Model averaging in Markov-switching models: Predicting national recessions with regional data

- We show how to perform dynamic model averaging with Markov-switching models.
- We introduce combination weights based on the models' ability to fit a discrete outcome.
- We combine forecasts from a large set of Markov-switching models to predict U.S. recessions.
- Forecasts obtained from our new combination schemes outperform competitive benchmarks.

This paper introduces new weighting schemes for model averaging when one is interested in combining discrete forecasts from competing Markov-switching models. In the empirical application, we forecast U.S. business cycle turning points with state-level employment data. We find that forecasts obtained with our best combination scheme provide timely updates of U.S. recessions in that they outperform a notoriously difficult benchmark to beat (the anxious index from the Survey of Professional Forecasters) for short-term forecasts.