Semi-Markov Dynamic Stochastic General Equilibrium Models:Theory, Solution, and Estimation

This paper develops a general framework for Dynamic Stochastic General Equilibrium (DSGE) models where structural parameters switch regimes according to a semi-Markov process---the Semi-Markov DSGE (SM-DSGE) model. Unlike standard Markov-switching (MS-DSGE) specifications, which assume constant, memoryless exit probabilities, our framework lets transition probabilities depend on how long the economy has remained in the current regime. This feature captures the duration dependence widely documented in monetary policy and business cycle data. Our main contributions can be summarized as a complete econometric toolkit---solution, filtering, stability verification, and Bayesian inference---for DSGE models with semi-Markov regime switching. The specific contributions are as follows. First, we embed the semi-Markov process into a DSGE model by constructing an augmented state space that tracks both the current regime and elapsed duration, and we derive first-order perturbation solutions that depend on duration. Second, we develop a Hamilton-Kalman filter that integrates over the augmented discrete state while applying the Kalman filter to the continuous state, producing an exact likelihood evaluator whose cost is linear in the sample size. Third, we establish conditions for mean-square stability of the SM-DSGE solution, extending the determinacy results of Davig and Leeper (2007) and Farmer, Waggoner and Zha (2009) to the semi-Markov setting. Fourth, we estimate the model using U.S. quarterly data on output growth, inflation, and the federal funds rate (1954 Q3--2007 Q4). The estimation uncovers two monetary policy regimes: a hawkish regime (Regime 1) with an inflation response coefficient \(\phi_{\pi} = 1.50\) [90\% CI: 1.38--1.61] and an accommodative regime (Regime 2) with \(\phi_{\pi} = 0.62\) [0.29--1.01]. The Weibull shape parameters for both regimes are estimated at \(\hat{\kappa} \approx 1.11\), indicating mildly increasing hazard rates and thus positive duration dependence: the longer a regime has persisted, the more likely it is to end. Duration-dependent impulse responses show that the real effects of a monetary policy shock vary substantially with regime tenure. These findings highlight the empirical relevance of moving beyond the geometric-duration assumption embedded in conventional Markov-switching models.

Authors

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-10313432/v1
Primary Topic
Monetary Policy and Economic Impact
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Semi-Markov Dynamic Stochastic General Equilibrium Models:Theory, Solution, and Estimation

Jian Ming Shi
Research Square
Monetary Policy and Economic Impact
preprint

Semi-Markov Dynamic Stochastic General Equilibrium Models:Theory, Solution, and Estimation

Jian Ming Shi
preprint en

Abstract

This paper develops a general framework for Dynamic Stochastic General Equilibrium (DSGE) models where structural parameters switch regimes according to a semi-Markov process---the Semi-Markov DSGE (SM-DSGE) model. Unlike standard Markov-switching (MS-DSGE) specifications, which assume constant, memoryless exit probabilities, our framework lets transition probabilities depend on how long the economy has remained in the current regime. This feature captures the duration dependence widely documented in monetary policy and business cycle data. Our main contributions can be summarized as a complete econometric toolkit---solution, filtering, stability verification, and Bayesian inference---for DSGE models with semi-Markov regime switching. The specific contributions are as follows. First, we embed the semi-Markov process into a DSGE model by constructing an augmented state space that tracks both the current regime and elapsed duration, and we derive first-order perturbation solutions that depend on duration. Second, we develop a Hamilton-Kalman filter that integrates over the augmented discrete state while applying the Kalman filter to the continuous state, producing an exact likelihood evaluator whose cost is linear in the sample size. Third, we establish conditions for mean-square stability of the SM-DSGE solution, extending the determinacy results of Davig and Leeper (2007) and Farmer, Waggoner and Zha (2009) to the semi-Markov setting. Fourth, we estimate the model using U.S. quarterly data on output growth, inflation, and the federal funds rate (1954 Q3--2007 Q4). The estimation uncovers two monetary policy regimes: a hawkish regime (Regime 1) with an inflation response coefficient \(\phi_{\pi} = 1.50\) [90\% CI: 1.38--1.61] and an accommodative regime (Regime 2) with \(\phi_{\pi} = 0.62\) [0.29--1.01]. The Weibull shape parameters for both regimes are estimated at \(\hat{\kappa} \approx 1.11\), indicating mildly increasing hazard rates and thus positive duration dependence: the longer a regime has persisted, the more likely it is to end. Duration-dependent impulse responses show that the real effects of a monetary policy shock vary substantially with regime tenure. These findings highlight the empirical relevance of moving beyond the geometric-duration assumption embedded in conventional Markov-switching models.

Research Square
Monetary Policy and Economic Impact
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon