Page 1 of 1

choosing deterministic trend specification in cointegration

Posted: Thu Feb 04, 2010 8:09 pm
by anjankorke
hi
I have following problem. pls help me
1. First i estimated unrestricted VAR
2. then, chossing lag lenght criteria, VAR lag order 2 is selected
3. i did cointegration test and found -- 1 cointegrating relationship for two types of options mentioned in deterministic trend specification-
Intercept(no trend) in CE- no intercept in VAR (option 2)
and
Intercept and trend in CE- no trend in VAR (option 4)
My problem is-
1. which option to choose
when i choose option 4, the result is very much different than the result from the option 2.
The coefficient representing long run normalized relation is positive with option 2 while it is negative with option 4. but in option 4 there is positive and significant trend. how to interpret the trend component?
thank you