Identification, estimation and testing of conditionally heteroskedastic factor models

We investigate the effects of dynamic heteroskedasticity on statistical factor analysis. We show that identification problems are alleviated when variation in factor variances is accounted for. Our results apply to dynamic APT models and other structural models. We also find that traditional ML estimation of unconditional variance parameters remains consistent if the factor loadings are identified from the unconditional distribution, but their standard errors must be robustified. We develop a simple preliminary LM test for ARCH effects in the common factors, and discuss two-step consistent estimation of the conditional variance parameters. Finally, we conduct a detailed simulation exercise.