|
| |
| Ying Lun Cheung |
| |
| ''Identification of matrix-valued factor models'' |
| ( 2024, Vol. 44 No.2 ) |
| |
| |
| The analysis of matrix-valued time series has been popular in recent years. When the dimensions of the matrix observations are large, one can use the matrix-valued factor model to extract information from the data. However, as in standard factor analysis, the common factors and factor loadings are not separately identifiable. This note considers two sets of identification restrictions that help exactly identify the model. |
| |
| |
| Keywords: Approximate factor models, Matrix-valued time series, Principal components, 2DSVD |
JEL: C3 - Cross-Sectional Models; Spatial Models; Treatment Effect Models; Quantile Regressions
G1 - General Financial Markets |
| |
| Manuscript Received : Sep 28 2023 | | Manuscript Accepted : Jun 30 2024 |
|
