International Studies of Economics
Volume 19, Issue 3 pp. 330-352
RESEARCH ARTICLE
Open Access

Ecology versus economic development: Effects of China's Yangtze River Economic Belt strategy

Guan Gong, 

Guan Gong

School of Economics, Key Laboratory of Mathematical Economics (SUFE), Ministry of Education, Shanghai University of Finance and Economics, Shanghai, China

Contribution: Funding acquisition, Resources, Writing - review & editing

Search for more papers by this author
Yu Zhao, 

Corresponding Author

Yu Zhao

School of Economics, Shanghai University of Finance and Economics, Shanghai, China

Correspondence Yu Zhao, School of Economics, Shanghai University of Finance and Economics, Shanghai 200433, China.

Email: [email protected]

Contribution: Data curation, Software, Writing - original draft

Search for more papers by this author
First published: 18 February 2024
https://doi.org/10.1002/ise3.75
Citations: 6

Abstract

This study employs China's Yangtze River Economic Belt strategy as a quasi-natural experiment to investigate the impact of prioritizing green development on economic growth. Our empirical findings show that the strategy significantly reduces urban industrial wastewater discharge. It helps transition the region's industries towards technology-driven service sectors while maintaining a steady economic growth rate. On average, cities in the Yangtze River Economic Belt see a 21.9% decrease in annual industrial wastewater discharge, a 1.9% increase in economic growth rate, a 4.9% rise in the proportion of service industries' contribution to GDP, and a 2.4% increase in the number of employees in productive service industries. Moreover, our empirical results highlight the heterogeneity in the effects of the strategy across different regions, which can be attributed to factors such as population density, infrastructure, levels of human capital, and government governance. The implementation of the Yangtze River Economic Belt strategy offers valuable insights for developing countries on how to balance between economic development and environmental protection.

1 INTRODUCTION

China is presently placing significant emphasis on prioritizing the advancement of green development to achieve a dual objective of environmental protection and economic growth. However, research on environmental regulation policies has demonstrated the persistent challenge faced by government in balancing economic development and environmental protection (Chen et al., 2018; Greenstone, 2002; Li, Qiao, et al., 2019; Lin & Zhu, 2020; Song, Wang, et al., 2018; Zhang et al., 2017). The reduction of pollution often entails significant economic costs, as evidenced by studies (He et al., 2020). In the case of developing countries, which heavily rely on low-end manufacturing industries characterized by high pollution emissions (Ebenstein et al., 2017; Greenstone & Hanna, 2014), a common approach to attract foreign direct investment is to relax environmental regulations (Yao et al., 2018). However, intensified pollution control measures may have adverse effects on economic development (Hwang & Kim, 2017; Pang et al., 2019). Empirical investigations of environmental policies in China have also indicated the potential for unintended economic consequences resulting from environmental regulation (Pang et al., 2019). Given the intricate relationship between economic development and environmental protection, it remains uncertain whether green development can effectively achieve both pollution mitigation and economic development in developing countries.

There exist two primary perspectives regarding the relationship between ecological governance and economic growth. The first perspective is the “cost-effectiveness” view, which maintains that environmental regulations raise enterprise production costs, reduce production efficiency, and consequently hinder economic growth (Chintrakarn, 2008; Jaffe & Palme, 1997). The second perspective is the “innovation compensation effect,” which posits that suitable environmental measures can exert external pressure on enterprise technological innovation. This not only mitigates the negative impacts of the “cost-effectiveness” but also stimulates technology diffusion and facilitates structural upgrading, ultimately fostering economic growth (Mazzanti & Zoboli, 2009). As a result, a disparity exists between government's economic growth objectives and the goals in environmental protection.

This study aims to investigate the relationship between environmental protection and economic development by examining the Yangtze River Economic Belt strategy in China. Our objective is to evaluate the effectiveness of the strategy in reducing pollution and fostering economic growth. Moreover, we seek to identify the cities that have experienced positive outcomes from this strategy and explore the factors that contribute to variations in green development performance.

Environmental protection is a major concern for the Chinese government, and they have implemented various polices to address it (Mu et al., 2014). However, previous research has shown that economic development often takes precedence over environmental concerns, leading to ongoing environmental problems and disagreements between local and central governments (He et al., 2020; Li et al., 2020; Lin & Zhu, 2020). In response, the Chinese government introduced the “Yangtze River Economic Belt” strategy in 2014, which places a paramount focus on restoring the environmental and controlling pollution. Encompassing 21.4% of China's total area and housing 40% of the nation's population and GDP, this region stands densely populated and harbors a notable concentration of chemical industries, including traditional sectors like oil, gas, and steel.

Being the primary locus for China's ambitious green development, the Yangtze River Economic Belt provides an exceptional milieu for examining the intricate interplay between ecological governance and economic development in developing countries. The implementation of this strategy initiative presents a propitious opportunity to appraise the efficacy of green development pursuits and gain invaluable insights into the prospects of balancing environmental and economic imperatives.

Following the implementation of the Yangtze River Economic Belt strategy, several studies have investigated regional development using diverse analytical frameworks. These investigations have examined aspects such as the level of green development (Guo et al., 2021; Hu et al., 2021), ecological efficiency (Liu et al., 2020), water pollution levels (Li & Lu, 2020), local ecosystem services (Luo et al., 2019), and green innovation. However, there is a dearth of research that quantitatively evaluate the effects of the policy. Previous quantitative evaluations of policy effects in the Yangtze River Economic Belt primarily rely on the difference-in-differences (DID) and synthetic control methods (SCM) to evaluate the average effect of policy implementation on pollution emissions, green total factor productivity, and economic growth at the regional level. Nevertheless, there has been a lack of exploration into the heterogeneous effects of the policy at the city level, leaving uncertainties regarding the achievement of the dual benefits of environmental governance and economic development.

We employ a panel data program evaluation method devised by Hsiao et al. (2012) to assess the heterogeneous impacts of the Yangtze River Economic Belt strategy on urban green development. Our analysis utilizes data from the China Urban Statistical Yearbook (1995–2019) and examine the effects from both ecological governance and economic development perspectives. Additionally, we employ a two-way fixed-effects model to examine the factors influencing variations in urban green development outcomes.

Our empirical study reveals significant reductions in industrial wastewater emissions in the middle and upper reaches of the Yangtze River following the implementation of the strategy. The Chengdu-Chongqing urban agglomeration and the middle reaches of the Yangtze River urban agglomeration have experienced an average annual decrease of over 12% in industrial wastewater discharge, while the impact in the Yangtze River Delta urban agglomeration is insignificant.

Furthermore, our findings indicate that ecological governance, as reflected by the reeducation in industrial wastewater emissions, has had no substantial impact on urban economic development. Among cities where industrial wastewater discharge has decreased significantly, more than 90% have not experienced negative changes in economic growth and industrial structure. Additionally, over 45% of these cities have witnessed positive outcomes in their industrial structure, indicating a successful transition toward a more sustainable and efficient industrial composition. Notably, the Chengdu-Chongqing urban agglomeration has demonstrated a clear shift toward the service industry and an increase in high technical productivity through the engagement of productive service workers. However, no significant industrial structure upgrading has been observed in the urban agglomerations in the middle reaches of the Yangtze River and the Yangtze River Delta.

Our analysis reveals heterogeneity in the effectiveness of green development across cities within city clusters. Further regression analysis indicates that cities with higher population density, better information technology infrastructure, superior supporting infrastructure, and more abundant human capital have benefitted more from the strategy. These factors indicate that cities with a strong foundation in terms of population, technology, and infrastructure have been better positioned to leverage the environmental measures and drive sustainable development.

Our paper contributes to the literature on green development in three main ways. First, while previous research mainly focuses on analyzing the effectiveness of policies at the regional and provincial levels, we conduct a more detailed analysis at the municipal level. This provides new evidence for the effectiveness and heterogeneity evaluation of policy effects, filling the gap in current research. Second, unlike existing studies that measure policy impacts using a single indicator, this paper evaluates policy effects from three dimensions: ecological governance, economic growth, and industrial structural adjustments. It conducts a more detailed exploration of policy impacts through a multidimensional analysis. Last, we investigate the factors influencing urban policy heterogeneity, offering valuable insights for local governments to address the imbalance in regional development.

The rest of the paper is structured as follows. Section 2 lays out the background of the strategy, while Section 3 presents the identification strategies. Section 4 details the empirical results, and Section 5 introduces the analysis of factors influencing the heterogeneous policy effects. Section 6 includes the robustness tests, and finally, Section 7 concludes the paper.

2 THE YANGTZE RIVER ECONOMIC BELT STRATEGY

2.1 Background

The Chinese government recognizes the significant conflict between economic development and environmental protection and has made efforts to address environmental issues (Deng et al., 2019; Li, Qiao, et al., 2019). Initially, environmental governance in the Yangtze River Economic Belt was primarily led by the central government. However, the existence of information asymmetries between the central and local governments resulted in unintended consequences for the economy (Yao et al., 2018). To leverage the information advantages of local governments in environmental protection, the central government pursued decentralization of environmental protection authority and integrated environmental responsibility into the evaluation and promotion system of local officials (Li, Qiao, et al., 2019; Lin & Zhu, 2020; Yu et al., 2019). Nevertheless, due to regional competition, promotion incentives, and insufficient supervision, the objectives of the central and local governments often diverge, with local economic development taking precedence over environmental protection.

In 2014, the government introduced the Yangtze River Economic Belt strategy, designating it as an “early demonstration belt for ecological civilization construction.” The Yangtze River Economic Belt is a major national strategic development region established along the Yangtze River (see Figure 1), covering 11 provinces and cities including Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Hubei, Hunan, Chongqing, Sichuan, Yunnan, and Guizhou. This strategic positioning emphasized the prioritization of the ecological environment in the region's development. Additionally, as part of this strategy, there were efforts to decentralize environmental protection authority and involve local governments in environmental governance. In subsequent years, multiple symposiums were organized to reiterate this priority and reinforce the significance of ecological conservation in the Yangtze River Economic Belt.

Details are in the caption following the image
Figure 1
Open in figure viewerPowerPoint
The Yangtze River Economic Belt. The red line represents the course of the Yangtze River, the black portion symbolizes the 11 provinces (and cities) along the Yangtze River Economic Belt, and the blue segment depicts the coastal region.

The Yangtze River Economic Belt Development Strategy distinguishes itself from other environmental protection policies by emphasizing its focus on the “ecological aspect.” The government has presented a range of proposals to prioritize ecological preservation and foster high-quality development in the region. The strategy consists of three main components:

Effective control and efficient management of water pollution in the Yangtze River. This involves the implementation of rigorous measures to curtail the discharge of pollutants into rivers or lakes. This initiative also focuses on enhancing the management of industries that contribute to pollution, including chemical, papermaking, printing and dyeing, and nonferrous industries located along the river. Additionally, the strategy includes developing a mechanism to either relocate or transform industrial parks posing significant environmental risks, thereby mitigating the emission of toxic and hazardous pollutants.

Facilitation of industrial transformation and upgrading. To align with the prevailing global trends in industrial transformation, there is an emphasis on actively nurturing emerging industries, expediting the transformation of traditional industries, eliminating outdated production capacities, and addressing pollution transfer and environmental risk accumulation. Furthermore, there is a vital push to increase the service industry's share substantially, establishing an industrial framework centered around the service sector in pivotal cities within the region.

Acceleration of the modern service industry's development. This strategy prioritizes the growth of productive service sectors, including finance and insurance, energy conservation and environmental protection, modern logistics, and shipping services. These measures aim to contribute to the growth of the real economy.

By the end of 2018, significant achievements had been made in environmental and economic aspects. Along the river, 1772 sewage treatment facilities were constructed, 6485 chemical enterprises were relocated or closed down, and the volume of odorous water in 12 key cities was reduced by 93.1%, effectively curbing water pollution. In terms of economic performance, the 11 provinces and cities in the region contributed to 44.76% of the national GDP. The GDP growth rate reached 7.7%, surpassing the national average by 1.1 percentage points, and the growth rate of industrial value added exceeded the national average by 0.72 percentage points.

2.2 Data

2.2.1 Sources

The data used in this study is extracted from the “China City Statistical Yearbook” spanning the years 1995 to 2019. To ensure a more granular analysis of urban changes and avoid smoothing effects on trends, the study analyzes the city-level data rather than larger research units such as provinces or regions. Due to the unavailability of pertinent data on industrial wastewater discharge and productive service workers before the year 2003, the research periods for the four variables analyzed in this study slightly vary.

It is worth noting that the GDP data used in this paper has been adjusted using the reduced index at the provincial level as presented in the “China Statistical Yearbook,” with the reference year being 1994. The productive service industry includes five categories of sectors: transportation, warehousing, postal, and telecommunications services; information transmission, computer services, and software industry; financial industry; leasing and business service industry; scientific research, technical services, and geological exploration sector.

2.2.2 Variables

This study focuses on evaluating the efficiency of green development in Yangtze River Economic Belt, specifically in terms of environment protection and economic growth. The assessment of the Yangtze River Economic Belt Strategy is based on its policy contents, considering outcome variables within three distinct areas: ecological governance, economic growth, and industrial structural adjustment. The selection of variables for analysis is structured to capture and delineate these three aspects.

Water pollution is a significant ecological governance issue in the development of the Yangtze River Economic Belt. As early as 2012, over 30 billion tons of wastewater had been discharged into the Yangtze River, and the majority of it was industrial wastewater. Therefore, industrial wastewater discharge is selected as the environmental protection measure for this study.

To examine the impact of prioritizing ecology in the Yangtze River Economic Belt, relevant data sources are used for analyzing the relationship between industrial wastewater discharge and economic indicators, as well as the progress made in adjusting the industrial structure.

This paper evaluates the economic development of cities by considering relevant variables associated with actual GDP changes and industrial structure adjustment. Two indicators are selected to measure industrial structure adjustment: the proportion of service industry output value in regional GDP and the proportion of productive service workers in service industry personnel.

Previous research has indicated that the movement of labor towards productive services and high-end service industries can contribute to social productivity growth. Therefore, the proportion of productive service workers not only reflects the improvement in the structure of service industry personnel but also serve as an indicator of changes in regional productivity level and economic efficiency. By analyzing the proportion of productive service workers, we can gain insights into the dynamics of regional economic development and the effective utilization of labor in various sectors.

2.2.3 Summary statistics

Table 1 shows the descriptive statistics of variables. It is evident that significant differences exist among the three urban agglomerations. The Yangtze River Delta has the largest economy, with the highest proportion of service industry output and the most optimized service industry personnel structure. Additionally, it has the highest industrial wastewater discharge. The average GDP of cities in the middle reaches of the Yangtze River urban agglomeration is similar to that of the Chengdu-Chongqing urban agglomeration. However, the middle reaches of the Yangtze River have a higher discharge of industrial wastewater, a greater proportion of service industry output, and a greater proportion of productive service workers compared to Chengdu-Chongqing.

Table 1. Descriptive statistics of variables for three urban agglomerations.
Urban agglomeration Mean Std. Min Max
Industrial wastewater discharge (log million tons) (2003–2018) Chengdu-Chongqing 3.574 0.391 2.895 4.518
Mid-Yangtze River 3.800 0.295 2.823 4.539
Yangtze River Delta 4.134 0.363 3.309 4.933
Real GDP (log million yuan) (1994–2018) Chengdu-Chongqing 6.690 0.485 5.781 8.076
Mid-Yangtze River 6.644 0.479 5.395 7.947
Yangtze River Delta 7.016 0.463 5.523 8.032
Proportion of service industry output (%) (1994–2018) Chengdu-Chongqing 34.420 7.642 20.900 54.120
Mid-Yangtze River 36.580 6.589 21.800 55.330
Yangtze River Delta 38.718 7.523 22.310 63.900
Proportion of productive service workers (%) (2003–2018) Chengdu-Chongqing 21.857 6.244 11.425 34.925
Mid-Yangtze River 22.490 5.367 13.120 37.080
Yangtze River Delta 28.510 4.872 20.300 44.880
  • Note: The variables “Industrial wastewater discharge” and “Real GDP” are expressed in logarithmic levels. The variable “Proportion of productive service workers” is represented as the percentage of the total number of workers in the service industry.

3 ESTIMATION STRATEGY

The key to evaluating the effectiveness of policy implementation lies in constructing a counterfactual scenario without policy implementation. Common methods to estimate counterfactual values include differences-in-differences (DID) model and propensity score matching (PSM). In the case of the Yangtze River Economic Belt strategy, the decision to implement it is influenced by a range of factors, including geographical advantages, resource concentration, population, market dynamics, industries, and the presence of cities along the Yangtze River. It is important to consider these factors when selecting the treatment group for analysis, as different cities within and outside the region may have distinct natural resource endowments, inland freight capabilities, geographical features, strategic locations, and social advantages. Ignoring these factors can lead to biased estimation outcomes. Therefore, it becomes crucial to identify appropriate proxy variables that can capture these hard-to-measure factors such as social endowment and natural resource endowment. However, selecting suitable proxy variables presents its own challenging.

To address these challenges, Hsiao et al. (2012) propose a comprehensive solution that leverages interindividual correlations to construct counterfactual trajectories, eliminating the necessity of selecting unobservable variables. This method enables the measurement of policy effects at the individual level and facilitates the examination heterogeneity in those effects.

3.1 Panel data program evaluation method

We employ the method introduced by Hsiao et al. (2012). Herein, we provide an overview of its setup and implementation.

3.1.1 Settings

Consider a vector y t = ( y 1 t , … y Nt ) ′ ${y}_{t}={({y}_{1t},\ldots {y}_{{Nt}})}^{^{\prime} }$ with dimensions N × 1 $N\times 1$ . In this context, assume there exists a city within the area where the development strategy is being implemented. Without loss of generality, let's designate this city as the first one. Here, y 1 t ${y}_{1t}$ denotes the outcome variable of city 1 at period t, while y 2 t , … y Nt ${y}_{2t},\ldots {y}_{{Nt}}$ represent the outcome variables of the other N − 1 $N-1$ cities outside the implementation area.

Let y 1 t 1 ${y}_{1t}^{1}$ and y 1 t 0 ${y}_{1t}^{0}$ denote the outcome variables of city 1 when the strategy is implemented and when it isn't, respectively, during period t. It is important to note that we cannot observe both y 1 t 1 ${y}_{1t}^{1}$ and y 1 t 0 ${y}_{1t}^{0}$ simultaneously. Instead, the observed values are expressed as: y 1 t = d 1 t y 1 t 1 + ( 1 − d 1 t ) y 1 t 0 ${y}_{1t}={d}_{1t}{y}_{1t}^{1}+(1-{d}_{1t}){y}_{1t}^{0}$ , where d 1 t ${d}_{1t}$ equals 1 when the implementation is in effect at period t, and d 1 t ${d}_{1t}$ equals 0 otherwise. This formulation allows us to capture the treatment effect and estimate the outcomes based on whether the strategy is implemented or not.

Let's assume that the development strategy is implemented starting from period T 1 + 1 ${T}_{1}+1$ , and for period t ( t ≥ T 1 + 1 $t\ge {T}_{1}+1$ ), the treatment effect of the strategy for city 1 can be represented as
△ 1 t = y 1 t 1 − y 1 t 0 . ${\bigtriangleup }_{1t}={y}_{1t}^{1}-{y}_{1t}^{0}.$ ()
To estimate the unobserved counterfactual value y 1 t 0 , ${y}_{1t}^{0},$ the model for y it 0 ${y}_{{it}}^{0}$ is based on the following the data generating process (DGP):
y it 0 = α i + b i ′ f t + u it i = 1 , … , N and t = 1 , … , T , ${y}_{{it}}^{0}={\alpha }_{i}+{b}_{i}^{^{\prime} }{f}_{t}+{u}_{{it}}\,i=1,\ldots ,N\,\text{and}\,t=1,\ldots ,T,$ ()
where α i ${\alpha }_{i}$ denotes individual fixed effect, f t ${f}_{t}$ denotes K × 1 $K\times 1$ unobserved common factors, b i ${b}_{i}$ denotes K × 1 $K\times 1$ factor loading matrix of constant varied across i $i$ , and u it ${u}_{{it}}$ denotes random idiosyncratic component . $.$
Pooling the data from N $N$ cities yield:
y t 0 = α + B f t + u t , ${y}_{t}^{0}=\alpha +B{f}_{t}+{u}_{t},$ ()
where y t 0 = ( y 1 t 0 , … , y Nt 0 ) ′ ${y}_{t}^{0}={({y}_{1t}^{0},\ldots ,{y}_{{Nt}}^{0})}^{^{\prime} }$ , α = ( α 1 , … , α N ) ′ $\,{\alpha =({\alpha }_{1},\ldots ,{\alpha }_{N})}^{^{\prime} }$ . B = ( b 1 , … , b N ) ′ ${B=({b}_{1},\ldots ,{b}_{N})}^{^{\prime} }$ is an N × K coefficient matrix that allows for differential city-level responses to the common factors; u t = ( u 1 t , … , u Nt ) ′ ${u}_{t}={({u}_{1t},\ldots ,{u}_{{Nt}})}^{^{\prime} }$ is an N × 1 heteroscedastic error term matrix.
Before the strategy implementation:
y t = y t 0 = α + B ′ f t + u t i = 1 , … , N and t = 1 , … , T 1 . ${y}_{t}={y}_{t}^{0}=\alpha +{B}^{^{\prime} }{f}_{t}+{u}_{t}\,i=1,\ldots ,N\,\text{and}\,t=1,\ldots ,{T}_{1}.$ ()
After the strategy implementation:
y 1 t = y 1 t 1 = y 1 t 0 + Δ 1 t = α 1 + b 1 ′ f t + Δ 1 t + u it t = T 1 + 1 , … , T . ${y}_{1t}={y}_{1t}^{1}={y}_{1t}^{0}+{{\rm{\Delta }}}_{1t}={\alpha }_{1}+{b}_{1}^{^{\prime} }{f}_{t}+{{\rm{\Delta }}}_{1t}+{u}_{{it}}\,t={T}_{1}+1,\ldots ,T.$ ()
For the control group:
y it = y it 0 = α i + b i ′ f t + u it i = 2 , … , N and t = T 1 + 1 , … , T . ${y}_{{it}}={y}_{{it}}^{0}={\alpha }_{i}+{b}_{i}^{^{\prime} }{f}_{t}+{u}_{{it}}\,i=2,\ldots ,N\,\text{and}\,t={T}_{1}+1,\ldots ,T.$ ()
The method proposed by Hsiao et al. (2012) on identifying the treatment effect relies on five key assumptions:
  • 1.

    Each city is affected by common factors f t ${f}_{t}$ , though the response to f t ${f}_{t}$ might vary across cities.

  • 2.

    The error term u t ${u}_{t}$ is an independent and stationary sequence with E ( u t ) = 0 ${E(u}_{t})=0$ and for any j ≠ i $j\ne i$ , u it ${u}_{{it}}$ is uncorrelated with u jt ${u}_{{jt}}$ .

  • 3.

    The error term u t ${u}_{t}$ is uncorrelated with the common factor f t ${f}_{t}$ .

  • 4.

    The number of observable cross-sectional units, N $N$ , is greater than the number of common time-varying factors, f t ${f}_{t}$ .

  • 5.

    The policy under assessment does not have an impact on the control group cities.

Among above assumptions, assumption five holds particular sensitivity for our identification. This assumption assumes the treatment effect doesn't extend to the control group, as referenced in prior works (Ke & Hsiao, 2021; Ke et al., 2017). To address this issue, we will select cities positioned away from the vicinity of the Yangtze River Economic Belt implementation area. Subsequently, we will rigorously test the robustness of our findings by using different control groups.

3.1.2 Treatment effect estimation and control group selection

In situations where the data's time series and cross-sectional dimensions are sufficiently large, the number of common factors K can be determined using the methods proposed by Bai and Ng (2002) and Bai and Li (2012). Estimation of f t ${f}_{t}$ and B $B$ can then be achieved through the maximum likelihood method. However, in cases like the Yangtze River Economic Belt, where the sample data fails to meet these requirements, an alternative approach introduced by Hsiao et al. (2012) can be employed. This approach involves forecasting the counterfactual values of y 1 t 0 ${y}_{1t}^{0}$ for city 1 postpolicy implementation by utilizing information from ( y 2 t , … , y Nt ) $({y}_{2t},\ldots ,{y}_{{Nt}})$ instead of estimating the unobservable factors f t ${f}_{t}$ and B $B$ (Bai et al., 2014; Ching et al., 2012; Du & Zhang, 2015).

Let y − 1 t = ( y 2 t , y 3 t , … , y N − 1 , t , y Nt ) ′ ${y}_{-1t}=({y}_{2t},{y}_{3t},\ldots ,{y}_{N-1,t},{y}_{{Nt}})^{\prime} $ , then the counterfactual value y 1 t 0 ${y}_{1t}^{0}$ can be expressed as:
y 1 t 0 = E y 1 t 0 | y − 1 t + ε 1 t , ${y}_{1t}^{0}=E\left({y}_{1t}^{0}|{y}_{-1t}\right)+{\varepsilon }_{1t},$ ()
where ε 1 t ${\varepsilon }_{1t}$ is the error term.
From Equation (3),
y 1 t 0 = α + b ′ y − 1 t + u 1 t − b ′ u − 1 t , ${y}_{1t}^{0}=\alpha +b^{\prime} {y}_{-1t}+{u}_{1t}-b^{\prime} {u}_{-1t},$ ()
where b = ( b 2 , … , b N ) ′ $b={({b}_{2},\ldots ,{b}_{N})}^{^{\prime} }$ , u − 1 t = ( u 2 t , … , u Nt ) ′ ${u}_{-1t}={({u}_{2t},\ldots ,{u}_{{Nt}})}^{^{\prime} }$ .
Let u 1 t * = u 1 t − b' u − 1 t ${u}_{1t}^{* }={u}_{1t}-{b\text{'}}{u}_{-1t}$ , then Equation (8) becomes:
y 1 t 0 = α + b ′ y − 1 t + u 1 t * . ${y}_{1t}^{0}=\alpha +b^{\prime} {y}_{-1t}+{u}_{1t}^{* }.$ ()
Under assumption 5, Hsiao et al. (2012) show that α $\alpha $ and b $b$ can be estimated by minimizing
1 T 1 ∑ t = 1 T 1 y 1 t 0 − α − b ′ y − 1 t ′ y 1 t 0 − α − b ′ y − 1 t . $\frac{1}{{T}_{1}}\sum _{t=1}^{{T}_{1}}{\left({y}_{1t}^{0}-\alpha -{b}^{^{\prime} }{y}_{-1t}\right)}^{^{\prime} }\left({y}_{1t}^{0}-\alpha -{b}^{^{\prime} }{y}_{-1t}\right).$ ()
The predicted value of y 1 t 0 ${y}_{1t}^{0}$ can be approximated as follows:
y ˆ 1 t 0 = α + b ′ y − 1 t , ${\hat{y}}_{1t}^{0}=\alpha +b^{\prime} {y}_{-1t},$ ()
where α $\alpha $ is the fixed effect of city 1, b $b$ is the coefficient matrix.
Then, a prediction for the treatment effect at time t ( t ≥ T 1 + 1 $t\ge {T}_{1}+1$ ) can be calculated as:
△ ˆ 1 t = y 1 t 1 − y ˆ 1 t 0 . ${\hat{\bigtriangleup }}_{1t}={y}_{1t}^{1}-{\hat{y}}_{1t}^{0}.$ ()
Since the strategy implementation took effect at time T 1 + 1 ${T}_{1}+1$ , the consistent estimate of the average treatment effect (ATE) over the whole evaluation period can be expressed as
1 ( T − T 1 ) ∑ t = T 1 + 1 T △ ˆ 1 t . $\frac{1}{(T-{T}_{1})}\sum _{t={T}_{1}+1}^{T}{\hat{\bigtriangleup }}_{1t}.$ ()
In selecting a control group, we must choose a subset of untreated cities. Often, the value of N is relatively small compared to T. In a study comparing six methods via Monte Carlo simulations with the data generating process of common factors, Hsiao et al. (2012) propose a two-step approach. This method aims to strike a balance between within-sample fit and postsample prediction error:
  • 1.

    In the panel data set consisting of N cities, the pre-processing period involves selecting m cities from the N − 1 $N-1$ control cities to align with the observed values of y 1 t 0 ${y}_{1t}^{0}$ . The goal is to find the optimal combination of control cities, denoted as M ( m ) * ${M(m)}^{* }$ , where m = 1 , … , N − 1 $m=1,\ldots ,N-1$ , based on the highest R 2 ${R}^{2}$ value. This step ensures that the control group closely matches the characteristic of the treatment group.

  • 2.

    The optimal M ( m ) * ${M(m)}^{* }$ is selected from M ( 1 ) * , M ( 2 ) * , … , M ( N − 1 ) * ${M(1)}^{* },{M(2)}^{* },\ldots ,{M(N-1)}^{* }$ using the Akaike information criterion corrected for small sample sizes (AICC) (Hurvich & Tsai, 1989). The AICC helps in choosing the best model while taking into account the sample size and avoiding overfitting.

3.2 Group settings

3.2.1 Three treatment groups

In this study, we select three urban clusters as the treatment groups. The first treatment group is the Chengdu-Chongqing urban agglomeration, which is located in the upper reaches of the Yangtze River. The second treatment group is the middle reaches of the Yangtze River, as known as the Mid-Yangtze River region. The third treatment group is the Yangtze River Delta, situated in the lower reaches of the Yangtze River. These urban clusters encompass a total of nine provinces and 73 cities, each characterized by distinct geographical locations, resource endowments, stages of industrialization, and levels of urbanization. These clusters are considered the “three poles” of the development pattern of the Yangtze River Economic Belt. The differences observed among these urban clusters provide an opportunity to analyze the effectiveness of the strategy.

Shanghai and Chongqing, as municipalities, exhibit distinct economic behaviors and influential factors compared to other local cities. Therefore, they have been excluded from the sample that was analyzed in this study. Additionally, Xiantao City, Qianjiang City, and Tianmen City have significant amounts of missing data, making them unsuitable for inclusion in the sample. Consequently, this paper utilized a sample size of 68 cities from the three treatment groups to assess the effectiveness of the strategy.

3.2.2 Control group selections

The selection of control group cities in this study is based on two criteria. First, the control group cities should closely resemble the treated cities before the implementation of the strategy. This ensures comparability between the two groups. Second, the control group cities should meet the key assumption of exogeneity, meaning they are unaffected by the implementation of the strategy.

To determine the optimal control group urban areas, this paper follows the two-step approach proposed by Hsiao et al. (2012). Initially, urban areas adjacent to the Yangtze River Economic Belt implementation area are excluded to mitigate spillover effects. Next, urban areas that underwent changes in their statistical caliber during the study period are eliminated to ensure comparability. To enhance the selection efficiency, urban areas with high predictive ability were included in the control group after a preliminary fit. Furthermore, robustness tests are conducted to verify the exogeneity of the selected control group urban areas, thereby strengthening the validity of the analysis.

4 EMPIRICAL RESULTS

In this section, we examine the effectiveness of the strategy in three dimensions: industrial wastewater discharge, economic growth, and industrial structure upgrading and adjustment. First, we report the overall impact of the strategy, and then we delve into individual cities, reporting the effects on the major cities where it has been implemented.

4.1 Treatment effects of the strategy

The results provided in Table 2 show the effectiveness of development in the urban agglomeration areas of the Yangtze River Economic Belt. The implementation of the strategy has led to significant reductions in industrial wastewater discharge, gradual improvement in the structure of the service industry, and no significant impact on service industry output and economic growth.

Table 2. Treatment effects of the strategy.
Urban agglomeration Ecological governance Economic development
Industrial wastewater discharge (log level) Real GDP (log level) Proportion of service industry output Proportion of productive service workers
(million tons) (million yuan) (%) (%)
Overall −0.078* 0.025 1.549 1.532**
Chengdu-Chongqing −0.146** 0.051 1.685* 3.700**
Mid-Yangtze River −0.119** 0.042 1.872 1.056
Yangtze River Delta −0.045 0.012 1.343 1.122
  • Note: ** and * indicate that actual value has been outside the 95% and 90% confidence intervals, respectively.

The treatment effects vary across different urban agglomerations. In the Chengdu-Chongqing city cluster, there was a remarkable decrease of 14.6% in industrial wastewater discharge. This reduction did not hinder regional economic development, as there was no significant change in overall real GDP. The industrial structure and the structure of service industry employees were adjusted, with a significant increase of approximately 1.7% in the proportion of service industry output and 3.7% in the proportion of productive service industry employees. This improvement contributed to enhanced regional productivity.

In the middle reaches of the urban agglomeration, there is a curated reduction of 11.9% in industrial wastewater discharge. However, no significant changes are observed in economic growth, the proportion of output from the tertiary industry, and the proportion of productive service workers in the Yangtze River Delta city cluster.

4.2 Effects on cities

We ranked the cities according to their actual GDP before the implementation of strategic measures and selected a total of 41 cities located within the intervals of 0%–20%, 40%–60%, and 80%–100% to report the empirical analysis results. These selected cities serve as a representative sample for reporting the findings. The geographical distribution of these 41 cities is depicted in Supporting Information S1: Figure A1.

4.2.1 Effects on industrial wastewater discharge

Table 3 shows the estimated results of the policy treatment effects on industrial wastewater discharge in each city.

Table 3. The average treatment effects (ATEs) on cities.
Urban agglomeration City Industrial wastewater discharge Real GDP Proportion of service industry output Proportion of productive service workers
Quantity Growth rate
Chengdu-Chongqing Chengdu −0.153** 0.063 0.016 1.795** 5.886**
Luzhou −0.030 0.084** 0.023 −0.404 −1.017
Neijiang 0.011 0.036 −0.004 5.417** −0.739
Zigong −0.036 −0.009 −0.009 4.660** 0.293
Mianyang −0.120 0.051 0.016 −1.434 2.007**
Suining −0.051 0.008 −0.012 3.853** 1.205**
Guanan −0.691** −0.002 0.003 −1.484* 1.220
Mid-Yangtze River Wuhan 0.005 0.053* 0.020 −0.124 1.842
Xiaogan −0.116** −0.005 0.004 −1.224 −3.189**
Yichang −0.129** 0.009 0.007 2.291** −0.342
Jingmen −0.140** 0.131 0.039 −0.207 0.306
Changsha −0.004 0.078** 0.029 0.495 1.056**
Yueyang −0.177** −0.015 −0.010 12.276** 0.386
Zhuzhou −0.134** 0.018 −0.002 −0.597 3.466**
Yiyang −0.086** 0.039** 0.013 4.785** 0.445
Xiangtan −0.120** 0.028 0.007 4.677** 1.967*
Changde −0.880** 0.051** 0.010 6.251** 1.244
Hengyang −0.414** 0.018 −0.004 3.814** 3.021**
Nanchang −0.096 0.001 0.005 0.656 3.121**
Jiujiang 0.042 0.073 0.016 5.490** −3.873**
Yingtan −0.275** −0.013 −0.016 −5.520* 1.353*
Xinyu 0.042** 0.138** 0.036 −1.444 −0.600
Pingxiang −0.121 −0.022 −0.016 3.104 −2.102**
Yangtze River Delta Hefei −0.046 0.050** 0.012 −0.745 1.027*
Wuhu 0.028 0.065** 0.027 6.535** 1.410
Tongling −0.116** 0.070** 0.026 3.317** 2.136**
Changzhou 0.187** 0.041** 0.015 1.301 2.939**
Nanjing −0.208** 0.050** 0.018 −0.381 −3.064
Wuxi 0.040 −0.007 0.009 −0.007 1.987**
Suzhou −0.002 −0.001 0.013 2.884 1.048*
Yancheng −0.218** 0.019** 0.004 0.469 −0.036
Yangzhou −0.197** 0.016 0.005 1.263 4.442**
Nantong 0.010 0.039** 0.008 5.237** −3.573**
Zhenjiang −0.133** −0.051** −0.024 0.301 2.048**
Hangzhou −0.060 −0.013** −0.006 4.001** 2.930**
Huzhou −0.028** 0.035** 0.016 3.894** 0.595**
Ningbo −0.101** −0.024** −0.006 −1.499 3.659**
Jiaxing 0.009 0.039** 0.023 −1.318 3.414**
Shaoxing 0.037 −0.009 −0.003 0.564 1.822**
Jinhua −0.010 0.007 0.001 0.768 −0.441
Taizhou −0.061** 0.011 0.004 −0.019 0.123
  • Note: This table presents the ATE throughout the entire policy evaluation period. The treatment effect is determined by calculating as the difference between the actual and predicted average values. Specifically, we measure the ATE of the GDP growth rate by comparing the actual growth rate to the forecasted growth rate. ** and * indicate that a city's actual value has been outside the 95% and 90% confidence intervals, respectively, for a minimum of three consecutive years.

The results indicate that a majority of cites, approximately 76%, experienced a decrease in industrial wastewater discharge, indicated by negative ATEs. This suggests that the implementation of the strategy has been effective in reducing industrial wastewater emissions in these cities. On the other hand, around 24% of cities showed positive ATEs, indicating an increase in industrial wastewater emissions. Further analysis reveals that among the cities with negative ATEs, around 35% did not exhibit a statistically significant impact, while the remaining 65% showed a significant reduction in industrial wastewater discharge.

Overall, the results show that 49% of cities demonstrated a significant decrease in industrial wastewater emissions, about 46% showed no significant change, and 5% experienced a significant increase in industrial wastewater emissions.

Notably, the cities that exhibited a substantial decrease in industrial wastewater discharge were primarily located in the middle and upper reaches of the Yangtze River. This can be attributed to the industrial structure of these urban agglomerations, as the middle and upper reaches have a longer history of development are dominated by traditional industries such as iron and steel, automobile manufacturing, and metallurgy. Consequently, the water environment management measures implemented as part of the strategy had a more pronounced impact on the midstream and urban Chengdu-Chongqing urban agglomeration.

At the city level, significant reductions in industrial wastewater emissions were observed in several cities. The majority of these cities experienced an average annual decrease of more than 10% in industrial wastewater emissions compared to the scenario where the strategy was not implemented. On average, the reduction in annual industrial wastewater emissions across all cities was approximately 21.9%.

4.2.2 Effects on economic growth

Table 3 presents the estimated results of the policy treatment effects on the economic growth of each city. The analysis reveals that approximately 71% of cities experienced positive ATEs, while roughly 29% displayed negative ATEs. Further examination of the data show that 35% of cities with positive ATEs exhibited statistically significant impact at the 5% level, and approximately 37% experienced no significant change. On the other hand, 75% of cities with negative ATEs did not show statistical significance at the 5% level.

In summary, about 37% of the cities recorded a notable rise in real GDP, approximately 56% of the cities experienced no significant change, and 7% of the cities, including Zhenjiang, Hangzhou, and Ningbo, witnessed a decline in economic growth. Concerning the growth rate of the economy, cities with a significant positive ATE witnessed an increase between 0.4% and 3.6%.

Based on the results from the previous sections, it can be concluded that the implementation of the strategy has led to a decrease in urban industrial wastewater discharges without significantly impeding the growth of the economy. Moreover, cities experiencing notable economic development have recorded an average increase of approximately 5.7% in real GDP, with an average annual economic growth rate increase of about 1.87%.

4.2.3 Effects on industrial structure adjustment

In terms of the effects on industrial structure adjustment, the implementation of the strategy has had varying impacts on different cities within the three major urban clusters. Table 3 presents the ATEs of the strategy, and the results indicate that approximately 63% of cities exhibit positive effects, while 37% show negative effects. Among the cities with positive effects, about 39% experienced a significant increase in the share of their tertiary industry output, while approximately 61% did not exhibit a statistically significant change. None of the cities with negative effects were significant at the 5% level.

In cities where the industrial structure has been significantly optimized, the implementation of the strategy has led to an average increase of 4.89% in the output share of the tertiary industry.

4.2.4 Effects on service personnel structure

Based on the results presented in Table 3, approximately 73% of the municipalities show a positive ATE, indicating a favorable impact of the strategy on the proportion of employees in productive service industries. On the other hand, around 27% of municipalities exhibit a negative ATE, suggesting a decline in the share of employees in these industries.

Further analysis of the confidence intervals reveals that about 41% of municipalities demonstrate a significant increase in the proportion of employees in productive services, with a statistical significance level of 5%. This indicates that the strategy has effectively contributed to the growth of productive service industries in these municipalities. Additionally, approximately 49% of municipalities show no significant change in the share of employees, while 10% of municipalities, including Xiaogan, Jiujiang, Pingxiang, and Nantong, experience a statistically significant decline.

It is noteworthy that cities in the Yangtze River Delta urban agglomerations, particularly in Zhejiang Province, have a higher proportion of municipalities with significant increases in the share of employees in productive service industries. This could be attributed to the “On Accelerating the Development of Productive Service Industry for Industrial Structure Adjustment and Upgrading” policy implemented by the Zhejiang government in 2015. Except for Jinhua and Taizhou, all other analyzed cities in the region have demonstrated significant optimization in the structure of service industry employees. On average, cities with significant increases experienced a growth of about 2.4% in the share of employees in productive service industries.

Overall, the results indicate that the strategy has been effective in promoting the growth and optimization of the productive service industry in the majority of municipalities, particularly in the Chengdu-Chongqing urban agglomeration and the Yangtze River Delta region.

According to Tables 2 and 3, while some cities in the Yangtze River Delta urban agglomeration achieves notable results in industrial wastewater treatment, economic growth, and industrial structure adjustment, the overall treatment effects for the entire urban agglomeration are not statistically significant.

Based on the data presented in Table 3, cities with significant reductions in industrial wastewater discharge are primarily located in the upper and middle reaches of the Yangtze River. Cities with significant increases in service industry output were mainly found in the middle reaches city agglomeration. The adjustment of the industrial structure in the Yangtze River Delta city agglomeration exhibited diverse patterns.

Overall, the analysis indicates that the protection and governance of the ecology in the Yangtze River region have not hindered regional economic growth but rather facilitated urban development and transition of the industrial structure toward the productive service industry. The majority of cities that experience significant reductions in industrial wastewater emissions also see increases or no significant changes in economic growth rates and service output shares. Additionally, a significant decrease in the share of productive service workers is not observed in the majority of cities. These findings suggest a positive relationship between ecological protection and regional development, supporting the transition towards a more sustainable and service-oriented urban economy.

4.3 Dynamic policy effects

The implementation of policies often entails a lag effect and dynamic consequences. Since the inception of the development strategy for the Yangtze River Economic Belt, the significance of this strategy has progressively gained prominence. Table 4 provides an overview of the dynamic influences exerted by the strategy on regional industrial wastewater discharge, economic growth, service industry output value, and the employment composition within the service industry.

Table 4. Dynamic policy effects.
Year Industrial wastewater discharge Real GDP Proportion of service industry output Proportion of productive service workers
2014 −0.015 0.010 0.212 0.676
2015 −0.021 0.015 0.825 0.874*
2016 −0.069* 0.016 1.712 1.797**
2017 −0.121** 0.037** 1.964** 2.168**
2018 −0.161** 0.048** 3.571** 2.142**
  • Note: ** and * indicate that a city's actual value has been outside the 95% and 90% confidence intervals, respectively.

The findings demonstrate that the role of the strategy gradually becomes prominent during the third year of policy implementation. The data indicates a significant and ongoing expansion of the policy effect at or above a statistical level of 10% after 2015, implying that the development strategy of the Yangtze River Economic Belt consistently promotes regional development. Furthermore, this promoting effect tends to amplify over time.

The occurrence of this phenomenon is reasonable since the effectiveness of the policy hinges on improving and upgrading the relevant supporting infrastructure. Following its official designation as a national strategy in 2014, the Yangtze River Economic Belt has witnessed the consecutive formulation and promulgation of implementation plans by various provinces and cities. However, organizing and implementing government actions, as well as planning and constructing facilities, may require a certain amount of time. Taking wastewater as an example, during the initial phase of policy implementation, local governments must establish prerequisites for reducing wastewater discharge by means of scientific planning, attracting investments, and creating centralized industrial wastewater treatment facilities. In the short term, constructing supporting infrastructure will not significantly impact regional industrial wastewater discharge. However, in the long term, infrastructure improvements will enhance wastewater treatment capacity, leading to a sustained reduction in regional industrial wastewater discharge. Therefore, it is evident that the regional development strategy has not had a significant short-term effect on the region due to the time required for policy implementation, indicating a delay in policy impact.

5 FACTORS ASSOCIATED WITH DISPARITIES IN STRATEGY IMPACTS

The preceding section highlights the effects of the strategy on urban development among different cities. This section aims to further examine the factors that are associated with this diversity of policy effects. Drawing on Ke et al. (2017), we analyze this problem using a linear regression model:
△ ˆ it = α + β X it + μ i + ν t + ε it , ${\hat{\bigtriangleup }}_{{it}}=\alpha +\beta {X}_{{it}}+{\mu }_{i}+{\nu }_{t}+{\varepsilon }_{{it}},$ ()
where i $i$ represents the city and t $t$ represents time. △ ˆ it ${\hat{\bigtriangleup }}_{{it}}$ is the estimated policy effect, as calculated in Table 3 of the previous section, which corresponds to the difference between y i t 1 ${y}_{{\rm{i}}t}^{1}$ and y ˆ i t 0 ${\hat{y}}_{{\rm{i}}t}^{0}$ . The explanatory variable X it ${X}_{{it}}$ captures variables characteristics of the city.

These characteristics include: (1) Foreign investment, depicted by the proportion of actual foreign capital utilization to regional GDP; (2) Trade size, gauged by the total value of imports and exports divided by regional GDP; (3) Human capital, measured by the ratio of students in regular colleges and universities to the permanent population of the region; (4) Population density; (5) Infrastructure, represented by per capita road area; (6) Informatization level, evaluated by the proportion of Internet users; (7) Local government fiscal expenditures; (8) The level of local financial development, quantified by the ratio of deposit and loan balances to regional GDP; (9) R&D expenditures; (10) Innovation capability, assessed by the number of invention patent applications; (11) Average wage. Supporting Information S2: Table A2 provides detailed definitions and descriptive statistics of these explanatory variables.

Table 5 exhibits the regression outcomes from the fixed-effect model used in this analysis. In examining the factors linked to policy effects regarding industrial wastewater disposal, insights are presented in column 1. The results indicate a noteworthy negative correlation between human capital stock and policy effects, signifying that cities endowed with abundant human capital witness a more substantial decrease in industrial wastewater discharge. Moreover, the coefficient of trade size demonstrates a positive significance, implying that increased trade openness is linked to a deteriorating condition in industrial wastewater treatment. This suggests that cities with higher levels of international trade may encounter added difficulties in effectively managing industrial wastewater and preserving environmental sustainability.

Table 5. Heterogeneity analysis of treatment effects.
Industrial wastewater Real GDP Proportion of service industry output Proportion of productive service workers
Foreign investment ratio 0.028 0.028 −2.487 1.217
(0.064) (0.064) (1.418) (1.351)
Trade size 0.025 −0.005 −0.412 0.143
(0.013) (0.002) (0.145) (0.147)
Human capital −0.280 −0.080 6.167 1.538
(0.131) (0.029) (1.756) (2.602)
Population density (log level) −1.907 1.009 26.225 −3.920
(2.108) (0.568) (43.652) (43.915)
Infrastructure level 0.185 0.162 1.094 9.545
(0.388) (0.047) (7.326) (5.891)
Informatization level 0.193 0.001 5.920 1.096
(0.189) (0.039) (2.595) (3.072)
Fiscal expenditure ratio −1.756 −0.283 37.388 56.664
(2.244) (0.304) (22.263) (29.583)
Financial development level 0.114 −0.035 1.254 −4.424
(0.212) (0.034) (2.125) (3.230)
R&D expenditure ratio −0.078 −0.016 −1.019 −1.775
(0.084) (0.017) (1.087) (1.081)
Innovation capability −0.117 0.033 3.029 −0.582
(0.130) (0.033) (2.077) (2.765)
Average wage (log level) −0.957 −0.403 −18.511 −0.195
(1.123) (0.235) (16.452) (19.889)
Observations 123 123 123 123
R2 0.281 0.385 0.348 0.262
Fixed effects City and year City and year City and year City and year
  • Note: Standard errors are reported in parentheses, clustered at city level.
  • * p < 0.10
  • ** p < 0.05
  • *** p < 0.01.

The insights revealed in column 2 of the table indicate a positive association between factors like higher population density and improved infrastructure with a city's economic growth. This underscores the significance of urban development strategies considering population dynamics and substantial investments in infrastructure. Furthermore, the proportion of imports and exports demonstrates a negative correlation with the policy's treatment effect on economic growth. This could be attributed to the bilateral opening-up effect of the Yangtze River Economic Belt, which stimulated trade development in cities with initially lower openness. As a result, cities with higher trade openness may exhibit relatively lower treatment effects on economic growth compared to those with lower trade openness. This finding emphasizes the importance of managing trade dynamics thoughtfully and balancing the benefits of openness with the aims of sustainable economic growth.

Columns 3 and 4 present the outcomes of the analysis exploring the relationship between a city's attributes and the treatment effect of service industry development.

In column 3, a significant and positive correlation emerges between higher levels of human capital and the augmentation of service industry output. Conversely, the coefficients of foreign investment and trade openness exhibit significant negative associations. Additionally, cities with heightened levels of informatization and government fiscal expenditures tend to have a positive association with the augmentation of service industry output.

Column 4 reveals that foreign investment and trade size are linked to the enhancement of the service industry personnel structure. Furthermore, the study suggests that cities equipped with robust infrastructure or convenient transportation tend to experience a more rapid upgrading in the service industry personnel structure.

6 ROBUSTNESS CHECK

To ensure the reliability of the estimated treatment effects, this study conducted robustness tests examining three key aspects: control group assessment, alterations in control group selection criteria, and placebo test.

6.1 Control group test

The Hsiao et al. (2012) method assumes that the control group is exogenous. To ensure the exogeneity assumption and prevent spillover effects from influencing the research, control group cities are selected specifically from areas without shared borders with the Yangtze River Economic Belt. This section conducts a further test of the control group to address the possibility of any potential impacts of the strategy. Each control group city is sequentially selected as a treatment group city, with the remaining cities serving as the control group. The estimating strategy proposed by Hsiao et al. (2012) in Section 3 of this paper is also used to estimate the impact of the strategy on the selected city. If the estimated counterfactual and actual values show no significant difference, then the exogeneity assumption is satisfied.

Supporting Information S2: Tables A3–A6 present the control group cities used for the analysis of industrial wastewater discharge, actual GDP, the share of tertiary industry output, and the share of productive service personnel discussed in the preceding section. These tables present the R 2 ${R}^{2}$ value of the sample fit, the ATE, and the level of significance set at 5%. The results indicate that the selected control group cities do not appear to be affected by the strategy, based on the statistical level of 5%.

6.2 Alternative control-group selection criteria

The AICC, utilized as the model selection criterion in the baseline analysis, aids in selecting the control group to establish counterfactual values. To evaluate the reliability of policy effects assessed through alternative selection methods, the BIC information criterion is applied in this section to identify the most suitable control group.

Supporting Information S2: Table A7 presents the outcomes, showing a strong correspondence between the effect identified by the BIC criterion and the estimated treatment effect derived from the AICC criterion. This strengthens the evidence supporting the significant impact of the Yangtze River Economic Belt policy.

6.3 Placebo test

A placebo test is conducted to examine whether factors other than the implementation of the strategy could explain the observed changes in the region. This test involves shifting the implementation time to 2010, creating a counterfactual analysis. If significant treatment effects are still observed in cities during this period, it may cast doubt on the established causal relationship between the strategy implementation and the observed development.

Supporting Information S2: Table A8 presents the R 2 ${R}^{2}$ value of the sample fit and provides an analysis of the significance of policy effects during the pseudo-treatment period. The findings indicate that the introduction of a pseudo-policy implementation time does not exert a significant impact on the measurement of policy effects. No evidence suggests a causal relationship between other policies and the observed development. This finding provides robust support for the conclusions of the earlier research.

7 CONCLUSION

The study utilizes data from the China Urban Statistical Yearbook (1995–2019) to quantify the impact of the Yangtze River Economic Belt development strategy on urban green development. Three key areas are analyzed: industrial wastewater discharge, economic growth, and the adjustment of industrial structure in relation to ecological governance and economic development. The findings of the study are as follows:

First, when considering the overall urban agglomeration, there are variations in the treatment effects among different regions. The implemented strategy leads to a significant reduction in industrial wastewater discharge of approximately 14.6% in Chengdu-Chongqing urban agglomeration and 11.9% in the middle reaches of Yangtze River urban agglomeration. However, there is no significant change observed in the treatment effect in the Yangtze River Delta urban agglomeration. In terms of real GDP, there are no statistically significant differences in the overall ATEs among the three major urban agglomerations. The only notable optimization is observed in the industrial structure and service structure of Chengdu-Chongqing urban agglomeration, with the proportion of service output and employees in productive services increasing by about 1.7% and 3.7%, respectively.

Second, the study finds that the implementation of the strategy has supported green development in specific cities. Over the evaluation period, there is an average reduction of approximately 21.9% in annual industrial wastewater emissions. Additionally, cities that experienced significant changes as a result of the strategy observed an average increase of approximately 5.7% in annual real GDP, 1.9% in the economic growth rate, 4.9% in the share of tertiary industry output, and 2.4% in the proportion of employees in productive services.

Third, the study reveals that ecological treatment does not have a detrimental effect on the regional economy. In particular, the industrial wastewater treatment effect in Chengdu-Chongqing urban agglomeration is effective, leading to optimization in the regional industrial structure and service industry personnel structure. Furthermore, economic growth and industrial structure optimization are not significantly negatively impacted in over 90% of cities that achieved significant reductions in industrial wastewater discharge.

Fourth, the study finds that in terms of optimizing economic structure, there is a significant increase in the proportion of output and employees in the service industry in Chengdu-Chongqing urban agglomeration. This suggests a trend towards the dominance of the industrial structure in the Yangtze River Midstream urban agglomeration and Yangtze River Delta urban agglomeration is not significant on overall level. However, certain cities within these agglomerations experience industrial structure optimization. Approximately 44% and 28% of cities in the Yangtze River Midstream urban agglomeration and Yangtze River Delta urban agglomeration, respectively, exhibits a significantly higher proportion of output in the service industry. Additionally, 37% and 66% of cities in these respective areas exhibits a significant increase in the proportion of employees in productive services.

Fifth, the analysis of factors influencing the disparity of policy effects reveals several key findings. Cities with dense populations tend to experience a greater treatment effect in terms of reducing industrial wastewater emissions. On the other hand, trade openness has a negative effect on the reduction of industrial wastewater emissions. In terms of the economy, cities with high population densities and complete infrastructure tend to experience greater economic benefits. Moreover, cities that possess abundant human capital, high levels of informatization, and strong government intervention in the economy are able to make a faster transition towards a dominant service industry structure. Last, cities with higher degrees of openness and complete infrastructure experience faster structural upgrades in their service industry workforce.

The findings of this study have important implications for future policy adjustments. First, with regard to the agglomerations, it is evident that the Yangtze River Delta urban agglomeration has not experienced significant treatment effects across all three dimensions of assessment. Therefore, there is a need to implement enhanced promotional policies to effectively reduce industrial wastewater discharge and optimize the industrial structure in the Yangtze River Delta region. In the midstream region, it is crucial to focus on promoting economic quality in addition to achieving ecological treatment outcomes. Efforts should be directed towards accelerating the transformation of the industrial structure towards a service-dominated framework, along with increasing the proportion of productive service workers in the region.

Second, there are cities along the Yangtze River that have not yet achieved notable treatment effects in terms of industrial wastewater discharge. This highlights the need for targeted interventions and improvements in these cities to enhance their performance in terms of green development.

Last, at the city level, it is important to prioritize enhancing supporting infrastructure, increasing levels of urbanization and informatization, and fostering the accumulation of human capital. These factors can significantly contribute to the realization of the objectives of the Yangtze River Economic Belt construction while promoting eco-friendly development.

It is worth noting that the research period of this study corresponds to the early stage following the implementation of the strategy. The treatment effects observed during this period can be influenced by various factors, including the implementation capabilities of city governments and the challenges faced by different types of industrial enterprises in their transformation. Therefore, it is essential to conduct further research to examine the long-term effects of the strategy in urban areas and to evaluate its sustainability and effectiveness over time.

AUTHOR CONTRIBUTIONS

Guan Gong: Funding acquisition; resources; writing—review and editing. Yu Zhao: Data curation; software; writing—original draft.

ACKNOWLEDGMENTS

This work is supported by the National Natural Science Foundation of China (grants no. 71873087 and 72373088) and the Fundamental Research Funds for the Central Universities (grant no. CXJJ-2022-326).

    CONFLICT OF INTEREST STATEMENT

    The authors declare no conflict of interest.

    ETHICS STATEMENT

    Not applicable.

    • 1 Studies, as seen in Sun et al. (2018), have explored green development, focusing primarily on three core elements: reducing pollutant emissions, improving resource efficiency, and promoting economic development. Moreover, in studying the relationship between economic growth and environmental protection, scholars often evaluate environmental policies by measuring factors such as industrial pollutant emission intensity and the rate of industrial pollutant removal (Wang & Shen, 2016).
    • 2 We also performed the regression specification error test (RESET test) to analyze whether the model's linear assumptions are reliable. Please check on Supporting Information S2: Table A1 for further details

    The full text of this article hosted at iucr.org is unavailable due to technical difficulties.