I carried out ARDL bounds test which shows cointegration but my long run coefficient is insignificant. when i generate the ardl-ecm however, these are significant. what do i conclude on cointegration? should i interprete the ecm and or do i need to use another method?
Assuming that the model is correctly specified and diagnostics are OK, it means that there is no relation between the dependent variable and the insignificant independent variable in the long-run.
I suggest extending the study period (the number of years) or adding other variables related to the study
Zaitri Sara Çağlar Karaduman
I am studying interest rate pass through in my country. Would it be okay to conclude on insignificant long run pass through and provide justifications based on that or do i add other variables. Other studies in the field however worked mostly on interpreting cointegrating variables but their models were bivariate.
Conluding that all the LR coefficients to be insignificant is not a good idea. Instead, you may try including new variables or changing the length of observations. If the conclusion does not change, then you can present all your results and argue that there are no LR relation. I suggest following the p value for the RESET test. The higher the p value for RESET test, the lower the probability of omission of other variable(s).
Btw, have you checked your independent variables for multicollinearity?
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation