A multivariate dynamic logit model for categorical panel data with latent Markov subject-specific parameters

For the analysis of multivariate categorical longitudinal data, we propose a model based on a marginal parametrization of the conditional distribution of each vector of response variables given the covariates, the lagged response variables, and a set of subject-specific random effects for the unobserved heterogeneity. The
latter ones are assumed to follow a first-order Markov chain. For performing maximum likelihood estimation we outline an EM algorithm. A Viterbi algorithm is described for estimation of the most likely sequence of latent states. The data analysis approach based on the proposed model is illustrated by an application to a dataset which
derives from the Panel Study on Income Dynamics and concerns fertility and female participation to the labor market.

Joint work with F. Bartolucci