I want to use VAR in levels, in case some series I(1) and some I(0)
1. Most papers first examine cointegration by Johansen with all I(1) variables. what if some variables are I(0) and I(1)? ARDL approach seems suitable but you have to determine dependent and independent variables, it does not show how many cointegration relations exist.
2.Another approach in Bayesian VAR, in Eviews impulse response of Bayesian VAR does not have confidence intervals, is there any theory behind it?
3. Which software do you suggest for estimating Bayesian VAR? I need impulse response with error bands?
Please refer to the article:
"Error Bands for Impulse Responses" by Christopher A. Sims and Tao Zha
at the website of
http://sims.princeton.edu/yftp/imperrs/m4Web.pdf
Good Luck.
This Matlab user-created toolbox might be useful to estimate Bayesian VAR.
https://www.mathworks.com/matlabcentral/fileexchange/32601-toolkit-on-econometrics-and-economics-teaching
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation