Unintended Results: Inter-Provincial Differences in Environmental Protection Tax Rates and Relocation Strategies of Polluting Enterprises

2.1 Institutional Background

The environmental tax system in a true sense has been long absent in China, which only has a pollution discharge fee system. In 2003, the State Council promulgated the Regulations on the Collection and Use of Pollutant Discharge Fees, along with Administrative Measures for the Collection Standards of Pollutant Discharge Fees, and Administrative Measures for the Collection and Use of Pollutant Discharge Fee Funds. Then, the system of collecting fees based on the total amount of pollutant discharge, similar to the existing EPT system, was put into practice. Since then, the pollutant discharge fees were collected on a growing scope of subjects and types, and the collection standards have become more refined and stringent. However, the effect of implementing the pollutant discharge fee system has been unsatisfactory due to the lack of tax rigidity and low collection efficiency (Huang and Li, 2018). In this context, China has been exploring the development of EPT system. On December 25th, 2016, the Standing Committee of the National People’s Congress passed the “Environmental Protection Tax Law of the People’s Republic of China”, changing the pollutant discharge fees into the EPT and incorporating the EPT into law. The Law was stipulated for nationwide implementation beginning in 2018. The EPT Law delivers a smooth transition from the pollution discharge fee system to the EPT system in the principle of “tax burden shifting”. Specifically, the taxable pollutants refer to solid wastes, noise, air pollutants, and water pollutants. The applicable tax rates for taxable solid wastes and noise should refer to the Table of Items and Amounts of Environmental Protection Tax as an annex to the EPT Law. The taxable air and water pollutants shall be determined based on the pollution equivalent. Each province could determine its local tax rates within the scope of the prescribed range, based on its own economic, social and ecological development goals, environmental carrying capacity and pollutant discharge situation. In practice, many provinces “shifted” the original pollution discharge fee standard to the EPT rate, but some provinces and cities gradually raised the tax rate.^[1] In this way, the existing EPT in China shows a distinctive feature—differentiated EPT rates across regions. Judging from the applicable tax rates for major air pollutants in various regions, 14 provincial-level regions, namely Inner Mongolia, Liaoning, Jilin, Heilongjiang, Anhui, Fujian, Jiangxi, Yunnan, Xizang, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang, implemented the minimum tax rates after the EPT reform in 2018, which is the lower bound of tax rate for taxable air pollutants set by the law at RMB 1.20 yuan per pollution equivalent. Except for Beijing that implemented the upper bound of tax rate, other provincial-level regions set the tax rates for air pollutants within the range of RMB 1.8~10 yuan per pollution equivalent. As for the applicable tax rates for water pollutants, 14 provincial-level regions, namely Inner Mongolia, Liaoning, Jilin, Heilongjiang, Anhui, Jiangxi, Shandong, Yunnan, Xizang, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang, implemented the lower bound of tax rates at RMB 1.40 yuan per pollution equivalent after the EPT reform in 2018. Except for Beijing that implemented the an upper bound of tax rates, other provincial-level regions set the tax rates for water pollutants within the range of RMB 1.5~12 yuan per pollution equivalent. In addition, Inner Mongolia, Yunnan, and Shanghai adjusted their local EPT rates in 2019, and Inner Mongolia and Liaoning raised their EPT rates in 2020.

2.2 Theoretical Models and Research Hypotheses

With reference to the research of Levinson and Taylor (2008) and Shapiro et al. (2018), this paper applies a partial equilibrium model to analyze the effect of regionally different EPT rates on industrial transfer or corporate migration. To simplify the analysis, it is assumed that there are two jurisdictions in a country: Jurisdiction 1 and Jurisdiction 2, representing the original location and the target location respectively, while the prices of production factors (labor and capital) and the unit EPT rate t are given exogenously. Assume that in the initial state, in the Jurisdiction 1, there exists a group of polluting industries η∈(0,1] that are sorted by polluting degree, and each polluting industry is composed of N enterprises that produce homogeneous products and discharge pollution. These enterprises move along with their industries, with i(η) referring to the representative enterprise of industry η. There’s no industry in Jurisdiction 2. Therefore, the pollution degree α_i(η)of enterprise i satisfies the conditions that α_i’(η)>0，0<α_i(η)<1, meaning that the enterprises of different industries produce varied degrees of pollution. Q_i(η) refers to the output of enterprise i in the industry η. The production costs for each unit of output of enterprise i in Jurisdiction 1 and Jurisdiction 2 are reflected as c₁ and c₂, respectively. The labor force and capital can flow freely among regions. The capital cost is the main reason for differences in production costs between jurisdictions. Due to the trans-regional management costs caused by transferring partial investment of enterprise i to Jurisdiction 2, we have c₁<c₂.

The production of each enterprise requires labor and capital of a specific industry, and pollution is a by-product of production. The enterprise will invest θ percent of production factors for emission reduction. Assuming that the production satisfies constant returns to scale and zero entry cost, the enterprises’ net output after investing θ percent of production factors for emission reduction can be calculated in the following formula:

The pollution emission E of representative enterprise i is a function of the production activities F and the emission reduction intensity θ, so the total emission E of enterprise i is:

Given that ϕ should be a decreasing function of emission reduction intensity θ, assume ϕ(θ)=(1−θ)^1/α, where 0<α<1, with reference to Copeland and Taylor (2004). Then, apply it to the above Formula (2) and it can be seen that enterprises choose θ to minimize the production cost under the conditions of given labor and capital prices .

To simplify the analysis, the Formula (3) sets the pollution emission E as a relative value. In the case of relatively low EPT rate, enterprises wouldn’t invest in pollution control. That is, θ = 0, and the pollution emission E equals the output Q. In the case of relatively high EPT rate, enterprises would actively engage in pollution control. That is, θ > 0 and the pollution emission decreases accordingly. Combining Formula (1) with Formula (3), we can get the following formula:

The above Formula (4) can be regarded as a Cobb-Douglas production function that regards pollution emission and net output as input factors, and the prices of the two are the unit EPT rate t and the finished product price c. Given the enterprise location in Jurisdiction 1 and based on the first-order condition for profit maximization, it can be known that the production cost of enterprise i(η) in the industry η in Jurisdiction 1 is:

Wherein, A i ( η ) = 1 − α α α − 1 + 1 − α α α . Similarly, if enterprise i(η) migrates to the target location in Jurisdiction 2, the production cost will become:

If and only if C_i1(η) > C_i2(η), that is, satisfying Formula (7), enterprise i(η) in Jurisdiction 1 would transfer investment to Jurisdiction 2.

Since c₁ < c₂ and 0 < α < 1, it comes to that c 1 c 2 1 − α α < 1 , making t₂ < t₁ an essential condition for the establishment of Formula (7). That is to say, when there is a difference in EPT rates between regions and the tax rate t₂ in the target location is lower than the tax rate t₁ in the original location, the polluting enterprises may choose to invest in migration to the target location. The specific range depends on the unit production costs c₁ and c₂ at the two locations and enterprises’ pollution degree α. Based on this, this paper puts forward Hypothesis 1 for verification.

Hypothesis 1: Inter-regional differences in EPT rates may lead to investment migration of polluting enterprises from regions with high EPT rates (original location) to regions with low EPT rates (target location).

If the above hypothesis 1 is established, let the function H η ; t 1 , t 2 = t 2 t 1 α ( η ) 1 − α ( η ) . There exists an industrial threshold η ¯ to establish H η ¯ ; t 1 , t 2 = c 1 c 2 , prompting enterprises of the industry η > η ¯ to migrate to Jurisdiction 2. If the EPT rate t₁ in the Jurisdiction 1 where the enterprise is located is raised to t 1 ∗ , at the same level of c 1 c 2 , the industrial threshold will change from η ¯ t o η ∗ . Since the function H(η;t₁,t₂) is a decreasing function of t₁, then η ∗ < η ¯ , indicating that more enterprises will choose to migrate their investment to the target location in Jurisdiction 2. This leads to the following Hypothesis 2.

Hypothesis 2: When there are inter-regional differences in EPT rates, if the EPT rate is raised in the region where the enterprise is located, the polluting enterprises’ investment migration effect of “avoiding high rates and opting for low rates” will be more significant.

3.1 Model Settings

To test Hypothesis 1, this paper draws on Ma et al. (2020) to employ the data of listed polluting companies in regions excluding Hong Kong, Macao, and Taiwan to establish a panel of data “listed companies c (with parent company located in region i) – the region of the subsidiary company j (j≠i) – year t” (hereinafter referred to as “company-region”), so as to better examine the impact of environmental differences between the original location and the target location on enterprises’ trans-regional migration. On this basis, the DID approach is applied to examine the impact of inter-provincial differences in EPT rates brought about by the EPT reform on enterprises’ location choice for investment. The empirical model is set as follows:

Wherein, the core explanatory variable is treat_cj×post_t. The treat_cj denotes whether the listed company c invest in the region j with the EPT rates lower than in the region where its parent company is located. If yes, it belongs to the treatment group, with the value at 1; if not, it belongs to the control group, with the value at 0. The dummy variable post_t denotes the implementation time of EPT reform, equaling 1 in the year of the reform and 0 in other years. The explained variable Invest_cjt represents trans-regional investment activities, measured by the number of subsidiaries established by listed company c in other regions j. If Hypothesis 1 is true, α₁ should be significantly positive. X is a set of control variables, including corporate size, asset-liability ratio, profitability, age, number of board directors, proportion of independent directors, staff of subsidiaries, as well as regional differences in GDP per capita, industrial structure, and road area. μ_cj controls the fixed effect at the level of the region j where the subsidiary of the listed company c is located. This fixed effect not only absorbs the factors that do not change over time at the company-region level (such as the historical relationship between the listed company and a certain region), but also controls factors that don’t change over time between two regions (such as the geographical distance between the two). At the same time, this paper controls the year fixed effect λ_t. ε_cjt is a stochastic disturbance, and the standard errors are clustered at the company-region level.

In order to further test the differential impacts of “raising” or “shifting” EPT (rate) reform on enterprises, this paper introduces the dummy variable increase_c on the basis of Formula (8). The specific model setup is shown as follows:

Wherein, increase_c denotes whether the EPT rate applicable to the region i where the parent company c is located is raised after the EPT reform. During the sample period, a total of 19 provinces in China raised the EPT rates. If the parent company of the listed company c is located in these regions, increase_c equals 1. The EPT rates in the other 12 provinces remained the same during the sample period, and increase_c equals 0. If Hypothesis 1 and Hypothesis 2 are both true, the coefficient treat_cj×post_t×increase_c in Formula (9) should be significantly positive.

3.2 Data Source and Description of Variables

Industrial production is the main source of environmental pollution. This paper selects listed companies (mainly industrial companies) under the first level of industry classification of B, C, and D among all A stocks from 2015 to 2020 as the initial samples. With reference to Xie et al. (2023), the information about subsidiaries is obtained from the notes to “long-term equity investment” in the annual report of the parent company of the listed company, and the detailed regional distribution is manually sorted out. This paper defines “trans-regional investment” as a domestic subsidiary in a province/region other than the registration place of the parent company. The initial samples are processed as follows. First, eliminate companies with special treatment (ST, ^*ST) during the sample period. Second, eliminate samples with relevant financial data missing. Third, eliminate sample companies with trans-regional changes in registration or operation place. Fourth, all continuous financial variables are winsorized at levels of 1% and 99 %. In the end, a total of 47000 “company—region” observations are obtained.^[1]

The information and financial data of listed companies in this paper are sourced from the CSMAR database. The data of regional EPT rate is obtained by manually sorting out the documents issued by the provincial development and reform commissions and environmental protection bureaus and the public information on official websites of the governments. Economic and social data at the provincial level is sourced from the National Bureau of Statistics, China Statistical Yearbook, and CEIC database.

4.1 Benchmark Regression: Inter-Provincial Differences in EPT Rates and Trans-Regional Investment

Table 1 presents the regression results at the company-region level obtained based on Formula (8). The explained variables in Columns (1) and (2) are the number of trans-regional subsidiaries invested by listed companies, and the control variables at the company and regional levels are gradually added. The results show that after the EPT reform, compared with regions with high EPT rates, the number of subsidiaries invested by polluting enterprises in regions with lower EPT rates increased by 0.101, accounting for 4.27% (=0.101/2.368×100%) of the average number of subsidiaries of listed companies in a single target region, and the effect was significant at the level of 1%. In other words, the inter-regional differences in EPT rates as a result of the EPT reform have indeed triggered the effect of migrating investment from regions with high EPT rates (original location) to regions with low EPT rates (target location). To avoid the influence of extreme values on the results, we also take the natural logarithm of the explained variables. The results shown in Columns (3) and (4) verify Hypothesis 1. The results of the control variables show that the polluting enterprise’s trans-regional investments increase with the growing number, size and age, and decreasing profitability of its subsidiaries.

4.2 Further Test: Differences in the Policy Effects of “Raising” and “Shifting” EPT Rates

Columns (5) and (6) of Table 1 display the impact of “raising” and “shifting” EPT rates after the EPT reform on the polluting enterprises’ location choice for trans-regional investment. The estimation results show that the coeficients for interaction term are all significantly positive, indicating that under the circumstance of interregional differences in EPT rates, if the EPT rate in the region where the polluting enterprise is located is raised (that is, the environmental regulation in the original location is stringent), the higher environmental protection costs will prompt the enterprise to adopt an environment-seeking strategy. It means transferring a part of investment to regions with lower EPT rates (that is, regions with less stringent environmental regulation) by setting up subsidiaries or other means of direct investment, so as to reallocate the production activities. The result supports Hypothesis 2.

4.3 Robustness Test

4.3.1 Parallel Trend Test

Satisfying the parallel trend is an essential premise of applying the DID model for causal inference. In other words, there is no difference in the trend of change between the treatment group and the control group before the policy implementation. This paper tests the parallel trend hypothesis with the event study methodology, with the model setup as follows:

(10)  Invest  c j t = α 0 + ∑ k = − 4 , k ≠ − 1 2 γ k D k ×  treat  c j + β X + μ c j + λ t + ε c j t

Wherein, D_k×treat_cj denotes the cross-product term between the dummy variable in the kth year before and after the policy implementation and the variable of whether to invest in region with low EPT rate. Taking the period before the policy implementation as the base period, compare the effects in each year before and after the policy implementation. The coefficients of the cross-product term in each year and their confidence interval boundaries at the level 95% are shown on the upper side of Figure 1. Before the policy implementation, there’s no significant difference in the number of trans-regional subsidiaries invested by listed companies in regions with relatively high or low EPT rates. However, beginning from the year of policy implementation, the number of subsidiaries in the treatment group (regions with relatively low EPT rates) has been significantly higher than that in the control group (regions with relatively high EPT rates). Such policy impact has increased year by year, thereby validating the parallel trend hypothesis.

Figure 1

Parallel Trend Test and Placebo Test

5.1 Heterogeneity Analysis

5.2 Extended Analysis: Economic Effect Test

As demonstrated above, the EPT reform exerts a significant positive impact on the migration of enterprises to areas with low EPT rates. The analysis into the economic effect brought by trans-regional migration of enterprises is of critical importance for green development of enterprises and green governance of government. Therefore, this paper carries out ex-post analysis into the cost-profit effect, innovation effect and environmental effect of trans-regional migration of polluting enterprises due to inter-provincial differences in EPT rates.