A0196
Title: Fixed-T estimation of matrix-valued factor models
Authors: Ying Lun Cheung - Capital University of Economics and Business (China) [presenting]
Abstract: Many data sets are best represented as time series of matrices, yet econometric models tailored for this data structure remain scant. The matrix-valued factor model (MVFM) is one of the few such models. Most existing methods operate under the ``large $N$, large $T$'' context as a natural extension of the high-dimensional factor model. However, many matrix-variate data sets have either a short time span or a low frequency. The estimation of the MVFM under the ``large $N$, fixed $T$'' setting is considered. It is shown that the 2DSVD procedure continues to work. The consistency and asymptotic normality of the estimator is proved. The estimator's performance is evaluated through simulations and applications with real data.