The Impacts of High-Speed Rail on Producer Service Industry Agglomeration: Evidence from China’s Yangtze River Delta Urban Agglomeration

Abstract

The construction of the high-speed rail (HSR) network in China has greatly weakened the spatial barriers to the flow of production resources, which has become a key factor affecting the spatial layout of the producer service industry. Based on the panel data of 26 cities in the Yangtze River Delta urban agglomeration from 2005 to 2018, this paper uses a multi-phase difference-in-difference (DID) model to examine the impacts of HSR services on the agglomerations of the producer service industry and its subdivision industries from two perspectives, namely, specialized agglomeration and diversified agglomeration. The results show that: (1) on the whole, the opening of an HSR has a significant positive effect on the specialized agglomeration of the producer service industry and a significant negative effect on the diversified agglomeration; (2) in terms of subdivision industries, there exists significant industrial heterogeneity in the agglomeration effect of the producer service industry under HSR services, regardless of whether it is a specialized agglomeration or a diversified agglomeration; among them, the financial industry belongs to the “highly significant promotion” industry, while the other four subdivision industries belong to the “highly significant inhibition” industry.

1. Introduction

In recent years, China’s high-speed rail (HSR) has developed rapidly and made great achievements, reshaping China’s transportation pattern. China has become the country with the longest operating mileage of HSR in the world. By the end of 2022, China’s HSR operating mileage exceeded 42,000 km. HSR is an important driving force of regional economic development in the country [1,2,3,4,5]. The influential effects of HSR on regional economic development mainly include a polarization effect and a diffusion effect [6,7,8,9,10]. In general, HSR directly and indirectly affects regional economic development. The direct impact is reflected in the investment pulling effect and the increase in operating income [11]. The indirect impact is more prominent. This impact is mainly reflected in the spatial (re)allocation of production resources due to the rapid flow of elements (information flow, capital flow, personnel flow, etc.), thereby promoting the realization of the influential effect of regional economic growth [12]. This is called the economic allocation effect of the HSR in the New Economic Geography [13]. Specifically, this reallocation effect is mainly reflected in the reallocation of production resources between cities along the HSR lines [14], as well as between cities along and not along HSR lines [11,15,16]. Based on the existing research results [10,17,18,19,20], this paper holds that HSR service changes the flow and direction of production resources by promoting accessibility, leading to an unbalanced agglomeration and the diffusion of production resources, thereby changing the spatial pattern of the regional economy. Thus, it is theoretically and practically significant to discuss the impact of HSR on economic resource agglomeration at the micro level.

Under the background of the new normal development of China’s economy, China is gradually entering the service economy era. In 2018, the added value of the service industry accounted for 52.2% of the GDP, and it is expected to account for 73.7% by 2030 (Source: Statistical Bulletin of National Economic and Social Development of the People’s Republic of China 2020 and CASS). The service industry, especially the producer service industry, has become the main force promoting China’s economic growth [21]. So far, there is still no unified definition of the scope of the producer service industry in academia. The existing studies mainly define the producer service industry from the following perspectives: service objects [22], service types [23], service objects and types [24], service activities [25], and industrial employment [11]. Weber put forward the view of agglomeration economy in 1929, and believed that agglomeration economy was an important factor in the location choice of enterprises. Service industry agglomeration is the agglomeration of interrelated service industry enterprises in a specific geographical space. Service industry agglomeration can significantly promote regional economic growth to some extent, but a higher degree of service industry agglomeration can hinder regional economic growth. The integration level of transportation in the Yangtze River Delta urban agglomeration affects the service industry agglomeration through the knowledge-spillover effect brought by the improvement in accessibility [26]. The HSR has promoted the upgrading of industrial structure and has a significant impact on service industry agglomeration [27]. Some studies believe that HSR service is more likely to attract more industrial enterprises and increase industrial output, but the impact on the service industry is not obvious [28]. That being the case, how does the opening of an HSR affect the spatial distribution of the producer service industry and its subdivision industries (agglomeration or diffusion)?

Urban agglomerations have become an important characteristic of China’s social and economic development. Therefore, this paper takes urban agglomeration as the research object to explore the spatial distribution of the producer service industry under the influence of HSR service. There are few studies on the development of rural areas. Many interrelated rural settlements have been established by taking advantage of the traditional spatial advantages of the riverbank location [29]. Tourism is an important industrial choice for rural revitalization. As a sustainable form of tourism, rural community-based tourism is highly valued for its role in the industry [30]. For HSR service and the development of rural areas, research has found that HSR service improves the accessibility to rural and peripheral areas compared to BRT [31]. Furthermore, HSR service affects the change in urban and rural spatial patterns to some extent [32]. It has also been found that HSR stations in rural areas have a greater spatial impact, so the government is encouraged to install HSR stations in those areas [33].

With respect to the impact of HSR service on producer service industry agglomeration, the existing studies mainly focus on: firstly, in terms of research content, exploring the spatial effect of the whole producer service industry or single service industry (tourism industry/finance industry/information industry, etc.) under the influence of the HSR [34,35,36]; secondly, in terms of research scale, a specific HSR line or a certain area (provincial-level/prefecture-level cities) is taken as the research object to explore the industrial spatial effect caused by HSR development [37,38]. The existing theoretical and empirical studies on the influencing factors of producer service industry agglomeration have not paid enough attention to the transportation factors, especially the HSR effect. There are relatively few studies that comprehensively investigate the impact of the rapid development of the HSR network on the spatial agglomeration of the producer service industry and its subdivision industries in urban agglomeration.

In order to address the existing research insufficiency, based on the panel data of 26 cities in the Yangtze River Delta urban agglomeration in China from 2005 to 2018, this paper measures the impact of the opening of an HSR on the producer service industry agglomeration and the differentiated effects of its subdivision industry agglomeration from the two dimensions of specialized agglomeration and diversified agglomeration.

The innovation points of this paper are as follows: (1) on the whole, this paper comprehensively examines the impact of HSR service on the agglomerations of the producer service industry and its subdivision industries; (2) this paper examines the impact of HSR service on the producer service industry agglomeration from the two dimensions of specialized agglomeration and diversified agglomeration; (3) further, for the producer service industry, this paper examines the heterogeneity of the impacts of HSR service on its subdivision industry agglomeration from the two dimensions of specialized agglomeration and diversified agglomeration.

The rest of this paper is organized as follows: Section 2 details theoretical framework and hypothesis. Section 3 shows the results of the empirical analysis and a further analysis of the research results. Finally, Section 4 presents the conclusions.

2. Theoretical Framework and Hypothesis

The industrial effect of the HSR refers to the impact of the HSR on the industrial structure, industrial development, and spatial distribution of cities along the lines. This effect exists both within and between cities. The agglomeration or diffusion of the spatial distribution of producer service industries is the result of enterprises’ location choices for the purpose of reducing production costs and improving profits; it is also an important manifestation of the space–time evolution of economic activities.

The service industry agglomeration is an effective way for regions to obtain competitive advantages [39]. Regional traffic conditions are an important influencing factor for service industry agglomeration, especially producer service industry agglomeration. The development level of local transportation affects industrial agglomeration mainly through the “cost effect” [40]. The impact of the HSR on producer service industry agglomeration is determined by the technical characteristics of the HSR and the industrial characteristics of the producer service industry. The impact of space–time distance on producer service industry agglomeration is more obvious than that of geographical distance. HSR service decreases space–time distances between cities, greatly promoting the spatial reallocation of production resources [41]. The changes in accessibility and connectivity caused by the space–time compression effect of the HSR provide convenience for the transmission of knowledge flow and information flow, which effectively promote the agglomeration effect and diffusion effect of the service industry. The producer service industry is dependent on high-end flow elements, such as labor flow, knowledge flow, and information flow, and thus more sensitive to the HSR [11]. Therefore, HSR service significantly affects producer service industry agglomeration through the space–time compression effect [11,42].

Nowadays, two opposing externality theories, the Marshall–Arrow–Romer type and the Jacobs type, are widely used to explain the causes of industrial agglomeration, resulting in two different agglomeration modes: specialized agglomeration (the same industrial sectors concentrated in a specific spatial geographical location) and diversified agglomeration (different industrial sectors concentrated in a specific spatial geographical location) [43,44,45,46]. Due to different agglomeration modes, the HSR may have different effects on the specialized and diversified agglomerations of the producer service industry. In particular, due to the influence of industrial characteristics and urban development level, there may exist significant differences in the influential effects of the HSR on different industrial types [11,47].

Thus, the theoretical framework of the impact of the HSR on producer service industry agglomeration is shown in Figure 1.

Based on this, this paper proposes the following hypotheses:

Hypothesis 1 (H1):

The opening of an HSR has different effects on the specialized and diversified agglomerations of the producer service industry.

Hypothesis 2 (H2):

The impact of the opening of an HSR on the producer service industry agglomeration exists despite significant industrial heterogeneity, whether it is a specialized or a diversified agglomeration.

2.1. Sample Description

China’s Yangtze River Delta Urban Agglomeration (YRD) consists of “three provinces and one municipality”. The planning scope includes 26 cities in Shanghai municipality, Jiangsu Province (Nanjing, Wuxi, Changzhou, Suzhou, Nantong, Yancheng, Yangzhou, Zhenjiang, Taizhou), Zhejiang Province (Hangzhou, Ningbo, Jiaxing, Huzhou, Shaoxing, Jinhua, Zhoushan, Taizhou), and Anhui Province (Hefei, Wuhu, Ma’anshan, Tongling, Anqing, Chuzhou, Chizhou, Xuancheng) (according to the development plan of the Yangtze River Delta urban agglomeration issued by China’s National Development and Reform Commission in June, 2016. See https://max.book118.com/html/2017/0607/112135355.shtm, accessed on 1 January 2023). The land area of the YRD is 211,700 km². In 2022, the total population of the YRD exceeded 245 million, and its total GDP exceeded 27 trillion yuan. In terms of the employment ratio of the service industry, in 2018, the service industry employment of the YRD was 21.5% lower than that of the secondary industry, accounting for 43.9% of the total employment in this region (68.9% in Shanghai, 33.4% in Jiangsu, 41.1% in Zhejiang, and 42% in Anhui) (Source: China City Statistical Yearbook 2018).

The first HSR line of the YRD opened in August 2008. By the end of 31 December 2018, 20 cities in the YRD had opened HSR service. These cities will be selected as the treatment group (marked in green). Six cities were still without HSR service, including Nantong, Yancheng, Yangzhou, Taizhou, Zhoushan, and Xuancheng. They will be selected as the control group (marked in red), as shown in Figure 2. Overall, the YRD is the region with the most intensive HSR network construction in China. Thus, this region provides a very appropriate research sample for the paper to investigate the impact of the HSR on producer service industry agglomeration.

According to the “National Economical Industry Classification” (revised version) from 2002, and also referring Shao et al., this paper further classifies the types of producer service industry into five subdivision industries, namely, “transportation, warehousing and postal service”, “information transmission, computer services and software”, “finance”, “leasing and business services”, and “scientific research, technical services and geological prospecting”. In 2018, the YRD’s producer service industry employment accounted for 37.9% of the total service industry employment in the region, of which employment in “transportation, warehousing and postal service” accounted for the highest proportion of producer service industry employment (24.9%), while employment in “scientific research, technical services and geological prospecting” accounted for the lowest proportion of producer service industry employment (13.1%) (Source: China City Statistical Yearbook 2018).

2.2. Model Specification and Data Description

2.2.1. Model Specification

At present, the difference-in-difference (DID) method has become one of the research methods most commonly used by scholars to evaluate policy or project effects [48]. However, due to the differences in urban geographical location, element resource endowments, and economic development levels, the opening time of an HSR varies in different regions. Thus, referring to Liu et al.’s (2022) method, this paper constructs a multi-phase DID model to analyze the net effect of the opening of an HSR on urban producer service industry agglomeration [49].

The benchmark regression model is set as Equation (1):

$$Y_{it}={\beta }_0+{\beta }_{1}cit{y}_{it}yea{r}_{it}+{\beta }_2{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

where $i$ denotes the city; t denotes the year; ${\mu }_i$ is the individual-fixed effect, which is used to control individual factors affecting producer service industry agglomeration; ${\lambda }_t$ is the time-fixed effect, which is used to control time factors affecting producer service industry agglomeration; ${\epsilon }_{it}$ is the random disturbances; ${\beta }_0$ is the intercept; and ${\beta }_1$ and ${\beta }_2$ are the coefficients.

(1) $Y$ is the dependent variable, denoting the spatial agglomeration level of the urban producer service industry. According to the analysis in Section 1, this paper measures urban producer service industry agglomeration from two aspects: specialized agglomeration and diversified agglomeration. The commonly used indicators to measure the industrial agglomeration include the local entropy index, the Herfindahl index, the spatial Gini coefficient, etc. Among them, the local entropy index can more realistically reflect the spatial distribution of geographical elements of the same industrial sector in a certain region, while the Herfindahl index can more comprehensively reflect the differences in the concentration degree of geographical elements of different industrial sectors in a certain region. Therefore, the local entropy index and the Herfindahl index are used to measure the specialized agglomeration and diversified agglomeration of the urban producer service industry, respectively [11,50].

The formula of the specialized agglomeration ($sag$) is set as Equation (2):

$$sa{g}_{ij}=\frac{x_{ij}/{\displaystyle \sum _j{x}_{ij}}}{{\displaystyle \sum _i{x}_{ij}}/{\displaystyle \sum _i{\displaystyle \sum _j{x}_{ij}}}}$$

(2)

The formula of the diversified agglomeration ($dag$) is set as Equation (3):

$$da{g}_{ij}={\displaystyle \sum _j\frac{x_{ij}}{{\displaystyle \sum _j{x}_{ij}}}}\left\{\frac{1/{\displaystyle \sum _{j^{\text{'}}\ne j}^n{\left[x_{i{j}^{\text{'}}}/({\displaystyle \sum _j{x}_{ij}}-x_{ij})\right]}^2}}{1/{\displaystyle \sum _{j^{\text{'}}\ne j}^n{\left[x_{j^{\text{'}}}/\left({\displaystyle \sum _i{\displaystyle \sum _j{x}_{ij}}}-{\displaystyle \sum _i{x}_{ij}}\right)\right]}^2}}\right\}$$

(3)

where variable $x_{ij}$ denotes the employment of industry $j$ in city $i$.

(2) $cit{y}_{it}yea{r}_{it}$ is the core independent variable; its coefficient denotes the net effect of the opening of an HSR on the producer service agglomeration. $cit{y}_{it}$ is a group dummy variable; it equals 1 for city $i$ and year t with HSR service, and 0 for city $i$ and year t without HSR service. $yea{r}_{it}$ is a time dummy variable; if HSR service exists in city $i$ at year $t^{\prime }$, then when $t\ge t^{\prime }$, $yea{r}_{it}=1$, and 0 otherwise [51].

(3) $X_{it}$ denotes a series of control variables.

Referring to the existing studies on the influencing factors of producer service industry agglomeration [11,15,47], this paper introduces five indicators as control variables: urban scale, wage level, informatization level, knowledge-spillover level, and openness.

The reasons for selecting the control variables are as follows:

(1)

Urban scale: A higher population density means a larger urban scale, which affects the market scale to some extent and thus affects the spatial layout of the producer service industry. Hence, the urban scale is measured by population density, as denoted by the population size per square kilometer.

(2)

Wage level: Higher wages can attract a labor force to create wealth for enterprises, thus improving the agglomeration degree. However, higher wages will also increase the production costs of enterprises, thus reducing the agglomeration degree. The wage level is represented by the workers’ average wage level.

(3)

Informatization level: The improvement in the urban informatization level affects the location choice of producer service industry enterprises. The informatization level is measured by the number of Internet broadband access users.

(4)

Knowledge-spillover level (teacher): Producer service industry enterprises tend to introduce knowledge-flow elements into the process of production, thereby forming market competitive advantages and finally improving the enterprises’ benefits. The knowledge-spillover level is represented by the number of full-time teachers in institutions of higher learning.

(5)

Openness: The improvement in the openness level can further meet the needs of diverse elements of the producer service industry enterprises and affect the spatial layout of the producer service industry. The openness is represented by the proportion of foreign direct investment (FDI) actually used in the GDP of the year.

2.2.2. Data Description

Considering the availability and validity of the data, as well as the inconsistent statistical quality of data indicators of the subdivision industries of the producer service industry before and after 2018, we selected 26 cities in China’s Yangtze River Delta urban agglomeration during the period 2005–2018 as a research sample.

The relevant data mainly include HSR operation data and city statistical data. The HSR operation data come from the National Railway Timetable and the HSR network’s website. The city statistical data come from the China City Statistical Yearbook, provincial and urban statistical yearbooks, and the Statistical Bulletin of National Economic and Social Development of each city for the period between 2005 and 2018.

In order to avoid spurious regression and reduce heteroscedasticity, we make a logarithmic treatment for the variables with larger values other than dummy variables and percentile variables in regression analysis.

Table 1. Definitions and descriptive statistics of variables.

Variable	Indicator	Definition	Mean	Standard Deviation	Mini Mum	Maxi Mum
Dependent variable	sag	Specialized index, see Equation (2)	0.807	0.318	0.269	2.153
	dag	Diversified index, see Equation (3)	0.229	0.103	0.064	0.655
Group dummy variable	city	If the city is in the treatment group, then city = 1, and 0 otherwise.	0.769	0.422	0	1
Time dummy variable	year	If HSR service exists in the city at year $t^{\prime },$ then when $t\ge t^{\prime }$, year = 1, and 0 otherwise.	0.398	0.490	0	1
Control variable	urban scale	The population size per square kilometer	6.408	0.484	5.242	7.743
	wage level	The workers’ average wage level	10.702	0.482	9.407	11.870
	informatization level	The number of Internet broadband access users	4.249	1.205	0.900	8.551
	teacher	The number of full-time teachers in institutions of higher learning	0.804	1.179	0.010	5.253
	openness	The proportion of FDI actually used in the GDP of the year	3.688	2.110	0.204	11.674

presents the definitions and descriptive statistics of the various variables.

3. Results and Discussion

3.1. Baseline Regression

The test results are shown in

Table 2. Baseline regression results.

	(1)	(2)	(3)	(4)
	Sag	Sag	Dag	Dag
$cit{y}_{it}yea{r}_{it}$	0.051 **	0.042 *	−0.027 ***	−0.029 ***
	(0.022)	(0.022)	(0.010)	(0.010)
urban scale		0.091		−0.067
		(0.134)		(0.064)
wage level		0.134		−0.135 ***
		(0.087)		(0.042)
informatization level		−0.090 ***		0.013
		(0.022)		(0.010)
teacher		0.017		−0.027
		(0.038)		(0.018)
openness		0.020 ***		−0.005 **
		(0.004)		(0.002)
_cons	0.820 ***	−0.902	0.220 ***	1.985 ***
	(0.023)	(1.126)	(0.011)	(0.540)
city-fixed effect	yes	yes	yes	yes
time-fixed effect	yes	yes	yes	yes
N	364	364	364	364
R-squared	0.076	0.183	0.225	0.276

Standard errors in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

, where columns (1) and (3) are estimation results that do not involve control variables and columns (2) and (4) are estimation results involving control variables.

The empirical results show that, whether the model includes control variables or not, the opening of an HSR has a significant positive effect on the specialized agglomeration of the producer service industry. The result in column (2) shows that the opening of an HSR increases the specialized agglomeration of the producer service industry in cities along the HSR by 0.042. The main reasons for this influential effect are as follows: One of the important characteristics of the producer service industry is its high added value. This tacit knowledge is the most important part of this added value. The transfer of tacit knowledge is not only affected by human capital but also by time costs. The space–time compression effect brought by the opening of an HSR reduces the time cost of communication between cities. Additionally, the market accessibility and connectivity of cities along the HSR lines are significantly improved. It is more conducive to connecting the commodity market, capital market, and labor market of different cities than reducing the degree of information asymmetry between enterprises. The production resources required by a large number of enterprises in the same industrial sector will be continuously gathered in the cities along the HSR within the spatial scope, thus generating agglomeration economic benefits to improve the production benefits of enterprises in the same industrial sector in the cities along the HSR.

Moreover, whether the model includes control variables or not, the opening of an HSR has a significant negative effect on the diversified agglomeration of the producer service industry. The result in column (4) shows that the opening of an HSR decreases the diversified agglomeration of the producer service industry in cities along the HSR by 0.029. The main reasons for this influential effect are as follows: Firstly, based on the active diffusion phenomenon caused by the “relative factor prices”, for the larger regional price differences of production resources, such as real estate and land, the producer service industry enterprises in different industrial sectors can take advantage of the traffic convenience brought by the opening of an HSR to actively transfer industries to cities not along the HSR lines, thus reducing transaction costs and forming industrial diffusion. Secondly, based on the passive diffusion phenomenon caused by the “crowding effect”, the agglomeration economic effect brought by HSR will have negative impacts, such as the rising cost of labor force, capital, and technology. The rising cost of production factors will force the producer service industry enterprises in different industrial sectors to spread to cities not along the HSR line with relatively low cost of production resources.

Additionally, the findings are inconsistent with those of Freeman (2007), Murakami and Cervero (2012), Shao et al. (2017), and Hu and Xu (2022), who believe that the HSR has a significant positive effect on producer service industry agglomeration [11,27,52,53]. The findings are also inconsistent with those of Tian et al. (2021), who believe that the HSR has a significant negative effect on producer service industry agglomeration [54]. Moreover, these studies did not further subdivide the dimensions of producer service industry agglomeration.

In addition, we also see that most of the R-squared values are relatively low. This is because producer service industry agglomeration is not only affected by the HSR but also by other factors such as highways, railways, and civil aviation.

3.2. Subdivision Industry Regression

Based on the definition scope of producer service industries in Section 2, this paper further explores the differentiation effects of HSR service on the agglomerations of different types of producer service industries. We find that the impact of the opening of an HSR on producer service industry agglomeration exists despite significant industrial heterogeneity. In addition, we also see that most of the R-squared values are relatively low. This is because the agglomeration degree of different types of producer service industries is not only affected by the HSR but also by other factors. Due to different industrial characteristics, the impacts of the HSR on the agglomeration degree of different types of producer service industries have a lag effect.

Columns from (1) to (5) of

Table 3. Empirical results of the specialized agglomeration of subdivision industry sample.

	(1)	(2)	(3)	(4)	(5)
	Sagtra	Saginf	Sagfin	Sagren	Sagsci
$cit{y}_{it}yea{r}_{it}$	0.029	−0.027	0.145 ***	0.032	0.011
	(0.031)	(0.050)	(0.036)	(0.049)	(0.033)
urban scale	0.124	0.454	0.061	−0.737 **	0.170
	(0.192)	(0.308)	(0.222)	(0.300)	(0.201)
wage level	0.100	−0.384 *	−0.150	0.593 ***	0.455 ***
	(0.124)	(0.199)	(0.144)	(0.194)	(0.130)
Informatization level	−0.080 **	−0.153 ***	−0.065 *	−0.175 ***	0.038
	(0.031)	(0.050)	(0.036)	(0.049)	(0.033)
teacher	−0.171 ***	0.564 ***	−0.140 **	0.167 *	−0.114 **
	(0.055)	(0.088)	(0.063)	(0.086)	(0.057)
openness	0.021 ***	0.027 ***	0.056 ***	−0.005	0.005
	(0.006)	(0.010)	(0.007)	(0.010)	(0.006)
_cons	−0.708	1.870	2.176	−0.088	−4.935 ***
	(1.609)	(2.580)	(1.861)	(2.513)	(1.686)
city-fixed effect	yes	yes	yes	yes	yes
time-fixed effect	yes	yes	yes	yes	yes
N	364	364	364	364	364
R-squared	0.093	0.344	0.232	0.140	0.173

Standard errors in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

present the empirical results of the impacts of the HSR on the specialized agglomerations of diverse producer service industries.

The results show that the opening of an HSR has a significant positive effect on the specialized agglomeration of the financial industry, and the financial industry belongs to the “highly significant promotion” industry (at a 1% significance level). The main reasons are as follows: The development of the financial industry in a region benefits from the local financial-market environment and financial agglomeration degree. The opening of an HSR can expand the scale of the local financial market. A better financial-market development environment can cultivate or attract a large number of the same type or similar types of financial enterprises to integrate into regions with a higher degree of financial agglomeration, which will lead to the effect of industrial specialized agglomeration.

However, the opening of an HSR has insignificant positive effects on the specialized agglomerations of “transportation, warehousing and postal service”, “leasing and business services”, and “scientific research, technical services and geological prospecting”, and insignificant negative effects on the specialized agglomeration of “information transmission, computer services and software”. The insignificant effects can be attributed to the fact that the opening of an HSR brings the flows of capital, knowledge, and technology to the local market and introduces external market competition, so the positive and negative effects ultimately offset each other.

Columns from (1) to (5) of

Table 4. Empirical results of the diversified agglomeration of subdivision industry sample.

	(1)	(2)	(3)	(4)	(5)
	Dagtra	Daginf	Dagfin	Dagren	Dagsci
$cit{y}_{it}yea{r}_{it}$	−0.011 ***	−0.004 ***	−0.003	−0.008 ***	−0.004 ***
	(0.004)	(0.001)	(0.010)	(0.003)	(0.001)
urban scale	−0.009	0.014 *	0.034	−0.095 ***	−0.011
	(0.024)	(0.008)	(0.059)	(0.018)	(0.009)
wage level	0.003	−0.005	−0.195 ***	0.038 ***	0.024 ***
	(0.016)	(0.005)	(0.038)	(0.012)	(0.006)
Informatization level	0.007 *	0.000	0.008	−0.004	0.001
	(0.004)	(0.001)	(0.010)	(0.003)	(0.001)
teacher	−0.008	0.014 ***	−0.035 **	0.002	0.001
	(0.007)	(0.002)	(0.017)	(0.005)	(0.003)
openness	−0.004 ***	−0.000 *	0.006 ***	−0.004 ***	−0.002 ***
	(0.001)	(0.000)	(0.002)	(0.001)	(0.000)
_cons	0.087	−0.028	1.781 ***	0.285 *	−0.140 *
	(0.201)	(0.069)	(0.494)	(0.152)	(0.076)
city-fixed effect	yes	yes	yes	yes	yes
time-fixed effect	yes	yes	yes	yes	yes
N	364	364	364	364	364
R-squared	0.218	0.278	0.177	0.316	0.298

Standard errors in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

present the empirical results of the impacts of the HSR on the diversified agglomerations of diverse producer service industries.

The results show that the opening of an HSR has a significant negative effect on the diversified agglomerations of “transportation, warehousing and postal service”, “information transmission, computer services and software”, “leasing and business services”, and “scientific research, technical services and geological prospecting”, and these four subdivision industries belong to the “highly significant inhibition” industry (at a 1% significance level). The main reasons are as follows: For these four subdivision industries, although the city’s higher infrastructure construction level, internet-information transmission speed, commercial development potential, and science and technology development level have comparative advantages, they also have negative effects, such as higher land prices, production costs, and labor costs, as well as serious environmental pollution. Although it can attract different types of diversified service industry enterprises in the short term, in the long term, the negative impact will lead to the transfer of service industry enterprises and make it difficult to form a joint force, which will eventually hinder the diversified agglomeration of the service industry.

However, the opening of an HSR has an insignificant negative effect on the diversified agglomeration of the financial industry. This insignificant effect can be attributed to the fact that enterprises with a higher financial development level can better integrate into the areas with a higher degree of financial agglomeration by taking advantage of the convenience brought by the opening of an HSR, while for the enterprises with an average or lower level of financial development, due to their own development restrictions, they can hardly integrate into the areas with a higher degree of financial agglomeration even if the opening of an HSR occurs, which makes it difficult to change the nature of the small size of the local financial market. Therefore, overall, the impact of HSR on the diversified agglomeration of the financial industry is statistically insignificant.

3.3. Further Analysis

3.3.1. Parallel-Trend Test

An important condition for a multi-phase DID model estimation is that the treatment and control groups have a parallel trend without policy intervention. This paper selects three years before and three years after the opening of an HSR for comparison. Figure 3 shows the estimation results of the three years before, the year of, and the three years after the opening of an HSR, which are used to test whether the parallel-trend test is passed. In the process of testing, in order to eliminate the influence of collinearity on the test results, the period prior to the policy implementation is taken as the benchmark group and will be dropped afterward. The results show that there is no significant difference in producer service industry agglomeration between the treatment group and the control group three years prior to the opening of an HSR. The year of the opening of the HSR, as well as one year and two years after the opening of the HSR, the impact of the HSR on producer service industry agglomeration is insignificant. It does not become significant until the third year after the opening of the HSR. Therefore, the parallel-trend test is passed.

3.3.2. Robustness Test

In the benchmark regression, the full sample contains four core cities, namely Shanghai, Nanjing, Hangzhou, and Hefei. These four core cities have greater core competitiveness, and an HSR may be opened earlier than in other cities. Additionally, the opening of the HSR will reallocate production resources between core cities and peripheral cities. Therefore, if these four core cities are included in the full sample, the reliability and accuracy of the test results may be affected. Thus, this paper further excludes the four core cities and does a robustness test of the estimation results of the baseline regression model.

Table 5. Sub-sample empirical results.

	(1)	(2)	(3)	(4)
	Sag	Sag	Dag	Dag
$cit{y}_{it}yea{r}_{it}$	0.064 ***	0.072 ***	−0.033 ***	−0.041 ***
	(0.021)	(0.021)	(0.012)	(0.012)
urban scale		−0.158		−0.025
		(0.127)		(0.072)
wage level		0.170 **		−0.145 ***
		(0.081)		(0.046)
informatization level		−0.078 ***		0.008
		(0.028)		(0.016)
teacher		0.378 ***		−0.231 ***
		(0.137)		(0.078)
openness		0.022 ***		−0.006 ***
		(0.004)		(0.002)
_cons	0.736 ***	0.100	0.237 ***	1.882 ***
	(0.022)	(1.026)	(0.012)	(0.581)
city-fixed effect	yes	yes	yes	yes
time-fixed effect	yes	yes	yes	yes
N	308	308	308	308
R-squared	0.166	0.289	0.260	0.325

Standard errors in parentheses, ** p < 0.05, *** p < 0.01.

shows the sub-sample empirical results excluding the four core cities in order to examine whether the regression results are consistent with those in the baseline regression. Columns (1) and (3) are empirical results that do not involve control variables; columns (2) and (4) are empirical results involving control variables. The results show that, whether the model includes control variables or not, the impact of the opening of an HSR on the specialized agglomeration of the producer service industry is significantly positive at a level of 1%; the impact of the opening of an HSR on the diversified agglomeration of the producer service industry is significantly negative at a level of 1%. Therefore, the sub-sample empirical results are consistent with the benchmark regression results, which verifies the robustness of the above research results.

In addition, we find that the absolute values of the sub-sample regression coefficients are all greater than those in the baseline regression, for both specialized and diversified agglomerations. In other words, the opening of an HSR makes the resources of the producer service industry spread from core cities to peripheral medium- and small-sized cities along the HSR. Therefore, the specialized agglomeration degree of the producer service industry is significantly improved, and the diversified agglomeration degree is significantly decreased in these medium- and small-sized cities.

3.3.3. Placebo Test

When using the multi-phase DID model to assess the policy effectiveness, the impact of potentially omitted variables on the estimation results is often ignored. In other words, although the model controls individual-fixed effects and time-fixed effects, some unobserved variables may have a significant impact on the estimation results that cannot be ignored. In order to avoid the influence of unobserved variables, a placebo test is also required in this paper. The method we used is to advance the opening time of an HSR in a city by seven years (denoted by PHSR). If the regression coefficients of the impact of the HSR on specialized and diversified agglomerations of the urban producer service industry are not significant, it indicates that the benchmark regression results are considered robust.

Table 6. Placebo test results.

	(1)	(2)	(3)	(4)
	Sag	Dag	Sag	Dag
PHSR	0.028	−0.021	0.003	−0.013
	(0.033)	(0.016)	(0.030)	(0.017)
urban scale	0.147	−0.108 *	−0.088	−0.073
	(0.136)	(0.065)	(0.129)	(0.073)
wage level	0.103	−0.113 ***	0.116	−0.115 **
	(0.085)	(0.041)	(0.081)	(0.046)
Informatization level	−0.096 ***	0.017	−0.093 ***	0.018
	(0.022)	(0.011)	(0.029)	(0.016)
teacher	0.030	−0.036 *	0.379 ***	−0.228 ***
	(0.039)	(0.019)	(0.141)	(0.079)
openness	0.020 ***	−0.004 **	0.022 ***	−0.006 **
	(0.004)	(0.002)	(0.004)	(0.002)
_cons	−0.947	2.023 ***	0.230	1.871 ***
	(1.144)	(0.551)	(1.061)	(0.600)
city-fixed effect	yes	yes	yes	yes
time-fixed effect	yes	yes	yes	yes
N	364	364	308	308
R-squared	0.175	0.262	0.258	0.297

Standard errors in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the results of the placebo test. Columns (1) and (2) show the regression results of the full sample. Columns (3) and (4) show the sub-sample regression results, excluding the four core cities (Shanghai, Nanjing, Hangzhou, and Hefei). The results show that the coefficients of the PHSR are all statistically insignificant in the four regressions. This indicates that the influence of unobserved, potentially omitted variables on the regression results can be ignored. Thus, the baseline regression results are reliable.

3.4. Limitations and Future Work

After completing the empirical examination, we believe that there are still some limitations in this paper. With regard to the selection of control variables, we believe that it is more scientific to use the median value of employee wage level and the number of locally employed college students to represent the local wage level and the knowledge-spillover effect, respectively. We cannot obtain relevant data at present. However, we will continue to search for the above relevant data to improve the scientific validity of this study. Moreover, some studies show that HSR service frequency has a more obvious impact on the flow of production resources and service industry agglomeration than the opening of an HSR [11,51]. Therefore, this provides a direction for future research. We will further explore the impact of HSR service frequency on producer service industry agglomeration in future studies.

4. Concluding Remarks

In recent years, the rapid development of HSR service has not only shortened space–time distances between cities but also significantly affected the spatial reallocation of production resources. The rapid development of the producer service industry is an important way to optimize the spatial structure of the regional economy and promote its sustainable growth. The economic allocation effect caused by the opening of an HSR will have an important impact on the spatial layout of the producer service industry. Thus, it is of great strategic significance to explore the spatial effect of the HSR in the producer service industry for promoting its better and faster development. In this paper, based on the panel data of 26 cities in the Yangtze River Delta urban agglomeration from 2005 to 2018, we established a multi-phase DID model in order to investigate the impact of HSR opening on the spatial agglomeration of the whole producer service industry and its subdivision industries.

The research conclusions are summarized as follows: (1) The opening of an HSR has a significant positive effect on the specialized agglomeration of the producer service industry, while having a significant negative effect on the diversified agglomeration. (2) There exists significant industrial heterogeneity in the agglomeration effect of the producer service industry under HSR service. The financial industry belongs to the “highly significant promotion” industry, while the other four subdivision industries (“transportation, warehousing and postal service”, “information transmission, computer services and software”, “leasing and business services”, and “scientific research, technical services and geological prospecting”) belong to the “highly significant inhibition” industry.

Based on the above research conclusions, this paper draws the following policy implications:

(1)

The planning and construction of HSR lines should be closely integrated with the development of the producer service industry and the coordinated development of the regional economy. When planning HSR lines and setting up HSR stations, we should not only consider the regional transportation demand and regional development planning but also the regional resource endowment, industrial base, industrial agglomeration modes, and other factors in order to stimulate the development potential of the producer service industry as much as possible through the planning and construction of HSR lines and promote the coordinated development of the regional economy.

(2)

Differentiated industrial development strategies for different industries based on a reasonable evaluation of the HSR industrial effects should be formulated. When formulating development policies for different industries and selecting regional leading industries, the agglomeration or diffusion effects of the HSR on different industries should be fully considered in order to give full play to the comparative advantages of different industries, plan industrial layout reasonably, and maximize the benefits of industrial development.

(3)

Differentiated industrial development strategies for core and peripheral medium- and small-sized cities based on a reasonable evaluation of the HSR industrial effects should be formulated. Considering the different industrial effects of the HSR on core and peripheral medium- and small-sized cities, the government should encourage regions with advantageous location conditions to make full use of their own advantageous resources to build a spatial agglomeration highland of producer service industries and form an industrial development pattern of “point to area” in geographical spatial distribution.

(4)

The development of the producer service industry is not only affected by the HSR but also by other factors such as highways, railways, and civil aviation. Based on the empirical results of this study, the government should also give full consideration to other factors affecting producer service industry agglomeration besides the HSR so as to effectively promote the development of the producer service industry.