How Does Vertical Fiscal Imbalance Affect CO₂ Emissions? The Role of Capital Mismatch

Abstract

Climate warming caused by greenhouse gases is an important practical issue. This study aims to explore the impact of the vertical fiscal imbalance (VFI) on CO₂ emissions from the perspective of theoretical analysis and empirical research. This study uses panel data from 30 provinces in China from 2004 to 2018 in order to test this issue. The results show that the VFI has a significant positive impact on CO₂ emissions and that the capital misallocation exacerbates the positive impact of the VFI on CO₂ emissions. These study results also have a significant temporal heterogeneity. The sample results dating after 2008 were more significant. These conclusions provide economic and political references for local governments in order to develop CO₂ neutrality and CO₂ peaking policy goals and to promote an in-depth reform of the fiscal system.

1. Introduction

In recent years, global warming caused by the greenhouse effect has become an important issue of general concern worldwide. Global warming will worsen the human living environment, thus cause the violation of the concept of sustainable development as well as having a potentially catastrophic impact on the future [1,2,3]. Carbon dioxide (CO₂) is the most important greenhouse gas. The Fifth Assessment Report by the United Nations Intergovernmental Panel on Climate Change reported that in 2011, the global atmospheric CO₂ equivalent concentration had reached 430 ppm. If not controlled, the CO₂ equivalent concentration will exceed 450 ppm by 2030 and exceed 750 ppm by the end of this century. Thus, determining how to effectively curb CO₂ emissions has become a major issue of concern among the international community. For China, solving this practical problem is an important guarantee in order to achieve a high-quality economic development in the future. In 2007, China’s total CO₂ emissions totaled 6.792 billion tons and surpassed the total CO₂ emissions of the United States, which were 5.795 billion tons during the same period, thereby becoming the world’s highest emitter of CO₂. The long-term development model of “pollution first, governance later” has facilitated China’s rapid economic growth while also consuming energy and paying high environmental costs. This model goes against the strategic goal of the high-quality economic development currently advocated by China. In order to solve this urgent practical problem, China has provided a number of relevant targets. According to the Paris Agreement and with 2005 as the base period, China’s CO₂ emissions per unit of the GDP needs to be reduced by 60–65% within 25 years. In September 2020, at the 75th United Nations General Assembly, the Chinese government reported that China’s CO₂ emissions would peak by 2030 and that CO₂ neutrality would be achieved by 2060. This realistic goal provides a good opportunity to study China’s CO₂ emissions.

The achievement of targets such as controlling CO₂ emissions and advancing CO₂ emission reductions in depth also depends on the relationship between governments, because CO₂ emissions are an environmental issue and have negative externalities for the outside world. Just as achieving the goal of CO₂ neutrality requires the cooperation of many countries around the world, China’s solution to the problem of high CO₂ emissions requires the coordination of the central government and the full cooperation of local governments. However, given the reform of the tax sharing fiscal system in 1994, from a vertical point of view, China’s intergovernmental fiscal relationship has been in an unbalanced state [4]. This specific governance structure with a prominent vertical fiscal imbalance (VFI) provides an interesting context in order to analyze the relationship between the VFI and CO₂ emissions. In this way, the present study considers whether and how the VFI affects CO₂ emissions. Despite the growing literature that explores how the VFI affects China’s CO₂ emissions through environmental regulation and industrial transformation [5], little is known about how the VFI affects CO₂ emissions via the capital mismatch. Varying degrees of capital mismatches occur among regions, industries, and companies in China [6,7,8]. Song and Wu [9] used the generalized ARP approach and empirically found that the capital misallocation caused a 20% loss in China’s production efficiency. This loss severely restricted the sustainable development of China’s economy [8,10,11]. The misallocation of capital leads to an excessive capital inflow into inefficient enterprises [12,13], thereby promoting the development of extensive industries and aggravating the environmental pollution. Therefore, the present study aims to investigate how the VFI affects CO₂ emissions via the capital mismatch.

On the basis of these problems, the contribution of this paper is mainly in the following three aspects. The existing literature has not yet explored the effect of the VFI on CO₂ emissions from the perspective of the capital mismatch. First, we explore this effect from the perspective of theoretical and empirical analyses, thereby enriching the relevant research. Second, we explore the temporal heterogeneity of the impact mechanisms from the perspective of financial crises. The relationship between financial crises and the capital misallocation is closely linked, and this result has various implications. Finally, we propose specific policy recommendations. These findings have a certain reference significance for local governments in order to achieve CO₂ neutrality, CO₂ peaking policy goals, and fiscal system reforms.

The remainder of the paper is structured as follows. Section 2 discusses the background of the literature and development of the hypotheses. Section 3 introduces the analysis strategy and data explanation. Section 4 follows with the empirical results. Section 5 summarizes the policy arguments.

2. Literature Review and Hypothesis Development

2.1. Literature Review

The existing research has produced many discussions on the economic consequences of VFIs. Eyraud and Lusinyan [14] used data from 28 OECD countries in order to quantify the VFI, they explored the relationship between the VFI and the financing structure of local governments. They found that reducing the VFI can help to increase the overall fiscal balance of local governments. Jia et al. [15] studied the impact of the VFI on local government expenditure policies based on China’s 1997–2006 county-level panel data and found that the decentralization of expenditures increased government expenditures; however, that study focused on production and constructive expenditures. Jia et al. [16] used China’s 1995–2014 county-level panel data in order to study the impact of the VFI on taxation policies and found that the VFI suppressed the taxation efforts of local governments. Liu and Zhang [4] used China’s 1998–2018 provincial panel data in order to study the impact of the VFI on government science and technology expenditures. The authors found that the VFI significantly suppressed local government spending on science and technology with a heterogeneity in time and space. Lin and Zhou [17] used China’s 1995–2017 provincial panel data in order to study the impact of the VFI on upgrading the industrial infrastructure. The authors found that the VFI was not conducive to the rationalization of the industrial structure between the different regions. Liu and Zhang [18] investigated the impact of the VFI on energy consumption. The authors found that the VFI indirectly increased energy consumption through upgrades to the industrial infrastructure.

At present, studies on the VFI and environmental pollution are limited. The most recent studies are Huang and Zhou [5], Li et al. [19], and Lin and Zhou [20]. Huang and Zhou [5] used China’s provincial panel data from 1999 to 2016 in order to empirically analyze the impact of the VFI on environmental pollution and to explore the path of its impact. The study found that the VFI will significantly aggravate the environmental pollution, mainly by influencing environmental regulation and industrial transformation and other intermediary pathways in order to promote CO₂ emissions, leading to the deterioration of the quality of the environment. Li et al. [19] used Pakistan’s provincial panel data from 2000 to 2018 in order to empirically analyze the impact of the VFI on CO₂ emissions. The authors found that the VFI promotes CO₂ emissions through intermediary paths such as the environmental supervision and industrial structure. This result is similar to the conclusions made by Huang and Zhou [5]. In general, this research has reached relevant conclusions through empirical studies based on samples from China and Pakistan. The studies found that the VFI has a positive effect on CO₂ emissions and will deteriorate the quality of the environment. At the same time, the studies all indicate that the environmental supervision and industrial structure are important intermediary paths. Using China’s provincial panel data from 2000–2017, Lin and Zhou [20] measured the energy and environmental performance (EEI) considering CO₂ emissions as an undesired output. The authors found that the VFI significantly reduced the EEI and that the main impact mechanisms were industrial structure upgrading, technological innovation, and the strengthening of government intervention. However, the authors did not consider the influence of other mechanisms, such as the moderating effect of capital mismatches.

In summary, the above literature provides empirical evidence for the economic consequences of the VFI by studying the relationship between the VFI and environmental pollution. Existing studies have examined the negative effects of the VFI on government behavior, industrial upgrading, energy, and environmental pollution and examined the influence channels of the VFI from various perspectives, thereby providing important policy suggestions for the reform of our fiscal system. However, there remain shortcomings in the existing research. First, published studies did not consider the impact of the VFI on CO₂ emissions in terms of the capital misallocation. Capital mismatches have an important impact on sustainable development such as the green total factor energy efficiency, energy intensity, and carbon emission reductions [8,11,12,21,22], which should not be ignored. Second, the included heterogeneity analysis usually lacks analysis of the influence channel. This situation provided an opportunity for the development of this article.

2.2. Hypothesis Development

Central governments in many countries around the world have decentralized fiscal power and delegated more expenditure responsibilities and revenue functions to lower-level local governments [14]. However, the balance between decentralized expenditure responsibilities and income power has shifted, and the expenditure responsibility is often greater than income, which directly leads to the VFI. In recent decades, China has implemented various fiscal system reforms, such as the highly centralized fiscal management system (before 1978), the fiscal contracting system (1979–1993), and the tax sharing system (1994–present). The reform of the fiscal system has had a profound impact on China’s economic transformation and development [23]. Unlike the previous two fiscal system reforms, the tax sharing system helps the central government to achieve a tax revenue concentration while delegating more expenditure responsibilities to local governments. In China, a fiscal imbalance between the central and local governments has gradually emerged. Generally, the central government bears the expenditure responsibilities for national public goods and services, while the local government bears the expenditure responsibilities for local public goods and services [24]. However, in reality, the VFI causes local governments to favor productive expenditures such as the construction of infrastructure [15] and the reduction of the provision of public goods and services [25], such as reducing investments in environmental governance. The environment is a public product with extremely strong negative externalities that require the government to invest a large amount of subsidy funds in order to intervene [26]. Reducing investments in environmental governance and allowing development will inevitably cause the deterioration of the quality of the environment.

First, under a political promotion championship, the promotion of local government officials is assessed based on the GDP [20,27]. Once the central government takes over part of the financial power, the local government can use fewer factor resources and bear a greater pressure over the financial expenditure, causing the local government to engage in short-sighted investment behaviors [15]. For example, concerning productive and innovative investments, local government officials tend to choose productive investment. The reason for this choice is because productive investment offers a short investment cycle, quick results, and low risk. Innovative investment, on the other hand, offers a long investment cycle, slow results, high risk, and great uncertainty [28,29]. Based on the hypothesis of a “political–economic man,” the choice of productive investment among government officials conforms to the maximization of both economic and political benefits. Therefore, the final result of the VFI is to promote an extensive economic growth model, aggravate the regional environmental pollution, and increase CO₂ emissions. Furthermore, due to resource limitations, the government puts more resources into productive expenditures, inevitably reducing expenditures on other public product projects. The expenditures originally invested in the environment will also be transferred and occupied [26], which is not optimistic for overall environmental governance. Finally, due to the previous vertical information gap between central and local governments, supervision requires huge costs [30,31]. A Chinese-style decentralization allows local governments and their officials to control the local economy, master various element resources, and have more autonomous decision-making powers [23,32,33]. This situation directly leads to an “absence of supervision” by the central government over local governments and their officials and is prone to deviations in policy objectives between the central and local governments.

Controlling CO₂ emissions requires local governments to invest in an environmental governance and other public product and service projects in order to maximize resource allocation and promote the development of a low-carbon economy. The rational behavior of local governments under the VFI will increase the level of CO₂ emissions, in addition to violating the strategic decisions of the central government. From a theoretical and practical point of view, the VFI is not conducive to the central and local governments reaching a consensus on achieving CO₂ neutrality or CO₂ peaking policy goals. Thus, we put forward the following hypothesis:

Hypothesis 1 (H1).

Ceteris paribus, the higher the VFI level, the higher the regional CO₂ emissions.

The phenomenon of capital mismatches is widespread in China [7]. Given that the establishment of the country’s capital market is relatively recent, the allocation of capital resources is largely subject to local government intervention. Wu [34] used data from the Chinese industrial enterprise database in order to establish a structural model and found that policy distortions and financial frictions cause capital mismatches. Moreover, capital mismatches create losses to production efficiency. In terms of green and sustainable development, the capital misallocation can exacerbate the economic consequences of VFIs on CO₂ emissions.

First, the capital misallocation distorts the price of capital factors. The direct result of the distortion of capital factor prices is inefficient capital allocation [6,9], potentially because capital elements cannot comply with changes in the basic supply–demand relationship under intervention and cannot supply capital to the sectors that need it most. At the same time, distortions in the price of capital factors will also induce companies to enter industries with overcapacity, leading to a waste of resources. Even if a higher economic output is ensured, this situation will eventually lead to the deterioration of the quality of the environment.

Second, the capital misallocation supports an extensive industrial development. Under the capital misallocation, too much capital flows into extensive industries with quick short-term results and low resource efficiency, thereby restricting upgrades to the industrial structure. The price of the short-term GDP rise is a waste of resources and environmental pollution, greatly restricting the high-quality development of the local economy [8,10,11]. Alam [35] used European Company data from 2005 to 2014 in order to examine the periodicity of the capital mismatch and its sources of funding and found that the degree of the capital mismatch is higher during economic downturns than during economic prosperity. A possible reason for this phenomenon is that the capital misallocation distorts capital factor prices, inducing companies to enter industries with overcapacity or invest in extensive growth projects. Such results will also affect the quality of the environment.

Third, capital mismatches allow zombie companies to survive, and market competition makes achieving “survival of the fittest” difficult. A large amount of capital is misallocated to low-efficiency enterprises, resulting in a large amount of low-efficiency output. At the same time, due to the lack of long-term capital for development and innovation, as well as the mismatch and distortion of short-term capital, innovative companies cannot easily maintain their R&D and innovation activities. More importantly, this process is not conducive to reducing energy consumption per unit of output, pollution treatment equipment, or clean technology innovation. Thus, this process will inevitably increase the environmental pollution and hinder the development of a low-carbon economy [12]. Finally, in order to attract the inflow of social capital, local governments often use policy measures in order to enhance their competitiveness [36], such as relaxing environmental regulations [37,38]. Such measures will aggravate the environmental pollution and lead to an increase in CO₂ emissions. In general, the capital misallocation will exacerbate the positive effect of the VFI on CO₂ emissions. Thus, we put forward the following hypothesis:

Hypothesis 2 (H2).

Ceteris paribus, the capital misallocation has exacerbated the impact of the VFI on regional CO₂ emissions.

The outbreak of the financial crisis in 2008 had a recessionary impact upon the global economy, causing huge expenditure losses [39] that exceeded the impacts of previous crises [40]. Prior to the financial crisis, local governments had begun to strongly intervene in the financial system in order to deal with problems such as shortages of financial resources. At the same time, the administrative control directly intervened in the allocation of funds in order to support the development of industries with short-term results. Following the 2008 financial crisis, China’s economic development was affected and the growth rate slowed. Financial resources also became scarcer, thereby leading to fierce competition for capital. In order to revive the economy and ensure the GDP performance of local officials during their tenure, local governments increased their intervention in the allocation of funds in the financial system, thus intensifying the degree of the capital misallocation; additionally, a large amount of funds flowed into low-efficiency enterprises [41,42,43]. At the same time, local governments increased the production expenditure inputs [44], supported the rapid recovery of extensive industries, and achieved a short-term economic recovery [4]. However, these practices are not friendly to environmental governance and will lead to a sharp rise in CO₂ emissions. Thus, we put forward the following hypothesis:

Hypothesis 3 (H3).

Ceteris paribus, after the financial crisis, the moderating effect of the capital allocation has become more significant.

3. Empirical Strategy, Variables, and Data

3.1. Econometric Model Specification

First, in order to test Hypothesis 1, we establish a benchmark regression model in order to verify the impact of the VFI on CO₂ emissions, as shown in Equation (1):

$$\ln {({\mathrm{CO}}_2)}_{\mathrm{i},\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{VFI}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{Control}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\tau }}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(1)

where $\mathrm{i}$ and $\mathrm{t}$ represent province and year, respectively. ${\mathrm{CO}}_2$ represents per capita CO₂ emissions, $\mathrm{VFI}$ means vertical fiscal imbalance, and $\mathrm{Control}$ represents a series of control variables. ${\mathsf{\alpha }}_0$, ${\mathsf{\alpha }}_1$, and ${\mathsf{\alpha }}_2$ are the unknown regression coefficients. $\mathsf{\tau }$ and $\mathsf{\mu }$ represent the province fixed effect (province FE) and the year fixed effect (year FE), respectively. Lastly, $\mathsf{\epsilon }$ is the error term.

In order to test Hypothesis 2, we further examine the impact of the capital misallocation. With reference to Bu et al. [45] and Lin and Zhou [20], based on the original model, we add a moderating variable that characterizes the capital mismatch and its interaction term with the VFI. Following Equation (1), we can construct Equation (2):

$$\ln {\left({\mathrm{CO}}_2\right)}_{\mathrm{i},\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{VFI}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\alpha }}_2{\mathrm{Kims}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\alpha }}_3{\mathrm{VFI}}_{\mathrm{it}}\times {\mathrm{Kims}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\alpha }}_4{\mathrm{Control}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\tau }}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(2)

where $\mathrm{Kims}$ refers to the capital misallocation index. Other variables are defined the same as those in Equation (1).

3.2. Variable Definitions

3.2.1. Measuring CO₂ Emissions

For the measurement of CO₂ emissions, we mainly measure the per capita CO₂ emissions by taking the natural logarithm (ln(CO₂)).

3.2.2. Measuring the VFI

Based on the characteristic facts of China’s fiscal decentralization, following Eyraud and Lusinyan [14], we calculate the coefficient of the VFI by constructing an index system:

$$\mathrm{VFI}=1-\frac{\mathrm{FRD}}{\mathrm{FED}}\left(1-\mathrm{LFG}\right)$$

(3)

where $\mathrm{FED}$ and $\mathrm{FRD}$ denote the fiscal spending decentralization and the fiscal revenue decentralization, respectively. According to the measurement method of the fiscal decentralization by Liu et al. [46] and Liu and Li [47], the $\mathrm{FED}$ ($\mathrm{FRD}$) is equal to the local per capita fiscal budget expenditure (revenue)/local per capita fiscal budget expenditure (revenue) and the central per capita fiscal budget expenditure (revenue). $\mathrm{LFG}$ is the local fiscal self-sufficiency rate, which is measured by the difference between the local fiscal budget expenditures and the local fiscal budget revenue/local fiscal budget expenditures. When the degree of asymmetry between the $\mathrm{FED}$ and $\mathrm{FRD}$ is higher, the degree of the VFI is greater.

3.2.3. Measuring the Moderator Variable

With reference to Hsieh and Klenow [6] and Chen and Hu [48], we calculate the capital mismatch index for each province. The calculation formula is shown in Equation (4):

$${\mathrm{Kmis}}_{\mathrm{i}}=\frac{1}{{\mathsf{\gamma }}_{{\mathrm{K}}_{\mathrm{i}}}}-1$$

(4)

where ${\mathsf{\gamma }}_{\mathrm{K}}$ is the absolute distortion coefficient of capital prices. The calculation method is shown in Equation (5):

$${\mathsf{\gamma }}_{{\mathrm{K}}_{\mathrm{i}}}=\left(\frac{{\mathrm{K}}_{\mathrm{i}}}{\mathrm{K}}\right)/\left(\frac{{\mathrm{s}}_{\mathrm{i}}{\mathsf{\beta }}_{{\mathrm{K}}_{\mathrm{i}}}}{{\mathsf{\beta }}_{\mathrm{K}}}\right)$$

(5)

where $\frac{{\mathrm{K}}_{\mathrm{i}}}{\mathrm{K}}$ represents the proportion of the capital of the region $\mathrm{i}$ in the total capital. ${\mathrm{S}}_{\mathrm{i}}=\frac{{\mathrm{P}}_{\mathrm{i}}{\mathrm{y}}_{\mathrm{i}}}{\mathrm{Y}}$ represents the share of output ${\mathrm{y}}_{\mathrm{i}}$ of region $\mathrm{i}$ in output $\mathrm{Y}$ of the entire economy. The output variable $\mathrm{Y}$ is represented by GDP. $\frac{{\mathrm{S}}_{\mathrm{i}}{\mathsf{\beta }}_{{\mathrm{K}}_{\mathrm{i}}}}{{\mathsf{\beta }}_{\mathrm{K}}}$ represents the theoretical proportion of the capital used by region $\mathrm{i}$ when the capital is effectively allocated. The amount of the capital input (${\mathrm{K}}_{\mathrm{i}}$) is expressed by the fixed capital stock of each province, calculated using the perpetual inventory method. The formula is as follows:

$$K_{i,t}=\raisebox{1ex}{$I_{i,t}$}\!\left/ \!\raisebox{-1ex}{$P_{i,t}$}\right.+\left(1-\delta \right)K_{i,t-1}$$

(6)

where $K_{i,t}$ is the current stock of the fixed capital, $I_{i,t}$ is the total amount of the nominal fixed assets formed in the current period, $P_{i,t}$ is the fixed asset investment price index, $\delta$ is the depreciation rate, and $K_{i,t-1}$ is the stock of the fixed capital in the previous period.

Finally, we calculate the capital mismatch index according to Equation (4) and take the absolute value. When the value of the index is larger, the degree of the capital mismatch is more serious.

3.2.4. Measuring Control Variables

Based on previous studies, we further consider the influencing factors of CO₂ emissions, including various control variables. First, we control the impact of regional economic development on regional CO₂ emissions and use (the log of) the GDP per capita (lnGDPPC) as a control variable. Second, we control the impact of government decisions on CO₂ emissions and use trade opening (Open) and (the log of) the amount of fixed asset investment in the whole society (lnInvest) as control variables. Finally, natural resources such as forests and water may have a certain impact on carbon emissions. It is well known that forest carbon sinks have great potential in order to mitigate climate change [49]. The utilization of water resources is helpful for both carbon emission reductions and carbon sequestration. Water is an important ecological factor that affects forest life activities and soil status and will have a certain impact on ecosystem carbon sequestration [50]. Abundant water resources are conducive to forest growth [51] and promote the benefits of high carbon sink ecosystems such as wetlands [52,53]. Therefore, in this study, we control the impact of natural resources on CO₂ emissions using (the log of) the total forestry investments (lnForestry) and (the log of) the water resources per capita (lnWater) as control variables.

3.3. Data Sources

Considering the availability of data, the present study uses data from 30 provinces (except for Hong Kong, Macau, Taiwan, and Tibet) from 2004 to 2018 in order to test the hypothesis. On this basis, we made a total of 450 observations. Panel data were obtained from the Statistical Yearbook of China, Finance Yearbook of China, and the WIND database.

Table 1. Descriptive statistics of each variable.

Var	Obs	Mean	Std. Dev.	Min	Median	Max
ln(CO2)	450	−2.757	0.645	−4.695	−2.779	−0.837
VFI	450	0.685	0.235	0.009	0.779	0.978
Kmis	450	0.234	0.179	0.001	0.200	1.471
lnGDPPC	450	10.298	0.675	8.353	10.369	11.925
Open	450	0.045	0.055	0.003	0.020	0.259
lnInvest	450	8.856	1.080	5.667	8.951	10.981
lnForestry	450	12.853	1.343	8.186	12.949	16.213
lnWater	450	7.028	1.242	4.288	7.379	9.691

displays a descriptive analysis of each empirical variable. The results show that the VFI is more serious, and CO₂ emissions (ln(CO₂)) are also relatively high. These findings provide a solid foundation for our follow-up research on the relationship between the VFI and CO₂ emissions. We also use the variance inflation factor test (VIF) in order to test whether a multicollinearity problem exists between variables. The test results show that the maximum VIF is 6.560, meaning that the data used in this study do not have a high degree of multicollinearity.

4. Empirical Results

We first empirically studied the influence of the VFI on CO₂ emissions according to Equation (1). Then, we analyzed the influence mechanism of the capital mismatch between the VFI and CO₂ emissions, according to Equation (2). Next, we tested the robustness of the basic regression results, including the use of alternative VFI measurement methods, adding control variables, potential endogenous testing, and eliminating periodic interference. Finally, we considered the heterogeneous impact of financial crisis events and technological innovation.

4.1. Baseline Regression Results

Table 2. Baseline regression results.

	(1) ln(CO2)	(2) ln(CO2)	(3) ln(CO2)
VFI	0.406 ** (2.03)	0.602 *** (3.05)	0.688 *** (3.47)
Kims			0.510 *** (3.98)
VFI × Kims			0.861 *** (3.21)
lnGDPPC	0.658 *** (6.43)	0.223 * (1.68)	0.281 ** (2.14)
Open	2.188 *** (2.85)	2.236 ** (3.00)	2.298 *** (3.06)
lnInvest		0.259 ** (4.80)	0.204 *** (3.62)
lnForestry		−0.059 (−3.13)	−0.056 *** (−3.06)
lnWater		0.059 (1.43)	0.068 * (1.66)
Province FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Constant	−9.867 *** (−10.26)	−7.612 *** (−6.99)	−7.993 *** (−7.37)
R2	0.666	0.689	0.701
F test	47.33	44.30	42.40
Observations	450	450	450

Notes: *, **, and *** denote the significance level at 10%, 5%, and 1%, respectively; t statistics are in parentheses.

shows the baseline regression results. Column (1) presents the regression results considering only the VFI and some control variables, and column (2) presents the regression result after adding all of the control variables. The explained variables in columns (1) and (2) are CO₂ emissions (ln(CO₂)). Compared with column (1), the control variables in column (2) are more comprehensive and provide a higher goodness of fit. Therefore, the results of column (2) can better explain the phenomenon. We also add the capital mismatch variable and its interaction term with the VFI (VFI × Kmis) in column (3) in order to observe the adjustment mechanism of the capital mismatch. The results of the F test reveal that in the above three models, the overall regression results are significant.

According to the estimation results (columns (1) and (2)), we find that the regression coefficient of the VFI is significantly positive, meaning that the VFI has a significant positive effect on CO₂ emissions. Thus, Hypothesis 1 is supported. This result shows that the VFI significantly promotes the growth of CO₂ emissions. On the one hand, the imbalance in fiscal revenue and expenditures has led local governments to favor productive investments and promote the growth of extensive industries. On the other hand, when fiscal expenditure responsibility is greater than income, as “political–economic” actors, decision makers in local governments tend to reduce the provision of public products and services such as environmental governance, hindering the environmental governance and increasing CO₂ emissions.

According to the estimation results in column (3), we find that the VFI and CO₂ emissions are still significantly positive, further supporting Hypothesis 1. The coefficient of capital mismatch and the coefficient of the interaction term are significantly positive—that is, the higher the degree of capital mismatch, the more significant the positive impact of the VFI on CO₂ emissions. This finding means that capital mismatches have a positive moderating effect. Thus, Hypothesis 1 is supported. This result shows that the capital misallocation exacerbated the impact of the VFI on regional CO₂ emissions, possibly because capital mismatches distort the price of capital factors and support extensive industries, thereby exacerbating the impact of VFIs on CO₂ emissions.

Next, we briefly describe the results of the control variables. The GDP per capita (lnGDPPC) has a significant positive effect on CO₂ emissions, meaning that the higher the level of regional economic development is, the greater the degree of CO₂ emissions becomes. A possible reason for this result is that the previous regional economic development mainly relied on the contribution of the secondary industry. The development of industrial enterprises is often accompanied by high energy consumption, leading to an increase in CO₂ emissions. The regression coefficients of the trade opening variable (Open) and the fixed asset investment variable of the whole society (lnInvest) are significantly positive. A possible reason for this result is that the opening of trade and the increases in social investment will further stimulate social demand and yield rapid development of the real economy, leading to increased CO₂ emissions. The regression coefficient of the forestry investment variable (lnForestry) is negative, and the regression coefficient of per capita water resources (lnWater) is positive. This finding shows that natural resources will have a certain impact on CO₂ emissions.

4.2. Robustness Checks

In this section, we use several methods in order to test whether the baseline regression results are robust. This robustness test is based on column (3) in Table 2. First, we change the way that the VFI is measured. For the calculation of the VFI, we replace the local fiscal gap rate with the national fiscal revenue and expenditure gap rate. The calculation method for the national fiscal revenue and expenditure gap rate is the ratio of the national budget gap (national public budget expenditures minus national public budget revenue) among national public budget expenditures. According to the estimation results in column (1) of

Table 3. Robustness test results.

	(1) ln(CO2)	(2) ln(CO2)	(3) ln(CO2)
VFI	0.764 *** (3.08)	0.520 *** (2.73)	1.287 ** (2.59)
Kims	0.419 *** (3.33)	0.550 *** (4.28)	0.625 *** (2.73)
VFI × Kims	0.949 ** (2.57)	0.835 *** (3.19)	0.958 ** (2.00)
lnGDPPC	0.281 ** (2.14)	0.161 (1.19)	0.076 (0.32)
Open	2.246 *** (2.99)	2.638 *** (3.56)	2.589 * (1.83)
lnInvest	0.220 *** (3.89)	0.202 *** (3.72)	0.289 *** (2.67)
lnForestry	−0.056 *** (−3.01)	−0.051 *** (−2.75)	−0.075 ** (−2.06)
lnWater	0.061 (1.48)	0.046 (1.13)	0.256 ** (2.15)
Province FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Constant	−8.007 *** (−7.28)	−6.500 *** (−5.00)	−8.166 *** (−4.04)
R2	0.696	0.649	0.752
F test	41.41	32.48	27.28
Observations	450	450	150

Notes: *, **, and *** denote the significance level at 10%, 5%, and 1%, respectively; t statistics are in parentheses.

, the regression coefficients of the VFI remain significantly positive, and the regression coefficients of the capital mismatch and the interaction terms also remain significantly positive. This finding shows that use of the new VFI indicator does not change the original results. In addition, the performance of the control variables remains consistent.

Second, in order to eliminate any endogenous interference, we re-regress the key variables after a period of lag. According to the estimated results of column (2) in Table 3, we still find that the regression coefficients of key variables after a lag of one period are consistent with the basic results.

Finally, in order to eliminate business cycle effects [4,54], we take the average of all of the variables with a period of three years and retest. In this way, the panel data become smooth data instead of annual data. Ultimately, our study object was reduced to 150 observations. According to the estimated results of column (3) in Table 3, the results shown by the smoothed data remain consistent with the previous ones.

The results did not change after the robustness tests, indicating that our conclusions are reliable. Therefore, the VFI has a significant positive relationship with CO₂ emissions, and the capital mismatch has a positive regulatory effect.

4.3. Heterogeneity Analysis

In this part, based on Equation (2), we further capture the heterogeneity in time.

Table 4. Heterogeneity test results.

	(1) 2004–2008 ln(CO2)	(2) 2009–2018 ln(CO2)
VFI	0.251 (0.87)	0.652 ** (2.21)
Kims	−0.699 * (−1.88)	0.468 *** (3.10)
VFI × Kims	−0.513 (−0.47)	0.720 ** (2.29)
lnGDPPC	0.689 (1.61)	0.049 (0.30)
Open	−4.270 * (−1.95)	2.119 *** (2.77)
lnInvest	0.012 (0.06)	0.091 (1.61)
lnForestry	−0.092 * (−1.95)	−0.030 (−1.48)
lnWater	0.066 (0.85)	−0.006 (−0.17)
Province FE	Yes	Yes
Year FE	Yes	Yes
Constant	−9.120 ** (−2.27)	−4.175 *** (−2.61)
R2	0.595	0.356
F test	13.24	8.216
Observations	150	300

Notes: *, **, and *** denote the significance level at 10%, 5%, and 1%, respectively; t statistics are in parentheses.

shows the test results of the heterogeneity analysis. The F test shows that the following model results are generally significant.

Here, we explore the impact of the time heterogeneity before and after the financial crisis. Column (1) reports the estimation results of the samples before the financial crisis (2004–2008), and column (2) introduces the estimation results of the samples after the financial crisis (2009–2018). According to the estimation results, there is a significant difference between the results of columns (1) and (2). The regression coefficient of the VFI in the results before the financial crisis was positive but not significant, while the regression coefficient of the VFI in the results after the financial crisis was significantly positive. This finding supports Hypothesis 1 and also indicates that the impact of the VFI on CO₂ emissions is temporally heterogeneous. Therefore, in order to promote the local economic development as soon as possible after the financial crisis, local governments should be more inclined to invest in industrial enterprises that yield quick results. To some extent, this practice has contributed to an increase in CO₂ emissions. Further observing the regression coefficients of the capital mismatch and interaction terms, we find that the results in column (1) are not significant. However, in the results of column (2), the regression coefficients of the capital mismatch and interaction terms are both significantly positive. This finding is consistent with the previous results, showing that after the financial crisis, the adjustment effect of the capital mismatch is more significant. Therefore, the more serious the capital misallocation is, the more significant the positive relationship between the VFI and CO₂ emissions becomes. This result supports Hypotheses 2 and 3 and also shows that the adjustment effect of the capital mismatch is heterogeneous in time. In areas with serious capital mismatches, in order to encourage economic recovery, local governments should pay more attention to investing in industrial enterprises with faster short-term performance improvements. Such policies are often accompanied by high energy consumption, which will promote an increase in CO₂ emissions [18].

5. Discussion and Conclusions

5.1. Discussion

Since the signing of the Kyoto Protocol, the issue of climate warming has received widespread attention. Determining how to effectively control CO₂ emissions has become the key to achieving sustainable development. For China, in order to effectively control CO₂ emissions, it will be necessary to change the extensive economic growth model. However, since the tax-sharing system reform in 1994, fiscal revenue has become concentrated, and expenditure responsibilities have been decentralized, thereby increasing the gap in local government revenue and expenditures. The VFI changed the decision-making behavior of local governments but promoted the extensive industrial growth, thus violating the Sustainable Development Goals and leading to an increase in CO₂ emissions.

This context provided a good opportunity for us to study the impact of the VFI on CO₂ emissions. The present study explored the impact of the VFI on CO₂ emissions from the perspective of the capital mismatch and further considered the impact of the time heterogeneity, enriching the related study. Based on the basic results, the VFI has a significant positive effect on CO₂ emissions. This conclusion is consistent with the results of Huang and Zhou [5] and Li et al. [19]. We believe that the VFI affects the quality of the environment and hinders sustainable development.

China is currently in a period of economic transformation. Thus, it is particularly critical to achieve high-quality development. CO₂ emissions have negative externalities and represent an environmental issue that requires tacit cooperation between governments in order to handle. However, the VFI has led to distorted fiscal behaviors among local governments, thereby obstructing this goal. Consequently, in line with the views of Huang and Zhou [5] and Li et al. [19], we also believe that reducing or eliminating the VFI will lead to a better environmental quality, which is beneficial to sustainable development.

Next, we explored the impact of the capital misallocation. When discussing issues in related fields, Bian et al. [12], Yang et al. [11], and Gao et al. [8] all discussed the capital misallocation as a key mechanism. However, Huang and Zhou [5] and Li et al. [19] ignored the key mechanism of the capital misallocation when discussing the relationship between the VFI and CO₂ emissions. We believe that the capital misallocation aggravates the positive effect of the VFI on CO₂ emissions, which is not conducive to energy conservation and emission reductions. This result provides new empirical evidence for exploring the relationship between the capital mismatch and energy conservation and emission reductions and enriches relevant research.

Finally, we explored the temporal heterogeneity of impact mechanisms from the perspective of financial crises. Based on the results, the role of the capital misallocation was more significant after the financial crisis due to impacts on the Chinese economy during this period. Under the dual stimulus of economic recovery and “political championships”, local governments may increase productive spending, thereby squeezing the environmental protection spending and resulting in an increase of CO₂ emissions. This phenomenon is especially prominent in regions with a severe capital misallocation.

5.2. Conclusions

In this paper, using the panel data of 30 provinces in China from 2004 to 2018, we obtained the following findings. (1) The VFI had a significant positive impact on CO₂ emissions. (2) The capital misallocation exacerbated the positive impact of the VFI on CO₂ emissions. The more serious the capital misallocation, the more positive the impact of the VFI on CO₂ emissions. (3) These study results had significant heterogeneous differences in time. In the sample results after 2008, the results were more significant.

Based on the empirical research conclusions, we propose the following policy recommendations. First, social planners should appropriately balance the fiscal revenue and expenditure gaps between the central and local governments, expand the local fiscal revenue sources, and reduce the positive impacts of the VFI on carbon emissions. Specifically, it is possible to appropriately reduce the proportion of local governments’ tax payments and increase their tax revenue. In order to strengthen the cooperation between the central and local governments, the central government could accept a portion of the fiscal expenditures of public goods. In this way, the financial pressure on local governments could be alleviated. Second, social planners should reasonably consider the impact of capital mismatches with local low-carbon economic development, formulate relevant policy guidelines, design and effectively implement relevant supervision and control systems, and guide local governments to maximize the efficiency of capital allocations. In regions with a severe capital misallocation, guidance and subsidies should be strengthened in order to further reduce the levels of the VFI. In addition, changing the “GDP-only” assessment system and guiding local officials to set sustainable development goals would help to solve this problem. Finally, social planners should focus on the impact of capital mismatches on sustainable development. China’s economy is in a stage of high-quality development. It is thus necessary to actively guide capital to flow to green industries, invest in energy-saving and emission-reduction activities, reduce CO₂ emissions caused by the VFI, and promote improvements to the quality of the environment.

This study provides a new perspective on and empirical evidence for the relationship between the VFI and environmental pollution. However, some limitations remain. First, we focused only on the impact of the VFI on carbon emissions and did not consider the impact of the VFI on carbon-emission intensity. Second, we only analyzed related issues at the provincial level. Due to the difficulties in obtaining certain data, we did not consider the impact of prefecture- or county-level cities. Third, due to the difficulties in obtaining corporate carbon-emission data, we did not consider the impact of the VFI on corporate carbon emissions from a micro-enterprise perspective, which should be explored in subsequent studies. Finally, due to the difficulties in obtaining certain data, we could only study the situation up until 2018, which represents another shortcoming.

Future research could be carried out based on the following three aspects. First, future work could study the impact of the VFI on CO₂ emission intensity and CO₂ emission efficiency. Second, if data from the prefecture-level cities, county-level cities, and enterprise levels are available, follow-up research could be carried out at those levels. Finally, after the data are updated, empirical results over longer periods could be examined.