A Counting Processes-based Model for the Analysis of the International Combat Aircraft Trade Network from 1950 to 2017
Cornelius Fritz, Paul W. Thurner, Goeran KauermannWe propose a novel tie-oriented model for the analysis of longitudinal event network data. The generating mechanism is assumed to be a multivariate Poisson process, that governs the onset and repetition of events with two separate intensity functions. The onset of an event indicates the start of a relation between two actors in the network, whereas repetition is defined as a continuation of the first occurrence within year-wise snapshots. Per assumption, the covariate effects differ for the onset and repetition of an event relation. As an application case, we apply the introduced method to a network consisting of the transactions of international combat aircraft trades between 1950 and 2017. The complexity, as well as the long time frame of observed networks make time-varying effects reasonable. Additionally, we include random effects on the country level to control for repeated measurements and uncaptured heterogeneity. The findings reveal strongly differing and time-varying effects of endogenous and exogenous covariates on the onset and repetition of aircraft trade events. Especially during the end of the cold war period, the time-varying network effects detect a fundamental structural change in the behavior of combat aircraft transactions.