Working Paper: CEPR ID: DP7447
Authors: Pablo A. Guerrón-Quintana; Atsushi Inoue; Lutz Kilian
Abstract: We show that in weakly identified models (1) the posterior mode will not be a consistent estimator of the true parameter vector, (2) the posterior distribution will not be Gaussian even asymptotically, and (3) Bayesian credible sets and frequentist confidence sets will not coincide asymptotically. This means that Bayesian DSGE estimation should not be interpreted merely as a convenient device for obtaining asymptotically valid point estimates and confidence sets from the posterior distribution. As an alternative, we develop new frequentist confidence sets for structural DSGE model parameters that remain asymptotically valid regardless of the strength of the identification.
Keywords: Bayes Factor; Bayesian Estimation; Confidence Set; DSGE Models; Identification; Inference; Likelihood Ratio
JEL Codes: C32; C52; E30; E50
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Weak Identification (C26) | Posterior Mode Consistency (C62) |
| Weak Identification (C20) | Posterior Distribution Normality (C46) |
| Weak Identification (C26) | Bayesian Credible Sets Validity (C11) |
| Likelihood Ratio Tests and Bayes Factors (C11) | Confidence Sets Validity (C12) |