Limited Information Bayesian Model Averaging for Dynamic Panels with Short Time Periods
IMF Working Papers, April 1, 2009
Source details
- Canonical URL
- Limited Information Bayesian Model Averaging for Dynamic Panels with Short Time Periods
Other formats
Bibliographic details
- Authors: Alin T Mirestean, Charalambos G Tsangarides, Huigang Chen
- Published: April 1, 2009
- Series: IMF Working Papers
Summary and purpose
- Extends the Bayesian Model Averaging (BMA) framework to panel data models that include the lagged dependent variable and endogenous regressors.
- Proposes a Limited Information Bayesian Model Averaging (LIBMA) methodology intended for dynamic panel data models with short time periods under model uncertainty.
- Tests LIBMA using simulated data to evaluate performance in model selection and parameter estimation.
Methodology
- LIBMA: a limited-information extension of BMA tailored to dynamic panels with endogenous variables and lagged dependent variables as regressors.
- Approach targets inference when time dimension is short and model uncertainty is present.
Simulation results and findings
- Asymptotic performance: simulation results suggest LIBMA performs well both in Bayesian model selection and in averaging.
- Data-generating process recovery: LIBMA "recovers the data generating process very well, with high posterior inclusion probabilities for all the relevant regressors, and parameter estimates very close to the true values."
- Implication: methodology is well suited for inference in dynamic panel data models with short time periods and endogenous regressors under model uncertainty.
Subjects and keywords
- Subject: Bayesian models, Data processing, Estimation techniques
- Keywords: mover accent, WP
Alin T Mirestean, Charalambos G Tsangarides, and Huigang Chen. "Limited Information Bayesian Model Averaging for Dynamic Panels with Short Time Periods", IMF Working Papers 2009, 074 (2009), accessed 9/17/2026