A dynamic network model with persistent links and node-specific latent variables, with an application to the interbank market

Abstract

We propose a dynamic network model where two mechanisms control the probability of a link between two nodes: (i) the existence or absence of this link in the past, and
\n(ii) node-specific latent variables (dynamic fitnesses) describing the propensity of each node to
\ncreate links. Assuming a Markov dynamics for both mechanisms, we propose an ExpectationMaximization algorithm for model estimation and inference of the latent variables. The estimated parameters and fitnesses can be used to forecast the presence of a link in the future. We
\napply our methodology to the e-MID interbank network for which the two linkage mechanisms
\nare associated with two different trading behaviors in the process of network formation, namely
\npreferential trading and trading driven by node-specific characteristics. The empirical results
\nallow to recognize preferential lending in the interbank market and indicate how a method that
\ndoes not account for time-varying network topologies tends to overestimate preferential linkage.