Examining the Association Between Network Properties and Departure Delay Duration in Japan’s Domestic Aviation

Delays are a global issue affecting both airports and airlines. Departure delays are particularly likely to propagate across airports, rendering the entire flight network susceptible to increased delay durations. The delay network and its duration fluctuate daily or even hourly across airports. This study investigates the association between departure delay duration and delay network properties. Using various network metrics, we apply a fixed-effect Prais–Winsten regression model within a panel data framework covering the period from 2018 to 2021 for two full-service carriers in Japan. The key findings reveal that higher in-degree centrality is associated with longer departure delays. Betweenness centrality suggests that, in addition to hub airports, some spoke airports may function as delay bridges, thereby increasing delay durations. Eigenvector centrality is linked to shorter but more frequent departure delays across the network, which are more likely to result in frequent delay propagations of shorter durations. The results indicate that some airports may form delay clusters among themselves, potentially extending departure delay durations among connected airports. During the COVID-19 pandemic, the state of emergency contributed to varying associations between network properties and departure delay durations. These outcomes are expected to provide valuable insights for airline delay and schedule management policymakers.