4. - 5. Juni 2024

Graduate Center Masterclasses

Bayesian Approaches to the Analysis of Panel Data

Termin

4. - 5. Juni 2024
Day 1: 09:30 - 17:15h

Day 2: 09:30 - 12:45h

Ort

Karl Popper Room

DIW Berlin

Room 2.3.020

Mohrenstr. 58

10117 Berlin

Sprecher*innen

Frank Schorfheide

Zum Kalender hinzufügen

This course provides a self-contained introduction to Bayesian analysis of panel data models. We will start with an introduction to Bayesian inference, covering the basic building blocks, which involve inference in a heteroskedastic linear regression model and Gibbs sampling to generate draws from the posterior distribution. Moreover, we will consider nonparametric inference. We then proceed with Bayesian and empirical Bayes inference on a vector of means with particular emphasis on the determination of hyperparameters of the prior distribution. We combine the pieces to analyze linear and nonlinear panel data models. Finally, we consider the estimation of two-way effects models for matched data sets, e.g., student-teacher or firm-employee data sets. Much of the course draws from my recent research on these topics, in particular the papers listed below.

Outline:

Day 1
Lecture 1 (90min): Introduction to Bayesian Inference and Gibbs Sampling I
Lecture 2 (90min): Introduction to Bayesian Inference and Gibbs Sampling II
Lecture 3 (90min): Bayesian and empirical Bayes inference on a vector of means, extensions to panel data settings (P2, WP2).
Lecture 4 (90min): Parametric and nonparametric approaches to correlated random effects models (P2, P3)

Day 2
Lecture 5 (90min): Application: forecasting with a panel Tobit model (P1)
Lecture 6 (90min): Empirical Bayes estimation of two-way effects models (WP2)

Reading List:

(WP1) Cheng, Xu, Sheng Chao Ho, and Frank Schorfheide (2024): “Optimal Estimation of Two-Way Effects under Limited Mobility,” Working Paper.

(P1) Liu, Laura, Hyungsik Roger Moon, and Frank Schorfheide (2020): “Forecasting with Dynamic Panel Data Models,” Econometrica, 88(1), 171-201.

(P2) Liu, Laura, Hyungsik Roger Moon, and Frank Schorfheide (2023): “Forecasting with a Panel Tobit Model,” Quantitative Economics, 14(1), 117-159.

(P3) Liu, Laura (2023): “Density Forecasts in Panel Data Models: A Semiparametric Bayesian Perspective,” Journal of Business & Economic Statistics, 41(2)

(WP2) Moon, Hyungsik Roger, Frank Schorfheide, and Boyuan Zhang (2023): “Bayesian Estimation of Panel Models with Potentially Sparse Heterogeneity,” Working Paper,

About the instructor:

Frank Schorfheide is a Professor of Economics at the University of Pennsylvania and has served as Chair of the Department of Economics from July 2018 to June 2021. He is also a Research Associate at the National Bureau of Economic Research (NBER) and a Research Fellow at the Centre for Economic Policy Research (CEPR). He has served on the editorial board (as co-editor) of the International Economic Review from 2005-2009 and as co-editor of Quantitative Economics from 2011-2018. He has been a Visiting Scholar at several central banks.

His research areas are econometrics and empirical macroeconomics. Much of his work can be classified as macroeconometrics and is related to the Bayesian analysis of dynamic stochastic general equilibrium (DSGE) models. His research provides a set of tools that are useful for empirical work with modern macroeconomic models, including forecasting and policy analysis. He has applied these methods to analyze the sources of business cycle fluctuations and to study the effects of monetary policy. In recent years, he has also worked on forecasting with dynamic panel data models and, more generally, on the estimation of models with unobserved heterogeneity.