I am currently doing a time series analysis using VAR/VEC model with 1 DV and 3 IV to evaluate the association among variables. Reason i choose VAR/VEC is because it more suitable to analyse multivariate time series data. My question is, if my data set does not fit the model or have an outlier, what should i do?should i analyse with different model or should i combine/elaminate the variable?Thank you in advance.
In general I'd say that your model is still unknown so investigate how it should be. You could analyze the two possibilities and decide which would work best for your purpose. AIC and BiC are two statistics that may help you .. you can find some beginning references in the attached document. Best wishes David Booth
Identify the time when this outlier began and apply a STRUCTURAL BREAK test. If the test fails, re-analyze the data past the structural break with a new set of coefficients. But this is just one of a possible ways to deal with the lack-of-fitting.
I would return to my data and ask the following questions
These are questions that your econometrics program can not answer. If you find that one or more of the above questions has a positive answer you need to adjust your model.
- Badges
- Science topic
- Similar topics
- Condensed Matter Physics
- Zero