Airport relationships are complex, influenced by factors like geographic proximity, air traffic patterns, ownership, partnerships, and regulations. Spatial dependence highlights mutual influences between geographically close airports, leading to competition or cooperation. Spatial heterogeneity accounts for variations in airport characteristics (location, size, services), impacting their relationships. Existing studies using spatial econometric models often lack detailed examination of how spatial dependence and heterogeneity influence these relationships, specifically whether interactions vary geographically or across airport hierarchies. This study addresses these gaps by investigating: 1) How spatial dependence changes with geographic distance and its effect on airport relationships; 2) Indicators reflecting spatial heterogeneity among airports; and 3) How spatial heterogeneity affects airport relationships. The study uses a spatial econometric model to quantify competition/complementarity, analyzes relationship dynamics with distance, identifies spatial heterogeneity indicators using an entropy-weighted TOPSIS method to classify airports hierarchically, and examines relationship variations across hierarchies. The empirical analysis uses data from 34 major Chinese airports over 13 years, chosen for their diverse and complex interactions, reflecting geographic proximity, traffic flow, ownership, collaborations, and regulatory environments. The findings aim to provide insights and a theoretical foundation for effective airport policy formulation.

Extensive research explores inter-airport relationships, examining competition, complementarity, market structure, and regional development. Methodological approaches include quantitative models analyzing passenger/airline preferences (e.g., conditional logit models), and multi-layer models examining competition and cooperation effects based on distance. Studies also link airport relationships to efficiency, using indices like the Herfindahl-Hirschman Index (HHI) to assess the impact of hub-spoke networks on transportation efficiency. Spatial econometric models are increasingly used to model airport relationships, accounting for spatial dependence and heterogeneity, using various spatial weight matrices. However, existing literature lacks a comprehensive understanding of the mechanisms through which spatial dependence and heterogeneity influence competition or complementarity between airports, particularly concerning how distance changes or airport attribute changes impact these relationships. This study aims to fill this gap by focusing on impact mechanisms and dynamic changes of spatial dependence and heterogeneity on airport relationships.

This study uses a spatial panel data model to analyze the relationships between 34 major Chinese airports from 2007-2019. The model incorporates spatial lag of dependent variables (passenger traffic), non-spatial covariates (GDP per capita, aviation employment, international tourism receipts, crude petroleum price), spatially lagged independent variables, unit-specific fixed effects, and spatially autocorrelated error terms. Three spatial interaction effects are included: spatial autoregressive (SAR), spatial error (SEM), and spatial Durbin (SDM). The model is estimated using quasi-maximum likelihood (QML) estimation to address endogeneity issues arising from the spatially lagged dependent variable. Two spatial weight matrices are employed: an inverse distance matrix (Wg) and an asymmetric economic distance matrix (We). Wg considers only geographic distance, while We incorporates the difference in average passenger traffic between airports, reflecting economic distance and weighting airports with higher passenger traffic more heavily. The economic distance matrix is asymmetric, reflecting the directional influence between airport pairs. The study uses Moran's I to analyze spatial autocorrelation before applying the spatial econometric model. Airports are classified into three hierarchical levels (large hub, regional trunk, and local branch airports) using an entropy-weighted TOPSIS method based on a comprehensive strength index system incorporating multiple indicators. The spatial relationships between different airport groups are then analyzed using the SDM model.

Moran's I index reveals significant positive spatial correlation in airport passenger traffic, stronger when using the economic distance matrix than the geographic distance matrix. The SDM model reveals a significant positive spatial autoregressive coefficient (δ), indicating a mutually driven aggregation effect in passenger traffic. Analysis of spatial relationships at different distances shows a shift from complementarity to competition as geographic distance decreases, with the turning point around 1400 km. When considering the asymmetric economic distance matrix, the spatial autoregressive coefficients are higher, reflecting the impact of spatial heterogeneity. Classification of airports into three hierarchies (hub, trunk, branch) reveals significant competition among hub airports (negative spatial autoregressive coefficient), and complementarity between trunk and branch airports. The absence of hub airports weakens the complementary relationships. Analysis of spatial relationships between different airport groups (hub-trunk, hub-branch, trunk-branch) further confirms these findings, showing stronger complementarity between hub-branch and trunk-branch airports than between hub-trunk airports, suggesting some level of competition between hub and trunk airports. Robustness tests using an inverse distance spatial weight matrix confirm the findings.

The findings address the research questions by demonstrating the significant impact of both spatial dependence and heterogeneity on airport relationships. The shift from complementarity to competition with decreasing distance highlights the influence of geographic proximity and the potential substitutability of airports, particularly with the expansion of high-speed rail. The stronger complementarity observed using the economic distance matrix emphasizes the role of spatial heterogeneity, such as differences in airport size and economic factors, in shaping relationships. The hierarchical analysis clarifies that while overall complementarity exists, competition is pronounced among hub airports, reflecting their role in international flight transfers. The importance of hub airports is further emphasized by the weaker complementarity observed in the absence of hubs. These findings are relevant to airport planning and policy, emphasizing the need to consider both competition and complementarity in infrastructure investment and development.

This study contributes to the understanding of airport relationships by demonstrating the interplay of spatial dependence and heterogeneity. It shows a mutually driven aggregation effect in passenger traffic, a distance-dependent shift from complementarity to competition, and the crucial role of hub airports in maintaining network efficiency and collaboration. Future research should extend this analysis to other countries, consider a broader range of airports, and explore additional analytical techniques to further refine the understanding of complex airport relationships.

The study's findings may not be generalizable beyond the context of Chinese airports due to their unique spatial, economic, and infrastructural characteristics. The focus on 34 major airports may not fully represent the diversity of all airports in China, potentially overlooking smaller or more remote airports. The spatial econometric model may not capture all unobservable factors influencing airport relationships.