Parametric estimation of hidden Markov models by least squares type estimation and deconvolution

This paper develops a simple and computationally efficient parametric approach to the estimation of general hidden Markov models (HMMs). For non-Gaussian HMMs, computation maximum likelihood estimator (MLE) involves high-dimensional integral that has no analytical solution can be difficult accurately. We develop new alternative method based on theory estimating functions deconvolution strategy. Our procedure requires same assumptions as MLE estimators. provide theoretical guarantees about performance resulting estimator; its consistency asymptotic normality are established. leads construction confidence intervals. Monte Carlo experiments investigated compared with MLE. Finally, we illustrate our using real data for ex-ante interest rate forecasts.