Dear all,
A quick one to those familiar with CUSUM (this is my first-time), how can I interpret the attached Stata output. Please anyone can contribute you may not be a Stata-user. Please take a look at the graph and advise whether the model is stable or not.
Thank you.
Ngozi. your model is unstable since it goes outside the 5% boundary lines. Definitely, you may be missing a dummy variable. Think of what happened within that area and include either a one-off or impulse dummies because the test statistics wanders off and comes back to stabliliy.
Dear Ngozi,
The graph shows that the coefficients of your model are not stable in the long-run @ 5%. The reason for this maybe from the variance of the data-set between other variables and itself. You can try to improve the predictive power (i.e. improving the stability of the coefficients) of he model by differentiating or by taking natural log of the variables before developing the model.
I do not see the need for Imokhai's comments, nevertheless taking natural logs are known by everyone as a solution for interpretation and minimization of variation. One has to assume that you have done that. Instability often arises due to the omission of variables in the short-run dynamics. So start by experimenting with the residuals from the UECM; get the SEE & divide "saved residuals" by SEE and that will give you and idea of the outlying observations.
Think of SAP beginning in 1986, and you want to remember interest rate deregulation thereafter. ADD d86 and D98. Look at this paper " Re-Examining Exchange-Rate Risk Effects and Export Trade Using
the ARDL Bounds Testing Approach " IJEF -- you can see my RG
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs