Evolutionary mechanism for subregional cooperation: Evidence from provincial subregion in Central China

Abstract

Against the backdrop of regional integration, subregions with border effects have become the bottleneck in securing the sustainable development of cross-border cooperation. Based on the discussion of the subregions and subregional cooperation in China, this article put forward three theoretical hypotheses with eight typical provincial subregions in Central China as examples. Combined with relevant models, variables and data, the study tested the evolutionary mechanism of subregional cooperation. The main conclusions include: (a) internal factors contribute to better provincial subregional cooperation; (b) geographical and institutional arrangements significantly affect subregional cooperation, with the shielding effect being the dominant border effect and (c) spatial heterogeneity of high-tech industries has an inverted U-shaped effect on subregional cooperation.

Keywords

Provincial subregion subregional cooperation border effect institutional barrier industry heterogeneity

Against the backdrop of globalization and regional collectivization, subregional cooperation has become an established practice around the world. Under the auspices of international or regional organizations (e.g., UN, ASEAN, EU, SADC [The Southern African Development Community], etc.), subregional cooperation or cross-border cooperation (CBC), initiatives have made progress to various extents, including the Greater Mekong Subregion (GMS), the South Growth Triangle of ASEAN, the South Asia Growth Quadrangle, the Tumen River Subregion, Oberrheinkonferenz (upper Rhine), Euregio Maas-Rhein, Agglomération franco-valdo-genevois (grand Geneva), Zambezi Basin Delta and Malawi Lake Basin Delta (Valentin et al., 2016). Instead of ‘going global’, subregional cooperation features territories or regions collaborating in a relatively wide range of areas while opening their internal markets to a certain extent (Hanson, 2001; Liang et al., 2012). Since the concept of a ‘subregion’ was first used in academic seminars of Asian Development Bank, its connotations, explained by different scholars from different perspectives, show great variations (Liang, 2013; Oehlers, 2006). Earlier studies on subregion mainly focussed on economic cooperation (Ocampo and Titelman, 2009), and later the research hotspots extended to the non-economic effect of economic cooperation, non-economic impacts and non-economic fields (Hiroyuki and Keiichiro, 2014; Hu et al., 2010; Nicholas, 2014). Although regional integration drives regional economic development, a series of bottleneck problems still exist, such as the imbalanced development despite rapid overall growth as well as excessive concentration of social and economic resources in regional centres. The two or more geographically adjacent border regions often become ‘the forgotten corner’ or an area of ‘economic depression’ (Han and Shi, 2014). Therefore, extensive research on the cooperation mechanism of subregions may help improve the efficiency of joint development in border regions. Meanwhile, the study on cross-border space bears theoretical and practical significance for the boundary regions to speed up urban agglomeration and explore a new way of urbanization (Kukovic and Hacek, 2018).

I Literature review

Conceptually, a ‘subregion’, delimited with a ‘region’ as a reference, lacks a clear definition (Liang et al., 2012; Mukherjee, 2014). Though often used in the context of economic cooperation, it is particularly necessary to explain ‘subregional cooperation’ from non-economic perspectives. Therefore, it is argued that, against the backdrop of regional integration, a ‘subregion’ refers to two or more adjacent territories joint together to form a specific geographical space with a certain homogeneity, due to their geographic locations, administrative divisions, resource conditions, cultural customs, industrial structures and development stages (Bat et al., 2013). Moreover, a subregion usually has three characteristics: (a) obvious boundary effects, (b) a conjugate relationship with coexistence of homogeneity and heterogeneity and (c) strong demands for integration. The ‘subregional cooperation’, in the process of subregional integration, refers to the comprehensive cooperation in fields, such as economy, politics, culture, environment and security, in order to achieve stability and development with the goal of 1 + 1 ≥ 2. Such cooperation manifests a type of regionalism with extensive fields. A subregional cooperation zone as well as any other cooperation organization would need mechanisms to ensure its effective operation.

The current research on subregions in China mainly focusses on two aspects. The first is transnational subregions, like the GMS and the Tumen River subregion, which attract more attention. The second is provincial subregions, like Huaihai economic zone, which are less studied. The concept of provincial subregions has been first put forward as ‘provincial border regions’ in Research on Economic Development of China’s Provincial Border Regions in 1993, which established a theoretical framework for the economic analysis of border regions from perspectives such as location, spatial organization and the size of administrative regions. As problems in border regions become more prominent and the regional economic theories become more developed, the notions of provincial border regions have expanded to include three aspects. First, the administrative regional economy theories emphasize the excessive administrative intervention in regions’ economic development, which results in the marginalization of the border economy and the segmentation of the regional market. Due to the economic game among the provincial border regions, the spatial game theory was introduced to describe how local governments deal with segmentation and integration of the market in the provincial border regions in the context of financial incentives and political competition (Zhou and Tao, 2011). However, these theoretical explanations magnified the negative role of administrative regions and exaggerated the positive role of economic zones and neglected the interactions between administrative regions and economic zones. Second, the practice analysis of the provincial border regions highlighted the development model, spatial structure optimization, urbanization, industrial development and institutionalization. Although such research studies have noticed the particularity of provincial border regions in cross-regional cooperation, most have not gone beyond the theme of regional cooperation, and consequently, the research hotspots have been mostly limited to regional economic cooperation and coordinated development from the macro level. What is more, the particularity of the provincial border regions has been involved in relevant research studies. Considering the characteristics of marginal areas, the border effect in ecology is used to redefine and classify the provincial border regions. In addition, the border effect, which is a key variable in the development of the provincial border regions, has gradually attracted scholars’ attention. Therefore, with the new economic geography as the theoretical framework, the mechanism of how the border effect impacts the economic development of provincial border regions has been analysed (Zeng, 2015). Besides, empirical analysis on the border effect has been conducted from the perspectives of interprovincial trade and convergence of economic growth (Wang et al., 2008; Xing and Li, 2010).

From the geographical and administrative perspectives, the problems of regional division and cooperation inhibition in border territories, arising from the shielding effect and institutional barriers, seem prevalent, especially in developing countries or territories with disparate development. The ‘shielding effect’ keeps the flow of passengers, commodity, information and capital from crossing borders, which impairs CBC. The institutional barriers with an additional regulatory function indirectly affect the geographical distribution of economic growth, reduce the efficiency of resource allocation, intensify the local competition and inhibition of regional cooperation, and leave administrative imprints in the generally market-based process of cooperation, which adds complexity in the relationship between the spatial structure and the subregional cooperation (Castanho et al., 2018; Kurowska-Pysz, 2016). At the same time, the geographical continuity means lower transportation costs (Castanho et al., 2017). However, due to the ‘mesomeric effect’, borders areas have become important ports for both sides as an economic, social and cultural link, while demands for transit have increased, which further activate the flow of elements across the border (e.g., Lin, 2012), thus enhancing the cooperation efficiency.

From the perspective of industrial heterogeneity, the classic ‘centre-periphery’ spatial structure is closely related to regional division and cooperation (Jacobs, 2016; Paul, 1991), in which the centre often focusses on management and R&D functions, while the periphery focusses on manufacturing and processing functions (Jacobs, 2016). Industrial division further widens the gap between the centre and periphery, for regional differences lead to the release of gradient potential energy to bring spillover effects and complementary effects, which promote regional cooperation and influence the cooperation mode from the perspective of comparative relations. However, some studies have proved that the industrial division will not increase the regional gap (Kurowska-Pysz, 2016), and the homogeneous size effect will strengthen regional cooperation. In summary, in the initial stage of industrial heterogeneity, the spillover effect and complementary effect allow for more room of regional cooperation, but if industrial heterogeneity continues to rise after reaching a critical value, the loss of scale efficiency will reduce the room for regional cooperation.

The aforementioned three geographical, administrative and industrial perspectives are important factors affecting the coordinated development of the subregion. If any of them goes wrong, the development of the subregion might slow down. According to the new economic geography, different spatial relations have different impacts on the operational efficiency of subregional cooperation, demonstrated in two main forms: spatial spanning and spatial continuity based on geographical proximity. Subregional cooperation based on spatial spanning only considers benefits of the subregional complementarity, neglecting the influence of space length, which is closely related to the costs of transportation, information transmission and transaction. Specifically, under the attenuation effect of space length, the farther the distance, the higher the cost of transportation and information transmission generated by interactions among subregions, which will harm the coordination of subregional cooperation. In addition to the higher transaction cost, the intensity of spatial spillover effect based on complementarity among subregions will also be weaker in the spatial spanning form. However, in the spatial continuity form, the administrative barriers and the market segmentation from administrative division have become the biggest obstacle to provincial subregional cooperation. In Central China, administrative barriers have severely hindered the integration process of provincial subregions. Due to a lack of unified planning and coordination, highly homogeneous industries lead to vicious competition, the subregional industrial clustering effect remains weak and the external scale effect of agglomeration cannot be given full play. In addition, due to the ‘Matthew effect’, various resources would concentrate in the central city, creating more competition and less coordination in subregions, which may lead to a convergent industrial structure. For trans-regional businesses, the absence of regional cooperation policies will lead to the separation of forms of production in the border regions. Without the coordination of regional cooperation policies, local governments tend to slow the development of trans-regional businesses to guarantee local employment and tax revenues first. Literature on subregional cooperation in Central China points to three challenges: giving full play to the advantages of spatial continuity and removing administrative barriers to promote the free flow and optimal allocation of production factors, strengthening industrial connection to avoid industrial homogenization and utilizing regional policy tools to define urban functions and effectively promote urban specialization and collaboration.

To sum up, from the provincial and prefectural level in Central China, few references have verified the relationship between subregional cooperation through hypothesis testing, and empirical analyses on the relationship influenced by space structure, industrial heterogeneity and regional policies are insufficient. However, the literature indicates that a series of questions remain unanswered: How efficient is provincial subregional cooperation in China? How about the effect binding and standardability of the operational mechanism? Which influence dominates the spatial structure based on the geographical-administrative perspectives of on subregional cooperation, the ‘shielding effect’ or the ‘mesomeric effect’? How significant is the main influence? Is there an inverted U-curve relationship between the industrial heterogeneity and subregional cooperation? If so, does this relationship apply to the subregion under the boundary effect? To address the above-mentioned questions, this article focusses on the comprehensiveness, coordination, responsiveness and complementarity of subregional cooperation. Moreover, the article will further illustrate the evolution trend of restraining factors and the effectiveness of regional policies on enhancing the subregional cooperation efficiency.

II Research hypothesis

1 Responsiveness and complementarity

Due to low administrative costs and high-level openness in subregional cooperation, connections between all sides become increasingly close, specialization increasingly deepens and behaviours of each side are increasingly territorialized. In addition to bringing enormous demands for transit, integration promotes behavioural agents, including governments, businesses and non-governmental organizations (NGOs), to expand the market and look for production factors on a large scale. However, original strength of the development is driving force, which is the primary cause of subregional cooperation. How the driving force forms and functions lies in the internal mechanism for revealing law, solving problems and developing strategies and policies of subregional cooperation (Lin, 2012). Depending on its sources, driving forces can be divided into endogenous dynamic ones and exogenous dynamic ones, in which the endogenous dynamic ones are the decisive factors for subregional cooperation. Their types are interchangeable under certain conditions. The effective system of motivation and restriction that encompass benefit sharing and compensation mechanisms tend to give subregional cooperation enough ‘stimulus’, ‘award the good, punish the bad’ and achieve the transfer and distribution of benefits, thus protecting subregional cooperation (Castanho et al., 2017, 2018; Kurowska-Pysz et al., 2017). Accordingly, the article put forward the following hypotheses:

Hypothesis 1a: Subjective essential factors, endogenous and exogenous dynamic factors and security factors are important parts of the coordinated development in subregional cooperation.

Hypothesis 1b: The joint responsiveness and complementarity of the four cooperation mechanisms tend to have an impact on the healthy development of the subregions and the efficiency of subregional cooperation.

Hypothesis 1c: The responsiveness and complementarity of cooperation mechanisms in different regions tend to vary greatly.

2 Geographical proximity and governance

Subregional cooperation is both geographical and administrative, and its operational efficiency varies according to the similarities and differences of spatial relations. Geographically, subregional cooperation might see two kinds of spatial relations: geographical proximity and geographical crossing. Administratively, there are also two kinds of spatial relations: local cooperation and CBC (Wang et al., 2018). The former perspective only considers the geographical distance and transportation costs, while the latter involves the influence of the boundary effect which might produce two outcomes: (a) obstruction of transit for shutdown of border (i.e., shielding effect) and (b) contact and communication for open of border (i.e., mediating effect). The attenuation of geographical distance and transportation cost will reduce transaction cost of subregional cooperation (Makkonen, 2018), while the border effect affects the cooperation through the information symmetry, transaction costs and institutional barriers. In this regard, we propose the following hypotheses:

Hypothesis 2a: Geographical crossing and institutional barriers tend to baffle the development and evolution of subregional cooperation mechanism.

Hypothesis 2b: Local cooperation with geographical proximity tends to be more conducive to the establishment of a co-operation zone.

3 Heterogeneity of high-tech industries

Functional specialization and the industrial structure are the key factors that determine the relationship between subregional cooperation partners: competitive or cooperative. High-tech industries’ division of labour, level of specialization and spatial structure can effectively represent the pattern in which the centre focusses on the function of management and R&D, while the periphery focusses on the function of manufacture and machining. Actually, interactions exist between the scale effect caused by high-tech industry agglomeration and gradient potential energy and complementarity caused by spatial heterogeneity. In the early stage of high-tech industry heterogeneity, gradient potential energy and complementarity caused by the potential industrial dissimilation are greater than the scale effect caused by specialization agglomeration, which promotes the development of subregional cooperation. Meanwhile, the space dissimilation of high-tech industries can effectively reduce the congestion cost of core industry area, which further enhances the centripetal force of cooperation and expands the scope and density of cooperation between the core industry district and the supporting industry district. In this way, spatial dependence bears great significance. However, in the process of gradual evolution, the mismatching of the transfer speed and the spatial misallocation of resources caused by excessive industrial spatial heterogeneity leads to the loss of efficiency and the rise of transaction costs and coordination costs, which further reduces the scale effect of subregional cooperation (Niu and Zhang, 2012; Ramani et al., 2017). The gradient potential energy and complementarity are gradually offset by the increasing transaction costs and coordination costs, which further reduces room for subregional cooperation, especially when the centripetal force of the core industry district tends to transform into the centrifugal force and results in a decrease of subregional cooperation ability. The following hypotheses are proposed:

Hypothesis 3a: Spatial heterogeneity of high-tech industries will have a non-linear (inverted U-shaped) effect on subregional cooperation.

Hypothesis 3b: Different subregions are located on both sides of the inverted U-shaped curve.

The subregional cooperation potential and the complementarity of the four restraining factors, illustrated in hypotheses 1a, 1b and 1c, are the core elements of the evolutionary mechanism used to describe the current situation and evolution trends of subregional cooperation. From the perspectives of spatial relations, institutional barriers and industrial heterogeneity, hypotheses 2a, 2b, 3a and 3b provide supplementary environmental factors affecting subregional cooperation.

III Samples, models and variables

Keeping in mind the aforementioned considerations, this study believes that, combining with spatial organization structure of subregional cooperation, a subregional cooperation zone refers to a cluster of border areas and adjacent areas that seek to cooperate with each other. Further, in the testing of evolutionary trends of the subregional cooperation mechanism, provincial subregions in Central China would be a typical sample in line with the conditions of most developing countries. Based on the aforementioned hypotheses and considering the restraining factors, spatial continuity, institutional arrangements and industrial structure of provincial subregional cooperation, variables are set, and the panel regression models are introduced to explore the internal mechanisms and relationships among variables (Judge and Zapata, 2015; Lu and Shang, 2017; Vikas et al., 2004).

1 Sample selection

Central China is a populous economic zone and an important part of the Belt and Road initiative, which connects the east and the west, the north and the south. However, within Central China, social and economic resources are more concentrated in the central regions, while two or more adjacent provincial border regions are often underdeveloped minority-based subregions. In order to grow the provincial subregions, the Thirteenth Five-Year Plan (2016) for Promoting the Rise of the Central China proposed that more efforts should be made to deepen subregional cooperation in the Golden Triangle of the Yellow River and Hunan-Jiangxi Open and Cooperation Experimental Zone. The Program for the Development of Urban Agglomerations in the Middle Reaches of the Yangtze River (2015) also proposed to promote the development of provincial adjacent cities groups, such as Xianning–Yueyang–Jiujiang Industrial Cooperation Zone and Jing-Yue-Chang-Yi Cooperation Zone.

Based on the provincial adjacent border regions identified in the regional policies in Central China, this study put forward eight typical subregions in Central China. These subregions meet the following standards (cf. Figure 1): (a) located in the geographically provincial border region; (b) closely connected in terms of landscape, history, culture and interpersonal relationship; (c) have homorganicity in resource endowment, integrity in environmental function and social and cultural homology and (d) are cooperative and complementary with each other.

Figure 1.

Source: The authors.

The eight typical provincial subregions in Central China include 24 prefectural level cities, six counties (districts) and one county-level forest district: (a) subregion A includes four prefecture cities from three provinces (Shanxi, Henan and Shaanxi); (b) subregion B includes four prefecture cities from two provinces (Henan and Anhui); (c) subregion C includes four prefecture cities from three provinces (Henan, Hubei and Anhui); (d) subregion D includes five prefecture cities and seven counties from three provinces (Chongqing, Hubei and Hunan); (e) subregion E includes three prefecture cities from three provinces (Hubei, Hunan and Jiangxi); (f) subregion F includes three prefecture cities from two provinces (Hubei and Jiangxi); (g) subregion G includes four prefecture cities from two provinces (Anhui and Jiangxi); and (h) subregion H includes four prefecture cities from two provinces (Hubei and Hunan). As the size of a county is different from that of a city, in order to unify the statistical calibre, six counties from Chongqing province are considered as one city in subregion D, which is called ‘Yudong’, and ‘Shennongjia’ county is also regarded as a part of ‘Yichang’ city, which is called ‘Yishen’ (cf. Table 1).

Table 1.

Geographic division of subregional cooperation in Central China

No.	Subregion (8)	Prefecture City (25)	County (7)
A	Golden Triangle of the Yellow River subregion	Linfen Sanmenxia Weinan Yuncheng	– – – –
B	The Middle Reaches of Huai River subregion	Bengbu Fuyang Huainan Xinyang (crossing area)	– – – –
C	Dabie mountain economic subregion	Anqing (crossing area) Huanggang (crossing area) Lu’an Xinyang (crossing area)	– – – –
D	The three gorges eco-economy subregion	‘Yudong’	Wanzhou
			Kaixian
			Yunyang
			Fengjie
			Wushang
			Wuxi
		‘Yishen’	Shennongjia
		Enshi	–
		Jingmen	–
		Jingzhou (crossing area)	–
		Zhangjiajie	–
E	‘Xian-Yue-Jiu’ Small Triangle subregion	Jiujiang (crossing area) Xianning Yueyang (crossing area)	– – –
F	‘Jiu-Huang-Huang’ subregion	Huanggang (crossing area) Huangshi Jiujiang (crossing area)	– – –
G	‘Jiu-An-Chi-Jing’ subregion	Anqing (crossing area) Chizhou Jingdezhen Jiujiang (crossing area)	– – – –
H	Dongting Lake eco-economy subregion	Changde Jingzhou (crossing area) Yiyang Yueyang (crossing area)	– – – –

Source: The authors.

2 Model selection

Gravity model: Derived from Newton’s law of gravity, this model originated from the monopolistic competition model based on increasing returns (Deardorff, 1998; Paul, 1991). This model believes that, for two economy units, their trade volume is directly proportional to their gross national economy (GDP) and inversely proportional to their distance. Based on Stephen and Christine’s study (Stephen and Christine, 1995), the gravity model is introduced to measure spatial interaction among subregions in Central China. Traditionally, the total population or the total GDP was often used in place of the ‘mass’ to make data collection and calculation easier, but in this way, the regional economy quality could not be appropriately measured (Sun and Xu, 2011). In order to distinguish the economy gravity model in this study from the previous one, the dependent variable ‘per capita GDP’ (economy quality) is used to replace the ‘mass’. Besides, this article measures the spatial distance from one region to another (the distance between subregions) as kilometrages of public roads, and the distance decay between cities is denoted as ‘2’ (James and Eric, 2008). Hence, the gravity model of economy cooperation is

C I_{ij,t} {=GQ}_{i,t} Q_{j,t} D_{ij}^{–2} ​

(1)

In Equation (1), CI_ij,t refers to the economy cooperation interaction between region i and j (subregional level or prefectural level), Q_i refers to the per capital GDP of region i, D_ij refers to the distance between region i and j, G refers to gravitational constant and t refers to the time of a year.

Grubel–Lloyd index: To reflect the differences of subregional cooperation conditions, the Grubel–Lloyd index (G–L index for short) is introduced, which indicates that trade takes place either within industries or between industries, and intra-industry trade index is associated with the aggregation of statistical data, with its value range being [0,1] (Grubel and Lloyd, 1975; Proença and Faustino, 2015; Yoshida, 2013). The common expression is:

G ​ - ​ L_{i} = i + M_{i} - |X_{i} - M_{i}| X_{i} + M_{i} \times 100 %

(2)

where X_i and M_i represent the export and import volumes of industry i, respectively. The extreme point 0 means that when a country has only import or only export in a certain industry, the country has no intra-industry trade, or the level of intra-industry trade is minimum. In this case, either X_j or M_j is zero and the G–L index equals to zero. However, the extreme point 1 means that when the import and export is equal in a certain industry, the level of intra-industry trade is maximum in this country. In this case, X_j equals to M_j and the G–L index equals to 1. Therefore, the closer the value of the G–L index is to 0, the lower the level of intra-industry trade is, while the closer the value is to 1, the higher the level of intra-industry trade is.

To describe the differentiation and complementarity of cooperation potential in subregional level and prefectural level (Xu et al., 2015), the G–L index is improved as

G ​ - ​ L_{i j, t, λ} = 1 \frac{| C_{i t, λ} - ​ C_{j t, λ} |}{C_{i t, λ} + C_{j t, λ}} ​

(3)

where i and j represent prefecture cities and λ (λ = 1, 2, 3, 4) represents the subjective factor (λ = 1), internal factor(λ = 2), external factor (λ = 3) and the security factor (λ = 4), respectively. C_{(i,t, λ)} represents the cooperation potential score of prefecture city i on factor λ, which will be measured by the second-class variables listed in Table 2, and t represents the time of a year.

Table 2.

Definitions of variables

	Variable	Definition
Cooperation potential	Subjective factors (Sub)	Local public financial expenditure ÷ total GDP
		Gross industrial output value above designated size
		R&D personnel
	Internal factors (Int)	Gross output value of inner funded enterprises
		Total retail sales of consumer goods
		Total imports
	External factors (Ext)	FDI
		Total export-import volume ÷ total GDP
	Security factors (Sec)	Internet service (100 subscribers)
		Number of students enrolled in institutions of higher education
		Number of patent applications examined
		Earnings from domestic tourism
		Deposits and loans balance of banking institutions ÷ total GDP
		Area of green land per 10,000 persons (total city)
		Public road network density (public road mileage ÷ total land area)
Spatial heterogeneity of high-tech industry (H-tSH)		$(H - t S H_{i j, t}) = (H - t I A_{i, t}) + (H - t I A_{j, t}) \| (H - t I A_{i, t}) Ô ø Ω (H - t I A_{j, t}) \|$
Control variables	Neighbour (Nei)	Dummy variable (0 or 1)
	Institution (Ins)	Dummy variable (0 or 1)
	Border (Bor)	Dummy variable (0 or 1)

Source: The authors.

Regression model: To deal with paired samples in many regions, according to the theoretical hypothesis, this study offers an adaptable regression model based on two spatial dimensions of panel data (subregional level or prefectural level), with G–L indexes of cooperation mechanism as explanatory variables to test hypothesized influences on economic connection (Nonthapot, 2014). Stepwise regression method is used to enhance the robustness of regression, while logarithmic quantization is used to maintain data stability and eliminate heteroscedasticity (Yang, 2018). Subregional cooperation is a dynamic evolution process with interactions between environmental influencing factors and regional cooperation. These considerations suggest the dynamic panel regression equation

\begin{matrix} \ln C I_{i j, t} = a_{0} + a_{λ} \sum_{λ = 1}^{4} \begin{matrix} \ln (G ​ - ​ L_{i j t, λ}) \\ + a_{5} \ln (H ​ - ​ t S H_{i j, t}) \end{matrix} \\ + a_{6} {[\ln (H ​ - ​ t S H_{i j, t})]}^{2} + a_{θ} X_{i j} + ε_{i j, t} ​ \end{matrix}

(4)

where i and j represent subregions or prefecture cities, t represents the time of a year, α₀-α_θ represent the undetermined parameter, ε represents a disturbance term that changes with individual and time and [H-tSH_ij,t]² is a binomial that reflects influences of the heterogeneity of high-tech industries on subregional cooperation. X_ij represents a sequence of control variables (dummy variables), in which Neighbour (Nei), Border (Bor) and Institution (Ins) represent spatial continuity, border effect and institutional barriers, respectively.

N e i_{i j} = ​ \{\begin{matrix} 1 ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ w h e n i a n d j a r e n e i g h b o u r \\ 0 ​ ​ ​ ​ ​ ​ ​ ​ ​ o t h e r w i s e \end{matrix}

(5)

I n s_{i j} = \{\begin{matrix} 1 ​ ​ ​ ​ ​ ​ ​ ​ ​ w h e n i a n d j b e l o n g t o d i f f e r e n t p r o v i n c e s \\ 0 ​ ​ ​ ​ ​ ​ ​ o t h e r w i s e \end{matrix}

(6)

B o r_{i j} = \{\begin{matrix} 1 ​ ​ ​ ​ ​ ​ ​ w h e n i a n d j a r e n e i g h b o u r b u t b e l o n g t o d i f f e r e n t p r o v i n c e s \\ 0 ​ ​ ​ ​ ​ ​ ​ ​ o t h e r w i s e \end{matrix}

(7)

Each variable is bilateral, in that it applies to both i and j (cf. Table 2 and Appendix A).

3 Data and definition of variables

In this article, the prefectural panel data of eight typical provincial subregions in China from 2011–2015 were collected. There were a total of 125 samples and 1,500 observation values ( $5 \times C_{25}^{2} $ ), with subregional data obtained by summing up cities’ data. The main sources of the above data include China City Statistical Yearbook (2012–2015), statistical bulletins, annual reports and statistical yearbooks of Shaanxi, Shanxi, Henan, Anhui, Hubei, Hunan, Chongqing and Jiangxi. Through the subjective factors (Sub), internal factors (Int), external factors (Ext) and security factors (Sec) that impact subregional cooperation, a comprehensive score of the four influencing factors was calculated to reflect the degree of cooperation of regions, Z_i,t = f(Sub_i,t, Int_i,t, Ext_i,t, Sec_i,t).

Subregional cooperation is essentially institutional innovation of the spatial structure with cross-border distribution of institutional innovation benefits as one main feature. The system design needs to ensure that each subregion benefits from such arrangements. However, as multiple stakeholders are involved, the institutional innovation will inevitably conflict with the original system and mechanisms, which goes against the synergistic effect between subregions (Fleisher et al., 2010). One of the institutional barriers that causes the institutional space conflict is the differences between regions’ institutional arrangements. The greater the institutional differences, the higher the cost of collaboration. Another institutional barrier is market segmentation and local protectionism, especially when the local government overly pursues its own interests under the decentralized system, the restrictive factors of connection between cities worsen the institutional obstacle of subregional cooperation. Therefore, this article introduces a set of control variables to reflect the influence of geographical and administrative factors on the efficiency of subregional cooperation.

4 Descriptive statistical analysis

The classification index of the comprehensive score consists of four secondary-level variables and 15 tertiary-level variables (cf. Table 2). Due to the length, the details of the measurement process are omitted in this article (cf. Table B1 and Figures C1 and C2).

In order to improve multicollinearity, this study used the statistical analysis software SPSS17 to analyze the level of cooperation of subregions in different years by means of generalized principal component analysis (GPCA) (Keum, 2010; Yin et al., 2017). The index weight obtained through this method is more objective than that through the traditional principal component analysis. First, based on the original observation data, the dimensionless variables are obtained using the method of Z-score to enable validity analysis, in which variables pass the Kaiser-Meyer-Olkin (KMO) test and the Bartlett spherical test. Second, in the principal component analysis, three principal components are extracted, while the explained variance accounted for 88.59 per cent of the total variance, which is suitable for factor analysis. Moreover, through the product of the explained variance from the total variance explained table and components from the rotated component matrix, the weights of the observation index are obtained. By accumulating the product of observation variables’ weights and Z-score, the score of the second-level index and the total score are calculated. Tables 3 and 4 present the statistical analysis results. The results indicate that the time and spatial differences of subregional cooperation in Central China are notable in the subregional level and prefectural level.

Table 3.
Comprehensive score of eight typical provincial subregions

Year
Subregions 2011 2012 2013 2014 2015

Golden Triangle of the Yellow River subregion 0.989 1.083 1.146 1.192 1.180

The Middle reaches of the Huai River subregion 0.843 0.942 1.042 1.119 1.204

Dabie mountain economic subregion 1.016 1.137 1.245 1.367 1.450

The three gorges eco-economy subregion 1.405 1.611 1.795 1.989 2.145

‘Xian-Yue-Jiu’small triangle subregion 0.887 1.021 1.143 1.261 1.356

‘Jiu-Huang-Huang’subregion 0.752 0.851 0.950 1.043 1.100

‘Jiu-An-Chi-Jing’subregion 0.794 0.891 0.970 1.064 1.126

Dongting Lake eco-economy subregion 1.314 1.504 1.667 1.845 1.990

Year Subregions	2011	2012	2013	2014	2015
Golden Triangle of the Yellow River subregion	0.989	1.083	1.146	1.192	1.180
The Middle reaches of the Huai River subregion	0.843	0.942	1.042	1.119	1.204
Dabie mountain economic subregion	1.016	1.137	1.245	1.367	1.450
The three gorges eco-economy subregion	1.405	1.611	1.795	1.989	2.145
‘Xian-Yue-Jiu’small triangle subregion	0.887	1.021	1.143	1.261	1.356
‘Jiu-Huang-Huang’subregion	0.752	0.851	0.950	1.043	1.100
‘Jiu-An-Chi-Jing’subregion	0.794	0.891	0.970	1.064	1.126
Dongting Lake eco-economy subregion	1.314	1.504	1.667	1.845	1.990

Source: The authors.

Table 4.

Comprehensive score of 25 prefecture cities

Year Prefectural Cities	2011	2012	2013	2014	2015
Yuncheng	1.003	1.054	1.125	1.186	1.158
Linfen	1.120	1.204	1.207	1.197	1.146
Sanmenxia	1.049	1.133	1.189	1.224	1.244
Weinan	1.015	1.196	1.331	1.441	1.449
Xinyang	1.260	1.390	1.560	1.734	1.853
Fuyang	0.842	0.950	1.048	1.131	1.250
Huainan	0.700	0.771	0.808	0.779	0.760
Bengbu	0.770	0.878	0.994	1.094	1.236
Huanggang	1.031	1.177	1.315	1.457	1.568
Anqing	1.199	1.341	1.399	1.524	1.592
Lu’an	0.810	0.906	0.997	1.072	1.128
‘Yudong’	1.155	1.270	1.394	1.550	1.680
‘Yishen’	2.126	2.492	2.797	3.111	3.362
Jinzhou	1.029	1.180	1.317	1.461	1.569
Jinmen	0.930	1.071	1.187	1.293	1.370
Enshi	0.413	0.476	0.545	0.604	0.662
Zhangjiajie	0.294	0.334	0.361	0.405	0.442
Yueyang	1.874	2.170	2.398	2.634	2.848
Jiujiang	1.240	1.401	1.580	1.756	1.877
Huangshi	0.914	1.026	1.129	1.202	1.212
Chizhou	0.368	0.412	0.456	0.497	0.537
Jingdezhen	0.557	0.620	0.671	0.728	0.762
Xianning	0.643	0.751	0.860	0.951	1.016
Changed	1.787	2.011	2.235	2.481	2.673
Yiyang	0.872	1.007	1.108	1.236	1.336

Source: The authors.

IV Regression results and discussion

Based on the improved model and relevant variables, the study carried out the goodness-of-fit testing to assess the overall fitting effect by using Eviews8.0. The testing results suggest all indicated good fit and model performance, with results of parameter estimation shown in Table 5.

Table 5.
Estimation results and robustness check of fixed effect model

Variable (1)
(2)
(3)
(4)
(5)
(6)

Subregional Level
Prefectural Level
Robustness Test

Section One-period Lag Section Section One-period Lag

LnG-L_Sub 0.334* 0.228* 0.305* 0.883* 0.663 0.522*

(1.756) (1.902) (1.842) (1.772) (1.208) (1.916)

LnG-L_Int 0.358* 0.509 1.482 0.992* 3.142* 1.332*

(4.062) (2.343) (2.152) (3.787) (5.224) (5.192)

LnG-L_Ext 1.307* 0.715 0.558 0.847 0.379 0.874**

(1.823) (2.238) (2.009) (2.035) (2.115) (2.229)

LnG-L_Sec −0.554* −0.256 0.534 −1.307 −1.258 −1.238*

(−1.833) (−1.461) (0.404) (−1.442) (−0.937) (−1.825)

LnG-L_Sub × LnG-L_Sec −1.399*
(−1.847) −1.207
(−0.497) 1.544
(0.708) −0.539*
(−1.911)

H-tCP 2.886** 1.205* 0.996* 1.025* 1.889* 1.334*

(2.359) (1.776) (1.798) (1.844) (1.703) (1.925)

(H-tCP) ² −1.223 −0.774 −0.146* −0.869 −0.352* 2.376*

(−2.323) (−2.038) (−4.722) (−2.118) (−4.580) (5.039)

Bor −2.115* −1.778* −3.116* −3.109*

(−5.322) (−5.97) (−7.242) (−6.221)

Bor × LnG-L_Ext −1.322*
(−5.229) −4.966
(−2.339) −2.772**
(−2.409)

Ins −2.002** −2.228* −3.224* −0.756**

(−2.117) (−1.802) (−1.772) (−2.312)

Nei 2.993* 4.515* 1.226* 4.754*

(7.803) (6.244) (3.407) (6.228)

C −2.883* −2.994* −2.111* −5.002 −3.009 −4.035*

(−4.223) (−4.967) (−3.228) (−4.118) (−5.882) (−5.117)

Adjusted R² 0.320 0.248 0.521 0.408 0.239 0.244

Obs 140 119 1,284 1,284 1,425 1,368

Variable	(1)	(2)	(3)	(4)	(5)	(6)
LnG-L_Sub	0.334*	0.228*	0.305*	0.883*	0.663	0.522*
	(1.756)	(1.902)	(1.842)	(1.772)	(1.208)	(1.916)
LnG-L_Int	0.358***	0.509**	1.482**	0.992***	3.142***	1.332***
	(4.062)	(2.343)	(2.152)	(3.787)	(5.224)	(5.192)
LnG-L_Ext	1.307*	0.715**	0.558**	0.847**	0.379**	0.874**
	(1.823)	(2.238)	(2.009)	(2.035)	(2.115)	(2.229)
LnG-L_Sec	−0.554*	−0.256	0.534	−1.307	−1.258	−1.238*
	(−1.833)	(−1.461)	(0.404)	(−1.442)	(−0.937)	(−1.825)
LnG-L_Sub × LnG-L_Sec		−1.399* (−1.847)	−1.207 (−0.497)		1.544 (0.708)	−0.539* (−1.911)
H-tCP	2.886**	1.205*	0.996*	1.025*	1.889*	1.334*
	(2.359)	(1.776)	(1.798)	(1.844)	(1.703)	(1.925)
(H-tCP) ²	−1.223**	−0.774**	−0.146***	−0.869**	−0.352***	2.376***
	(−2.323)	(−2.038)	(−4.722)	(−2.118)	(−4.580)	(5.039)
Bor			−2.115***	−1.778***	−3.116***	−3.109***
			(−5.322)	(−5.97)	(−7.242)	(−6.221)
Bor × LnG-L_Ext				−1.322*** (−5.229)	−4.966** (−2.339)	−2.772** (−2.409)
Ins			−2.002**	−2.228*	−3.224*	−0.756**
			(−2.117)	(−1.802)	(−1.772)	(−2.312)
Nei			2.993***	4.515***	1.226***	4.754***
			(7.803)	(6.244)	(3.407)	(6.228)
C	−2.883***	−2.994***	−2.111***	−5.002**	−3.009**	−4.035***
	(−4.223)	(−4.967)	(−3.228)	(−4.118)	(−5.882)	(−5.117)
Adjusted R²	0.320	0.248	0.521	0.408	0.239	0.244
Obs	140	119	1,284	1,284	1,425	1,368

Source: The authors.

Notes: *p < 0.1; **p < 0.05; ***p < 0.01. T-statistic are provided in parentheses.

1 Impacts of the responsiveness and complementarity

Across the fixed effect model, the results of the provincial subregional cooperation mechanism suggest that the complementarity of endogenous and exogenous mechanisms better explains the rationale behind the provincial subregional cooperation in Central China. The two variables are significant in regions with both subregion levels and prefectural levels, and in particular, the significance of the endogenous mechanism is at least 5 per cent. The complementarity of the subjective mechanism and the security mechanism, however, failed to pass the significance test, indicating low synchronicity of these two variables in provincial subregional cooperation and separation in the interaction between the two variables. Therefore, this article further examined the changes needed in the influence of the complementarity of the subjective mechanism and the security mechanism on provincial subregional cooperation in Central China. After the interaction term was introduced, the results showed that the coefficient of the security mechanism still failed to pass the significance test and that the interaction items had a negative effect on the explained variable. This means that the subjective mechanism on security mechanism is in need of promotion, with the responsiveness needed to be further reinforced and optimized.

In addition, the differences of standardized regression coefficients and T-statistic between provincial subregions in Central China were obvious, which indicates that the influences of the four cooperation mechanisms and high-tech industries’ development on subregional cooperation vary depending on the mode, direction and intensity of action. The complementarity of the subjective mechanism and the internal mechanism is not significant, and the release of internal market potential promoted by behavioural agents in the process of subregional cooperation is limited, which further shows that the government did not give full play to its leading role. However, the deepening market-oriented reform has enabled the release of this vitality, which significantly promotes subregional cooperation, and the internal mechanism plays a key role in provincial subregional cooperation. For the external mechanism, the underdeveloped private economy, the incomplete and backward market, the insufficient openness and ineffective use of foreign capital led to weak driving forces of subregional cooperation and failed to create an effective external mechanism for cooperation. To foster and promote cooperation, external resources need further integration. The security mechanism was supposed to have a positive impact on provincial subregional cooperation, but the results failed to confirm this hypothesis. The possible explanation is that the incentive and restraint mechanism in provincial subregions is still absent. What is more, there is a vacuum zone in the security mechanism, which means insufficient support to provincial subregional cooperation.

2 Border effect, resource allocation efficiency and institutional barriers

In terms of the border effect factor, the variable Bor is significantly negative (p < 0.01), indicating that provincial subregional co-operation in Central China is strongly impacted by the shielding effect of borders. Due to local protectionism, each city focusses more on the economic development of their own region when planning the cooperation layout and often neglects the interactions of subregional cooperation. The interaction term between the boundary effect and the external mechanism further explains that the ineffective cross-border subregional cooperation is caused by obstacles to the reconfiguration of resource elements, especially the local allocation efficiency of external resources obviously weakened by the boundary. Therefore, policy innovation is necessary, such as institutional changes of the administrative and management systems, and design of benefit sharing and compensation mechanisms. This article also tested institutional barriers between prefectural cities, which are the most difficult in institutional changes and policy innovation. The significantly negative simulation results indicate that due to administrative intervention, institutional changes lack motivation and policy innovation is deficient, which hinder the creation and development of CBC mechanisms in provincial subregions. The cooperation efficiency in cross-border prefectures is obviously lower than that in inner prefectures, which further forms a ‘forced’ mechanism to promote the spatial evolution of provincial subregional cooperation switch from the governance structure of traditional administrative division to that of CBC.

3 Geographical proximity and spatial extension

In Model (3), the estimated result of the geographical proximity effect is 2.993 (p < 0.01), which suggests that geographical proximity effect contributes to the creation and development of subregional cooperation. Compared to geographical crossing, geographical proximity leads to a decrease in transport costs and an increase in scale benefits, which is an important internal mechanism for the evolution of subregional cooperation. In this study, geographical proximity can be expressed as inner provincial proximity or interprovincial proximity. Previous analysis indicates that both the shielding effect and institutional barriers often restrain spatial expansion of subregional cooperation, which makes the spatial structure of subregional cooperation obviously ‘track the boundary’. However, when the opening boundary generates a mediating effect and when institutional barriers are mitigated, one subregion will be forced to expand to an adjacent subregion.

4 Inverted U-shaped curve of heterogeneity of high-tech industries

According to the regression results, the coefficient of H-tCP is positive, while that of its quadratic term is negative (p < 0.05) in Model (1) to Model (5), which proves that the spatial heterogeneity of high-tech industries has an inverted U-shaped effect on subregional cooperation in both subregional level and prefectural level. In other words, with the evolution of spatial heterogeneity of high-tech industries, the cooperation performance in both subregional level and prefectural level will significantly increase, but as spatial heterogeneity rises above the critical value, cooperation performance tends to worsen. According to a further calculation, only 9 subregional samples and 83 prefectural samples are on the right side of the critical value, indicating that most samples are still in the stage where subregional cooperation develops with the expansion of spatial heterogeneity of high-tech industries.

V Conclusions

As a subregional cooperation zone is the product of collaboration among a series of cooperation mechanisms, overall planning is necessary to include four mechanisms, industrial layout and the cooperation framework agreement. It is also necessary to improve the subjective mechanism, especially the responsiveness of the security mechanism in provincial subregional cooperation. The functional positioning of different mechanisms in the process of subregional cooperation should be clear, and the mechanisms should have targeted improvement according to their functional differentiation. The complementarities between the dynamic mechanism, the security mechanism and the subjective mechanism are significant incentives for provincial subregional cooperation. However, the regression results indicate that the complementarity between the subjective mechanism and the dynamic mechanism is significant at a low level, which needs to be further improved, while the complementarity between the subjective mechanism and the security mechanism is insignificant, suggesting that horizontal linkages between cooperative mechanisms need to be strengthened. First, in subregional cooperation, this study proposed to establish a regional joint conference mechanism and a working mechanism to regularly negotiate and arrange significant problems in the cooperation mechanism together. Second, regional cooperation framework agreements could improve the government subjective mechanism. Third, the study suggests that guidance from and improvement of participation and enthusiasm of non-governmental organizations has the potential to promote subregional cooperation. Fourth, the demonstration effect of the model of other subregional cooperation could be further exploited. Fifth, based on laws and policies to standardize rules, the incentive and restraint mechanism could be improved to ensure compliance with the rules for reward and punishment. Sixth, subregional cooperation needs to coordinate the conflicts of interests, explore various means of benefit distribution and improve efficiency and equity through seamless integration of benefit sharing and compensation. Finally, subregional cooperation needs to speed up the development of integrated mechanisms, especially innovation in investment and financing, risk control and incentives for acceptance of cooperative projects.

As a cross-border spatial structure, subregional cooperation zones are plagued by institutional barriers, which go against regional coordination. Therefore, to develop cross-border (provincial level) subregional cooperation zones, this article proposed to establish provincial coordinating mechanisms led by the central government, while for cross-border (prefectural level) subregional cooperation zones, the prefectural coordinating mechanisms should be introduced by the provincial government. Development plans for subregional cooperation zones would enable resource reallocation. Furthermore, by breaking administrative boundaries and strengthening strategic coordination of technology, human resources, credit system, market access, mutual quality certification, social management and government services, the governance structure of subregional cooperation zones will be established, and the spatial allocation efficiency of resources will be improved continuously. Provincial subregional cooperation zones could be upgraded and transformed from a subregional cooperation backward zone to a subregional cooperation advancement zone and to a subregional cooperation demonstration zone.

The evolution of the spatial network structure of subregional cooperation is a gradual process from simple (low-level) to complex (high-level), which not only increases internal elements and expands in space but also optimizes the system structure and function. To effectively expand subregional cooperation zones, a top-level design is essential for the subregional cooperation network structure, drawing lessons from the success of the Yangtze River Delta and the Pearl River Delta. At the same time, breaking barriers posed by the original boundary in subregions, improving market integration and reshaping borders are all conductive to the continuous expansion of coverage based on the eight typical provincial subregional cooperation zones. For example, findings of this article support the establishment of the Xinyu–Yichun–Pingxiang Open Cooperation Zone and incorporation of Anqing into the Xian-Yue-Jiu Small Triangle subregion. Moreover, it is important to understand the comparative strengths and weaknesses of each area in the subregional network to identify growth opportunities, deal with relationships among areas and strengthen the interactions, mutual visits, mutual growth and mutual assistance within and beyond subregions. Our findings may further suggest that the demonstration effect of provincial subregional cooperation zones can be leveraged in other developing countries and regions to put in place a reasonable, dense and overlapped subregional cooperation network structure.

Overall, our findings indicate that subregions should actively leverage the industrial gradient potential energy released by urban agglomerations and economic belts with external demonstration effect, develop dominant industrial clusters, promote industrial transfer of cross-border subregions and put in place mechanisms for coordinated industrial development. Subregional cooperation strong areas should be the spillover centres in subregional high-tech cooperation, with comprehensive infrastructure, advantageous geographical conditions, better industrial base and higher resource agglomeration. Subregional cooperation weak areas should actively leverage the transfer of both the industrial chain and human resource chain, improve their own industrial structures and optimize potential differentiation through the externality brought by the gradient potential energy from the spillover centre.

As the global economic growth pattern shifts, the macro environment for subregional cooperation and division has changed, making a cooperative and coordinated horizontal governance structure more necessary to transform the subregional development from a government-based vertical governance structure to a criss-cross governance structure. Strategic cooperation framework agreements within regions should in place to secure institutional arrangements for integration and a network governance structure. They will also pave the way for further opening up of markets, elimination of institutional barriers and decrease of transaction costs in subregions.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

This work was supported by the Fundamental Research Funds for the Central Universitiesunder Grant 2017-YB-016. We gratefully appreciate these supports.

Formula appendix

However, in this study, spatial heterogeneity of high-tech industrial agglomeration (H-tSH_ij,t for short) is used to describe differentiation and complementarity of high-tech industrial agglomeration among eight subregions or 25 prefectural cities. According to the Balassa index (Bela, 1965) and ‘measurement of complementarity of the Four Modernizations Level’ (Xu et al., 2015), the H-tSH_ij,t can be revealed as follows: (A.1)

(H ​ - ​ t S H_{i j, t}) = \frac{(H ​ - ​ t I A_{i, t}) + (H ​ - ​ t I A_{j, t})}{| (H ​ - ​ t I A_{i, t}) Ô ø Ω (H ​ - ​ t I A_{j, t}) |} ​

where H-tIA_i,t or H-tIA_j,t represents the high-tech industrial agglomeration of each subregion or prefectural city.

(A.2)

H ​ - ​ t I A_{i, t} = \sum_{m = 1}^{5} \frac{Y_{i m, t}}{\sum_{i = 1}^{I} Y_{i m, t}} ​

where Y refers to the output value of high-tech industry. m refers to the high-tech industrial classification, which are generally divided into five categories: pharmaceutical industry, aerospace industry, electrocommunication and communications equipment industry, electronic computer and office equipment industry, medical equipment and instrument industry. i refers to the subregion (I = 8) or prefectural city (I = 25) and t refers to the time of year. For econometry, the data were mainly obtained from ‘China City Statistical Yearbook (2012–2015)’ and ‘China’s High-tech Industry Statistical Yearbook (2012–2015)’.

Appendix B

Table B1.

Classification of 25 prefecturalities’ comprehensive scores

Year Prefecturalities	2011	2012	2013	2014	2015
Yuncheng	W	W	W	W	W
Linfen	W	S	S	W	W
Sanmenxia	W	W	W	S	S
Weinan	W	W	S	S	S
Xinyang	S	S	S	S	S
Fuyang	W	W	W	W	S
Huainan	W	W	W	W	W
Bengbu	W	W	W	W	S
Huanggang	W	W	S	S	S
Anqing	W	S	S	S	S
Lu’an	W	W	W	W	W
‘Yudong’	W	S	S	S	S
‘Yishen’	S	S	S	S	S
Jinzhou	W	W	S	S	S
Jinmen	W	W	W	S	S
Enshi	W	W	W	W	W
Zhangjiajie	W	W	W	W	W
Yueyang	W	W	W	W	W
Jiujiang	S	S	S	S	S
Huangshi	S	S	S	S	S
Chizhou	W	W	W	S	S
Jingdezhen	W	W	W	W	W
Xianning	W	W	W	W	W
Changde	S	S	S	S	S
Yiyang	W	W	W	S	S

Source: The authors.

Notes: The threshold of comprehensive score was defined as 1.2, when Z_i,t > 1.2, the prefecturality is defined as a cooperation strong area, denoted by S. When Z_i,t ≤ 1.2, the prefecturality is defined as a cooperation weak area denoted by W. In Figures C1 and C2, red colour indicates area S, while green colour indicates area W. The figures are available in colour in the online version.

Appendix C

References

Bat

, Hitt

M.A.

and Tsui

A.S.

2013: Institutional polycentrism, entrepreneurs’ social networks, and new venture growth. Academy of Management Journal, 56(4), 1024–49.

Bela

1965: Trade liberalisation and ‘revealed’ comparative advantage. The Manchester School, 33(2), 99–123.

Castanho

R.A.

. 2017: Accessibility and connectivity—Movement between cities, as a critical factor to achieve success on cross-border cooperation (CBC) projects. A European analysis. Sustainable Cities and Society, 32, 181–90.

Castanho

R.A.

2018: Identifying critical factors for success in cross border cooperation (CBC) development projects. Habitat International 72, 92–99.

Deardorff

V.A.

1998: Determinants of bilateral trade: Does gravity work in a neoclassical world? University of Chicago Press, 7–22.

Fleisher

B.M.

. 2010: Human capital, economic growth and regional inequality in China. Journal of Development Economics, 92(2), 215–31.

Grubel

H.J.

and Lloyd

P.J.

1975: Intra-industry trade: The theory and measurement of international trade in differentiated products. John Wiley & Sons.

Han

C.Z.

and Shi

Y.H.

2014: Bottlenecks in east Asia’s regional cooperation. CIR, 24, 29–37.

Hanson

G.H.

2001: U.S.-Mexico integration and regional economies: Evidence from border-city pairs. Journal of Urban Economics, 50(2), 259–87.

10.

Hiroyuki

and Keiichiro

2014: Trade integration of Yunnan and Guangxi with the Greater Mekong Sub-region re-visited. China Economic Policy Review, 3, 1–14.

11.

Z.D.

. 2010: Sub-regional cooperation and the causes of its development: Analysis on an interdisciplinary perspective. World Regional Studies, 19, 34–41.

12.

Jacobs

2016: Spatial planning in cross-border regions: A systems-theoretical perspective. Planning Theory, 15(1), 68–90.

13.

James

E.A.

and Eric

V.W.

2008: Gravity with gravitas: A solution to the border puzzle. American Economic Review, 93, 170–92.

14.

Judge

T.A.

and Zapata

C.P.

2015: The person-situation debate revisited: Effect of situation strength and trait activation on the validity of the big five personality traits in predicting job performance. Academy of Management Journal, 58(4), 1149–79.

15.

Keum

2010: Tourism flows and trade theory: A panel data analysis with the gravity model. The Annals of Regional Science, 44(3), 541–557.

16.

Kukovic

and Hacek

2018: Cross-border cooperation as a tool of escaping crisis conditions in Slovenia. Transylvanian Review of Administrative Sciences, 53E, 54–66.

17.

Kurowska-Pysz

2016: Opportunities for cross-border entrepreneurship development in a cluster model exemplified by the Polish-Czech border region. Sustainbility, 8, 21.

18.

Kurowska-Pysz

. 2017: The analysis of the determinants of sustainable cross-border cooperation and recommendations on its harmonization. Sustainability, 9, 24.

19.

Liang

S.L.

2013: An analysis of the classification of economic integration and the sub-regional international economic integration. Reformation & Strategy, 29, 36–40.

20.

Liang

S.L.

. 2012: An research on the mechanism of regional agglomeration in the course of sub-regional economic integration. South China Journal of Economy, 9, 143–155.

21.

Lin

2012: Dynamics of cross border industrial development in Mekong Sub-region: A case study of Thailand. Singapore Journal of Tropical Geography, 33, 270–81.

22.

G.Y.

and Shang

G.Z.

2017: Impact of supply base structural complexity on financial performance: Roles of visible and not-so-visible characteristics. Journal of Operations Management, 53–56, 23–44.

23.

Makkonen

2018: Developing survey metrics for analysing cross-border proximity. Geografisk tidsskrift-Danish Journal of Geography, 118, 114–21.

24.

Mukherjee

2014: The South Asian Association for Regional Cooperation: Problems and prospects. Progress in Development Studies, 14(4), 373–81.

25.

Nicholas

N.N.

2014: International trade and agricultural production: Evidence from the Southern African development community sub-region. Development Southern Africa, 31(3), 494–514.

26.

Niu

C.H.

and Zhang

2012: On the relationship among science and technology talent accumulation, high-tech industry agglomeration and regional technologic innovation. Science & Technology Progress and Policy, 29, 46–51.

27.

Nonthapot

2014: The relationship between tourism and economic development in the Greater Mekong Sub-region: Panel cointegration and granger causality. Journal of Advanced Research in Law and Economics, 5(1), 44–51.

28.

Ocampo

J.A.

and Titelman

2009: Subregional financial cooperation: The South American experience. Journal of Post Keynesian Economics, 32(2), 249–68.

29.

Oehlers

2006: A critique of ADB policies towards the Greater Mekong Sub-region. Journal of Contemporary Asia, 36(4), 464–78.

30.

Paul

1991: Increasing returns and economic geography. The Journal of Political Economy, 99(3), 483–99.

31.

Proença

and Faustino

H.C.

2015: Modelling bilateral intra-industry trade indexes with panel data: A semiparametric approach. Computational Statistics, 30(3), 865–884.

32.

Ramani

S.V.

. 2017: High-value hi-tech product introduction in emerging countries: The role and construction of legitimacy. Qualitative Market Research Journal, 20(2), 208–25.

33.

Stephen

F.W.

and Christine

A.W.

1995: Forecasting tourism demand: A review of empirical research. International Journal of Forecasting, 11(3), 447–75.

34.

Sun

and Xu

C.Z.

2011: The construction and application of the ‘economic gravity’ model on the perspective of spatial economics: Example from the economic data of Yangtze River Delta in 2010. Economist, 7, 37–44.

35.

Valentin

. 2016: In search of emerging polities: Thematic agendas of selected European cross-border cooperation structures. Romanian Journal of European Affairs, 16(1), 57–71.

36.

Vikas

. 2004: Geographic patterns in customer service and satisfaction: An empirical investigation. Journal of Marketing, 68(3), 48–62.

37.

Wang

Z.B.

. 2008: Measurement of the border effect of border regions: A case study of Huai-hai economic zone. Economic Geography, 28(5), 765–70.

38.

Wang

Z.F.

. 2018: Research on micro-mechanism of cross-border tourism cooperation based on transport network: A case study of Wuling Mountain Area. Geographical Research, 37, 250–62.

39.

Xing

W.B.

and Li

S.T.

2010: Gravity model, border effect and interregional trade in China: Evidence from regional input-output data. Journal of International Trade, 10, 32–41.

40.

Xu . 2015: Research on formation and evolution mechanism of ‘four modernizations synchronous zone’ based on spatial continuity. China Industrial Economic, 5, 18–31.

41.

Yoshida

2013: Intra-industry trade, fragmentation and export margins: An empirical examination of sub-regional international trade. North American Journal of Economics and Finance, 24, 125–38.

42.

Yang

2018: More efficient local polynomial regression with random-effects panel data models. Econometric Reviews, 37(7), 760–76.

43.

Yin

Z.Y.

. 2017: Pubic security expenditure, education investment, and social stability: An empirical analysis based on provinicial panel data from China. Revista de cercetare si interventie sociala, 59, 239–60.

44.

Zeng

2015: Border effect and economic development of the provincial border region: From the prospective of new economic geography. Finance & Economics, 9, 87–97.

45.

Zhou

and Tao

2011: Bureaucratic promotion competition and border effect: Evidence from economic development in China’s border regions. Journal of Financial Research, 3, 15–18.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Subregional Level		Prefectural Level			Robustness Test
	Section	One-period Lag	Section	Section	One-period Lag	Robustness Test
LnG-L_Sub	0.334*	0.228*	0.305*	0.883*	0.663	0.522*
	(1.756)	(1.902)	(1.842)	(1.772)	(1.208)	(1.916)
LnG-L_Int	0.358***	0.509**	1.482**	0.992***	3.142***	1.332***
	(4.062)	(2.343)	(2.152)	(3.787)	(5.224)	(5.192)
LnG-L_Ext	1.307*	0.715**	0.558**	0.847**	0.379**	0.874**
	(1.823)	(2.238)	(2.009)	(2.035)	(2.115)	(2.229)
LnG-L_Sec	−0.554*	−0.256	0.534	−1.307	−1.258	−1.238*
	(−1.833)	(−1.461)	(0.404)	(−1.442)	(−0.937)	(−1.825)
LnG-L_Sub × LnG-L_Sec		−1.399* (−1.847)	−1.207 (−0.497)		1.544 (0.708)	−0.539* (−1.911)
H-tCP	2.886**	1.205*	0.996*	1.025*	1.889*	1.334*
	(2.359)	(1.776)	(1.798)	(1.844)	(1.703)	(1.925)
(H-tCP) ²	−1.223**	−0.774**	−0.146***	−0.869**	−0.352***	2.376***
	(−2.323)	(−2.038)	(−4.722)	(−2.118)	(−4.580)	(5.039)
Bor			−2.115***	−1.778***	−3.116***	−3.109***
			(−5.322)	(−5.97)	(−7.242)	(−6.221)
Bor × LnG-L_Ext				−1.322*** (−5.229)	−4.966** (−2.339)	−2.772** (−2.409)
Ins			−2.002**	−2.228*	−3.224*	−0.756**
			(−2.117)	(−1.802)	(−1.772)	(−2.312)
Nei			2.993***	4.515***	1.226***	4.754***
			(7.803)	(6.244)	(3.407)	(6.228)
C	−2.883***	−2.994***	−2.111***	−5.002**	−3.009**	−4.035***
	(−4.223)	(−4.967)	(−3.228)	(−4.118)	(−5.882)	(−5.117)
Adjusted R²	0.320	0.248	0.521	0.408	0.239	0.244
Obs	140	119	1,284	1,284	1,425	1,368