Random and Markov switching exponential smoothing models

In this paper we report results from Bayesian analysis of random switching exponential smoothing models. The new methods are robust and easy to implement. In a Monte Carlo setting it is shown that the results are particularly encouraging and the methods perform well with real data sets. Moreover, we extend the basic model under a Markov chain assumption on the slope of the stochastic trend, and we provide tools for model comparison and model selection in terms of out-of-sample behavior. The models are applied to a number of U.S. time series.

Automated summary

This paper presents results from Bayesian analysis of random switching exponential smoothing models, showing that the new methods are robust and easy to implement. Monte Carlo simulations and real data sets demonstrate the methods' strong performance, especially when extended with a Markov chain assumption on the slope of the trend. Model comparison and selection tools are also provided for out-of-sample behavior analysis.