My data are I(2). So, after first difference they become I(1). My optimal lag length is 5. I run Johansen co-integration test, and result shows they are co-integrated. After that, I run VECM. As procedure suggest, I used 4 lag length in VECM calculation. But it returned with following message "Insufficient Number of Observation". It should be noted that, I have 28 observation , from 1990 to 2017.
The question is how many of these four variables is endogenous. If all are endogenous and the lag-length is four, then the short-run model for each of them involves estimation of 18 parameters (including the constant term and the error correction term). Thus, the number of parameter estimates in total will be 72. You can not estimate this much parameters with a sample size of just 28!
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running