The Government’s Environmental Attention and the Sustainability of Environmental Protection Expenditure: Evidence from China

Abstract

Transformation to a green economy depends on long-term, continuous government financial investments in the environmental protection field. This study used the budget and final accounts of provincial environmental protection expenditure in China from 2007 to 2019 to establish a two-way fixed-effects model to verify the impact of government attention on environmental protection expenditure. The sustainability of Chinese local governments’ environmental protection expenditure is weak, and the fluctuations conform to the punctuated equilibrium characteristics. Fluctuations in government attention have a significant positive impact on changes in fiscal environmental protection expenditure, and the characteristics of campaign governance are relatively obvious. Fiscal transparency can strengthen the relationship between government attention and the environmental protection expenditure budget; however, the adjustment effect on the final accounts of environmental protection expenditure, which depend more on the budget implementation process, is not significant. To improve the effectiveness of environmental governance, government departments must pay long-term and sustained attention to environmental issues, improve the transparency of the implementation process of environmental protection expenditure budgets, strengthen the role of public supervision, and ensure investment in environmental governance.

1. Introduction

With China’s structural economic transformation, the traditional extensive economy has gradually transformed into a “high-quality development” economy that pursues quality and efficiency. Environmental governance, which is necessary to establish a green and low-carbon circular economic development system, has become increasingly important. In the 2021 Chinese Government Annual Work Report, the frequency of environment-related words was 55. The priority of environmental governance on the government agenda is constantly advancing. The central government pays substantial attention to energy savings, emission reduction, environmental protection, and other related fields [1].

Compared to the central government’s focus on environmental issues, local governments in China pursue short-term, steady, and fast goals. Only when major environmental pollution accidents and events occur and strategic policies are implemented with unconventional administrative coercion will local governments control and shut down all types of enterprises to reduce pollutant emissions [2,3]. In the early stages, the lack of authority and the effectiveness of local environmental protection departments prompted the central government to conduct comprehensive and central environmental protection inspections [4]. Campaign-style governance breaks conventional institutionalized governance, allocates resources through political mobilization, and concentrates all forces to complete a specific task; however, the sustainability of the effect cannot be guaranteed in the short term. In the context of inconsistent adjustments, China’s environmental governance has shown a long-term abnormal and non-institutionalized pattern. The intensity of environmental regulation and investment in environmental governance was largely determined by local governments in different periods. Public financing plays an important role in environmental governance. Investment in environmental protection investment has direct and indirect effects on environmental governance [5], which can not only directly act on energy savings, emission reduction, and pollution control but also have an important effect on social capital flow toward environmental protection. As a major systematic project, environmental governance requires long-term, sustained, and stable financial investments for China’s economic transformation.

Government attention provides a unique perspective on the sustainability of environmental governance in China and helps explain long-term campaign-style environmental governance. The basic view of government attention is based on the assumption of finite rationality, which states that the attention of decision makers in any organization is a scarce resource. In the era of abundant information, what is truly scarce is attention rather than information [6,7]. Policymakers must sort competing issues because simultaneously focusing on all public affairs is impossible. As a limited economic resource, fiscal budget allocation is a direct monetary measure of government investment. Its structure reflects the change in government priorities, which is a comprehensive embodiment of the changes in the attention of decision makers [8].

This study focused on the sustainability of China’s environmental protection expenditure and used the annual budget and final accounts of provincial environmental protection expenditure from 2007 to 2019 to answer the following questions from the perspective of government attention:

• Is the Chinese provincial government’s environmental protection expenditure sustainable?
• Do fluctuations in government attention affect fluctuations in environmental protection expenditure?
• Does the relationship between fluctuations in government attention and environmental protection expenditures in provincial governments have heterogeneity?

2. Literature Review and Research Hypothesis

2.1. Government Attention

Herbert Simon first introduced the concept of attention to the field of management to explain irrational decision-making behavior based on traditional economic theory with basic assumptions of consumer preferences [9]. Simon believes that decision makers’ attention is scarcer than abundant external information and that actual decisions are highly influenced by how decision makers allocate their attention to external information [7]. Ocasio proposed the firm’s attention-based view, whereas Baumgartner and Jones first introduced attention to the field of public policy and budget decision making, arguing for the limited attention paid to policy decisions and the infinity of external information from the punctuated-equilibrium mode of policy and budget change [6,10].

Government attention refers to the level of attention the government pays to a particular field. As the government’s resources and attention are limited, a higher level of government attention in a specific field implies that the government will allocate more resources and policy support to that field, prioritizing its development [11,12,13].

The main method for measuring government attention is text analysis, which is mainly based on many policy documents and public information depicting the attention distribution levels of different policy subjects in different fields using the word frequency statistical method. Mortensen, based on US hearings before Congress documents of the Policy Agendas Projects, studied the relationship between congressional attention changes and federal budget allocation changes [14]. Jennings and John encoded the policy agendas mentioned in speeches delivered by the Queen before Parliament from 1996 to 2001, measuring the government’s attention in different policy areas [15]. Sagarzazu and Klueve used quantitative text analysis of over 40,000 press releases to construct an issue concern dataset that examined the dynamics of coalition party issue concerns [16]. Owing to government attention to the field of environmental protection, Shi et al. analyzed statements on environmental protection in some 600 government work reports from 30 provinces in China between 1998 and 2017 and examined how the government’s attention actually shifted [17].

The government’s environmental attention measurement was based on word frequency statistics from the analysis of government work reports. Such indices are more susceptible to subjective factors and lack reliability and validity. Furthermore, attention measures based on government work reports are prone to show “planning effects”, which cause errors in the measurement of attention in each year of the beginning of the “Five-Year Plan” due to intensive policy information surges. Based on previous studies, this study used the per 10,000 population of environmental protection proposals from the Provincial People’s Congress (PPC) adopted by each province each year to represent the government’s attention to the field of environmental protection. This index is objective and can effectively eliminate the interference of various subjective factors and facilitate longitudinal or horizontal comparative analyses.

2.2. Budget Expenditure Sustainability

Punctuated-equilibrium theory provides theoretical support to verify the sustainability of environmental protection spending. According to Jones and Baumgartner, data on budget expenditure variation following the punctuated-equilibrium pattern show an approximately characteristic leptokurtic or power-law distribution [8]. In contrast, in the ordinary budget progressive model, the difference among budget years approximately follows a normal distribution with a zero mean, whereas in the growth budget progressive model, the difference among budget years approximately follows a normal distribution with a positive mean [18].

The punctuated equilibrium in budgets has much empirical evidence from democratic countries, and testing budget data consistent with punctuated-equilibrium patterns has become popular in academic circles [19,20,21,22,23,24]. Breuning and Koski studied 18-year budget data from 10 budget categories and found that US state budgets mostly followed a punctuated-equilibrium pattern [25]. Breuning used the random process method to test the national budgets of Denmark, Germany, the United Kingdom, and the United States from the mid-1960s to 1989. The results showed that the budget changes in the four countries follow a punctuated-equilibrium model [26]. Baumgartner and Foucault found that the French national budget during 1820–2002, 7 sectors during 1868–2002, and 10 sectors during 1947–2002 all fit the leptokurtic distribution of the punctuated-equilibrium pattern [27]. Martin and Streams examined government spending on global health in 17 OECD countries during 1990–2009 using punctuated-equilibrium theory and found that their annual changes were subject to a leptokurtic distribution [28]. Sebők and Berki examined year-over-year changes in final account expenditure in Hungary from 1991–2013, showing that it was consistent with the punctuated-equilibrium pattern, and expanded the sample to 2580 expenditure categories during 1868–2013 [29,30].

Some studies have also indicated that under China’s special authoritarian system, the punctuated equilibrium has different characteristics [31]. Yao et al. used Chinese final accounts from 31 provinces during 1996–2018 and found that economic construction spending and science, education, culture, and health (SECH) spending were more discontinuous, whereas public security spending showed more progressive characteristics [32]. Lam and Chan used Hong Kong data to verify that the centralization process aggravates the instability of government attention and the discontinuity of policy output [33].

The adaptive issue of the punctuated-equilibrium model in budgets remains to be tested by different government departments. Following the steps of general punctuated equilibrium empirical research and combining the characteristics of China’s campaign-style governance in the environment, this study first examined the sustainability of local government environmental protection expenditures and proposed hypothesis H1:

H1: 

Government environmental protection expenditures conform to the punctuated-equilibrium model.

The punctuated-equilibrium theory proposed by Baumgartner and Jones provides a complete theoretical link between “attention → policy” behavior and insight into the sustainability of environmental governance investment [10]. The core view of punctuated-equilibrium theory is that policy behavior is in two states, long-term stagnant stability and occasional change, the reason for which is the disproportionate information processing model in policy decision-making systems [18]. Political agendas are driven by the import of information into decision-making agencies to acquire external attention. The government receives extensive external information from different sources integrated into implicit indicators. The information weights of the different domains are relatively balanced and independent of each other under ideal conditions, and the output of the policy decision systems after summing the information conforms to a normal distribution. However, unlike an ideal parallel processing method, finite rational decision agents can only perform serial processing. Governments must prioritize matters and obtain new information by constantly updating their attention. This mechanism of renewal of attention among different types of information ultimately leads to policy outputs that manifest as a leptokurtic distribution [18].

Due to limited government attention and fiscal resources, when government attention shifts, the change in the policy agenda drives the structural adjustment of fiscal expenditure and tilts the limited fiscal resources to areas with higher attention. The more violent the fluctuation in government attention, the stronger the fluctuations in the corresponding fiscal expenditure. Therefore, H2 was proposed as follows:

H2: 

The fluctuation in government attention to environmental protection leads to the fluctuation in environmental protection expenditures.

3. Methodology

3.1. Model Specification

This study used the following model to verify the relationship between fluctuations in government attention and environmental protection expenditure:

$${CRE}_{it+1}={\beta }_0+{\beta }_1{dpersug}_{it}+\alpha X_{it}+{\delta }_i+{\tau }_t+{\epsilon }_{it}$$

(1)

CRE presents the rate of change in environmental annual expenditure, $dpersug$ is the annual change in the number of the PPC’s proposals adopted by the government in the environmental area, $X_{it }$ is the control variable, ${\delta }_i$ is the provincial fixed effect, ${\tau }_t$ is the time fixed effect, and ${\epsilon }_{it}$ is a random error term.

3.2. Variables and Data Resource

3.2.1. Variables

In 2007, China adjusted its classification of fiscal revenue and expenditure and built a new government revenue and expenditure classification system. To ensure consistency of environmental protection expenditure data for each year, this study used the budget and final accounts data of 30 provinces from 2007 to 2019, excluding Hong Kong, Macao, Taiwan, and the Tibet Autonomous Region due to a lack of data.

The dependent variable ${CRE}_{it}$ in Equation (1), based on Jones and Baumgartner and Li et al. [18,34], is computed as follows:

$${CRE}_{it}=\frac{({EP}_{it}-{EP}_{it-1})}{{EP}_{it-1}}$$

(2)

${EP}_{it}$ is the proportion of provincial environmental protection expenditure to public budget expenditure. As China’s fiscal system adopts a calendar year system, the fiscal budgeting work for the next year is conducted in the current year, earlier than the PPC sessions, and the dependent variable, CRE, should be pushed forward for the first period.

The core independent variable $dpersug$ in Equation (1) represents the fluctuations in government attention. The centralized up-bottom management mode of Chinese government departments and the stability and various appeals from the bottom to the top of the public led to the coexistence of stability and policy fluctuations [35]. One reason for changes in environmental protection expenditure is that the government attention caused by public appeal fluctuates. Due to the finite rationality of government processing information and the traditional tournament promotion system, the government’s attention to environmental protection fluctuates annually. This study uses the per 10,000 population of environmental protection proposals adopted by each province each year to measure the government’s attention to the field of environmental protection, $persug$. For example, in 2009, the population of Anhui Province was 61.31 million and the number of environmental protection proposals in the province was 86. Therefore, the government’s attention to the field of environmental protection in Anhui Province in 2009 is calculated as 86/6131 = 0.014. In China, the number of delegates in the PPC is strictly based on the population of each province. Therefore, the total number of proposals from provincial delegates each year is closely related to the population of each province. Proposals from provincial delegates refer to constructive opinions on relevant government work, which can further develop into various policies after being adopted by the government. A higher $persug$ value indicates that the government prefers to adopt proposals in the field of environmental protection, reflecting a higher level of government attention to the environmental protection field. Furthermore, $dpersug$ measures annual changes in the government’s attention to environmental issues.

$${dpersug}_{it}=\frac{\left({persug}_{it}-{persug}_{it-1}\right)}{{persug}_{it-1}}$$

(3)

Based on existing research [32,36], the control variables are fiscal decentralization (FDecen), fiscal pressure (FPre), economic industrialization (EIndust), fiscal transparency (FTrans), provincial urbanization (PUrban), per capita income (PInc), population density (PDen), and internet development (Int).

The fiscal decentralization is computed as FDecen = per capita provincial financial expenditure/per capita national financial expenditure [37]. Fiscal pressure is general provincial public budget expenditure minus public budget income. Fiscal pressure reflects the constraints on financial fund allocation. The higher the fiscal pressure, the more intense the competition for financial funds among expenditures [38]. Economic industrialization is expressed as the ratio of the secondary industry GDP to the provincial GDP, which is used to measure the degree of dependence of each provincial economy on industrial industries. Provincial fiscal transparency data are from the China Fiscal Transparency Report released by the Shanghai University of Finance and Economics. Provincial urbanization is expressed as the proportion of the urban population to the total population in each province. Internet development is measured based on the number of internet access ports in each province.

3.2.2. Data Source

The panel data in this study cover 30 provincial regions in China, considering the availability of sample data, and the time span was set at 10 years, from 2009 to 2018. The data were obtained mainly from the China Environmental Yearbook, China Finance Yearbook, and China Statistical Yearbook.

3.3. Descriptive Statistics

Table 1. Descriptive statistics.

Variable Name	Obs.	Mean	Std. Dev.	Min.	Max.
${CRE}_1$	300	0.004	0.168	−0.365	0.761
${CRE}_2$	299	0.018	0.187	−0.362	0.897
dpersug	300	0.08	0.45	−0.89	2.86
FDecen	300	0.999	0.419	0.536	2.366
FPre	300	7.33	0.664	5.476	8.665
EIndust	300	0.464	0.086	0.186	0.594
FTrans	300	3.422	0.438	2.639	4.353
PUrban	300	0.559	0.128	0.299	0.896
PInc	300	5.451	1.276	2.042	8.25
PDen	300	10.133	0.346	9.387	11.128
Int	300	6.892	1.023	3.492	9.006

presents the descriptive statistics. The number of observations was 300, and the annual change in the final accounts removed outliers greater than 2; thus, the number of observations was 299. ${CRE}_1$ represents the annual change in the government budget for environmental protection. ${CRE}_2$ represents the annual change in the government’s final environmental protection expenditure accounts.

4. Empirical Results

4.1. Sustainability of Environmental Protection Expenditure

According to public budget decision-making theory, if environmental protection expenditure fluctuates according to gradual budget theory, CRE will be normally distributed by various random factors; if environmental protection expenditure fluctuates and conforms to the punctuated-equilibrium model, CRE will be leptokurtic and have a fat-tail distribution [18]. The distribution characteristics of the CRE were tested using histograms and normality tests.

The descriptive statistics and normality test results for the rate of change in environmental protection expenditures from 2008 to 2019 are shown in

Table 2. CRE normal distribution test.

CRE	Obs.	Mean	Std. Dev.	Kurtosis	Skewness	Shapiro–Wilk W	Shapiro–Francia W’
CRE1	360	0.03	0.19	5.23	1.20	0.0000 ***	0.0000 ***
CRE2	359	0.04	0.21	6.34	1.37	0.0000 ***	0.0000 ***

*** p < 0.01. Data source: The budget and final accounts for 2007–2018 came from the China Finance Yearbook. The budget and final accounts for 2019 were obtained from the government websites of each province.

. It can be observed that ${CRE}_2$ had a larger standard deviation than ${CRE}_1$. Stata 15 software was used to perform the Shapiro–Wilk and Shapiro–Francia W tests. The results showed that the normality test for the changes in both the budget and the final accounts failed, and the change in environmental protection expenditure in each province did not conform to the characteristics of the gradual distribution. Changes in environmental protection expenditure are drawn as histograms (Figure 1). The provincial environmental protection budget and final accounts in China from 2008 to 2019 showed leptokurtic and fat-tail distribution characteristics. The discontinuity of the final environmental protection accounts was higher than the environmental protection budget, further verifying that the distribution of environmental protection expenditures in each province conforms to the characteristics of punctuated equilibrium changes, which supports H1.

4.2. Benchmark Regression Results

According to

Table 3. Benchmark regression results.

	(1)	(2)	(3)	(4)	(5)	(6)
	M1	M2	M3	M4	M5	M6
	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$
dpersug	0.0499 *	0.0563 **	0.0531 **	0.0589 **	0.0578 **	0.0636 **
	(0.0246)	(0.0264)	(0.0238)	(0.0252)	(0.0253)	(0.0281)
FDecen			0.502 ***	0.404 **	0.385 ***	0.383 *
			(0.101)	(0.166)	(0.137)	(0.191)
FPre			−0.310 ***	−0.146 *	−0.217 *	−0.144 *
			(0.0798)	(0.0857)	(0.117)	(0.0770)
FTrans					−0.0245	0.0232
					(0.0316)	(0.0390)
EIndust					−0.127	0.438
					(0.344)	(0.404)
PUrban					1.095	−1.295
					(0.816)	(0.964)
PDen					−0.560	−0.902 *
					(0.376)	(0.456)
PInc					−0.0763	0.157
					(0.238)	(0.290)
Int					0.0600	0.125
					(0.0855)	(0.0894)
Province	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−0.0154	0.0507	1.585 ***	0.645	4.081	3.648
	(0.0343)	(0.0422)	(0.510)	(0.551)	(3.148)	(2.880)
Obs. 1	296	295	296	295	296	295
R-squared	0.105	0.154	0.151	0.171	0.171	0.190

*** p < 0.01, ** p < 0.05, * p < 0.1. ¹ Observations exclude five outliers of $dpersug$.

, the regression coefficient of $dpersug$ was positive, passing the 5% significance test for both M5 and M6. For each unit of government attention fluctuation in environmental protection, ${CRE}_1$ increased by 0.0578 and ${CRE}_2$ increased by 0.0636. ${CRE}_2$ levels were slightly higher than those of ${CRE}_1$. This is related to the fact that the environmental protection final accounts lag the budget, making them more vulnerable to fluctuations in government attention. Hypothesis H2 holds that the higher the government’s attention to environmental protection, the higher its environmental protection expenditure volatility. Therefore, financial investments in local government environmental governance are vulnerable to changes in government attention. When the government’s attention to environmental protection decreases, the scale of financial investment in environmental governance decreases. The more intensely the government attention drops, the more serious is the corresponding reduction in the financial investment scale.

4.3. Robustness Test

To verify the reliability of the model, this study performed the robustness test in three ways, that is, removing some sample data from Beijing, Shanghai, Tianjin, and Chongqing; replacing fiscal expenditure decentralization with the dependence of local governments on the central transfer payments; and adding the “Five-Year Plan” variables to further test the robustness in different political cycles.

Table 4. Results of the robustness test.

	(1)	(2)	(3)	(4)	(5)	(6)
	M7	M8	M9	M10	M11	M12
	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$
dpersug	0.0550 **	0.0573 **	0.0581 **	0.0649 **	0.0578 **	0.0636 **
	(0.0236)	(0.0277)	(0.0263)	(0.0282)	(0.0253)	(0.0281)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Constant	7.302 **	2.737	4.638	4.695	4.081	3.648
	(3.157)	(3.858)	(3.061)	(2.878)	(3.148)	(2.880)
Obs.	256	256	296	295	296	295
R-squared	0.201	0.229	0.162	0.196	0.171	0.190

** p < 0.05.

presents the results of the robustness test. M7 and M8 show the regression results after excluding sample municipalities, and M9 and M10 show the regression results after replacing the data on the decentralization of fiscal expenditures. Therefore, the coefficient sign of government attention fluctuations did not change and remained significant at the 5% level, indicating the model is robust. M11 and M12 are added planning variables mainly used to characterize the five-year planning period in which the observed values lie.

The model sampling time was 2009–2018, across the 11th, 12th, and 13th Five-Year Plan periods; to test the robustness of different political cycles, the corresponding planning variables were set to 1 (11th Five-Year Plan period), 2 (12th Five-Year Plan period), and 3 (13th Five-Year Plan period). The sign of the government attention fluctuation coefficient was consistent with that of the original regression model and passed the robustness test.

4.4. Moderating Effect Test of Government Fiscal Characteristic Variables

When government attention fluctuations affect changes in environmental protection expenditure or are affected by budget management and financial characteristics, it may be heterogeneous. For example, increased fiscal pressure increases the cost of government budget allocation decisions [39], fiscal decentralization reflects the discretion of local and central governments to use and adjust financial resources [40], and the improvement of fiscal transparency is conducive to strengthening hard budget constraints and consistently maintaining central and local environmental protection attention [41].

This study selected three typical variables characteristic of financial markets, fiscal decentralization, fiscal pressure, and fiscal transparency, to explore whether there was a moderating effect. The results of the calculation are listed in

Table 5. Moderating effect test regression results.

	(1)	(2)	(3)	(4)	(5)	(6)
	M13	M14	M15	M16	M17	M18
	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$	${CRE}_1$	${CRE}_2$
dpersug	0.0564 **	0.0601 **	0.0569 **	0.0551 **	0.0561 **	0.0626 *
	(0.0249)	(0.0277)	(0.0264)	(0.0259)	(0.0254)	(0.0309)
FDecen	0.385 ***	0.387 **	0.384 ***	0.383 **	0.407 ***	0.391 *
	(0.137)	(0.175)	(0.137)	(0.174)	(0.136)	(0.193)
FPre	−0.213 *	−0.136	−0.217 *	−0.146 *	−0.231 *	−0.150 *
	(0.122)	(0.0801)	(0.118)	(0.0788)	(0.114)	(0.0758)
FTrans	−0.0247	0.0226	−0.0244	0.0246	−0.0217	0.0250
	(0.0317)	(0.0392)	(0.0314)	(0.0393)	(0.0313)	(0.0389)
dpersug × FDecen	0.0289	0.0713
	(0.0546)	(0.0623)
dpersug × Fpre			−0.00648	−0.0614
			(0.0497)	(0.0495)
dpersug × Ftrans					0.0911 **	0.0557
					(0.0441)	(0.0522)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Constant	4.283	4.148	4.138	4.196	4.122	3.667
	(3.007)	(2.625)	(3.180)	(2.807)	(3.242)	(2.931)
Obs.	296	295	296	295	296	295
R-squared	0.172	0.195	0.171	0.198	0.182	0.194

*** p < 0.01, ** p < 0.05, * p < 0.1.

. M13 and M14 are the interaction models for fluctuations in government attention and fiscal decentralization, respectively. M15 and M16 are the interaction models for fluctuations in government attention and fiscal pressure, respectively. M17 and M18 are the interaction models for fluctuations in government attention and fiscal transparency, respectively. The interaction coefficients of government attention fluctuation and fiscal decentralization were positive but not significant. The interaction coefficients of government attention fluctuation and fiscal pressure were negative but not significant. The interaction coefficients of government attention fluctuation and fiscal transparency were positive, passing the 5% significance test in M17 and failing the significance test in M18. Therefore, fiscal transparency has a significant positive moderating effect on the relationship between government attention fluctuations and changes in environmental protection expenditure budgets. The higher the fiscal transparency, the greater the environmental protection expenditure budget fluctuations caused by fluctuations in government attention. The reasons for the insignificant positive adjustment effect in the relationship between fiscal transparency on $dpersug$ and ${CRE}_2$ may be that changes in environmental final accounts depend more on the budget implementation process and are vulnerable to the compound impact of other potential factors.

5. Conclusions

This study used the budget and final accounts of provincial environmental protection expenditure in China from 2007 to 2019 to test the fluctuation characteristics of environmental protection expenditure and constructed a two-way fixed-effects model to verify the influence of government attention fluctuations on changes in environmental protection expenditure. The main conclusions are as follows.

First, the sustainability of provincial government expenditure on environmental protection is weak and the fluctuations conform to the characteristics of punctuated equilibrium. Both the budget and the final accounts data for environmental protection expenditure support this conclusion. The skewness of final environmental protection accounts data is higher than that of the budget, indicating that the final accounts have higher discontinuity and more instability characteristics. Second, fluctuations in local government attention have a positive and significant impact on changes in China’s environmental protection expenditure. When the government’s attention to the field of environmental protection changes, its financial input in environmental governance also fluctuates, resulting in the failure to form long-term sustainable financial input. These conclusions have passed a series of robustness tests. Third, the impact of government attention fluctuation on the change in the environmental protection expenditure budget is heterogeneous. Fiscal transparency has a significant and positive adjustment effect on fluctuations in government attention and changes in the environmental protection expenditure budget. The fluctuation in government attention in regions with high fiscal transparency has a stronger impact on the change in environmental protection budget expenditure. However, the positive adjustment effect on the final accounts of changes in environmental protection expenditures, which depend more on the budget enforcement process, is not significant.

Therefore, to ensure the sustainability and effectiveness of environmental governance, government departments should focus on environmental protection, monitor the environmental governance investment scale and policy implementation for a long time, and enhance effective long-term attention to environmental protection issues. By ensuring the continuity of environmental protection financial resource input, the rebound of environmental problems caused by campaign-style governance can be avoided. Furthermore, government departments should improve the disclosure of financial information, improve the transparency of the environmental protection expenditure budget implementation process, strengthen public supervision, and ensure long-term sustainability of investment in environmental governance.

6. Prospects

This study proposes that current government attention fluctuations in China will have a significant influence on the changes in environmental protection expenditure. This research can be a useful complement to the study of the punctuated equilibrium characteristics of budget allocation in China and other authoritarian countries. This study also provides methodological references for researching punctuated-equilibrium theory in other expenditure areas in China. However, the generation and transfer mechanism of government attention are not adequately discussed in research. Government information processing capacity is limited. In the internet era of information overload, governments can select and filter external information raised on the policy agenda. Information processing and transformation mechanisms must be further analyzed and discussed. In addition, the environmental protection budget and final accounts data were analyzed as explanatory variables in this study. The regression results showed that the significant influence on core variable changes and the environmental protection expenditure budget is heterogeneous. The internal reasons for this heterogeneity should be explored further in subsequent studies.