Munich Personal RePEc Archive

A grouped factor model

Chen, Pu (2010): A grouped factor model.

WarningThere is a more recent version of this item available.

Download (362kB) | Preview


In this paper we present a grouped factor model that is designed to explore clustering structures in large factor models. We develop a procedure that will endogenously assign variables to groups, determine the number of groups, and estimate common factors for each group. The grouped factor model provides not only an alternative way to factor rotations in discovering orthogonal and non-orthogonal clusterings in a factor space. It offers also an effective method to explore more general clustering structures in a factor space which are invisible by factor rotations: the factor space can consist of subspaces of various dimensions that may be disjunct, orthogonal, or intersected with any angels. Hence a grouped factor model may provide a more detailed insight into data and thus also more understandable and interpretable factors.

Available Versions of this Item

Logo of the University Library LMU Munich
MPRA is a RePEc service hosted by
the University Library LMU Munich in Germany.