Spatial Agglomeration and Innovation Capacity: Evidence of Spatial Allocation of Construction Land Resources by Provincial Governments in China

Abstract

The Chinese government has proposed enhancing the land element guarantee capacity in advantageous regions. At the provincial level, this essentially means tilting and allocating construction land resources toward central cities with comparative advantages and increasing the spatial concentration of construction land (hereinafter referred to as “SCCL”) within the province, thereby maximizing the positive spatial agglomeration externalities of provincial central cities. Against the backdrop of China’s vigorous promotion of new-quality productive forces, whether SCCL can enhance innovation capacity (hereinafter referred to as “IC”) is a topic worthy of discussion. The share of the primate city is used to identify the degree of SCCL and the logarithm of patent grants per unit of built-up area to represent innovation capacity. Building on this foundation, this paper utilizes panel data from 23 provinces (autonomous regions) in China spanning 2000 to 2022 to examine the impact of SCCL on IC. The research finds the following: SCCL has an enhancing effect on IC. R&D investment, human capital, and marketization negatively moderate the relationship between SCCL and IC. The positive impact of SCCL on IC is enhanced with the increase in environmental regulation levels. The effect of SCCL on IC has a non-linear characteristic of “strengthening at both ends and collapsing in the middle.” The effect of SCCL on IC has spatial heterogeneity. Regarding different functional uses of construction land, only the spatial concentration of commercial and service land has a significant promoting effect on IC. This study provides provincial governments with decision-making support for enhancing IC through SCCL.

1. Introduction

Innovation serves as the primary engine driving sustainable development and the core impetus behind China’s current advancement of new-quality productive forces. From the perspective of government behavior, how to enhance regional innovation capacity (hereinafter termed “IC”) by optimizing resource allocation represents a critical issue for China, particularly as a developing nation. Land is one of the most important resources owned by the government. As a crucial factor of production and the carrier of all production activities, the allocation method of construction land undoubtedly determines the innovation potential of a region over a relatively long period [1]. From the perspective of the territorial spatial system, agricultural land, ecological land, and construction land mutually constrain one another. To safeguard food security and ecological security, China faces extreme scarcity in construction land resources. Consequently, the state implements an annual plan-based quota control system characterized by “central allocation preceding local allocation, total volume preceding structural allocation”. Once quotas are allocated from the central government to provincial governments, the latter are responsible for secondary distribution among their subordinate prefectural-level cities. In this process, on the one hand, provinces coordinate on the basis of economic scale, population capacity and farmland-protection obligations; on the other, under the stimulus of GDP-oriented performance evaluation, municipal governments vie “from the bottom up” for land with great intensity; ultimately, this process leads to a highly concentrated and uneven spatial distribution of construction land (Figure 1). This phenomenon has been termed “concentrated spatial allocation of construction land” in scholarly discussions [2], from which emerges the important concept of “the spatial agglomeration effect of land resources”. From one perspective, the scarcity of land resources necessitates concentrated spatial allocation to exert agglomeration effects. From another perspective, agglomeration is a key characteristic of spatial economic phenomena, making spatial agglomeration an essential feature that construction land resources should possess in their spatial allocation. Meanwhile, as mentioned earlier, construction land serves as the spatial carrier of human economic activities, and its spatial concentration allocation will lead to the spatial agglomeration of population, capital, and industrial economic activities. Unlike the spatial agglomeration of traditional factors such as physical capital and labor, which is formed through their own explicit spatial movement, the spatial agglomeration of construction land as a production factor is the result of allocation by provincial governments and spatial competition among municipal governments, with a relatively implicit character. Given that agglomeration theory holds that agglomeration influences innovation activities [3], a critical question emerges: From the perspective of provincial government actions, how will strengthening the concentrated spatial allocation of construction land and increasing the spatial concentration of construction land (hereinafter termed “SCCL”) affect IC?

On the one hand, scholars hold divergent views on agglomeration and its innovation effects, with much of the literature focusing on how agglomeration externalities shape IC [3]. The core and basic view is that agglomeration promotes the construction of formal and informal relationship networks among various entities and promotes innovation through knowledge spillovers, innovation diffusion, and reduced innovation costs. In comparison, a large number of studies have focused on industrial agglomeration. Within this framework, the literature can be divided into two branches. One branch mainly examines the effect of agglomeration on IC in terms of location entropy, including the agglomeration of manufacturing, producer services [4,5], and digital economy [6]. The other branch mainly examines the impact of agglomeration structure on regional and urban IC from the perspective of industrial diversification and specialization [7,8,9]. Some scholars have incorporated spatial factors into the analysis framework of agglomeration. One branch of the literature studies the relationship between economic density [10,11] and innovation, which reveals the geographical proximity of economic entities and affects innovation through interactive learning [12,13,14]. Related to this is another branch of the literature, which finds that firms in larger cities have higher IC [15] and that the spatial agglomeration of human capital [16] and high-tech enterprises [17] has a positive impact on regional innovation.

On the other hand, land is a fundamental element of innovation and production activities, and its utilization and allocation methods are closely linked to innovation. Existing literature mainly focuses on two aspects: one part of the literature examines the relationship with IC through the lens of urban land marketization, land financing, and land finance [18,19]; the other part of the literature focuses on the misallocation of land resources within cities (the allocation ratio of construction land between industrial and service land use) [20,21,22] to study its adverse effects on innovation.

Overall, prior research mainly concentrates on how functional land misallocation affects IC from the perspective of municipal government behavior at the prefecture level, with few studies examining the influence of the uneven allocation of construction land within a province on IC from the perspective of provincial government behavior. As mentioned earlier, innovation stems from the geographic concentration of economic or innovative entities, and the spatial concentration of construction land provides a material basis for the spatial concentration and geographic proximity of these entities. In fact, the spatial concentration of construction land is not simply the “spatial accumulation of land factors,” but rather the reallocation of construction land quotas to guide the agglomeration of innovative factors such as population, capital, technology, and public services to advantageous regions, thereby stimulating innovation. It is important to highlight that the fourth meeting of the Central Committee for Comprehensively Deepening Reform of the Communist Party of China emphasized the need to “enhance the ability of land factors to support high-quality development in advantageous regions.” One important manifestation of high-quality development in advantageous regions is innovative development, which involves increasing spatial innovation intensity or knowledge concentration per unit of land. At the provincial level in China, intensifying the concentrated allocation of construction land resources to advantaged regions is an inherent requirement of this mandate. Meanwhile, the Third Plenary Session of the 20th Central Committee of the Communist Party of China proposed to “improve the institutional mechanisms for developing new types of productive forces in accordance with local conditions,” which means forming production relations that are adapted to the quality of advanced productive forces. Promoting differentiated spatial allocation of land factors and enhancing the centralized allocation of scarce construction land resources to central cities with development advantages is a basic requirement. From a theoretical perspective, the spatial agglomeration of construction land within a province essentially reflects the spatial structure of construction land resources. Based on the structuralist methodology [23,24], optimizing the spatial structure of construction land within a province and enhancing the level of SCCL may be an important way to improve provincial IC.

Given the aforementioned analysis, this paper aims to investigate the influence of SCCL on IC in Chinese provinces. This study will provide new empirical evidence on the innovation effects of spatial agglomeration through the lens of SCCL, thereby offering decision-making references for provincial governments to enhance provincial-level innovation capacity through optimized allocation of construction land resources. In fact, the phenomenon of spatial concentration of construction land resources is a topic of common concern in regional/spatial economics and land resource management. Examining its impact on innovation capacity, on the one hand, enriches the research on the spatial allocation effects of construction land resources; on the other hand, it enriches the mechanisms driving innovation at the provincial level and further supplements decision-making references for optimizing construction land spatial allocation to boost provincial innovation. Building upon previous studies, this research makes the following contributions: (1) It studies the impact of spatial agglomeration of construction land resources formed by cross-city provincial construction land resource centralization on IC, which makes up for the deficiency in prior studies regarding how land allocation (misallocation) relates to IC. (2) It enriches the research on the relationship between spatial agglomeration and IC, with a focus on SCCL. (3) It reveals the environmental regulation threshold effect of construction land spatial allocation on IC. (4) It reveals the R&D investment substitution effect of construction land spatial allocation on IC. (5) It reveals the innovation effects of spatial concentration of different functional land uses.

2. Research Hypotheses

The spatial allocation of provincial construction land essentially optimizes and reshapes provincial spatial development rights by adjusting the spatial structure of the most basic land factors, thereby establishing new spatial production relations. Based on the structuralist methodology and the theoretical logic that “spatial production relations react on spatial productive forces”, the spatial agglomeration of resource factors brought about by this unbalanced centralized spatial allocation of construction land will influence the overall IC of the province.

From the viewpoint of the external innovation-friendly environment, the spatial agglomeration of innovation resource and economic activities is the key to stimulating innovation, and spatial agglomeration itself is a core feature of innovation activities. Large cities, with their “high-quality soil” for promoting innovation, naturally become regional innovation centers, thus, there is a coupling relationship between resource spatial agglomeration and regional innovation. However, a favorable innovation environment or the agglomeration of innovation resources depends on the effective coordination of government intervention and market laws.

In terms of the formation mechanism of spatial agglomeration, the spatial agglomeration of production factors such as population and capital is more dominated by market mechanisms, but such factor agglomeration bears a close connection to the spatial allocation of land factors [25]; the allocation of land factors is more subject to government intervention. Therefore, the degree of concentrated allocation of construction land to central cities will indirectly affect the factor agglomeration level of innovation centers, thereby playing an important role in provincial innovation output.

This paper briefly sorts out the analytical framework for the impact of SCCL on IC at the provincial level in China (Figure 2). Firstly, the SCCL drives IC through four microscopic-level paths and one macroscopic-level path. Secondly, R&D investment, human capital and marketization moderate the relationship between SCCL and IC. Thirdly, the intensity of environmental regulation constitutes a threshold moderating factor for the intensity of their interaction.

2.1. Direct Effects of SCCL on IC

At the micro level, SCCL enhances provincial IC through the R&D enclave effect (i.e., peripheral regions establishing R&D institutions in central cities), knowledge spillover effect, local market effect, and spatial competition effect. The detailed analysis is as follows:

Firstly, there is the R&D enclave effect. SCCL provides a broader space for resource allocation for innovative activities, driving the formation of R&D enclaves and creating innovation growth poles [26], thereby enhancing the IC of the province. The SCCL makes construction land quotas scarce in peripheral cities, leaving little space for high-cost, high-risk innovative activities. In contrast, central cities in the province have inherent advantages in innovation. The flow of construction land resources towards central cities provides them with relatively abundant land resources, facilitating the agglomeration of scientific and technological research and development activities. Consequently, peripheral cities will adopt an innovation model of “R&D elsewhere local transformation” to achieve development by leveraging external strengths. This cross-city innovation cooperation model triggered by SCCL will greatly enhance the level of innovation clusters and innovation growth poles for the entire provincial region, thereby improving its overall IC.

Secondly, there is the knowledge spillover effect. SCCL forces the flow of factors such as labor and capital, as well as industries, across cities [23]. Under the influence of geographical proximity, this also accelerates the spatial dissemination of technical knowledge [27]. On the one hand, the geographical agglomeration of innovative talents helps to increase the frequency of face-to-face communication and enhance the efficiency of tacit knowledge dissemination. On the other hand, the geographic concentration of related or unrelated industries promotes the spatial diffusion of industrial innovation knowledge. Firms acquire new technologies and knowledge through interactions with suppliers, customers, and other firms, which reduces the risks and R&D costs of corporate innovation, encourages firms to engage in technological innovation, and promotes inter-firm innovative cooperation and collaborative innovation.

Thirdly, there is the local market effect. SCCL will lead to the agglomeration of diverse-demand populations and industrial economic activities in central cities, forming a local market effect and stimulating innovation [28,29]. On the one hand, large-scale local diversified market demand can provide enterprises with a broad market space, which spurs enterprises to actively innovate, augment investment in R&D, develop new products and conduct research on new technologies to meet the diversified demands of the local market. At the same time, diversified market demand will also give birth to new enterprises and foster various new business forms. On the other hand, a large-scale market can provide enterprises with a basis for demand information for technological innovation. A large number of diverse local consumers can provide timely feedback to enterprises, and enterprises can improve their products based on the feedback information, thereby stimulating continuous innovation.

Fourthly, there is the spatial competition effect. SCCL means that these resources display a clear “core–periphery” distribution structure. This further intensifies the scarcity of construction land resources and, in turn, imposes a certain degree of spatial regulation on the layout of industrial enterprises, restricting their spatial mobility. Due to the limited availability of spatial resources, spatial competition compels enterprises to upgrade their industries and improve their land use efficiency. In this process, enterprises that exceed environmental pollution standards and have high carbon emissions, lack technological IC, and are unable to meet market demands will be forced to improve their production processes and enhance IC [30].

At the macro level, SCCL guides various resource factors and economic activities to agglomerate in central cities, and improves the province’s overall economic efficiency to further boost its IC. The detailed analysis is as follows:

The spatial centralized allocation of construction land affects IC through economic efficiency. Specifically, this allocation model—where construction land indicators are tilted toward provincial central cities—enables central cities to further leverage the mechanism of increasing returns to scale. Driven by the local market effect and the price index effect, it enhances the overall economic efficiency of the province. On one hand, improved economic efficiency reduces the input of resource factors such as labor and physical capital per unit of output, freeing up more resources to flow into the field of technological innovation and directly boosting IC. On the other hand, it provides more financial support for research and development (R&D), strengthening the sustainability of innovation investment. The growth of corporate profits and increased government tax revenue will feed back into R&D capital investment, promote the improvement of education levels, cultivate more high-quality R&D talents, and improve innovation infrastructure, as well as reduce the coordination costs of innovation. In addition, improved economic efficiency drives the growth of residents’ income and the industrial transformation and upgrading of the entire society. This fosters new demands, providing clear application scenarios and profit expectations for innovation.

Hypothesis 1.

SCCL promotes the overall IC of the province.

2.2. The Impact of SCCL on IC: The Moderating Role of R&D Investment and Marketization

With respect to the logic of the endogenous growth framework and the knowledge production function [31], R&D investments, including R&D funding and the quality of R&D labor force or human capital level, are direct drivers of innovation. However, from the perspective of agglomeration economy theory, the spatial agglomeration of factors and economic activities promotes innovation through mechanisms such as learning and competition. Given the basic fact of resource scarcity and the bias towards central cities, SCCL, on the one hand, guides the geographical agglomeration of innovation entities, forming external economies of scale in innovation space and saving R&D funding costs. On the other hand, it gives rise to the geographical concentration of industrial economic activities, which, under the influence of the local market effect and spatial competition effect, induces innovation, saves R&D investment costs, and improves R&D efficiency.

From the angle of the government versus the market, SCCL reveals the rational choice made by an effective government in accordance with economic laws, which will guide the spatial agglomeration of resources such as population, thereby promoting innovation. Meanwhile, marketization, by optimizing resource allocation and improving efficiency, also benefits innovation. However, marketization can bring negative effects. For instance, local governments, seeking to maximize fiscal revenue, may over-rely on land conveyance income, leading to excessive concentration of land resources in high-yield sectors like real estate development, while neglecting support for innovation industries. City governments, aiming to boost short-term economic growth and attract enterprises, often allocate large quantities of construction land quotas to industry at low prices. Yet, this model may prove unsustainable for supporting innovative development. Simultaneously, the marketization process drives up land prices, which in turn causes housing prices to rise. High housing costs increase the living expenses of innovative talent, reducing their willingness and capacity to innovate. Businesses may also become more inclined to pursue short-term profits rather than invest resources in innovation. These negative effects dilute the positive impact that SCCL has on IC.

Hypothesis 2.

R&D investment and marketization negatively moderate the promotional effect of SCCL on IC.

2.3. The Environmental Regulation Threshold Effect of SCCL on IC

Based on the Porter Hypothesis, moderate environmental regulation promotes enterprise technological innovation [32]. Some studies have found that a relatively higher level of environmental regulation reduces the constraining effect of improper land resource allocation on green technological innovation [33]. In fact, differences in the intensity of environmental regulation may also cause the impact of SCCL on the innovation capacity of a province to exhibit non-linear characteristics. SCCL reflects spatial regulation of economic activities, which forces a large number of enterprises to concentrate spatially. When the overall environmental regulation intensity of a province is low, the market entry threshold for pollution-intensive enterprises is relatively low, and enterprises’ motivation to engage in green technological innovation is insufficient, thereby weakening the agglomeration innovation effect generated by the spatial concentration of construction land resources. As environmental regulation reaches a high intensity, the market entry threshold for pollution-intensive enterprises increases. With the aim of surviving in the market, these enterprises are forced to carry out technological innovation. Under the influence of geographical proximity, this helps to generate knowledge spillover effects, reduce innovation costs and risks, and enhance the province’s IC.

Hypothesis 3.

The impact of SCCL on IC is subject to an environmental regulation threshold effect.

3. Research Design

3.1. Establishment of Empirical Model

3.1.1. Dual Fixed Effects Panel Model

Based on Hypothesis 1, this paper constructs a static panel model as shown in Equation (1) to empirically test the impact of SCCL in Chinese provinces on IC. The specific model is as follows:

$$\ln I_{it}=a_0+a_1{S}_{it}+{\displaystyle {\sum }_{k=1}^m{\varphi }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(1)

Herein, i and t are subscripts denoting the provincial-level region and the corresponding year, respectively, which applies to the following as well; lnI is the dependent variable, representing the IC of the province; S is the core explanatory variable, representing the level of SCCL; X_k represents a set of control variables, with ϕ_k denoting their coefficients; α₀ is the intercept term; α₁ is the coefficient of the explanatory variable; μ_i and τ_t denote individual and time fixed effects, respectively; ε_it denotes the random error term.

3.1.2. Four-Stage Mediation Effect Model

Taking into account the transmission mechanism analysis of Hypothesis 1 and the data availability, in the empirical research section, this paper focuses on exploring the economic efficiency improvement mechanism through which SCCL affects IC. The four-stage mediating effect test method is employed, with Models (2) to (4) constructed to test the economic efficiency mediating mechanisms underlying the impact of SCCL on IC, where E represents economic efficiency (ee). Other variable notations are the same as above.

$$E_{it}=b_0+b_1{S}_{it}+{\displaystyle \sum _{k=1}^{\mathrm{m}}{\theta }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(2)

$$\ln I_{it}={\beta }_0+{\beta }_1{E}_{it}+{\displaystyle \sum _{k=1}^m{\varphi }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(3)

$$\ln I_{it}=c_0+c_1{S}_{it}+c_2{E}_{it}+{\displaystyle \sum _{k=1}^m{\varphi }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(4)

Equation (2) serves to test the impact of SCCL on the mediating variable (economic efficiency). If the impact is significant, proceed to the next step. Equation (3) is employed to test the influence of the mediating variable (economic efficiency) on IC. As above, where the impact is significant, we progress to the next stage. Equation (4) incorporates both SCCL and the mediating variable into the determinants of IC. In the empirical analysis, we observe the changes in the impact of SCCL on IC (lnI) before and after the mediating variable is included. Per the mediation effect model principles, a mediating effect may exist if the following basic conditions are satisfied: coefficients b₁, β₁, and c₂ are all significant; the product of b₁ and c₂ has the same sign as c₁; and coefficient c₁ is smaller than α₁ or its significance level decreases. On this basis, the Sobel test and Bootstrap test are further conducted.

3.1.3. Moderation Effect Model

Based on Hypothesis 2, the moderating effect model is established as illustrated in Equations (5) and (6) to identify the role of the moderator in the process through which the SCCL at the provincial level affects IC.

$$\ln I_{it}=a_0+a_1{S}_{it}+a_2{T}_{it}+{\displaystyle {\sum }_{k=1}^m{\varphi }_k}{X}_{kit}\hspace{0.17em}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(5)

$$\ln I_{it}={\rho }_0+{\rho }_1{S}_{it}+{\rho }_2{T}_{it}+{\rho }_3(S_{it}\times T_{it})+{\displaystyle {\sum }_{k=1}^m{\varphi }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(6)

Equation (5) is the benchmark regression model for the impact of SCCL and the moderator variable (T) on provincial IC. Equation (6) introduces an interaction term (S × T) based on Model (5), where both S and T in the interaction term have been centered. The ρ₃ is coefficient of the interaction term, which is used to identify the moderating role of T in the process through which SCCL affects IC. In the regression results, if ρ₃ is significant, then the moderating effect is significant. If ρ₃ has the same positive or negative sign as ρ₁, it indicates that the T enhances the effect of SCCL on lnI. If ρ₃ has the opposite sign to ρ₁, this result indicates that T significantly negatively moderates the relationship between SCCL and IC.

3.1.4. Threshold Effect Model

Based on Hypothesis 3, this paper employs a double threshold model as shown in Equation (7) to examine the non-linear characteristics of SCCL on IC under different environmental regulation thresholds.

$$\ln I_{it}=a_0+{\eta }_1{S}_{it}\cdot I(\ln e{r}_{it}\le \gamma )\hspace{0.17em}\hspace{0.17em}+{\eta }_2{S}_{it}\cdot I(\ln e{r}_{it}>\gamma )\hspace{0.17em}+{\displaystyle {\sum }_{k=1}^m{\varphi }_k}{X}_{kit}+{\mu }_i+{\tau }_t+{\epsilon }_{it}$$

(7)

where lner is the threshold variable, representing the intensity of environmental regulation. γ denotes the threshold value awaiting estimation. I(⋅) is the indicator function. η₁ and η₂ correspond to the impact coefficients of the SCCL on IC under different levels of environmental regulation, respectively. For the other variables and symbols, their meanings are the same as previously stated.

3.2. Variable Description

3.2.1. Explained Variable

Innovation capacity (lnI). Generally, the indicators for measuring IC include the number of patent applications and the number of patent grants. This paper adopts the number of patent applications. For robustness tests, the number of patent grants is used. To eliminate the impact of the scale of construction land in provinces, this paper uses the logarithm of the number of patent applications per unit of built-up area to represent it. Naturally, this study adopts this approach for additional considerations: if the logarithm of patent applications (or grants) is used to characterize the explained variable, and given that the core explanatory variable is the SCCL, the control variables must include the logarithm of the provincial construction land scale (ln(∑Area_iRt)). To simplify the model, this paper moves the factor of the logarithm of construction land scale from the right-hand side to the left-hand side of the model, thereby obtaining the current explained variable.

3.2.2. Explanatory Variable

The spatial concentration of construction land (SCCL), SCCL, not only reflects the extent of uneven or unequal distribution of construction land resources across cities by provincial governments, but also identifies the inherent spatial agglomeration phenomenon of construction land as a special resource, which is subject to provincial government intervention.

In regional economics, there are many indicators to characterize the spatial inequality or spatial imbalance, such as the Spatial Gini Coefficient, Location Quotient, the share of the primate city, Herfindahl–Hirschman Index (hereinafter abbreviated to as “HHI”), and Zipf Index. This paper focuses on the spatial centralized allocation (spatial agglomeration) of construction land within a provincial jurisdiction, or the degree of unbalanced distribution of construction land resources inside a province. Therefore, the Spatial Gini Coefficient and Location Quotient are not suitable for identifying this degree of unbalanced distribution, while the primacy city share, HHI, and Zipf Index can effectively fulfill this purpose.

Among these indicators, the share of the primate city, which measures the SCCL within a province, is relatively simple and straightforward. Specifically, a higher share of construction land in the largest city is associated with a stronger spatial agglomeration of construction land resources within the province. In contrast, the Herfindahl Index and Zipf Index are relatively comprehensive and somewhat complex indicators. In other words, these two indicators can explain the information related to spatial agglomeration in a more comprehensive manner. They not only take into account the primacy city share but also the shares of non-primate cities. Higher values reflect a stronger degree of spatial agglomeration. From a statistical perspective, the primacy city share is a principal component of the HHI and Zipf Index. Regarding the identification approaches, the share of the primate city and the HHI adopt non-parametric estimation methods, while the Zipf Index employs a parametric estimation method. From the perspective of policy implications, the primacy city share has more direct policy relevance. Decision-makers can use this indicator to quickly determine whether construction land resources within a province are excessively tilted toward the provincial capital or core cities, thereby providing clear quantitative basis for formulating policies related to coordinated provincial development.

Considering the above-mentioned analysis, the share of the primate city is mainly used to identify SCCL in this study. A larger value signifies a more distinct degree of SCCL. The specific calculation is shown in Equation (8), where pr refers to the percentage of urban construction land in the primary city, Area_i1t denotes the built-up area of the primary city, and Area_iRt denotes the urban built-up area of the r-th ranked city within the province.

$$p{r}_{it}={\displaystyle \frac{\max (Are{a}_{i1t},Are{a}_{i2t},\dots ,Are{a}_{\mathrm{int}})}{{\displaystyle \sum _{R=1}^{n}Are{a}_{iRt}}}}={\displaystyle \frac{Are{a}_{i1t}}{{\displaystyle \sum _{R=1}^{n}Are{a}_{iRt}}}}$$

(8)

3.2.3. Mechanism Variables

Moderating Variables: R&D intensity (rd), indicated by the share of provincial R&D expenditure in GDP; human capital level (edu), represented by the weighted sum of the proportion of educational attainment of the population aged 6 and above, with educational years serving as the weight; marketization (market), where the degree of marketization is measured by the Fan Gang Marketization Index [34]. Threshold Variable: Environmental regulation intensity (lner), represented by the logarithm of the investment amount for industrial pollution control per unit of industrial value added. Mediating Variable: Economic Efficiency (ee), characterized by real GDP per unit of provincial construction land area, where provincial actual GDP is calculated using 1999 as the price base year.

3.2.4. Control Variables

Based on this study, the measurement methods for the control variables are as follows: R&D intensity (rd), human capital level (edu), environmental regulation intensity (lner), marketization level (market): measured using the same methodologies described previously. Foreign Direct Investment (fdi) is characterized by the real FDI per unit of provincial construction land area, converted into RMB at the current exchange rate. Specifically, the FDI converted into RMB is deflated using the GDP deflator with 1999 as the base year to eliminate the impact of price factors, resulting in the real FDI. Urbanization level (urb): defined as the ratio of urban population to the entire population. Industrial agglomeration (agg): calculated as the ratio of a province’s industrial added value to its GDP, divided by the ratio of the nation’s industrial added value to GDP.

3.3. Data Description

This study calculates the annual provincial SCCL, encompassing 23 provinces and 273 cities. The distribution of the number of cities in each province is presented in Figure 3. Specifically, this study initially excludes China’s four centrally administered municipalities (Beijing, Tianjin, Shanghai, and Chongqing) and the Hong Kong and Macao regions. Second, it excludes provinces and autonomous regions where the number of cities is small or data is seriously missing, including Hainan Province, Qinghai Province, Taiwan Province, Xinjiang Uygur Autonomous Region, and Tibet Autonomous Region. Data are collected over the period 2000–2022, with a total of 529 observational samples. The innovation data are derived from the “Chinese Patent Database” under the State Intellectual Property Office (SIPO). The data on built-up area are from the “China Urban Construction Statistical Yearbook.” The data for the remaining variables mainly come from the “China Statistical Yearbook,” “China Industrial Statistical Yearbook,” “China Environmental Statistics Yearbook,” and the CSMAR Database in China. The basic statistics of the variables adopted in this study are presented in

Table 1. The descriptive statistical information of the variable.

Variable	Definition	Sample Size	Mean	Standard Deviation	Minimum	Maximum
lnI	Innovation Capacity	529	2.750	1.242	0.118	5.391
pr	The primacy city share based on construction land	529	0.322	0.127	0.134	0.721
rd	R&D Intensity	529	1.531	1.200	0.103	7.042
edu	Human Capital Level	529	8.490	0.879	5.438	10.174
market	Marketization level	529	6.959	2.042	2.813	12.390
fdi	Foreign Direct Investment	529	1.123	0.634	0.037	2.333
urb	Urbanization Level	529	0.496	0.130	0.167	0.786
agg	Industrial Agglomeration Level	529	0.982	0.143	0.588	1.501
lner	Environmental Regulation Intensity	529	1.914	0.633	0.940	3.585

.

4. Empirical Results and Analysis

4.1. Basic Facts

Figure 4 reports the temporal evolution characteristics of the SCCL and innovation capacity (IC) in Chinese provinces. It can be observed that the two exhibit a clear co-evolutionary trend. Over time, as the level of SCCL continuously increases, IC also shows a trend of continuous improvement.

Figure 5 further presents the temporal evolution of the relationship between the SCCL and IC across different provinces. First, IC generally shows an upward trend, with its fluctuation range being larger than that of the SCCL. In recent years, the SCCL has shown a downward trend in Hebei, Shanxi, Inner Mongolia, Jiangsu, Guangdong, and Guangxi, while most provinces have experienced an upward trend.

Second, regarding the co-evolution of the SCCL and IC in most provinces, such as Liaoning, Jilin, Hei longjiang, Anhui, Hunan, Sichuan, Guizhou, Shaanxi, Shandong, and Guangxi, they show a co-evolutionary trend. In a few provinces, such as Hebei, Inner Mongolia, Jiangsu, and Guangdong, the SCCL and IC move in opposite directions. In some provinces, there is a complex relationship between the SCCL and IC, with both co-evolutionary and counter-evolutionary trends. For example, in Shanxi Province, the two indicators showed a co-evolutionary trend from 2000 to 2014, but a counter-evolutionary trend after 2014. In Zhejiang Province, there was a certain counter-evolutionary trend before 2015, but a co-evolutionary trend after 2015. In Gansu Province, there was a counter-evolutionary trend before 2015, but a co-evolutionary trend after 2015. In Yunnan Province, the two indicators showed a co-evolutionary trend from 2000 to 2006 and from 2018 to 2022.

4.2. Benchmark Regression

Based on the stepwise regression method, the impact of SCCL on IC is estimated using dual fixed effects model, with the results presented in

Table 2. Baseline estimates.

Variable	(1) lnI	(2) lnI	(3) lnI	(4) lnI	(5) lnI	(6) lnI	(7) lnI	(8) lnI
pr	1.026 ***	1.064 ***	1.370 ***	1.235 ***	1.279 ***	1.301 ***	1.223 ***	1.086 ***
	(3.00)	(3.13)	(3.94)	(3.61)	(3.66)	(3.74)	(3.55)	(3.00)
rd		0.132 ***	0.120 ***	0.127 ***	0.121 ***	0.115 **	0.154 ***	0.156 ***
		(2.92)	(2.67)	(2.88)	(2.68)	(2.57)	(3.37)	(3.41)
edu			0.253 ***	0.246 ***	0.241 ***	0.241 ***	0.306 ***	0.309 ***
			(3.46)	(3.44)	(3.33)	(3.36)	(4.16)	(4.21)
fdi				0.099 ***	0.098 ***	0.081 ***	0.086 ***	0.078 ***
				(4.50)	(4.46)	(3.61)	(3.83)	(3.35)
urb					0.155	0.141	0.270	0.256
					(0.61)	(0.56)	(1.07)	(1.02)
agg						0.364 ***	0.817 ***	0.801 ***
						(2.85)	(4.44)	(4.34)
lner							0.351 ***	0.351 ***
							(3.39)	(3.39)
lnmarket								0.201
								(1.21)
Constant	1.055 ***	0.929 ***	−0.922 *	−0.826	−0.851	−1.198 **	−2.820 ***	−3.067 ***
	(8.80)	(7.34)	(−1.68)	(−1.53)	(−1.57)	(−2.18)	(−3.89)	(−4.07)
Time and individual effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.938	0.939	0.940	0.943	0.943	0.944	0.945	0.945

Note: Parenthetical values correspond to t-statistics; *, **, and *** signify significance at the 10%, 5%, and 1% levels, respectively.

. Column (1) shows that, without considering other control variables, The variable pr shows a significantly positive coefficient at the 1% level. This preliminarily indicates that SCCL can help to enhance the overall IC of a province. Based on this, other control variables were gradually included. As observed, the coefficient of pr continues to be significantly positive at the 1% significance level. Its effect on IC is stable, indicating that the concentration of construction land resources in provincial central cities can effectively enhance the overall IC of the province. Thus, Hypothesis 1 is established.

4.3. Robustness Tests

4.3.1. Replacing the Core Explanatory Variable

The SCCL reveals the essence of spatially unbalanced distribution of construction land resources. Without loss of generality, this paper employs the Zipf Index (q) on one hand and the Herfindahl–Hirschman Index (HHI) on the other to identify SCCL.

The calculation formula for the HHI is shown in Equation (9). The specific implications of relevant symbols and variables are detailed in Equation (8) of this study.

$$\begin{array}{l}HH{I}_{it}={\left(Are{a}_{i1t}/{\displaystyle \sum _{R=1}^{n}Are{a}_{iRt}}\right)}^2+{\left(Are{a}_{i2t}/{\displaystyle \sum _{R=1}^{n}Are{a}_{irt}}\right)}^2+\dots +{\left(Are{a}_{int}/{\displaystyle \sum _{R=1}^{n}Are{a}_{int}}\right)}^2\\ \hspace{0.33em}\hspace{1em}\hspace{1em}\hspace{0.17em}={\displaystyle \sum _{R=1}^n{\left(Are{a}_{iRt}/{\displaystyle \sum _{R=1}^{n}Are{a}_{iRt}}\right)}^2}\end{array}$$

(9)

As a key tool to disclose the imbalanced spatial distribution of resource factors, the power law encompasses typical mathematical power functions such as Zipf’s Law. Its detailed expression is given in Equation (10).

$$Are{a}_{iRt}=Are{a}_{i1t}\times {R_{iRt}}^{-q_{it}}$$

(10)

Here, q is referred to as the Zipf Index, which is greater than 0. In this paper, Area_iRt represents the scale of construction land with a size rank of R. Among the parameters, Areat_i1t, R, and Area_iRt are known values. Logarithm transformation of both sides of Equation (10) results in Equation (11).

$$\ln Are{a}_{iRt}=\ln Are{a}_{i1t}-q_{it}\ln R_{iRt}$$

(11)

The ordinary least squares (OLS) method is applied to calculate parameter q. As q increases, the concentration of the annual spatial allocation of provincial construction land becomes higher.

Columns (1)–(2) of

Table 3. Robustness test.

Variable	(1) lnI	(2) lnI	(3) lnI	(4) lnI	(5) lnI_g	(6) lnI_g	(7) lnI_g	(8) lnI_g	(9) lnI_g	(10) lnI_g
HHI	1.648 ***	1.756 ***					1.744 ***	1.709 ***
	(3.47)	(3.65)					(3.81)	(3.70)
q			0.529 ***	0.779 ***					0.691 ***	0.915 ***
			(3.18)	(4.87)					(4.39)	(6.10)
pr					0.854 **	0.855 **
					(2.57)	(2.46)
Control Variable	NO	Yes	NO	Yes	NO	Yes	NO	Yes	NO	Yes
Time effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	1.069 ***	−3.158 ***	0.894 ***	−1.703 ***	0.974 ***	−3.221 ***	0.917 ***	−3.381 ***	0.614 ***	−2.264 ***
	(10.26)	(−4.21)	(5.56)	(−2.85)	(8.38)	(−4.45)	(9.13)	(−4.71)	(4.02)	(−4.05)
R2	0.938	0.946	0.914	0.928	0.937	0.945	0.938	0.946	0.908	0.924

Note: Parenthetical values correspond to t-statistics; ** and *** signify significance at the 5% and 1% levels, respectively.

report the estimation results of the HHI, while columns (3)–(4) of Table 3 report the estimation results of the q. Whether other variables are controlled or not, the sign of HHI and q remains significantly positive at the 1% level.

4.3.2. Replacing the Dependent Variable

The dependent variable is replaced by the logarithm of the number of patents granted per unit of urban built-up area (lnI_g). In Table 3, columns (5) and (6) report the estimation findings, where the variable pr shows a positive and statistically significant result at the 5% level with or without controls.

4.3.3. Replacing Both the Dependent and Core Explanatory Variables

This paper combines the above two robustness testing methods and simultaneously replaces both the core explanatory variable and the dependent variable. The estimation results are shown in columns (7) to (10) of Table 3. Regardless of whether other variables are controlled or not, both q and HHI exhibit significant positive effects at the 1% level.

In summary, regardless of whether the explanatory variable, the explained variable, or both the explanatory and explained variables are changed, and whether other variables are controlled for or not, SCCL significantly enhances IC at least at the 5% level. This indicates that the conclusion of the baseline regression is robust.

4.4. Endogeneity Discussion

This paper employs instrumental variable (hereinafter termed “IV”) estimation to resolve potential endogeneity problems such as omission of important explanatory variables and reverse causality. Drawing on previous literature, we construct an IV for SCCL by interacting the variables of topographic relief [35], river density distribution [36], exchange rate [37] and the reciprocal of linguistic diversity [38]. The justifications for selecting IV are set out as follows:

Topographic relief and river density distribution, as geographical variables, have natural exogeneity. Linguistic diversity, with its historical objectivity, can objectively describe regional characteristics. Areas with high topographic relief tend to have higher development costs, leading to a concentration of urban construction in flat areas, thereby promoting the spatial concentration of construction land. Cities with high river density usually have larger populations and more concentrated construction land resources. Linguistic diversity reflects the degree of regional segmentation; the more severe the segmentation, the more dispersed the construction land resources. In summary, linguistic diversity, topographic relief, and river density maintain a close relationship with SCCL, complying with the relevance assumption. Simultaneously, topographic relief and river density distribution are clearly unrelated to innovation. Although dialect diversity may have a weak correlation with innovation, it is an outcome of historical development and bears no inevitable association with current IC, thus satisfying the exogeneity assumption.

As topographic relief, river density, and dialect diversity from geographical and historical dimensions are time-constant, a time-varying variable is necessary. The exchange rate, an exogenous shock at the macro level, is employed as an IV, and a positive association exists between the exchange rate and the agglomeration of economic activities [37].

Table 4. Instrumental variable (IV) estimation.

Variable	(1) pr	(2) lnI	(3) lnI_g	(4) HHI	(5) lnI	(6) lnI_g	(7) q	(8) lnI	(9) lnI_g
IV	0.117 ***			0.111 ***			0.333 ***
	(7.38)			(9.76)			(8.49)
pr		4.809 ***	5.206 ***
		(3.85)	(4.14)
HHI					5.038 ***	5.454 ***
					(4.08)	(4.52)
q								1.828 ***	1.925 ***
								(4.16)	(4.60)
Control Variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	0.292 **	−3.099 ***	−3.831 ***	0.187 **	−2.636 **	−3.330 ***	0.956 ***	−2.936 **	−3.740 ****
	(2.48)	(−2.61)	(−3.22)	(2.20)	(−2.45)	(−3.17)	(3.21)	(−2.41)	(−3.22)
Time and Individual effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.367	0.933	0.927	0.309	0.940	0.939	0.232	0.929	0.924
Anderson LM	51.892			84.330			65.777
Cragg-Donald Wald F	54.394			95.195			72.081

Note: Parenthetical values correspond to t-statistics; ** and *** signify significance at the 5% and 1% levels, respectively. Due to space constraints, the estimation results for the control variables are omitted.

reports findings estimated by the IV estimation, where the core explanatory variables are the primacy city share, the Herfindahl–Hirschman Index (HHI), and the Zipf Index (q). The first stage (columns 1, 4, and 7) shows that the coefficients of the instrumental variables are all significant at the 1% level. The minimum value of the Anderson-LM statistic is 51.892 (p = 0.000), and the null hypothesis is rejected. The Cragg–Donald Wald F statistic is at least 54.394, far exceeding the 10% critical value, indicating the lack of weak IV issues. The second stage results demonstrate that after controlling for endogeneity, regardless of whether the explained variable is the number of patent applications or granted patents, the coefficients of all three explanatory variables remain positive and statistically significant at the 1% level. In conclusion, the baseline findings prove robust after accounting for endogeneity.

4.5. Transmission Mechanism Analysis

Table 5. Mechanism of mediation analysis.

Variable	(1) lnI	(2) ee	(3) lnI	(4) lnI
pr	1.086 ***	0.285 **		0.943 ***
	(3.00)	(2.48)		(2.62)
ee			0.544 ***	0.502 ***
			(3.81)	(3.51)
Control Variable	Yes	Yes	Yes	Yes
Time and individual effect	Yes	Yes	Yes	Yes
Constant	−2.820 ***	0.253	−2.873 ***	−3.195 ***
	(−3.89)	(1.06)	(−3.89)	(−4.29)
R2	0.945	0.922	0.946	0.946
Sobel_Z		2.027 **
Bootstrap		[0.0014]	[0.6640]

Note: Parenthetical values correspond to t-statistics; ** and *** signify significance at the 5% and 1% levels, respectively.

presents the estimation findings of the economic efficiency (ee) improvement mechanism of SCCL affecting IC. Column (1) shows the total effect of SCCL on IC, consistent with Column (8) of Table 3. Column (2) indicates that SCCL significantly enhances economic efficiency (ee) at the 5% level of significance, in line with theoretical expectations. As shown in column (3), the coefficient of ee is positive and statistically significant at the 1% level, suggesting that it promotes IC. Column (4) further incorporates both ee and SCCL into the model determining IC. The coefficient of ee remains significantly positive at the 1% level, while that of SCCL also stays significant at the 1% level, though its effect intensity declines by 13.17% compared to column (1), which suggests that ee is an important transmission path for the impact of SCCL on IC. According to these results, the Sobel_Z statistic is significant at the 5% level, and as revealed by the Bootstrap test with 1000 iterations, the 95% confidence interval for the mediation effect does not include 0. To sum up, the path “SCCL → economic efficiency → IC” holds true.

4.6. Testing the Moderating Effects of R&D Investment and Human Capital

Table 6. Substitution effect test.

Variable	(1) lnI	(2) lnI	(3) lnI	(4) lnI
pr	1.086 *** (3.00)	1.184 *** (3.33)	1.074 *** (3.04)	0.896 ** (2.53)
rd	0.156 *** (3.41)	0.151 *** (3.37)	0.058 (1.19)	0.067 (1.40)
pr × rd		−0.624 *** (−4.71)
edu	0.309 *** (4.21)	0.398 *** (5.36)	0.414 *** (5.53)	0.417 *** (5.61)
pr × edu			−0.624 *** (−4.92)
lnmarket	0.201 (1.21)	0.401 ** (2.39)	0.298 * (1.82)	0.428 ** (2.56)
pr × lnmarket				−2.215 *** (−5.28)
Control Variable	Yes	Yes	Yes	Yes
Time and individual effects	Yes	Yes	Yes	Yes
Constant	−3.067*** (−4.07)	−3.978 *** (−5.22)	−3.312 *** (−4.49)	−3.438 *** (−4.67)
R2	0.945	0.948	0.948	0.948

Note: Parenthetical values correspond to t-statistics; *, **, and *** signify significance at the 5% and 1% levels, respectively. Due to space constraints, the estimation results for the control variables are omitted.

reports the estimation findings of the moderation effect tests for R&D intensity (rd), human capital level (edu) and marketization (lnmarket). Among them, column (1) is the baseline regression, with results consistent with column (8) in Table 2. Columns (2) to (4) display the estimates of the moderating effects. The results show that all three exhibit a moderating pattern where the main effects are positive while the interaction terms are negative. This indicates that the three factors have a negative moderating effect on the IC enhancement of the SCCL.

For R&D intensity and human capital, a core rationale lies in the substitution effect between SCCL and these two factors. Specifically, when provincial-level R&D investment and human capital are insufficient, SCCL enhances provincial IC by aggregating dispersed resources (e.g., land, capital, labor) to establish innovation growth poles. Conversely, as R&D intensity and human capital levels increase, SCCL’s indirect innovation-promoting effect diminishes. This implies that robust R&D and human capital can functionally substitute for the innovation-driving role of SCCL. As can be seen from column (2), after incorporating the interaction term (pr × rd), the coefficient of pr is significantly higher than that in column (1), while the coefficient of rd is slightly lower than that in column (1). Similarly, upon the inclusion of the interaction term (pr × edu), relative to column (1), there is a reduction in the coefficient of pr, while the coefficient of edu exhibits a significant increase. In addition, another possible explanation is that when the intensity of SCCL is high, it will lead to the excessive agglomeration of R&D and human capital in central cities, thereby weakening the innovative effect of SCCL itself.

For marketization, one possible explanation is that under a highly marketized environment, the probability of short-term and flexible utilization of land resources by urban governments and enterprises increases. Such short-term behavior may deviate from the long-term needs of innovative development, conflicting with the “long-term stability and efficient utilization” model emphasized by intensive allocation. This conflict may lead to a misalignment between spatial agglomeration and the pace of marketization, thereby generating a negative moderating effect. Consequently, although marketization and SCCL individually contribute positively to IC, their interaction may inhibit innovation due to the pursuit of short-term interests. In addition, there exists a substitution effect between marketization and SCCL in the process of influencing IC. Fundamentally, it embodies the mutual substitution between governmental and market forces in shaping IC at the provincial level. As can be seen from column (4), when the interaction term (pr × lnmarket) is incorporated into the regression model, compared with column (1), the coefficient of pr decreases significantly, whereas that of rd increases. For instance, in the east region, although the level of SCCL—an indicator reflecting government intervention—is not high and the agglomeration-driven innovation effect it generates is weak, a high degree of marketization can facilitate the agglomeration of capital and talent, thereby fostering innovation.

4.7. Threshold Effect Test

To explore the threshold effect of environmental regulation on how SCCL affects IC, this paper successively conducted significance tests for single, double, and triple thresholds. First, using the grid search method, the sample interval was divided into 400 grid points for searching to determine the threshold value, and the truncation method was used to eliminate 1% of the extreme values to enhance the threshold value’s reliability; second, the Bootstrap resampling method was used 300 times to obtain the F-statistic and the corresponding p-value.

Table 7. Threshold number test.

Number of Threshold	Fstat	Prob
Single	66.69	0.000
Double	49.10	0.007
Triple	32.16	0.617

reports the number of thresholds and their corresponding probability values.

The p-values of both the single-threshold and double-threshold panel models are significant with a significance level of 0.01; in contrast, the triple-threshold effect does not achieve statistical significance. Drawing on these results, this study confirms that the model contains two threshold values and therefore employs a double-threshold panel model for subsequent analyses.

To enhance the credibility of the threshold estimation results, the likelihood ratio (LR) test method was used to test the statistical significance of the threshold effect. The estimated values and confidence intervals for the single and double thresholds are presented in

Table 8. Threshold estimates and confidence intervals.

Model	Threshold	95% Confidence Interval
Th-1	2.6741	(2.6471, 2.6807)
Th-2	1.3137	(1.2807, 1.3218)

, and the LR function for the threshold estimates is presented in Figure 6. Based on the principle of the threshold model, the threshold estimates correspond to the values that the likelihood ratio statistic approaches 0. The lowest points of the statistic correspond to the true threshold values, and the two lowest points are almost consistent with the threshold estimates in value, located between (1.28, 1.32) and (2.65, 2.68), respectively. Meanwhile, the LR value corresponding to the threshold value γ is significantly below the 5% critical line; the threshold values determined in this paper have passed the rigorous statistical test and are reliable and valid.

The findings of the double-threshold effect test of environmental regulation (lner) on the impact of SCCL on IC are shown in

Table 9. Environmental regulation threshold effect test.

Variable	lnI
pr·I (lner ≤ 1.3137)	0.987 (1.04)
pr·I (1.3137 < lner ≤ 2.6741)	2.911 *** (3.12)
pr·I (lner > 2.6741)	4.463 *** (4.59)
Control Variable	Yes
Constant	−7.448 *** (−9.70)
Time effects	Yes
Individual effects	Yes
R2	0.907

Note: Parenthetical values correspond to t-statistics; *** signifies significance at the 1% levels. Due to space constraints, the estimation results for the control variables are omitted.

. It is clear that as the degree of lner increases, the positive effect of SCCL on IC gradually increases. When the lner is less than or equal to 1.3137, the variable pr has a positive but insignificant effect on IC. When the lner is in the range of [1.3137–2.6741], at the 1% significance level, pr has a notable positive influence on IC, with an estimated coefficient of 2.911. When the lner is greater than 2.6741, pr continues to exert a significant influence on IC at the 1% level, with an estimated coefficient of 4.463. In summary, environmental regulation exhibits a significant threshold effect in the process of SCCL influencing IC; with the gradual escalation of environmental regulation, the promotional effect of SCCL on IC continues to strengthen. Based on this, Hypothesis 3 proposed in this study is verified and thus holds.

5. Extension Analysis

5.1. Non-Linear Relationship Test

The results of quantile regression estimating the effect of SCCL on IC are presented in

Table 10. Quantile Regression.

Variable	(1) 0.1	(2) 0.25	(3) 0.5	(4) 0.75	(5) 0.9
pr	1.115 ***	0.383	0.505	1.072 **	0.732 **
	(4.28)	(0.89)	(0.93)	(2.27)	(2.35)
Control Variable	Yes	Yes	Yes	Yes	Yes
Time effects	Yes	Yes	Yes	Yes	Yes
Individual effects	Yes	Yes	Yes	Yes	Yes
Constant	−0.523	−1.080	−2.027 *	−2.782 ***	−3.250 ***
	(−1.01)	(−1.26)	(−1.87)	(−2.95)	(−5.24)
N	529	529	529	529	529
R2	0.837	0.825	0.823	0.826	0.835

Note: Parenthetical values correspond to t-statistics; *, **, and *** signify significance at the 10%, 5%, and 1% levels, respectively. Due to space constraints, the estimation results for the control variables are omitted.

. It can be seen that in the 10% quantile group of provinces with the lowest IC, the variable pr is positively correlated with IC at the 1% significance level. The reason is that in provinces with lower innovation capabilities, the insufficient intensity of R&D investment and the inadequate accumulation of human capital directly inhibit the advancement of IC. From the government’s perspective, strengthening the degree of SCCL will enhance the overall scale economy effect of the province, save resources, and direct more resources towards innovation. At the same time, SCCL helps guide the spatial agglomeration of innovation factors. Through knowledge spillover effects, it can reduce the fixed costs of innovation, disperse innovation risks, and thereby initiate weak innovation activities. Meanwhile, in the 75% to 90% quantile group of provinces with strong IC, the variable pr remains significantly positive at the 5% significance level, which indicates that in regions with strong IC, SCCL will further attract highly mobile innovation elements (such as R&D funds and researchers) to continue flowing in, forming a virtuous cycle of “agglomeration → innovation enhancement → re-agglomeration of elements”.

Meanwhile, at the 0.9 quantile of IC, the intensity of the effect of SCCL is significantly lower than that at the 0.75 quantile, and the reason is that the provinces at the top of the innovation-capability pyramid generally maintain a high marketization level, R&D investment, and human capital, which, to some extent, substitute for the role of SCCL in enhancing IC. For the provincial group with medium-level IC (ranging from the 0.25th to the 0.5th quantile), the impact of SCCL is not significant. A possible reason for this is the “siphon effect” of talents in developed regions [39,40], which leads to insufficient innovation momentum in these medium-level regions. Relying solely on the SCCL therefore struggles to produce a noticeable effect. In summary, the impact of SCCL on IC is not uniform across the entire conditional distribution. Instead, it exhibits a non-linear characteristic of “strengthening at both ends and collapsing in the middle”.

5.2. Heterogeneity Analysis

As shown in

Table 11. Heterogeneity test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Variable	East	Central	West	South	North	Provincial Capital	Non-Provincial Capital	High	Medium	Low	High	Medium-Low
pr	−2.873 ***	−0.287	3.576 ***	1.430 ***	0.806 *	1.261 ***	−7.398 ***	1.675 ***	2.858 ***	−2.019	0.018	1.143 ***
	(−3.09)	(−0.40)	(7.08)	(2.71)	(1.69)	(3.28)	(−3.35)	(3.12)	(3.60)	(−1.45)	(0.02)	(2.97)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−1.875	−3.948 ***	−3.377 ***	−2.141 *	−5.221 ***	−2.444 ***	2.153	−4.390 ***	−3.379 *	−2.621	−3.797	−2.804 ***
	(−1.17)	(−2.68)	(−3.14)	(−1.68)	(−5.01)	(−2.87)	(0.95)	(−3.66)	(−1.76)	(−1.59)	(−1.19)	(−3.38)
N	161	184	184	276	253	472	57	184	184	161	92	437
R2	0.966	0.964	0.967	0.950	0.959	0.939	0.990	0.960	0.945	0.962	0.980	0.947

Note: Parenthetical values correspond to t-statistics; * and *** signify significance at the 10% and 1% levels, respectively. Due to space constraints, the estimation results for the control variables and the non-linear regression estimation results for the eastern region are omitted.

, the heterogeneity of SCCL’s impact on IC is presented. First, the samples are divided into the east (Hebei, Liaoning, Shandong, Zhejiang, Jiangsu, Fujian and Guangdong), the central (Shanxi, Henan, Hubei, Anhui, Hunan, Jiangxi, Heilongjiang and Jilin) and the west (Shaanxi, Sichuan, Gansu, Ningxia, Yunnan, Guizhou, Guangxi and Inner Mongolia).

As can be seen from columns (1) to (3), the impact of SCCL on IC exhibits obvious spatial heterogeneity, and the direction of its effect shows a changing trend of “significantly negative correlation–negative correlation–significantly positive correlation” from the east to the west. Specifically, in the east, there is a significantly negative correlation between the two variables at the 1% significance level, which is contrary to the conclusion of the full sample. The underlying reason lies in the relatively low level of SCCL, which has failed to foster a striking agglomeration-driven innovation effect through such concentrated allocation. Specifically, the crowding effect currently prevailing in central cities outweighs the agglomeration effect—an imbalance that hinders innovation development. This argument is further supported by data observations: provincial-level regions in eastern China exhibit relatively strong overall IC, yet their SCCL remains comparatively low, indicating a lack of effective coupling and coordination between these two factors.

On this basis, this paper probes the non-linear effect of the SCCL on IC and finds that SCCL and IC exhibit a significant U-shaped correlation, which is verified by the Utest. The value of the inflection point equals 0.296, and only some years of provincial-level regions such as Fujian, Jiangsu, and Guangdong have exceeded this inflection point. The sub-sample regression reveals that for the left part of the sample (i.e., when the concentration degree is below the inflection point), the impact coefficient of −4.181 is statistically significant at the 1% level, which implies that only when the SCCL reaches a certain level can the agglomeration innovation effect be generated.

In the central region, the relationship between the two is also negative but statistically insignificant. Compared with the east region, the central region has a higher degree of SCCL, but it also fails to generate the agglomeration innovation effect through such concentrated allocation. Similarly, further examination of the non-linear relationship between the two reveals that it is not statistically significant.

In the west region, SCCL exerts a statistically significant positive impact on IC at the 1% significance level, which is consistent with the conclusion of the full sample but obviously opposite to that of the eastern region. The reason lies in the high degree of SCCL in the west region, which can effectively generate the spatial agglomeration innovation effect. The rationale is that, given the historically dispersed innovation resources and weak infrastructure in western regions, SCCL facilitates the creation of innovation growth poles. This disrupts the original low-level equilibrium of innovation, avoiding the inefficient “scattered allocation” of innovation resources. Consequently, it fosters the high-density, coordinated clustering of essential factors, including talent, capital and technology, in provincial central cities. Once such coordinated clustering emerges, it readily cultivates a robust “innovation ecosystem,” enhances innovation efficiency, and generates concentrated agglomeration economies. Leveraging spatial spillover effects, this process elevates the overall IC across the province. In addition to this, the SCCL is often linked to policies such as industrial parks and relocation demonstration zones. On one hand, this approach facilitates the relocation and attraction of advanced manufacturing and eco-friendly industries from eastern regions, thereby elevating the overall level of technological innovation. On the other hand, it enables the implementation of unified environmental standards and shared innovation platforms, driving research, development, and application of green technologies.

Second, the sample was divided into two major regions: the south and the north. the dividing line between the south and the north is based on the Qinling–Huaihe Boundary Line. The outcomes presented in columns (4)–(5) reveal that the positive influence coefficient of SCCL is statistically significant at the 1% level for southern regions and at the 10% level for northern regions, respectively. Relatively speaking, the intensity of the impact of SCCL on IC is higher in the southern regions. The reason is that the degree of SCCL is higher in the southern provinces, and the market economy is more active. The spatial agglomeration of population, industry, and economic activities brought about by the SCCL is more evident, thereby generating a greater innovation effect.

Third, based on the location orientation of SCCL, generally speaking, provincial capitals are generally recognized as the central cities and advantageous regions in provincial administrative areas, serving as the primary agglomeration places for resource elements and the main origins of innovation activities. Of course, there are exceptions, and SCCL within a province may also be oriented toward non-provincial capital cities. Based on this, according to the spatial orientation of the SCCL, the samples are divided into two categories, which are provincial capital city-oriented and non-provincial capital city-oriented. Columns (6) and (7) report the estimation results, which show obvious heterogeneity. It can be seen that with a provincial capital city orientation of SCCL, the variable pr passes the 1% significance test with a positive sign; with a non-provincial capital city orientation of the SCCL, the coefficient of pr is statistically significant and negative at the 1% level. The reason for this phenomenon is that provincial capital cities are natural innovation hubs, and the concentration of construction land in non-provincial capital cities leads to spatial misallocation of resources, which impedes the full exertion of the innovative effect of agglomeration.

Fourth, the test is conducted by grouping according to the level of variables. On the one hand, grouping is based on the degree of SCCL. As shown in columns (8) to (10), medium–high SCCL promotes IC at the 1% level, while low SCCL restricts IC, which is, however, not significant. On the other hand, grouping is based on the level of IC. Columns (11) and (12) indicate that in provinces with high IC, the IC effect of SCCL is positive but not significant; in contrast, in provinces with medium–low IC, the SCCL is significantly positive at the 1% level. The reason behind this estimate is that provinces with low SCCL are mostly eastern provinces with relatively strong IC.

5.3. The Impact of the Spatial Concentration of Different Types of Construction Land on IC

The functions of construction land can generally be divided into production functions, residential functions, and commercial service functions that support enterprise operations and residents’ daily lives. Among these, the production function is the basic function, while the residential and commercial service functions are non-basic functions. When construction land quotas are allocated by provincial governments to municipal governments, in the context of competition among municipal governments, the municipal governments, in order to boost urban economic growth, intensify the functional allocation of construction land, which leads to the formation of spatial patterns of different types of land use within provincial jurisdictions. The spatial concentration of these different types of land use will have varying impacts on IC. In this study, Equation (8) has been applied to identify the SCCL-related level with distinct functions.

Table 12. Test of the heterogeneity of the effects of different functional land uses.

Variable	(1) lnI	(2) lnI	(3) lnI	(4) lnI	(5) lnI
pr_industry	0.021 (0.08)
pr_reside		−0.548 ** (−2.07)
pr_service			0.945 *** (3.32)
pr_public				0.524 *** (2.72)
pr_otherservice					0.676 *** (2.80)
Control variable	Yes	Yes	Yes	Yes	Yes
Time effects	Yes	Yes	Yes	Yes	Yes
Individual effects	Yes	Yes	Yes	Yes	Yes
Constant	−2.686 *** (−3.57)	−2.603 *** (−3.49)	−2.920 *** (−3.93)	−2.894 *** (−3.87)	−2.784 *** (−3.74)
R2	0.944	0.945	0.945	0.945	0.945

Note: Parenthetical values correspond to t-statistics; ** and *** signify significance at the 5% and 1% levels, respectively. Due to space constraints, the estimation results for the control variables are omitted.

reports the influence of the spatial concentration of different types of land use on IC. Columns (1)–(2) present that the spatial concentration of industrial land (pr_industry) is positive but not significant, while the spatial concentration of residential land (pr_reside) inhibits IC at the 5% statistical level. The reason is that urban economic development is an important performance indicator for municipal governments. In the process of functional land allocation, in order to increase the “bargaining chips” for land resources in the process of attracting investment, municipal governments implement a dual-track system for land prices [41]. They auction, bid, and hang residential land for sale, while industrial land tends to be sold at zero price or through negotiated transfer, and they reduce the supply of residential land to achieve “killing two birds with one stone.” On the one hand, they raise the price of residential land to compensate for the loss of industrial land transfer. On the other hand, they increase the “bargaining chips” for land resources for local governments to attract investment. This leads to a low degree of spatial agglomeration of residential and industrial land, which is not conducive to the agglomeration of population and enterprises and makes it difficult to form agglomeration innovation effects. Additionally, the low price of industrial land reduces the internal motivation of enterprises to innovate and develop, while the high price of residential land leads to increased living costs, which will reduce human capital investment and thus be not conducive to innovation.

Column (3) presents that the spatial allocation concentration of commercial service land (pr_service) is significantly positive at the 1% level, which suggests that the concentrated allocation of commercial and service land helps to enhance IC. As the spatial carrier of the service industry, commercial service land serves as the main material carrier for producer services, such as R&D and design industries and R&D service departments. It inherently exhibits an obvious threshold effect and a tendency to cluster in central cities. Therefore, the concentrated spatial allocation of commercial service land can directly reduce the land use costs of producer services, promote the agglomeration of relevant entities—including producer service enterprises, clients, universities, scientific research institutions, local governments, and innovation service institutions—and thereby boost innovation. Meanwhile, within commercial service land, public facility land undertakes the hard environment for innovation. As a direct carrier of R&D and innovation, innovation infrastructure (such as institutions of higher education, scientific research institutions, R&D and design industries, and R&D service departments) exerts a direct impact on IC. The agglomeration of such land helps to induce the spatial concentration of relevant innovation entities, thereby generating knowledge spillover and collective learning effects.

Column (4) reports that the variable pr_public is also significantly positive at the 1% level, which demonstrates that the concentrated allocation of public service land will leverage the location advantages of leading cities in innovation development. Column (5) shows that the concentrated spatial allocation of other service land (pr_otherservice) is also significantly positive at the 1% level. The reason lies in the fact that other commercial service land undertakes and influences the urban quality of life—a key component of the soft environment for innovation. The agglomeration of such land in central cities with strong IC will help promote the transformation of consumer services from basic public services to inclusive and high-quality services, strengthen the attraction of high-end talents, and thus exert an indirect promotional effect on IC.

6. Discussion

In continuation of the foregoing, as for the relationship between construction land resource allocation (or construction land use) and IC, the academic community has mainly examined the negative effect of urban construction land misallocation on urban IC from the perspective of functional mismatches (between commercial and industrial uses) within cities [20,21,22]. From the standpoint of China’s construction land allocation system, which follows the principle of “central government first, local government second; region first, use second,” existing literature has predominantly focused on the “use second” aspect. This means that after the central government allocates construction land resources to provincial governments and provincial governments further allocate these resources to various cities, municipal governments allocate construction land resources according to their own development needs. In other words, existing studies have failed to give adequate attention the “region first” aspect. Therefore, from the perspective of government hierarchy, existing research has mainly focused on municipal governments, while there is a lack of discussion on the dimension of provincial governments allocating construction land resources to cities. This paper attempts to fill this research gap.

Of course, the scarcity of attention in existing literature to the impact of provincial-level SCCL, which is formed by the spatial allocation of construction land resources, on IC can be attributed to at least two factors. First, the concept of SCCL is an interdisciplinary concept that is not easy to grasp. On the one hand, concepts such as spatial allocation and agglomeration are specialized terms in the field of regional economics. On the other hand, the research object “construction land resources” falls within the scope of land resource management. In this discipline, while there is also a focus on the optimal allocation of construction land resources, more attention is paid to the allocation system, and insufficient attention has been given to the issue of unequal allocation, spatial inequality and spatial synergy of land resources. To propose the concept of “spatial agglomeration of construction land resources”, one must possess knowledge from multiple disciplines. Second, from the perspective of empirical research, studies at the provincial scale have certain limitations in terms of sample size compared to those at the urban scale. Most current studies, in pursuit of fine-grained research, have overlooked the importance of the provincial spatial unit. In fact, given the scarcity of construction land resources in China, researching the spatial optimization of construction land resources within the province and its economic, social, and ecological effects is of great significance for the scientific decision-making of provincial governments.

In fact, how to optimize the scientific allocation of provincial construction land resources among different cities is a question involving the provincial spatial development strategy. In essence, this is also a debate on whether the provincial development path should be balanced or unbalanced. So, what impact will different provincial development paths have on the provincial IC? Meanwhile, in both academia and the political sphere, the concept of a “strong provincial capital” has been proposed, which essentially represents an unbalanced development strategy. Therefore, some scholars have studied the impact of a strong provincial capital on the provincial IC and found that a strong provincial capital strategy based on population or GDP (the proportion of the provincial capital’s population or GDP in the province) will promote the improvement of IC [42,43,44]. Fundamentally, the above conclusion coincides with the findings of this paper, that is, spatial agglomeration in the province will promote the improvement of IC. However, unlike these existing studies, this paper abandons the provincial capital orientation of SCCL. Only in the extended analysis and split-sample regression does this paper divide SCCL into provincial capital-oriented and non-provincial capital-oriented types. Overall, the provincial capital-oriented type accounts for the vast majority of cases in SCCL. From the view of research, specifically, this paper starts from two aspects: on the one hand, it explores the underlying logic behind the spatial agglomeration of population or the economy; on the other hand, it is based on the perspective of government behavior. The reason is that spatial agglomeration of population or economy in the province needs to be supported by SCCL that are adapted to it, and as an important resource owned by the government, the degree of population and economic spatial agglomeration demonstrates a strong relationship with the spatial allocation of construction land resources based on government behavior. Therefore, this study, from the perspective of government behavior, provides more targeted decision-making references for population and economic spatial agglomeration under the goal of improving provincial IC.

Additionally, compared with the existing literature, this study has also made some new progress. For example, through the threshold effect test, it is revealed that the rise in the environmental regulation threshold strengthens the impact of SCCL on IC. Essentially, this is consistent with the Porter Hypothesis [45]. Meanwhile, this paper also found through the moderation effect test that the centralized spatial allocation of construction land would negatively moderate the promoting effects of R&D intensity, human capital, and marketization on IC. That is to say, the promoting effects of R&D investment, human capital, and marketization on innovation would gradually weaken with the increase in the degree of SCCL. This tells us that even if R&D investment is insufficient and the levels of human capital and marketization are low, the IC can still be improved by enhancing the level of regional spatial agglomeration. These findings are also in line with our intuition.

7. Conclusions

7.1. Research Conclusions

This study employs multiple methods, such as the two-way fixed effects model, instrumental variable (IV) estimation method, the mediation effect model, moderating effect model, threshold effect model, quantile regression and subsample regression, to empirically investigate how SCCL (spatial concentration of construction land) affects IC (innovation capacity), utilizing panel data covering 23 provinces in China over the period 2000–2022. The study’s conclusions are as follows:

First, SCCL has a significant positive impact on IC. This conclusion still holds after replacing the explanatory variable and the explained variable, as well as conducting IV estimation.

Second, at the macro level, economic efficiency a key transmission mechanism for SCCL to promote IC.

Third, R&D investment, human capital, and marketization negatively moderate the relationship between SCCL and IC.

Fourth, the positive impact of SCCL on IC is enhanced as the level of environmental regulation improves.

Fifth, the influence of SCCL on IC has a non-linear characteristic of “strong at both ends and weak in the middle”. In other words, the lower the IC of a region, or the higher the IC of a region, the more pronounced the positive impact of SCCL on IC.

Sixth, the influence of SCCL on IC has spatial heterogeneity. From east to west, there is a significant shift from diseconomies of agglomeration to economies of agglomeration. That is, only in the west region does SCCL exhibit a significant promotional effect on IC. SCCL with a provincial capital orientation has a significant effect on innovation enhancement. In provinces with a high level of SCCL, the promoting effect of SCCL on IC is significant. In provinces with a low level of IC, the promoting effect of SCCL is significant.

Seventh, in terms of different types of construction land use, only the spatial concentration of commercial service land exerts a significant positive impact on IC. Further subdividing commercial service land into public facility land and other service land, the concentration of these subcategories still has a significant positive effect on IC. The spatial concentration of industrial land does not have a noticeable positive effect on IC, while the residential land has a certain negative impact on IC.

7.2. Policy Recommendations

According to the empirical results, this paper proposes the following policy implications from the perspective of provincial government behavior:

First, lay emphasis on the role of the “spatial agglomeration” factor in enhancing IC. While provincial governments focus on R&D investment and human capital to drive IC, they also need to unlock the role of “space” as a factor in boosting IC. In this process, provincial governments should strive to strengthen spatial governance, actively optimize the provincial spatial structure, and attach importance to spatial organizational innovation that supports innovation. They should also actively guide the concentration of both direct innovation factors and indirect innovation-inducing population and economic activities toward advantageous provincial central cities. Leveraging geographical proximity, this concentration will promote innovation through mechanisms such as R&D enclave effects, knowledge spillover, local market effects, and spatial competition.

Second, take construction land resources as a key lever to realize “promoting innovation through land management”. Land is not merely a “container”. By optimizing the spatial allocation of land resources, it can become a significant engine for enhancing the IC of a province. Provincial governments should focus on the centralized allocation of scarce construction land to provincial capitals with innovation advantages, thereby enhancing the innovation effects of the agglomeration scale of these capitals. Currently, there is a significant imbalance in regional development in China, especially in the inland western regions, where innovation resources are scarce, R&D investment intensity is low, and there is a shortage of human capital. It is necessary to form agglomeration innovation effects through the centralized allocation of construction land to offset the adverse impacts of low R&D investment and human capital shortages on innovation.

Third, adhere to the principle of spatial agglomeration, strengthen the cross-city cooperative construction of industrial parks, promote the centralized allocation of industrial land, and facilitate the innovative development of industry in the process of spatial agglomeration. Provincial governments should actively establish cross-city coordination mechanisms, and improve the pattern of fragmented allocation of industrial land behind industrial competition by promoting inter-city cooperation in building “enclave parks”. On this basis, efforts should be made to realize the cross-municipal clustered development of homogeneous industries, ultimately forming a new provincial industrial pattern characterized by orderly division of labor, mutual complementarity and win-win cooperation, and thereby injecting sustained impetus into the high-quality development of the province.

Fourth, establish a coordinated linkage mechanism between population changes and land supply. As can be seen from Table 12, the spatial concentration of residential land inhibits IC at the 5% significance level, and a key reason is the spatial mismatch between population mobility and the corresponding allocation of residential land. Consequently, during the spatial allocation of land resources, provincial governments should focus on linking population size with the total land supply, optimize the spatial layout of land based on population distribution, and avoid the imbalance of “population inflow with insufficient land” or “idle land with population outflow”. A regular evaluation and adjustment mechanism should be established to dynamically fine-tune the land supply plan according to population changes, ultimately achieving the precise matching of “people following industries and land following people”. Meanwhile, in cities with population agglomeration, priority should be given to residential, commercial, and land for education and medical services to form an integrated “industry–urban” spatial pattern that promotes innovative development.

Fifth, establish a targeted environmental regulation constraint mechanism to enhance the positive effect of SCCL on IC of Chinese provinces. It is necessary to make good use of the SCCL as a spatial management tool, while strengthening environmental regulation and promoting the coordinated interaction between environmental regulation and spatial management to force enterprises to innovate. Through the dual drive of “spatial guidance + environmental constraints,” it is possible to avoid the fragmentation of innovation factors caused by the dispersed layout of construction land, and to transform environmental pressure into the internal driving force for enterprise innovation. Ultimately, this will form a virtuous cycle of “intensive land allocation–enterprise innovation breakthrough–regional innovation upgrading,” continuously enhancing the effect of SCCL on IC.

7.3. Research Outlook

Agglomeration and innovation are a fundamental proposition that has long been of interest to the academic community. Innovation cannot be separated from the market and the government, but few studies have focused on the role of provincial governments. This paper, from the perspective of spatial political economy, takes the spatial allocation of construction land resources by Chinese provincial governments as the starting point to examine its mechanism of action on IC. This essentially involves the issue of the counter-effect of spatial production relations on spatial productive forces. To elaborate further, cities, as the direct spatial subjects of provincial innovation, are directly influenced by the provincial governments in terms of the equilibrium of resource allocation among cities, which in turn determines the overall innovation efficiency of the province. Fundamentally, in the process of shaping provincial IC, the important function of provincial governments is to build an environment for innovation agglomeration and reshape the provincial innovation landscape through spatial governance. On the one hand, land, as an important resource owned by the government, plays a significant role in macro-control and serves as the material basis for spatial governance. On the other hand, talent is the core resource for innovative development, and the agglomeration of talent in advantageous regions is a fundamental law.

The spatial coordination of population and land is an essential requirement for innovation. By concentrating land resources in advantageous regions, provincial governments can provide these regions with spatial carriers for innovation, guarantees for the agglomeration of innovative elements, and support for innovation efficiency. Therefore, optimizing the spatial allocation structure of provincial construction land is an important means of promoting local innovative development. By “allocating quotas” to advantageous regions, it is possible to achieve “recreation of innovation space and optimization of the innovation spatial pattern” within the province. This paper attempts to study this topic from multiple dimensions, but there are still the following issues that need further discussion.

First, the empirical analysis of the mediating mechanisms remains to be explored. Drawing on the “structure–conduct–performance” analytical paradigm, this study merely examines the direct impact of SCCL on IC, and only conducts a qualitative analysis of the micro-level behavioral mechanisms in four aspects, without empirical testing. Of course, one reason for this is the difficulty in obtaining data that reflect the innovative behavior of micro entities. Future research can further deepen the empirical study of the mechanisms.

Second, the moderating effects need to be further examined. In the process of analyzing moderating effects, this paper only considers the interaction between R&D investment intensity and human capital level with the SCCL from the perspective of the R&D production function. In fact, the interaction between population spatial agglomeration and construction land spatial agglomeration, which reveals the relationship between population and land, as well as the interaction between industrial agglomeration and construction land spatial concentration, and their impacts on IC, all deserve further discussion.

Third, future research should further strengthen the exploration of innovative development in provincial advantageous regions. According to the estimation results in Table 11, it can be observed that SCCL biased toward provincial capitals promotes IC, while SCCL biased toward non-capital cities inhibits IC. This indicates that provincial capitals serve as the primary advantage regions within provinces. As the dominant areas within their respective provinces, the superior status of provincial capitals is rooted in the interaction between administrative hierarchy and the laws of the market economy, giving rise to the unique phenomenon of “provincial capital advantage.” At the same time, provincial capitals serve as vital spatial carriers for development factors, how to give full play to the “provincial capital advantage” to drive the enhancement of overall provincial IC is a topic that warrants in-depth discussion.