DUMMY VARIABLES - ZIVOT ANDREWS / Sign restrictions

[tdabs]

Hi all,

Thanks for taking the time to read this - I am new to this forum.

I am currently undertaking a project looking at the responses of inflation and output to monetary policy. When using the ADF and PP test to check my data is stationary, it provides an argument to take the first difference of most of the variables. However, this is biased if there is a a structural break and as my data spans over the financial crisis, this is very likely.

I therefore used the Zivot Andrews add in and ran that test on both the levels and the first differences. It provides me with a break point from minimising the t statistic. I am confused by two things, if there seems to be more than one break point - and i create a dummy, which date should i create it for?

Also, the null hypothesis on eviews is that there is a unit root and a structural break, however, some papers postulate the null is that there is a unit root with no structural break. I was wondering if people had any idea on this?

Also, another point- does anyone know how to impose SIGN RESTRICTIONS on my VAR model?

Thank you!

Tom

[dakila]

Use the srvar add-in.

[tdabs]

I have just found the add in thank you. a couple of questions on this if you could help:

- What are the % of confidence band
- What should be inputted into 'sign restriction vector'? Should I create a vector before? How do i specify the sign?

TY!

[dakila]

Did you read the instruction? Did you run the example?
The confidence band is something which is similar to the confidence interval. It is Bayesian terminology.