Intercity asymmetrical linkages influenced by Spring Festival migration and its multivariate distance determinants: a case study of the Yangtze River Delta Region in China

As open systems, cities establish intricate connections with other cities and regions through various means, such as economic, social, and cultural interactions. With the rapid advancement of information technology and the widespread implementation of transportation infrastructure supported by physical and virtual networks, intercity linkages are evolving from hierarchical to networked connections. Additionally, linkages based on urban functions are gradually being replaced by those driven by frequent flows of factors between cities. These flows, encompassing different natures and levels, serve as the tangible drivers of intercity linkages, reflecting the connections between cities from various perspectives. A growing focus has recently been on population flows as a critical lens to illustrate intercity linkages. However, transportation and communication technology advancements have resulted in significant changes in human interactions. People can now travel longer distances using automobiles, high-speed rail, and airplanes. Consequently, the frictional impact of geography on human mobility and social interaction has been argued to be diminishing—referred to as the "death of distance"—suggesting that modern information and communication technologies (ICT) have made geographic distance less restrictive for intercity social connections. Critics argue that this exaggerates the role of technological advancements: while technology expands social mobility, individuals may not fully leverage it, and people tend to rely on existing social relationships. This debate has spurred empirical studies using emerging data sources (location-based services, online social networks, cell phone communication) to illuminate the topic. In mobility, spatially dispersed areas are increasingly interconnected and communicated across borders. While mobility contributes to the homogenization of different places, it also underscores the significance of geographical differences in everyday life and the functioning of social, political, and economic forces. The pursuit and construction of unique, distinct spaces remain strong, leading to a constant interplay between connected and differentiated spaces, characterized by a tug-of-war of equilibrium and differentiation. This dialectical process reshapes original geographical differences and imbalances, transforming urban spaces beyond their original spatial extent. Networks in mobile contexts can reinforce, terminate, or disrupt geospatial differentiation and the spatial structures it produces, altering existing socio-spatial disparities. Geospatial differences and socio-spatial unevenness can be partially expressed through the multidimensional concept of "distance", which may exist in various forms. Population flow represents intercity interaction—an artificial spatial reconfiguration of production factors. Agglomeration and dispersion of factors in city development are reflected in population flow patterns, so intercity linkages based on population flow also mirror the spatial connections of economic and social factors among regional cities. Population urbanization in China features large-scale movement from rural to urban, developed regions. The Chinese Spring Festival migration reflects this phenomenon and provides a short-period big-data window into intercity linkages amid urbanization and social change. Higher intensity of intercity linkages drives regional development and underpins dense urban areas and regional integration. In China’s push for regional integration, competition shifts towards cooperative competition, tightening socioeconomic linkages. The YRDR, known for integration and rapid growth, exemplifies China’s intercity linkage network. Since the 2018 designation of "Yangtze River Delta integration" as a national strategy, intercity cooperation and regional coordination have strengthened. This paper explores the "near" and "relevant" linkage between cities in the YRDR to inform high-quality integrated development. It examines the influence of multivariate distance factors—geographic distance and attribute homogeneity distances—on intercity linkage asymmetry. The paper proceeds with a literature review, hypotheses and analytical framework with data, results on asymmetry and determinants, and discussion and conclusion.

Spatial structural effects like spatial autocorrelation and dependence have been studied since the 1970s. Prior work shows intercity asymmetry linkages are associated with geographic distance: closer cities tend to have more symmetric linkages, consistent with distance decay and Tobler’s first law of geography—"near" and "related" as foundations for spatial autocorrelation and urban interactions. "Near" has evolved from Euclidean to combined concepts incorporating time and monetary cost distances with transportation advances. Reductions in time and cost distances alter relative city positions, foster co-location effects, and strengthen intercity linkages, especially via high-speed rail, airline, and highway expansion. Non-hub or off-corridor cities face accessibility challenges that impede linkages. The notion of "related" extends beyond geographic space to attribute spaces. Following Tobler’s principle equating similarity with distance in multidimensional scales, "related" is treated as proximity in multi-attribute space. Distance measures in socioeconomic attribute spaces include administrative divisions, city classifications, industrial structures, and economic scales. Institutional distance: Administrative boundaries can impede factor flows and intercity linkages; inter-provincial borders create market segmentation effectively increasing "distance" between cities beyond geographic separation. Hierarchical distance: City hierarchy influences spillovers—higher-hierarchical cities possess greater radiation capacity and broader hinterlands, inducing passive linkages from lower to higher tiers; linkages among lower-tier cities are weaker due to limited complementarity. Economic (scale) distance: Agglomeration economies allow larger cities to attract factors from smaller ones, promoting flows and linkages, with larger cities inducing and spreading linkages more broadly. Structural (industrial) distance: Similar structures foster cooperation, exchange, and learning, but shifts to knowledge- and technology-intensive industries alter labor demand, spurring population flows. Both similarity and disparity in industrial structures can promote linkages depending on context. Overall, factors influencing intercity asymmetric linkages can be framed as proximity in geographic and multivariate spaces. Smaller distances correspond to less pronounced asymmetry in intercity linkages, and larger distances to greater asymmetry. The conceptual illustration frames multivariate distance (geographic, time, cost, institutional, hierarchical, economic, structural) as jointly shaping intercity linkages, spatially expressed as degrees of asymmetry between core and peripheral cities.

Conceptual design and hypotheses: The linkage symmetry index measures asymmetry of intercity linkages: higher values indicate more prominent asymmetry and pronounced unidirectional ties; lower values indicate more mutual, bidirectional connections. Potential determinants of asymmetry are represented as distances between cities in geographic and multivariate attribute spaces. The dependent variable is the intercity asymmetric linkage index; independent variables are geographic distance and multivariate distances (time, cost, institutional, hierarchical, economic, structural). Expected effects are positive for all distance measures. Variables: - Geographical distance (GD): shortest Earth-surface distance between cities (km). - Time distance (TD): average travel time for trains on intercity highways and railways (minutes). - Cost distance (CD): average travel cost for trains on intercity highways and railways (yuan). - Institutional distance (ID): provincial administrative boundary effect measured by the number of straight-line connections crossing provincial boundaries. - Hierarchical distance (HD): absolute difference in administrative level (municipalities=4, provincial capitals=3, metropolitan core=2, general prefecture-level=1). - Economic distance (ED): absolute difference in GDP between city districts. - Structural distance (SD): absolute difference in the ratio of secondary to tertiary production between cities. Study area and data: The YRDR covers Shanghai and Jiangsu, Zhejiang, and Anhui provinces (approx. 358,000 km²). Shanghai is a centrally governed municipality; Nanjing, Hangzhou, and Hefei are provincial capitals. Location Based Services (LBS) data from Tencent (via apps like WeChat, QQ, etc.) provide spatiotemporal trajectories (time, longitude, latitude, usage frequency). The Spring Festival migration, one of the world’s largest short-term migrations, is well captured by such data. The "Tencent Migration" dataset includes prefecture-level-and-above cities in the YRDR, listing the top ten daily inflow and outflow cities and migration volumes, totaling 7,543 migration records. A two-way matrix is constructed to analyze intra-day population migration patterns. Network core–periphery structure analysis: Using k-shell decomposition, periphery nodes (k=1) and their edges are iteratively removed, then nodes with degree ≤K (K≥2) are recursively eliminated to identify core nodes that occupy more central positions in the population mobility network. Link symmetry index: Quantifies symmetry/asymmetry in dyadic linkages, capturing balance or imbalance in direction and strength of connections. MRQAP (Multivariate Regression Quadratic Assignment Procedure): A nonparametric approach to assess associations between matrices with complex correlations. Standard OLS is first run between dependent and independent matrices, then rows and columns of the dependent matrix are randomly permuted many times to obtain empirical significance for coefficients while accounting for network dependencies. In this study, a 41×41 matrix records the asymmetric linkage index between 41 YRDR cities; independent variables are corresponding dyadic distances. The MRQAP model is: A_ij = β0 + β1 GD_ij + β2 TD_ij + β3 CD_ij + β4 ID_ij + β5 HD_ij + β6 ED_ij + β7 SD_ij + ε_ij. Geographic distance thresholds for asymmetry: Average GDs for five asymmetry degrees (A–E) are used to define categories I–V of geographic distance to analyze how attribute distances influence asymmetry under controlled geographic ranges. Example GD ranges mapped to asymmetry degrees: 38.17–142.26 km (A), 142.27–233.50 km (B), 233.51–334.78 km (C), 334.79–458.96 km (D), 458.97–683.13 km (E).

Core–periphery and asymmetry patterns: - Core cities identified: Shanghai, Suzhou, Nanjing, Hangzhou, Hefei, Wuxi, Ningbo, Jiaxing, Changzhou. These dominate the YRDR population linkage network. - Degrees of asymmetry (A–E, low→high) vary with distance: A and B primarily occur at short geographic distances (often neighboring cities, both intra- and inter-provincial). C occurs at short distances between neighbors and at some longer distances between non-neighbors, dominated by intra-provincial links. D and E are predominantly over long distances; intra-provincial links with such asymmetry are rare, while inter-provincial links are common across core cities. MRQAP results and effect sizes: - Model fit: R-squared = 0.857, explaining 85.70% of variance; number of observations = 1,640. - Geographic distance (GD) is the most decisive factor for asymmetry: coefficient ≈ 0.961 (strong positive effect; greater GD → greater asymmetry), consistent with spatial autocorrelation and Tobler’s first law. - Additional influences (descending after GD): cost distance (CD ≈ 0.219), economic distance (ED ≈ 0.187), structural distance (SD ≈ 0.151), time distance (TD ≈ 0.149), institutional distance (ID ≈ 0.004), hierarchical distance (HD ≈ 0.003). Time and cost both positively influence asymmetry, with cost effects exceeding time in this setting of improved accessibility but rising cost differentials. - Economic distance: greater disparities in economic scale associate with more pronounced asymmetry; more symmetric (A-degree) linkages are likelier among cities with similar socioeconomic levels. - Structural distance: larger industrial structure disparities increase asymmetry; similarity fosters cooperation and frequent contact, lowering asymmetry. - Institutional and hierarchical distance effects are small on average, reflecting progress in YRDR integration and coordination. Distance-threshold-specific effects (categories I–V by GD): - Categories I–II (short GD): time and cost distances have the most pronounced influence, highlighting accessibility and the small-world effect within metropolitan proximities. - Category III (medium GD): marked increase in time and economic distance influence; cost and hierarchical effects decline. Economic distance shows a strong decay effect on symmetry (reported peak influence ≈ 0.844), suggesting spillover from large to small cities. - Category IV (long GD): time and economic influences diminish; economic distance remains most influential (≈ 0.295), followed by hierarchical (≈ 0.145), indicating the growing role of city scale and rank disparities at longer distances. - Category V (longest GD): hierarchical and institutional distances increase markedly (peaks ≈ 0.491 and ≈ 0.320, respectively), along with rising cost effects, indicating E-degree asymmetry results from combined administrative boundary barriers and hierarchy disparities. Overall, attribute-distance effects on asymmetry become more complex as geographic distance increases beyond certain thresholds, with shifting dominance across time, cost, economic, institutional, and hierarchical dimensions.

The study provides empirical evidence from the YRDR that, despite advances in transportation and ICT, geographic distance continues to play a decisive role in shaping intercity linkage asymmetry—countering simplistic notions of the "death of distance." By conceptualizing distance multidimensionally (geographic plus attribute dimensions such as time, cost, institutional, hierarchical, economic, and structural), the analysis reconciles debates by showing that proximity in multiple spaces governs intercity interactions and that attribute distances interact with geographic thresholds. Linking China’s regional integration policy discourse to global debates on urban networks, the findings underscore the continued relevance of physical spatial distance in population mobility and urban interactions, even in highly connected, infrastructure-rich regions. The framework generalizes beyond the YRDR, offering a basis to study how multivariate distances shape regional integration and urban development elsewhere. Policy implications: Effective intercity linkage development should prioritize transportation and logistics infrastructure to improve accessibility; strengthen intergovernmental coordination under unified policy frameworks; and promote industrial complementarity and cooperation (e.g., through innovation platforms and inter-firm collaboration) to mitigate constraints imposed by geographic and attribute distances and to facilitate balanced, collaborative regional growth.

Using Tencent Migration data on Spring Festival flows in the YRDR, the study analyzes intercity linkage symmetry and tests the "death of distance" hypothesis across multiple distance dimensions. An intercity linkage matrix and MRQAP framework show that geographic distance remains the most decisive determinant of asymmetry, while attribute distances (time, cost, economic, institutional, hierarchical, structural) exert positive but context-dependent effects that shift with geographic thresholds. These results challenge the idea that distance is losing importance and highlight the enduring role of spatial proximity alongside socioeconomic proximity. For practice, the findings suggest policies that enhance regional coordination, invest in transport and logistics networks, and foster industrial complementarity to reduce effective distances, promote intercity mobility, and support integrated development. The multidimensional distance framework offers an alternative perspective for understanding intercity linkages and is applicable to other regions. Future research should incorporate additional attribute dimensions—especially cultural and linguistic distance—and finer controls on demographic group attributes to better capture how different proximities influence linkage symmetry across varying geographic contexts.

The study acknowledges potential group bias in Tencent Migration data and the absence of socioeconomic attributes of mobile individuals. Cultural and linguistic differences (dialect areas) are not explicitly modeled but may affect intercity linkages and their asymmetry. Future work should control for demographic group attributes and incorporate measures of cultural distance (e.g., linguistic similarity, dialect regions) to refine understanding of how non-spatial proximities shape intercity connections.