Bayesian_TVPVAR:TVP-VAR模型的贝叶斯估计

# Bayesian_TVPVAR Bayesian Estimation of a TVP-VAR Model This repo contains information on how to conduct Bayesian analysis using a TVP-VAR model. Before diving into the code, you should look is the Bayes_TVPVAR_Presentation file. This will give you a baseline understanding of How the TVP-VAR differs from the normal VAR model and how we can cunduct Bayesian analysis in the TVP-VAR setting. Once you have a decent grasp on this, the code should be a little easier to follow. Before moving on to the code, I want to cite the authors that the code was obtained from. The main source was from "Bayesian Multivariate Time Series Methods for Empirical Macroeconomics" (Koop and Korobilis, 2010) and only slight modifications were made to this. This was the main textbook that most of the information for this project was obatined from. There are many different files and functions contained in the file TVP_VAR_CK, so to avoid being overwhelmed by the MATLAB code, restrict your focus to three main files. One, the Homo_TVP_VAR.m file is set up to read in data from Korobilis (2008) and reproduce the impulse response functions from his paper. This will also do the bulk of our homoskedastic TVP-VAR Bayesian analysis, which is the section that we will focus on. Lines 260-263 is where the posteriors of our parameters show up in this file. Two, the ts_prior.m function (where the ts stands for training sample, not time-series) sets the priors on our variables for our training sample. Three and finally, the carter_kohn_hom.m function runs the Kalman filter forwards so we can obtain the draws for our parameters via backwards recurssion.