We introduce multivariate state space models for estimating and forecasting fertility rates that are dynamic alternatives to logistic representations for fixed time points. Strategies are provided for the Kalman filter and for quasi-Newton algorithm initialization, that assure the convergence of...