Background and Study Area

To control environmental pollution, the Chinese government has done a lot of work and formulated a series of environmental policies. Some environmental regulations and policies have also achieved certain results (Yu et al., 2019). For example, the State Council promulgated the “Two Control Zones” policy for 175 cities, and proposed phased control targets for the “Two Control Zones” in 1998. In 2005, the State Council of China promulgated the “Decision on Implementing the Scientific Concept of Development and Strengthening Environmental Protection” to incorporate environmental protection into the assessment indicators or KPI for local officials. Due to the complexity of environmental issues, the main responsibility of pollution control tends to be attributed to the government (Konisky, 2011). Firms with more pollution are often the main contributors to local GDP, employment and taxation. Therefore, prior to the implementation of restrictive pollution control, key polluting units were often tacitly approved or even protected by local governments to avoid the environmental regulations (Jia & Nie, 2017).

During the “Twelfth Five-Year Plan” period, the severe atmospheric environment and regional air pollution problems were becoming pressing issues. China's industrialization and urbanization were still in a period of rapid development, and resource and energy consumption kept increasing. Thus, in accordance with the Law of the People’s Republic of China on Air Pollution Control” and the “Outline of the Twelfth Five-Year Plan for National Economic and Social Development of the People’s Republic of China”, China formulated the “Twelfth Five-Year Plan for Air Pollution Prevention and Control in Key Areas” (the 12th plan hereafter), which was immediately put into effect. Based on geographical characteristics, socioeconomic development, air pollution, urban spatial distribution, and the air pollutants in the area, the 12th plan categorized cities in different areas into two groups: one is the planning area, namely Key Control Areas (KCA), the other is the general control areas (non-KCA). If a city was designated as a KCA city, by 2015, the total emissions of SO₂, nitrogen oxides, industrial smoke and dust must decline by 12%, 13%, and 10%, respectively. The average concentration of inhalable particulate matter (PM₁₀), sulfur dioxide, nitrogen dioxide, and fine particulate matter (PM_2.5) decreased by 10%, 10%, 7%, 5%. The annual average concentration of fine particulate matter in the Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta areas decreased by 6%. According to the “The KCA policy”, the goals, tasks, and governance were decomposed and implemented at the levels of municipal and county-level governments and related entities. The central government used rewarding polies such as “substituting awards for replacement”, “promoting prevention with awards” “promoting governance with awards”, etc. to support pollution-reduced projects.

Since the last two decades of 20th century, China's economy has continued to grow, and the city transformation has gradually become an inevitable trend. The economic growth in urban area can no longer be blindly and disorderly expanding, and wastewater and exhaust emissions were gradually reducing. From 2000 to 2017, the average GDP per square kilometers showed an upward trend, especially in the Yangtze River Delta and Pearl River Delta regions. Foshan, Shanghai, Changzhou, Yangzhou, and Wuhan had relatively rapid growth, respectively, up to $82,400/km², $36,450/km², $28,660/km², $25,900/km², $24,710/km². There were also areas where the relative growth rate was low, and even showed negative growth. For instance, Zhongshan was negative $13,160/km². The possible reason was that the GDP growth rate was slow, but the urban construction area growth was too high, then the relative growth rate of GDP per square kilometer got a negative growth. The regions with relatively low growth rates of GDP per square kilometer were mainly Bazhong, Nanyang, Zhoukou, and Huangshan, respectively, $1,460/km², $2,870/km², $3,040/km², $3,670/km².

From 2000 to 2017, the amount of SO₂ emissions basically showed a decreasing trend, especially in Chengdu, Laibin, Shanghai among others. The reductions in emissions were relatively large: 524,000 tons, 335,000 tons, and 314,000 tons, respectively. However, some cities, especially those in the western region, their emissions increased instead. Comparing to eastern coastal regions, those western cities performed much worse in economic development, so their environmental regulations were much looser. Therefore, the western region attracted a large number of polluting enterprises transferred from the eastern region (Shen et al., 2017). Emissions in cities in the Midwest of China have not decreased but increased in recent years. Among them, the emissions in Zunyi, Yuxi, Pu'er increased to 78,000 tons, 43,000 tons, 23,000 tons.

From 2000 to 2017, wastewater discharge showed a decreasing trend. The largest drop in emissions were Chongqing, Nanjing, and Shanghai, with 650 million tons, 490 million tons, and 400 million tons, respectively. The discharge of wastewater from Guangdong, Dongguan, Foshan, etc. has increased significantly, with increases by 1.489 billion tons, 787 million tons, and 258 million tons, respectively (Figure 1).

Figure 1.

Changes in cgdp, SO₂, and Waste Water Discharge of Tropical and Subtropical Cities in China From 2000 to 2017(Review Number: GS(2016)2929).

Figure 2.

Logic Diagram of the Transmission Mechanism of Environmental Regulation to Promote Urban Development and Transformation.

Hypothesis

The city transformation is mainly influenced by factors such as resource and environmental constraints, economic development stage, and technological progress. The development and transformation of a city depends on the changes in the allocation of factors within a city, such as transformation of the production model of enterprises and adjustment of urban industrial structure. The transformation of high-end and service-oriented directions has shifted from simply pursuing the expansion of the economic aggregate to a more scientific and adaptable model for urban development. Only the success of city transformation could bring the improvement of city efficiency.

As the main body of city production, firms' production models directly affect the development and transformation of cities. The strengthening environmental regulations would increase the cost of compliance (Kneller & Manderson, 2012). Since existing equipment, technology, and system have the “lock-in effect”, regulations would also increase the internal cost of firms. Appropriate but strict environmental regulations help to promote technological innovation, and the resulting first-mover advantage could compensate for environmental costs and achieve a win-win situation between economy and environment (Lee et al., 2011; Porter & van der Linde, 1995). Environmental regulations also motivate companies to increase their investment in fixed capital, which in turn leads to efficiency gains resulting from technological improvements (Yuan & Zhang, 2017). Simultaneously, the introduction of technology will bring a certain technology spillover effect (Takao & Noriaki, 2014). Certainly, internal adjustments brought about by environmental regulations would also have internal technology spillover effects as well (Galloway & Johnson, 2016). The compliance costs also made the environmental regulation have an extrusion effect on corporate investment, resulting in a decline in scale and output (Burton et al., 2011). Melitz and Polanec (2015) found that the dynamic effect of enterprise resource allocation would also squeeze some firm out of the market. Wang et al. (2019) found that environmental regulation mainly optimizes the allocation of resources among enterprises by affecting the entry and exit of firms. There is also a certain product market risk in the transfer of enterprise costs, such as higher product pricing may lead to consumption transfer or substitution, which is detrimental to a firm’s productivity (Clarkson et al., 2015). The impact of environmental regulations on enterprise production has not always been beneficial, Jiang et al. (2013) found that the relationship between environmental regulation and enterprise technological innovation presents a U-shaped pattern, that is, descending first and then rising. Albrizio et al. (2017) found that the increase in enterprise productivity may be contributed to environmental regulations that promote enterprise innovation, or it may be because environmental regulations have pushed resource allocation to tilt high-productivity firms (Figure 2).

Of the government's performance in environmental policies, studies often discussed industrial structure changes (Fan & Liu, 2017; Zheng & Shi, 2017). The improvement of environmental regulations would promote the R&D, environment-friendly and innovative technologies of enterprises, and improve the productivity of enterprises to further increase their performance (Feng et al., 2017; Kumar, 2019; Long & Wan, 2017). Rubashkina et al. (2015) suggest that “appropriate” environmental regulation is conducive to technological innovation and technological progress based on European manufacturing data. The strengthening of environmental regulations would lead to green technology-oriented innovation (Acemoglu et al., 2012). Environmental regulations can also encourage companies to discover new opportunities for innovation (Tang et al., 2016). An improvement in the intensity of environmental regulations was a compulsory “refinement cleaning” or selecting the superior firms and eliminating the inferior ones, leading to industrial transformation and upgrading (Berman & Bui, 2001). The KCA policy was concentrated in the secondary industry, by implementing stricter environmental access conditions in key control areas, setting special discharge limits for pollutants in key industries, taking stronger pollution control measures, and speeding up the elimination of outdated production process. The KCA policy strictly reviewed investment projects involved with environment protection and suspended the approval and filing procedures for construction projects in key industries such as thermal power, steel, non-ferrous petrochemicals, cement, and chemical industries in areas if their elimination tasks were not completed on time. For firms that were not eliminated on schedule, the pollutant discharge permits and production permits shall be revoked according to the policy (Figure 2).

Data and DID Method

We complied a panel dataset including 155 tropical and subtropical cities in China from 2000 to 2017 from the “China City Statistical Yearbook”, “China City Construction Statistical Yearbook”, “China Regional Economic Statistical Yearbook”, and the China Economic Information Network statistical database. For those missing values, we searched the regional city statistical yearbooks and statistical bulletins to replenish them. Since the observations last 18 years, to consider the impact of inflation, all nominal variables were adjusted using the provincial consumer price index (Li et al., 2016).

Our study applies a simple difference-in-differences (DID) method. DID is an econometric method based on experimental design or quasi-experimental design, which uses the idea of random experiment as the basis of causal effect and could effectively solve the endogeneity problem in regression (Angrist & Pischke, 2008; Ashenfelter, 1978). The implementation of KCA policy targets only the KCA cities, which provides a “perfect” natural experiment. So, the treatment group includes those designated KCA cities in 2012, and control group is consisting of non-KCA cities. The regression model is specified as follows:

where Y_itis the measurement of development transformation flows in city i at year t;

KCA_i = 1 If i is a KCA city and = 0 if a non KCA city and post_t indicates the post-treatment period, i.e., 2012 and = 0 otherwise η_i and γ_t are city fixed effects and time fixed effects, respectively, capturing city i' s time-invariant characteristics such as geographic features, climate, natural endowment and year-invariant factors. ε_it is the error term. β₁ represents the effect of implementing KCA policy.

Dependent variable includes GDP per square kilometers (cgdp), sulfur dioxide emissions (SO2) and wastewater discharge (water), measuring the development transformation of cities in China.

X_itrepresents control variables. Specifically, the first one is urban fixed asset investment. Since reduction of fixed asset investment improves air quality (Li & Zheng, 2016), as a part of city construction it might relate to urban development transformation. The second control variable is the index of openness to the outside world (fdi). Foreign direct investment as part of urban development, our measurement is consistent with Chen and Chen (2018) that uses the ratio of the level of foreign direct investment to the actual GDP of the city’s jurisdiction. Because municipal governments in China play an important role in guiding urban industrial development and urban transformation (Zhou et al., 2015), so we select the ratio of the government's general budget expenditure to the population (govpo) as a control as well. The public services would affect the convenience and quality of life for urban residents through public education resources, comprehensive medical and health services, and transportation and communication facilities, thereby affecting the distribution of the population (Da Silva et al., 2017), so we control the level of public services by including log number of teachers in higher education, hospital beds per 10,000 population, and public shutter bus numbers per 10,000 population. we also control for production efficiency of industrial enterprises, a measurement of output value of enterprises above designated size/number of enterprises greater than designated size. The pattern of industrial development shows that the three industrial structure evolutions could be divided into two stages (Herrendorf et al., 2014). The advanced stage is the service-oriented process. The service industry gradually replaces manufacturing industry and becomes the largest industry in the national economy (Gao, 2019), so we use the output value of the tertiary industry over secondary industry to measure the urban industrial structure. In order to make the data stable and address the collinearity and heteroscedasticity, we take log values of Fai, Fdi, Teacher, Industry. Table 1 shows the summary statistics of major variables and interpretation of variables.

Table 1.

Summary Statistics and Description of Variables.

Pre-Policy Common Trend Illustration

In Figure 3, both the treatment group and the control group show similar trend for all the three measures of dependent variables in the period 2000–2012. After 2012, however, the treatment group and the control group illustrate significant differences. It can be seen that the economic status of the treatment group is better than the control group, and the urban transformation and development of the treatment group is faster than the control group. The purpose of the preliminary illustration is to enhance the intuitive experience. The following part will also make a more accurately empirical test on the hypothesis of common trends.

Figure 3.

cgdp, SO₂, Water Parallel Trend Test.

Main Results

As an important part of the “Twelfth Five-Year Plan”, the KCA policy inevitably has a certain impact on the development and transformation of cities. The division of air pollution prevention and control in key areas provides a natural experiment for the study.

In Table 2, the regression (1), (3), (5) have no time fixed effects and city fixed effects, which shows that comparing to those non-KCA cities, those KCA-cities endure significant effects from the policy after its implementation in 2012. Similar effects still hold in regression (2), (4), (6) with time fixed effects and city fixed effects. It indicates that the KCA policy brings about the growth of cgdp and reduction of SO₂ and waste water emission, consistent with the discussion in our Hypothesis1–the strict environmental regulations promote city transformation. Fixed asset investment promotes the city transformation, but the impact becomes insignificant after adding the fixed effects of cities and periods. The ability of cities to attract foreign direct investment has no significant impact on the city transformation. Government's emphasis on urban development has a significant positive impact on the city transformation when time and regional fixed effects are not included, but has a significant negative impact on city transformation after taking in time and city fixed effects.

Table 2.

Basic Regression.

Heterogeneity Analysis

Cities vary due to their heterogeneity of the economic base, factor endowment structure, geographical environment, among others. In 1998, 175 cities in China were designated into Two Controlled Zones(TCZ) based on their severity of SO₂ and acid rain pollution, and the government applied a much draconian environmental policies in both controlled zones due to their serious pollution problems. The regression results in Table 3 show that the impacts of environmental regulations on GDP per square kilometer are significant in both TCZ and non-TCZ. The coefficients for discharge of sulfur dioxide and waste water in the TCZ are significantly positive, but they are not significant when a city belong to a non-TCZ, which implies that pollution status or climate is closely connected to urban development (Mi et al., 2019).

Table 3.

Heterogeneity Analysis1.

In addition, residents are the basic and major activity subject of a city, and they are producers and consumers. Human activities affect the implementation of environmental regulations and the city transformation. Therefore, we did the following heterogeneity analysis: identify whether a city belongs to TCZ, and identify whether a city is a small (population less than 1 million) or a medium-sized (population more than 1 million) city according to the 2012 population data.

The impacts of environmental regulations on city transformation by population size are more significant when environmental regulations are further strict in cities with larger populations. The impacts on GDP per square kilometer are greater in large cities than in smaller ones. The impacts on sulfur dioxide and waste water discharge are not significant in small-scale cities but significantly negative in large-scale cities which means significant suppression of sulfur dioxide emissions. The impacts of setting of key pollution areas on waste water discharge are greater in large-scale cities (Table 4).

Table 4.

Heterogeneity Analysis2.

Robustness Check

Based on above results, the strengthening of environmental regulations could inevitably promote the transformation of cities. However, this conclusion may still be affected by missing variables. In order to verify the robustness of the DID identification strategy, we conduct a robustness analysis as follows.

In order to further test the common trends prior to the implementation of environmental regulation in 2012 and observe whether the policy has a time lag effect, we apply an event study to discover the dynamic effects of KCA. The research framework is from Li et al. (2016). More specifically, we replace DID variable in Equation (1) with the a few time dummies including 9 years before and 5 year after the implementation.

D₀ is a dumb variable in the year when the KCA policy was implemented,

s<0, these are the years before policy implementation,

s > 0, these are the years after policy implementation.

Figure 4 reports the estimated parameters with the corresponding 90% confidence interval. Figure 4 reports the dynamic effects of the estimated coefficients of the explained variables cgdp, SO₂ and water respectively. The estimation results not only verified the common trends, but also showed that the policy have positive impacts on urban transformation. It has a certain continuity, including promoting the growth of GDP per square kilometer and reducing the discharge of sulfur dioxide and wastewater.

Figure 4.

The Dynamic Impact of Environmental Regulations (cgdp, SO₂, Water).

In order to ensure the robustness of the above regression results, this paper further uses the PSM-DID method to analyze the policy effects of the establishment of atmospheric prevention and control planning in key regions. For comparison, we apply above control variables to predict the probability of air pollution prevention and control in each city as a key area based on a Logit regression. Then, kernel matching, radius matching, and the nearest neighbor matching methods are applied to match the samples (treatment groups) of air pollution control in the key areas with the control group, and the results show that there is no significant difference between the treatment group and the control group before the establishment of the key prevention and control area. Therefore, the DID method was used to identify the net impact of the division of key areas on city transformation of development. Because the propensity score can solve the deviation of the observable covariate to the greatest extent, and the difference-in-difference method can eliminate the effect of unobserved variables such as time invariance and time synchronous change, so the combination of the two methods can better Identify policy effects. The regression results are shown in Table 5. The estimation results are consistent with the benchmark regression, and the establishment of air pollution prevention and control in the key areas estimated in this paper is stable to promote the transformation of city development (Table 5).

Table 5.

Robustness Test Results1.

Our second robustness check is to change sample period. The regression in this article is mainly based on the sample of tropical and subtropical cities from 2000 to 2017, and the policies for the prevention and control of atmospheric pollution in key regions occurred in 2012, so the period before the policy may be too long. Moreover, in order to avoid the impact of the 2008 financial crisis, we selected a new sampling period 2009 to 2017, that is, three years before and five years after the reform. As shown in columns (4)–(6) of Table 6, the empirical results are still consistent with the baseline regression.

Table 6.

Robustness Test Results2.

The third is to adjust the sample size. In baseline regression, cities with good economic development in both treatment group and control group, such as cities in the Yangtze River Delta and the Pearl River Delta, have higher level of industrialization in the post-industrial period due to their rapid economic development. Many policies from central government would be tilted to them somehow, which could lead to biased results. So, we remove cities in the Yangtze River Delta and the Pearl River Delta, and run the DID estimation again. The regression results are shown in Table 6. Compared with the baseline regression, the results are still significant.

The last is to add lagged one period of control variables. Considering that there may be a reverse effect between the selected variables and the establishment of air pollution control areas in key areas, to identify such a potential endogenous problem, all control variables are lagged by one period. The empirical results are shown in columns (7), (8) and (9) in Table 6. Results still hold similar to the baseline regression results.

The Mechanism of Environmental Regulation Promoting City Transformation

To better implement the environmental regulation and test the impact of environmental regulation on the urban transformation, the implementation of environmental regulations mainly promotes the change of business and industrial. The strict environmental regulations promote City Transformation, which internalizes production cost, increases firm’' technology substitution to adjust to the implementation. Some firms increase technology investment and expand business, while other firms might not be able to afford the extra cost and choose to reduce production and even shut business. City transformation is a function of industrial restructuring, that is, as the implementation of environmental regulation, secondary industrial activities reduce, service activities increase. The manufacturing upgrading and service industries expansion are the result of emission limitation, then the declining of relative proportion of manufacture and the industrial structure has been improvement, which ultimately promotes the city transformation. By “manipulating” the data, we try to identify the intermediate mechanism and transmission process in which policies affect urban transformation.

The empirical results of the first step are shown in Table 7. Column (1) reports the policy effects in product market. The establishment of air pollution in key regions has increased the productivity. Column (5) shows that the implementation of key regions promotes the increase in the ratio of tertiary industry to secondary industry. In the second step, to explore whether the policy has indirectly promoted economic performance through the construction of the corresponding key areas, we refer to the intermediary effect test method proposed by Wen et al. (2004).On the basis of the first step of identifying policy effects, we construct an intermediary effect test model. The results show that it meets the expectations of the intermediary variables. Curitiba has proved to the world with its achievements, that is, economy and environment are not in conflict, and survival and life are not in conflict. 80% of the employed population in Curitiba is invested in the service industry, especially the pollution-free and technology-intensive information service industry. Curitiba has the highest human development index in Brazil and becomes one of the most livable cities in the world (Deng, 2015; Gustafsson & Kelly, 2012). Economic development and environmental governance are not contradictory, and they have even achieved rapid development because of environmental governance.

Table 7.

Mechanism Validity Test.