ESTIMATION AND PREDICTION FOR A CLASS OF DYNAMIC NONLINEAR STATISTICAL-MODELS

A class of nonlinear state-space models, characterized by a single sou rce of randomness, is introduced. A special case, the model underpinni ng the multiplicative Holt-Winters method of forecasting, is identifie d. Maximum likelihood estimation based on exponential smoothing instea d of a Kalman filter, and with the potential to be applied in contexts involving non-Gaussian disturbances, is considered. A method for comp uting prediction intervals is proposed and evaluated on both simulated and real data.