase environmental supervision to meet public environmental demands, and public supervision also provides a public basis for the government’s precise environmental governance. The cooperation between the public and the government will increase the cost of environmental violations by enterprises and force them to reduce pollutant emissions as much as possible. At the same time, talent-gathering facilitates enterprises’ quicker acquisition of suitable R&D and management talents, improvement in enterprise operation efficiency, and reduction in enterprise costs. The talent aggregation effect satisfies the diverse talent needs of enterprises and cities, improves the green transformation ability of cities and enterprises, and is conducive to reducing urban pollutant emissions from the source [50]. Therefore, Hypothesis 3 is proposed.

Hypothesis 3.

The level of environmental data disclosure affects urban pollutant emissions by changing the regional talent aggregation effect.

Environmental data disclosure increases urban environmental constraints, and in the era of digital economy, the phenomena of urban industrial restructuring and industrial agglomeration are intensifying. The environmental pressure brought by environmental data disclosure motivates cities to use various digital technologies to amplify the positive effects of industrial agglomeration and improve resource utilization efficiency as much as possible [51,52]. Many traditional heavy-industry cities with unreasonable industrial structures continue to optimize industrial layouts, strengthen industrial collaborative agglomeration, increase urban clean investment, promote the exit of pollution-oriented enterprises from the market [29], and improve the level of green technology innovation through the introduction and development of high-tech industry and a knowledge-intensive service industry to enhance the urban green total factor productivity [53]. Therefore, although urban environmental pollution will be aggravated due to concentrated pollution emissions and low green production efficiency at the beginning of the industrial adjustment and industrial agglomeration [54], the improvement in urban green total factor productivity in the later period is conducive to reducing urban pollution emissions. Therefore, Hypothesis 4 is proposed.

Hypothesis 4.

Environmental data opening level reduces urban pollutant emissions by improving regional green total factor productivity.

3. Methods and Materials

3.1. Data and Samples

This paper focuses on the impact of the level of environmental data disclosure on urban pollutant emissions. Sulfur dioxide, as a major by-product of industrial production activities, is a major component of urban pollution emissions and is the main cause of the formation of pollutants such as acid fog and haze, so sulfur dioxide emissions can objectively represent urban pollution emissions; urban pollution emissions as a dependent variable are represented by UPE, which is derived from the China City Statistical Yearbook. At present, there is no unified standard for measuring the openness level of urban environmental data. Urban environmental data are mainly used to reflect urban pollution and other information. In 2013, the IPE and International NRDC jointly released the new standard of PITI, which is a continuous, systematic, and scientific evaluation of the pollution source information disclosure status of 120 cities in China. To a certain extent, these data have reached the international stage and can objectively reflect the urban environmental data disclosure situation. Therefore, this index was used to represent the level of environmental data disclosure in cities. A higher value of PITI indicates a higher level of environmental data disclosure. Referring to the existing literature, we selected the control variables of urban urbanization level (UrbLev), energy structure (EnStr), population size (PopSize), consumption level (ConLev), and greening coverage rate (GreCov). The urbanization level is measured by the proportion of urban population to the total population, the energy structure is measured by the ratio of coal consumption to total energy consumption, the population size is expressed by the logarithm of the total population of the city at the end of the year, the consumption level is measured by the logarithm of total retail sales of social consumer goods, and the greening coverage rate is drawn directly from the China City Statistical Yearbook.

The above data were obtained from the China City Statistical Yearbook, the China Stock Market and Accounting Research Database (CSMAR), and the 120 Cities Pollution Source Regulatory Information Disclosure Index Report in previous years. The time span used was 2013–2018, involving 120 cities, and some missing data were filled by the linear interpolation method. Descriptive statistics are shown in

Table 1. Descriptive statistics.

Variables	Obs	Mean	SD	Min	Median	Max
UPE	720	1.52	0.727	0.08	1.53	3.92
ENVData	720	47.35	15.913	8.30	47.50	82.40
UrbLev	720	57.69	9.233	37.83	56.79	89.60
EnStr	720	0.40	0.124	0.02	0.40	0.67
PopSize	720	6.07	0.701	3.40	6.19	8.13
ConLev	720	16.27	0.994	13.19	16.26	18.66
GreCov	720	41.37	5.177	0.39	41.40	92.87

.

3.2. Econometric Model

In order to enhance the accuracy of the model estimation and correct or eliminate the adverse effects caused by the heteroscedasticity of cities and periods as much as possible, we used the panel fixed-effect model to evaluate the impact of environmental data disclosure on urban pollution emissions based on the results of the robust Hausman test. The following econometric model was constructed:

$$UP{E}_{it}={\alpha }_0+{\alpha }_{1}ENVDat{a}_{it}+{\alpha }_{2}ENVDat{a}_{it}^2+{\alpha }_{n}Con{t}_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(7)

In Equation (7), i denotes the city, t denotes the year, the dependent variable UPE_it represents urban pollution emissions and is mainly measured using the logarithm of urban sulfur dioxide emissions, the independent variable ENVData_it represents environmental data disclosure as measured by the PITI, Cont_it is the control variable, δ_i and λ_t denote the individual and time effects, respectively, and ε_it is a random error term.

4. Results and Discussion

4.1. The Effect of Environmental Information Disclosure on Urban Pollutant Emissions

Table 2. Benchmark regression results.

	(1)	(2)	(3)	(4)
Variables	UPE	UPE	UPE	UPE
ENVData	−0.6142	0.0978	2.0509 **	1.9002 **
	(−0.6829)	(0.1193)	(2.3459)	(2.2417)
ENVData2	−0.0280 ***	−0.0202 **	−0.0199 **	−0.0180 **
	(−2.8812)	(−2.2905)	(−2.2495)	(−2.1043)
UrbLev		−2.0952 *		0.0093
		(−1.9098)		(0.0025)
EnStr		376.7925 ***		−20.5851
		(6.4178)		(−0.2308)
PopSize		85.9520 ***		−53.0806
		(5.1670)		(−0.5956)
ConLev		−43.7999 ***		4.5584
		(−3.3700)		(0.3533)
GreCov		0.3607		0.6565
		(0.5698)		(1.3407)
_cons	1.1 × 103 ***	1.2 × 103 ***	1.1 × 103 ***	1.3 × 103 **
	(53.5509)	(8.7504)	(59.2825)	(2.0201)
City FE	NO	NO	YES	YES
Year FE	NO	NO	YES	YES
Obs	720	720	720	720
R-squared			0.719	0.718

Notes: The t-statistic is in parenthesis. *, **, and *** denote statistical significance at the 10%, 5%, and 1% significance levels, respectively.

reports the results of baseline regressions. Columns (1) and (2) of Table 2 do not control the year and city fixed effect and only the quadratic coefficients are significant. Columns (3) controls the year and city fixed effects, and the coefficients of the primary and quadratic terms of explanatory variables are significant. In Column (4), control variables are added on the basis of Column (3), and the significance remains the same. The inflection point of the explanatory variable increases from 51.53 to 52.78, indicating that the control variables better absorb the influence of other factors on urban pollution emissions. The inflection point is calculated by the regression coefficient of Equation (7) through the equation $-{\alpha }_1/\left(2{\alpha }_2\right)$. In Column (4), the coefficient of the primary term is significantly positive, at a significance level of 5%, while the coefficient of the secondary term is significantly negative, indicating that the opening level of urban environmental data and urban pollution emissions form an inverted U-shaped curve. From this, hypothesis 1 is verified.

There are several reasons for this result. When the level of urban environmental data disclosure is low, the government and the public cannot accurately grasp the situation of urban pollution, and the cost of pollutant discharge of enterprises is low, so they will have an opportunistic mentality and use the time window to expand production and obtain more profits as much as possible, resulting in the increase in urban pollution emissions. When the level of environmental data disclosure is high, supervision is easier, the cost of pollutant emissions continues to increase, and stronger constraints force polluters to gradually reduce pollutant emissions.

4.2. Robustness Test

Benchmark regression proves the relationship between the dependent variable and the independent variable. To further verify the robustness of these results, alternative indicators of the pollutant emission type and open data standard are used to re-estimate the benchmark model. Smoke dust and nitrogen oxides are also common air pollutants, and the industrial wastewater associated with the production process is the main component of liquid pollution discharge in cities. Therefore, the discharges of smoke dust (lnDust), nitrogen oxides (lnNO_X), and industrial wastewater (lnWastewater) are used to replace the dissolved variables; the regression results are shown in Columns (1), (2), and (3) in

Table 3. Robustness test.

	(1)	(2)	(3)	(4)
Variables	lnDust	lnNOX	lnWastewater	UPE
ENVData	2.4214 **	1.8312 *	0.5256 **
	(2.2906)	(1.8070)	(2.2298)
ENVData2	−0.0287 ***	−0.0192 *	−0.0056 **
	(−2.8708)	(−1.8775)	(−2.2295)
DigitalE				19.8070 *
				(1.7281)
DigitalE2				−1.8628 *
				(−1.6629)
_cons	668.5363	−9.1 × 102	109.8557	1.5 × 103 **
	(1.0123)	(−1.1908)	(0.7558)	(2.3300)
Controls	YES	YES	YES	YES
City FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Obs	720	720	720	720
R-squared	0.411	0.253	0.343	0.717

Notes: The t-statistic is in parenthesis. *, **, and *** denote statistical significance at the 10%, 5%, and 1% significance levels, respectively.

. The coefficient of the primary term is significantly positive, while the coefficient of the secondary term is significantly negative, indicating an inverted U-shaped curve relationship between the open level of urban environmental data and pollution discharge.

The disclosure of urban environmental data is closely related to the level of urban digital development. The development of the urban digital economy is better, and the data openness is generally timelier and more comprehensive. Therefore, the dependent variable is replaced by the development level of urban digital economy (DigitalE), and the development level of the digital economy is calculated by the principal component method. The result of re-regression is Column (4) in Table 3, which is consistent with the result of basic regression.

4.3. Endogeneity Test

The instrumental variable method is used to alleviate endogeneity issues. The instrumental variables used in this article are the installation density of industrial robots (InstallCsmd) at the city level and the number of fixed telephones per 100 people (AveTelephone) in cities in history. The calculation of the installation density of industrial robots at the city level refers to the existing literature [55,56], matches the data of the International Federation of Robots (IFR) with the data of the second economic census of China, obtains the installation data of industrial robots in each industry, and calculates the weight of the robot density of each industry in each China city, using 2008 as the benchmark. Based on this, the robot installation density of each city over the years is further calculated. The reason for the use of historical urban per-capita telephone data is that China’s Internet popularity started with fixed telephones. Cities with a high telephone penetration rate in history generally have a higher level of information openness at present, and historical telephone data have little impact on current urban pollution emissions. Therefore, the historical fixed telephone data from 2003 to 2008 are used as the instrumental variable. Industrial robot installation density is chosen because regions with higher robot usage density generally have a higher level of digitization and of open urban environmental data, and because it is difficult for urban pollution emissions to have an inverse effect on robot usage density; therefore, these two types of variables are used as instrumental variables.

Table 4. Results of endogeneity test.

	First-Stage	Second-Stage
Variables	ENVData	UPE
ENVData		58.5007 ***
		(17.4134)
ENVData2		−0.5713 ***
		(0.1959)
InstallCsmd	0.6712 **
	(0.3215)
AveTelephone	5.4130 **
	(2.1420)
_cons	−81.3472	70.2560
	(9.7173)	(604.3464)
Controls	YES	YES
Obs	702	702
City FE and Year FE	YES	YES
F	102.12

Notes: The t-statistic is in parenthesis. **, and *** denote statistical significance at the 5%, and 1% significance levels, respectively.

reports the results of the 2SLS regression method. The first-stage regression results’ instrumental variables are significantly positively correlated with the level of environmental data disclosure, indicating that instrumental variables have good explanatory power for the dependent variable. The test of weak instrumental variables shows that the F-statistic is far greater than 10, indicating that there are no weak instrumental variables. The results of the second-stage regression show that the primary term and quadratic form are significant at the 1% level. According to the coefficients of the primary term and quadratic form, the environmental data disclosure value at the turning point is 51.19, which is within the range of the environmental data disclosure level of the sample [8.3, 82.4], proving that there is a significant inverted U-shaped curve relationship between environmental data disclosure and pollution emissions.

4.4. Heterogeneity Analysis

As a soft constraint to control urban pollutant emissions, environmental data disclosure has a significant U-shaped effect on urban pollution. China has a vast territory, and the stages and modes of urban development are quite different from one another, so whether there is heterogeneity is also worth discussing. Therefore, heterogeneity is discussed from three aspects: urban location, urban vitality, and urban citizen characteristics.

Regarding regional heterogeneity, in eastern China, cities with rapid development generally took early measures to control the environment. Before the release of environmental data, relatively strict pollution emissions standards were adopted. For example, heavy industries with high energy consumption and high pollution were moved out of the core cities and clean enterprises were introduced. Most cities in the Midwest are resource-dependent, with slower urban development, a higher share of heavy industry, more total pollutant emissions, and more room for adjustment, making it easier for environmental data disclosure to quickly have an impact on pollutant emissions in the Midwest. Columns (1) and (2) of

Table 5. Regression results of the heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Variables	UPE	UPE	UPE	UPE	UPE	UPE	UPE	UPE
ENVData	1.7477	3.8457 ***	2.3516 *	0.3843	2.1161 **	1.9597	2.1436 **	1.9661
	(1.5518)	(3.0692)	(1.9359)	(0.2853)	(2.1332)	(1.2485)	(2.0035)	(1.3835)
ENVData2	−0.0150	−0.0405 ***	−0.0204 *	−0.0018	−0.0195 **	−0.0183	−0.0202 *	−0.0195
	(−1.4790)	(−2.8419)	(−1.7623)	(−0.1362)	(−2.0012)	(−1.0663)	(−1.9070)	(−1.2359)
_cons	920.0465	1.3 × 103 **	672.3029	−3.1 × 102	1.5 × 103	1.6 × 103	1.4 × 103	1.6 × 103
	(1.0047)	(2.4969)	(0.7127)	(−0.1272)	(1.4082)	(1.6658)	(1.1925)	(1.6498)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES	YES
Obs	318	402	372	348	453	267	453	267
R-squared	0.889	0.821	0.751	0.658	0.702	0.734	0.691	0.743

Notes: The t-statistic is in parenthesis. *, **, and *** denote statistical significance at the 10%, 5%, and 1% significance levels, respectively.

show the regression results for the eastern and central-western regions, respectively. The results show that the impact of environmental data disclosure on urban pollution emissions demonstrates the same trend, and both have an inverted U-shaped relationship, but the trend in the central-western region is more significant than that in the eastern region, and there is strong regional heterogeneity.

Regarding the heterogeneity of city dynamics, open and inclusive cities generally have stronger entrepreneurial dynamics and entrepreneurship brings about the flow of talent and technology, creating good conditions for green urban development. Under the guidance of mainstream social values, new companies generally choose green development concepts and adopt green technologies, so the impact of environmental data opening on urban pollution emissions may be different in cities with different entrepreneurial vitalities. Cities are divided into high-vitality cities and low-vitality cities according to the median of urban entrepreneurial activity. The index of urban entrepreneurial activity is measured by the annual number of start-ups per 100 people in the city, which is obtained by using the Qichacha Database. The regression results of high-activity cities are shown in Column (3) of Table 5, and the regression results of low-activity cities are shown in Column (4) of Table 5. High-vitality cities are more significant than low-vitality cities, on the one hand, because high-vitality cities have a faster flow of talents and technology, which provides convenience for urban industrial transformation and green development, and, on the other hand, are more significant because high-vitality cities have a higher proportion of middle-aged and young people, a higher per-capita literacy level, and stronger environmental protection concepts, so environmental data disclosure can significantly affect urban decision making and corporate environmental costs, and have a significant impact on urban pollution emissions.

Regarding the heterogeneity of public environmental participation, the higher the public’s environmental attention, the more stringent the requirements on environmental quality and the more likely the public is to urge the government to strengthen environmental supervision through complaints and other means to force enterprises to make green transformation; thus, we ask, is there heterogeneity in the impact of urban environmental data disclosure on pollution emissions under different levels of public environmental concern? Using the number of searches for environmental-pollution- and haze-related terms by the urban public on the internet to measure public environmental involvement, cities with a median number of searches above the median are classified as cities with high public environmental involvement, and those with a median number below the median are classified as cities with low involvement. In Table 5, Columns (5) and (6) show the regression results of searching environmental pollution; (7) and (8) are the regression results of searching haze-related terms. Columns (5) and (7) show the regression results of high public environmental participation with significant regression results, while Columns (6) and (8) indicate that the effect of environmental data opening on urban pollution emissions is not significant under low public environmental participation. This heterogeneity is in line with common sense and consistent with expectations.

4.5. Mechanism Analysis

To further investigate the mechanism of the effect of environmental data disclosure on urban pollution emissions, we refer to the existing literature [57,58] to construct the mediating effect models (8) and (9) and make arguments for the transmission mechanism from the channels of green innovation, talent aggregation, and green production.

$$M_{it}={\alpha }_0+{\alpha }_{1}ENVDat{a}_{it}+{\alpha }_{2}ENVDat{a}_{it}^2+{\alpha }_{n}Con{t}_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(8)

$$UP{E}_{it}={\beta }_0+{\beta }_1{M}_{it}+{\beta }_{2}ENVDat{a}_{it}+{\beta }_{3}ENVDat{a}_{it}^2+{\beta }_{n}Con{t}_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it }$$

(9)

M_it is the mechanism variable in Equations (8) and (9). The improvement in supervision ability brought by accurate and detailed environmental data will directly increase the environmental illegal cost of polluters and facilitate the public’s environmental supervision. Under a series of soft and hard constraints, enterprises, as major pollutant emitters, have a strong desire to reduce pollution emissions, carry out green technology innovation, and adopt technology as the main way to reduce pollutant generation and emissions. The number of green patents for inventions (GrePatInv) and green patents for utility models (GrePatMod) applied for by the city in that year are used to measure the overall level of green innovation in the city. Columns (1) and (3) of

Table 6. Regression results of mediation test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Variables	GrePatInv	UPE	GrePatMod	UPE	TalAggreg	UPE	TechExp	UPE	GreenTFP	UPE
ENVData	−61.6642 ***	1.1419	−48.6923 ***	1.1133	−0.0003 **	1.5591 *	−6.9 × 103 **	1.6805 **	−0.0020 *	1.6983 *
	(−3.1921)	(1.3765)	(−4.9209)	(1.3727)	(−2.3443)	(1.9205)	(−2.1541)	(2.0121)	(−1.8157)	(1.9738)
ENVData2	0.5903 ***	−0.0108	0.5183 ***	−0.0097	0.0001 **	−0.0143 *	65.0595 **	−0.0160 *	0.0001 *	−0.0158 *
	(3.0796)	(−1.2759)	(4.8493)	(−1.1594)	(2.5975)	(−1.7511)	(2.0336)	(−1.8849)	(1.9081)	(−1.8249)
GrePatInv		−0.0123 ***
		(−4.4862)
GrePatMod				−0.0162 ***
				(−2.9063)
TalAggreg						−1.1 × 103 ***
						(−2.7806)
TechExp								−0.0001 *
								(−1.6621)
GreenTFP										−1.0 × 102 **
										(−2.5213)
_cons	−4.1 × 104 **	789.2931	−3.0 × 104 **	805.6035	−0.1610	1.1 × 103 *	−1.2 × 107	921.1252	2.1439 ***	1.5 × 103 **
	(−2.2295)	(1.2169)	(−2.2636)	(1.1909)	(−1.5616)	(1.7160)	(−1.4833)	(1.3265)	(4.7105)	(2.2819)
Controls	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Obs	720	720	720	720	720	720	720	720	720	720
R-squared	0.430	0.729	0.561	0.725	0.142	0.723	0.274	0.723	0.172	0.723

Notes: The t-statistic is in parenthesis. *, **, and *** denote statistical significance at the 10%, 5%, and 1% significance levels, respectively.

show the results of the first-stage regression, indicating that the opening of urban environmental data beyond the critical value can significantly promote green technology innovation at the 1% level, and Columns (2) and (4) of the second stage regression show that the mediating variables are significantly negative, proving that the enhanced ability to apply green technology innovation can significantly reduce pollutant emissions. The independent variables are not significant, indicating a full mediating effect.

In recent years, environmental issues have been given more and more attention in China, and environmental control has become increasingly strict. In order to improve urban environmental quality under strict data supervision, local governments will promote urban industrial upgrading, increasingly welcome the investment and expansion of green environmental protection enterprises, and tend to transform and expel enterprises with high energy consumption and high pollution. Low-pollution enterprises are mostly knowledge-intensive or high-tech enterprises, whose development requires more senior intellectuals and that attract many talented employees to their cities. Talent aggregation will increase the overall cultural level of the city, improve public participation in environmental protection, and increase the pressure of local governments and polluting enterprises on environmental protection. On the other hand, talent aggregation provides abundant human resources for the green transformation of cities and enterprises, facilitating their rapid strategic adjustment, which ultimately affects urban pollution emissions. However, the low level of environmental data disclosure, weak environmental regulation, and poor urban environment will drive talent away, and the talent leaving will intensify urban pollution emissions [49]. By referring to the common practice of the academic circle, the proportion of the number of employees in scientific research, technical service, and the geological survey industry, and the sum of the number of employees in information transmission, computer service, and the software industry in the total urban employees in the statistical yearbook are used to measure the level of urban talent aggregation. The knowledge and technical level of employees in these industries are higher, and the yearbook data are more complete. It is feasible and reasonable to use these data to represent the level of talent aggregation in cities [59]. The regression results are shown in Columns (5) and (6), where the first stage shows an inverted U-shaped effect of urban environmental data opening on talent aggregation (TalAggreg), and the second stage shows that talent aggregation significantly reduces urban pollution emissions at the 1% level and that the mechanism is proven.

Green production is represented by urban science and technology expenditure and green total factor productivity. The opening of environmental data alleviates the information asymmetry between the upper and lower levels of government and increases the pressure on local governments to protect the environment. While urging enterprises to develop green, local governments will increase scientific research expenditure to promote technology upgrading and improve urban green total factor productivity. The regression results of the model with urban science and technology expenditure (TechExp) as the mechanism variable are shown in Columns (7) and (8), and the regression results of the model with urban green total factor productivity (GreenTFP) as the mechanism variable are shown in Columns (9) and (10). The regression results indicate that when the level of environmental data disclosure is low, it cannot have a positive and substantial impact on local government behavior, while the level of openness reaches a certain level that can substantially increase local government environmental protection pressure, significantly promote local government scientific research expenditure, and significantly improve urban green total factor productivity.

5. Conclusions

Environmental data disclosure is the main way to guarantee citizens’ right to know about the environment and the important information support for citizens’ participation in environmental supervision. Environmental data are also the main basis for higher-level governments to assess the quality of lower-level governments’ environmental supervision. This paper uses the data of 120 cities in China from 2013 to 2018 to study the impact of environmental data disclosure on urban pollution emissions. The research results show that there is a significant inverted U-shaped relationship between the level of environmental data disclosure and urban pollutant emissions. When the level of environmental data disclosure is low, the ability of urban environmental supervision is limited, enterprises will try their best to use the time window to expand production, and pollutant emissions will increase together with production; but when the level of environmental data disclosure exceeds the critical value, environmental regulatory capacity is substantially improved, talent aggregation increases, early green technology investment begins to be effective, pollution prevention capacity is improved, and the high level of environmental data disclosure significantly reduces urban pollutant emissions. Heterogeneity analysis found that cities with higher vitality and a stronger public awareness of environmental protection had a more significant impact on urban pollutant emissions, but in cities with less heavy industry, lower urban vitality, and a weaker public awareness of environmental concerns, the relationship was not clear. Mechanism analysis found that environmental data disclosure mainly influences urban pollutant emissions by promoting government research investment and enhancing the overall level of green technology innovation in cities, gathering senior talents, and improving urban green total factor productivity pathways. This study counters the traditional notion that environmental data disclosure necessarily reduces urban pollution emissions. Further, the study provides an important reference for controlling urban pollution emissions and promoting green and sustainable urban development. According to the above research results, the following suggestions are put forward.

First, digital technology should be used to improve the scope and quality of environmental data disclosure. The development of digital technology in the digital age has provided great convenience for data collection, cleaning, and publishing, making the content of publicly available environmental data more accurate; thus, the quality of data is higher. Therefore, the advantages of advanced digital technologies such as the Internet of Things, cloud computing, and artificial intelligence should be further exploited to realize real-time reporting and error correction of environmental data and improve data quality and the timeliness of publication.

Second, environmental accountability mechanisms should be strengthened. The key to environmental governance is accountability; without accountability, environmental governance is impossible. Data openness leads to more accurate and efficient environmental accountability. The Report to the 20th National Congress pointed out that “Tightening resource and environmental constraints and environmental pollution were pronounced”, “we will prioritize ecological protection”, and that “ecological conservation and environmental protection remain a formidable task”. Therefore, it is necessary to strengthen the environmental accountability mechanism, raise the fines for violating environmental protection laws, urge local governments to pay attention to environmental problems, and force enterprises to not risk polluting.

Third, innovation in green technologies should be encouraged to achieve the more efficient transfer of green research results. At present, there are many achievements in green technology research and development, but the transformation of achievements is slow and the application of green technology is infrequent. Therefore, it is necessary to further promote the organic integration of industry, university, and research, accelerate the transformation of scientific research results while promoting the scientific research innovation of universities and institutes, promote the orderly and rapid upgrading of production equipment, reduce the generation of pollutants, reduce pollution from the source, and promote the green development of the city.

This paper has certainly contributed to filling the academic gap regarding the relationship between environmental data disclosure and urban pollution emissions. There are still some limitations that warrant further research. First, the study is limited by the lack of adequate data. The currently selected sample only contains 120 cities, spanning 2013–2018, leaving open the questions of whether the impact of environmental data disclosure on urban pollution emissions is related to the administrative level of cities and whether there is a different impact of environmental data disclosure on urban pollution emissions over a longer period of time. With the update of research data in the future, more in-depth discussion can be carried out. Second, the current research is mainly based on Chinese cities, and whether the impact of environmental data disclosure on urban pollution emission is related to national cultural habits is also worth discussing. In the future, quantitative studies and case studies can be carried out by introducing cities in Western countries, which will help to reveal the deep internal mechanism of the impact of environmental data disclosure on urban pollution emissions.