Frank Schorfheide

Overview

Frank Schorfheide is a macroeconometrician at the University of Pennsylvania (Department of Economics). His research focuses on Bayesian estimation of DSGE models, VAR specification, and econometric methods for forecasting. He is particularly known for the Del Negro-Schorfheide DSGE-VAR framework that uses a DSGE model to construct prior distributions for reduced-form VARs.

Key Contributions / Features

• Del Negro and Schorfheide (2004): DSGE-VAR methodology — uses equilibrium conditions from a DSGE model to construct a prior distribution for a VAR, parameterized by $\lambda$ (weight on DSGE-implied restrictions); $\lambda \to \infty$ recovers the DSGE, $\lambda \to 0$ recovers an unrestricted VAR.
• Schorfheide (2005): VAR forecasting under misspecification; local misspecification framework (DGP drifts toward VAR(p*) at rate $T^{-1/2}$); MLE vs. LFE predictor comparison; Prediction Criterion (PC) for joint lag/estimator selection.

Related

• Vector Autoregression
• Lag-Order Selection Criteria
• Schorfheide (2005)