Inference in matrix-valued time series with common stochastic trends and multifactor error structure

Abstract: We develop an estimation methodology for a factor model for high-dimensional matrix-valued time series, where common stochastic trends and common stationary factors can be present. We study, in particular, the estimation of (row and column) loading spaces, of the common stochastic trends and of the common stationary factors, and the row and column ranks thereof. In a set of (negative) preliminary results, we show that a projection-based technique fails to improve the rates of convergence compared to a "flattened" estimation technique which does not take into account the matrix nature of the data. Hence, we develop a three-step algorithm where: (i) we first project the data onto the orthogonal complement to the (row and column) loadings of the common stochastic trends; (ii) we subsequently use such "trend free" data to estimate the stationary common component; (iii) we remove the estimated common stationary component from the data, and re-estimate, using a projection-based estimator, the row and column common stochastic trends and their loadings. We show that this estimator succeeds in refining the rates of convergence of the initial, "flattened" estimator. As a by-product, we develop consistent eigenvalue-ratio based estimators for the number of stationary and nonstationary common factors.