Is there a difference between cointegration and long-term relationship? Example (ARDL) The results of the Bound test indicate a co-integration while the long-term relationship is not significant
co integration the first step to insure of sationearity in levels or to do first difference, after you can use the Schurtez criterion,or H-Q criterion , the to the co integration ,when its not significant ,you may use the SVAR model.
cointegration indicates integration of two or more variables at the same order.
in Econometric terminology integration refers to the order (ie., level, first differenced, second differenced etc.,)at which a variable becomes stationary.
if two or more variables are integrated at same order, it is treated as those variables are having long run equilibrium relationship i.e, they are cointegrated.
Long run relationship is a very wider term which includes cointegration also.
In ARDL, we test the dependence relationship between the two variables one is dependent variable and other one is explanatory variable and/or its lagged values.
In cointegration testing, there is no such dependence relationship between the variables.
i hope this reply will be useful to you.
further discussion in this regard is most welcome.
Dear Mohammed
Some variables are stationary at the level and the other in the first difference how can used SVAR?
There is many difference between co integration and ARDL (auto regressive distributed lags models.
Well, my simple contribution is that "cointegration" implies that a "long-run relationship" exist between/among the variables. If the variables are not cointegrated, no long-run relationship exists...hence, estimate only the ARDL model. But if there's cointegration, then estimate the ECM/VECM.
Dear Ngozi Co-integration is a sign of a great potential for a long-term relationship, but empirical evidence may have been found on that relationship.
If the variables are cointegrated, they are a long term relationship but it is possible that the coefficients are not significant. If you can a long term relationship you can have variables I(0) no cointegrated.