Markov-Switching DSGE Modeling in RISE

Many important episodes in modern macroeconomics are defined by temporary shifts between different economic conditions: monetary policy may switch between dovish and hawkish stances, external shocks between high and low volatility, financial markets between periods of tight and loose frictions, and so on. Standard linear DSGE models cannot accommodate such shifts in behavior. A natural extension is multiple-regime models, in which an economy at any given time is in one of several regimes and selected parameters take different values in each regime. One popular way to model transitions between regimes is via a finite-state Markov process.This framework captures recurrent episodes parsimoniously while preserving the structural discipline of DSGE modeling. The main challenge for researchers is computational: a Markov-switching rational expectations model is considerably more complex to solve and estimate than its standard single-regime counterpart. Expectations must be treated consistently across regimes, and econometric inference requires specialized f ilters, which estimate both the probability of the economy being in each regime and the values of unobserved (latent) variables, such as the output gap. The RISE toolbox for MATLAB is designed to make this workflow straightforward. It allows users to declare Markov chains and regime-specific parameters, solve switching models by perturbation methods, and estimate them using dedicated switching filters. The outputs—regime probabilities (updated and smoothed), latent variables, and regime-dependent impulse responses—are precisely what applied macroeconomists need for empirical work.