*2.2. HQD Index System*

#### *2.2. HQD Index System*  2.2.1. Meaning of HQD

2.2.1. Meaning of HQD The current research has not yet uniformly defined the connotation of HQD. Starting from the goal of HQD, the connotation of HQD is efficient, fair, green, and sustainable development aimed at meeting people's growing needs for a better life [10]. HQD is the economic development mode, structure, and dynamic state that meet the real needs of people's growth [7]. From the perspective of the "five development concepts" and the main social contradictions, HQD is defined by identifying imbalances and inadequacies in economic and social development [45]. As a typical river basin flowing through China's nine major provinces and regions, the YRB requires HQD based on the full consideration of various factors, such as the natural ecological environment and economic structure characteristics of the basin, guided by systematization, integrity, and relevance, as well as the benign interaction and coordinated development of economy, society, and the ecology The current research has not yet uniformly defined the connotation of HQD. Starting from the goal of HQD, the connotation of HQD is efficient, fair, green, and sustainable development aimed at meeting people's growing needs for a better life [10]. HQD is the economic development mode, structure, and dynamic state that meet the real needs of people's growth [7]. From the perspective of the "five development concepts" and the main social contradictions, HQD is defined by identifying imbalances and inadequacies in economic and social development [45]. As a typical river basin flowing through China's nine major provinces and regions, the YRB requires HQD based on the full consideration of various factors, such as the natural ecological environment and economic structure characteristics of the basin, guided by systematization, integrity, and relevance, as well as the benign interaction and coordinated development of economy, society, and the ecology in the whole basin [2].

#### in the whole basin [2]. 2.2.2. Calculation of HQD

The entropy method is an objective weighting method, and it analyzes the role of the comprehensive evaluation by comparing the information entropy of the indicators [46]. Chen et al. [47] used the entropy weight method to calculate the weight of each index and to evaluate the urban ecological level on the basis of analyzing the characteristics of the entropy weight method in different stages in detail. Thus, this paper also uses the entropy

weight method [46–48] to measure the HQD level of the 78 prefecture-level cities in the YRB from 2004 to 2019. The specific steps are as follows:

First, this paper performs extreme value standardization on the original dataset. The positive index is *X* 0 *ij* = (*Xij* − min(*Xij*))/(max(*Xij*) − min(*Xij*)), and the negative index is *X* 0 *ij* = (max(*Xij*) − *Xij*)/(max(*Xij*) − min(*Xij*)), where *Xij* is the index value of the original data, and *X* 0 *ij* is the standardized index value. Then, it calculates the contribution of the *i* evaluation object under the *j* index with the formula *Pij* = *Xij*/ *n* ∑ *i*=1 *Xij*. Next, it calculates the entropy value with the formula *E<sup>j</sup>* = −*k n* ∑ *i*=1 [*Pij* × ln(*Pij*)], where *k* = 1/ ln(*n*). Later, it calculates the weight of the *j* indicator with the formula *W<sup>j</sup>* = (1− *Ej*)/( *n* ∑ *i*=1 (1 − *Ej*)). Lastly, it calculates the HQD index of the *i* evaluation object with the formula *Y<sup>i</sup>* = *n* ∑ *i*=1 *<sup>W</sup><sup>j</sup>* <sup>×</sup> *<sup>P</sup>ij* .

#### 2.2.3. Index System Construction

The selection of indicators in this paper is based on the principles of comprehensiveness, systematicness, objectivity, and data availability. Drawing on the research ideas of Liu et al. [24] and Lin et al. [49], this paper constructs 25 evaluation indicators from the four dimensions of HQD, including the driving force, structure, method, and achievement, and establishes a scientific, fair, objective, and practical indicator system for HQD in the YRB, as shown in Table 1.



In terms of the driving force of HQD, it is divided into two element layers: technological progress and human capital. It mainly reflects the transformation of economic development from factor-driven to innovation-driven relying on human capital, which is an important symbol of HQD and the cornerstone of ensuring green, fair, and sustainable development. Therefore, the level of technological progress and the level of human capital are measured here. The level of technological progress is measured by the intensity of R&D expenditures and the proportion of scientific and technological expenditures in fiscal expenditures to the total population of the region.

In terms of the structure of HQD, the proportion of the added value of the tertiary industry in the regional GDP is used to reflect the changes in the industrial structure. The proportion of deposits and loans of financial institutions in the GDP is used to measure the changes in the financial structure. The ratio of the per capita income of urban and rural residents and the urbanization rate reflects the urban and rural structure. The proportion of foreign investment in the regional GDP reflects the level of economic opening to the outside world.

In terms of the method of HQD, it is divided into two element layers: resource conservation and environmental protection. In terms of resource conservation, the energy consumption per unit of GDP and electricity consumption per unit of GDP are selected to represent the main indicators of resource conservation by economic activities. The per capita area of park green space represents the main indicator of economic activities for environmental protection.

In terms of the results of HQD, the per capita GDP is used to measure the level of economic development, the proportion of fiscal revenue to GDP is used to measure the quality of economic operation, and the urban registered unemployment rate is used to measure the impact of economic fluctuations on people's living and welfare. Indicators, such as the number of public libraries per 10,000 people and the number of public transport vehicles per 10,000 people, are used to measure multi-dimensional social life. Indicators such as the number of beds in medical institutions per 1000 people and the number of people insured by endowment insurance are used to measure social security. In terms of environmental cost, the amount of wastewater discharged per unit of output, the amount of sulfur dioxide discharged per unit of output, and the amount of smoke emissions (dust) are used to measure the damage to the environment caused by economic activity.

#### *2.3. Empirical Strategy*

## 2.3.1. Benchmark Regression Model

Based on the above theoretical analysis, to empirically explore the impact of environmental regulation and local government competition on HQD, this paper uses the data of prefecture-level cities in the YRB from 2004 to 2019 to construct the following measurement model:

$$HDQ\_{\rm il} = \mu\_0 + \beta\_1 ER\_{\rm il} + \beta\_2 ER\_{\rm il} \times \ln{GOV\_{\rm il}} + \beta\_3 \ln{GOV\_{\rm il}} + \sum \delta \ln{X\_{\rm il}} + \mu\_{\rm il} \tag{1}$$

where *i* represents the prefecture-level city, and *t* represents the time. *HDQ* is the level of high-quality development; *ER* is the environmental regulation; *GOV* is the local government competition; and *ER* × *GOV* is the interaction term between environmental regulation and local government competition. *Xit* is the control variable that affects the level of HQD; and *µ* is a random disturbance term. ln *GOV* is in logs.

## 2.3.2. Threshold Regression Model

The relationship between environmental regulation and HQD is also different depending on the intensity of the competition between local governments. Existing studies mostly draw linear conclusions [22,24]. According to the above theoretical analysis, environmental regulation, local government competition, and HQD have interactive effects. Therefore, it is not accurate to test the effect between them with a simple linear relationship. In order to verify the nonlinear relationship between environmental regulation, local government competition, and the HQD of the YRB, this paper uses a nonlinear threshold panel model for this research. The threshold regression model was developed by Tong in 1978 and further improved by Hansen in 2000 [50,51]. This paper further uses local government competition as the threshold variable and adopts the method of Hansen [51] and Ding et al. [52] to test the threshold effect. When the model only has a single threshold,

$$HQD\_{\rm il} = \mathfrak{a}\_0 + \sum \delta \ln X\_{\rm il} + \mathfrak{f}\_1 ER\_{\rm il} \times I(\ln GOV\_{\rm il} \le r\_1) + \mathfrak{f}\_2 ER\_{\rm il} \times I(\ln GOV\_{\rm il} > r\_1) + \varepsilon\_{\rm il} \tag{2}$$

In many cases, there are multiple thresholds, so the extended multi-threshold model is constructed as follows:

$$\begin{cases} HQD\_{\text{il}} = \mathfrak{a}\_0 + \sum \delta \ln X\_{\text{il}} + \mathfrak{f}\_1 \mathbb{E}R\_{\text{il}} \times I(\ln GOV\_{\text{il}} \le r\_1) + \mathfrak{f}\_2 \mathbb{E}R\_{\text{il}} \times I(r\_1 < \ln GOV\_{\text{il}} \le r\_2) + \\ \mathfrak{f}\_3 \mathbb{E}R\_{\text{il}} \times I(\ln GOV\_{\text{il}} > r\_2) + \varepsilon\_{\text{il}} \end{cases} \tag{3}$$

where ln *GOV* is the threshold variable, *r* is the threshold value, and *ε* is the residual item.

#### *2.4. Variable Selection and Descriptive Statistics*

#### 2.4.1. Explained Variable

The explanatory variable in this paper is the high-quality development level (HQD). This paper uses the entropy method to construct an indicator system for HQD in the YRB from four dimensions, namely, the driving force, structure, method, and achievement of HQD, as shown in Table 1.

#### 2.4.2. Core Explanatory Variables

Environmental regulation (ER): This is a general term for the "policies, regulations, measures, and means" promulgated and implemented by the government or related organizations. Currently, the measurement of environmental regulation is mainly divided into two categories: the single index method and the comprehensive index method. Single indicators mainly include pollution fee collection [53], single pollutant discharge or treatment efficiency [54,55], environmental treatment costs [56,57], and environmental protection regulations and standards [58,59]. The comprehensive index method selects indicators from different angles. It constructs comprehensive indicators, such as various pollutant removal rates [60,61], environmental taxes and fees [62], and environmental input [63], by weighting using the entropy weight method and factor analysis method. This paper combines the availability and accuracy of data and refers to the construction methods of relevant empirical research [57,64]. It calculates the discharge of industrial wastewater, industrial waste gas, and industrial solid waste using the entropy weight method to obtain a comprehensive environmental regulation index.

Local government competition (GOV): Most of the literature uses the ratio of productive expenditure to total regional budget expenditure [65], FDI per capita, FDI per unit of GDP, and the share of FDI in national FDI [66] as proxy variables. However, this paper suggests motives for chasing and surpassing neighboring prefecture-level cities in the whole region. Therefore, referring to the research method of Miu et al. [67], this paper adopts the level of economic catching up as a proxy variable of local government competition.

First, this paper calculates the highest per capita GDP of neighboring cities divided by the highest per capita GDP of decision-making units. Next, it calculates the highest per capita GDP of all the regions and cities divided by the highest per capita GDP of decision-making units. Finally, it multiplies the two to obtain the economic catch-up level.

#### 2.4.3. Control Variables

This paper refers to the existing literature research [20,24] and selects the following control variables: (1) urban population density (DEN), measured by the proportion of the urban population in the area of administrative divisions; (2) the level of informatization (INO), measured by the proportion of regional post and telecommunications business revenue to GDP; (3) infrastructure (INF), measured by the per capita urban road area; (4) industrialization level (IND), measured by the proportion of secondary industry output value in total production; (5) human capital (HU), measured by the number of college students per 10,000 people; and (6) industrial structure (IS), measured by the proportion of the output value of the tertiary industry to the output value of the secondary industry. The meaning of the variables and a descriptive statistical analysis are shown in Table 2.

**Table 2.** Descriptive statistics.


## **3. Results**

#### *3.1. Benchmark Regression*

The estimated results of the benchmark regression are shown in Table 3. Column (1) is the OLS estimation result. The estimated coefficient of environmental regulation is 0.459, which is significant at the 1% statistical level. Columns (2)–(5) control the fixed effects of city and year, and they introduce the control variables one by one. The estimated coefficients of environmental regulation are still significantly positive, and they are significant at the 1% statistical level, indicating that environmental regulation can significantly improve the HQD level of the YRB. The research hypothesis H1 is validated. As shown in column (5) of Table 3, local government competition has a significantly negative impact on HQD, indicating that, in order to catch up with the economic level of the surrounding cities in the region, the policies implemented by the local government will reduce the HQD level of the local city. On the one hand, the "promotion championship" hypothesis holds that local officials tend to focus on the economy and that they neglect the environment for their political performance and promotion opportunities, resulting in the lack of effective protection of local environmental quality [38]. On the other hand, under the development goal of "only GDP", local governments will relax environmental regulations. The region will absorb highpolluting, high-energy-consuming industries in developed regions. Intensified competition benefits GDP growth, but environmental pollution intensifies, and the negative externality of environmental pollution is significant [68]. This competition will also cause ecological damage and reduce the level of HQD. As shown in column (5) of Table 3, the coefficient of the interaction term between environmental regulation and local government competition is −0.020, which is significant at the 1% statistical level. This regression result shows that, with the improvement of local government competition, the role of environmental regulation in promoting HQD is weakened. The research hypothesis H2 is validated.

In addition, regarding the control variables, the level of informatization, the improvement of human capital, and the optimization of the industrial structure have a significantly positive impact on HQD. The increase in urban population density and the proportion of secondary industries have a significant negative impact on the level of HQD. This

shows that the HQD of the YRB will be affected not only by the environmental and local government competition but also by other factors.


**Table 3.** Benchmark regression results.

Notes: \* *p* < 0.10, \*\* *p* < 0.05, \*\*\* *p* < 0.01; numbers in parenthesis are robust standard error.

#### *3.2. Threshold Effects Regression*

This paper uses the threshold effect bootstrapping method (bootstrap) to test whether there is a threshold value and the number of thresholds in the model (2). The results are shown in Table 4. When the threshold variable is local government competition, the F statistic is 58.07 in the single-threshold effect estimate, which is significant at the 1% level and rejects the assumption of a linear relationship; in the double-threshold effect estimate, the F statistic is 19.95, which is not significant. The result of the significance test shows that there is no double threshold. Therefore, a single threshold is more appropriate.

**Table 4.** Results of threshold conditions test and double threshold estimated value.


Table 4 shows a threshold value of 3.037 with local government competition as the threshold effect. The regression results in Table 5 show that, when the local government competition level LnGOV ≤ 3.037, the relationship between environmental regulation and HQD is significantly positively correlated at the 1% level, with a coefficient of 0.260. When the local government competition level is greater than 3.037 (LnGOV > 3.037), the impact of environmental regulation on HQD is significantly positive. This result still passes the significance test at the 1% level, but the coefficient is reduced to 0.239. The above analysis shows that, with the improvement of the competition level of local governments, the positive impact of environmental regulation on the HQD of the YRB is weakened, which is consistent with the above research conclusions and verifies H3.


**Table 5.** Estimation results and tests of threshold regression mode.

#### *3.3. Heterogeneity*

According to the above test of the threshold effect, it is confirmed that environmental regulation has a nonlinear relationship with HQD. However, differences in resource endowment, ecological environment, and economic development among different regions of the YRB lead to heterogeneity [69]. This paper further investigates the heterogeneous impact of environmental regulation on the HQD level in the different regions of the YRB. This paper divides the sample into three subsamples: "upstream", "midstream", and "downstream". The results are shown in Table 6.

**Table 6.** Heterogeneity analysis based on different regions.


Notes: \* *p* < 0.10, \*\* *p* < 0.05, \*\*\* *p* < 0.01; numbers in parenthesis are robust standard error.

As shown in columns (1)–(3) of Table 6, environmental regulation has a significant promoting effect on the HQD level of the upper, middle, and lower regions, respectively, and all of them are significant at the 1% statistical level. By comparing coefficients, environmental regulation has a more significant positive impact on HQD for the lower reaches of the YRB. The lower reaches of the Yellow River are rich in various resources and have a good foundation for development [6]. However, environmental pollution is severe due to the over-exploitation of energy and mineral resources and the development of heavy

chemical industries in the middle and upper reaches [70]. Environmental regulation has led to an increase in the cost of pollution reduction for enterprises and a lack of innovation motivation. As a result, it has a weaker impact on HQD. The interaction term of environmental regulation and local government competition in the upper reaches of the YRB is significantly negative at the 5% level, the lower reaches are significantly negative at the 1% level, and the middle reaches are insignificant. The stronger the environmental regulation is, the lower the pollution emissions and the higher the level of HQD. However, as the level of competition between local governments intensifies, the role of environmental regulation in promoting the level of HQD is weaker.

#### *3.4. Robustness Test*

#### 3.4.1. Endogenous Processing

To alleviate the possible endogeneity problem in the benchmark model, according to Arellano and Bover [71], this paper uses the lag one period of the HQD index as an instrumental variable to perform a systematic generalized method of moments (GMM). The results are shown in Table 7. The AR(2) value is greater than 0.1, and the value of Hansen's test is greater than 0.1, indicating that the instrumental variable selected by the model is reasonable. After removing the endogeneity, the lag term coefficient of the HQD index is significant at the statistical level of 1%. The HQD level has a strong trend, and the HQD level of the previous period affects the current period; the environment regulation still has a significant positive impact on the HQD level of the YRB at the statistical level of 1%.

**Table 7.** Robustness test results: endogenous processing.


Notes: <sup>1</sup> The lag period of HQD. \* *p* < 0.10, \*\*\* *p* < 0.01; numbers in parenthesis are robust standard error.
