Short-courses and Tutorials


Bayesian Econometrics


Department of Economics, University of Pretoria,
South Africa, December 5th to 9th 2011/h4]

Course syllabus

Detailed description of lectures and lab sessions


Lecture notes

Lecture 1: Introduction to Bayesian thinking

Lecture 2: Monte Carlo methods

Lecture 3: Hierarchical models

Lecture 4: Dynamic Models

Lecture 5, part I: Sequential Monte Carlo methods for state learning (pure filter)

Lecture 5, part II: Sequential Monte Carlo methods for state and parameter learning


R code

Lab session 1: Bayesian linear regression (logwages-yearseducation.txt)

Lab session 2: Student’s t regression

Lab session 2: Logistic regression

Lab session 2: Heteroskedastic regression

Lab session 3: First order dynamic linear model

Lab session 3: Nonlinear dynamic model

Lab session 3: Stochastic volatility model

Lab session 4: AR(1) plus noise – pure filter

Lab session 4: AR(1) plus noise – pure filter (subroutines)

Lab session 4: AR(1) plus noise – parameter learning

Lab session 4: AR(1) plus noise – parameter learning (subroutines)


More R code

Piecewise linear regression (SA household health & education expenditures)
(Results)

DLM (SA stock market index (prices))
(Results)

SV-AR(1) model (SA stock market index (returns))
(Results 1)
(Results 2)
(Results 3)


Coded in the lab sessions

Bayesian linear regression via Gibbs sampler

Kalman filter and kalman smoother (SA CPI)

Mixture of two normals

AR(1) plus noise via BF and APF


Other useful links

MCMC: Stochastic Simulation for Bayesian Inference

Markov Chain Monte Carlo Preprint

Sequential Monte Carlo Methods Page

The R project

WinBUGS

R2WinBUGS