Environmental quality and the ebb and flow of urban innovation in China: An Explanation from the Perspective of Human Capital Mobility

doi:10.21203/rs.3.rs-4938910/v1

Download PDF

Article

Environmental quality and the ebb and flow of urban innovation in China: An Explanation from the Perspective of Human Capital Mobility

https://doi.org/10.21203/rs.3.rs-4938910/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

As urban competition intensifies, city quality increasingly becomes a key factor in determining human capital distribution and fostering urban innovation. Does environmental quality, as a crucial aspect of urban quality, lead to the mobilization of human capital and drive urban innovation? A definitive answer to this question remains unclear. This study begins with urban air pollution and constructs a framework of “environmental quality - human capital mobility - urban innovation efficiency.” It verifies this framework by integrating “Baidu Migration” big data with urban panel data. The findings reveal that: (1) Air pollution significantly inhibits urban innovation efficiency. This conclusion holds even after using instrumental variable techniques to address endogeneity concerns and conducting robustness analyses. (2) Mechanism tests show that reductions in both the quantity and quality of human capital, along with the outflow of high-skilled labor, are key mechanisms underlying the innovation-dampening effects of air pollution. (3) Heterogeneity analysis reveals that the negative effects of air pollution on innovation are more pronounced in inland cities with high-speed rail connections. This implies the reinforcement of China's regional innovation pattern, with lower innovation levels in the western regions and higher levels in the eastern regions. Additionally, enhancing economic agglomeration, providing high-quality public services, fostering cultural diversity, and strengthening digital infrastructure can increase residents' attachment to their local areas, thereby mitigating the negative impact of air pollution.

Earth and environmental sciences/Environmental sciences

Earth and environmental sciences/Environmental social sciences

Earth and environmental sciences/Natural hazards

air pollution

Urban innovation efficiency

Human capital flows

Simultaneous equations in space

Urban sustainable development has recently become a prominent topic in both domestic and international research. From a historical perspective, the rise and fall of cities are influenced by various factors such as resource endowment, geographical features, institutional environment, and policy orientation. According to Schumpeter's theory of innovation and endogenous growth, innovation is the primary driving force of development. With the intensification of the great power game and the new round of urban competition, more research has shifted from focusing on urban innovation itself to its underlying causes. Discussions have explored various angles, including economic scale, collaborative innovation, population agglomeration, land resource allocation, and innovation policies (Bai and Jiang, 2015; Cai et al., 2023; Sedgley and Elmslie, 2011; Xie and Hu, 2020). However, compared to these indirect or short-term inducements, high-quality human capital—as an essential carrier of knowledge and skills and a crucial element in knowledge production—serves as a core force in aggregating innovative factors, supporting industrial structure upgrading, and promoting high-quality development (Liu et al., 2018). Its crucial role in innovation activities should receive more attention. Considering China's aging and declining population structure, the shortage of high-quality human capital, and the massive labor mobility caused by household registration system reform, attracting and retaining top talent has become a long-term and critical task for local governments.

Research in urban spatial equilibrium theory and creative class theory emphasizes the importance of urban quality in the employment choices of high-skilled labor (Storper and Scott, 2009). In recent years, “smoggy” days have become frequent in China, with PM2.5 levels in some cities even reaching “off the charts.” In the current context of severe environmental pollution, increased environmental awareness, and promotion of high-quality development, environmental quality, as an essential dimension of urban quality, is not only a core element affecting urban comfort and residents' well-being but also a crucial factor in determining the spatial distribution of human capital, shaping the spatial structure of regional innovation, and driving long-term urban development, potentially leading to changes in the rise and fall of cities. However, existing literature mostly focuses on the short-term impacts of environmental quality on human health, labor productivity, labor supply, population migration, and other micro-level aspects, neglecting its importance for the long-term development of cities. Studies on the effects of environmental quality on urban innovation and its mechanisms are lacking, making it difficult to evaluate the economic effects of environmental quality from a macro or long-term perspective, and underestimating the long-term benefits of environmental governance (Wang et al., 2021).

In the context of the common problem of “quantity, quantity and low quality” in China's innovation activities and the restriction of limited innovation elements, this paper takes urban air pollution as the breakthrough point to explore whether environmental quality affects urban innovation efficiency and to what extent. What role do human capital flows play in this? What factors enhance or hinder the innovative impact of environmental quality? Addressing these issues provides a theoretical and empirical basis for understanding the internal relationship between local environmental governance, human capital, and urban innovation in the context of continuous pollution emissions, ecological damage, and significant cross-regional labor mobility in China.

This paper relates closely to three main categories of literature. Firstly, the impact of human capital on innovation. According to endogenous growth theory, substantial literature argues that human capital is both the engine of technological innovation and a core element of long-term growth (Romer, 1990; Zou and Dai, 2003). However, there is a scarcity of systematic research addressing what kind of human capital better promotes innovation. Secondly, the influence of environmental pollution on labor mobility. Environmental quality, an important indicator of urban quality, has recently garnered academic attention for its impact on labor mobility. Existing literature predominantly examines general labor, using the health effects of environmental pollution as the research entry point. The prevailing view suggests that as economic development and living standards improve, population migration is affected by environmental pollution (Li et al., 2021). Thirdly, the impact of environmental pollution on innovation. Existing literature largely revolves around theories like the “Porter Hypothesis,” examining how environmental regulation-induced technological innovation affects pollution (Chen et al., 2022), with few systematic analyses on how pollution reversely affects innovation. Wu et al. (2021) found that environmental pollution can inhibit the positive impact of human capital quality in company management on the future financial performance of enterprises. Cao et al. (2022) discovered that haze pollution negatively impacts the productivity of low-tech manufacturing enterprises, both indirectly reflecting the negative externality of environmental pollution on innovation.

Although the existing literature provides valuable insights for this study, there is room for improvement in several aspects. First of all, existing studies tend to focus on the short-term and micro impacts of air pollution, ignoring the importance of environmental quality in the long-term development of cities. Secondly, there are few studies on the impact, mechanism and heterogeneity of air pollution on urban innovation efficiency in the existing literature. Third, when examining the economic effects of environmental pollution, the existing literature mainly relies on simple estimation, ignoring the spatial spillover effects of air pollution and potential model-setting bias. Fourth, based on the endogenous economic growth framework and the Porter hypothesis, bidirectional causality may exist between environmental pollution and innovation activities, forming an interactive feedback loop. However, the endogeneity problem caused by this relationship has not been effectively addressed in most of the literature.

Compared to existing literature, this paper may contribute in the following aspects: Firstly, in terms of research perspective, it focuses on the human capital mechanism by which air pollution affects urban innovation efficiency. Using urban air pollution as a starting point, this paper constructs a logical framework of environmental quality → human capital mobility → urban innovation efficiency, reflecting China's new development stage. This framework not only extends the research boundaries of human capital and innovation development but also provides theoretical support for urban talent aggregation and innovative development to promote green and low-carbon transformation.

Secondly, in terms of research methodology, this paper combines theoretical analysis with empirical testing to effectively integrate classical environmental economics theory with characteristics of China's new development stage, such as the free cross-regional mobility of skilled labor due to the relaxation of household registration restrictions. This approach provides useful analytical tools and testing methods for research in environmental quality and economic development. Particularly in analyzing the economic effects of air pollution, previous studies often utilized mediation models. However, considering that air pollution is not localized and can create externalities on neighboring cities, and the possibility of feedback mechanisms between air pollution and innovation activities, this paper constructs spatial panel simultaneous equations. These equations can control for spatial spillover effects and interaction feedback relationships among variables, providing a more comprehensive empirical investigation. Additionally, for causal inference, air flow coefficients and atmospheric inversions are used as instrumental variables for air pollution to further control endogeneity, ensuring the reliability of core findings.

Thirdly, in terms of heterogeneity analysis, existing literature often examines the differentiated economic consequences of environmental pollution based on demographic characteristics such as age, gender, income, education, or traditional factors like geographical location and economic development. In contrast, this paper closely aligns with the characteristics of China's practices, delving into the complexity of how air pollution affects urban innovation based on the differences in economic agglomeration, high-speed rail networks, public service quality, cultural diversity, and digital infrastructure across regions. These more unique perspectives on heterogeneity analysis contribute to expanding theoretical understanding of the long-term impact of environmental quality on urban development.

3.1 Human capital mechanism through which air pollution affects urban innovation efficiency

3.1.1 Innovation effect of air pollution: the perspective of human capital stock and quality

Drawing upon existing literature and theoretical deductions, this article elucidates the pathways through which air pollution affects urban innovation efficiency, viewed through the lens of human capital stock and quality. Firstly, air pollution poses significant health risks to residents, increasing the incidence of respiratory and cardiovascular diseases. This deterioration in health reduces both the quality and quantity of human capital, as well as labor supply, thereby stifling local innovative development. Secondly, from an environmental psychology perspective, air pollution induces negative emotional states such as depression, anxiety, and stress. It also diminishes reaction times, attention spans, and cognitive abilities, weakening individual learning capacities and lowering labor productivity. This, in turn, hinders engagement in innovative activities. Thirdly, air pollution deters investment in human capital, further suppressing technological innovation. Innovation critically depends on a skilled labor force and the human capital embodied within it. Given the non-transferability and specificity of human capital, investments in it bear greater risks compared to material capital investments. Individuals exposed to adverse environmental conditions for extended periods have lower life expectancy, exhibit myopic behavior, and are inclined to reduce investments in education. This exacerbates the constraints that the shortage of high-quality human capital places on innovation development.

3.1.2 Innovation effect of air pollution: the perspective of high-skilled human capital flow

From the perspective of the spatial allocation of innovation elements, the urban innovation effect of air pollution is not solely based on the mechanism of human capital stock and quality but also depends on its influence on the mobility of high-skilled human capital. Drawing upon the Rosen-Roback spatial equilibrium theory, under the premise of labor mobility, differences in housing prices among cities can be explained by differences in wages and livability, tending towards spatial equilibrium. This implies that, within an open urban system, the labor force will balance the wage levels, livability, and living costs of different cities, subsequently migrating to the city that maximizes their utility. According to Tiebout's “voting with their feet” theory, higher urban quality reduces the migration rate of the population with a university education. As a crucial dimension of urban quality, air pollution increases the migration intentions of high-skilled human capital. Whether these intentions translate into actions depends on the marginal benefits and costs of migration. To avoid short-term mild air pollution, individuals may adopt adaptive protective behaviors such as reducing outdoor activities, using anti-smog masks, and purchasing air purifiers. However, severe air pollution can cause significant damage to the respiratory system and even induce malignant diseases like lung cancer, significantly increasing the probability of death. It is reasonable to speculate that if exposed to air pollution over the long term, residents may choose to migrate due to severe health threats. Compared to the general labor force, high-skilled human capital, with a higher demand for health and environmental quality, is more likely to migrate to cities with better air quality to maximize their welfare, leveraging their higher knowledge, skills, and broader range of choices.

Based on endogenous growth theory and empirical evidence, the mobility of high-skilled human capital can propel urban innovation through the following channels:

Serving as a crucial conduit for technical knowledge, particularly tacit knowledge, the interregional mobility of high-skilled labor facilitates proximity among human capitals, enhancing face-to-face contact and interaction, reducing the cost of knowledge exchange, and fostering the diffusion and spillover of knowledge.

As the complexity of research problems and technological bottlenecks in the real world intensifies, the process of individual invention and innovation has progressively shifted to team creation, characterized by diverse backgrounds and a multiplicity of knowledge perspectives. This implies that knowledge production and innovation constitute an open and complex process. The mobility of high-skilled labor across regions not only introduces and applies foreign ideas, insights, and experiences to the local market but also accelerates the cultural and conceptual integration between incoming human capital and local human capital, promoting reciprocal learning and collaborative innovation.

The inflow of high-skilled labor can augment the human capital supply in the receiving area, creating a robust and diverse shared labor pool, strengthening the agglomeration economy effect of the city. This enhances the quantity and quality of matches between enterprises and human capital, providing higher-quality human capital support for corporate innovation activities. It also aids in nurturing local entrepreneurs, further aggregating innovation elements and activities, promoting the cross-regional optimization of resources, activating the potential of various elements, and thereby driving urban innovation.In the context of China, facing a shortage of high-quality human capital, the heightened sensitivity of high-skilled human capital to environmental quality, and the significant relaxation of household registration system restrictions on population mobility, this paper speculates that urban air pollution accelerates the outflow of high-skilled talent, altering the urban human capital structure, reducing its average level of human capital, and exacerbating the constraint of human capital shortage on innovation-driven development. In summary, the following hypothesis is proposed:

Hypothesis 1

Air pollution not only has a negative externality on the stock and quality of human capital, but also induces cross-regional mobility of highly skilled human capital, which ultimately inhibits urban innovation efficiency.

3.2.1 Heterogeneous impact of air pollution under urban characteristics

Existing studies have shown that urban characteristics significantly affect labor mobility (Shang et al., 2023; Whisler et al., 2008). A review of existing literature and the characteristics of Chinese practices indicates that differences in high-speed rail networks, economic agglomeration, public services, cultural diversity, and digital infrastructure in cities may lead to varying innovation effects of air pollution.

First, consider the impact of the high-speed rail (HSR) network. As a major artery of the national economy and a significant livelihood project, the opening of high-speed trains has compressed the space-time distance between cities, reduced the cost of human capital flow, expanded the scope of individual job searches, accelerated the spatial allocation of innovation resources, and improved regional innovation efficiency (Wang et al., 2020). Sun and Zhang (2020) found that the high-speed rail network promotes the flow of highly educated talents to scientific research positions in manufacturing enterprises along the Belt and Road, thereby fostering high-quality innovative development in these industries. Considering that eastern cities usually have a higher level of economic development, which can provide residents with better public services, broader development prospects, and higher returns on human capital, residents in central and western cities are less attached to local jobs and medical conditions.

In the pursuit of easier migration of economic income and development opportunities, the high-speed rail network can promote the relocation of high-quality labor around transportation routes (Bian et al., 2019; Knaap and Oosterhaven, 2011). This paper speculates that with the development of high-speed rail network construction, air pollution may accelerate the migration of human capital from central and western cities to eastern cities, thereby strengthening China's regional innovative development pattern of “low in the west and high in the east.”

Second, we consider the impact of economic agglomeration. Economic agglomeration, where economic activities are relatively concentrated in a specific geographical area, can generate economic centripetality conducive to urban development. This agglomeration attracts human capital inflow through mechanisms such as labor pools, intermediate input sharing, and knowledge spillover, thus promoting urban innovation and mitigating the negative effects of air pollution (Lin and Tan, 2019). Firstly, the scale effect brought by economic agglomeration allows individuals to enjoy inexpensive public infrastructure and achieve higher returns on human capital, which attracts human capital inflow. Secondly, the agglomeration of diverse economic activities helps the labor force with different professional backgrounds achieve better matches with enterprises, providing more job opportunities and greater income growth potential. This also promotes cross-industry knowledge spillover, cross-innovation, and entrepreneurial success (Zhang, 2018), forming a pull effect on human capital inflow.

Third, consider the impact of public services. Based on Tiebout's (1956) “voting with their feet” theory, residents will choose suitable areas to live in according to their preference for the combination of public goods and taxes. This means that better public services such as education, housing, pensions, and medical care will attract more highly skilled talents while restraining the outflow of human capital. This enhances the ability of enterprises to absorb and develop new knowledge, thereby promoting their innovation efficiency. Additionally, the inflow of highly skilled talents, attracted by high-quality public services, can promote knowledge diffusion through sharing and learning mechanisms, further fostering urban innovation and development (Zhang, 2019).

Fourth, consider the impact of cultural diversity. Studies have shown that as an informal institution, regional cultural characteristics profoundly affect people's belief preferences and decision-making, significantly impacting entrepreneurship and economic development (Zhang, 2020). According to cultural economics, the integration and collision of diverse cultures are more likely to break conventions and promote innovation. Cultural diversity includes the diversity and inclusiveness of living habits, temperaments, thoughts, and cognition, which can promote complementary effects between labor forces with different technologies and abilities, leading to higher incomes (Liu et al., 2015). It also encourages frequent exchanges between people from different regions and cultural backgrounds, increasing mutual trust and cooperation. This contributes to knowledge spillover and diffusion, generating new ideas (Elia et al., 2019). Thus, it promotes innovation and growth and further attracts labor inflows.

Fifth, consider the impact of digital infrastructure. As the hardware foundation of the digital economy, the impact of digital infrastructure on social equity and common prosperity has garnered increasing attention from scholars. From the perspective of labor supply, digital infrastructure, with its information richness and low copying cost of knowledge products, helps increase communication opportunities between labor with disadvantaged preendowed factors and labor with high levels of social capital. This improves the social capital of the former and brings higher income. It also reduces information asymmetry between workers and jobs, enabling better job matches and providing convenient learning opportunities. This alleviates the constraints of insufficient educational resources, improves the return on human capital investment, and ultimately realizes digital empowerment of the labor supply side (Fang et al., 2023). From the perspective of labor demand, digital infrastructure features such as network interconnection and information transparency reduce market information asymmetry, expand the scope for enterprises to obtain technology spillovers, and increase labor demand by expanding the innovation boundary of enterprises and improving market vitality (Shen et al., 2023). It also enhances social supervision, protects the interests of vulnerable groups, and promotes fairness and the rule of law. This provides a good guarantee for vulnerable groups to achieve upward mobility, ultimately realizing digital empowerment for the labor demand side.

There is no doubt that economic factors and the living environment are core factors of labor mobility. Based on the above logic, combined with the push and pull theory of population migration and endogenous growth theory, economic agglomeration, high-quality public services, cultural diversity, and the strengthening of digital infrastructure are important factors that improve the economic situation, development opportunities, or living environment of workers. These factors not only increase the attachment of workers to their location, making them less susceptible to the impact of air pollution and less likely to migrate, but also attract more highly skilled talents and promote urban innovation, thus mitigating the negative innovation effect of air pollution. The following hypothesis is proposed:

Hypothesis 2

The negative impact of air pollution on urban innovation efficiency through the channel of human capital flow is mainly reflected in inland cities with high-speed trains. However, this impact will be mitigated by improvements in economic agglomeration, high-quality public services, cultural diversity, and the strengthening of digital infrastructure.

4.1 Design of model

4.1.1 Baseline regression configuration

This paper initially constructs a streamlined model to investigate the comprehensive effects of air pollution on urban innovation efficiency. Existing literature indicates that due to factors such as knowledge spillovers and the mobility of air, both air pollution and innovative activities demonstrate significant spatial correlations (Bai and Jiang, 2015; Shao et al., 2019). Ignoring these inherent spatial spillover effects could lead to biased estimations. Additionally, macroeconomic variables in reality often exhibit path-dependent characteristics. Consequently, following the methodology of Zhu and Lee (2021), we establish a dynamic spatial Durbin model that is capable of capturing both temporal effects and spatial dynamics:

Uieit = α + τUieit − 1 + pWUieit + βPolit + γWPolit + θX1it + φWX1it + ϑi + ιt + εit (1)

In the model, i and t represent the city and year, respectively. W denotes the spatial weights matrix.Uie_it, Uie_{it −1}, and WUie_it correspond to the urban innovation efficiency, its first-order lag, and spatial lag, respectively. Pol_it and WPol_it pertain to air pollution and its spatial lag. X comprises control variables, which, according to existing literature, include the level of economic development, research and development intensity, infrastructure, financial development, and openness to foreign trade. ϑ_i and ι_t are the city-specific and time-specific fixed effects, respectively. This structure facilitates a nuanced exploration of the inter dependencies and dynamic interactions among these variables within the spatial econometric framework.

4.1.2 Spatial simultaneous equations model configuration

Building on the existing literature and the analysis presented, there is a recognized interactive influence among air pollution, human capital mobility, and urban innovation efficiency. Employing a simplified model would not only make it challenging to discern the interrelationships among these variables but would also struggle to control for the endogeneity resulting from bidirectional causality. Taking into account the significant spatial spillover effects evident in urban innovation efficiency and air pollution, and drawing inspiration from Yang and Lee (2017), we construct a spatial simultaneous equations model that comprehensively controls for endogeneity and spatial effects. This model aims to elucidate the mechanisms by which air pollution impacts urban innovation efficiency from the perspective of human capital.

Uieit = α0 + α1WUieit + α2Polit + α3WPolit + α4Hcsit + α5 Hcqit

+α6 Hcmit + α X2it + κi + ηt + ε1it (2)

Hcsit = β ₀ + β₁Polit + β₂WPolit + βX3it + σi + ρt + ε2it (3)

Hcqit = θ ₀ + θ₁Polit + θ2WPolit + θX4it + vi + πt + ε3it (4)

Hcmit = λ0 + λ ₁ Polit + λ2WPolit + λX5it + Ψi + ζt + ε4it (5)

Polit = γ0 + γ1WPolit + γ2Uipit + γX6it + ϖi + υt + ε5it (6)

The simultaneous equations model comprises five fundamental equations, among which Eq. (2) serves as the urban innovation efficiency driver equation, employed to examine the direct effects of air pollution and its spatial lag on urban innovation efficiency. To test Hypothesis 1, which posits that human capital factors significantly influence urban innovation efficiency, variables for human capital stock (Hcs), human capital quality (Hcq), and human capital mobility (Hcm) are incorporated into the equation. Given the potential spatial spillover effects of urban innovation efficiency, its spatial lag is also included in the model. X2it represents the control variables, which include research and development intensity, infrastructure, financial development, and openness to foreign trade. κi and ηt respectively denote the city-specific and time-specific fixed effects. This model configuration allows for a comprehensive analysis of the variables influencing urban innovation, while accounting for spatial dependencies and fixed effects.

Equation (3) is the human capital stock equation, designed to test whether air pollution and its spatial lag can impact urban innovation efficiency through human capital stock. The control variables X_3it include geographical location, initial human capital level, educational expenditure, and income disparity.σ_i and ρ_t represent the city-specific and time-specific fixed effects, respectively.

Equation (4) is the human capital quality equation, which assesses whether air pollution and its spatial lag can affect urban innovation efficiency through human capital quality. The control variables, X4it, encompass the level of economic development, industrial structure, average annual urban temperature, and policies for attracting high-level talent. vi and πt serve as the city-specific and time-specific fixed effects, respectively.

Equation (5) is the human capital mobility equation, aimed at examining whether air pollution and its spatial lag can influence urban innovation efficiency by triggering human capital mobility. The control variables X_5it include the level of economic development, average wages, public services, housing prices, and unemployment rate. Ψ_i andζ_t are the city-specific and time-specific fixed effects, respectively.

These equations are integral components of a comprehensive model that connects environmental factors with the dynamics of human capital and urban innovation, embedding spatial and temporal dimensions into the analysis.

Equation (6) is the air pollution equation, designed to explore the feedback effects of urban innovation efficiency on air pollution and to address estimation biases caused by reverse causality among core variables. The control variables, X6it, are based on Shao et al. (2019) and include economic development level, industrial structure, energy consumption, openness to foreign trade, and environmental regulation.ϖ_i and υ_t represent the city-specific and time-specific fixed effects, respectively.

4.2 Variable Definition

4.2.1 Dependent variable: Urban innovation efficiency

In this study, R&D personnel and R&D capital stock are used as the input indicators for innovation activities, while per capital patent applications serve as the output indicator of urban innovation activities. Regarding the R&D capital stock, following the approach of Wu (2006), the perpetual inventory method is employed for calculation. For the choice of measurement method, we utilize Data Envelopment Analysis (DEA), which does not require a priori specification of the production function form and can handle multiple inputs and outputs. This method calculates urban innovation efficiency (Uie_ 1) as detailed in Li and Yang (2018).

To enhance the robustness of the conclusions, this study also employs the Urban Innovation Efficiency Index (Uie_2) from the “China Urban and Industrial Innovation Report” to measure the innovation efficiency across different cities. Following the method of Kou et al. (2017), this paper updates the Urban Innovation Index for the years 2017 to 2022.

The measurement of spatial lag for urban innovation efficiency necessitates the appropriate configuration of the spatial weights matrix. This paper considers both geographical and economic distances by constructing a nested weights matrix that combines these two dimensions. This matrix is used to calculate the weighted average of neighboring areas' innovation efficiencies. The calculation method is as follows:

WRipit = ∑(ω_ij x Ripjt) (7)

In this context, ω_ij represents the elements of the nested weights matrix combining geographical and economic distances, defined as ω_ij = 1/2 (d_ij¹ + d_ij²). Here, d_ij¹ is the reciprocal of the shortest highway mileage between cities i and j, representing geographical distance, while d_ij² is the reciprocal of the absolute difference in the average annual per capita real GDP between cities iand j, representing economic distance. This formulation reflects a balanced integration of both geographical proximity and economic similarity in assessing spatial interactions between cities.

4.2.2 Core explanatory variable: Air pollution

This variable is measured using PM2.5, which is widely recognized as a primary culprit of haze pollution. Drawing from Shao et al. (2019), this study utilizes the global PM2.5 grid data monitored via satellite and provided by the Socioeconomic Data and Applications Center at Columbia University. Furthermore, using ArcGIS technology, this data is disaggregated to yield PM2.5 concentration data for 285 prefectural-level cities and above in China. This approach ensures precise and geographically specific measurement of air pollution levels across the urban landscape.This study employs the aforementioned spatial weights matrix to compute the weighted average of neighboring haze pollution, thereby characterizing the spatial lag of haze pollution. This methodological approach facilitates a nuanced understanding of the spatial dynamics of environmental degradation, enhancing the analysis of regional pollution influences.

4.2.3 Mediating variable

(1) Human capital stock. In view of the lack of health data at the household level, referring to Wang et al. (2008), we assume that the key to health level lies in health investment, and use the number of hospital beds per 10,000 people to measure health investment, so as to measure the urban health level.

(2) Quality of human capital. Referring to Schultz's definition of the connotation of human capital quality, from the three dimensions of knowledge and skills, health quality and creative ability,

The factor analysis method is used for measurement, and the proportion of people with higher education in the total population is used to measure knowledge and skills, the number of health technicians per 1,000 people is used to measure health quality, and the number of patents granted per 10,000 people is used to measure creativity.

(3) Human capital flow. Referring to Wang et al. (2020), this paper uses the “Baidu Migration” big data based on the location service geolocation system.

To measure the scale of human capital mobility, this study selects data from the “Baidu Migration” network on the migration of individuals with college education or higher between cities around the time before and after the Chinese New Year. The rationale for this choice is that the motives for inter-city population movements vary significantly at different times. Daily inter-city movements typically revolve around official business, commuting, tourism, and family visits, occurring sporadically under normal circumstances without external shocks. In contrast, migrations for changing employment or residence show distinct temporal patterns, often taking place around the Chinese New Year. Individuals typically return to their hometowns before the New Year and then return to their cities of residence to continue working after the holiday. This universally observed behavioral pattern offers a viable basis for measuring population migration. Specifically, if an individual moves from city i to city j before the New Year, it is likely for the celebration, making i the city of immigration and j the city of emigration. Conversely, if the move occurs from city j to city i after the New Year, it indicates a return to the city of residence for work and life. Consequently, the observation points of individuals moving from city i to city j before the Chinese New Year, and from city j to city i after the Chinese New Year, may both indicate emigration from city j to city i. To reduce statistical errors, following the methodology of Xu and Yao (2018), the average of the migration figures from these two observation points is used to measure the scale of migration from city j to city i. Summing the total number of all incoming migrants to a specific city provides the total volume of population migration into that city. The calculation method is as follows:

Migrantj − i = 1/2 (Pop_flowi − j |_before +Pop_ flowj − i |_after) (8)

Mgri = ∑Migrantj − i (9)

In this context, “Pop_flow” refers to the volume of high-quality human capital migrating from city j to city i around the Chinese New Year. “Mgr” represents the total influx of high-quality human capital into city i, reflecting its attractiveness to talent from other regions.

4.2.4 Control variables

Economic development level is represented by per capita real GDP; research and development intensity is measured by the proportion of science and technology expenditure to GDP; industrial structure is gauged by the share of secondary industry's added value in GDP; infrastructure is quantified by per capita urban road area; financial development is expressed as the proportion of value added by the financial sector to GDP; openness to foreign trade is assessed by the ratio of total imports and exports to GDP; environmental regulation is measured using the comprehensive utilization rate of industrial solid waste; high-level talent attraction policies are represented by a dummy variable indicating whether the city's province implemented such policies in the given year; average wages are measured by the average salary of employed workers; geographical location is measured by the road distance to the nearest port; initial human capital level is quantified by the stock of human capital in each city in 2014; housing prices are measured by the average sale price of commercial housing in each city; public services are based on the number of teachers per thousand people in primary and secondary schools, the number of doctors per thousand people, and the number of books per hundred people in public libraries, constructed into an index of public service supply capacity using principal component analysis; unemployment rate is measured by the registered urban unemployment rate.

4.3 Data Sources

Given the availability and consistency of statistical standards, this paper utilizes panel data from 285 prefecture-level cities in China for the years 2014–2022. Apart from haze pollution and human capital mobility, which are derived from “Baidu Migration” big data, the sources of data include the “China Urban Statistical Yearbook,” the “China Science and Technology Statistical Yearbook,” the CEIC China Economic Database, and the “China Urban and Industrial Innovation Capability Report.” To eliminate the effects of price volatility, all variables related to prices are deflated using 2014 as the base year. Missing values are imputed using linear interpolation. To mitigate the effects of heteroscedasticity, logarithmic transformations are applied to all variables. To reduce the impact of outliers on estimation results, tail-trimming is performed on all continuous variables at the 1% and 99% quantiles. This comprehensive approach ensures the reliability and accuracy of the data used in the analyses.

5.1 Benchmark result: the comprehensive impact of air pollution on urban innovation

This paper first regresses Eq. (1), and the dynamic spatial panel Durbin model contains the spatial lag term and time lag term of the explained variable, which does not conform to classical assumptions. The Han-Phillips Generalized method of moments (GMM) proposed by Han and Phillips (2010) can not only effectively solve the problem of weak instrumental variables in difference GMM estimation, but also avoid the problem of estimation inconsistency caused by the time lag coefficient approaching 1 in system GMM estimation. In addition, this method requires less on the number of sample sections N and time T, and can still produce unbiased and consistent estimates in the case of small samples. Therefore, this method is used for estimation in this paper. In order to test the robustness of the regression results, the estimation results of non-spatial OLS and non-spatial dynamic panel model system GMM (SYS-GMM) are also reported.

Table 1 reports the estimation results of Eq. (1) based on the nested weight matrix of geographical distance and economic distance. It can be seen that in Models 1 and 4 without considering endogeneity, spatial spillover effect and dynamic effect, the estimated coefficients of air pollution have opposite signs and are not significant. From the perspective of measurement, these unrobust results indicate that not considering endogeneity, spatial spillover effect and dynamic effect will lead to estimation bias. The dynamic spatial Durbin model (Models 3 and 6) controls endogeneity, spatial spillover effect and dynamic effect at the same time, and the estimated results have good statistical characteristics and robustness, and are in line with theoretical expectations. Therefore, this paper focuses on the analysis of the regression results of the dynamic spatial Durbin model based on the Han-Phillips GMM.

Table 1

Benchmark regression results: urban innovation effect of air pollution
	OLS	Uie_ 1 SYS-GMM	Han-Phillips GMM	OLS	Uie_2 SYS-GMM	Han-Phillips GMM
	Model1	Model2	Model3	Model4	Model 5	Model 6
L.Uie		0.325*** (3.31)	0.307*** (3.21)		0.284*** (3.08)	0.261*** (2.74)
WUie			0.225** (2.07)			0.249*** (2.68)
Pol	-0.179 (− 1.56)	-0.157 (− 1.31)	-0.208*** (-2.62)	0.207 (1.34)	-0.133 (− 1.46)	-0.184*** (-2.56)
WPol			-0.352*** (-4.31)			-0.317*** (-4.65)
Variable of control	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES
Time FE	YES	YES	YES	YES	YES	YES
AR(1)-P a			0.003			0.002
AR(2)-P b			0.183			0.224
Hansen-P c			0.347			0.328

It can be seen from the results of Model 3 and Model 6 that the coefficients of the time-lag term of urban innovation efficiency are significantly positive, indicating that the urban innovation efficiency has obvious path-dependent characteristics. The coefficient of spatial lag term of urban innovation efficiency is significantly positive at least at the level of 5%, which not only means that the economic competition and mutual imitation between regions make the innovation activities of adjacent cities have a demonstration and driving effect on the local area, but also reflects that with regional economic integration, the economic connection between adjacent cities is increasingly close, and it is easier to form industrial clusters with high correlation between upstream and downstream. It promotes cross-regional technology spillover and knowledge sharing, leading to obvious spatial spillover effect of urban innovation efficiency.

The estimated coefficients of local air pollution are all significantly negative at the level of 1%, indicating that the increase of air pollution will inhibit urban innovation efficiency, which initially verifies Hypothesis 1, reflecting that air pollution may inhibit local innovation efficiency by weakening human capital accumulation and triggering human capital migration. The coefficients of the spatial lag term of air pollution are all significantly negative at the level of 1%, indicating that the diffusion of local air pollution will weaken the innovation efficiency of neighboring areas. The reason is that local air pollution has obvious spatial spillover effect under the action of atmospheric circulation, integration of regional economy, transfer of polluting industries and other mechanisms, which will inhibit the innovation efficiency of neighboring areas by reducing labor supply, crowding out the R&D investment of enterprises, reducing the productivity of enterprises and other mechanisms. Therefore, in order to achieve a win-win situation between pollution prevention and innovation and development, it is necessary to have a global perspective and general equilibrium thinking, and build and improve the regional pollution joint prevention, control and collaborative treatment mechanism.

5.2 Robustness test

To further ensure the reliability of the benchmark conclusions, the following robustness tests are conducted:

5.2.1 Further control the endogenous bias

From the logic of interaction between air pollution and innovation efficiency, the seemingly exogenously given effect of air pollution may have endogeneity caused by bidirectional causality. On the other hand, higher innovation efficiency means that cities can better achieve the coordinated development of economic growth, resource conservation and environmental protection, which helps to curb air pollution. Although the simultaneous equation model has been used to control this endogeneity problem above, in order to obtain more reliable results, we refer to Hering and Poncet (2014) and use the air flow system

Number is used as the first instrumental variable for air pollution. The reason is that, on the one hand, the larger the coefficient is, the stronger the air mobility is, which means that it is negatively correlated with the local air pollution level, which is in line with the correlation conditions of instrumental variables. On the other hand, as an objective natural phenomenon determined by meteorological system and geographical conditions, air flow coefficient is affected by wind speed and atmospheric boundary layer height, which does not directly affect urban innovation efficiency except through air pollution, which meets the exogeneity condition of effective instrumental variable. Referring to Hering and Poncet (2014), the calculation method of air circulation coefficient is as follows:

VC = WS _{it ×} BLH_it (10)

Where, VC, WS and BLH are air flow coefficient, wind speed and atmospheric boundary layer height, respectively. According to the global 0.75×0.75 issued by the European Centre for Medium-Range Weather Forecasts (ECMWF). For the height wind speed and boundary layer height data of the grid, this paper calculates the VC of each grid in each year by ArcGIS software, and then matches each grid with the cities during the sample investigation period based on longitude and latitude to obtain the VC data at the prefecture-level city level over the years.

Referring to Wu et al. (2021), atmospheric inversion is used as the second instrumental variable of air pollution. The reason is that on the one hand, temperature inversion is a naturally formed atmospheric phenomenon. On the other hand, temperature inversion is independent of economic activities, and if temperature inversion affects urban innovation efficiency, it will only meet the exclusion restriction of instrumental variables through the only transmission channel of affecting air pollution level. Inversion data were obtained from the National Aeronautics and Space Administration (NASA) database. This paper uses the above two instrumental variables to identify the impact of air pollution on urban innovation efficiency through the two-stage least squares (2SLS) method.

5.2.2 Replace the spatial weight matrix

The geographical distance weight matrix and 0–1 adjacency spatial weight matrix are used to replace the nested weight matrix of geographical distance and economic distance for estimation. In addition, the estimation method mentioned above is still used to test Eq. (1).

5.2.3 Changing the measurement method of core explanatory variables

Since 2013, the Ministry of Ecology and Environment has begun to disclose the air quality index (AQI) including PM10, SO2, NO2 and so on. The data are from the official website of the Ministry of Ecology and Environment.

5.2.4 Changing the measurement method of core mediating variables

Referring to Bai et al. (2017), this paper uses the modified gravity model to measure the flow of human capital between cities, so as to replace the above indicators of human capital migration. The measurement method is as follows:

$$Hc{m_{ijt}}=G \times \frac{{\sqrt {Hc{s_n} \times Hc{s_H} \times ({w_H}/{w_H}) \times ({p_H}/{p_H})} }}{{{d_y}}}$$

Where i (j) and t are city and year respectively; Hcmijt is the amount of human capital flowing from city j to city i; G is the gravity coefficient, generally valued at 1; Hcs, w and p are the stock of human capital, average wage and number of scientific research projects respectively; The measurement methods and data sources of human capital stock and average wage are consistent with those above. The connotation of Eq. (11) is that the flow scale of human capital between cities is directly proportional to the stock of human capital between the two places, and inversely proportional to the distance between the two places.

5.2.5 Consider other potential influencing factors

In order to further alleviate the estimation errors caused by the omission of important variables, the national innovative city pilot policy may affect both urban air quality and innovation efficiency, leading to spurious causality. This paper adds the dummy variable of the pilot policy of innovative cities to Eq. (1), so as to control the impact of innovation policy on the estimation results.

From the results of the robustness test, except for the changes in the signs and significance of some control variables, the signs of the coefficients of the core explanatory variables are all in line with theoretical expectations and show good statistical significance, indicating that the previous conclusions are robust.

6.1 Test of action mechanism

Based on the above analysis, from the perspective of human capital flow, the spatial linkage model (Equations (2)–(6)) is used for mechanism analysis. This paper uses GS3SLS to estimate the spatial simultaneous equation model, and the results are shown in Table 2.

Table 2

Test on the human capital mechanism of air pollution affecting urban innovation efficiency
		Uie 1				Uie 2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Uie	Hcs	Hcq	Hcm	Uie	Hcs	Hcq	Hcm
WUie	0.264** (2.16)				0.236** (2.14)
Pol	-0.103* (− 1.78)	-0.067** (-2. 11)	-0.190** (-2.14)	-0.073** (-2.12)	-0.110* (− 1.77)	-0.059* (− 1.80)	-0.205*** (-2.69)	-0.077** (-2.10)
WPol	-0.227** (-2.13)	-0.029* (− 1.73)	-0.145*** (-2.68)	-0.088** (-2.09)	-0.185** (-2. 12)	-0.025* (− 1.66)	-0.170*** (-2.74)	-0.084** (-2.08)
Hcs	0.118* (1.75)				0.127** (2.10)
Hcq	0.151*** (2.84)				0.216*** (2.73)
Hcm	0.089** (2.13)				0.069* (1.82)
Variable of control	YES	YES	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES	YES	YES
Time FE	YES	YES	YES	YES	YES	YES	YES	YES
Adjust R square	0.703	0.478	0.559	0.609	0.638	0.474	0.587	0.627
Note: (1) and (5) correspond to the regression results of Eqs. (2), (2) and (6) correspond to the regression results of Eqs. (3), (3) and (7) correspond to the regression results of Eq. (4), (8) corresponds to the regression results of Eq. (5).

From columns (1) and (5) of Table 2, it can be seen that the coefficient of Pol is significantly negative at the 10% level, and its absolute value decreases compared with Models 3 and 6 of Table 1. According to the identification idea of mediating effect, this means that human capital plays a partial mediating role between air pollution and urban innovation efficiency. The coefficients of Hcs, Hcq and Hcm are all significantly positive, indicating that the increase of human capital stock, the improvement of human capital quality and the immigration of human capital can all improve urban innovation efficiency, which is consistent with the conclusions of most empirical studies. Therefore, human capital stock, human capital quality and human capital flow can all be the transmission channels through which air pollution affects urban innovation efficiency.

From columns (2)–(4) and (6)–(8) of Table 2, it can be seen that the influence coefficients of air pollution on human capital stock, human capital quality and human capital flow are all significantly negative, which means that air pollution inhibits the increase of human capital stock, the improvement of human capital quality and the inflow of human capital, which is in line with theoretical expectations. The reasons are as follows: first, air pollution will not only damage the health of individuals, bring negative effects on their emotions, learning ability and labor efficiency, but also increase the risk of human capital investment, which is not conducive to human capital investment, and further weaken the stock and quality of urban human capital. Second, residents who are exposed to heavy smog for a long time choose to migrate due to the threat to their survival. In particular, compared with the general labor force, high-quality human capital is more sensitive to air pollution. In order to further discuss the heterogeneous impact of air pollution on the migration of different human capital groups and strengthen the verification of Hypothesis 1, this paper also selects the migration network data of groups with a junior college degree or below between cities at two time points around the Spring Festival of “Baidu Migration” over the years, and constructs the migration quantity index of non-high-quality human capital in the city, so as to replace Hcm for regression 6. The results show that air pollution has no significant impact on the immigration of non-high-quality human capital, indicating that air pollution mainly leads to the loss of high-level human capital compared with the general labor force.Combined with the regression results in columns (1) and (5) of Table 2, it can be seen that air pollution not only has negative externalities on the stock and quality of human capital, but also induces it.The flow of human capital ultimately inhibits urban innovation efficiency, and Hypothesis 1 is fully verified.

It can also be seen from Table 2 that the estimated results of the impact of the spatial lag term of air pollution on human capital stock, human capital quality and human capital flow are significantly negative, that is, the exacerbation of haze pollution in the neighboring area has a negative effect on the local human capital stock, human capital quality and human capital inflow. This result is similar to the findings of Wang and Miao (2019). On the one hand, neighboring pollution diffusion will aggravate local pollution and cause loss of human capital, thus inhibiting local innovation efficiency; On the other hand, under the joint action of atmospheric circulation, nearby transfer of polluting industries and regional economic integration, China's regional air pollution has the characteristics of high emission club agglomeration (Shao et al., 2019). High-quality human capital exposed to the haze environment for a long time will “vote with its feet” and migrate to areas with more advantageous environmental quality rather than adjacent cities with similar pollution levels on a large scale in space, so it is difficult to offset the inhibitory effect of pollution diffusion in neighboring areas on local innovation efficiency.

6.2 Quantitative decomposition of mechanism

Referring to the method of Gelbach (2016), this paper quantitatively decompose the above mechanism within the framework of spatial linkage model, and the decomposition formula is as follows (taking the human capital flow mechanism of air pollution affecting local innovation efficiency as an example):

Interpretation Effect of Hcm = α₆λ₁ /(α₂ + α₄β₁ + α₅θ₁ + α₆λ₁) (12)

Where α₄β₁, α₅θ₁ and α₆λ₁ are the explanatory effects of the three mechanisms of human capital stock, human capital quality and human capital flow on the innovation effect of air pollution, and the remaining unexplained part is α₂. Therefore, Eq. (12) measures the proportion of human capital flow mechanism in explaining the innovation effect of air pollution.

Figure 1 reports the quantitative decomposition results of the mechanism, and it can be seen that in the total effect of air pollution on urban innovation efficiency, the explanatory proportions of human capital flow, human capital stock and human capital quality are 4.36%, 5.23% and 19.91%, respectively. Therefore, the mediating effect of the three human capital factors explains 29.51% of the total effect, and other mediating mechanisms explain 70.49% of the total effect. This means that although other mechanisms found in the existing literature (such as crowding out firms' R&D funds, reducing firms' productivity and increasing firms' labor costs) largely explain the innovation effect of environmental pollution, the human capital mechanism still has strong credibility and explanatory power. In particular, with the increasingly fierce competition for talents between cities, the increasing demand of high-quality human capital for environmental quality, and the increasing role of human capital quality in promoting innovation and development, the intermediary role of human capital flow between environmental pollution and urban innovation should not be underestimated.

This paper conducts heterogeneity analysis to better understand the differences in the impact of air pollution on urban innovation efficiency in different scenarios. In order to verify Hypothesis 2, Eq. (5) in the spatial simultaneous equation is further set as:

Hcm _it = λ₀ + λ₁Pol_it + λ₂Pol_it ×Moderator_it + λ₃WPol_it + λX_5it + Ψi + ζt + ε_4it (13)

Where Moderator_it is the variable that plays a moderating role between air pollution and human capital flow in city i in year t. According to Hypothesis 2, The interaction term between high-speed rail network and geographical location (Hn), economic agglomeration (Ea), public service quality (Pq), cultural diversity (Cd) and digital infrastructure (Di) are selected as moderating variables. The meanings of other variables and symbols are similar to those of Eq. (5).

It can be seen from Table 3 that, except for the interaction term between high-speed rail network and geographical location as the moderating variable, the coefficient of the interaction term between air pollution and other moderating variables is significantly positive, which confirms the conjecture in this paper that the negative effect of air pollution on urban innovation efficiency through the channel of human capital flow is particularly significant in inland cities with high-speed trains. However, this effect will be mitigated to a certain extent by the improvement of economic agglomeration, quality public services, cultural diversity, and the strengthening of digital infrastructure. The reasons are as follows: first, the superior geographical location of coastal cities means that individuals can obtain more professional matching and more satisfactory job opportunities, broader development prospects and higher return on human capital from the local area. Therefore, with the development of high-speed rail network construction, air pollution may promote the migration of human capital from inland cities to coastal cities, thus strengthening the Matthew effect of the innovative division of labor pattern among cities.

Table 3

Heterogeneity analysis of the innovation effect of air pollution: from the perspective of human capital flow
Moderating variable	(1) Hn	(2) Ea	(3) Pq	(4) Cd	(5) Di
Pol	-0.054* (− 1.82)	-0.069** (-2.13)	-0.061* (− 1.79)	-0.072** (-2.13)	-0.082*** (-2.57)
Pol × Moderator	-0.027* (− 1.76)	0.017* (1.85)	0.030** (2.16)	0.034* (1.89)	0.024** (2.12)
WPol	-0.074*** (-2.60)	-0.077*** (-2.58)	-0.069** (-2.13)	-0.065* (− 1.77)	-0.078* (− 1.86)
Variable of control	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES
Time FE	YES	YES	YES	YES	YES
Adjust R square	0.613	0.652	0.634	0.635	0.626

Secondly, the scale effect brought by economic agglomeration enables individuals to share local facilities at a lower cost and obtain higher living comfort and return on human capital. As the key for the labor force to obtain social welfare, high-quality public service is an important factor affecting the livability and happiness of the city. Cultural diversity not only means the diversity and inclusiveness of thought and technology, which is easy to promote the complementary effect between labor forces and obtain higher income, but also promotes the communication between people from different regions and cultural backgrounds, and increases the trust and knowledge spillover among residents. Digital infrastructure relies on information enrichment, low cost of knowledge product replication, network interconnection and other characteristics to help realize the digital empowerment of both sides of labor supply and demand. Therefore, the improvement of economic agglomeration, high-quality public services, cultural diversity, and the strengthening of digital infrastructure will not only increase the viscosity of labor to the location, but also attract more highly skilled talents and promote urban innovation, thus slowing down the negative innovation effect of air pollution.

This paper uses urban air pollution as an entry point to construct a logical framework of “Environmental Quality → Human Capital Mobility → Urban Innovation Efficiency.” It tests this framework by matching “Baidu Migration” big data with urban panel data, revealing that: (1) Air pollution significantly suppresses urban innovation efficiency. This conclusion holds even after using instrumental variables to address endogeneity and conducting other robustness analyses. (2) Mechanism tests show that declines in human capital stock and quality, and the outflow of high-skilled human capital, are important mechanisms through which air pollution affects innovation. (3) Heterogeneity analysis finds that the negative innovative effects of air pollution are more pronounced in inland cities with high-speed rail connections, suggesting that air pollution exacerbates the “low west, high east” regional innovation pattern in China. Economic agglomeration, high-quality public services, cultural diversity, and strengthened digital infrastructure can increase residents' attachment to their locality, mitigating the negative impacts of air pollution.

The policy implications of this study include: First, by embracing the concept of green development, the transition from green dividends to talent and innovation dividends can be realized. Second, by accelerating the construction of a talent-strong nation, increasing the stock and quality of human capital, and breaking down barriers that hinder the mobility of human capital, the key role of human capital in urban innovative development can be further unleashed. Third, by adapting to the laws of talent mobility and aggregation, and targeting urban characteristics with top-level design and comprehensive planning, precise supporting policies can systematically address air pollution and its economic consequences.

Author Contribution

Z is responsible for the concept definition of the paper and the provision of financial resources; ;L Responsible for information management and investigation, project management and supervisionQ was responsible for writing the first draft of the paper, formal analysis and visualization of the software, data management and processing;All of them contributed to the revision of the paper

Data Availability

he data supporting the findings of this study are available from the "Baidu Migration" database and the CEIC China Economy Database, but the availability of these data is limited and these data are used under the permission of this study and therefore are not publicly available. However, the data are available to the authors upon reasonable request and by permission of the database. For data from this study, please contact the corresponding author of this paper.

Bai, J. H. & Jiang, F. X. Collaborative innovation, spatial linkages, and regional innovation performance. Econ. Res. J. 7, 174–187 (2015).
Bai, J. H., Wang, Y., Jiang, F. X. & Li, J. R&D factor mobility, spatial knowledge spillover, and economic growth. Econ. Res. J. 7, 109–123 (2017).
Bian, Y. C., Wu, L. H. & Bai, J. H. Has the opening of high-speed rail promoted regional innovation? J. Financ Res. 6, 132–149 (2019).
Cai, Q. F., Wang, S. J., Liu, H. & Shu, S. W. Does population agglomeration in urban agglomerations promote corporate innovation? China Ind. Econ. 3, 152–170. https://doi.org/10.19581/j.cnki.ciejournal.2023.03.006 (2023).
Cao, Y., Wang, Q. & Zhou, D. Does air pollution inhibit manufacturing productivity in Yangtze River Delta, China? Moderating effects of temperature. J. Environ. Manage. 306, 114492. https://doi.org/10.1016/J.JENVMAN.2022.114492 (2022).
Chen, Y., Yao, Z. & Zhong, K. Do environmental regulations of carbon emissions and air pollution foster green technology innovation: Evidence from China's prefecture-level cities. J. Clean. Prod. 350, 131537. https://doi.org/10.1016/J.JCLEPRO.2022.131537 (2022).
Elia, S., Messeni Petruzzelli, A. & Piscitello, L. The impact of cultural diversity on innovation performance of MNC subsidiaries in strategic alliances. J. Bus. Res. 98, 204–213. https://doi.org/10.1016/J.JBUSRES.2019.01.062 (2019).
Fang, F. Q., Tian, G. & Zhang, X. Digital infrastructure and intergenerational upward income mobility: A quasi-natural experiment based on the broadband China strategy. Econ. Res. J. 5, 79–97 (2023).
Han, C. & Phillips, P. C. B. GMM estimation for dynamic panels with fixed effects and strong instruments at unity. Econom Theory. 26 (1), 119–151. https://doi.org/10.1017/S026646660909063X (2010).
Hering, L. & Poncet, S. Environmental policy and exports: Evidence from Chinese cities. J. Environ. Econ. Manage. 68 (2), 296–318. https://doi.org/10.1016/J.JEEM.2014.06.005 (2014).
Knaap, T. & Oosterhaven, J. Measuring the welfare effects of infrastructure: A simple spatial equilibrium evaluation of Dutch railway proposals. Res. Transp. Econ. 31 (1), 19–28. https://doi.org/10.1016/J.RETREC.2010.11.004 (2011).
Kou, Z. L., Liu, X. Y. & Liu, J. Has industrial policy led to overcapacity? An empirical study based on China's industrial sectors. Fudan J. Humanit. Soc. Sci. 5, 148–161 (2017).
Li, D., Zhang, Y., Ma, S. & Shao, S. The effects of air pollution on regional labor reallocation and stock effects. Econ. Res. J. 5, 127–143 (2021).
Li, Z. & Yang, S. Y. Fiscal decentralization, government innovation preference, and regional innovation efficiency. Manage. World. 12, 29–194. https://doi.org/10.19744/j.cnki.11-1235/f.2018.0031 (2018).
Lin, B. Q. & Tan, R. P. China's economic agglomeration and green economic efficiency. Econ. Res. J. 2, 119–132 (2019).
Liu, Y. Y., Xu, X. X. & Xiao, Z. K. The inverted U-shape pattern of cross-dialect labor migration. Econ. Res. J. 10, 134–146 (2015).
Liu, Z. Y., Li, H. Z., Hu, Y. Y. & Li, C. H. The advancement of human capital structure and economic growth: On the formation and narrowing of the gaps between eastern, central, and western regions. Econ. Res. J. 3, 50–63 (2018).
Romer, P. M. Human capital and growth: Theory and evidence. Carnegie-Rochester Confer Ser. Public. Policy. 32, 251–286. https://doi.org/10.1016/0167-2231(90)90028-J (1990).
Sedgley, N. & Elmslie, B. Do we still need cities? Evidence on rates of innovation from count data models of metropolitan statistical area patents. Am. J. Econ. Sociol. 70 (1), 86–108. https://doi.org/10.1111/J.1536-7150.2010.00764.X (2011).
Shang, Y. P., Pan, Z. & Meng, M. X. Research on the innovation performance of China's urban polycentric spatial strategy: From the perspective of agglomeration economy and comfort. Q. J. Econ. 3, 965–982. https://doi.org/10.13821/j.cnki.ceq.2023.03.09 (2023).
Shao, S., Li, X. & Cao, J. H. China's urbanization and haze governance. Econ. Res. J. 2, 148–165 (2019).
Shen, K. R., Lin, J. W. & Fu, Y. H. Network infrastructure construction, information accessibility, and corporate innovation boundaries. China Ind. Econ. 1, 57–75. https://doi.org/10.19581/j.cnki.ciejournal.2023.01.014 (2023).
Storper, M. & Scott, A. J. Rethinking human capital, creativity and urban growth. J. Econ. Geogr. 9 (2), 147–167. https://doi.org/10.1093/JEG/LBN052 (2009).
Sun, W. H. & Zhang, J. Can high-speed rail networks promote high-quality innovation in the manufacturing industry? J. World Econ. 12, 151–175. https://doi.org/10.19985/j.cnki.cassjwe.2020.12.008 (2020).
Tiebout, C. M. A pure theory of local expenditures. J. Polit Econ. 64 (5), 416–424. https://doi.org/10.1086/257839 (1956).
Wang, C. Y., Lan, Z. M., Zhang, C. & Hou, X. S. High-speed rail construction, human capital migration, and regional innovation. China Ind. Econ. 12, 102–120. https://doi.org/10.19581/j.cnki.ciejournal.2020.12.005 (2020).
Wang, D. H., Gong, L. T. & Li, H. Y. Health human capital, health investment, and economic growth: Based on China's interprovincial data. Manage. World. 3, 27–39. https://doi.org/10.19744/j.cnki.11-1235/f.2008.03.004 (2008).
Wang, H. & Miao, L. China's domestic and international migration development (Springer, 2019).
Wang, Z. H., Ma, J. H., Zhang, B. & Wang, B. Air pollution and urban population migration: Evidence from big data on household smart meters. Manage. World. 3, 19–33. https://doi.org/10.19744/j.cnki.11-1235/f.2021.0033 (2021).
Whisler, R. L., Waldorf, B. S., Mulligan, G. F. & Plane, D. A. Quality of life and the migration of the college-educated: A life-course approach. Growth Change. 39 (1), 58–94. https://doi.org/10.1111/j.1468-2257.2007.00405.x (2008).
Wu, C. P., Li, A. & Zhang, Q. Does air pollution affect the quality of human capital in corporate management? J. World Econ. 2, 151–178. https://doi.org/10.19985/j.cnki.cassjwe.2021.02.008 (2021).
Wu, Y. B. R&D and productivity: An empirical study based on China's manufacturing industry. Econ. Res. J. 11, 60–71 (2006).
Xie, C. Y. & Hu, H. H. China's land resource allocation and urban innovation: Mechanism discussion and empirical evidence. China Ind. Econ. 12, 83–101. https://doi.org/10.19581/j.cnki.ciejournal.2020.12.004 (2020).
Xu, T. & Yao, Y. Intercity population migration and housing price changes: An empirical study based on census and Baidu migration data. J. Jiangxi Univ. Financ Econ. 1, 11–19. https://doi.org/10.13676/j.cnki.cn36-1224/f.2018.01.002 (2018).
Yang, K. & Lee, L. Identification and QML estimation of multivariate and simultaneous equations spatial autoregressive models. J. Econom. 196 (1), 196–214. https://doi.org/10.1016/J.JECONOM.2016.04.019 (2017).
Zhang, C. What makes cities more conducive to entrepreneurship? Econ. Res. J. 4, 151–166 (2018).
Zhang, C. External human capital, cultural diversity, and urban innovation in China. J. World Econ. 11, 172–192. https://doi.org/10.19985/j.cnki.cassjwe.2019.11.009 (2019).
Zhang, C. Clans, entrepreneurship, and development of the private sector in China. J. Comp. Econ. 48 (1), 100–123. https://doi.org/10.1016/J.JCE.2019.08.008 (2020).
Zhu, C. & Lee, C. C. The internal and external effects of air pollution on innovation in China. Environ. Sci. Pollut Res. 28 (8), 9462–9474. https://doi.org/10.1007/S11356-020-11439-Y (2021).
Zou, W. & Dai, Q. Technological imitation, human capital accumulation, and economic catch-up. Soc. Sci. China. 5, 26–206 (2003).

No competing interests reported.

Download PDF

Editorial decision: Revision requested
13 Nov, 2024
Reviews received at journal
13 Nov, 2024
Reviewers agreed at journal
19 Oct, 2024
Reviews received at journal
06 Sep, 2024
Reviewers agreed at journal
31 Aug, 2024
Reviewers agreed at journal
31 Aug, 2024
Reviewers invited by journal
31 Aug, 2024
Editor assigned by journal
30 Aug, 2024
Editor invited by journal
30 Aug, 2024
Submission checks completed at journal
30 Aug, 2024
First submitted to journal
19 Aug, 2024

You are reading this latest preprint version

Environmental quality and the ebb and flow of urban innovation in China: An Explanation from the Perspective of Human Capital Mobility

Status:

Version 1

Abstract

Figures

1. Introduction

2. Related literature and contribution

3. Theoretical framework

3.1 Human capital mechanism through which air pollution affects urban innovation efficiency

3.1.1 Innovation effect of air pollution: the perspective of human capital stock and quality

3.1.2 Innovation effect of air pollution: the perspective of high-skilled human capital flow

3.2.1 Heterogeneous impact of air pollution under urban characteristics

4. Empirical specifications

4.1 Design of model

4.1.1 Baseline regression configuration

4.1.2 Spatial simultaneous equations model configuration

4.2 Variable Definition

4.2.1 Dependent variable: Urban innovation efficiency

4.2.2 Core explanatory variable: Air pollution

4.2.3 Mediating variable

4.2.4 Control variables

4.3 Data Sources

5. Empirical results and analysis

5.1 Benchmark result: the comprehensive impact of air pollution on urban innovation

5.2 Robustness test

5.2.1 Further control the endogenous bias

5.2.2 Replace the spatial weight matrix

5.2.3 Changing the measurement method of core explanatory variables

5.2.4 Changing the measurement method of core mediating variables

5.2.5 Consider other potential influencing factors

6. Mechanism testing

6.1 Test of action mechanism

6.2 Quantitative decomposition of mechanism

7. Analysis of heterogeneity

8. Conclusions and implications

Declarations

Author Contribution

Data Availability

References

Additional Declarations

Status:

Version 1