Impact of the Coupling Coordination Degree of Human Capital and Infrastructure on High-Quality Economic Development: Empirical Evidence from Chinese Cities

Abstract

China’s economy is transitioning from high-speed to high-quality development, making coordination between economic factors increasingly important. Human capital (HC) and infrastructure (INFRA) are important elements of economic development, and the coupling coordination of the two can have a non-negligible impact on regional economies’ high-quality development. Therefore, this study measures the coupling coordination degree (CCD) of HC and INFRA based on data of 184 prefecture-level cities in China from 2011 to 2019, and comprehensively examines the impact effect and mechanism of the CCD on high-quality economic development (HQED). The relevant results are threefold. (1) Compared with single factors, the degree of HC-INFRA CCD can promote HQED more efficiently. (2) A significant single-threshold effect is evident in the promotion of HQED by the degree of HC-INFRA CCD, and a smaller city size and more extensive higher education institutions promote a stronger threshold effect. (3) Factor allocation and factor efficiency are significant transmission mechanisms of the HC-INFRA CCD affecting HQED, and the mediating effect of the capital-labor force allocation path accounts for 66.28% of the total indirect effect. The results provide empirical evidence verifying that HC-INFRA CCD promotes HQED, along with insightful reference suggestions for formulating relevant macro policies and promoting high-quality urban economic development.

1. Introduction

Following the reform and opening up, China relied on large-scale factor inputs to achieve rapid economic growth [1]. While social productivity increased dramatically, China’s economic efficiency remained low, which led to the waste of various resource elements, and made it difficult to achieve sustainable economic growth [2,3]. Development economics contends that the social division of labor to advance economic development cannot be separated from the scale of factor inputs, but the factor allocation (FA) coordination is more valued to achieve a sustained model of economic growth [4,5]. The 19th Communist Party of China (CPC) Congress Report asserted that China’s economy must be transformed from high-speed growth to high-quality development, and that the crude economic growth model was unsustainable. The 20th CPC Congress Report also proposed the promotion of high-quality economic development (HQED) as China’s primary goal, in order to reduce resource waste and thus realize sustainable economic growth. In the process of transitioning the nation’s economic development from quantitative increase to qualitative change, it is essential to focus on the interaction of internal factors within the system and emphasize the coordinated development of factors [6,7]. Human capital (HC) and infrastructure (INFRA) are the key factors for advancing HQED [8,9]. The state of HC-INFRA coordinated development and its impact on HQED is worthy of further in-depth research.

As the primary contributors to economic production, laborers are the most active factor in economic activities [10]. However, according to the National Bureau of Statistics of China, the number and proportion of China’s working-age population has experienced a double-decline for 10 consecutive years since 2012. With a dwindling population, it is no longer possible to rely on large-scale labor factors to drive economic development. However, HC can compensate for the negative impact of the declining workforce and aging population [11]. While establishing a new development pattern and promoting HQED, it is crucial to prioritize raising the nation’s HC level. The Outline of the 14th Five-Year Plan and 2035 Vision for National Economic and Social Development of the People’s Republic of China highlighted the necessity of expanding the demographic quality dividend to raise the nation’s HC level. Meanwhile, in the new technological revolution, HC is gradually replacing physical capital as the most central force driving economic development [12]. These circumstances require policymakers and scholars to strategically investigate the coupling coordination between HC and other factors and advance the efficient promotion of HC for economic development.

INFRA is an indispensable material guarantee for promoting China’s economic development, and supporting cities’ competitiveness, residents’ income, industrial upgrading, and sustainable development [13,14,15]. INFRA investment and construction have continued to advance in the process of China’s rapid urbanization; however, due to economic disparities, geographic factors, and population mobility, some cities struggle with a mismatch between INFRA construction and economic development needs [16]. INFRA construction in some cities lags behind economic demand, which limits urban economic development. For example, Taian is located in the central region of Shandong Province, with the provincial capital Jinan to the north, but its INFRA differs from the overall level of the province. Nearly 80% of the land was unused in 2020, and the per capita road area is half that of the province, making it difficult to advance the development of urban industries and retain HC, which restricts economic development. This city’s gross domestic product (GDP) per capita in 2020 was $7318, representing only 70% of the province’s. INFRA construction in some cities is detached from actual demand, consuming large amounts of economic resources for excessive construction and crowding out other production investments. For example, the GDP growth rate of Neijiang and Lanzhou in 2015 did not exceed 5%, but fixed asset investments were up to 30% or more. The cities’ excessive investment in fixed assets did not effectively promote economic development, and the return on investment was much lower than the average level. At the same time, large-scale construction, maintenance, and operation of INFRA requires considerable financial support. Local governments have raised taxes and increased government debt to balance fiscal revenue and expenditure, which inhibits economic growth, increases economic volatility, and does not promote HQED [17].

As noted above, HC and INFRA are both closely related to HQED; however, HC and INFRA both interact with and constrain one another. HC inputs promote the continuous improvement of INFRA, and high-quality HC development can help local governments rationally plan the overall layout of urban INFRA, while reducing construction and maintenance costs through technological innovation. As the funds allocated for INFRA ultimately come from labor force tax revenue, HC can enhance local governments’ revenue and alleviate INFRA construction investment and financing constraints [18]. Furthermore, stable INFRA operations provide residents with social public services, establishing an important guarantee for HC optimization [19]. For example, information communication INFRA can enhance information flow and reduce indexing and sharing costs [20]. Transportation INFRA improves HC by increasing population movement efficiency and attracting population agglomeration [21,22]. According to the coupling coordination theory, systems interact and co-evolve to enhance the external effects on HQED [23]. The relationship between HC and INFRA aligns with these definitions, and this study contends that coupling coordination is inherent between the two in the promotion of cities’ economic development. When coupling coordination between the two occurs, a city can realize INFRA reinvestment and improve INFRA construction in addition to promoting steady HC development toward scale and quality, releasing the full potential of HC productivity and further promoting HQED.

To investigate the impact of the coupling coordination of HC and INFRA on HQED, this study quantifies HC-INFRA coupling coordination degree (CCD) and HQED indicators using panel data from 184 prefectural-level cities in China covering the period from 2011 to 2019, constructs and empirically examines the mechanism pathway of HC-INFRA CCD for urban HQED, and further explores the threshold effect of city scale (CS) and higher education institutions (HEI) on HC-INFRA CCD for advancing urban HQED. This study makes three main contributions. First, we construct a comprehensive evaluation system to quantify urban HQED, INFRA, and HC. Combining China’s new development philosophy and the concept of HQED, we construct an indicator system with 6 indicators to measure and examine urban HQED. From the perspective of urban INFRA functionality, we constructed 9 dimensions and 18 indicators to measure the level of INFRA. Unlike the majority of previous studies that have used education level to measure HC, this study takes a multidimensional approach by including age, employment, space, education, and security. Second, we present a new perspective for examining the drivers of HQED. In this paper, we adopt HC-INFRA CCD to examine the interaction relationship between HC and INFRA, and to theoretically analyze and empirically test the role mechanism of the CCD of the two in influencing HQED. Third, this study constructs a detailed path model (PM). At the theoretical level, this study examines FA and factor efficiency (FE) as significant paths through which HC-INFRA CCD promotes HQED. At the empirical level, this study refines the mediating variables specific to FA and FE and constructs a PM that includes six mediating variables of capital-economic allocation (CEA), labor-economic allocation (LEA), capital-labor allocation (CLA), labor efficiency (LRE), capital efficiency (CE), and land efficiency (LE). The empirical results contribute to a deeper understanding of the pathways of HC-INFRA CCD on HQED.

The remainder of this paper is organized as follows. Section 2 introduces a review of the relevant literature. Section 3 presents our theoretical analysis and research hypotheses. Section 4 introduces the research methodology and data description. Section 5 details the analysis of the empirical results. Section 6 presents the relevant conclusions and policy recommendations.

2. Literature Review

2.1. HC and Economic Development

Studies have found that HC has positive effects on the economy, such as improving efficiency, promoting growth, and reducing disparities [12,24,25,26]. Previous research has also determined that HC does not drive the economy in isolation, using three main approaches to explore the impact of HC in conjunction with other factors. The most common approach has been regressing a sample grouped according to economic characteristics such as development and economic structure [27,28,29]. Other studies have measured the degree of association by constructing interaction terms. For example, Li et al. [30] introduced an interaction term between renewable energy consumption and HC to explore the joint impact of the two factors on economic development in G10 countries. Rahim et al. [31] introduced an interaction term between natural resource fecundity and HC to examine the effect of these factors on economic growth. Impulse response and variance decomposition have also been used to investigate the interaction between factors [8].

2.2. INFRA and Economic Development

Research on INFRA and economic development has focused on examining INFRA’s impact on economic development and evaluations of INFRA subject to economic constraints. In the former approach, scholars have tended to focus on specific functional INFRA such as education, transportation, logistics, and internet access [32,33,34,35,36]. In the latter, scholars have established comprehensive systems to investigate CCD and economic development. Cui et al. [37] employed 41 indicators to measure INFRA and used a meta-frontier slacks-based measurement model to measure the economic carrying capacity of INFRA in China’s Beijing–Tianjin–Hebei region.

2.3. HC, INFRA, and Economic Development

The coupling coordination degree model (CCDM) has commonly been used to analyze the CCD of varying economic systems. However, scholars have not directly used the CCDM to explore the relationship between HC, INFRA, and economic development. Three primary types of relevant research perspectives have prevailed. The first involves incorporating the three factors as important elements in other systems. For example, Lin et al. [38] integrated HC into the social system to analyze the economic–social–INFRA–ecological resilience of the Yangtze River Economic Belt using the CCDM. Lu et al. [39] combined the rural population and agricultural output into the socioeconomic system to study the CCD between rural socioeconomics and road INFRA. Studies have also explored the state of HC and INFRA CCD under economic constraints. For example, Lu et al. [40] measured population quality and INFRA CCD in Qingdao, Weihai, and Yantai in Shandong Province, China from an urban integration perspective. Moreover, research has investigated the impact of INFRA on HC-economic CCD. For example, Sun and Cui [41] examined the impact of INFRA on the three-system CCD by integrating population health into the social system and measuring the socioeconomic–environmental benefits of CCD in four autonomous regions in China. Thereafter, Sun and Cui [14] included education level and residents’ health in the social system to explore the impact of urban public transportation INFRA on the CCD of economic, social, and environmental benefits. In addition, in econometric studies, Tsaurai and Ndou [42] explored the impact of the interaction between HC and INFRA on economic growth in transition economies by introducing a cross-term between the two into the regression equation. While the study made an important contribution and presented a valuable perspective regarding the interaction between HC and INFRA on economic development, it did not systematically present relevant mathematical modeling and deeper investigation, leaving room for the research in this study.

As noted above, previous research has deeply studied the relationships between HC, INFRA, and economic development, but has left room for additional investigation in three ways: (1) the majority of existing studies concerning the synergies between HC and other factors have employed basic econometric methods without using specific models; (2) in econometric studies, researchers can construct comprehensive HC and INFRA evaluation systems to explore the impact on economic development; and (3) while CCDM has been widely used to quantify the relationship between HC, INFRA, and economic development, scholars have not examined these factors from the perspective of HC and INFRA CCD on economic development. Therefore, this study applies the CCDM for the first time to examine the coupling coordination relationship between HC and INFRA and explore its impact on HQED.

3. Theoretical Analysis and Research Hypotheses

HC-INFRA CCD affects HQED through FA and FE, and there are threshold effects of CS and HEI for this process. The mechanism path diagram is shown in Figure 1.

3.1. The Influence Mechanism of HC-INFRA CCD on HQED

FA and FE are mechanistic pathways through which HC-INFRA CCD affects HQED. In this study, FA refers to the overall CCD of urban factors to measure the reasonable degree of the amount of inputs of various factors of production. FE refers to the economic outputs that can be brought about by various factors. Notably, the mainstream assumption has been that FA drives FE upgrading [6,43,44]; therefore, we consider FA and FE to be moving in the same direction. Economic development requires the accumulation of production factors [45]. However, it is difficult to achieve HQED simply by relying on improving factor quantities. Optimizing FA and improving FE can ensure the optimal use of factors and improve economic output, which is crucial for promoting HQED [46].

A considerable body of literature has demonstrated that optimizing HC and improving INFRA are the main channels for improving FA and FE [47,48,49]. When the HC-INFRA CCD is high, the two enhance one another in a process of mutual optimization, helping cities to rationally use factors. When HC-INFRA CCD is low, the positive economic effect of a high-level side will be weakened, and may even inhibit regional FA and FE.

Specifically, HC-INFRA CCD can optimize FA and enhance FE through homogeneous factor agglomeration, cost reduction, and internal factor mobility. HC-INFRA CCD attracts selective factor agglomeration and promotes homogeneous factor aggregation. With superior INFRA and high-quality HC, cities with high HC-INFRA CCD can promote the establishment and improvement of factor platforms such as financial institutions and talent markets, further concentrating various high-quality factors. Furthermore, high HC-INFRA CCD can establish demand regarding factor quality, which limits the entry of inferior elements into a city, subsequently helping homogeneous factors to be allocated appropriately and promoting FE upgrade. HC-INFRA CCD also reduces labor costs and factor market distortion. More residents in the coordinated coupling state will have access to the low-cost services provided by INFRA, reducing economic costs. If the INFRA level is too low, service quality will decline, causing more residents to opt for private services. If the INFRA level is too high, the local government will raise the price of services to maintain fiscal balance. Therefore, coupling dissonance forces the price of labor to increase due to the high cost of living. Inflated labor prices cause firms to overestimate labor productivity and excessively substitute labor with other factors, which reduces FA and FE. Finally, reasonable HC-INFRA CCD ensures that INFRA provides positive externalities. Transportation and communication INFRA breaks down spatial constraints on factor mobility and promotes efficient factor mobility, which improves FA and FE.

Based on the above analysis, we propose the following research hypotheses:

Hypothesis 1:

HC-INFRA CCD promotes HQED.

Hypothesis 2:

FA and FE are mechanistic pathways through which the action of HC-INFRA CCD on HQED takes place.

3.2. Threshold Effects of CS

Excessive CS will reduce the impact of HC-INFRA CCD on FA and FE, which subsequently reduces the positive effect on urban HQED. The geographical distance of oversized cities increases the transaction costs of factors and restricts factor mobility. In addition, the cost of living for residents in oversized cities will be higher. These circumstances will reduce FA and discourage HQED. Therefore, we propose the following research hypothesis:

Hypothesis 3:

CS has a reverse threshold effect on HC-INFRA CCD for advancing HQED.

3.3. HEI Threshold Effects

HEI influence HC-INFRA CCD and have an important role in promoting HQED. First, HEI have dual efficacy in HC enhancement and advancing INFRA construction, which enhances HC-INFRA CCD. As potential quality HC inputs, high school graduates are expected to be the main contributors to HQED in the future [50]. HEI serve as a carrier, preparing the region to provide high-quality HC output [51]. HEI will also promote other INFRA construction such as transportation and healthcare in the region in the process of promoting education INFRA development, which promotes comprehensive INFRA. With cutting-edge expertise, specialized research departments, and abundant research talent, HEI are the sites of basic knowledge development, research, and scientific and technological innovation [52]. HEI actively participate in knowledge creation and scientific activities through the integration of industry, academia, and research, prompting the fusion of various high-quality production factors and promoting HQED. Therefore, we propose the following research hypothesis:

Hypothesis 4:

HEI have a positive threshold effect on HC-INFRA CCD for advancing HQED.

4. Methodology and Data

This section describes the measures of the study’s main variables, the research models, and the data sources. Section 4.1 constructs a comprehensive evaluation system for HC, INFRA, and HQED. Section 4.2 describes the main research models used in this study with corresponding purposes. We first measured the comprehensive utility values of HC, INFRA, and HQED using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) model. We then calculated the HC-INFRA CCD based on an optimized CCDM. Next, we constructed a benchmark regression model to examine the effect of HC-INFRA CCD on HQED, and then constructed a threshold model to verify the threshold effects of CS and HEI. Finally, we built the PM to verify the mediating effects of specific FA and FE. Section 4.3 describes the data sources for this study.

4.1. Indicators

4.1.1. HC and INFRA

Considering that HC and INFRA refer to multidimensional and complex systems, a single indicator cannot comprehensively measure their development. By summarizing the related research in high-level academic journals within the past 5 years [37,53,54,55,56,57,58,59], and based on the principles of comprehensiveness, scientificity, and availability [60], we constructed an HC evaluation system containing 6 indicators and an INFRA evaluation system containing 18 indicators. The specific systems are presented in

Table 1. Human capital (HC) evaluation system.

Dimensions	Indicators	Clarification	Effect
Age	Old-age dependency ratio	Ratio of unemployed to employed persons in units at the end of the year	−
Employment	Manpower Advancement Index	(Employees in primary sector + 2× employees in secondary sector + 3× employees in tertiary sector)/total employees	+
Space	Population density	Ratio of resident population to area	+
Educational	Urbanization rate	Ratio of urban population to total population	+
	Proportion of higher education qualifications	Ratio of the number of people with university education or above to the total population	+
Security	Basic pension insurance coverage	Ratio of the number of urban workers insured to the number of employed people	+

Note: + indicates positive data and − indicates negative data.

and

Table 2. Infrastructure (INFRA) evaluation system.

Dimensions	Indicators
Water supply	Length of water pipeline
	Total water supply
Gas supply	Length of gas supply pipeline
	Total gas supply
Sewage treatment	Sewage treatment rate
	Centralized sewage treatment rate
Garbage disposal	Domestic waste disposal rate
	Non-hazardous treatment rate of domestic waste
Greener	Green coverage
	Green ratio
Transportation	Percentage of road length
	Percentage of road area
Medical services	Hospitals per capita
	Doctors per capita
Education	Schools per capita
	Teachers per capita
Telecommunication	Telecommunications per capita
	Share of cell phone subscribers

Note: All indicators are positive.

, respectively.

4.1.2. HQED

China’s new development philosophy is an important connotation of HQED, which continues to feature economic growth as an important goal. Therefore, referencing multiple studies [61,62,63,64], we constructed a HQED indicator system that included six dimensions: economy, innovation, coordination, greenness, openness, and sharing. The specific indicators are detailed in

Table 3. High-quality economic development (HQED) Indicator System.

Dimension	Indicators	Clarification	Effect
Economy	Economic growth rate	Ratio of GDP value added to current year’s GDP	+
	GDP per capita	GDP/population	+
Innovation	Patent output	Number of patents granted	+
	Science and education input	Ratio of financial expenditure on science, technology, and education	+
Coordination	Industrial structure	Ratio of tertiary industry value added to that of secondary industry	+
	Rural–urban income ratio	Ratio of urban and rural residents’ disposable income per capita	−
Greenness	Carbon intensity	Carbon dioxide emissions/real GDP	−
	Emissions from industrial waste	Ratio of industrial wastewater, sulfur dioxide, smoke (powder), and dust emissions to industrial output value	−
Openness	Foreign capital use	Ratio of total imports and exports to GDP	+
	Actual rate of foreign capital used	Ratio of actual foreign capital used to GDP	+
Sharing	Wage level	Average wage of employed workers	+
	Population consumption	Ratio of social consumption retail sales to GDP	+

Note: + indicates positive data and − indicates negative data.

.

4.2. Research Model

4.2.1. TOPSIS Model

The TOPSIS model relies on the differences in sub-indicators’ data to calculate the composite indicator and circumvent subjective assignment errors. This model reflects the weights and differences of sub-indicators with high accuracy [65]. Therefore, we chose this technique to measure HC, INFRA, and HQED. The specific calculation process standardized processing, established indicators’ weights, and calculated the composite value.

Step (1) Standardized processing.

To avoid the impact of differences in indicators’ measurement units, we standardized the indicators using the extreme deviation method. Different calculation formulas were used for positive and negative indicators, as shown below:

$$X_{ij}=\left\{\begin{array}{l}\left(x_{ij}-\min x_j\right)/\left(\max x_{ij}-\min x_j\right),x_{ij}\mathrm{is} \mathrm{a} \mathrm{positive} \mathrm{indicator}\hfill \\ \left(\max x_{ij}-x_j\right)/\left(\max x_{ij}-\min x_j\right),x_{ij}\mathrm{is} \mathrm{a} \mathrm{negative} \mathrm{indicator}\hfill \end{array}\right.$$

(1)

where X_ij is the standardized indicator value, X_ij denotes the value of indicator j for city i, and max x_j and min x_j denote the maximum and minimum values of indicator j.

Step (2) Establishing indicator weights.

The steps of the entropy weighting method to establish indicator weights are detailed below.

The weights (P_ij) were calculated as follows:

$$p_{ij}={\displaystyle \frac{X_{ij}}{{\displaystyle \sum _n^{i=1}}{X}_{ij}}}$$

(2)

The information entropy of each indicator G_j was calculated as follows:

$$G_j=1-{\displaystyle \frac{1}{\ln n}}\sum _n^{i=1}{P}_{ij}\ln P_{ij}$$

(3)

Entropy weights (W_j) were established as follows:

$$W_j={\displaystyle \frac{G_j}{{\displaystyle \sum _m^{j=1}}{G}_j}}$$

(4)

Step (3) Calculating the composite value.

The formula for calculating the combined utility value is as follows:

$$U_i=\sum _m^{j=1}{W}_{j}Pij$$

(5)

4.2.2. Optimized CCDM

Coupling was originally a physical term referring to the state of two or more systems interacting, influencing, and coordinating with one another, and has since been widely used in various research fields [66]. CCD can only measure the effect of interactions between systems and reflect the effect and coordination relationship of the system as a whole [67]. Therefore, we employed the CCDM to measure the coupling coordination relationship between HC and INFRA as follows:

$$C=2\sqrt{{\displaystyle \frac{U_1{U}_2}{{\left(U_1+U_2\right)}^2}}}$$

(6)

$$D=\sqrt{C\times T}$$

(7)

$$T=a{U}_1+b{U}_2$$

(8)

where U₁ and U₂ denote the utility values of HC and INFRA, respectively, C denotes the combined coupling degree of the two, D denotes the CCD of the two, T denotes the integrated coordination index, a and b are coefficients that measure the influence of the HC and INFRA subsystems on the whole system, where the sum of the influence coefficients is 1, and each is generally considered to be equally important in most models, both taking a value of 0.5 [68,69]. Dong et al. [70] used the combined utility value of the subsystems to measure the CCD of both for the whole system. In this study, HC and INFRA are explanatory variables in the underlying regression analysis, and their regression coefficients are considered to reflect their impact on the system more accurately. Therefore, we use the ratio of the coefficients to the sum of their coefficients to measure their weights in the whole system. After subsequent measurement, a and b take values of 0.508 and 0.492, respectively.

4.2.3. Benchmark Regression Model

Based on the theoretical analysis above, this study constructs a fixed-effects regression model to explore the effect of HC-INFRA CCD on HQED as follows:

$$HQE{D}_{i,t}={\beta }_0+{\beta }_{1}HC-INFRA CC{D}_{i,t}+{\beta }_{2}F{D}_{i,t}+{\beta }_{3}I{A}_{i.t}+{\beta }_{4}G{I}_{i.t}+{\beta }_{5}M{E}_{i,t}+C_t+u_{i,t}$$

(9)

where HQED is the explanatory variable; HC-INFRA CCD is the core explanatory variable; and FD, IA, GI, and ME are the control variables. FD denotes financial development, which is expressed by the ratio of the balance of deposits and loans in the financial industry [71,72]. Financial industry development ensures that savings can be efficiently used to provide financial support for enterprise development, which promotes HQED. IA denotes industrial agglomeration, which is expressed by the total output value of industries above scale. Industrial agglomeration promotes industrial technological innovation and improves productivity through scale and technology effects, which drives the development of the whole economy [73]. GI represents government intervention, which is expressed as the ratio of expenditure to revenue in government finance. Governments use financial expenditure to provide economic services to residents, enterprises, and foreign investors, and rely on policies to promote economic development [74,75]. ME is the market environment, which is expressed as the ratio of fixed assets of the whole society to GDP. A good market environment produces more efficient economic output and is a crucial aspect of achieving regional HQED [76]. β₀ is a constant term, β₁–β₅ are the impact coefficients of the explanatory variables, C_t is city fixed effects, u denotes a random error term, and i and t denote city and year, respectively.

4.2.4. Threshold Models

Referencing the threshold model [77], we constructed the model as follows:

$$\begin{array}{c}HQE{D}_{i,t}={\alpha }_{1}HC-INFRACC{D}_{i,t}•I(CS\le Y_{CS})+{\alpha }_{2}HC-INFRACC{D}_{i,t}•I(CS>Y_{CS})+\\ {\beta }_0+{\beta }_{2}F{D}_{i,t}+{\beta }_{3}I{A}_{i.t}+{\beta }_{4}G{I}_{i.t}+{\beta }_{5}M{E}_{i,t}+C_{t+ui}\end{array}$$

(10)

$$\begin{array}{c}HQE{D}_{i,t}={\alpha }_{3}HC-INFRACC{D}_{i,t}•I(HEI\le Y_{HEI})+{\alpha }_{4}HC-INFRACC{D}_{i,t}•I(HEI>Y_{HEI})+\\ {\beta }_0+{\beta }_{2}F{D}_{i,t}+{\beta }_{3}I{A}_{i.t}+{\beta }_{4}G{I}_{i.t}+{\beta }_{5}M{E}_{i,t}+C_t+u_{i,t}\end{array}$$

(11)

where CS and HEI are threshold variables. We use cities’ population size and the number of universities to measure CS and HEI, and Y_CS and Y_HEI are the corresponding thresholds. I (.) denotes the exponential function. α_1–4 are the regression coefficients of HC-INFRA CCD. All other variables are consistent with Equation (9).

4.2.5. PM

The stepwise method of testing mediating effects suffers from insufficient causal identification, which biases the parameter estimates. The PM can incorporate multiple paths into a single analytical model to identify the overall mediating paths. This can produce estimated parameters that reflect the causal relationships of the variables more accurately [78]. Therefore, this study used the PM to explore the multiple paths of HC-INFRA CCD for HQED.

In Section 3, FA and FE refer to the overall allocation and total efficiency of all factors, respectively; however, the effect of HC-INFRA CCD on specific FA and FE will differ, primarily due to factors’ characteristics. For example, an increase in the HC-INFRA CCD implies that the HC level will also increase, and labor allocation and efficiency should also be stronger. Since INFRA construction requires occupation of urban land, the impact of HC-INFRA CCD on land allocation and efficiency will also differ from other factors. Therefore, to further explore the mechanism path, we subdivided FA into CEA, LEA, and CLA, and FE into LRE, CE, and LE. We used GDP, capital stock, number of employees, and area of urban built-up area to measure economy, capital, labor force, and land, respectively. Since the CCDM can also measure the degree of allocation among factors more accurately [79], we used this model to measure FA, and we used the ratio of economic outputs to factor inputs to measure FE. Finally, this study constructed a PM containing one explanatory variable and six mediating variables. Figure 2 illustrates the initial paths of the HC-INFRA CCD affecting HQED.

4.3. Data Description

This study used 184 cities in China from 2011 to 2019 as the research sample. The data were obtained from the China Statistical Yearbook, the China Urban Statistical Yearbook, the China Urban Construction Statistical Yearbook, the Economic Prediction System Database, and Chinese Research Data Services. Individual missing data were filled in using linear interpolation. In addition, all data were logarithmized to exclude the effect of heteroskedasticity.

Table 4. Descriptive statistical analysis.

Variable	Observations	Mean	Standard Deviation	Minimum	Maximum
HQED	1656	0.1044	0.074	0.028	0.620
Human capital-infrastructure coupling coordination degree (HC-INFRA CCD)	1656	0.354	0.087	0.225	0.758
Financial development (FD)	1656	1.572	0.416	0.208	4.471
Industrial agglomeration (IA)	1656	37.898	49.551	0.315	324.451
Government intervention (GI)	1656	0.471	0.224	0.070	1.541
Market environment (ME)	1656	0.799	0.273	0.026	2.296

presents the descriptive statistics of the study’s variables.

5. Empirical Results and Analysis

5.1. Benchmark Regression

Table 5. Baseline regression results.

Variable	(1)	(2)
Ln HC	0.322 ***
	[0.063]
Ln INFRA	0.312 ***
	[0.064]
Ln HC-INFRA CCD		1.266 ***
		[0.163]
Ln FD	−0.246	−0.246
	[0.127]	[0.128]
Ln IA	0.069 **	0.069 **
	[0.022]	[0.022]
Ln GI	−0.269 ***	−0.269 ***
	[0.046]	[0.046]
Ln ME	0.103 ***	0.103 ***
	[0.025]	[0.025]
Constant	−2.352 ***	−2.343 ***
	[0.432]	[0.428]
N	1656	1656
R-squared	0.451	0.451

Note: **, and *** indicate significance at 5%, and 1% levels, respectively; t-values in [].

presents the results of the baseline regression of HC and INFRA affecting HQED. Column (1) shows the regression results introducing HC and INFRA as variables. The coefficients of HC and INFRA are 0.322 and 0.312, respectively, both of which are significant at a 1% level. This indicates that HC and INFRA can drive HQED. Column (2) shows that the coefficient of HC-INFRA CCD is 1.266, passing the significance test at the 1% level. This indicates that the development of coupling coordination between HC and INFRA is a much greater driver of HQED than the single factors, confirming H1.

Among the cities in the study sample, Hefei’s HC-INFRA CCD efficiently drives HQED. The HC-INFRA CCD increased by 0.074 during the sample years, while the HQED indicator increased by 0.139, representing the third highest growth rate. This is attributable to Hefei prioritizing HC optimization and INFRA construction. Hefei focuses on education investment and talent cultivation. The average growth rate of the city’s share of financial expenditures on education is 3.15%, ranking sixth and well above average. Correspondingly, residents’ education level has significantly improved, with a growth of highly educated residents that is three times higher than the average. Hefei also actively improves INFRA construction in response to population and industry distribution, which improves HC and economic operation. The city had up to 1370 new major construction projects in 2019, with an investment of up to $8.25 billion, accelerating INFRA construction, which narrowed the gap between INFRA and HC. During the 9 years examined in our study, the INFRA level increased by 55%, which is 2.97 times higher than that of HC, and improved the circumstances of INFRA lagging behind HC. Therefore, the economic development of Hefei is moving toward HQED by upgrading and coupling coordination of HC and INFRA.

To further analyze the impact of HC-INFRA CCD on HQED segmentation dimensions, based on the indicator system constructed above, we next regressed the economic, innovative, coordinated, green, open, and shared development indices as explanatory variables, presenting the results in

Table 6. Regression results of HQED segmentation dimensions.

Variable	(1) Economy	(2) Coordination	(3) Greenness	(4) Innovation	(5) Openness	(6) Sharing	(7) Scale of Openness
Ln HC-INFRA CCD	2.135 ***	1.895 ***	0.220 **	2.540 ***	−0.954 ***	1.631 ***	0.432 *
	[0.249]	[0.243]	[0.096]	[0.337]	[0.357]	[0.212]	[0.246]
Ln FD	−0.501 **	−0.276	−0.091 *	−0.431 **	0.098	−0.279	−0.264 ***
	[0.217]	[0.271]	[0.055]	[0.207]	[0.251]	[0.201]	[0.085]
Ln IA	0.123 ***	0.010 ***	0.023 ***	0.125 **	−0.022	0.129 ***	0.104 ***
	[0.036]	[0.031]	[0.007]	[0.049]	[0.054]	[0.031]	[0.032]
Ln GI	−0.421 ***	−0.430 ***	−0.045 ***	−0.503 ***	0.310 ***	−0.370 **	−0.081
	[0.071]	[0.072]	[0.014]	[0.086]	[0.098]	[0.059]	[0.078]
Ln ME	0.136 ***	0.118 ***	0.014 *	0.150 ***	0.171 ***	0.111 ***	0.167 ***
	[0.038]	[0.040]	[0.008]	[0.041]	[0.060]	[0.033]	[0.044]
Constant	0.269	0.431	−0.863 ***	0.697	0.308	−0.367	−6.273 ***
	[0.683]	[0.642]	[0.120]	[0.944]	[1.045]	[0.646]	[0.618]
N	1656	1656	1656	1656	1656	1656	1656
R-squared	0.500	0.462	0.119	0.423	0.060	0.517	0.096

Note: (1) *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively; t-values are in []. (2) The Hausman test for the model on green indicators failed the original hypothesis and we chose a random effects regression. Unless otherwise specified, other regressions default to the fixed-effects model.

. The results show that HC-INFRA CCD promotes economic, innovative, coordinated, green, and shared development indices to different degrees, all of which pass the significance test at the 5% level.

HC-INFRA CCD has the strongest effect on enhancing the innovative development index, with a regression coefficient as high as 2.540. This is because the positive effect of INFRA affects the entire process of knowledge innovation through creation, accumulation, exchange, and agglomeration overflow, with high-quality HC as the main body of innovation. Specifically, water supply, power supply, medical care, education, and other INFRA functions provide residents with convenience in executing production and life functions, optimizing the allocation of time, and guaranteeing the accumulation of knowledge, which subsequently improves overall knowledge innovation capabilities and processes. Digital INFRA promotes information dissemination and knowledge diffusion under near-zero marginal communication costs, which improves the efficiency of innovation factor exchange. Transportation INFRA construction and upgrading enhances the free flow of innovation factors in the region, which promotes innovation factor transport between regions, advancing factor agglomeration and knowledge spillover effects. HC-INFRA CCD contributes less to the green development index, with a regression coefficient of only 0.220. The probable rationale for this is that the HC upgrading and INFRA construction process consumes various resources and increases pollutant emissions, which weakens the contribution to green development.

In addition, HC-INFRA CCD suppresses the open development index, which may be because this study used the ratio of total imports and exports and real foreign capital use to GDP to measure the level of openness to eliminate the interference of economies’ size. China’s GDP increased by 202.17% during the sample years; however, total imports and exports and actual foreign capital use increased by 125.69% and 117.33%, respectively. As a result, the pulling effect on these factors was smaller than that on GDP, producing a negative regression coefficient. To verify this conjecture, we normalized the total import and export and foreign capital data and constructed an open-scale development index according to the measured weights. Column (7) in Table 6 presents the regression results of HC-INFRA CCD using this open-scale development index. The coefficient is significantly positive, confirming that enhancing HC-INFRA CCD can effectively increase the scale of openness.

We next conducted further empirical analysis due to the regression model of the development indices for green, open, and open-scale potentially having insufficient explanatory power. The regression results for these indices show that R-squared is only 0.119, 0.060, and 0.096, respectively, and are far from the other dimensions, which are basically at the same level. There is no uniform standard in the economic field, but in contrast, the credibility of the regression results deserves a subsequent note. Therefore, we employ the Monte Carlo method to test the credibility of the three models. The variables are set to follow a normal distribution with a variance of 1, and 10,000 sets of random numbers are generated. The regression results in

Table 7. Monte Carlo method test results.

Variable	(1) Greenness	(2) Openness	(3) Scale of Openness
Ln HC-INFRA CCD	−0.492 ***	0.104 ***	0.209 ***
	[0.009]	[0.007]	[0.007]
Ln FD	0.054 ***	−0.0404 ***	−0.126 ***
	[0.007]	[0.007]	[0.007]
Ln IA	0.011 ***	0.012 ***	0.0516 ***
	[0.0001]	[0.0001]	[0.0005]
Ln GI	0.159 ***	−0.0200 ***	−0.0373 ***
	[0.007]	[0.007]	[0.007]
Ln ME	0.0858 ***	0.0020	0.0779 ***
	[0.007]	[0.007]	[0.007]
Constant	0.476 ***	−0.794 ***	0.175 ***
	[0.014]	[0.014]	[0.018]
N R-squared	10000	10000	10000
	0.495	0.500	0.495

Note: *** indicate significance at 1% levels; t-values are in [].

show that the impact coefficients are all in the same direction as the original model and have similar values, passing the 1% significance level test, while the models’ R-squared values are similar to the remaining dimensions in Table 6. These results improve the explanatory power of the regression model.

5.2. Threshold Effect Analysis

5.2.1. Threshold Estimation

This study uses the bootstrap method to examine threshold effects, presenting the test results in

Table 8. Threshold effect test.

Variant	Threshold	F-Statistics	Prob	Crit10	Crit5	Crit1
City scale (CS)	Single	37.34	0.097	37.093	43.940	53.595
	Double	18.62	0.390	31.821	39.949	49.849
Higher education institutions (HEI)	Single	77.17	0.001	27.555	34.284	43.734
	Double	21.40	0.153	24.506	30.843	41.580

Note: Crit10, Crti5, and Crit1 indicate the critical values of the F statistics at the 10%, 5%, and 1% confidence levels, respectively.

. The p-values of the single-threshold effects of CS and HEI are 0.097 and 0.001, respectively, which are less than 1. The p-values of the double threshold effects are all greater than 0.1. In summary, a single-threshold effect of HC-INFRA CCD on HQED is evident, with CS and HEI as the threshold variables.

Table 9. Threshold estimation results.

Threshold Variables	Threshold Value	Confidence Interval
CS	54.070	[53.145, 54.200]
HEI	26.000	[25.000, 31.000]

Note: confidence interval means the interval at the 95% confidence level.

shows the estimated thresholds and confidence intervals. The threshold for CS is 54.07 and the confidence interval is located at [53.145, 54.200], and the threshold for HEI is 26 and the confidence interval is located at [25,31]. The likelihood ratio (LR) functions for CS and HEI are plotted in Figure 3a and Figure 3b, respectively. The dashed line is the LR threshold corresponding to the 5% significance level. The results confirm that the thresholds are lower than the critical values and pass the LR test, indicating that the estimated thresholds are valid.

5.2.2. Analysis of Threshold Effect Results

The regression results with CS as the threshold are shown in column (1) of

Table 10. Single-threshold panel model regression results.

Variable		(1)	(2)
Ln HC-INFRA CCD	(CS ≤ 54.07)	1.310 ***
		[0.085]
	(CS > 54.07)	1.221 ***
		[0.085]
	(HEI ≤ 26)		1.049 ***
			[0.088]
	(HEI > 26)		2.925 ***
			[0.217]
Ln FD		−0.250 ***	−0.248 ***
		[0.029]	[0.029]
Ln IA		0.070 ***	0.067 ***
		[0.012]	[0.012]
Ln GI		−0.262 ***	−0.256 ***
		[0.022]	[0.022]
Ln ME		0.0964 ***	0.122 ***
		[0.013]	[0.013]
Constant		−2.383 ***	−2.407 ***
		[0.233]	[0.230]
N		1656	1656
R-squared		0.395	0.410

Note: *** indicate significance at 1% levels; t-values are in [].

. When CS is less than the threshold value of 54.07, the coefficient of HC-INFRA CCD is 1.310, which passes the significance test at the 1% level. When CS is above the threshold, the coefficient drops to 1.221, which passes the test of significance at the 1% level. The results indicate that CS expansion reduces the driving effect of HC-INFRA CCD on HQED, confirming H3. Notably, when CS crosses the threshold, the effect of HC-INFRA CCD on urban HQED only decreases slightly, which may be attributable to a larger CS, implying a larger body of factors that can be more easily searched for and identified. This reduces the negative effect of high transaction costs on FA and safeguards the promotion of HQED by HC-INFRA CCD.

Column (2) of Table 10 shows the regression results with HEI as the threshold. When HEI ≤ 26, the coefficient of HC-INFRA CCD is 1.049, which passes the significance test at the 1% level. When HEI exceeds the threshold, the coefficient rises to 2.925, which is nearly three times higher and passes the significance test at 1% level. This indicates that HEI efficiently drives the positive effect of HC-INFRA CCD on HQED, confirming H4. The 2019 sample data reveal that 18 cities have HEI numbers above the threshold. Among them, Beijing ranked first, with 93 universities. Among the sample cities, Beijing ranked seventh from the bottom in terms of HC-INFRA CCD growth. The city’s HQED index ranks third in terms of growth rate, which is attributable to the city’s emphasis on HEI development and ensures that a low-growth HC-INFRA CCD leads to a high rate of HQED. Specifically, Beijing’s education expenditure has continued to grow, from $11.41 billion in 2011 to $21.42 billion in 2019, representing a 187.66% increase. Education spending in 2019 accounted for 2.86% of the nation’s total, which was well above the average. High education funding supports HEI’s teaching and research quality and promotes continuous high-level talent cultivation. In addition, Beijing has established more research platforms such as the Engineering Research Center of the Ministry of Education and the Technology Transfer Alliance of Universities. These approaches are effective in promoting the allocation of high-level HC from universities to high-quality elements of high-tech industries, promoting the development of Beijing’s economy in the direction of HQED.

5.3. Robustness Tests

5.3.1. Endogeneity Testing

There are two main endogeneity concerns in this study. First, the choice of Chinese urban panel data as the sample makes the data less available than provincial and cross-sectional data, which may incur the problem of omitting key variables. Second, while HC-INFRA CCD influences HQED, HQED will also drive INFRA construction and attract talent concentration, subsequently increasing HC-INFRA CCD in a bidirectional causal relationship. Therefore, this study employs the two-stage least squares (2SLS) approach to mitigate the endogeneity problem.

We used the CCD of the number of scholars in China’s Ming Dynasty and the degree of terrain leveling as instrumental variables (IVs). In terms of the variables’ correlation, HC is intergenerationally heritable, and high-quality HC in the past carries over to the present [80]. Considering the Jinshi system as a means of selecting talent in ancient China, the number of Jinshi can comprehensively reflect the past HC level. Therefore, the number of Jinshi in the Ming Dynasty has a positive correlation with present day HC. For INFRA, leveled terrain reduces the complexity of development and the cost of construction and is more likely to result in superior regional INFRA, indicating that INFRA has a positive correlation with terrain leveling. In terms of variable exogeneity, as a time gap occurred between the Ming Dynasty Jinshi and the current HQED stage, no direct relationship exists. As a natural geographic factor, the degree of terrain leveling does not change with economic changes and does not have a direct relationship with HQED. In summary, the selected IVs satisfy relevance and exogeneity.

Table 11. 2SLS estimate results.

Variable	First-Stage	Second-Stage
Instrumental variable		4.734 ***
		[0.487]
Ln HC-INFRA CCD	19.004 ***
	[0.916]
Ln FD	−0.486 ***	0.160 *
	[0.178]	[0.093]
Ln IA	−0.001	0.008
	[0.005]	[0.258]
Ln GI	−0.007	0.010
	[0.010]	[0.581]
Ln ME	0.011 **	0.002
	[0.006]	[0.360]
Constant	−131.189	2.436
	[6.257]	[0.469]
Cragg–Donald Wald F statistic	430.660
	(16.38)
Kleibergen–Paap rk LM statistic	48.702
	{0.000}

Note: (1) *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively; t-values are in []. (2) The value in () denotes the critical value of the Stock–Yogo weak identification test at the 10% level. The value in {} denotes the p-value of the Kleibergen–Paap rk LM statistic.

presents the 2SLS regression results. The Cragg–Donald Wald F statistic is 430.660, which is well above the 10% threshold. The Kleibergen–Paap rk LM statistic is 48.702 with a p-value of 0.000. These results indicate that the IV passes weak identification and underidentification tests, confirming that the selected IV is appropriate. The results of the first-stage 2SLS regression show that the IV has a positive correlation with the explanatory variables at a 1% significance level. The second-stage regression results show that the significance and direction of the coefficients are consistent with the baseline regression, confirming that the conclusion that HC-INFRA CCD promotes HQED remains robust after considering endogeneity issues.

5.3.2. Other Robustness Tests

Constructing an interaction term between two variables is the most common method to examine the influence of coupling coordination [81,82]. Therefore, this study constructs the interaction term of HC and INFRA instead of HC-INFRA CCD as a robustness test. We also construct an evaluation analysis based on matching degree rather than HC-INFRA CCD, referencing Zuo et al. [83]. Columns (1) and (2) of

Table 12. Results of other robustness tests.

Variable	(1)	(2)	(3)	(4)	(5)
Ln HC-INFRA CCD	0.317 ***	0.634 ***	0.503 ***	0.632 ***	0.999 ***
	[0.041]	[0.081]	[0.072]	[0.096]	[0.151]
Ln FD	−0.246	−0.246 *	−0.106	−0.124 *	−0.445 ***
	[0.128]	[0.128]	[0.055]	[0.072]	[0.074]
Ln IA	0.069 **	0.069 ***	0.041 ***	0.053 ***	0.046 **
	[0.022]	[0.022]	[0.010]	[0.012]	[0.021]
Ln GI	−0.269 ***	−0.269 ***	−0.112 ***	−0.129 ***	−0.224 ***
	[0.046]	[0.046]	[0.018]	[0.022]	[0.039]
Ln ME	0.103 ***	0.103 ***	0.036 ***	0.052 ***	0.168 ***
	[0.025]	[0.025]	[0.010]	[0.014]	[0.030]
Constant	−2.347 ***	−2.347 ***	−1.200 ***	−3.149 ***	−2.110 ***
	[0.428]	[0.428]	[0.194]	[0.246]	[0.417]
N	1656	1656	1656	1656	1656
R-squared	0.4506	0.4691	0.4778	0.493	0.456

Note: *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively; t-values are in [].

present the regression results for these two methods. The results reveal that the significance and direction of the coefficients are the same as those of the benchmark regression. Columns (3) and (4) of Table 12 show the regression results of HQED measured using linear weighting and mean square deviation methods in contrast to the entropy weighting method. The results demonstrate that HC-INFRA CCD still drives HQED significantly. Column (5) shows the results of the regression after bilaterally winsorizing the data by 5%. Once again, the results show that the HC-INFRA CCD coefficient is significantly positive. The above analyses further validate the robustness of the conclusion that HC-INFRA CCD drives urban HQED.

5.4. Mechanism Testing

The PM was first proposed by Wright as an extension of multiple linear regression [84]. The model is based on a covariance matrix, using covariance to reflect the linear relationship between variables [85]. This approach enables multiple paths to be identified in the same analytical framework. The model has commonly been used to analyze the mediating effects of multiple variables in social science fields such as economics, psychology, and sociology [86]. The Sobel test and bootstrap method are commonly used to test for mediating effects. The premise of the Sobel test is that the variable being tested conforms to a normal distribution, and the bootstrap method is a non-parametric statistical method [87]. This approach constructs a dataset by repeated sampling of the original data, with no parametric distribution requirements for the original data. Notably, the product of path coefficients in this study does not necessarily obey a normal distribution; therefore, we employed the bootstrap method to estimate the path coefficients to obtain robust results more accurately. The specific steps are as follows: (1) examine the initial model fit test, (2) establish the final model that meets the fit index, and (3) estimate the path coefficients using the bootstrap method.

5.4.1. Model Fit Test

This study used SRMR, RMSEA, TLI, GFI, and CFI indices to measure the fit between the model and the sample data. We regressed the PM using the maximum likelihood method, revealing that individual fit indices did not pass the reference standard. Therefore, the residual terms of the relevant mediating variables were established based on the Modification Indices reported by the AMOS28.0 software. In addition, considering that some individual path coefficients did not pass the significance test, we excluded the relevant paths. The fitting indices after optimizing the model paths are presented in

Table 13. Fitting indices for PM.

Fitness Index	SRMR	RMSEA	TLI	GFI	CFI
Model parameter	0.001	0.014	0.999	0.999	0.999
Reference standard	<0.05	<0.05	>0.9	>0.9	>0.9

Note: (1) SRMR, RMSEA indicate standardized root mean square residual and root mean square error of approximation, respectively. The smaller their value, the better the fitness of the model. (2) TLI, GFI, and CFI indicate Tucker–Lewis index, goodness-of-fit index, and comparative fit index, respectively. the larger their value means the better the fitness of the model.

. The selected fit indices all satisfy the reference standard, indicating a good fit between the model and the sample data.

5.4.2. Analysis of PM Results

Table 14. Overall path estimation results.

Effect Type	Estimate	Standard Error	Lower	Upper
Direct effect	0.330 ****	0.023	0.286	0.373
Total indirect effect	0.261 ****	0.019	0.224	0.299
Total effect	0.591 ****	0.016	0.557	0.621

Note: (1) **** indicates a 1‰ significance level, implying a p-value < 0.001. (2) Lower and upper denote the lower and upper limits of the confidence interval at the 95% level, respectively.

and

Table 15. Path-specific estimation results.

Number		Specific Path			Estimate	Standard Error	Lower	Upper
1	HC-INFRA CCD	Capital-labor allocation		HQED	0.173 ****	0.012	0.143	0.207
2		Land efficiency (LE)			−0.065 ****	0.007	−0.080	−0.051
3		Capital efficiency (CE)			0.090 ****	0.012	0.067	0.116
4		Labor efficiency (LRE)			0.024 ****	0.005	0.015	0.035
5		Capital-economic allocation	LE		0.046 ****	0.006	0.035	0.059
6			CE		0.012 ****	0.003	0.006	0.018
7			LRE		0.061 ****	0.008	0.047	0.077
8		Capital-labor allocation	CE		−0.023 ****	0.004	−0.032	−0.016
9			LRE		−0.065 ****	0.009	−0.083	−0.049
10			LE		0.024 ****	0.006	0.013	0.037
11		Labor-economic allocation	LE		−0.004 ****	0.001	−0.006	−0.003
12					−0.013 ****	0.002	−0.018	−0.009

Note: (1) **** indicates a 1% significance level. (2) Lower and upper denote the lower and upper limits of the confidence interval at the 95% level, respectively.

present the results of the PM estimation using the bootstrap method. The results reveal that the mediating effects of HC-INFRA CCD on HQED were all significant (p-value < 0.001, 95% confidence interval not including 0), which confirms H2. The direct and indirect effects validating the mediating effect on PM are 0.330 and 0.261, accounting for 55.84% and 54.16% of the total effect, respectively. The indirect effect estimate for path 1 is the largest, accounting for 66.28% of the total indirect effect. CLA is the most significant mediating path.

We next analyze the remaining paths. The results of paths 2–4 show that HC-INFRA CCD can drive HQED by boosting CE and LRE, but it hinders HQED by suppressing LE to a small extent. The rationale for this may be that INFRA expansion takes up land but generates poor economic outcomes, which leads to a decline in LE and subsequently reduces the quality of economic development. Paths 5 and 10 reveal that HC-INFRA CCD still acts positively on HQED by indirectly raising LE through CEA and CLA. The estimates of paths 5–7 are all positive, indicating that HC-INFRA CCD can raise FE through CEA, which promotes HQED. Paths 8–10 use CLA and FE as mediators, indicating that when FE is LE, HC-INFRA CCD can drive HQED through these mediators. In contrast, when FE is CE or LRE, HC-INFRA CCD suppresses HQED. The reasons for this may be twofold. First, HC-INFRA CCD drives HQED directly through the efficient allocation of capital and labor, weakening its role by first affecting FA and then FE. Second, due to the competitive relationship between CE and LRE, rational allocation does not ensure continuous improvement of their efficiency. Paths 11 and 12 both contain LEAs, showing that both paths inhibit HQED, which may be because regional development is not sufficiently driven by labor and economic allocation alone to promote HQED, but rather requires quality capital to be rationally allocated to the economy and the labor force.

6. Conclusions and Policy Recommendations

6.1. Conclusions

Scholars have reached a consensus on the importance of HC and INFRA for HQED [8,27,88,89]. However, previous research has neglected to investigate the economic effects concerning their CCD. First, this study adopts the CCDM to measure the CCD between HC and INFRA. Second, we use fixed effects, threshold effects, and pathway models to verify the effect of HC-INFRA CCD on HQED. We draw the following four relevant conclusions. (1) HC-INFRA CCD efficiently drives HQED compared with a single factor. The baseline regression verifies that the coefficient of HC-INFRA CCD is almost twice as large as the sum of the coefficients of HC and INFRA. (2) HC-INFRA CCD can promote economic, coordinated, green, innovative, and shared development, with the strongest effect on promoting innovative development and the weakest effect on promoting green development. In addition, although HC-INFRA CCD reduces open development, after removing the effect of cities’ economic size to measure the open development index, HC-INFRA CCD increases the size of cities’ openness. (3) The HC-INFRA CCD promotes HQED with a single-threshold effect of CS and HEI. Specifically, if CS is too large it increases the difficulties of factor aggregation and cost of living and slightly diminishes the driving effect of HC-INFRA CCD. In contrast, HEI strengthens the driving effect of HC-INFRA CCD. The coefficient of HC-INFRA CCD is nearly tripled when HEI crosses the threshold. (4) HC-INFRA CCD affects HQED by influencing FA and FE. Among the multiple pathways of segmented FA and FE, CLA has the strongest positive effect, accounting for 66.28% of the total mediating effect. Interestingly, HC-INFRA CCD does not promote all FA and FE, and the correlation between different FAs and FEs is not perfectly positive. For example, CLA slightly inhibits HQED via CE and LRE.

6.2. Policy Recommendations

To fully leverage the positive effect of HC-INFRA CCD on HQED demonstrated in this study, we propose the following three reference suggestions for regional policymakers:

(1)

Optimize all aspects of HC and improve regional INFRA construction. To optimize HC, first, increase support for education and technical training to improve the overall worker quality. Second, strengthen subsidies for talent introduction and improve welfare benefits to enhance regional attractiveness to quality HC. For INFRA construction, the government should coordinate the spatial structure of INFRA construction based on regional characteristics such as population distribution and industrial development. The development of functional INFRA should also be strategically coordinated to avoid weakening the role of the economy due to the associated short-term effects.

(2)

Promote the coordinated development of HC and INFRA. Policymakers should rationally allocate financial resources and guide the flow of social funds according to development gaps between HC and INFRA in the region. This can efficiently leverage their synergistic effects to promote regional HQED.

(3)

Continuously deepen the reform of higher education and improve the quality of HEI construction. As HEI is a talent training base, the government should carefully manage and oversee the system of higher education talent training, developing scientific and comprehensive talent training programs for different disciplines and specialties. Policymakers should also fully understand current and future labor market demand and improve programs to develop students’ corresponding professional capabilities. In addition, the government should build an effective platform for cross-border cooperation; strengthen the integration and development of the higher education industry, academia, and research; and promote universities’ participation in fields of regional economic management and enterprise technological innovation. This approach will fully rationalize the contact between university resources and high-quality capital, technology, and other elements; improve the conversion rate of scientific research; and enhance the promotion of HQED.

6.3. Research Limitations

This study provides insights into the impact of HC-INFRA CCD on HQED, providing a novel perspective; however, it is important to acknowledge that the study has some limitations. (1) Due to a lack of data, the selected cities did not include all provinces in China. With improvements to city databases, we will subsequently be able to conduct more comprehensive research based on the city data for all provinces. (2) Due to the unavailability of data, we chose 2011 to 2019 as the time period for the study, and the robustness of the empirical results could be verified by extending the study period (e.g., 2011 to 2023). (3) The promotional effect of HC-INFRA CCD on HQED is also affected by inter-regional HC and other factor flows, with presumed spatial spillover characteristics. In future studies, we will use spatial panel modeling to explore the impact of spatial spillover effects.