Geographical Accessibility and Corporate Technological Innovation—Evidence from a Quasi-Natural Experiment

Abstract

Geographic accessibility is an important determinant of the quality of interactions between firms and stakeholders and has a significant impact on the technological innovation of enterprises. By using a quasi-natural experiment implemented with China’s high-speed rail service and designing a high-speed rail network centrality indicator using social network analysis, we examine the impact of geographic accessibility on corporate technological innovation. The results show that geographic accessibility significantly promotes the technological innovation of enterprises, especially for enterprise exploratory innovation. Mechanism analysis indicates that geographic accessibility promotes enterprise technological innovation by reducing financing constraints and increasing technicians’ mobility. Cross-sectional analysis reveals that the prompting effect is more obvious in high-tech firms and firms in less-developed regions. This study enriches the research on geographic accessibility and corporate technological innovation, and has significant implications for enhancing the core competitiveness and sustainable development of enterprises.

1. Introduction

Geographic location affects the economic decisions of individuals and entities. Investment managers have a greater investment preference for local firms due to their greater informational advantage [1]. There are also geographic peer effects in surplus management earnings forecasting by firms, and this effect increases as the firm’s exposure to local institutional investors increases [2]. With the development of modern science and technology, accessibility between regions has been greatly improved, and geographic accessibility still plays an important role.

High-speed rail (HSR) has greatly improved regional accessibility and shortened the spatial and temporal distances between regions, thus reducing transaction and communication costs in different cities [3,4], facilitating the flow and integration of capital, information and knowledge, and playing an important role in optimizing resource allocation. Technological innovation is an important source of a country’s long-term economic growth and of sustainable competitive advantage for firms [5]. Technological innovation can foster the emergence of new industries and new business models, create new market demand and employment opportunities, bring new momentum for economic growth, and enhance economic competitiveness and sustainability [6]. At the same time, technological innovation can also promote resource utilization efficiency and ensure sustainable economic development [7]. Therefore, determining whether the information and resource advantages brought about by improved geographic accessibility can motivate enterprises to carry out technological innovation activities is crucial to the promotion of sustainable socio-economic development.

The existing literature has examined the impacts of transportation infrastructure at both the regional and microeconomic levels. At the regional economic level, scholars have primarily studied the growth effects of HSR [8,9,10,11], the siphoning effect [12], and urban specialization patterns [3,13]. At the microeconomic level, scholars have focused on the spatial compression, time savings, and increased accessibility brought about by HSR, which have thus affected venture capital investment [14], compensation policies [15], and board oversight and advisory services [16], and have led to the underpricing of IPOs [17]. Furthermore, they have affected relationship banking [18], the cost of corporate debt [19], analyst forecasts [20,21,22], innovation [23], trade credit [24], tax avoidance [25], reductions in corporate fraud incidents [26], improvements in firm performance [27], and increases in share price efficiency [28]. These studies expand the economic consequences brought about by geographic proximity. Although some scholars have observed a relationship between geographic proximity and innovation [23], they have overlooked whether the different degrees of geographic accessibility may lead to different outcomes. In other words, areas with high geographical accessibility differ from those with low accessibility in terms of factors such as foot traffic, the degree of consumption stimulation, and the amount of capital inflow. These differences are an important factor influencing the aggregation and flow of regional resources. Therefore, using only dummy variables will not adequately capture these differences, leading to limitations in the research. Additionally, existing research has overlooked the significant differences in the demand for information and resources among exploratory innovation and exploitative innovation [5,23]. Moreover, resource distribution is highly regional, raising questions about how geographic accessibility in developed versus less developed regions affects technological innovation and ambidextrous innovation. In addition, what are the different impacts on high-tech and non-high-tech firms?

HSR in China provides an ideal environment in which to study the impact of geographic accessibility on the technological innovation of enterprises. First, with the advantages of safety, speed and convenience [24] can greatly shorten spatial and temporal distance [4], improve the accessibility between regions [3], and further strengthen communication between enterprises in different cities. Therefore, the implementation of HSR services in China can allow us to observe changes in proximity [26]. Second, HSR projects in China are initiated by the government [21], so the emergence of HSR in China causes drastic changes to proximity, providing us with the opportunity to carry out a quasi-natural experiment for a DID analysis exploring the impact of proximity on firm innovation. Furthermore, innovation is the driving force behind a company’s sustainable competitive advantage [29] and national economic growth [30,31]. Although China is the second-largest and fastest-growing major economy, Chinese companies still lag behind those in developed countries in terms of innovation [32]. Given that many developing countries face the same issue, understanding the driving factors of technological innovation in Chinese companies can provide valuable insights for the technological innovation of companies in other developing countries. In recent years, China’s high-speed rail has developed rapidly, with a total length of 36,000 km in operation [4], and the “Eight Vertical and Eight Horizontal” HSR construction is 80% complete [26]; the latter can achieve a 1–4 h traffic circle between neighboring large and medium-sized cities, and a 0.5–2 h traffic circle within urban agglomerations. China’s “Eight Vertical and Eight Horizontal” HSR structure, along with surrounding branch lines, has formed a huge HSR network. Therefore, we apply social network analysis methods to the HSR network so as to construct a HSR network centrality index describing different degrees of geographic accessibility.

Therefore, this study analyzes data from Chinese A-share-listed companies from 2009 to 2019, using social network analysis to examine the impact of geographic accessibility on corporate technological innovations and ambidextrous innovation. The empirical results show that the HSR network can significantly promote corporate technological innovation. The HSR network plays the role of an “information highway”, which can overcome the information barriers caused by geographical distance, promote the flow of information, capital and talents, and thus promote the technological innovation of firms. The results also show that the impact of the HSR network on technological innovation is mainly reflected in exploratory innovation. Compared with exploitative innovation, exploratory innovation requires more innovation resources, and it is more sensitive to the innovation resources introduced by the HSR network. Furthermore, mechanism analysis indicates that the HSR network promotes technological innovation by aiding in the alleviation of financing constraints and promoting the mobility of technicians. Financing constraints and technicians’ mobility play an intermediary role in the impact of the HSR network on technological innovation and exploratory innovation. Additionally, cross-sectional analysis reveals that compared with non-high-tech firms and developed regions, the HSR network can promote the technological innovation of high-tech firms and firms in less-developed regions.

Our study contributes to the existing literature. First, it contributes to the research on the economic consequences of geographic accessibility. Previous studies have only used dummy variables to examine the economic consequences of HSR [4,26], neglecting the impact of the different degrees of geographic accessibility, which may lead to different outcomes. Using only dummy variables will not allow us to adequately capture the differences in the aggregation and flow of regional resources between high geographical accessibility and low geographical accessibility, which is what may have led to limitations in previous studies. We combine social network analysis with new economic geography theory to construct an index of HSR network centrality, which provides a more comprehensive and reliable measure of geographic accessibility, and helps to enrich the study of geographic accessibility. Second, this study expands the research on the mechanism of geographic accessibility on technological innovation. The mechanism in previous studies only examined the influence of whether HSR is opened or not [3,22]. However, in this study, by constructing HSR network indicators, we reveal the roles of financing constraints and technician mobility in the impact of geographic accessibility on technological innovation, extend the scope of application of the mechanism from dummy variables to the HSR network, and deepen research on the effects of the mechanism of geographic accessibility on technological innovation. Third, our study enriches the research on the determinants of ambidextrous technological innovation. Based on ambidextrous innovation theory [33,34], exploratory innovations and exploitative innovations differ significantly in terms of knowledge bases, risk, and competitive advantage [35,36,37]. Little empirical evidence has been provided on the role of geographic accessibility in ambidextrous innovation research. Exploratory innovation is characterized by high risk, high investment, and high return, while exploitative innovation is characterized by low risk and high predictability [35,37], implying that exploratory and exploitative innovations are not affected in the same way when faced with the resources generated by the HSR network. We analyze the differential impact of the HSR network on exploratory innovation and exploitative innovation, and provide a piece of evidence suggesting the important role of geographic accessibility in promoting exploratory innovation.

The paper proceeds as follows. Section 2 develops the hypotheses. Section 3 describes the empirical design. Section 4 reports the empirical results. Section 5 contains additional analyses, and Section 6 contains the conclusions.

2. Development of Hypotheses

2.1. HSR Network and Technological Innovation

Firm innovation involves risks and uncertainties, and the innovative behaviors of firms can only be sustained effectively with the support of sufficient innovative elements. First of all, technological innovation requires strong financial support [38], and the HSR network can promote the smooth inflow of funds. The HSR network acts as an “information highway” for firms [22], helping reduce the information communication barriers between regions [3] and the information asymmetry between investment and financing parties, which in turn can alleviate the financing constraints of firms. A lack of information and financing constraints are the major problems faced by firms in technological innovation [39]. The increase in HSR network centrality improves the spatial accessibility of cities along the route, and allows the cities along the route to become more open, which reduces the degree of information asymmetry within and outside the enterprise, and then encourages the external capital to flow into the cities along the HSR route more smoothly [26].

Secondly, innovation activities rely on outstanding innovative personnel, and HSR networks can help firms attract more technical talent. HSR networks optimize the existing transportation system, making urban transit more convenient and improving infrastructure [40], which effectively facilitates talent aggregation and serves as an important carrier and medium for talent mobility. At the same time, due to the existence of threshold effects, it has restricted workers engaged in ordinary and repetitive jobs from entering the city to some extent [41].

Moreover, talent mobility is often accompanied by technology flow. A significant amount of knowledge closely related to innovation is tacit knowledge that cannot be easily obtained through advanced network technologies; it typically requires face-to-face interaction to acquire [42]. The HSR network significantly reduces travel time between regions and firms, lowers the social costs of transportation along the routes, and thus fosters closer business interactions among companies. As a result, face-to-face communication becomes more frequent, and information exchange and technological diffusion occur more rapidly [5], ultimately accelerating the flow and generation of new knowledge.

Finally, HSR networks can accelerate the formation of regional innovation systems and stimulate innovation investment behavior. The knowledge spillovers, technological connections, and diffusion of innovation activities brought about by HSR networks can promote the spatial agglomeration of enterprises [11]. This regional concentration of firms further accelerates technology diffusion, facilitates the formation of regional innovation networks, and enhances informal exchanges among employees within the region [3], thereby helping enterprises to better develop technological innovation. Based on the analysis above, we propose hypothesis 1:

H1. 

The high-speed rail network can promote the technological innovation of enterprises.

2.2. HSR Network and Ambidextrous Innovation

Exploitative innovation and exploratory innovation are two different types of technological innovation in enterprises [33,34]. Exploratory innovation is the innovation behavior of firms based on new knowledge and technology in a new target market, which is a more radical innovation strategy with high innovation intensity, high investment and high risk. On the other hand, exploitative innovation is a more moderate and gradual method of innovation adopted by firms on the basis of their original knowledge and technology, and their products may be new products in the original product portfolio of the firm, but they are not new products for the market [35,36]. Due to the differences in information disclosure, the demand for capital and the heterogeneous knowledge between different types of technological innovations [37], there are bound to be differences in the impacts of the HSR network on the two types of technological innovations. There is a competitive relationship between exploratory and exploitative innovation [43], and the two will compete with each other for innovation resources. Therefore, although the HSR network can introduce resources needed for corporate innovation, exploratory innovation and exploitative innovation will compete for scarce resources due to the limited availability of resources. Compared with exploitative innovation, exploratory innovation demands greater resources support.

First, exploratory innovation is subject to stronger financing constraints. Due to the high uncertainty and risk, exploratory innovation does not make it easy to obtain support in the form of external funds and comes with more serious financing constraints [44]. Hottenrott and Czarnitzki (2011) found that there was a negative correlation between exploratory innovation and firm credit rating, while there was no significant relationship between exploitative innovation and enterprise credit rating [45]. This suggests that external financing is more important for firms’ exploratory innovations than for their exploitative innovations. Financing generated by the HSR network for firms is prioritized to be used for exploratory innovations, which are more sensitive to finances.

Second, exploratory innovation relies more heavily on heterogeneous knowledge, placing higher demands on innovators. Travel time affects the extent and effectiveness of knowledge exchange and collaboration [5]. HSR reduces travel time, prompting easier communication between firms and technical personnel, along with the exchange of knowledge and technology [3]. Through interactions and exchanges with the technical staff of firms, it is more likely to stimulate the generation of new knowledge and new technologies. Therefore, compared with development innovation, the HSR network is more conducive to firms carrying out exploratory innovation. As a result, we propose hypothesis 2:

H2. 

The HSR network is more conducive to exploratory innovation than exploitative innovation.

3. Empirical Design

3.1. Sample

We used companies listed on the Shanghai and Shenzhen Stock Exchange from 2009 to 2019 as an initial sample. The sample period ends in 2019 to avoid the effects of the change in the macroeconomic environment. Information on HSR stations and the opening day of each route is obtained from the website of the Nation Railway Administration of China (www.12306.cn). We obtain the financial data of samples from the China Securities Market and Accounting Research Database (CSMAR). We exclude the following samples: (1) financial and insurance listed companies; (2) ST (special treatment) and other T-class listed companies; (3) samples with missing data. Our final sample includes 10,931 observations.

3.2. Technological Innovation

In our main test, we measure technological innovation via R&D investment divided by average total assets [46]. According to China Accounting Standards for Business Enterprises No.6—Intangible Assets (2006) [47], research and development projects within enterprises can be divided into the research stage and development stage. The research stage is more inclined toward exploratory expenditure, which has higher risk and uncertainty than the development stage. Therefore, we measure exploratory innovation by scaling expenditure to average total assets in the research stage, and we measure exploitative innovation by scaling expenditure to average total assets in the development stage [48].

3.3. HSR Network Centrality

We designate firms’ connection to HSR as an exogenous shock for firms’ geographical accessibility. Specifically, HSR services do not change the real spatial distances, but compress the time distance between firms and their stakeholders, and therefore improve the geographical accessibility of firms. Based on this, we calculate HSR network centrality to measure geographical accessibility in firms’ locations.

We focus on HSR network degree centrality in our main test [49,50]. The degree centrality is the sum of the number of HSR lines associated with a particular station, representing the accessibility of the station. A higher degree centrality indicates that the station is more centrally located within the entire HSR network.

The calculation process is as follows: First, we identify the data for HSR lines and stations that are in service from the website of the Nation Railway Administration of China (www.12306.cn: accessed on 1 March 2022). For rigor, when determining the opening dates of the HSR, lines opened between January and June of the current year are assigned to that year, while those opened between July and December are assigned to the following year. Second, we organize the extracted relevant information into matrix form and use Pajek 5.14 to calculate the network centrality for each station in each year. Third, we assign each station to its corresponding city and calculate the total network centrality of HSR stations within each city to obtain the city-level HSR network centrality. And forth, by matching the location of firms with each city, we obtain the HSR network centrality at the firm level. Last, for statistical regression purposes, the calculated centrality values are log-transformed to obtain the final indicators (C_D). The formula for calculating in Pajek is as follows:

Degree centrality (C_d):

$${\mathrm{C}}_{\mathrm{d}}=\sum \mathrm{Xij}\left(\mathrm{i}\ne \mathrm{j}\right)$$

(1)

where (i) represents a certain station; (j) represents other stations different from (i) in that year. (X_ij) denotes a network connection, which equals 1 if station (i) and station (j) share at least one HSR line; otherwise, it equals 0.

In addition, in the robustness test, we use the HSR network’s betweenness centrality to measure the geographic accessibility. The HSR network’s betweenness centrality refers to the ability of city A with an HSR service to be located in the “middle” of or in an “intermediary” position relative to multiple other cities with HSR services, and measures city A’s control over the HSR element. The formula for calculating betweenness is as follows:

Betweenness centrality (C_b):

$${\mathrm{C}}_{\mathrm{b}}={\displaystyle \frac{2{\sum }_{\mathrm{j}}^{\mathrm{m}}{\sum }_{\mathrm{k}}^{\mathrm{m}}{\mathrm{g}}_{\mathrm{jk}}\left({\mathrm{n}}_{\mathrm{i}}\right)}{{\mathrm{g}}_{\mathrm{jk}}\left(\mathrm{n}-1\right)\left(\mathrm{n}-2\right)}}$$

(2)

where g_jk stands for the number of sites through which site j connects with site k, and g_jk(n_i) represents the number of paths that include site i from site j to site k. The calculated centrality values are log-transformed to obtain the final indicators (C_B).

3.4. Control Variables

We follow prior innovation research [51], and control for the following variables: Lev (total liabilities/total assets), Roa (ebit/average total assets), Size (the natural logarithm of total assets), Age (number of years since a firm was listed), Cash (total cash and cash equivalents/assets), Tobins’Q (market capitalization/total assets), Mkt (marketization index of China’s provinces (2018)), Divid (dividends/earnings per share), and Inctrl (internal control index issued by Shenzhen Dibo Company/1000). Finally, we include industry and year fixed effects, controlling for characteristics that are invariant within the industry and year. All variables are defined in

Table 1. Definitions of variable.

Variable	Definition
Innovation	Innovation investment/average total assets
Exploration	Research stage innovation investment/average total assets
Exploitation	Innovation investment in development phase/average total assets
CD	Ln(Cd + 1)
CB	Ln(Cb + 1)
Lev	Total liabilities/total assets
Roa	Ebit/average total assets
Size	The natural logarithm of total assets
Age	Number of years since a firm was listed
Cash	Total cash and cash equivalents/assets
Tobins’Q	Market capitalization/total assets
Mkt	Marketization index of China’s provinces (2018)
Divid	Dividends/earnings per share
Inctrl	Internal control index issued by Shenzhen Dibo Company/1000
Ind	Industry dummy variable
Year	Annual dummy variable

.

3.5. Statistical Model

To examine the relation between technological innovation and HSR network centrality (H1) and which type of technological innovation is more influenced (H2), we estimate the following models:

Innovation = α₀ + α₁ ∗ C_D + α₂ ∗ Lev + α₃ ∗ Roa + α₄ ∗ Size + α₅ ∗ Age + α₆ ∗ Cash +
α₇ ∗ Tobin’Q + α₈ ∗ Mkt + α₉ ∗ Divid + α₁₀ ∗ Inctrl + ∑Ind + ∑Year + ε₁

(3)

Explotation/Exploitation = β₀ + β₁ ∗ C_D + β₂ ∗ Lev + β₃ ∗ Roa + β₄ ∗ Size + β₅ ∗ Age +
β₆ ∗ Cash + β₇ ∗ Tobin’Q + β₈ ∗ Mkt + β₉ ∗ Divid + β₁₀ ∗ Inctrl + ∑Ind + ∑Year + ε₂

(4)

where, in Model (3), we take HSR network degree centrality (C_D) as the independent variable, and enterprise technological innovation as the dependent variable, while α₁ reflects the impact of HSR network degree centrality on firm technological innovation. If the results show that α₁ is significant and positive, H1 is supported. Model (4) tests the impact of HSR network degree centrality (C_D) on exploratory and exploitation innovation. We conducted an inter-group coefficient difference test to verify whether H2 is supported.

4. Descriptive Statistics and Empirical Results

4.1. Descriptive Statistics

Table 2. Descriptive statistics of variables.

Variable	N	Mean	Std. Dev.	Median	Min	Max
Innovation	10,931	0.024	0.019	0.020	0	0.105
Exploration	10,931	0.020	0.017	0.017	0	0.089
Exploitation	10,931	0.003	0.007	0	0	0.042
CD	10,931	1.800	1.019	2.079	0	3.761
Lev	10,931	0.358	0.181	0.345	0.047	0.825
Roa	10,931	0.150	0.213	0.074	0.005	1.142
Size	10,931	21.811	1.025	21.701	19.918	25.381
Age	10,931	7.836	5.786	6	1	25
Cash	10,931	1.089	1.801	0.459	0.035	11.499
Tobins	10,931	2.453	1.82	1.902	0.353	10.159
Mkt	10,931	8.478	1.577	9.02	3.49	10.173
Divid	10,931	0.314	0.330	0.239	0	2.072
Inctrl	10,931	0.658	0.093	0.674	0	0.799

presents the descriptive statistics. All continuous variables were winsorized at the 1st and 99th percentiles to reduce the influence of outliers. The mean value of technological innovation is 0.024, higher than the median value of 0.020, indicating that the technological innovation of most firms is below that of the overall market, and that the technological innovation of Chinese firms still needs to be improved. The maximum value of technological innovation is 0.105, and the minimum value is 0, indicating that there are great differences in technological innovation among firms. The mean value of the exploration innovation of listed companies is 0.020, higher than that of exploitation innovation, which is 0.003. The mean value of exploratory innovation and exploitative innovation is greater than the median, indicating that the exploratory and exploitative innovation levels of more than half of the listed companies in China are below those of the overall market. The mean and median of HSR network centrality are 1.800 and 2.076.

4.2. Empirical Results: HSR Network Centrality and Technological Innovation

Table 3. Effect of HSR network centrality on technological innovation.

Variable	(1)	(2)	(3)
	Innovation	Exploration	Exploitation
CD	0.001 ***	0.001 ***	0.000
	(7.55)	(7.97)	(1.21)
Lev	0.000	−0.000	−0.000
	(0.10)	(−0.37)	(−0.13)
Roa	0.015 ***	0.017 ***	−0.003 ***
	(10.70)	(13.55)	(−4.06)
Size	0.000	−0.001 ***	0.001 ***
	(0.14)	(−3.70)	(7.20)
Age	−0.000 ***	−0.000 ***	−0.000
	(−4.84)	(−5.08)	(−0.83)
Cash	−0.000	−0.000 ***	0.000 **
	(−1.26)	(−3.39)	(1.96)
Tobins’Q	0.002 ***	0.002 ***	0.000 ***
	(20.37)	(18.31)	(7.05)
Mkt	0.001 ***	0.002 ***	−0.000 ***
	(12.17)	(16.81)	(−6.42)
Divide	0.001	0.001 **	−0.000 **
	(1.11)	(2.01)	(−2.26)
Inctrl	0.009 ***	0.009 ***	0.001
	(5.53)	(5.56)	(0.86)
Constant	−0.014 ***	0.000	−0.013 ***
	(−2.92)	(0.06)	(−5.93)
Ind	YES	YES	YES
Year	YES	YES	YES
N	10,931	10,931	10,931
F	122.7	115.1	31.63
Adj. R2	0.292	0.279	0.0940

Note. ** p < 0.05, *** p < 0.01.

reports the results for H1 and H2, examining the impact of HSR network centrality on technological innovation and ambidextrous innovation. Our dependent variables are technological innovation in Column (1), and exploration and exploitation in Columns (2) and (3).

The results show that the coefficient of HSR network degree centrality (C_D) on technological innovation is positive and significant (p-value < 0.01 in Column (1)), indicating that the HSR network can promote firm technological innovation, thereby supporting H1. The coefficient of HSR network degree centrality (C_D) on exploration innovation is positive and significant (p-value < 0.01 in Columns (2)), while the coefficient of HSR network degree centrality (C_D) on exploitation innovation is not significantly different from zero (see Columns (3)). These finding shows that the HSR network can promote exploratory innovation rather than exploitative innovation. Overall, the results suggest that a firm’s technological innovation benefits from its superior geographical proximity, and that the benefits derived from geographical proximity most heavily impact exploratory innovation.

4.3. Robustness Tests

In this section, we conduct several robustness checks to ensure the validity of our inferences. First, we replicate our analysis using alternative proxies for HSR network centrality and technological innovations. Then, we remove the samples in central cities. Finally, we use propensity score matching (PSM) to reduce the systematic bias of firms across regions.

4.3.1. Alternative Measures of HSR Network Centrality

First, for the definition of HSR network centrality, we use HSR network degree centrality in the main regression, and in the robustness test, we re-estimate the main test results using the calculated HSR network betweenness centrality (C_B).

Table 4. Alternative measures of HSR network centrality and technological innovation.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Innovation	Exploration	Exploitation	Innovation2	Exploration2	Exploitation2
CB	0.009 *** (6.65)	0.008 *** (6.57)	0.001 (1.58)
CD				0.003 *** (9.35)	0.003 *** (9.69)	0.000 ** (2.18)
Lev	0.000 (0.33)	−0.000 (−0.13)	−0.000 (−0.09)	−0.021 *** (−7.69)	−0.018 *** (−7.86)	−0.003 *** (−2.69)
Roa	0.015 *** (10.60)	0.017 *** (13.44)	−0.003 *** (−4.06)	−0.057 *** (−18.03)	−0.043 *** (−16.29)	−0.013 *** (−8.43)
Size	0.000 (0.16)	−0.001 *** (−3.67)	0.001 *** (7.19)	0.001 *** (3.18)	−0.000 (−0.64)	0.002 *** (7.23)
Age	−0.000 *** (−4.90)	−0.000 *** (−5.15)	−0.000 (−0.82)	−0.000 *** (−6.17)	−0.000 *** (−8.16)	0.000 (0.18)
Cash	−0.000 (−0.96)	−0.000 *** (−3.07)	0.000 ** (2.01)	0.004 *** (16.01)	0.003 *** (12.78)	0.001 *** (7.70)
Tobins’Q	0.002 *** (20.72)	0.002 *** (18.68)	0.000 *** (7.11)	0.005 *** (19.49)	0.004 *** (18.38)	0.001 *** (7.27)
Mkt	0.001 *** (12.15)	0.002 *** (16.87)	−0.000 *** (−6.51)	0.001 *** (3.93)	0.002 *** (9.59)	−0.001 *** (−7.53)
Divide	0.001 (1.22)	0.001 ** (2.13)	−0.000 ** (−2.24)	−0.003 *** (−2.74)	−0.002 * (−1.93)	−0.001 *** (−2.66)
Inctrl	0.009 *** (5.38)	0.008 *** (5.42)	0.001 (0.81)	0.006 (1.47)	0.005 * (1.67)	−0.001 (−0.38)
Constant	−0.013 *** (−2.80)	0.001 (0.14)	−0.013 *** (−5.86)	−0.028 *** (−2.58)	0.001 (0.14)	−0.026 *** (−5.12)
Ind	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
N	10,931	10,931	10,931	10,931	10,931	10,931
F	122.2	114.3	31.66	174.2	153.0	43.21
Adj. R2	0.291	0.277	0.094	0.370	0.340	0.125

Note. * p < 0.1, ** p < 0.05, *** p < 0.01.

reports the results. We find that the coefficient of HSR network betweenness centrality (C_B) on technological innovation in Column (1) is positive and significant (p-value < 0.01), suggesting that increased geographic access encourages firms to engage in technological innovation. The coefficient of HSR network betweenness centrality (C_B) on exploratory innovation in Column (2) is positive and significant (p-value < 0.01) while the coefficient of HSR network betweenness centrality (C_B) on exploitative innovation in Column (3) is not significantly different from zero, suggesting that the increase in geographic access mainly enhances firms’ exploratory innovation rather than their exploitative innovation. These results suggest that our inferences are robust against other measures of geographic proximity.

4.3.2. Alternative Measures of Corporate Technological Innovation

Second, we use R&D investment divided by sales to measure technological innovation and re-examine the effect of the HSR network centrality on firms’ technological innovation. R&D investments, expenditures in the research stage and expenditures in the development stage divided by sales are denoted by Innovation2, Exploration2 and Exploitation2, respectively. Columns (4), (5) and (6) of Table 4 show the results. We find that the coefficient of HSR network centrality on technological innovation (Innovation2) in Column (4), on exploratory innovation (Exploration2) in Column (5) and on exploitative innovation (Exploitation2) in Column (6) are consistent with our main results, suggesting that our inferences are robust against other measures of corporate technological innovation.

4.3.3. Remove the Samples in Center Cities

As the layout of HSR may be influenced by national strategies and the functions of cities along the line, the development of regional economics may affect the selection of HSR construction sites which could produce selectivity bias and cause biased estimates. Therefore, to address this concern, we remove the sample of sub-provincial cities, other provincial capitals, and municipalities, and use this new dataset to rerun our main tests. The results are shown in

Table 5. Results of removing central cities.

Variable	(1)	(2)	(3)
	Innovation	Exploration	Exploitation
CD	0.001 *** (4.98)	0.001 *** (5.45)	−0.000 (−0.14)
Lev	0.002 (1.06)	−0.001 (−0.57)	0.002 ** (3.25)
Roa	0.020 *** (11.45)	0.022 *** (13.30)	−0.002 * (−2.03)
Size	−0.000 (−1.35)	−0.001 ** (−3.11)	0.000 * (2.51)
Age	−0.000 * (−2.08)	−0.000 (−1.65)	−0.000 (−1.30)
Cash	−0.000 * (−2.11)	−0.000 ** (−2.75)	0.000 (0.59)
Tobins’Q	0.002 *** (13.52)	0.002 *** (11.99)	0.000 *** (3.76)
Mkt	0.001 *** (7.22)	0.001 *** (11.38)	−0.000 *** (−6.88)
Divide	0.000 (0.38)	0.001 (1.85)	−0.001 ** (−2.63)
Inctrl	0.009 *** (4.46)	0.009 *** (4.87)	−0.000 (−0.46)
Constant	−0.007 (−1.17)	−0.000 (−0.02)	−0.004 (−1.61)
Ind	YES	YES	YES
Year	YES	YES	YES
N	5259	5259	5259
F	40.32	47.70	7.60
Adj. R2	0.207	0.237	0.042

Note. * p < 0.1, ** p < 0.05, *** p < 0.01.

. We find similar results to the main regression, indicating that the findings above are robust.

4.3.4. Propensity Score Matching Method (PSM)

To reduce the systematic biases between firms in different cities, we perform PSM between firms in cities with HSR services and cities without HSR services. Specifically, we construct a dataset that includes both the treatment and control groups for each event (the launch of HSR services). We set a dummy variable according to whether or not HSR services are available. We designate firms in the treatment group as those with HSR services and restrict the control group to firms that do not have HSR services. We use firm size (Size), leverage ratio (Lev), return on assets (Roa), firm age (Age), cash holdings (Cash), dividend distribution (Divid), growth capability (Tobins’ Q), internal control (Inctrl), and degree of marketization (Mkt) as matching variables to match the treatment group.

Table 6. Result of PSM.

Variables	Samples	Treat Group	Control Group	Difference	Std. Dev.	T-Value
Innovation	Unmatched	0.0252	0.0183	0.0069	0.000435	15.93
	ATT	0.0252	0.0202	0.0050	0.000568	8.83
Exploration	Unmatched	0.0216	0.0155	0.0061	0.000386	15.85
	ATT	0.0216	0.0178	0.0039	0.000519	7.93
Exploration	Unmatched	0.0034	0.0026	0.0007	0.000173	4.28
	ATT	0.0035	0.0023	0.0011	0.000243	4.52

shows the results after PSM. The table shows that the ATT of firms’ overall technological innovation (Innovation) after matching is 0.0050, the ATT of exploration is 0.0039, and the ATT of exploitation is 0.0011, and all of them are significant (p-value < 0.01), indicating that the HSR network centrality does have a significant impact on the overall technological innovation and ambidextrous innovation of firms. We then perform Logit regression using the matched samples. The results are shown in

Table 7. Effect of HSR network centrality on technological innovation after PSM.

Variables	(1)	(2)	(3)
	Innovation	Exploration	Exploitation
CD	0.005 *** (8.22)	0.004 *** (7.21)	0.001 *** (3.58)
Lev	−0.004 (−1.36)	−0.004 (−1.45)	−0.001 (−0.37)
Roa	0.011 *** (4.82)	0.014 *** (6.17)	−0.003 *** (−4.27)
Size	−0.001 (−1.10)	−0.001 (−1.46)	0 (0.37)
Age	0 (−1.04)	0 (−1.48)	0 (0.61)
Cash	0 (−0.79)	−0.001 * (−2.56)	0.000 * (2.06)
Tobins’Q	0.002 *** (4.9)	0.002 *** (5.25)	0 (−0.58)
Mkt	0.002 *** (7.13)	0.002 *** (8.94)	0 (−1.04)
Divide	−0.001 (−0.76)	0 (−0.32)	−0.001 (−1.25)
Inctrl	0.008 ** (2.87)	0.009 ** (2.99)	−0.001 (−0.56)
Constant	0.011 (0.69)	0.01 (0.7)	0.003 (0.54)
Ind	YES	YES	YES
Year	YES	YES	YES
N	3508	3508	3508
F	30.7	39.95	12.07
Adj. R2	0.143	0.182	0.018

Note. * p < 0.1, ** p < 0.05, *** p < 0.01.

, in which it can be seen that our inferences are still robust.

5. Additional Analyses

5.1. Mechanism Analyses

5.1.1. Financing Constraints

According to resource-based theory, the survival and development of firms depend on external resources. Without sufficient funds, technological innovation cannot be initiated or sustained, but financing constraints are a common phenomenon faced by firms, restricting their innovation activities [52,53,54]. HSR networks can generate resource effects and information effects, reducing the information asymmetry between firms and external investors, alleviating financing constraints, and thereby promoting technological innovation in firms.

Geographical distance affects the information costs among trading agents. Investment institutions and investors can better obtain information about nearby firms through visiting them. Distant firms will face higher information asymmetry due to the high information gathering costs for investors and other stakeholders [55], which leads to fewer investments [1,56]. HSR networks improve the accessibility of firms, greatly shortening the travel time between regions and promoting more face-to-face communication between firms in different cities, improving the timeliness and reliability of information transfer and enabling timely access to “soft information”, thus alleviating information asymmetry between investors and firms.

For external investors, HSR networks can reduce the information acquisition costs caused by long distances [26], increase the information acquisition efficiency of external investors, and enable investors to better assess the feasibility and profitability of firms’ projects, which allows firm to obtain investment funds from high-quality innovation projects, therefore so alleviate the financing constraints they face. For analysts, HSR networks enable analysts to not only gain an information advantage for local stocks, but also to obtain real and effective “soft information” about distant companies through field inspections [22], thereby improving their predictive accuracy [20]. This provides references for investors and other stakeholders, alleviating information asymmetry in firms and helping to reduce information acquisition costs, ultimately easing the financing constraints of firms. In terms of banks, an HSR network can facilitate continuous and effective communication between banks and firms, aid in the transmission of non-public information on firms, and enhance the awareness and trust of firms, thus helping to alleviate firms’ financing difficulties.

To summarize, an HSR network can alleviate firms’ financing constraints and then facilitate the smooth development of technological innovation. Exploratory innovations come with more severe financing constraints, so we expect the mediating effect of financing constraints to also be reflected in exploratory innovations.

We use FC to represent the corporate finance constraint. Specifically, we first compute the SA index [57]. The SA index is calculated based on the size and age of the company, which can avoid endogenous issues caused by financial indicators. The SA values are negative, and the higher the absolute value, the more severe the financing constraints. To make the empirical results more intuitive, we then take the absolute value of the SA index as FC, with a higher value indicating more severe financing constraints [54]. The calculation of SA is as follows:

$$\mathrm{SA}=-0.737\ast \mathrm{Size}+0.043\ast {\mathrm{Size}}^2-0.040\ast \mathrm{Age}$$

(5)

Since the impact of HSR networks on exploitative innovation is not significant, we test the role of financing constraints in the impact of HSR networks on technological innovation and exploratory innovation. Columns (1)–(3) in

Table 8. Mechanism tests on HSR network and technological innovation/exploratory innovation.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	FC	Innovation	Exploration	Rdpersonratio	Innovation	Exploration
CD	−0.002 ***	0.001 ***	0.001 ***	1.061 ***	0.001 ***	0.001 ***
	(−3.13)	(7.49)	(7.86)	(7.56)	(4.69)	(5.13)
FC		−0.005 **	−0.008 ***
		(−2.01)	(−3.55)
Rdpersonratio					0.001 ***	0.0005 ***
					(37.10)	(32.07)
Lev	−0.054 ***	−0.000	−0.001	−4.692 ***	0.004 ***	0.002 *
	(−11.17)	(−0.12)	(−0.75)	(−4.45)	(2.71)	(1.77)
Roa	0.034 ***	0.015 ***	0.018 ***	−9.621 ***	0.020 ***	0.021 ***
	(6.04)	(10.79)	(13.74)	(−9.36)	(14.39)	(16.02)
Size	0.019 ***	0.000	−0.001 ***	0.157	−0.000	−0.001 ***
	(22.62)	(0.56)	(−2.87)	(0.89)	(−1.03)	(−4.43)
Age	0.040 ***	0.000	0.000 *	−0.141 ***	0.000	−0.000 *
	(318.34)	(0.40)	(1.79)	(−5.36)	(0.36)	(−1.91)
Cash	−0.006 ***	−0.000	−0.000 ***	0.610 ***	−0.001 ***	−0.001 ***
	(−13.89)	(−1.52)	(−3.83)	(4.82)	(−3.85)	(−4.62)
Tobins’Q	−0.009 ***	0.002 ***	0.002 ***	0.923 ***	0.002 ***	0.001 ***
	(−20.95)	(19.58)	(17.27)	(9.79)	(14.25)	(12.67)
Mkt	0.002 ***	0.001 ***	0.002 ***	0.472 ***	0.001 ***	0.001 ***
	(3.56)	(12.23)	(16.93)	(5.18)	(7.30)	(10.85)
Divide	−0.001	0.001	0.001 **	−1.892 ***	0.002 ***	0.002 ***
	(−0.56)	(1.10)	(1.99)	(−4.69)	(4.24)	(4.30)
Inctrl	−0.027 ***	0.009 ***	0.008 ***	2.508 *	0.007 ***	0.007 ***
	(−4.03)	(5.45)	(5.42)	(1.76)	(3.79)	(4.18)
Constant	2.728 ***	−0.001	0.021 ***	1.596	−0.013 **	0.001
	(144.13)	(−0.08)	(2.90)	(0.37)	(−2.26)	(0.22)
Ind	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
N	10,931	10,931	10,931	6951	6951	6951
F	4398	119.6	112.5	110.5	137.1	123.1
Adj. R2	0.937	0.292	0.279	0.342	0.400	0.374

Note. Due to technician data disclosure limitations, the sample size for technician mobility is reduced. * p < 0.1, ** p < 0.05, *** p < 0.01.

show the results of the mediating effect of financing constraints. The coefficient of financing constraints (FCs) on HSR network centrality (C_D) in Column (1) is −0.002 and is significant (p-value < 0.01), indicating that the HSR network has a mitigating effect on firms’ financing constraints. Furthermore, the coefficient of technological innovation (Innovation) on financing constraints (FCs) in Column (2) is significant and negative (p-value < 0.05), indicating that the alleviation of financing constraints promotes firms’ technological innovations. Therefore, we conclude that the HSR network is able to promote firms’ technological innovations by alleviating financing constraints. Similarly, the coefficient of exploratory innovation (Exploration) on financing constraints (FCs) in Column (3) is significant and negative (p-value < 0.01), indicating that the alleviation of financing constraints promotes exploratory innovation by firms, showing that financing constraints also play a mediating role in the impact of the HSR network on exploratory innovation.

5.1.2. Technicians’ Mobility

Excellent technical personnel are indispensable to the innovation activities of enterprises, and the HSR network can attract more senior talents for firms. The HSR network optimizes the original transportation system and becomes an important carrier and medium for technician mobility, thereby promoting technological innovation.

Firstly, the HSR network, with its advantages of high speed, safety, comfort, and high punctuality, increases regional accessibility and shortens the time–space distance [58], which allows for the more frequent movement of time-sensitive laborers, especially high-skilled laborers [3], and enables face-to-face communication between the high-skilled laborers in firms in different regions, thus facilitating the generation of new knowledge [5]. At the same time, the HSR network can increase accessibility for individuals searching for jobs [58], and areas with HSR services are often accompanied by good infrastructure and services, which contributes to a favorable employment environment, attracting specialists to flow into the region [3]. Due to the existence of a threshold effect which restricts the entry of ordinary workers, specialists are more likely to move to HSR cities.

Secondly, HSR networks as well as other infrastructure and services promote regional economic development [13], further expanding economic growth through the investment multiplier effect, which will attract more firms to settle in the region, gradually increasing market size. As competition among firms intensifies, cities will require more support in the form of human capital [58]. Only through innovative products or services can firms stand out among their competitors; therefore, areas with HSR services will further increase the demand for specialists, attract the flow of technical personnel, and then promote technological innovation in firms.

We use the ratio of the technical personnel of listed firms (Rdpersonratio) to measure technician mobility. Columns (4)–(6) in Table 8 show the results of the mediating effect of technician mobility on firms’ overall technological innovation and exploratory innovation. The coefficient of HSR network centrality (C_D) in Column (4) is 1.061, and it is significant (p-value < 0.01), indicating that the HSR network can promote the mobility of specialists; the coefficient of technological innovation (Innovation) on technicians’ mobility (Rdpersonratio) in Column (5) is significant and positive (p-value < 0.01), indicating that technician mobility can prompt firms to carry out technological innovation. Combining the results above, we conclude that the HSR network is able to promote technician mobility, and then prompt firms to pursue technological innovation. Similarly, the coefficient of exploratory innovation (Exploration) on technicians’ mobility (Rdpersonratio) in Column (6) is significant and positive (p-value < 0.01), also showing that the mediating role of specialist mobility is reflected in exploratory innovation.

5.2. Cross-Sectional Tests

In this section, we examine whether firms in less-developed regions and high-tech firms experience a more significant increase in technological innovation.

5.2.1. Developed or Less Developed Regions

As China is a vast country, the resource endowments of different regions are vary vastly between each other, resulting in unbalanced economic development between regions, so the innovation resources introduced by the HSR network will certainly have a heterogeneous impact. The eastern coastal regions are developed in terms of economics and serve as hubs for capital and labor, while the central and western regions are relatively underdeveloped, with limited access to information and a lack of capital investment and labor inflow [22]. The HSR network overcomes the information barriers in the central and western regions, and makes investors shift their attention to these regions, thus leading to the inflow of capital. The HSR network has a facilitating effect on local economic development, promoting infrastructure construction in the central and western regions, and through the effect of investment multiplier, the HSR network further promotes the development of the local economy [13], and then attracts the inflow of talents [3]. Therefore, compared with eastern regions, the resources needed for technological innovation are relatively scarce in the central and western regions, so the HSR network is more capable of promoting firms’ technological innovation activities in the central and western regions.

We divide the samples into the eastern and the central and western regions, representing the developed and less developed areas, respectively. The results are shown in

Table 9. HSR network and technological innovation/exploratory innovation: cross-sectional tests on the regional level.

Variable	Innovation		Exploration
	(1)	(2)	(3)	(4)
	Developed Regions	Less Developed Regions	Developed Regions	Less Developed Regions
CD	0.001 ***	0.002 ***	0.001 ***	0.002 ***
	(5.55)	(6.79)	(6.36)	(6.61)
Lev	−0.000	−0.000	−0.001	−0.001
	(−0.31)	(−0.07)	(−0.87)	(−0.44)
Roa	0.018 ***	0.007 **	0.020 ***	0.010 ***
	(9.14)	(2.15)	(11.00)	(3.45)
Size	0.000	−0.000	−0.001 ***	−0.001
	(0.46)	(−0.67)	(−2.82)	(−1.62)
Age	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***
	(−2.92)	(−3.26)	(−2.91)	(−4.27)
Cash	−0.000	−0.000 *	−0.000 ***	−0.000 ***
	(−0.87)	(−1.85)	(−2.95)	(−3.04)
Tobins’Q	0.002 ***	0.002 ***	0.002 ***	0.001 ***
	(13.09)	(7.35)	(12.16)	(6.41)
Mkt	0.001 ***	0.002 ***	0.002 ***	0.001 ***
	(3.42)	(5.39)	(9.33)	(5.38)
Divide	−0.000	0.003 ***	0.001	0.001 **
	(−0.32)	(3.26)	(1.33)	(2.13)
Inctrl	0.011 ***	0.007 ***	0.011 ***	0.005 ***
	(4.88)	(3.03)	(5.22)	(2.76)
Constant	−0.017 **	−0.001	−0.004	0.008
	(−2.44)	(−0.12)	(−0.61)	(1.03)
Ind	YES	YES	YES	YES
Year	YES	YES	YES	YES
N	8219	2712	8219	2712
F	89.27	39.60	81.16	37.08
Adj. R2	0.284	0.256	0.265	0.240
Difference	2.29 ***		2.80 ***

Note: referring to the National Bureau of Statistics of China, we define the eastern regions as Beijing, Tianjin, Hebei Province, Shanghai, Jiangsu Province, Zhejiang Province, Fujian Province, Shandong Province, Guangdong Province, and Hainan Province. * p < 0.1, ** p < 0.05, *** p < 0.01.

. The coefficient of the HSR network on technological innovation in less developed regions is 0.002, and it is significant (p-value < 0.01), while this value is larger than the coefficient in the developed regions (the coefficient is 0.001, p-value < 0.01). And the difference between the developed and less developed areas is significant (p-value < 0.01), indicating that the promoting effect of the HSR network on technological innovation is more significant for firms in less developed areas. Similarly, the promoting effect of the HSR network on exploratory innovation is stronger for firms in less developed areas.

5.2.2. High-Tech or Non-High-Tech Firms

Theoretically, the more accessible the HSR network is, the more technological innovations should be carried out by firms that rely more on innovation resources. Compared with non-high-tech firms, high-tech firms have a clearer motivation for technological innovation and a higher reliance on innovation resources, so the utilization of resources brought about by the HSR network should be more significant. According to the Administrative Measures for the Recognition of High-tech firms in China, we divide the samples into high-tech firms and non-high-tech firms, and perform regression separately for each group.

The results are shown in

Table 10. HSR network and technological innovation/exploratory innovation: cross-sectional tests on the firm level.

Variable	Innovation		Exploration
	(1)	(2)	(3)	(4)
	High-Tech Firms	Non-High-Tech Firms	High-Tech Firms	Non-High-Tech Firms
CD	0.001 ***	0.001 ***	0.001 ***	0.001 ***
	(5.22)	(4.03)	(5.22)	(5.24)
Lev	−0.000	−0.000	−0.000	−0.002
	(−0.07)	(−0.19)	(−0.05)	(−1.37)
Roa	0.023 ***	0.015 ***	0.025 ***	0.016 ***
	(10.91)	(9.02)	(13.40)	(10.59)
Size	0.001 *	−0.001 ***	−0.000	−0.001 ***
	(2.11)	(−4.04)	(−1.21)	(−5.94)
Age	−0.000 **	−0.000 ***	−0.000 ***	−0.000 **
	(−3.18)	(−4.38)	(−4.35)	(−2.71)
Cash	−0.000	−0.000 **	−0.000 **	−0.000 **
	(−0.79)	(−2.77)	(−2.67)	(−3.09)
Tobins’Q	0.003 ***	0.001 ***	0.002 ***	0.000 **
	(17.46)	(4.17)	(15.74)	(3.26)
Mkt	0.002 ***	0.001 ***	0.002 ***	0.001 ***
	(10.66)	(7.63)	(14.90)	(9.64)
Divide	0.001	0.001	0.001	0.002 ***
	(1.09)	(1.83)	(1.15)	(3.40)
Inctrl	0.012 ***	0.007 ***	0.010 ***	0.007 ***
	(4.91)	(3.68)	(4.59)	(4.05)
Constant	−0.020 **	0.018 **	−0.001	0.023 ***
	(−3.07)	(3.18)	(−0.16)	(4.60)
Ind	YES	YES	YES	YES
Year	YES	YES	YES	YES
N	6894	4037	6894	4037
F	105.39	28.76	98.57	37.36
Adj. R2	0.267	0.194	0.254	0.240
Difference	0.001 ***		0.001 ***

Note. Standards for recognition of high-tech enterprises in China. For firms with a sales revenue of less than CNY 50 million in recent years, the ratio of their R&D investment to sales revenue should not be less than 6%; for firms with sales revenue of CNY 50 million and CNY 200 million, the ratio of their R&D investment to sales revenue should not be less than 4%; for firms with a sales revenue of more than CNY 200 million, the ratio of their R&D investment to sales revenue should not be less than 4%. For firms with sales revenue higher than CNY 200 million for recent years, the ratio of R&D investment to sales revenue should not be less than 3%. * p < 0.1, ** p < 0.05, *** p < 0.01.

. The table shows that in both high-tech firms and non-high-tech firms, the HSR network (C_D) has a significantly promoting effect on technological innovation (Innovation), and the difference between high-tech firms and non-high-tech firms is 0.001; this is significant (p-value < 0.01), indicating that the promoting effect of the HSR network on technological innovation is more significant in high-tech firms. The promoting effect of the HSR network (C_D) on exploratory innovation (Exploration) is similar.

6. Conclusions

In this study, we use a sample of listed Chinese firms from 2009 to 2019, treating HSR services in China as an exogenous shock on firms’ geographical accessibility, and examine the role of geographic accessibility in technological innovation. We find that the HSR network can promote firms’ technological innovation, and compared with exploitative innovation, the impact of the HSR network on firms’ technological innovation is mainly reflected in exploratory innovation. Further analysis reveals that HSR networks can stimulate technological innovation by facilitating the alleviation of firms’ financing constraints and increasing the mobility of specialists, and this relationship is more pronounced among high-tech firms and firms in undeveloped regions. These findings contribute to the ongoing discussions in the field and help enhance the core competitiveness and sustainability of enterprises, offering valuable practical implications.

The results of this study are productive in terms of offering recommendations for policymakers and corporate managers. We believe that policymakers and managers should take into account the interaction between geographical accessibility and resource flows when making strategic choices. Policymakers, especially in developing countries, should fully recognize that the HSR network provides a feasible path for the implementation of the innovative development strategy. They should make full use of the optimizing effect of HSR construction on resource allocation and rationally select locations for key transportation positions, so as to guide the redistribution of resources, to promote the inflow of capital, talents and other resources [5], and then to optimize the industrial layout, develop the potential of the market, and realize the high-quality development of the regional economy [10]. Furthermore, policymakers can make use of geographic transportation’s guiding effect on resources to promote the redistribution of resources in less developed regions. At the same time, policymakers should also speed up the construction of supporting infrastructure in less developed areas and promote the synergistic effect between HSR and other infrastructures [59], so as to fully harness the economic advantages HSR can bring. Regional economic conditions and development potentials should be taken into account in the planning of transportation infrastructure routes, so that the spillover effect of the economic circle can fully play out. In addition, policymakers should reduce barriers to the mobility of capital and labor to stimulate corporate vitality and promote innovation activities in enterprises. They can improve infrastructure and enhance the physical conditions for inter-regional capital allocation, thereby lowering the cost of capital flow. By simplifying policies and regulations related to talent mobility and improving inter-regional talent exchange platforms, they can optimize the allocation of high-skilled labor across regions, ultimately promoting innovation activities in enterprises.

Firm managers, especially those of high-tech firms, should pay attention to the opportunity for transportation infrastructure construction, utilizing network connections fostered by transportation, joining extensive collaborative exchange networks, and seeking to yield the synergistic effects of internal and external innovation resources. Firstly, enterprises can promote the diffusion of technology and the exchange of tacit knowledge through the HSR network. For example, by frequently participating in industry conferences and technical salons, enterprises can actively integrate into regional innovation networks, fostering the creation of new knowledge and technologies. Secondly, managers should make full use of the advantages of geographic access to attract the inflow of high-quality talents, so as to foster core competitive advantages for firms [3] and promote the high-quality development and sustainable development ability of firms.

Since innovation is a key driver for sustainability [7], our research is particularly important for promoting sustainable economic and social development. First, geographical accessibility promotes technological innovation and enhances the core competitiveness of enterprises. Sustainability requires sustainable economic growth that focuses on the quality of economic development in the long term [60]. Our research finds that improving geographical accessibility promotes the technological innovation of enterprises, especially exploratory innovation, which helps them to keep up with and adapt to market demand and technological changes and to develop new businesses or carry out innovative transformations in a timely manner. The technological innovation of enterprises greatly enhances their core competitiveness, gives them an edge in market competition and helps them realize sustainable profitability and development [7].

Second, geographical accessibility reduces the loss of high-quality projects and improves the efficiency of resource utilization. On the one hand, geographic accessibility reduces the cost of information acquisition brought about by long distances, increases the efficiency of information acquisition by external investors [26], and enables investors to find more investment opportunities and better assess the feasibility and profitability of enterprise projects. This enables enterprises to obtain high-quality investment funds for innovative projects, thus improving the efficiency of resource utilization by both enterprises and external investors, and reduces the possible waste of resources. On the other hand, improved geographic accessibility can facilitate communication and interaction among specialists [3], integrate multidisciplinary knowledge and technology, identify technological deficiencies, promote technological improvements, and accelerate the development of new technologies. In addition, specialists can understand the demand for and utilization of resources in different regions and industries, optimize the allocation of resources, and achieve the efficient utilization of resources. Therefore, geographic accessibility helps to realize the efficient utilization of resources and a positive interaction between economic development and the resource environment.

Third, geographic accessibility can alleviate imbalances in regional economic development and promote synergistic and sustainable regional economic development. Sustainable development is concerned with social equity and is committed to eradicating poverty and reducing inequality [61]. Cross-sectional tests show that geographic accessibility significantly enhances the technological innovation of firms located in relatively underdeveloped regions. Information in underdeveloped regions is relatively isolated, and the HSR network breaks down the information barriers faced by underdeveloped regions [5,26], bringing in an inflow of capital and manpower that allows for the better development of underdeveloped regions. In addition, an increase in geographic accessibility will also be accompanied by the construction of other infrastructure and services, which will expand the economic development of the less developed regions through the effect of the investment multiplier [59], thus narrowing the gap with the developed regions and promoting the balanced development of all regions. Geographically more efficient links make it possible to improve the level of technological innovation of enterprises in less developed regions and to form a larger pattern of economic cooperation, thus promoting the synergistic and sustainable development of the regional economy.

This paper still has some limitations. First, our sample includes only listed companies. The panel data analysis used in this study requires a large amount of continuous, structured data to support it, and it is difficult for unlisted company data to meet the methodology’s requirements for data quality and sample size. Future research could attempt to incorporate samples of non-listed companies through methods such as enterprise surveys and data from industry associations, or make comparisons across countries to extend the findings further. Second, we do not explore the impact of the development of virtual interaction on our research. The use of internet-based technologies can reduce communication costs between distant individuals [62], which may reduce the role of transportation in innovation production. This is unlikely to have a substantial impact on our results, as virtual interactions were not popularly developed and applied during our sample period. The literature suggests that the soft information gained through face-to-face communication is a key driver of knowledge creation [63], because telecommunications technology and face-to-face interaction are complementary, not substitutes [64,65]. With the development and popularization of the digital economy and artificial intelligence [5], whether these technological advancements can replace the promoting effect of face-to-face communication on technological innovation in firms is a potential direction for future research. Additionally, it is possible to combine the hard networks formed by HSR and other transportation infrastructures with soft networks such as alumnus relations and hometown connections, and study whether their interaction can contribute to the economic development of enterprises and regions in future research.