Does Regional Innovation Environment Have an Impact on the Gathering of Technological Talent? An Empirical Study Based on 31 Provinces in China

Abstract

Driven by innovation, the implementation of a strategy for developing a quality workforce is the key to promoting the high-quality development of China’s economy. Based on the panel data of 31 provinces on the Chinese mainland from 2013 to 2020, a spatial econometric model is used to explore the impact of the regional environment, regional heterogeneity and its spatial effect on the gathering of technological talent. The results show that: (1) The improvement of the regional innovation environment can significantly promote the gathering of technological talent in a region; (2) The regional innovation environment has an obvious regional effect on the gathering of technological talent, which is manifested in the strong promotion of the eastern and western regions, and an obvious siphon effect in the eastern region, but it is not the key factor affecting the gathering of technological talent in the central region; (3) The gathering of technological talent has a significant spatial effect among neighboring provinces. Therefore, improving the regional innovation environment, adapting measures to local conditions in different regions and strengthening economic cooperation among provinces have become the key to the rational allocation of technology talent resources.

1. Introduction

China has become the second largest economy in the world, and its total economic volume continues to increase. How to maintain the high quality of China’s economic development has become the priority at this stage. As a key factor of innovation-driven development, the regional innovation environment has an important impact on the agglomeration of technological talent. There are great differences in the degree of improvement of the innovation environment among provinces in China, and the degree of agglomeration of technological talent is different. The problem of unbalanced allocation caused by regional brain drain and the excessive concentration of talent is particularly prominent. Therefore, the relationship between the regional innovation environment and the spatial effect of various provinces in the gathering of technological talent still need to be further discussed.

By studying articles in the Web of Science and ScienceDirect databases, we categorize past articles according to three aspects: the importance of population quality, the factors influencing talent gathering, and the development of a regional innovation environment, so as to clarify the research objectives of this study. Scholars generally believe that the change in population plays an important role in a country’s economic development, which is mainly manifested in the fact that a sufficient population size can deepen the social division of labor, thus liberating productive forces and realizing economic growth [1,2]. However, with the growth in the educational level in various countries, improving the quality of the population and realizing the accumulation of human capital have become important ways to promote economic growth [3]. Paul M. LeVasseur also pointed out that the source of technological progress and economic growth comes from the overflow of human capital, which is characterized by the fact that workers have mastered special professional skills after receiving education [4]. Innovation-talent agglomeration is mainly manifested in the accumulation and qualitative change in innovation-talent elements in a certain space [5]. The agglomeration effect and spatial spillover effect produced by regional innovation-talent agglomeration are the key to improving regional innovation performance and economic development [6]. Technological talent is the product of innovative talent gathering, and its gathering level will directly affect the technical performance of a region and become a source of power for promoting regional economic innovation and development.

Previous studies have shown that the gathering of technological talent is affected by many factors. From the perspective of regional development, talent gathering is driven by the economic level, employment opportunities and development prospects. The level of economic development improves the national income and personal disposable income of the region, which enhances the attraction of technological talent [7,8,9]. However, with the process of modernization and changes in people’s own needs, the regional environment and welfare benefits, including employment opportunities and development prospects, have also become the key factors affecting the gathering of technological talent [10]. From the perspective of policy guidance, the government’s investment in regional innovation will promote regional technological innovation, thereby enhancing the agglomeration effect of regional talent [11,12,13]. From the perspective of micro enterprises, the R&D investment intensity and scientific research achievements within an enterprise can create a good enterprise-innovation environment and promote the gathering of technological talent within an enterprise [14,15]. From the perspective of the living environment, urban infrastructure, including the public service and urbanization level, influences the concentration of technological talent in the region to a certain extent [16]. The construction of the internet and other related innovation infrastructure has further promoted the willingness of scientific and technological personnel to stay [17,18]. Most scholars are committed to studying the impact of a single factor on the gathering of technological talent. Few scholars study the internal factors of the gathering of technological talent by building a comprehensive system of regional innovation environments from a multi-dimensional perspective.

Since 1990s, the development of human society has shown obvious regional characteristics [19], and the regional innovation environment has become key to research. The concept of a regional innovation environment was first put forward by the European Innovation Research Group (GREMI) in 1985. They defined the regional innovation environment as an informal and complex social relationship established by various innovative subjects in a certain region through joint action and learning, and this social relationship is conducive to innovation. After 40 years of development, the regional innovation environment has been continuously innovated in connotation and dimension. According to its connotation, a regional system, regional rules, and regional innovation practices that can induce and promote innovation are collectively referred to as the regional innovation environment [20]. Moreover, the regional innovation environment includes two dimensions: the micro-enterprise environment and macro-social environment. On the one hand, the innovation environment in micro-enterprises can be expressed as the technological capability and technological market environment of enterprises [21]. On the other hand, the macro-social environment, such as innovative resources, human capital, and infrastructure, which determines innovation activities, also affects the level of the regional innovation environment [22]. Based on the analysis of the factors influencing technological talent agglomeration by scholars, this study establishes a comprehensive evaluation index to measure the regional innovation environment from six perspectives: the policy innovation environment, R&D investment intensity, R&D achievements, innovation-market environment, innovation-culture environment and innovation-infrastructure environment, and studies the relationship of influence between the regional innovation environment and technological talent agglomeration from a comprehensive perspective.

On the one hand, the innovation environment causes the innovation process and innovation performance to show different characteristics in different regions [23], and the same regional policy will have different spatial impacts on different regions [24]. On the other hand, the gathering of technological talent also has an obvious spatial effect [25,26], which is mainly manifested in the demonstration effect, cooperation effect, learning effect and spillover effect formed by the accumulation and gathering of talent in the region, thus improving the regional innovation ability, economic growth and regional coordinated development. This study explores the spatial effect of technological talent in various provinces in China from two aspects. On the one hand, based on the differences in economic development and policy guidance in the east, central and western regions of China, the heterogeneity analysis of technological talent gathering in the east, central and western regions is carried out to explore the different factors influencing scientific and technological talent gathering; on the other hand, based on the possible siphon and spillover effects of technological talent gathering in various provinces and regions, a spatial econometric model is introduced to explore the spatial effects of technological talent gathering in the context of the regional innovation environment.

Taking 31 provinces on the Chinese mainland as the object, this paper studied the influence of the regional innovation environment on the agglomeration of scientific and technological talent, as well as the regional heterogeneity and spatial effect in each province. The marginal contribution of this study may be reflected in the following aspects: (1) Exploring the indicators that can measure the regional innovation environment from various aspects, establishing a comprehensive and reasonable indicator system and obtaining the comprehensive influencing factors and influencing paths that affect the gathering of technological talent; (2) On the basis of benchmark regression, considering the heterogeneity of different regions and the spatial effect among provinces, studying the different factors affecting the agglomeration of technological talent in different regions; (3) In terms of data processing, using the entropy method to determine the weight of the index system, which avoids the phenomenon of underestimation and overestimation caused by the difference of the impact of each index on the comprehensive index. The use of the location quotient to measure the degree of talent concentration avoids the difference in the number of talents caused by the different area and population density of each province. The follow-up structure of this paper is arranged as follows: the Section 2 is the theoretical framework and research hypothesis; the Section 3 is variable-description, data-source and model construction; the Section 4 is empirical analysis; and the Section 5 is the research conclusion and policy suggestions.

2. Theoretical Framework and Research Hypothesis

2.1. Regional Innovation Environment and Technological Talent Gathering

Talent-gathering force refers to the ability of a region to continuously attract talent, retain talent, form a talent-gathering state, and produce economic and social gathering effects. The regional innovation environment emphasizes the synergy generated by the innovation subject, collective efficiency and innovation behavior in the industrial zone. There is an inseparable relationship between the regional innovation environment and talent agglomeration. Schmidt believes that the regional innovation environment can influence the agglomeration mode of technological talent by promoting R&D project expenditure [27]. By means of a spatial econometric model, Jing and Chan Hailan calculated that the innovation environment is the key factor affecting the flow of technological talent [28]. On the one hand, the innovative environment enables talent to accumulate rich innovative knowledge and skills, and the effective transmission of skills promotes the agglomeration effect of technological talent [29,30]. On the other hand, the innovation environment can promote the agglomeration of technological talent within enterprises by promoting regional economic growth [31,32]. As far as the main body of scientific and technological innovation enterprises is concerned, creating a good innovation environment within the enterprise will improve the innovation efficiency of the enterprise, and thus take the popularity and influence of the enterprise higher. The brand effect generated by higher enterprise popularity and enterprise influence will further attract a large number of scientific and technological talents to gather in the enterprise. This study considers that the regional innovation environment can describe the improvement degree of the regional innovation environment from six perspectives: the policy-innovation environment, R&D investment intensity, R&D achievements, innovation-market environment, innovation-culture environment and innovation-infrastructure environment. Based on this, we propose Hypothesis 1:

Hypothesis 1 (H1).

Creating a good regional innovation environment can stimulate the innovation kinetic energy of enterprises in the region and promote the gathering of technological talent within enterprises.

2.2. Regional Heterogeneity of Technological Talent Gathering

There are significant differences in the concentration of technological talent in the eastern, central and western regions of China. The data show that by the end of 2021, the R&D full-time equivalent of the eastern region accounted for 70.60% of the 31 provinces on the Chinese mainland, the R&D full-time equivalent of the central region accounted for 21.09% of the 31 provinces on the Chinese mainland and the R&D full-time equivalent of the western region accounted for 8.30% of the 31 provinces on the Chinese mainland. The distribution of technological talent in the eastern, central and western regions is extremely uneven; the impact of the regional innovation environment on the eastern, central and western regions should also be heterogeneous. First, the difference of economic level in different regions leads to the different quality of urban development in different regions, and the quality of urban development significantly affects the choice of talent to settle down, resulting in heterogeneity [33]; Secondly, scholars usually divide the regional environmental characteristics attracting talent into soft and hard factors. Among them, the soft factors mainly include the urban cultural atmosphere, population diversity, inclusiveness and development, while the hard factors mainly include regional economic development level, employment opportunities and innovative market environment [34,35]. Different regions have different comparative advantages in the soft and hard factors of the regional environment, which leads to different abilities to attract talent in different regions [36]. Thirdly, the innovation policy, regulatory quality and legal rules in the region have a significant positive impact on the regional innovation capability [37,38], while different regions have different regional innovation-policy-design schemes and implementation quality, along with different innovation policies [39]. Based on the above three factors, combined with the basic national conditions of China’s regional development, this study holds that the regional innovation environment has an obvious heterogeneous influence on the gathering of technological talent in the eastern, central and western regions of China. The eastern region has a high level of economic development, which attracts the inflow of technological talent from various regions all year round. Therefore, the regional innovation environment is a positive driving force for the gathering of technological talent in the eastern region. However, due to its relatively perfect economic development and high talent saturation, its gathering strength is relatively weak. The economic development level of the central region is lower than that of the eastern region, but obviously higher than that of the western region, which has great development potential and prospects. Therefore, the regional innovation environment has a strong concentration of technological talent in the central region. However, due to the low level of economic development and the remoteness of the western region, most talent gradually gathers in the central and eastern regions, and the brain drain in the region is serious. Therefore, enhancing the perfection of the regional innovation environment has the greatest influence on the gathering of technological talent in the western region, and can maintain the stability of regional technological talent to the greatest extent. Based on this, we propose Hypothesis 2:

Hypothesis 2 (H2).

The degree of perfection of the regional innovation environment has an obvious regional effect on the gathering of technological talent, with the influence in the western region being higher than that in the eastern region, but it is not the key factor affecting the gathering of technological talent in the central region.

2.3. The Spatial Effect of Technological Talent Gathering

With the talent power strategy and the theory of high-quality economic development put forward, the number of technological talents in China’s provinces continues to increase, and the ability of talent-driven economic development continues to improve. However, there are great differences in talent allocation and talent-gathering momentum among China’s provinces. By the end of 2020, Guangdong’s full-time equivalent of R&D had reached nearly 700,000 person times, accounting for 20.23% of the 31 provinces in the mainland, while Tibet’s full-time equivalent of R&D is only 190 person times, accounting for 0.01% of the country. The huge regional gap reflects the unreasonable and unbalanced allocation of scientific and technological talent nationwide. Exploring the spatial effects among provinces to promote the rational allocation of technological talent has become the key to coordinated regional development at this stage.

As far as the innovation environment is concerned, it has obvious spatial autocorrelation characteristics. By the end of 2020, China’s national expenditure on science and technology has reached 580.186 billion RMB, of which more than 55% is concentrated in the eastern region, and China’s national R&D expenditure has reached 153 million RMB, of which more than 65% is concentrated in the eastern region, showing a phenomenon of “clustering” in space. This shows the spatial interaction between the perfection of the innovation environment in one region and that in other regions, that is, if a region’s regional innovation environment is relatively perfect, the regional innovation environment in the region closely related to its economy and geography may also be relatively perfect. On the contrary, if the perfection degree of the regional innovation environment in a certain region is low, the perfection degree of the regional innovation environment in a region that is closely related to its economy and geography may also be low. Jaakko and Philip defined this effect as the knowledge-spillover effect, and considered that knowledge spillover is the direct or indirect interaction and unconscious knowledge dissemination between different subjects [40]. Xu Xilei also affirmed this spillover effect, pointing out that if the intellectual capital structure of a core area and a peripheral area is similar, the spillover effect will be more obvious [41]. Therefore, exploring the spatial effect among provinces to promote the rational allocation of scientific and technological talent has become the key to regional coordinated development at this stage.

As far as technological talent is concerned, its agglomeration within a province will have siphon effect and spillover effect on neighboring provinces [42]. On the one hand, due to the siphon effect, the more developed provinces in the economic cluster can attract technological talent from neighboring provinces by improving the innovation environment and employment environment within the province, thus inhibiting the development potential of neighboring provinces and causing a severe brain drain [43,44]. On the other hand, due to the spillover effect, the more developed provinces in the economic cluster can carry out technological innovation cooperation with neighboring provinces to generate technological spillover, which will further promote the economic development of neighboring provinces. The less developed provinces in the economic cluster can also learn from the development model of developed provinces, improve the innovation environment in the province step by step and improve their talent gathering strength [45,46,47]. Based on this, we propose Hypothesis 3:

Hypothesis 3 (H3a).

The gathering of technological talent has an obvious spatial effect among neighboring provinces.

Hypothesis 3 (H3b).

Strengthening technological innovation cooperation among provinces is the key to the rational allocation of regional technological talent.

3. Variable Description, Data Source and Model Construction

3.1. Variable Description

(1) Explained variable:

Concentration of technological talent ($STcon$): In order to eliminate the difference of “agglomeration degree” caused by different population sizes in different provinces, using Chen Lianfang [48] for reference, we use the idea of a location quotient to define the agglomeration degree of technological talent. The specific definition method is as follows:

$$STco{n}_{i,t}=\frac{s{t}_{i,t}/s{t}_t}{po{p}_{i,t}/po{p}_t}$$

(1)

In Equation (1), $STco{n}_{i,t}$ indicates the concentration of technological talent in province i in year t; $s{t}_{i,t}$ indicates the R&D full-time equivalents of province i in year t; $s{t}_t$ represents the total number of R&D full-time equivalents of 31 provinces in mainland China in year t; $po{p}_{i,t}$ represents the year-end population of province i in year t; $po{p}_t$ refers to the total population of 31 provinces in mainland China at the end of the year t.

(2) Core explanatory variables: Comprehensive indicators of the regional innovation environment (inn). The regional innovation environment is a relatively stable network system formed by the long-term interaction of various elements within the region to enhance regional innovation capability. Wang Hongwei measured the level of the regional innovation environment in six respects: innovation resource-support environment, innovation subject-growth environment, innovation-output environment, system and governance environment, innovation cultural environment and economic and social development environment [19]. Zhang Na added the impact of technology spillovers to the regional innovation environment indicator system [49], Feng Xiaoxiao added education and network environment to the indicator system [50] and Ye Dan and Dai Shufen specifically pointed out the importance of the financial market environment in the indicator system [51,52]. On the basis of many evaluation index systems of the regional innovation capability, this study establishes a comprehensive evaluation index to measure the regional innovation environment in six respects: the policy innovation environment, R&D investment intensity, R&D achievements, innovation-market environment, innovation cultural environment and innovation-infrastructure environment, and determines the weight level of each aspect of the measurement index through the maximum entropy method. The index system structure and weight distribution are shown in

Table 1. Regional innovation environment indicator system.

Dimension	Target	Index	Unit	Weight
Regional innovation environment (inn)	Policy innovation environment	Proportion of technology expenditure in fiscal expenditure	%	0.1490
	R&D investment intensity	Ratio of R&D input to GDP	%	0.1554
		Level of new product development expenditure	-	0.1615
	R&D achievements	Number of valid patents per 10,000 people	individual	0.1181
	Innovative market environment	Technical market turnover per 10,000 people	100 million RMB	0.1166
	Innovative cultural environment	Public library holdings per capita	book	0.1459
	Innovative infrastructure environment	Number of computers used per hundred people	hundred	0.1535

:

The sub-indicators of the index system in Table 1 have been adjusted to the proportion and per capita form. The level of new product-development expenditure is measured by location quotient, and the construction method is as follows:

$$LN{P}_{i,t}=\frac{n{p}_{i,t}/n{p}_t}{R{D}_{i,t}/R{D}_t}$$

(2)

In Equation (2), $LN{P}_{i,t}$ indicates the level of new product-development expenditure in province i in year t; $n{p}_{i,t}$ indicates the expenditure on new product development in province i in year t; $n{p}_t$ represents the sum of expenditures for new product development in 31 provinces of mainland China in year t; $R{D}_{i,t}$ indicates the R&D expenditure of province i in year t; $R{D}_t$ indicates the sum of expenditures for new product development of 31 provinces in mainland China in year t.

(3) Relevant control variables: Considering that the residents’ living standard (inc), medical service level (med), education level (edu), economic development level (eco), employment environment level (une), cultural activity level (cul), public transport level (inf) and environmental greening level (gre) in provinces and regions will enhance the willingness of technological talent to stay, thus affecting the level of scientific and technological talent concentration, this study will include the above variables in the control variables of the model, improve its rationality.

3.2. Data Source

The China Statistical Yearbook does not have comprehensive statistical data on science and technology before 2013, but the eight consecutive years from 2013 to 2020 can largely show the change in talent. Based on the panel data of 31 provinces on the Chinese mainland from 2013 to 2020, this study empirically analyzes the impact of the regional innovation environment on the gathering of technological talent. The original data mainly come from the 2013–2021 China Statistical Yearbook and the 2013–2021 China Statistical Yearbook on Science and Technology. Part of the data is obtained by relevant calculations based on the original data. The descriptive statistics of data and variables are shown in

Table 2. Descriptive statistics of variables.

Variable Type	Variable Name	Metrics	Symbol	Max	Min	Standard Deviation	Mean	Sample Size
Explained variable	Concentration of technological talent	Full time equivalent location entropy of R&D in each province	$STcon$	3.440	0.007	0.779	0.807	248
Explanatory variable	Regional innovation environment	Comprehensive indicators of regional innovation environment	$inn$	7.711	0.608	1.181	1.880	248
Control variable	Living standard of residents	Per capita disposable income (10,000 RMB)	$inc$	7.223	0.975	1.121	2.511	248
	Medical service level	Health technical personnel per thousand population (person)	$med$	27.677	6.052	2.796	11.705	248
	Educational level	Number of ordinary high schools (hundred)	$edu$	10.350	0.290	2.247	4.372	248
	Economic development level	Per capita GDP (10,000 RMB)	$eco$	16.489	2.315	2.769	5.926	248
	Employment environment	Unemployment rate (%)	$une$	4.600	1.210	0.629	3.199	248
	Cultural and recreational activities	Number of art performances (10,000 sessions)	$cul$	42.440	0.000	6.126	3.395	248
	Public transport level	Own buses per ten thousand people (vehicles)	$inf$	26.554	6.201	2.971	12.872	248
	Environmental greening level	Park green space area per capita (m2)	$gre$	21.049	5.850	2.791	13.317	248

below:

3.3. Model Construction

3.3.1. Spatial Durbin Model

There may be a spatial effect of mutual influence among Chinese provinces. Technological talent will flow across neighboring provinces. Under the guidance of policies, the regional innovation environment of neighboring provinces may reflect the siphon absorption and spillover effect between regions. Therefore, this study constructs a spatial Durbin model to analyze the spatial effects of each province. The specific model is as follows:

$${STcon}_{i,t}={\beta }_0+\rho w{STcon}_{i,t}+{\beta }_{1}in{n}_{i,t-1}+{\displaystyle \sum }_{i=1}^8{\gamma }_i{X}_{i,t}+{\mu }_i+{\nu }_t+{\epsilon }_{i,t}$$

(3)

In Equation (3), ${STcon}_{i,t}$ indicates the concentration of technological talent in province i in year t; $in{n}_{i,t-1}$ is the core explanatory variable, representing the improvement degree of the regional innovation environment of province i in year t − 1; $X_{i,t}$ represents the control variables, specifically including residents’ living standard (inc), medical service level (med), education level (edu), economic development level (eco), employment environment level (une), cultural activity level (cul), public transport level (inf) and environmental greening level (gre); $w$ is 31 × 31 spatial weight matrix; $\rho$ is the spatial autocorrelation coefficient; and ${\epsilon }_{i,t}$ is a random disturbance term. In order to eliminate the influence of temporal and regional heterogeneity, this paper controls both temporal and regional effects in the spatial Durbin model. ${\mu }_i$ and ${\nu }_t$ are the regional fixed effect and time fixed effect.

In order to eliminate the endogenous problem between the gathering of technological talent and the regional innovation environment, we adjust the regional innovation environment by one period of lag. It is believed that the regional innovation environment in the year t − 1 may have an impact on the gathering of technological talent in year t, while the gathering of technological talent of technological talent in year t will not have an impact on the regional innovation environment in year t − 1. The endogenous problem that the explanatory variable and explained variable are mutually causal is effectively addressed.

3.3.2. Spatial Weight Matrix

When using the spatial model to empirically study the effect of spatial talent aggregation, it is necessary to determine an appropriate spatial weight matrix based on the relevance between regions. Most scholars focus on the proximity of regions, and use the 0–1 spatial weight matrix and the inverse distance matrix to measure the relevance of regions. However, the two matrices only measure the geographical spatial relevance of regions, and do not consider the economic relevance between regions. On the basis of building a 0–1 spatial weight matrix and an inverse distance matrix, this study introduces regional spatial relevance factors, establishes a distance and economic nested spatial weight matrix, and measures the impact of the regional innovation environment on the concentration of scientific and technological talent in the case of economic spillovers. The weight matrix of economic nested space is constructed as follows:

$$W_d=\{\begin{array}{c}\frac{1}{d_{ij}}, i\ne j\\ 0,i=j\end{array}$$

(4)

$$W=W_d·diag\left({\overline{G}}_1/\overline{G},{\overline{G}}_2/\overline{G},\dots ,{\overline{G}}_n/\overline{G},\right)$$

(5)

$${\overline{G}}_i={{\displaystyle \sum }}_{t_0}^{t_1}{G}_{i,t}/\left(t_1-t_0+1\right)$$

(6)

$$\overline{G}={{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{t_0}^{t_1}{G}_{i,t}/n·\left(t_1-t_0+1\right)$$

(7)

where $W_d$ is the inverse distance matrix, $d_{ij}$ is the geographic distance calculated according to the longitude and latitude data between provinces, $G_{i,t}$ represents the GDP of province i in year t, ${\overline{G}}_i$ refers to the average regional GDP of province i during 2013–2020, and $\overline{G}$ refers to the average annual GDP of 31 provinces on the Chinese mainland in the eight years from 2013 to 2020. $W$ is the weight matrix of economic nested space that introduces regional economic factors based on the inverse distance matrix.

4. Empirical Analysis

4.1. Spatial Correlation Analysis

The global Moran index is used to measure the spatial autocorrelation between regional variables, and to measure the spatial correlation intensity of provinces. The calculation formula is as follows:

$$\mathrm{I}=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{{\displaystyle \sum }}_{i=1}^n{\left(x_i-\overline{x}\right)}^2}$$

(8)

The value of Moran index I is between −1 and 1. If Moran index I is greater than 0, it means a positive correlation, that is, a high value adjacent to a high value, or a low value related to a low value; a Moran index I less than 0 indicates negative correlation, that is, a high value adjacent to a low value. If the Moran index I is close to 0, it indicates that the spatial distribution is random and there is no spatial autocorrelation.

Table 3. Moran index value.

Variable	Moran Index	p-Value
$STcon$	0.030 * (0.019)	0.040
inn	0.031 * (0.019)	0.033

Note: Robust standard error is shown in brackets, * p < 0.05.

shows the Moran index values calculated according to Equation (8) from 2013 to 2020:

By calculating the Moran index of the gathering degree of technological talent and the comprehensive index of the regional innovation environment in eight years from 2013 to 2020, we can find that the Moran index values of the technological talent concentration degree and the comprehensive index of the regional innovation environment are significantly positive. As a result, there is a clear spatial correlation between the degree of technological talent concentration and the regional innovation environment in all provinces of China, and it is necessary to build a corresponding spatial model for analysis under the condition of considering spatial dependence.

4.2. Model Selectivity Test

There are three choices for the spatial econometric model of panel data. One is whether the spatial Durbin model can be degenerated into a spatial error model and a spatial lag model. The other is the choice of fixed effects and random effects. The third is whether the panel data need to double fix time and space. In this study, the LM test, Hausman test and joint significance test were used to test the above three aspects, so as to select an appropriate spatial econometric model. The test results are shown in

Table 4. Model-selection test.

Inspection Method		Statistic	p-Value
LM inspection—Spatial error	Moran’s I	2.308	0.021
	Lagrange multiplier	3.458	0.063
	Robust Lagrange multiplier	2.162	0.141
LM inspection—Spatial lag	Lagrange multiplier	1.405	0.236
	Robust Lagrange multiplier	0.109	0.741
Hausman inspection	H0: Select random effect	67.07	0.000
Joint significance test	both and ind	H0: Select fixed region	0.001
	both and time	H0: Select fixed time	0.000

below. The LM test results show that the Lagrange multiplier of the spatial error model passed the significance test at the 10% level, but failed the robust Lagrange multiplier test, while the spatial lag model failed the LM test, indicating that the spatial Durbin model is the optimal choice. The results of the Hausman test show that statistics reject the original hypothesis of random effects at the 1% significance level, indicating that fixed effects are the best choice. The results of the joint significance test showed that the statistic accepted the fixed effect of time and region at the 1% significance level, so the two-way fixed effect of time and region became the best choice. Consequently, this study chooses the spatial Durbin model with fixed time and region to conduct an empirical analysis of the gathering of technological talent.

4.3. Spatial Regression Analysis

We compared the regression of the spatial error model, the spatial lag model and the spatial Durbin model, and the results are shown in

Table 5. Model regression results.

Variable	(1)	(2)	(3)
	SDM	SAR	SEM
$inn$	0.228 *** (0.027)	0.161 *** (0.027)	0.169 *** (0.027)
$inc$	−0.558 *** (0.073)	−0.569 *** (0.078)	−0.633 *** (0.083)
$med$	0.006 (0.008)	0.006 (0.009)	0.004 (0.009)
$edu$	0.010 (0.030)	−0.001 (0.031)	0.003 (0.031)
$eco$	0.041 ** (0.019)	0.104 *** (0.017)	0.107 *** (0.016)
$une$	−0.003 (0.029)	−0.033 (0.028)	−0.034 (0.029)
$cul$	0.006 ** (0.003)	0.008 ** (0.003)	0.008 ** (0.003)
$inf$	0.017 ** (0.007)	0.033 *** (0.007)	0.030 *** (0.007)
$gre$	0.017 (0.012)	−0.004 (0.010)	−0.004 (0.010)
Spatial autocorrelation coefficient (ρ)	−1.372 *** (0.296)
Space-lag parameter (λ)		−0.071 (0.182)
Spatial error parameters (β)			−0.610 ** (0.304)
Observations	248	248	248

Note: Robust standard error is shown in brackets, ** p < 0.01, *** p < 0.001.

below:

From the significant level of the spatial coefficients of the three models, it can be seen that compared with the other two models, the spatial Durbin model is more suitable and the fitting and regression effect is better. The spatial autocorrelation coefficient ($\rho$) under the Durbin model is significantly negative at the 1% significance level, which indicates that the technological talent and the regional innovation environment have an obvious negative spatial effects in 31 provinces of Chinese mainland, and H3a has been verified. The provinces with higher gathering of technological talent and higher regional innovation level will have a strong siphon effect on the gathering of technological talent and the regional innovation level of neighboring provinces, and the talent is continuously concentrated in the provinces with better regional development, and the brain-drain phenomenon in some provinces and cities is serious. Therefore, increasing the economic cooperation between neighboring provinces is of great significance to coordinated economic development, and H3b has been verified.

After considering the spatial effects among provinces, the regression coefficients of the comprehensive indicators of the regional innovation environment reported by the spatial error model, the spatial lag model and the spatial Durbin model are significantly positive, which indicates that the improvement of the regional innovation environment in China’s provinces has effectively promoted the agglomeration of technological talent into the region; H1 is verified. The reasons for this are as follows: First of all, the relatively complete regional innovation environment has increased the influence of provincial cities. Under the environment of incentives, more innovative enterprises gather in the province, and the increased demand for high-level labor has promoted the gathering of technological talent. Secondly, the complete regional innovation environment has increased the employment opportunities and welfare level of technological talent, increased the attractiveness of technological talent and finally promoted the gathering of talent.

What is worth mentioning is that among the control variables, the level of regional economic development is significantly positive, which indicates that technological talent pays more attention to the development potential and market opportunities of their regions when choosing their residences. A high quality-of-living standard has become the key to affect the gathering of technological talent. In addition, the level of the public transport environment and recreational activities is significantly positive, which further shows the importance of strengthening the construction of urban transport, culture and other infrastructure for attracting talent to stay.

5. Further Analysis

5.1. Robustness Check

The economic geography nested matrix comprehensively considers the economic and geographical distance factors, and gives weights to the elements of the inverse distance matrix based on the different regional economic levels. To test the robustness of the empirical model and change the weight matrix of the model further, the regression results of the economic geography nested matrix are compared with the regression results of the 0–1 matrix focusing on adjacent situations. The 0–1 matrix is constructed as follows:

$$W_{ij}=\{\begin{array}{c}1,i \mathrm{adjacent} \mathrm{to} j\\ 0,i is not \mathrm{adjacent} \mathrm{to} j\\ 0,i=j\end{array}$$

(9)

In Equation (9), i and j represent any two of the 31 provinces on the Chinese mainland. The two matrix regression results are shown in

Table 6. Robustness test result.

Variable	(1)	(2)
	Economic Geography Nested Matrix	0–1 Matrix
$inn$	0.228 *** (0.027)	0.151 *** (0.030)
$inc$	−0.558 *** (0.073)	−0.329 *** (0.085)
$med$	0.006 (0.008)	0.014 (0.009)
$edu$	0.010 (0.030)	0.014 (0.030)
$eco$	0.041 ** (0.019)	0.074 *** (0.017)
$une$	−0.003 (0.029)	−0.027 (0.029)
$cul$	0.006 ** (0.003)	0.007 ** (0.003)
$inf$	0.017 ** (0.007)	0.023 ** (0.008)
$gre$	0.017 (0.012)	−0.007 (0.011)
Spatial autocorrelation coefficient (ρ)	−1.372 *** (0.296)	0.066 (0.087)
Observations	248	248

Note: Robust standard error is shown in brackets, ** p < 0.01, *** p < 0.001.

below:

The influence coefficient of the core explanatory variable only has a slight change in value, and there is no significant positive and negative change. The rest of the control variables are consistent with the empirical results of the economic geography nested matrix, which proves that the above model is robust and the empirical results are also convincing. What is more, the improvement of the regional innovation environment really has a significant role in promoting the gathering of regional technological talent, and this promotion will have a siphon effect on the surrounding provinces with close geographical and economic ties.

5.2. Heterogeneity Analysis

Considering the huge differences in economic development, policy implementation and residents’ living standards among the eastern, central and western regions of China, this study regresses the data samples by region to study the impact of regional heterogeneity on the concentration of technological talent. The results of spatial measurement regression by region are shown in

Table 7. Regression results of regional heterogeneity.

Variable	(1)	(2)	(3)
	East	Center	West
$inn$	0.234 ** (0.076)	0.067 (0.101)	0.238 ** (0.094)
$inc$	0.215 (0.258)	0.103 (0.240)	0.075 (0.310)
$med$	0.010 (0.014)	0.032 ** (0.015)	−0.018 (0.011)
$edu$	−0.073 (0.168)	−0.037 (0.022)	0.112 *** (0.026)
$eco$	−0.079 ** (0.039)	0.133 *** (0.036)	0.033 (0.025)
$une$	0.063 (0.144)	0.060 * (0.035)	0.023 (0.019)
$cul$	0.011 ** (0.005)	−0.017 ** (0.007)	−0.024 *** (0.007)
$inf$	0.046 ** (0.020)	−0.007 (0.012)	0.027 *** (0.007)
$gre$	0.021 (0.039)	0.062 (0.016)	−0.001 (0.007)
Spatial autocorrelation coefficient (ρ)	−0.892 *** (0.246)	−1.410 *** (0.297)	−0.606 (0.376)
Observations	88	80	80

Note: Robust standard error is shown in brackets, * p < 0.05, ** p < 0.01, *** p < 0.001.

:

It can be seen from Table 7 that the regional innovation environment can promote the gathering of technological talent in the eastern and western regions, but it is not significant in the central region. Compared with the eastern region, the improvement of the innovation environment has a more obvious effect on the gathering of talent in the western region.

As far as the western region is concerned, due to the low level of regional economic development and the remoteness of the region, most of the talent is gradually concentrated in the central and eastern regions, and the brain drain within the region is serious. Therefore, the improvement of the regional innovation environment plays a significant role in maintaining the stability of technological talent in the western region and promoting the gathering of technological talent. Among all the control variables, the education level and public transport level in the western region are significantly positive, indicating that in the current development stage of the western region, infrastructure construction such as education and transport is the key to talent gathering in the western region. Therefore, in the future, the country should first focus on the construction of other aspects in the western region, such as increasing the construction of transportation and other infrastructure in the western region to reduce commuting time, increasing investment attraction and enterprise project introduction to increase the amount of technological employment in the western region, and increasing education investment and education construction to increase the number of local technological talents. Only by gradually improving the innovation environment in the western region on this basis can we effectively increase the willingness of technological talent to stay.

When it comes to the central region, its economic development level is higher than that of the western region but lower than that of the eastern region. The perfection of the regional innovation environment is not the key to the gathering of technological talent in the central region, and the development opportunities that technological talent have at a moderate cost of living may be more important when they choose to settle down. In all the control variables, the regional development level is significantly positive, suggesting that the technological talent in the central region pays more attention to the economic development prospect of the region in terms of employment-location selection. Therefore, in the future, the country should focus on increasing the economic construction of provinces and cities in the central region, give play to the advantages of provinces in the eastern region, promote economic cooperation between the eastern and central regions and enhance the economic strength of the central region.

As far as the eastern region is concerned, its economic development level is in a leading position in the whole country. Compared with the central and western regions, its innovation environment is relatively perfect, and it plays a strong role in promoting the gathering of technological talent. Compared with the western region, the higher level of economic development grants the eastern region a strong siphon effect on the surrounding provinces all year round, and the spatial effect of talent gathering is obvious. Among all control variables, the level of recreational activities and public transport are significantly positive, indicating that the technological talent pays more attention to the cultural and recreational activities within the region, and the demand for a certain spiritual and cultural level is the key to causing technological talent in the eastern region to choose to stay. In addition, due to the expansion of the population, the demand for public transport construction in the eastern region has also increased, which becomes an important factor restricting the gathering of technological talent in the eastern region. Therefore, in the future, the country should continue to increase the cultural construction in the eastern region, meet the growing spiritual and cultural needs of the people in the region, and increase regional cooperation to drive the coordinated development of the central and western regions. H2 has been verified.

6. Conclusions

6.1. Research Conclusion

Using the panel data of 31 provinces on the Chinese mainland from 2013 to 2020, this study analyzes the relationship between the regional innovation environment and technological talent agglomeration by controlling the spatial Durbin model of time and region, and demonstrates and analyzes the spatial effect of technological talent agglomeration in more developed provinces on adjacent provinces. In the further analysis, the data samples are divided into eastern, central and western regions, and the regional differences in the impact of the regional innovation on the gathering of technological talent are analyzed. Based on the above empirical research, the following three research conclusions are drawn:

(1) The improvement of the regional innovation environment can significantly promote the gathering of scientific and technological talent in the region.

(2) The regional innovation environment has an obvious regional effect on the agglomeration of technological talent, and has significant heterogeneity in the agglomeration of talent in the east, middle and west. Specifically, it plays a strong role in promoting the eastern and western regions, and the east region has an obvious siphon effect, but it is not the key factor affecting the gathering of technological talent in the central region. The key to promote the concentration of talent is formulating different talent-introduction policies for different regions.

(3) The gathering of technological talent among neighboring provinces has an obvious spatial effect. The existence of the siphon effect and spillover effect indicates that strengthening the technological innovation cooperation among provinces is the key to the rational allocation of regional scientific and technological talent and the promotion of regional coordinated development.

However, some limitations should be noted. First, the data sample size is not large enough. We selected only panel data from 2013 to 2020 as the research samples, and we could not find data before 2012 that matched the latitude of this research variables, so the selection of sample data still needs to be considered. Second, the solutions to endogenous problems still need to be optimized. We use one period of lag to solve the endogenous problems, and it seems that a more reasonable and optimized method can be created to solve the endogenous problems of the model. Thirdly, the latitude of the weight matrix is not comprehensive. We only consider the economic and geographical relationship between neighboring cities. In the next step, we can take the financial and network continuity between cities as the breakthrough point, and construct the corresponding weight matrix for empirical research. Finally, we hope that this paper may serve as a reference study and provide some instructional guidance.

6.2. Policy Suggestions

According to the above research conclusions, this paper proposes the following policy recommendations:

(1) Optimize policies, gradually improve the regional innovation environment, and enhance the attractiveness for technological talent. The government and enterprises should attach importance to the improvement and creation of the regional innovation environment to enhance the regional innovation vitality. First, increase investment in innovation funds, strengthen the construction of the regional innovation infrastructure, and create a good innovation environment for the region. Second, create a benign, competitive innovation-market environment, encourage enterprises in the region to compete in relevant innovative technology products and promote enterprises to carry out innovative research and development, and thus improve the attractiveness for technological talent. Thirdly, perform well in publicity for and education about innovation culture, encourage talent in the province to improve its innovation awareness and cultivate a new generation of scientific and technological talent.

(2) Implement different, targeted technological innovation policies for different regions, and promote the rational allocation of technological talent in the eastern, central and western regions. For the eastern region, with a relatively high economic level, efforts should be made to promote its cooperation and linkage with the central and western regions to stimulate the economic development of the central and western regions. For the central region, with strong development potential, efforts should be made to improve the regional innovation environment, seize policy opportunities and promote the attractiveness of the region to technological talent. For the western regions, with a weak development level, we should pay attention to strengthening the construction of infrastructure such as in transportation and education, gradually narrow the gap with the central and eastern regions, seize the historical opportunity of the western development and comprehensively enhance our talent attraction.

(3) Bearing in mind win-win cooperation, strengthen economic cooperation between regions and enhance the ability of coordinated development between regions. Strengthen the coordinated development and construction of urban agglomerations, increase the project linkage and cooperation between cities in developed provinces and cities in neighboring provinces, and steadily improve the reasonable allocation of human resources within urban agglomerations.