High-Speed Railway Opening and Urban Innovation Dividend—A Case Study of Guangzhou Zhuhai Intercity High-Speed Railway

Abstract

Based on the panel data of 41 cities along and around the Guangzhou Zhuhai intercity railway (GZR) from 2005 to 2018 and taking the opening of the GZR as a natural experimental scenario, the difference-in-difference (DID) method was used to empirically test the impact of the opening of the GZR on the innovation of cities along the line. The research results show that the opening of GZR has had a statistically positive effect on the innovation of cities along the route—that is, the innovation dividend of high-speed railway (HSR)cities exists, which improves the level of urban innovation—and further, there is time dynamics and regional heterogeneity in the innovation dividend. The research results of this paper have significant policy implications for optimizing China’s HSR network.

1. Introduction

In the past 40 years of reform and development, China has brought economic prosperity by relying on the demographic dividend and the continuous high investment in transportation infrastructure. Transportation infrastructure, as an underlying system of public works designed to facilitate movement, has played a decisive role in promoting economic growth, facilitating resource integration, as well as coordinating social and scientific development [1,2,3,4]. It is suggested that there could be huge dividend yielded from transportation infrastructure [5].

One important transportation infrastructure is HSR, which has expanded rapidly over the last two decades in China. With a total operating mileage of over 40,000 km of high-speed rail in 2022, China now is home to the largest and fastest high-speed rail network in the world. However, with the gradual disappearance of the demographic dividend and the continuous strengthening of resource and environmental constraints, the traditional development mode of relying on large-scale investment in transportation infrastructure is difficult to maintain [6]. Instead, there is a shift toward innovation-driven development, and thus questions arise [7]. As the most important transportation infrastructure, can HSR play a role in the transformation of the economic development mode to high-quality? If so, what are the mechanisms for its creation?

Past research has studied the impact of HSR on cities and found that the opening of high-speed rail can promote urban economic growth, enhance urban accessibility, optimize tourism environment¹, etc., among which the examples of literature mainly focus on exploring the impact of HSR on economic growth. Further, scholars believe that HSR is one of the important factors to promote regional economic growth [8,9], as HSR drives the flow of production factors, improves the efficiency of resource allocation [10], promotes industrial development [11], and has an obvious pulling effect on the development of industries such as real estate, construction, and building materials in particular [12]. A few studies have also investigated the impact of HSR on urban industrial structure, for example, Li Shaokai et al. argued from the perspective of promoting factor flow and reallocation that there is a facilitating effect of HSR on regional industrial structure upgrading [13], which is enhanced with the improvement of the communication capacity of node cities’ transportation networks [14].

In summary, the existing literature, which mainly revolves around the impact of HSR on economic and industrial structures, has not yet touched on the most central network feature of HSR—the differences between cities connected by HSR, the interoperability and learning between cities, the two-way flow of innovation resources, and the increased concentration of advantageous resources, thus generating an urban innovation dividend. Based on the high cost of HSR and the urgency of high-quality development and transformation, an in-depth analysis of the innovation dividend and its generation mechanism of cities along the high-speed railway (hereinafter referred to as HSR cities) is of great significance for spatially optimizing the allocation of HSR resources and promoting the coordinated development of China’s regional economy.

2. Literature Review

In this paper, the innovation dividend of high-speed rail cities refers to the fact that the opening of high-speed rail has increased the integration of innovation factors, including labour, knowledge, and talent, in cities along the route [15], and innovation factors such as technology, knowledge, talent, and capital with different characteristics have been disseminated, exchanged, and learned among cities, thus improving the urban innovation environment, enhancing innovation capacity, improving innovation quality, and realising the “diffusion” and spillover benefits of innovation. The HSR city innovation dividend is the net benefit of innovation. If the innovation resources of cities along the line are siphoned off because of the opening of HSR, resulting in brain drain and decline in innovation capacity, the city innovation dividend is negative at this point; otherwise, it is positive. Figure 1 depicts the innovation dividend generation mechanism in HSR cities.

2.1. HSR and the Gathering of Talent Elements

Unlike traditional transportation modes, HSR is mainly passenger-oriented and connects mainly human resource elements, and with the movement of talents, it means an accompanying movement of scientific, technological, academic, and investment activities and patented achievements—and knowledge has spatial spillover benefits. Information and knowledge-intensive industries such as finance, education, and tourism can benefit more from HSR connectivity [16]. The opening of HSR has increased inter-regional accessibility and met the demand for high value-added innovation that relies on face-to-face interaction [17]. High-tech talents, as the main creators of innovative output, are highly sensitive to time. The characteristics of HSR, such as speed and punctuality, can effectively reduce the travel time of high-tech talents, accelerate the flow of talents and knowledge between cities, and promote the depth and breadth of knowledge transmission, which is particularly significant for the improvement of innovation ability of high-tech enterprises [18]. Yang Siying et al. have investigated the employment and residence preferences of the population along the HSR in the Beijing–Tianjin–Hebei urban agglomeration and discovered that the population with a high education level tends to migrate to the Beijing–Tianjin–Hebei city [19]. The research findings suggested that HSR is attractive to scientific and technological talents. Under the premise of the continuous influx of innovative talents, the city’s innovation ability and innovation vitality continue to improve, promoting the city’s continuous iterative accumulation to form a virtuous circle of innovation paths [15].

2.2. HSR and Gathering of Capital Factor

Without capital inflow and continuous capital input, urban innovation struggles. HSR compresses the space–time distance between cities, bringing the convenience of face-to-face communication and reducing the information asymmetry between investors and entrepreneurs and firms. This makes the chances of venture capital investment for firms in HSR cities much higher, especially in the more information-sensitive start-up and expansion stages, and conducive to urban innovation output. Long Yu et al. found that the opening of HSR in China reduced the spatiotemporal constraint of geographical distance in HSR cities and significantly increased the amount of venture capital investment in HSR cities, in which innovative talent, the level of science and technology, and the high rate of return on project investment were all important factors in attracting capital concentration [20].

2.3. HSR and Gathering of Scientific and Technological Elements

The opening of HSR has enhanced talent interchange and flow between cities, and more talent flow will facilitate the flow of knowledge and technology, thus fostering the agglomeration of scientific and technological among innovation elements. By using the “innovation potential” indicator, Berliant et al. quantified the scientific and technical agglomeration capacity of 30 HSR cities in Southwest China [21]. The results revealed that the quantitative indicators of HSR cities were greater than those of cities without HSR. This perspective is also supported by Hongyu Fan [22], who examined the cities along the Zhengzhou—Wuhan HSR as the study object and came to the conclusion that the opening of the HSR was the primary cause of the growth in scientific and technical agglomeration.

The above analysis demonstrates that HSR, as a typical rapid transportation infrastructure, can effectively drive the redistribution of information and knowledge among cities by reducing the transportation costs between cities. The essence of HSR urban innovation dividend is that the opening of HSR brings knowledge spillover [22], which has spillover effect on urban innovation [23,24,25]. The flow of innovation elements among cities is restricted by spatial location and the flow of innovation elements among cities with relatively close distances is faster. Based on the basic understanding of the spillover effect of HSR on urban innovation, many countries in the world put HSR in an important position in promoting the flow of innovative elements between cities and stimulating the innovation vitality of cities [4]. Due to the huge differences between HSR cities in terms of population size, industrial structure, and economic development level, the impact of HSR connectivity on urban innovation dividends also differs. This paper takes the Guangzhou Zhuhai Intercity Railway (GZR) as an empirical object to study the innovation dividend of GZR cities and analyzes its spatial differences and temporal characteristics.

As a transportation infrastructure, the opening of HSR brings convenience, reduces people’s transportation time cost, and facilitates the flow and gathering of scientific and technological talents. Meanwhile, the flow and gathering of scientific and technological talents promote the flow, interaction and study of knowledge, science, and technology. Smart capital pursues science and technology and talent resources, and the elements of the innovation system promote each other, promoting the evolution of the innovation system of HSR cities. The mechanism of the innovation dividend of HSR cities is the principle that the opening of HSR will gather talents, capital, and scientific and technological elements in cities along the line, and the elements will promote each other.

3. Materials and Methods

3.1. Study Sample

GZR is an intercity railway connecting Guangzhou and Zhuhai in Guangdong Province, China, and is one of the main lines of intercity rapid rail transit in the Pearl River Delta. GZR runs through the Pearl River Delta, passing through Guangzhou, Foshan, Zhongshan, Jiangmen, and Zhuhai, with a total length of 132.24 km and total investment of 19.5 billion yuan. GZR started construction in December 2005 and opened to traffic in December 2012, shortening the travel time from Guangzhou to Zhuhai by 46%. In order to analyze the effects of GZR on urban innovation and estimate the innovation dividend effect, this paper utilizes the opening of the GZR as an experimental event, the opening time of the GZR as the policy time point, and the cities along the GZR as the processing group. According to the availability of GZR data, the research period span for this paper is 2005–2018; 2012 is the “policy time point”, 2005–2012 is the 7-year construction period, and 2013–2018 is the 7-year operation period.

This paper uses the panel data of 41 cities along and around GZR from 2005 to 2018 for empirical analysis. Five cities along the GZR are set as processing groups. According to the comprehensive conditions such as urban population size and economic development level, control groups are selected from other cities in Guangdong Province, as well as cities in neighboring provinces such as Hunan, Fujian, and Jiangxi. There are 36 cities in the control group, such as Guangdong Province (15), including Chaozhou, Huizhou, etc.; Hunan Province (8), including Changde, Hengyang, etc; Fujian Province (7), including Nanping, Longyan, etc.; Jiangxi Province (7), including Pingxiang, Ji’an, etc.

3.2. Research Hypothesis

Hypothesis 1: The opening of GZR can bring urban innovation dividends, which is mainly due to the concentration of scientific and technological resources and talent resources and the inflow of capital brought about by the opening of GZR.

Hypothesis 2: The effect of the GZR’s opening on innovation in the cities along the line is spatially heterogeneous and temporally evolving.

3.3. Model Setting

In order to analyze the innovation dividend effect and the characteristics in the spatial dimension of the opening of GZR for a total of 41 cities along and around the route, respectively, this paper applies the double difference model (DID) and the spatial dual difference model (SDID), and verifies the reliability of the model and data through parallel hypothesis experiments and spatial correlation tests, etc. The DID model is as follows.

(1)$Y_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1+G{D}_{it}+{{\displaystyle \sum }}_{m=1}^M{\delta }_m{Z}_{it,m}+{\lambda }_i+{\alpha }_t+{\epsilon }_{it}$

where, $i$ represents the analysis city object ($i$ = 1, 2, 3, 4, 5 represents the processing group, $i$ = 6, 7, 8, 9... 41 represents the control group); $t$ represents the year of analysis ($t$ = 2005, 2005, 2007, 2008... 2018); Yit is the explained variable, representing the urban innovation dividend; ${\beta }_0$ is a constant term; ${\beta }_1$ indicates the net effect of the opening of HSR on the urban innovation dividend; $G{D}_{it}$ is the treatment effect, denoting the difference in the impact of the opening of the HSR on the processing and control groups, $D_t$ denoting the time dummy variable; $G_i$ denotes the city dummy variable, $G{D}_{it}=G_i\times D_t$; $Z_{it,m}$ represents the set of control variables, including 9 control variables such as scientific and technological level and industrial organization; ${\delta }_m$ denotes the effect size of the m-the control variable on the independent variable; ${\lambda }_i$ and ${\alpha }_t$ represent the fixed effects in terms of both individual city and time selected for the processing group and the control groups, respectively; ${\epsilon }_{it}$ is a random error.

3.4. Indicators Selection

3.4.1. The Explained Variable

This paper selects talent level, capital level, and science and technology level as explained variables (Seen in Table 1). Talent level is an important variable of GZR urban innovation dividend. As an explanatory variable, talent level is calculated as the sum of employees in six industries (finance, computer services and software, scientific research, education, culture, sports and entertainment, leasing and commercial services) divided by the total area of the city [26,27]; Capital level, which implies whether the city’s economic output has a higher technological content, is equal to the real GDP divided by the city area; The level of urban science and technology, which is assessed in this study using the urban innovation indicators [28], determines the extent of the urban innovation dividend.

3.4.2. The Core Explanatory Variable

In this paper, $G{D}_{it}$, the product of the time virtual variable $D_t$ and the city virtual variable $G_i$, is selected as the core explanatory variable, where $G{D}_{it}=G_i\times D_t$ (Seen in Table 1). $D_t=0$ during GZR construction; $D_t=1$ when GZR is opened; The processing group city is recorded as $G_i=1$, and the control group city is recorded as $G_i=0$.

3.4.3. The Control Variable

The control variable refers to the factors that may have a certain impact on urban innovation. As the urban innovation dividend is not only affected by the opening of GZR, but its formation is also a complex process. This paper takes government science and education support [29], wage level [30], urban population density [30], urban employed population [30], fixed assets [31], market opening environment [31], market financial environment and industrial structure [32] as control variables, and comprehensively considers the impact of various factors on urban innovation dividends (Seen in Table 1).

4. Experimentation and Results

4.1. Data Source and Description

4.1.1. Data Source

All the data sources in this paper are the statistical yearbooks of Guangdong, Jiangxi, Hunan, and Fujian provinces, the report on China’s cities and industrial innovation, and the statistical bulletin on national economic and social development of 41 sample cities. To ensure the balance of panel data, a few missing data shall be filled by averaging before using the model for measurement [33,34]. We also conducted the analysis without the missing samples, which produced the consistent results.

4.1.2. Data Processing

The entire GZR is divided into two sections, with the opening year being 2011 and 2012, respectively. In this paper, the opening year is determined by whether the cities along the GZR are opened to HSR before June of that year; if not, the opening year is postponed by one year. Except for the market financial environment and the government science and education support index, all indicators are taken as logarithms to meet the basic assumptions and reduce the degree of heteroskedasticity in the model [33,34]. Taking logarithms of all data can reduce the absolute difference between the data and avoid the impact of individual extreme values. It can also meet the classical linear model assumption, avoid collinearity, avoid heteroscedasticity, meet the basic assumption of the same variance, and conform to the normal distribution as much as possible. Table 2 is descriptive statistical data such as extreme value and mean value of variables substituted into the model.

4.2. Propensity Score Matching Result

The parallel trend hypothesis [35], which states that before the inauguration of GZR, the innovation dividends of cities along the line and those not along the line should have the same changing trend, must be satisfied by the study sample for the DID model to hold. The score matching probability density graphs of the treatment group and the control group before and after Propensity Score Matching (PSM) treatment (Figure 2 and Figure 3). The control group and treatment group are in the range of 0–0.5 and 0.8–1.0, respectively, and the gap is narrowed, which indicates that after PSM score matching, the variation trend difference between covariates in GZR treatment group and control group is minor, and there is no significant difference. “Parallel trend” hypothesis is established.

Before PSM score matching, the difference between the treatment and control groups was large in the 0–1.0 interval; whereas, after PSM propensity score processing, the standard deviations of the covariates all decreased to varying degrees, and the standardized deviations of all covariate indicators decreased to within 30% (Seen in Table 3). Further, the matched data had relatively balanced performance, indicating that the selected control variables were reasonable.

In addition, the propensity score matching also needs to meet the common support conditions. In order to improve the quality of sample matching and increase the effectiveness of propensity score matching estimation, it is necessary to further test the overlapping area of propensity scores of the processing group and the control group. As can be seen from Figure 4, most of the samples are within the common range of values and the distribution approximately obeys the positive terrestrial distribution, satisfying the common support condition.

4.3. Overall Empirical Analysis

The PSM-DID model is then applied to analyze the urban innovation dividend brought by the opening of GZR in terms of talent level, science and technology level, and capital level. Considering the difference of city scale, the treatment group is divided into mega-city, large city, and mid–minor city according to the population scale of 10 million, 5 million, and 1 million, respectively. Among them, Guangzhou is a mega-city, Foshan is a large city, and Zhuhai, Jiangmen, and Zhongshan are mid–minor cities.

4.3.1. In Terms of Talent Level

• According to the second column of Table 4, the urban innovation dividend regression results are significant at the level of 5% without controlling factors.

• The regression result after the control variable is added is shown in the third column of Table 4. The urban innovation dividend shows a positive impact, indicating that the control variable has a good promotion effect on the talent concentration of HSR cities.

• According to columns 4, 5, and 6 of Table 4, the cross-term coefficient of innovation dividend of mega-cities is significantly 13.02, slightly larger than that of large cities and much higher than that of mid–minor cities. It shows the opening of GZR, mega-cities, and large cities have brought huge innovation dividends in terms of talents, attracted a large number of scientific and technological talents, and produced a talent-gathering effect.

4.3.2. In Terms of Capital Level

• According to the second column of Table 5, the regression result of urban innovation dividend is significant at the level of 1% without controlling variables.

• The third column of Table 5 is the regression result after adding the control variables. The regression coefficient of urban innovation dividends is 0.940, which is significant at the level of 1%, indicating that the control variables have a positive impact on urban innovation dividends in terms of capital and are basically significant. Among them, the cross-term coefficient of innovation dividends in large cities is significantly 1.242, which is larger than that of mega-cities and mid–minor cities. This suggests that the innovation dividend in terms of capital is higher in large cities than in mega-cities and mid–minor cities, which is probably due to Guangzhou’s restricted geographical control as a mega-city, industrial structure specialising in services, and the slowdown in economic growth, while smaller cities—despite the economic boom that will be brought about by the opening of the GZR—do not have a large scale of capital investment compared to mega-cities and large cities.

4.3.3. In Terms of Scientific and Technological Level

• The second column of Table 6 is the regression result of urban innovation bonus without controlling variables, which is significant at the level of 1%; the third column of the table is the regression result after adding the control variables, which are significant at the level of 5%. In terms of science and technology level, the regression coefficient of urban innovation bonus is 11.40, which is reduced when compared with the benchmark DID model. The result is positive and significant at the level of 5%, indicating that the control variables in terms of technology level and urban innovation dividends are positive and significant.

• It is observed that the coefficients of the urban innovation dividend cross multiplier terms for cities of different sizes in terms of science and technology levels are significantly positive, and that mid–minor cities < large cities < mega-cities, indicating that all cities in the processing group have improved their science and technology level after the opening of the GZR, generating a science and technology dividend, and that the “Matthew effect”, “siphon effect”, and “cannibalisation effect” of HSR cities have occurred, but the effect of mega-cities is more pronounced.

4.4. Time Analysis

Numerous empirical research has examined the time effect of HSR in fostering economic growth. The project dividend has a long-term nature, and the investment income will last for 25 years or even longer [36,37]. Does the urban innovation dividend also have some characteristics of the time dimension? On the basis of Formula (1), this paper introduces the time dummy variable $G{T}^j$ ($j\in \left[-3,6\right]$ and is an integer, which can be adjusted to other time periods) and constructs a multi-stage DID model as Formula (2), in order to analyze whether the innovation dividend of GZR cities has a time trend phenomenon.

(2)$Y_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1+G{D}_{it}+{{\displaystyle \sum }}_{m=1}^M{\delta }_m{Z}_{it,m}+{\lambda }_i+{\alpha }_t+{\epsilon }_{it}$

where, $G{T}^j$ is a dummy variable, representing the |$j$|-th year of 2005–2018. $\rho$ represents the impact of $G{T}^j$ on urban innovation dividends. See Formula (1) for the meanings of the rest variables.

4.4.1. Talent Level

Table 7 shows that, in terms of talent, the innovation dividend effect of GZR cities is positive over the study period, and that the influence coefficient is rising annually, suggesting that the opening of GZR has drawn a considerable number of talents and created a talent highland.

4.4.2. Capital Level

Table 8 shows that, in terms of capital, GZR’s urban innovation dividend effect is positive and significant over the study period, but $G{T}^j$ starts to decrease significantly from the third year after GZR’s opening, which indicates that the urban innovation dividend observed in terms of capital after GZR’s opening shows a gradual increase first, but there is a tendency not to converge with time.

4.4.3. Scientific and Technological Level

According to Table 9, the regression coefficient of urban innovation dividend of GZR has been on the rise since the first year after its opening, and has become significant since the third year, indicating that it takes a certain time to improve the scientific and technological level of cities in the Pearl River Delta. With the opening of GZR, the urban innovation dividend observed in science and technology shows a time lag effect.

4.5. Spatial Analysis

In the process of exploring the innovation effect of GZR cities, spatial factors should be considered. This paper adds spatial parameters based on DID model, constructs a dual difference model considering spatial (SDID) effects, and analyzes the spatial effects of GZR urban innovation dividend.

(3)$Y_{it}=\beta 0+\psi WYit+W({\displaystyle \sum _{m=1}^M\delta mZ\mathrm{it},\mathrm{m}})\theta +\beta 1\times (Dt\times Gi)+W(Dt\times Gi)\varphi +{\displaystyle \sum _{m=1}^M\delta mZ\mathrm{it},\mathrm{m}}+\epsilon \mathrm{it}$

where, $\psi$ is the spatial matrix, and $W{Y}_{it}$, $W({\displaystyle \sum _{m=1}^M\delta m{Z}_{\mathrm{it},\mathrm{m}}})$ represents the lag term of $Y_{it}$, ${\displaystyle \sum _{m=1}^M\delta mZ\mathrm{it},\mathrm{m}}$, respectively, $\psi$ and $\varphi$ is the size of the innovation dividend of the control group and the processing group, respectively.

Table 10 shows the regression results: (1) It can be seen that the $G{D}_{it}$ results of SDID regression are less than that of PSM-DID regression. Therefore, spatial factors should be considered in the process of measuring urban innovation dividends, otherwise the measurement results will tend to be bigger; (2) Applying SDID to measure the $G{D}_{it}$ of urban innovation dividend in talent, capital, and science and technology, the figures are 4.228, 0.450, and 5.28, respectively, which are significant, indicating that the overall GZR urban innovation dividend is positive.

Table 11 shows the direct effect of the innovation dividends of the processing group, the indirect effect of the innovation dividends of the control group, and the total effect of the innovation dividends of 41 cities. It can be seen that:

• The $G{D}_{it}$ of direct, indirect, and total effects of urban innovation dividends in talent, capital, and science and technology are positive and significant, indicating that the opening of GZR has a positive and significant impact on the innovation dividends of cities along the line and surrounding cities.

• The spatial parameters $\psi$ are significantly 0.5841, 0.6612, and 0.6021, respectively, which indicates that after the opening of GZR, cities along the line have generated positive spillover of talents, capital, and technology; At the same time, the relatively big spatial parameter $\psi$ indicates the direct effect of innovation dividend is higher than the indirect effect, suggesting the opening of GZR has a positive impact on the innovation integration of urban agglomeration along the line and in the Pearl River Delta.

• The spatial parameters $\theta$ are significantly 0.0003, 0.0016, and 0.0009, respectively, which indicates that the opening of GZR has a positive impact on the talent, capital, and technology levels of the 36 surrounding cities; Meanwhile, the relatively small spatial parameter $\theta$ indicates weaker positive impact relationship, suggesting that the flow of innovative elements among cities is subject to spatial location, and the flow of innovation factors is more fully developed between cities relatively close in spatial location.

• Among the control variables, wages, the tertiary industry structure, and the government’s support for science and education have the most significant impact on the innovation dividend of talents; meanwhile, the tertiary industry structure, the market financial environment, and fixed asset investment have the most significant impact on the innovation dividend of capital; and the science and education resources and the government’s support have the most significant impact on the innovation dividend of technology. The above results are consistent with the regression results of PSM-DID model.

5. Discussion and Conclusions

As an important carrier for the spatial flow of human resources, capital, and science and technology, the construction of GZR has significantly improved the accessibility of the Pearl River Delta, enhanced the flow efficiency of information, human resources, capital, and other production factors, played a very important role in promoting urban innovation, and brought innovation dividends to cities. Based on the existing theoretical and empirical research results, this paper uses the panel data of 41 cities along and around the GZR in the Pearl River Delta to empirically analyze the impact of the opening of the GZR on urban innovation and verify the establishment of hypotheses 1 and 2, utilizing the double difference model. The main research findings of this paper include:

• (1). Overall, the opening of GZR has significantly improved the innovation of cities along the line; nevertheless, the positive effect of capital on urban innovation dividends does not weaken with time, and there is a significant lag effect in science and technology and talents. The existence of innovation dividends mainly benefits from the increase in population flow and capital inflow brought by GZR connectivity.

• (2). From the perspective of city scale, the opening of GZR has not only promoted the innovation dividends in talent, capital, and science and technology of cities of various sizes along the line, but also widened the gap in innovation dividends among cities of different sizes along the line. The comprehensive income of large-scale cities is higher.

• (3). From the perspective of space, the opening of GZR has a direct effect on innovation dividends of cities along the line and an indirect effect on innovation dividends of surrounding cities, but the positive spillover effect on cities along the line is higher than that on surrounding cities.

The findings of this paper have important policy implications. The Chinese economy has shifted from high-speed growth to high-quality development, and technological progress has taken the place of rapid expansion as the main factor advancing China’s economic development. As a transportation infrastructure, HSR accelerates the integration of resources in space, attracts high-tech talents, and improves the city’s innovation ability. Therefore, before planning the transportation infrastructure, the government should comprehensively evaluate the existing transportation facilities, including economic development, environmental inclusion, technological innovation, sustainable development innovation, etc., and optimize the layout of the HSR network on this basis.

6. Limitations

The paper has one limitation in that we have a relatively small sample size, although our sample is representative to some extent. The GZR is one of the main lines of intercity rapid rail transit in the Pearl River Delta region, which has a huge impact on the economic and social development of the Guangdong–Hong Kong–Macao Greater Bay Area. This area always serves as a nice demonstration of China’s reform and opening-up policy. We choose the panel data of 41 cities along and around GZR in this area from 2005 to 2018 for empirical analysis to demonstrate our hypothesis. The sample size is restricted but sufficient to support our results since treatment effects are large enough in this semi-demonstration project. Nevertheless, in the future, we hope we will have more chances to look into the research questions with richer data as China vastly develops its high-speed railway network.

Author Contributions

Conceptualization, Y.Q.; Methodology, Y.Q. and J.X.; Formal analysis, J.L. and J.X.; Resources, J.X. and S.C.; Writing—original draft, S.C. and J.L.; Writing—review & editing, Y.Q.. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

All data, models, and code generated or analyzed during this study are included in the submitted article. Code can be available from the corresponding author on reasonable request.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

Notes

1. According to the Economic Blue Book Summer No.: China’s Economic Growth Report (2017~2018), China’s total factor productivity contributed 20.83% to economic growth from 2013 to 2017. The increase in total factor produc-tivity means that economic growth gradually gets rid of factor input drive and turns to endogenous technology drive.

Footnotes

Disclaimer/Publisher’s Note: Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Generation mechanism of innovation dividend in high-speed rail cities.

Figure 2. Probability density before matching.

Figure 3. Probability density after matching.

Figure 4. Common value range of covariates before and after matching.

Table 1

Model variables, calculation methods and meanings.

Type	Variable	Symbol	Calculation Formula
main explanatory variables	the opening of high-speed railway	$G{D}_{it}$	$G{D}_{it}=G_i\times D_t$
explained variable	talent level	perland	total number of people in six industries in perland/City area
	capital level	e-density	ln(actual GDP/urban area)
	scientific and technological level	technology	urban innovation index
control variable	market financial environment	fin	proportion of the balance of financial institutions in the actual GDP at the end of the year
	market opening environment	fdi	external investment/regional real GDP
	government support for science and education	gov	total expenditure on science and education/financial expenditure
	science and education level	edu	number of college students/urban population
	salary level	sal	average salary of employees in urban units
	urban employed population	labor	$\mathrm{employees}\mathrm{in}\mathrm{urban}\mathrm{units}\mathrm{at}\mathrm{the}\mathrm{end}\mathrm{of}\mathrm{the}\mathrm{year}$
	urban population density	p-density	urban resident population/urban land area
	industrial structure	Industry2	output value of secondary industry/GDP
		Industry3	output value of tertiary industry/GDP
	fixed assets	asset	urban fixed assets

Table 2

Descriptive statistics of model variables.

Variable	Minimum	Maximum	Mean	Standard Deviation	Observed Value
GDit	0	1	0.06	0.24	574
perland	14.4	47.7	29.97	10.72	574
e-density	4.79	10.17	7.12	1.29	574
technology	1.42	252.89	16.17	13.28	574
fin	0.12	0.58	0.32	0.14	574
lnfdi	9.81	12.7	9.82	0.20	574
gov	0.053	1.711	0.16	0.15	574
lnedu	0.044	0.767	0.17	0.08	574
lnasset	13.75	18.11	15.85	0.99	574
lnsal	9.27	15.47	10.55	0.82	574
lnlabour	2.47	6.19	3.75	0.74	574
lnp-density	4.07	7.88	6.24	0.71	574
lndustry2	0.27	0.69	0.51	0.08	574
lndustry3	0.26	0.72	0.43	0.08	574

Table 3

PSM covariate data balance test.

Covariant/before and after Matching		Mean		T-Test		Standard Deviation (%)	Deviation Reduction (%)
		Processing Group	Control Group	t-Value	p > |t|
market financial environment	Unmatched	11.659	10.223	13.42	0.000	187.7	84.1
	Matched	11.659	11.888	−1.41	0.160	−19.9
market opening environment	Unmatched	9.9293	9.81	4.74	0.000	29.7	8.8
	Matched	9.9293	9.81	1.76	0.103	19.7
government support for science and education	Unmatched	0.1001	0.1715	−3.78	0.000	−62.5	84.5
	Matched	0.1001	0.1112	−1.82	0.173	−9.7
science and education level	Unmatched	0.2463	0.1639	9.22	0.000	116.2	81.2
	Matched	0.2463	0.2619	−0.99	0.326	−21.9
salary level	Unmatched	10.753	10.521	2.22	0.027	33.3	98.1
	Matched	10.753	10.757	−0.05	0.960	−0.6
labor	Unmatched	2.2800	2.12	2.86	0.004	37	13.5
	Matched	2.2800	2.23	0.76	0.451	13.3
Population density	Unmatched	6.7869	6.167	7.15	0.000	109.9	69.9
	Matched	6.7869	6.9737	−2.28	0.124	−23.1
tertiary industry structure	Unmatched	0.4525	0.4281	2.27	0.024	25.3	19.7
	Matched	0.4525	0.4720	−1.20	0.232	−20.7
fixed assets	Unmatched	16.068	15.310	6.73	0.000	86.1	87.5
	Matched	16.068	16.162	−0.60	0.552	−10.8

Table 4

Regression analysis of urban innovation bonus (talent level).

Variable	Basic DID	PSMDID	Mega-Cities	Large Cities	Mid–Minor Cities
$G{D}_{it}$	15.08 **	9.87 ***	13.018 ***	12.42 ***	7.16 **
	(2.58)	(3.32)	(10.85)	(13.03)	(2.13)
fin		12.05 *	10.126 ***	21.75 ***	1.17 *
		(1.91)	(7.57)	(6.17)	(1.72)
Infdi		3.872 *	2.187 **	2.023 ***	3.232
		(1.82)	(2.52)	(−3.41)	(0.41)
Inasset		10.27 **	10.12 ***	12.83 ***	9.33 ***
		(2.26)	(11.69)	(21.54)	(12.57)
gov		2.27 ***	1.65 ***	4.17 ***	6.24 ***
		(4.98)	(3.08)	(3.20)	(5.50)
Inedu		8.7 **	2.45 *	10.83 ***	1.88 **
		(2.38)	(2.21)	(5.70)	(2.70)
Insal		10.47 ***	8.54 ***	12.47 ***	4.90 **
		(6.17)	(6.24)	(6.17)	(3.07)
Inlabor		9.02 *	9.11 **	8.85 **	2.85
		(1.92)	(3.47)	(3.20)	(1.00)
Inp-density		2.02 **	4.52 **	1.85 **	3.12 **
		(2.06)	(2.40)	(1.99)	(2.09)
Industry2		1.1053 **	1.101 **	1.006 **	2.20
		(2.32)	(2.52)	(2.45)	(1.32)
Industry3		10.942 ***	1.098	2.28 ***	1.093
		(6.19)	(0.90)	(12.16)	(1.26)
_Cons	−10.221 ***	−11.783 ***	−208.97	−189.08 ***	−199.15 ***
	(−8.09)	(−20.11)	(−0.26)	(−16.12)	(−15.47)
R-squared	0.945	0.969	0.812	0.868	0.884
City FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 5

Regression analysis of urban innovation dividend (capital level).

Variable	Basic DID	PSMDID	Mega-Cities	Large Cities	Mid–MinorCities
$G{D}_{it}$	0.928 ***	0.940 ***	1.018 **	1.242 ***	0.716 **
	(4.38)	(10.32)	(2.55)	(3.03)	(2.43)
fin		0.1205 ***	0.187 *	0.075 ***	0.117
		(7.03)	(1.70)	(5.87)	(0.34)
Infdi		0.3672 ***	0.130 **	0.203 ***	0.012
		(4.59)	(2.57)	(4.41)	(0.71)
Inasset		0.072 **	0.032 **	0.083 **	−0.090 ***
		(2.60)	(2.01)	(2.54)	(3.30)
gov		−0.008 **	−0.85	−0.107 ***	0.025 **
		(−2.18)	(0.45)	(3.02)	(2.59)
Inedu		0.047 **	0.021 *	0.103 ***	0.083
		(2.53)	(1.93)	(4.17)	(0.72)
Insal		0.097 **	0.54 *	0.041 ***	0.090 *
		(2.57)	(1.74)	(3.17)	(1.83)
Inlabor		0.932 **	0.910 **	0.095 **	0.275
		(2.52)	(2.49)	(2.61)	(0.70)
Inp_density		0.101 **	0.052 *	0.005 *	0.032
		(2.46)	(1.76)	(1.99)	(0.39)
Industry2		0.054 **	0.180 ***	0.104 **	0.220 *
		(2.31)	(2.92)	(2.55)	(1.92)
Industry3		0.122 ***	0.118 ***	0.203 ***	0.029 **
		(4.09)	(2.61)	(3.16)	(2.43)
_Cons	12.291 ***	1.794 ***	−1.44 **	−9.38 ***	−5.15 **
	(4.19)	(5.11)	(−3.06)	(−4.92)	(−3.07)
R-squared	0.930	0.918	0.847	0.827	0.833
City FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 6

Regression analysis of urban innovation bonus (science and technology level).

Variable	Basic DID	PSMDID	Mega-Cities	Large Cities	Mid–MinorCities
$G{D}_{it}$	19.12 ***	11.40 **	12.18 **	10.40 **	7.73 *
	(6.48)	(2.53)	(2.61)	(2.49)	(1.97)
fin		1.205 ***	1.800 ***	1.170 **	2.107 *
		(4.03)	(3.10)	(2.57)	(1.94)
Infdi		16.22 **	17.20 ***	10.13 ***	14.12 **
		(2.59)	(4.27)	(4.01)	(3.17)
Inasset		9.70 *	−10.02 **	18.43 **	20.09 ***
		(1.80)	(−2.70)	(2.54)	(4.13)
gov		18.12 ***	22.50 **	20.07 **	7.02 *
		(5.21)	(4.32)	(2.44)	(1.25)
Inedu		11.00 ***	10.01 ***	9.00 **	12.03 ***
		(6.33)	(4.11)	(4.17)	(5.20)
Insal		9.97 *	10.94 **	8.01 *	11.00
		(1.89)	(3.04)	(2.30)	(0.13)
Inlabor		3.62 **	4.10 **	6.00 *	4.25 *
		(2.40)	(2.49)	(2.21)	(1.99)
Inp_density		10.11	10.02 *	10.00	7.32
		(0.26)	(1.86)	(0.99)	(0.49)
Industry2		2.54 *	2.80 **	2.04 *	2.20 *
		(1.87)	(2.60)	(1.85)	(1.79)
Industry3		11.22 **	12.18 ***	14.43 ***	10.07 *
		(2.09)	(4.61)	(5.76)	(2.22)
_Cons	12.291 ***	1.794 ***	−1.44 **	−9.38 ***	−5.15 **
	(4.19)	(5.11)	(−3.06)	(−4.92)	(−3.07)
R-squared	0.770	0.738	0.789	0.801	0.707
City FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 7

Time effect analysis of GZR urban innovation dividend (talent level).

Variable	GT−3	GT−2	GT−1	GT1	GT2	GT3	GT4	GT5	GT6	GT7
$G{T}^j$	2.314	3.192	3.581	4.191	4.061	4.28 **	6.17 ***	7.16 ***	9.57 ***	9.87 **
	(0.26)	(1.59)	(1.42)	(1.25)	(1.41)	(3.44)	(4.30)	(5.11)	(5.09)	(3.02)
city	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
year	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
control	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
R-squared	0.808	0.817	0.904	0.909	0.923	0.916	0.918	0.906	0.910	0.912

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 8

Time effect analysis of GZR urban innovation dividend (capital level).

Variable	GT−3	GT−2	GT−1	GT1	GT2	GT3	GT4	GT5	GT6	GT7
$G{T}^j$	0.012	0.019	0.025	0.045 **	0.050 ***	0.056 ***	0.063 **	0.071 **	0.083 **	0.099 **
	(0.76)	(0.69)	(0.93)	(2.55)	(3.91)	(4.24)	(2.20)	(2.11)	(2.42)	(2.30)
city	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
year	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
control	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
R-squared	0.799	0.697	0.714	0.869	0.872	0.906	0.892	0.773	0.790	0.680

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 9

Time effect analysis of GZR urban innovation dividend (scientific and technological level).

Variable	GT−3	GT−2	GT−1	GT1	GT2	GT3	GT4	GT5	GT6	GT7
$G{T}^j$	5.10	6.66	6.29	7.01	7.25	8.50 **	11.79 ***	14.05 **	15.84 *	16.00 **
	(0.79)	(1.09)	(0.92)	(1.35)	(1.41)	(4.24)	(5.12)	(2.67)	(2.08)	(2.63)
city	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
year	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
control	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
R-squared	0.529	0.612	0.643	0.665	0.712	0.816	0.802	0.713	0.706	0.810

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 10

Regression results of SDID.

Variable	Spatial Double Difference SDID Model			PSM-DID Dual Difference Model
	Talent	Capital	Technology	Talent	Capital	Technology
$G{D}_{it}$	4.228 **	0.450 ***	5.28 **	9.87 ***	0.940 ***	11.40 **
	(2.11)	(13.02)	(2.49)	(3.32)	(10.32)	(2.53)
fin	12.05 *	0.0233 ***	1.205 ***	12.05 *	0.1205 ***	1.205 ***
	(1.91)	(7.03)	(4.03)	(1.91)	(7.03)	(4.03)
Infdi	1.377 *	0.0612 ***	3.119 **	3.872 *	0.3672 ***	16.22 **
	(1.78)	(5.29)	(2.43)	(1.82)	(4.59)	(2.59)
Inasset	3.18 **	0.019 **	1.78 *	10.27 **	0.072 **	9.70 *
	(2.50)	(2.40)	(1.83)	(2.26)	(2.60)	(1.80)
gov	0.922 ***	−0.002 **	8.33 ***	2.27 ***	−0.008 **	18.12 ***
	(10.44)	(−2.09)	(10.23)	(4.98)	(−2.18)	(5.21)
Inedu	2.09 **	0.0097 **	2.40 ***	8.7 **	0.047 **	11.00 ***
	(4.79)	(2.20)	(7.54)	(2.38)	(2.53)	(6.33)
Insal	5.25 ***	0.011 **	2.56 *	10.47 ***	0.097 **	9.97 *
	(7.22)	(2.32)	(1.90)	(6.17)	(2.57)	(1.89)
Inlabor	4.77 *	0.238 **	1.78 **	9.02 *	0.932 **	3.62 **
	(1.88)	(2.56)	(2.33)	(1.92)	(2.52)	(2.40)
Inp-density	0.779 **	0.078 **	3.99	2.02 **	0.101 **	10.11
	(2.10)	(2.46)	(0.90)	(2.06)	(2.46)	(0.26)
Industry2	0.1222 **	0.019 **	0.522 *	1.1053 **	0.054 **	2.54 *
	(2.01)	(2.51)	(1.72)	(2.32)	(2.30)	(1.87)
Industry3	1.942 ***	0.028 ***	2.78 **	10.942 ***	0.122 ***	11.22 **
	(10.39)	(6.56)	(2.14)	(6.19)	(4.09)	(2.09)
R-squared	0.712	0.911	0.825	0.969	0.918	0.738
City FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

Table 11

Direct and indirect regression results of SDID.

	Talent Innovation Bonus			Capital Innovation Dividend			Science and Technology Innovation Dividend
	Direct	Indirect	SUM	Direct	Indirect	SUM	Direct	Indirect	SUM
$G{D}_{it}$	4.17 **	7.64 **	11.81 **	0.109 **	1.223 **	1.332 **	2.091 **	11.229 **	13.320 **
	(2.18)	(2.33)	(2.28)	(2.17)	(2.12)	(2.43)	(2.08)	(2.32)	(2.58)
fin	4.25 *	10.11 *	14.36 *	0.0272 ***	0.1300 ***	0.1572 ***	0.1058 *	1.120 *	1.2258 *
	(1.65)	(1.70)	(2.00)	(15.09)	(14.20)	(17.23)	(1.87)	(1.94)	(2.13)
Infdi	0.370 *	4.111 *	4.481 *	0.0118 **	0.0803 **	0.0921 **	3.110 **	14.028 **	17.138 **
	(1.92)	(1.91)	(1.80)	(2.16)	(2.28)	(2.42)	(2.23)	(2.27)	(2.61)
Inasset	2.07 **	9.32 **	11.39 **	0.0366 ***	0.3756 ***	0.4122 ***	2.21 *	8.09 *	10.30 *
	(2.42)	(2.29)	(2.27)	(8.37)	(19.02)	(26.25)	(1.80)	(1.64)	(1.90)
gov	0.971 ***	1.881 ***	2.952 ***	−0.0019 **	−0.0077 **	−0.0096 **	4.661 ***	16.023 ***	20.684 ***
	(14.08)	(12.19)	(14.67)	(−2.08)	(−2.10)	(−2.30)	(15.01)	(14.32)	(9.09)
Inedu	1.07 **	7.54 **	8.61 **	0.0036 *	0.0580 *	0.0616 *	1.933 ***	12.008 ***	13.941 ***
	(15.03)	(27.33)	(14.33)	(1.79)	(1.88)	(1.83)	(16.03)	(9.45)	(10.13)
Insal	2.44 ***	10.01 ***	12.45 ***	0.017 **	0.133 **	0.150 **	2.07	8.22	10.29
	(16.19)	(16.22)	(15.90)	(2.54)	(2.49)	(2.27)	(1.34)	(1.03)	(0.98)
Inlabor	1.12 *	3.76 *	4.88 *	0.102 *	0.894 *	0.996 *	0.457 *	4.089 *	4.546 *
	(1.71)	(1.82)	(1.90)	(1.72)	(1.94)	(2.03)	(1.70)	(1.81)	(1.92)
Inp_density	0.82 **	4.08 **	4.90 **	0.0117 *	0.1003 *	0.1020 *	1.022	10.809	11.831
	(2.19)	(1.78)	(2.01)	(1.88)	(1.69)	(1.79)	(0.17)	(0.56)	(1.09)
Industry2	0.1013 **	1.3244 **	1.4257 **	0.0094 **	0.0520 **	0.0614 **	0.299 *	4.541 *	4.840 *
	(2.23)	(2.31)	(2.39)	(2.50)	(2.58)	(2.53)	(1.47)	(1.63)	(1.86)
Industry3	1.010 ***	18.410 ***	19.420 ***	0.0223 ***	0.1779 ***	0.2002 ***	1.821 **	11.109 **	12.930 **
	(15.10)	(15.33)	(16.09)	(14.29)	(13.99)	(10.76)	(2.29)	(2.09)	(2.38)
R-squared	0.712	0.712	0.712	0.911	0.911	0.911	0.825	0.825	0.825
$\psi$	0.5841 ***	0.5841 ***	0.5841 ***	0.6612 ***	0.6612 ***	0.6612 ***	0.6021 ***	0.6021 ***	0.6021 ***
	(36.22)	(36.22)	(36.22)	(20.87)	(20.87)	(20.87)	(9.54)	(9.54)	(9.54)
$\theta$	0.0003 ***	0.0003 ***	0.0003 ***	0.0016 ***	0.0016 ***	0.0016 ***	0.0009 ***	0.0009 ***	0.0009 ***
	(12.09)	(12.09)	(12.09)	(38.74)	(38.74)	(38.74)	(18.07)	(18.07)	(18.07)
City FE	YES	YES	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES	YES	YES

Note: ***, **, * are significant at 1%, 5%, 10% levels, respectively, and values in brackets is t-value.

References

1. Wang, Y.F.; Ni, P.F. Economic Growth Spillover and Spatial Optimization of High-speed Railway. China Ind. Econ.; 2016; 2, pp. 21-36.

2. Wang, C.; Wang, G.H.; Zhang, X. Regional differences and structural change: China 1978~2016. J. Manag. World; 2019; 35, pp. 11–26+194.

3. Sheng, S.H.; Xue, X.L.; An, S. Constructing Theoretical System and Discourse System of Mega Infrastructure Construction Management with Chinese Characteristics. J. Manag. World; 2019; 35, pp. 2-16.

4. Zeng, S.X.; Chen, H.Q.; Jin, Z.Z.; Sun, Q.K. The Evolution of Mega project Innovation Ecosystem and Enhancement of Innovative Capability. J. Manag. World; 2019; 35, pp. 28-38.

5. Banister, D.; Thurstain-Goodwin, M. Quantification of the non-transport benefits resulting from rail investment. J. Transp. Geogr.; 2011; 19, pp. 212-223. [DOI: https://dx.doi.org/10.1016/j.jtrangeo.2010.05.001]

6. Shi, Q.Q.; Zhu, J.B.; Sheng, S.H. Study on Profit Distribution of Collaboration in Mega Project Supply Chain. Chin. J. Manag. Sci.; 2017; 25, pp. 42-51.

7. Sheng, S.H. A study on some issues of major project management in China. Proceedings of the NSF 79th Issue of the Shuangqing Forum; Zhenjiang, China, 6 September 2012.

8. Qin, Y. ‘No county left behind?’ The distributional impact of high-speed rail upgrades in China. J. Econ. Geogr.; 2017; 17, pp. 489-520. [DOI: https://dx.doi.org/10.1093/jeg/lbw013]

9. Qin, Y.; Zhang, X. The road to specialization in agricultural production: Evidence from rural China. World Dev.; 2016; 77, pp. 1-16. [DOI: https://dx.doi.org/10.1016/j.worlddev.2015.08.007]

10. Faber, B. Trade integration, market size, and industrialization: Evidence from China’s National Trunk Highway System. Rev. Econ. Stud.; 2014; 81, pp. 1046-1070. [DOI: https://dx.doi.org/10.1093/restud/rdu010]

11. Engwall, M. No project is an island: Linking projects to history and context. Res. Policy; 2003; 32, pp. 789-808. [DOI: https://dx.doi.org/10.1016/S0048-7333(02)00088-4]

12. Efthymiou, D.; Antoniou, C. How do transport infrastructure and policies affect house prices and rents? Evidence from Athens, Greece. Transp. Res. Part A Policy Pract.; 2013; 52, pp. 1-22. [DOI: https://dx.doi.org/10.1016/j.tra.2013.04.002]

13. Li, S.K.; Zhu, W.P.; Liu, D. Can High-speed Railway Promote Industrial Structure Upgrading: Based on Resource Reconfiguration Perspective. South China J. Econ.; 2022; 2, pp. 56-72.

14. Luo, N.S.; Xiao, N.F.; Li, J.M. On Whether the High-speed Railway Can Promote the Industrial Structure to Upgrade and Optimize: An Analysis Based on Quasi-natural Experiments. J. Manag.; 2022; 33, pp. 38-49.

15. Lin, Y. Travel Costs and Urban Specialization Patterns: Evidence from China’s High Speed Railway System. J. Urban Econ.; 2017; 98, pp. 98-123. [DOI: https://dx.doi.org/10.1016/j.jue.2016.11.002]

16. Dong, X.; Zheng, S.; Kahn, M.E. The role of transportation speed in facilitating high skilled teamwork across cities. J. Urban Econ.; 2020; 115, 103212. [DOI: https://dx.doi.org/10.1016/j.jue.2019.103212]

17. Acemoglu, D.; Jacob, M.; James, A. Robinson state capacity and American technology: Evidence from the Nineteenth Century. Am. Econ. Rev. Pap. Proc.; 2016; 106, pp. 61-67. [DOI: https://dx.doi.org/10.1257/aer.p20161071]

18. Li, T. Research on Technological Innovation Mechanism of High-speed Rail Industry and Technology Spillover Effect-Based on “Diamond Model”. Master’s Thesis; Nanjing University: Nanjing, China, 2014.

19. Yang, S.Y.; Li, Z. Evolution of Regional Innovation Patterns and Their Functional Mechanisms under Space-time Compression. South China J. Econ.; 2022; 5, pp. 49-64.

20. Long, Y.; Zhao, H.L.; Zhang, X.D.; Li, Y. High-speed Railway and Venture Capital Investment. Econ. Res. J.; 2017; 52, pp. 195-208.

21. Berliant, M.; Reed, R.R.; Wang, P. Knowledge exchange, matching, and agglomeration. J. Urban Econ.; 2020; 60, pp. 69-95. [DOI: https://dx.doi.org/10.1016/j.jue.2006.01.004]

22. Hong, Y.F. Impacts of High-Speed Rail Stations Opening on the Regional Innovation Level—Analysis Based on Scoring Tendency Matching Model. Mod. Econ.; 2020; 11, pp. 383-398.

23. Guo, Y. The Impact of Transport Infrastructure on Innovation Output-Based on Provincial Road Panel Data Model in China from 1997–2012. Sci. Technol. Econ.; 2015; 3, pp. 36-39.

24. Xuegang, C.; Chuanglin, F.; Zhenbo, W.; Chao, B. Spatial relationship of high-speed transportation construction and land-use efficiency and its mechanism: Case study of Shandong Peninsula urban agglomeration. J. Geogr. Sci.; 2019; 29, pp. 549-562.

25. Tan, K.H. Innovation spillover and regional spatial optimization under the influence of high-speed railway. China Ind. Econ.; 2018; 2, pp. 21-36.

26. Zhang, S.D.; Hu, L.N.; Zhou, L.Q. Research on the Measurement and Influence of Talent Gather in the Center Cities of Metropolitan Areas. Sci. Technol. Prog. Policy; 2019; 36, pp. 54-61.

27. Han, L.; Chen, S.; Liang, L.L. Digital economy, innovation environment and urban innovation capabilities. Sci. Res. Manag.; 2021; 42, pp. 35-45.

28. Kou, Z.L.; Liu, X.Y. FIND Report on City and Industrial Innovation in China (2017); Fudan Institute of Industrial Development, Fudan University: Shanghai, China, 2017.

29. Li, J.K.; Zhong, C.B. High-speed Rail Opened, Urban Accessibility and Employment. Soft Sci.; 2020; 34, pp. 21-26.

30. Makkonen, T.; Merisalo, M.; Inkinen, T. Containers, facilitators, innovators? The role of cities and city employees in innovation activities. Eur. Urban Reg. Stud.; 2017; 25, pp. 106-118. [DOI: https://dx.doi.org/10.1177/0969776417691565]

31. Hu, J.T. Research on the Spatial Effect of Innovation Agglomeration. Ph.D. Thesis; China University of Geosciences: Wuhan, China, 2018.

32. Shwuing, W. The Influence of Innovation Strategy and Organizational Innovation on Innovation Quality and Performance. Int. J. Organ. Innov.; 2019; 3, pp. 45-66.

33. Han, J.K.; Kim, N.; Srivastava, R.K. Market Orientation and Organizational Performance: Is Innovation a Missing Link?. J. Mark.; 1998; 62, pp. 30-45. [DOI: https://dx.doi.org/10.1177/002224299806200403]

34. Fei, F.; Zhang, X. Transformation effect of resource-based cities based on PSM-DID model: An empirical analysis from China. Environ. Impact Assess. Rev.; 2021; 91, 106648.

35. LU, J. The performance of performance-based contracting in human services: A quasi-experiment. J. Public Adm. Res. Theory; 2015; 26, pp. 277-293. [DOI: https://dx.doi.org/10.1093/jopart/muv002]

36. Kamps, C. The dynamic effects of public capital: VAR evidence for 22 OECD countries. Int. Tax Public Financ.; 2005; 12, pp. 533-558. [DOI: https://dx.doi.org/10.1007/s10797-005-1780-1]

37. Melo, P.C.; Graham, D.J.; Brage-Ardao, R. The productivity of transport infrastructure investment: A meta-analysis of empirical evidence. Reg. Sci. Urban Econ.; 2013; 43, pp. 695-706. [DOI: https://dx.doi.org/10.1016/j.regsciurbeco.2013.05.002]