A1492
Title: Model averaging of multi-layer time-varying network vector autoregressions
Authors: Degui Li - University of Macau (China) [presenting]
Abstract: A time-varying multi-layer network vector autoregression (VAR) model framework is introduced for large-scale time series, allowing agents in dynamic systems to interact through multiple channels and incorporating multiple adjacency matrices to capture network spillover effects. A penalized model averaging method is proposed to determine a time-varying optimal combination of multi-layer network VAR candidate models whose number is allowed to be divergent. Under regularity conditions, asymptotic properties such as asymptotic optimality and convergence rates of the proposed time-varying weight estimation are derived in the contexts of both in-sample fitting and out-of-sample prediction. Additionally, conformal prediction is introduced to construct interval forecasts. Monte-Carlo simulation and empirical studies illustrate reliable finite-sample estimation and predictive performance of the developed methodology.
