Park Test - Need Help!!

[meinhaj]

Hi, I am trying to do the Park test using Eviews 5 and can't figure it out. Tried googling for it, reading the manual, browsing over books, but nothing, zit.

[meinhaj]

The form I am using for the regular OLS regression is:
RD c sales

I am taking the residuals and then trying:

log(resid01^2) c log(Sales)

The answer won't come up right. I tried the White test and that comes out perfect to the dot.

[startz]

check that all your values of Sales are positive.

[meinhaj]

Hi startz, thanks for the reply. I checked, its all positive.

The stochastic behavior of the Park Test is making me very stochastic, showing collinearity between the Park Test and me, please help :)

[startz]

As a double check, make sure that the original OLS and the test equation are using the same number of observations. Past that, tell us why you think the results you're getting are wrong.

[meinhaj]

Hi Startz, thanks for the fast reply. I am trying to shadow the answers in Basic Econometrics (Porter and Gujarati). I can get to the White test fine (Because its given in Eviews) but when I try to derive the Park test it comes out wrong. Maybe my equation is wrong or I am not specifying it correctly then. Maybe it would help if I could find out how I can do the White test without using the given View>residuals>White test built in function. That is, what form my OLS regressed against the residual would be.

[startz]

EViews 7 shows you exactly what equation is being estimated for the heteroskedasticity test and would let you do a custom test such as the Park test. I don't know if these features are available in EViews 5.