Original Coverage & Source Attribution: bioengineer.org
A long-term macro-level panel investigation examining China and ten member countries of the Association of Southeast Asian Nations from 2000 to 2023 has established that domestic digital infrastructure correlates positively with the structural resilience capacity of agricultural supply chains, operating through both direct national channels and positive cross-border spatial spillovers. Published in Agricultural and Food Economics by researchers Jing Wang, Jiancheng Chen, Xiangzheng Deng, and Jingyu Wang, the empirical study analyzes a balanced country-level panel consisting of eleven economies and 264 country-year observations. Employing a two-way fixed-effects Spatial Durbin Model to account for systemic cross-border economic interdependence, the authors found that digital infrastructure investment exhibits a positive baseline relationship with structural agricultural supply chain resilience capacity, accompanied by a statistically significant positive spatial indirect effect transmitted across economically connected regional partners under a time-averaged economic-distance weighting matrix.
Conceptually, the authors explicitly distinguish structural agricultural supply chain resilience capacity, abbreviated as structural ASCR, from short-term shock absorption or post-disruption recovery velocity. Rather than tracking high-frequency recovery dynamics following specific disruption events, structural ASCR operationalizes the durable, long-term systemic readiness and adaptive capacity embedded within national production and distribution systems. To measure this multidimensional capacity, the research team constructed a composite index comprising five distinct analytical dimensions: production structure, logistics and circulation, resource management, demographic and human-resource conditions, and land-use and ecological conditions. The authors emphasize that while these dimensions capture foundational readiness to sustain functional continuity under persistent environmental and economic stress, the proxy reflects potential structural capacity and agricultural development configurations rather than realized recovery performance following isolated historical shocks.
To avoid subjective scoring biases across diverse national indicators, the empirical investigation utilized the objective entropy weight method to calculate composite index scores across the balanced pooled panel. Within the structural ASCR framework, ten underlying indicators were standardized using min-max normalization, with weights derived directly from sample dispersion and information entropy. The production structure dimension incorporates cereal yield as a reflection of land productivity, the share of agriculture, forestry, and fisheries in gross domestic product to reflect sectoral economic positioning, and the livestock production index. The logistics and circulation dimension integrates port container traffic as a proxy for international shipping connectivity alongside rural electricity access rates supporting cold chains and digital services. Resource management is captured through cultivated land area and the agricultural irrigation ratio, while demographic capacity is measured through rural population growth. Finally, the ecological dimension tracks national forest cover and the proportion of agricultural land, establishing a standardized baseline metric that remains directly comparable across countries and over time.
The principal explanatory variable, digital infrastructure (DigInfra), was similarly synthesized into a comprehensive composite index via the entropy weight method, drawing upon standardized telecommunications data compiled from the International Telecommunication Union and EPSDATA. In accordance with established evaluation frameworks, the DigInfra composite incorporates four foundational connectivity metrics: the percentage of individuals using the internet, fixed-broadband internet subscriptions, fixed-telephone subscriptions, and mobile-cellular subscriptions. Crucially, none of these four telecommunications indicators overlap with the variables comprising the structural ASCR composite. To reduce potential omitted-variable bias, the econometric models simultaneously adjust for seven socioeconomic control variables: permanent cropland area, airline passengers carried, personal remittances received, gross fixed capital formation, trade in transport services, gross national expenditure, and air freight volume.
Methodologically, the authors investigated spatial dependence patterns using both Global Moran’s I and Local Moran’s I diagnostics before specifying a spatial panel econometric architecture. Spatial interdependence was operationalized through an economic-distance spatial weight matrix constructed from reciprocal differences in time-averaged per capita gross domestic product across the 2000 to 2023 sample period. The researchers justified prioritizing an economic-distance formulation on the theoretical premise that economies sharing comparable developmental trajectories and living standards frequently exhibit similar consumer demand structures, regulatory frameworks, and digital technology adoption capabilities. Within the baseline two-way fixed-effects linear regression without spatial lags, digital infrastructure demonstrated a positive coefficient of 0.1005 (p < 0.01), which moderated to 0.0399 (p < 0.10) upon introducing the complete vector of socioeconomic control variables.
Upon estimating the complete two-way fixed-effects Spatial Durbin Model, digital infrastructure maintained a robust positive conditional association with structural ASCR. In the spatial specification controlling for socioeconomic covariates, the direct coefficient on domestic digital infrastructure was estimated at 0.0298 (p < 0.05), while the coefficient on the spatially lagged digital infrastructure variable across economically linked neighboring nations was 0.1463 (p < 0.01). Furthermore, the spatial autoregressive coefficient on structural resilience capacity was positive and statistically significant at 0.2337 (p < 0.01), confirming substantial spatial clustering and systemic interdependence among agricultural supply chains across the integrated China–ASEAN macroeconomic network.
The authors consequently performed partial-differential decompositions to separate the overall relationship into direct, indirect, and total spatial effects. The resulting estimates revealed a statistically significant direct effect of 0.0384 (p < 0.05), representing the net within-country association after factoring in regional spatial feedback. The indirect effect, capturing cross-border spatial spillovers transmitted from connected trading partners, was substantially larger at 0.1999 (p < 0.01), yielding a cumulative total effect of 0.2383 (p < 0.01).
Extensive heterogeneity analyses demonstrated that this positive relationship is not uniform across temporal sub-periods or distinct developmental and geographic settings. A temporal split comparing 2000–2011 against 2012–2023 revealed striking parameter divergence: while the direct, indirect, and total associations were all positive and significant during the initial twelve-year period (0.1020, 0.1361, and 0.2382, respectively; p < 0.01), the corresponding estimates shifted to negative values during the 2012–2023 period (−0.0584, −0.2617, and −0.3201; p < 0.05). In full-sample interaction specifications, the overall association was significantly weaker for China relative to ASEAN economies, evidenced by a negative total interaction coefficient of −0.2610 (p < 0.01). Furthermore, when dividing countries by median economic development, positive interaction differentials were concentrated exclusively within the spatial spillover channel (W×HighDev×DigInfra = 0.1401, p < 0.01), whereas higher median digitalization exhibited positive interaction differentials across direct, indirect, and total effect channels.
To capture institutional divergence across Southeast Asia, the researchers integrated the World Bank Worldwide Governance Indicators, segmenting ASEAN member states into high- and low-institutional quality cohorts based on long-term historical averages of Government Effectiveness and Rule of Law. In the Government Effectiveness specification, the baseline association was statistically insignificant for low-effectiveness economies, but the interaction with high effectiveness yielded a strong positive total effect differential of 0.4268 (p < 0.01). Similarly, under the Rule of Law classification, economies characterized by established legal frameworks exhibited positive, statistically significant interaction differentials across direct, indirect, and total dimensions, underscoring that institutional predictability, administrative enforcement, and transparent legal environments substantially enhance the productive integration of digital infrastructure within agricultural networks.
Moderation analyses examining digital governance (DigGov) and digital trade (DigTrade) identified contrasting structural transmission mechanisms across local and spatial domains. Digital governance—evaluated across regulatory independence, dispute resolution mechanisms, consumer protection laws, and cybersecurity mandates—exhibited uniformly positive moderation, producing a direct interaction effect of 0.0598 (p = 0.002) and an indirect interaction effect of 0.3883 (p < 0.001). Conversely, digital-trade development conditioned the local and spatial channels in opposite directions: the within-country direct interaction was negative (−0.0862, p < 0.10; with an unstandardized interaction of −0.1225, p < 0.01), whereas the cross-border spatial lag interaction was strongly positive (W×DigInfra×DigTrade = 0.6519, p < 0.01), driving a positive indirect interaction effect of 0.8009 (p < 0.01).
The baseline findings withstood a battery of six rigorous econometric robustness checks. Replacing contemporaneous digital infrastructure with one-period and two-period temporal lags maintained positive and statistically significant direct effects (0.0392 and 0.0515) and indirect effects (0.1940 and 0.2108). Substituting the economic-distance matrix with a fixed geographic-economic distance matrix combining geographic coordinates and gross domestic product retained significant direct (0.0810) and indirect (0.3621) effects. Applying sample winsorization to minimize the influence of extreme data values confirmed stability, as did excluding the disruptive COVID-19 pandemic year of 2020. Finally, recalculating the dependent variable using a Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) closeness aggregation score confirmed consistent positive direct (0.0795) and indirect (0.3081) associations, affirming that the underlying empirical patterns are robust against alternative aggregation methodologies and spatial weighting formulations.
Despite the statistical consistency of the model estimates, the authors outline several critical analytical caveats and boundaries of inference. Because the structural ASCR index measures aggregate national capacity rather than event-specific recovery trajectories, the findings should not be interpreted as direct evidence of post-shock physical recovery or crisis mitigation. Additionally, observational panel regressions and two-way fixed-effects structures can account for unobserved time-invariant heterogeneity, but they do not eliminate potential reverse causality, unobserved time-varying shocks, or establish strict causal mechanisms. The authors caution that macro-level country aggregates cannot capture micro-level farmer decisions, enterprise logistics behaviors, or subnational institutional nuances, and the benchmark time-averaged spatial weight matrix cannot capture evolving real-time bilateral trade routing. Consequently, the empirical results provide macro-level structural associations rather than prescriptive evaluations of specific isolated policy interventions.
Subject of Research: Macro-level spatial econometric analysis of digital infrastructure and agricultural supply chain resilience in China and ASEAN
Article Title: The structural resilience capacity of agricultural supply chains in China–ASEAN: the role of digital infrastructure and spatial spillovers
Article References: Wang, J., Chen, J., Deng, X., & Wang, J. (2026). The structural resilience capacity of agricultural supply chains in China–ASEAN: the role of digital infrastructure and spatial spillovers. Agricultural and Food Economics, 14(1), Article 90.
Keywords: agricultural supply chains, structural resilience capacity, digital infrastructure, spatial spillovers, Spatial Durbin Model, China-ASEAN, entropy weight method, digital governance, digital trade, economic distance matrix, two-way fixed effects, indirect effects, food security, institutional quality, regional integration, spatial econometrics, TOPSIS robustness




