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

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 (ii) node-specific latent variables (dynamic fitnesses) describing the propensity of each node to create links. Assuming a Markov dynamics for both mechanisms, we propose an Expectation-Maximization 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 apply our methodology to the e-MID interbank network for which the two linkage mechanisms are associated with two different trading behaviors in the process of network formation, namely preferential trading and trading driven by node-specific characteristics. The empirical results allow to recognise preferential lending in the interbank market and indicate how a method that does not account for time-varying network topologies tends to overestimate preferential linkage.

[1]  Trevor Hastie,et al.  The Elements of Statistical Learning , 2001 .

[2]  Kevin S. Xu.  Stochastic Block Transition Models for Dynamic Networks , 2014, AISTATS.

[3]  Pavel N Krivitsky,et al.  A separable model for dynamic networks , 2010, Journal of the Royal Statistical Society. Series B, Statistical methodology.

[4]  Ruey S. Tsay,et al.  Analysis of Financial Time Series , 2005 .

[5]  João F. Cocco,et al.  Lending Relationships in the Interbank Market , 2003 .

[6]  Gen Li,et al.  Varying-coefficient models for dynamic networks , 2017, Comput. Stat. Data Anal..

[7]  Guido Caldarelli,et al.  Reconstructing topological properties of complex networks from partial information using the Fitness Model , 2014 .

[8]  Giulio Cimini,et al.  Systemic Risk Analysis on Reconstructed Economic and Financial Networks , 2014, Scientific Reports.

[9]  A. Moore,et al.  Dynamic social network analysis using latent space models , 2005, SKDD.

[10]  Falk Bräuning,et al.  The Dynamic Factor Network Model with an Application to Global Credit Risk , 2016 .

[11]  Francesco Bartolucci,et al.  Dealing with reciprocity in dynamic stochastic block models , 2018, Comput. Stat. Data Anal..

[12]  G. Weisbuch,et al.  Market Organisation and Trading Relationships , 2000 .

[13]  Thomas Lux,et al.  Network analysis of the e-MID overnight money market: the informational value of different aggregation levels for intrinsic dynamic processes , 2013, Comput. Manag. Sci..

[14]  Alfred O. Hero,et al.  Dynamic Stochastic Blockmodels for Time-Evolving Social Networks , 2014, IEEE Journal of Selected Topics in Signal Processing.

[15]  E. Xing,et al.  Discrete Temporal Models of Social Networks , 2006, SNA@ICML.

[16]  Thomas Lux,et al.  Network formation in the interbank money market: An application of the actor-oriented model , 2017, Soc. Networks.

[17]  P. Holland,et al.  An Exponential Family of Probability Distributions for Directed Graphs , 1981 .

[18]  Fabrizio Lillo,et al.  Detectability thresholds in networks with dynamic link and community structure , 2017, ArXiv.

[19]  Fabrizio Lillo,et al.  The organization of the interbank network and how ECB unconventional measures affected the e-MID overnight market , 2015, Comput. Manag. Sci..

[20]  Yihong Gong,et al.  Detecting communities and their evolutions in dynamic social networks—a Bayesian approach , 2011, Machine Learning.

[21]  Vincent Miele,et al.  Statistical clustering of temporal networks through a dynamic stochastic block model , 2015, 1506.07464.

[22]  M. Newman,et al.  Random graphs with arbitrary degree distributions and their applications. , 2000, Physical review. E, Statistical, nonlinear, and soft matter physics.

[23]  Yuguo Chen,et al.  Latent Space Models for Dynamic Networks , 2015, 2005.08808.

[24]  M. A. Muñoz,et al.  Scale-free networks from varying vertex intrinsic fitness. , 2002, Physical review letters.

[25]  Peter A. W. Lewis,et al.  Discrete time series generated by mixtures III: Autoregressive processes (DAR(p)) , 1978 .

[26]  Fabrizio Lillo,et al.  Effects of memory on spreading processes in non-Markovian temporal networks , 2018, New Journal of Physics.

[27]  Stephen E. Fienberg,et al.  Maximum lilkelihood estimation in the $\beta$-model , 2011, 1105.6145.

[28]  Daniele Durante,et al.  Bayesian dynamic financial networks with time-varying predictors , 2014, 1403.2272.

[29]  James G. Scott,et al.  Bayesian Inference for Logistic Models Using Pólya–Gamma Latent Variables , 2012, 1205.0310.

[30]  M. Newman,et al.  Statistical mechanics of networks. , 2004, Physical review. E, Statistical, nonlinear, and soft matter physics.

[31]  Roma,et al.  Fitness model for the Italian interbank money market. , 2006, Physical review. E, Statistical, nonlinear, and soft matter physics.

[32]  Jari Saramäki,et al.  Temporal Networks , 2011, Encyclopedia of Social Network Analysis and Mining.

[33]  Fabrizio Lillo,et al.  Methods for Reconstructing Interbank Networks from Limited Information: A Comparison , 2017 .

[34]  Michael Bacharach,et al.  Estimating Nonnegative Matrices from Marginal Data , 1965 .

[35]  G. Kapetanios,et al.  Estimating the Dynamics and Persistence of Financial Networks, with an Application to the Sterling Money Market , 2016 .

[36]  Diego Garlaschelli,et al.  Fitness-dependent topological properties of the world trade web. , 2004, Physical review letters.

[37]  Mathias Staudigl,et al.  Evolution of Social Networks , 2012 .

[38]  Stéphane Gaïffas,et al.  Link prediction in graphs with autoregressive features , 2012, J. Mach. Learn. Res..

[39]  Marco Di Maggio,et al.  The Value of Trading Relationships in Turbulent Times , 2017 .

[40]  Fabrizio Lillo,et al.  Disentangling group and link persistence in dynamic stochastic block models , 2017, Journal of Statistical Mechanics: Theory and Experiment.

[41]  Eric R. Ziegel,et al.  The Elements of Statistical Learning , 2003, Technometrics.

[42]  T. Yan,et al.  A note on asymptotic distributions in directed exponential random graph models with bi-degree sequences , 2017 .

[43]  Peter A. W. Lewis,et al.  Discrete Time Series Generated by Mixtures Ii: Asymptotic Properties , 1978 .

[44]  G. Caldarelli,et al.  A Network Analysis of the Italian Overnight Money Market , 2005 .

[45]  Adrian E Raftery,et al.  Interlocking directorates in Irish companies using a latent space model for bipartite networks , 2016, Proceedings of the National Academy of Sciences.

[46]  Martin Rosvall,et al.  Modelling sequences and temporal networks with dynamic community structures , 2015, Nature Communications.

[47]  Peter D. Hoff,et al.  Latent Space Approaches to Social Network Analysis , 2002 .

[48]  M. Tumminello,et al.  Networked Relationships in the e-MID Interbank Market: A Trading Model with Memory , 2014, 1403.3638.

[49]  Guido Caldarelli,et al.  Bootstrapping Topological Properties and Systemic Risk of Complex Networks Using the Fitness Model , 2012, Journal of Statistical Physics.

[50]  Arne Gabrielsen.  Consistency and identifiability , 1978 .

[51]  Zoubin Ghahramani,et al.  Dynamic Probabilistic Models for Latent Feature Propagation in Social Networks , 2013, ICML.

[52]  Mathias Staudigl,et al.  Co-evolutionary dynamics and Bayesian interaction games , 2013, Int. J. Game Theory.

[53]  James G. Scott,et al.  Sampling P olya-Gamma random variates: alternate and approximate techniques , 2014, 1405.0506.

[54]  Fragkiskos Papadopoulos,et al.  Link persistence and conditional distances in multiplex networks , 2018, Physical review. E.

[55]  Daniele Durante,et al.  Locally Adaptive Dynamic Networks , 2015, 1505.05668.

[56]  T. Yan,et al.  A central limit theorem in the β-model for undirected random graphs with a diverging number of vertices , 2012, 1202.3307.

[57]  Cristopher Moore,et al.  Detectability thresholds and optimal algorithms for community structure in dynamic networks , 2015, ArXiv.

[58]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[59]  M. Tumminello,et al.  Quantifying preferential trading in the e-MID interbank market , 2013 .

[60]  Kevin Lee,et al.  A review of dynamic network models with latent variables. , 2017, Statistics surveys.

[61]  D. Rubin,et al.  On Jointly Estimating Parameters and Missing Data by Maximizing the Complete-Data Likelihood , 1983 .

[62]  Xiao Zhang,et al.  Random graph models for dynamic networks , 2016, The European Physical Journal B.

[63]  Allan Sly,et al.  Random graphs with a given degree sequence , 2010, 1005.1136.

[64]  J. Neyman,et al.  Consistent Estimates Based on Partially Consistent Observations , 1948 .