Introduction

The high pollution and high energy consumption seriously hinder China's environmental protection and green development1. As a public good, the quality of the environment is directly influenced by the government provision2. However, China has long practised an environmental decentralization system of public finance3, which means that regional environmental protection and pollution control are in the hands of local governments in each region. Some studies have found that local governments will indulge in developing high-polluting industries for rapid economic development4,5. The "race to the bottom" strategy is a reason for the failure of regional pollution control and the excessive low-end production capacity6. Additionally, It has been suggested that governments have the same selfish tendencies as public service providers7. Local governments with economic autonomy tend to neglect public inputs, thus lacking efficiency in improving regional environmental governance capacity and productivity8.

Since 2016, the central government has decided to implement the reform of vertical environmental governance (VEG) at the national level in China. The reform aims to consolidate environment-related governance and fiscal revenues and expenditures at the county and municipal levels to the provincial governments, thereby strengthening environmental centralization and enhancing the role of public finance in ecological governance and low-end capacity reduction. This article focuses on whether the environmental centralization system is more conducive to achieving technological progress and productivity improvement at the enterprise level. This paper measures technological progress and output efficiency by total factor productivity (TFP). The implementation of VEG is adopted as a quasi-natural experiment to compare the differences in enterprise TFP between environmental centralization regions and environmental decentralization regions. First, the result of the benchmark regression shows that enterprises in environmental centralization regions have significantly higher TFP than those in environmental decentralization regions. Moreover, the baseline conclusion still holds after considering robustness issues, including the common trend issue, heterogeneous treatment effects, the methods of TFP measurement, the self-selection problem, and the placebo test. Second, theoretical and empirical analyses of the impact mechanisms indicate that environmental centralization promotes enterprise TFP by increasing the intensity of environmental regulations and the willingness of firm green innovation. Finally, this study explores regional, industry, and enterprise heterogeneity, respectively. At the regional level, the effect of VEG is more significant in eastern regions, non-resource-based regions, and regions with low fiscal burdens. Technology-intensive, manufacturing, and low-energy-consuming industries are more susceptible to environmental centralization at the industry level. At the enterprise level, small and medium-scale enterprises, growth period enterprises, and non-state-owned enterprises are more adept at leveraging fiscal resources to realize enterprise output efficiency improvement.

The remainder of this paper is arranged as follows. Section "Policy background and literature review" provides an overview of the marginal contributions of relevant literature and articles. Section "Research hypotheses" proposes research hypotheses based on theoretical analysis. Section "Research design" is the research design of this article. Section "Empirical results and analysis" is the empirical analysis. Section "Conclusions and policy implications" shows the conclusions and policy implications.

Policy background and literature review

Policy background

After reform and opening up, China attempted gradual fiscal decentralization between 1978 and 1993. To further unleash local economic vitality, the central government implemented a tax-sharing reform in 1994, which became an important beginning of fiscal decentralization in China. The mainstream academic viewpoint believes that the main purpose of the tax-sharing system reform is to standardize the tax structure and management authority of the central and local governments, as well as significantly improve the autonomy of local governments in environmental governance9. However, while the reform of the tax-sharing system has increased the freedom of local governments, it has also triggered issues of local protectionism and resulted in the transfer of pollution emissions between regions10.

Therefore, to further optimize the fiscal system in environmental governance, the central government gradually implemented VEG in 2016. First, in terms of overall environmental governance design, the identity of the central government as the chief designer has not changed. Secondly, VEG has significantly increased the environmental governance authority of provincial governments, including supervisory power, financial allocation power, personnel appointment and removal power, and performance evaluation power. Thirdly, local governments have reduced their autonomy in environmental governance and become law enforcement agencies in environmental governance. It can be observed that VEG has significantly improved the concentration of provincial government environmental governance, and provincial governments have gradually become the main responsible institutions for environmental governance. The list of provinces involved in the reform from 2016 to 2019 was obtained from the websites of provincial governments and reported in Table 1.

Table 1 List of reform provinces.
Full size table

Literature review

Studies on environment-related fiscal regimes

Current research on fiscal regimes in the environment contains three main aspects. One is to focus on the discussion of the optimal fiscal regimes. Some scholars argued that compared to environmental centralization, the advantage of environmental decentralization lies in maximizing local information advantages and significantly optimizing government decision-making and administrative efficiency11,12. Another group of scholars hold that environmental decentralization lacks adequate supervision of local governments13,14. Therefore, environmental centralization can better ensure administrative efficiency. Secondly, some scholars focused on studying the causal relationship between fiscal systems and corporate environmental governance7. Thirdly, some scholars actively explored how fiscal systems affect economic development. The decentralization regime is beneficial in stimulating local governments' investment and promoting rapid regional economic growth15. However, environmental decentralization grants local governments too much financial autonomy9. Low-end industries with low risks, stable returns, and short payback cycles are favored by local governments, while the development of science and technology industries and public services lag behind16. This disadvantage of environmental decentralization is prevalent in developing countries and has become an important cause of ineffective environmental governance and difficulties in improving economic efficiency17.

Studies on TFP

Total factor productivity (TFP) reflects the amount of goods and services that can be created per unit of factor input18. TFP is an indicator of technological progress, economic efficiency, and development quality. The research on TFP mainly focused on two perspectives: indicator measurement and analysis of influencing factors. TFP measurement methods include econometric, index, data envelopment, and random boundary19. Extensive literature has estimated TFP at the regional, industrial, and enterprise levels20,21,22. Enterprise TFP is mainly measured using econometric techniques and parameter estimates23. The available literature explored the factors that influence enterprise TFP at different levels. From the macro perspective, the literature examined the external economic environment that affects enterprise TFP from the viewpoints of environmental regulations24, resource endowment25, and financing conditions26. Some scholars have added research on the impact of subsidies and tax burdens on firm TFP27,28. However, their studies focused on the economic behavior of the governments. Fewer studies have examined the fiscal system, which is the purpose of our focus on environmental centralization reform. At the meso level, the differences in enterprise TFP under different industry types and structures are the research hotspots29. At the micro level, the literature explored the perspectives of enterprises' technological innovation capability30, equity and board structure31, and managerial capability32.

A summary

Existing studies have extensively examined fiscal regimes in the environment and TFP. However, few scholars have examined the correlation between environmental concentration and enterprise TFP. Enterprises are the basic units of the national economy. The relationship between environmental regimes and enterprise TFP has become an essential topic of current research. The active participation of the governments in environmental governance is conducive to motivating enterprises to improve production technologies and management modes, thereby reducing energy consumption and pollution emissions, eventually leading to the upgrading of production capacity33.

Based on the above analysis, this article has four main contributions compared to the existing literature:

First, this study provides micro evidence of the economic effects of environmental centralization by focusing on enterprise TFP. The mechanisms of environmental centralization are explored from the perspective of environmental regulations and firm green innovation. The study provides empirical support for enriching relevant literature and developing more effective programs for fiscal reform. The analyses in this paper also provide an empirical basis for clarifying the relationship between environmental governance regimes and enterprise performance.

Second, this paper comprehensively explores the heterogeneous impact of environmental centralization regime at the macro, meso and micro levels. The analyses provide concrete options for adapting to environmental centralization reform in different regions, at different economic levels, in different industries and for different types of firms.

Thirdly, a staggered difference-in-differences (DID) model is used to compare the differences in TFP between enterprises in environmental centralization regions and environmental decentralization regions. Limited by the availability of economic data, most studies reflect the degree of fiscal decentralization with the help of proxy indicators such as budget expenditure structure and budgetary transparency34,35. Relative to using proxy indicators for fiscal decentralization, the model in this paper is better at avoiding endogeneity problems, thus resulting in more accurate estimated results.

Finally, most studies ignore the issue of heterogeneous treatment effects when applying the staggered DID model36,37. However, some recent literature has demonstrated that heterogeneous treatment effects will lead to biased estimated coefficients of the core explanatory variable38,39. This study also adopts several recently developed alternative estimators that address this issue to enhance the reliability of the results in this paper.

Research hypotheses

VEG centralized the environmental law enforcement and fiscal revenues and expenditures of the prefecture and municipal governments to the provincial governments, thereby enhancing the participation of the governments in regional environmental issues. With government support, reform regions' financial and social resources were more likely to concentrate on environmental governance13,16. These measures encouraged regions and enterprises to expand the research and introduction of green production technologies, which in turn led to the upgrading of regional industries and the transformation of enterprises. Therefore, the article proposes the following hypothesis.

Hypothesis 1 The vertical reform of environmental administrations (VEG) could promote enterprise TFP.

Environmental regulations are the means by which the government utilizes administrative law enforcement to restrict the pollution behaviors of enterprises40. Before VEG, law enforcement of environmental protection was distributed among local governments. Due to insufficient supervision of law enforcement behavior, regional environmental regulations lacked practical effectiveness41. Some local officials even facilitated rent-seeking and evasion of environmental penalties for enterprises9,42. VEG centralized the environmental enforcement power of local governments to the provincial governments, which prevented local governments from taking advantage of information to loosen environmental controls and harbor enterprises. The generally accepted opinion holds that environmental regulations will significantly increase the costs of corporate compliance, thus reducing the amount of capital spent on production and innovation43. However, after Porter's hypothesis was proposed, the academic perspective between environmental regulations and corporate TFP changed44. According to the innovation compensation theory, environmental regulations will raise the market access threshold, thus forcing firms to strengthen production technology research and development45. Moreover, the focus of the governments on environmental regulations is also helpful to the emergence of novel green industries, which motivates enterprises to improve their innovation capacity to meet the needs of industrial development46.

Furthermore, the high cost, high risk and long payback period associated with technological innovation act as significant barriers to enterprises taking the initiative to implement green innovation47. While environmental centralization served to intensify government environmental regulation, it simultaneously serves to reinforce firms' requirements for pollution control and green innovation. The Porter's hypothesis also suggests that, in the event that compliance costs are unavoidable, firms will reinforce green innovation and increase investment in green technology for the purpose of reducing pollution control spending44. Furthermore, environmental centralization served to reduce the likelihood that local governments would shield firms from polluting behaviour48, thus forcing firms to implement green innovations and strengthen pollution control.

Hypothesis 2 VEG promoted enterprise TFP by increasing the intensity of environmental regulations and the willingness of firm green innovation.

Location differences make enterprises face different regional economic development levels, resource endowment conditions, and fiscal burdens. Developed regions have more mature economic and technological conditions, which are more convenient for enterprises to improve their production technology and management capacity49. Resource endowment impacts regional industrial structure and enterprise development patterns50. According to the "resource curse" theory, affluent resource areas are prone to resource dependence, resulting in an industrial structure locked into low-end resource extraction and insufficient incentives for enterprises to transform and upgrade51. From the perspective of fiscal balance, environmental centralization reform means that the governments should assume more environmental responsibilities and thus face more significant fiscal pressure52. Therefore, it is likely that the reform would be more effective in regions with less budgetary pressure. Accordingly, this paper proposes the following hypothesis.

Hypothesis 3a. Enterprises situated in developed areas, areas with less resource endowment, and areas with less budgetary pressure were more affected by VEG.

Capital-concentrated and technology-concentrated industries are more likely to realize an increase in enterprise TFP with the advantage of funds and technology53. The main input of labor-concentrated industries is labor resources, making it more challenging to improve output efficiency54. The "Baumol's disease" theory states that non-manufacturing industries exhibit lagging productivity and slower TFP growth than manufacturing industries55,56. Hence, relative to the non-manufacturing sectors, manufacturing enterprises are more likely to achieve TFP growth. High energy-consuming industries depend more on the traditional energy extraction development model and lack incentives for technological innovation, making it more difficult to achieve TFP growth57. Accordingly, this paper proposes the following hypothesis.

Hypothesis 3b The effect of VEG on enterprise TFP was greater when the enterprises belong to technology-concentrated, capital-concentrated, manufacturing, and low-energy-consuming industries.

Enterprises' business practices and production decisions are influenced by their sizes, capabilities, and property rights. Large and mature firms have sufficient financial resources to develop green technologies58. State-owned enterprises (SOEs) have the advantage of state financial support59. Moreover, such enterprises also have stronger social influence and corporate reputation, which force them to fulfill more environmental responsibilities60. As a result, they are usually in a better position to improve output efficiency. However, such enterprises also have higher market power. From the perspective of maximizing enterprise benefits, these enterprises are more willing to earn excess profits by monopoly advantages and have less incentive for innovation59. To acquire competitive advantage and market share, small and medium-scale enterprises, growth period enterprises, and non-SOEs will be more motivated to realize technological upgrading and improve output efficiency61. Accordingly, this paper proposes the following hypothesis.

Hypothesis 3c. The effect of VEG on enterprise TFP was greater when the enterprises are large-scale enterprises, mature enterprises, and SOEs.

Hypothesis 3d. The effect of VEG on enterprise TFP was greater when the enterprises are small and medium-scale enterprises, growth period enterprises, and non-SOEs.

Research design

Model settings

This article leverages the staggered implementation of VEG to compare the changes in enterprise TFP between the treatment group and the control group from 2016 to 2022. Enterprises in regions where the reform is implemented are categorized as a sample of the treatment group with an environmental centralization regime. Enterprises in the remaining areas are classified as a control group sample with an environmental decentralization regime. An identification strategy of staggered DID is adopted to estimate the model as follows:

$$\begin{array}{c}{TFP}_{it}=\alpha +\beta {Reform}_{it}+\gamma {Control}_{it}+{\lambda }_{i}+{\mu }_{t}+{\varepsilon }_{it},\end{array}$$
(1)

where \({TFP}_{it}\) denotes total factor productivity (TFP), \(i\) and \(t\) respectively represent the enterprise and year. The core explanatory variable is \({Reform}_{it}\). \({Control}_{it}\) is the control variable that may affect enterprise TFP. The fixed effects of enterprises and years are represented by \({\lambda }_{i}\) and \({\mu }_{t}\). \({\varepsilon }_{it}\) indicates the random error term.

Variables and data

Variables and definitions

The explanatory variables of this article include enterprise TFP measured by several measurement methods: ordinary least squares (OLS), two-way fixed effects (TWFE), Olley-Pays (OP), and Levinsohn-Petrin (LP)23,62,63. Among these, the LP method adopts enterprise intermediate inputs as proxy variables, effectively avoiding sample loss and simultaneity deviation23. This article mainly uses the LP method for empirical analysis, and the other measurement methods are employed for the robustness test. The model of enterprise TFP measurement is set as follows:

$$\begin{array}{c}{\text{ln}Y}_{it}={\alpha }_{1}{\text{ln}L}_{it}+{\alpha }_{2}{\text{ln}I}_{it}+{\alpha }_{3}{\text{ln}K}_{it}+{\lambda }_{i}+{\mu }_{t}+{\epsilon }_{it},\end{array}$$
(2)

where \({Y}_{it}\), \({I}_{it}\), and \({K}_{it}\) represent the main business income, intermediate inputs, and fixed assets, respectively. To eliminate the influence of price level, the above three variables were adjusted using the industrial product, producer, and fixed asset price indexes based on 2012, respectively. \({L}_{it}\) is the number of enterprise employees. The logarithmic value of the residual \({\epsilon }_{it}\) is the enterprise TFP we want.

Core explanatory variable:\({Reform}_{it}\) indicates whether the enterprise \(i\) is intervened by VEG in year \(t\). If the province to which enterprise \(i\) belongs introduces VEG in year \(t\), then \({Reform}_{it}=1\); otherwise, \({Reform}_{it}=0\).

Following the relevant research64,65, this study mainly considers control variables that may affect enterprise output efficiency, including enterprise size, measured by total assets; enterprise age; enterprise operating performance, measured by return on equity; enterprise investment value, measured by Tobin’s Q ratio; and enterprise liquidity, measured by total liability ratio. Variable definitions are given in Table 2.

Table 2 Variable definitions.
Full size table

Data and descriptions

Based on the data of Chinese A-share listed firms from 2012 to 2022, this article provides a detailed analysis of the impact of VEG on enterprise TFP. Referring to Wang et al.21, this paper takes the enterprise registration location as the classification criterion. If located in provinces with environmental centralization reform, enterprises are classified into the treatment group, otherwise into the control group. Enterprise data is obtained from the CSMAR database. If not stated, all other regional data in the article are sourced from the China Statistical Yearbook. To improve the accuracy of sample data analysis, we exclude ST enterprises, *ST enterprises, PT enterprises, and delisted enterprises during the sample period, as these enterprises have significant problems with their operating performance. Moreover, this paper also excludes firms with abnormal values and enterprises listed after 2012. We finally obtained a sample of 1999 enterprises and 21,510 observations. Descriptive statistics of variables are shown in Table 3.

Table 3 Descriptive statistics.
Full size table

Empirical results and analysis

Benchmark regression

The empirical results of the impact of VEG on enterprise TFP are shown in Table 4. Column (1) shows that the regression coefficient is not significant with only year fixed effects considered. In column (2), control variables are added for analysis, and the result shows that the coefficient is significantly positive at the 5% level. From columns (3) and (4), it can be concluded that after controlling for fixed effects in provinces and enterprises, respectively, the empirical results are significantly positive at the 1% level. Therefore, the basic regression results show that VEG could improve firm TFP. Since the VEG is conducted at the province level and this paper focuses on the TFP performance of firms, the TWFE model in column (4) is adopted as the preferred model, and the regressions are clustered at the province level.

Table 4 Benchmark regression.
Full size table

Robustness test

Event study figures

Parallel trends and heterogeneous treatment effects are the keys to determining unbiased estimated coefficients of the staggered DID model. This section explores these two questions based on event study figures. The parallel trend hypothesis is first tested based on the TWFE model. The model is set as follows:

$$\begin{array}{c}{TFP}_{it}=\alpha +\sum_{s=-5}^{3}{\beta }_{s}{Reform}_{it}^{s}+\gamma {Control}_{it}+{\lambda }_{i}+{\mu }_{t}+{\upvarepsilon }_{it},\end{array}$$
(3)

where \({Reform}_{it}^{s}\) is a set of dummy variables relative to the "event" of implementing VEG. Suppose \({k}_{i}\) is the year of the introduction of VEG in the province of firm \(i\), \({Reform}_{it}^{s}\) represents whether \(t-{k}_{i}=s\) in year \(t\). To ensure the estimation is realizable, we set the previous year of VEG as the base period.

However, some studies of staggered DID have shown that the TWFE estimator will use the already-intervened samples as control samples for the newly-intervened samples, thus making the parallel trend test implausible38,39. Therefore, we simultaneously plot the event study figure based on several robust estimators66,67,68,69. These estimators yield unbiased estimates by avoiding the erroneous use of the treated group sample as a control group.

Figure 1 presents the estimators' results and shows no significant difference between the control and treated groups before the intervention of VEG. Therefore, the parallel trend hypothesis holds whether heterogeneous treatment effects are considered. Moreover, the average treatment effects exhibit a clear increasing trend over time in the post-reform periods. This result has two possible reasons: (1) The intervention effects increased over time at the firm level. As a result, the average treatment effect is lower in the current intervention period than in the three postperiods. (2) The intervention effects increased over time at the national level. For instance, in our data sample, the intervention effect of the current period is a four-year average from 2016 to 2019, and thus the treatment effect is smaller. The effect of the third postperiod is the average of 2019. Hence, the treatment effect is larger. However, the identification strategy based on event study figures cannot demonstrate from which this effect comes. In Sect. “Effects for the length of exposure to VEG”, this paper will explore this issue by leveraging a new identification strategy that estimates the average treatment effects based on the length of exposure to VEG.

Figure 1
figure 1

Event study figures.

Full size image

Effects for the length of exposure to VEG

This section further explores the reason for the increase in average treatment effects over time. The estimated model is constructed as follows:

$$\begin{array}{c}{TFP}_{it}=\alpha +\sum {\beta }_{l}{Reform}_{it}^{l}+\gamma {Control}_{it}+{\lambda }_{i}+{\mu }_{t}+{\upvarepsilon }_{it},\end{array}$$
(4)

where \({Reform}_{it}^{l}\) is a set of dummy variables relative to the "event" of implementing VEG. \({k}_{i}\) still indicates the year of the introduction of VEG in the province of firm \(i\). Under the condition that \({Reform}_{it}^{s}=1\), if \(2019-{k}_{i}=l\), then \({Reform}_{it}^{l}=1\). For instance, Chongqing implemented the reform in 2016, so \({Reform}_{it}^{3}=1\) for the year 2016 and later.

Figure 2 presents the regression coefficients and shows that the average effects increase over treatment length. This result suggests that the improvement in intervention effects comes from the duration of the intervention at the firm level. To better illustrate this, we assume that the intensity of the reform increases over time at the national level. Figure 2 shows that the estimated result is not significant for regions that only implemented VEG in 2019. Therefore, this implies that any period should be insignificant. However, the opposite is true, suggesting that the reform effects only increase based on the length of exposure to VEG.

Figure 2
figure 2

Effect on enterprise TFP by the length of exposure to VEG.

Full size image

PSM-DID test

Differences in enterprise nature may affect the accuracy of the results. This section uses the propensity score matching (PSM) method to rematch the treatment group and control group, thus making the two sample groups as similar as possible before the reform is implemented21. The method of 1:1 neighbor matching is adopted to exclude endogeneity problems stemming from self-selection bias, thereby enhancing the robustness of our results in the benchmark regressions. Table 5 reports the estimated results and shows that the effect of VEG is further improved compared to our preferred model. Therefore, the findings in this paper can rule out the self-selection problem and remain robust.

Table 5 PSM-DID.
Full size table

Alternative TFP measurement methods

To mitigate the concern of cherry-picking TFP measurement methods, this study also considers several other commonly used econometric models, including Olley-pakes (OP), ordinary least squares (OLS), and TWFE methods. Table 6 reports the estimated coefficients of the three measurement methods and shows little difference from the LP method in the preferred model. Therefore, the empirical results have not significantly changed after adopting different TFP measurement methods.

Table 6 Alternative TFP measurement methods.
Full size table

Placebo test

To eliminating the interference of unobservable variables in the reform effect, this section performs a placebo test by randomly assigning the treatment and control groups. Specifically, this study generates pseudo-reform variables through 500 repeated samples and estimates them in place of the original explanatory variable in the baseline regression. Figure 3 shows the placebo test results. This paper finds that the estimated coefficients of the pseudo-reform variables are centrally distributed in the plus and minus 0.02 range and significantly dissimilar to the value estimated in the preferred model.

Figure 3
figure 3

Placebo test.

Full size image

Therefore, the results above support Hypothesis 1.

Mechanism tests

Based on the above research results, the article further analyses the impact mechanisms of VEG on enterprise TFP. According to Hypothesis 2, this paper selects the two variables of government environmental penalties and firm green patents to test the mediating effect. Referring to Chen et al.70, this study adopts a two-stage model. First, we examine whether the core explanatory variable directly affects the mechanism variables. Second, we regress firm TFP on the intermediary variables.

Environmental regulations

The implementation of environmental regulations can reflect the government's concern for environmental issues71. This article uses government environmental penalties (EP) to measure environmental regulations. The results are shown in Table 7. As can be seen from column (1), VEG has significantly increased the intensity of regional environmental regulations. This indicates that the effectiveness of government oversight is more pronounced in environmental centralization areas, while governments in environmental decentralization regions are more inclined to set lower environmental regulation thresholds. The results in column (2) show that the estimate of EP is significant, and the coefficient is positive, which shows that environmental regulations have promoted enterprise TFP. This result has been confirmed by studies based on Porter's hypothesis45,72, implying that administrative supervision is favorable in forcing enterprises to strengthen the application of green technology and improve production management.

Table 7 Mechanism tests.
Full size table

Green patents

Traditional industries are highly dependent on an extensive development approach. The improvement of green innovation capacity is an important condition for enterprises to strengthen environmental governance. In this study, the number of green patent applications is employed as an indicator for measuring the willingness of enterprises to engage in green innovation. Based on column (3) of Table 7, it can be found that VEG has significantly increased the number of corporate green patent applications. The finding lends further support to Porter's hypothesis that government environmental oversight will encourage firms to pursue environmentally-related innovations. However, Column (4) of Table 7 shows that the empirical results of green patents are not significant. This may be attributed to the time lag inherent in the conversion of patent technology. It is challenging for patents to be transformed into effective technology investments within a relatively short timeframe.

Generally, the above results support Hypothesis 2.

Heterogeneity tests

The analyses above focus only on the average treatment effects of environmental centralization reform. This section further explores the heterogeneity of reform effects.

Region heterogeneity

From the economic development level perspective, China is characterized by obvious regional agglomeration. Specifically, the economic development in the eastern and central regions is significantly better than that in the western regions73. Panel A of Table 8 reports the sub-group estimates for the three regions and shows that the reform effect is only observed in the eastern and central regions. Less developed areas are more inclined to pursue higher economic growth rates. Compared to the eastern and central regions, the weak environmental centralization in the western regions makes it difficult to stimulate enterprises to improve output efficiency significantly.

Table 8 Region heterogeneity.
Full size table

According to the National Sustainable Development Plan for Resource-based Cities, this study categorizes sample enterprises into resource- and non-resource-based regions. Panel B of Table 8 shows the coefficients for the two areas and shows that VEG only significantly contributed to enterprise TFP in the non-resource-based regions. The findings suggest a marked resource curse in China. Regions with abundant natural resources instead form a drag on output efficiency due to the excessive reliance on low-end resource exploitation industries.

From the perspective of the fiscal balance, the extent to which a region can engage in environmental centralization reform is directly determined by fiscal pressure. This article considers the fiscal pressure index to reflect regional fiscal pressure. Regional fiscal pressure increases with the index. Panel C of Table 8 reports the estimated results for the two sample firms based on the median division of the index and shows that VEG had a better implementation effect in low-fiscal-pressure regions. This is because low-fiscal-pressure regions have more abundant fiscal resources for environmental expenditures.

Industry heterogeneity

In terms of industry factor aggregation, the samples can be divided into labor-concentrated, capital-concentrated, and technology-concentrated industries74. Panel A of Table 9 shows that TFP in labor-concentrated industries did not improve in response to environmental centralization reform. This is because capital-concentrated and technology-concentrated industries have the advantage of improving productivity through capital inputs and technological innovation. Labor-concentrated industries are less likely to improve output efficiency since factor inputs are labor-based.

Table 9 Industry heterogeneity.
Full size table

In terms of industry type, Panel B of Table 9 shows that VEG only significantly contributed to manufacturing TFP. The findings confirm that the presence of Baumol's disease will be detrimental to realizing the effects of environmental centralization reform in China. Therefore, it is necessary to increase fiscal support for the non-manufacturing industries to incentivize their participation in green production and services, thereby forming economies of scale in the non-manufacturing sectors.

This paper classifies the samples as high-energy-consuming enterprises and low-energy-consuming enterprises based on the industry code to which the enterprises belong. The estimated coefficients are reported in Panel C of Table 9. It shows that the reform effect of low-energy-consuming enterprises was stronger than that of high-energy-consuming industries. The productivity growth of high-energy-consuming industries tends to be slower than that of other industries due to the difficulty of eliminating the dependence on energy consumption in the short run. Moreover, the space for high-energy-consuming industries to survive is limited due to environmental regulations and the development of environmental industries. Shrinking market size and profitability have also led to a lack of innovation in high-energy-consuming enterprises.

Enterprise heterogeneity

Since enterprises of different sizes face different market environments, there may be differences in their business strategies. This study distinguishes large-sized enterprises from small and medium-sized enterprises based on the median asset size. Based on the results of Panel A in Table 10, the effect of VEG on promoting enterprise TFP was influenced by the size of the enterprise, with a more significant impact on small and medium-sized enterprises. This result indicates that the improvement of TFP in large-sized enterprises relies more on market position and scale advantages.

Table 10 Enterprise heterogeneity.
Full size table

The median year of establishment was used to classify the sample firms into growth period enterprises and mature enterprises. Based on the results of Panel B in Table 10, VEG had a more significant promoting effect on the TFP of growth period enterprises. In order to gain a higher market share and improve enterprise competitiveness, growth period enterprises are more proactive than mature enterprises in improving output efficiency.

This study further discusses the impact of property rights heterogeneity on the results. Table 10, Panel C shows that after distinguishing between state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs), the reform effect of SOEs is more significant in non-SOEs. This is due to the fact that SOEs have to consider both economic efficiency and political relations in their operations, thus lacking operating autonomy. However, non-SOEs are more flexible in their operations and more effective in utilizing fiscal resources to improve output efficiency.

Based on the whole empirical analysis in Sect. "Heterogeneity tests", Hypothesis 3a, Hypothesis 3b, and Hypothesis 3d hold, while Hypothesis 3c is not true.

Conclusions and policy implications

Using the reform of vertical environmental governance (VEG) in China as a quasi-natural experiment, this paper empirically examines the impact of environmental centralization on enterprise total factor productivity (TFP) by adopting the staggered DID method. This study shows that: (1) Compared to environmental decentralization regions, enterprises in environmental centralization regions have significantly higher TFP. (2) The effect of environmental centralization increases with the duration of exposure to the reform. (3) VEG increases the intensity of government environmental regulations and the willingness of corporate green innovation. (4) The effects of VEG are heterogeneous at the regional, industry, and enterprise levels.

The findings contribute to understanding the relationship between fiscal regimes and enterprise TFP, and the findings have several applications for reforming governance. First, it is necessary to accelerate the roll-out of environmental centralization reform across the country. Some regions have insufficient incentives for reform because of the long-term dependence on traditional high-polluting industries. Therefore, there is great value in strengthening fiscal interventions in unreformed areas. Furthermore, customized reform measures should be adopted for different regions. For regions with lax local government supervision, environmental regulations of the provincial governments should be strengthened, and administrative orders should be applied to force regional industries to transform. For regions with high fiscal pressures and insufficient environmental governance capacity, the central government is required to increase expenditures and exert the role of central budgetary support for local finances. Fiscal transfers are an essential lever to eliminate regional disparities. More transfer payments should be provided to enterprises in regions with backward economies and high fiscal pressure.

Second, traditional and non-manufacturing industries have insufficient technological innovation capacity to improve output efficiency through their own development. Fiscal attention to traditional and non-manufacturing industries should be strengthened to promote the upgrading of such enterprises. Therefore, environmental expenditures should focus on investing in environmental protection industries and constructing environmental infrastructure. By supporting environmental technological innovation through government support, the cost of enterprise inputs can be reduced, thereby stimulating the transformation of traditional and non-manufacturing industries.

Third, market monopolization relying on the size of enterprises and the nature of property rights is detrimental to output efficiency. The governments should strengthen market supervision and carry out administrative crackdowns on monopolistic behaviors and market barriers. As market regulators, the governments should improve support and protection for small and medium enterprises and newly established enterprises, thus maintaining the order of fair competition in the market and stimulating enterprise technological innovation. Moreover, large-scale and mature enterprises have higher market power and innovative capacity. It is also necessary to increase environmental advocacy and encourage enterprises with market power to actively participate in social responsibility.