Abstract 1 We investigate in this paper how to estimate the density function of a random variable using a parametric ARMA model for its characteristic function. The choice of this model is motivated by the fact that this type of density characterizes the duration of staying at an N-states Markov chain, but the approach is general enough to be applied to many practical problems. Both ML and moment-based linear estimates are derived, the former being based on the optimization of a highly non-linear function. 1.

Peer Reviewed