Structural VAR analysis with cointegration
Posted: Wed Jan 23, 2013 11:06 am
Hello everyone!
I'm doing structural VAR analysis for three countries: Canada, Mexico, Norway. My variables are log of industrial production, log of cpi, interest rates and the oil price. Could anybody tell me if I'm doing structural VAR analysis with cointegration in a right way? I'm doing it in the following steps:
1) I tested my variables for unit roots and found that in the case of Mexico and Norway my CPI variable is stationary.
2) I test for lag length and for cointegration in variables log industrial production, log cpi, interest rates and the log oil price for Canada; log ind prod-n, interest rates and log oil prices for Mexico and Norway.
3) I find the rank of cointegration for each country and also the deterministic trend assumption.
4) I construct the VECM model for the three countries (separately). For Mexico and Norway I do it without the CPI
5) I save the cointegrating equations for the three models as separate groups
6) I construct an unrestricted VAR with the variables in the following order: dlog oil prices, d interest rates, dlog cpi, dlog ind prod-n. Then I incorporate trend and cointegrating equations from the VECM as exogenous variables
7) finally I impose structural orthogonalisation and do structural analysis on this model
Hope I explained everything clearly and will be grateful if someone replied. Thanks!
I'm doing structural VAR analysis for three countries: Canada, Mexico, Norway. My variables are log of industrial production, log of cpi, interest rates and the oil price. Could anybody tell me if I'm doing structural VAR analysis with cointegration in a right way? I'm doing it in the following steps:
1) I tested my variables for unit roots and found that in the case of Mexico and Norway my CPI variable is stationary.
2) I test for lag length and for cointegration in variables log industrial production, log cpi, interest rates and the log oil price for Canada; log ind prod-n, interest rates and log oil prices for Mexico and Norway.
3) I find the rank of cointegration for each country and also the deterministic trend assumption.
4) I construct the VECM model for the three countries (separately). For Mexico and Norway I do it without the CPI
5) I save the cointegrating equations for the three models as separate groups
6) I construct an unrestricted VAR with the variables in the following order: dlog oil prices, d interest rates, dlog cpi, dlog ind prod-n. Then I incorporate trend and cointegrating equations from the VECM as exogenous variables
7) finally I impose structural orthogonalisation and do structural analysis on this model
Hope I explained everything clearly and will be grateful if someone replied. Thanks!