Chow and Cross Sectional Regression

[mscarlatos]

Does Eviews 6 permit cross sectional Chow tests? I have time series for 7 countries with a LHS and 4 RHS variables. I would like to run a CHow on 3 of the 7 and the same on the remaining 4 of seven. If so, could you tell me what tabs I look for?

[EViews Gareth]

Not 100% certain I follow what you're asking here. You seem to be saying that you have a panel workfile, and you want to test whether the coefficients are the same for three of the cross-sections? The only way to do this is manually with dummy variables (for a Wald test), or by estimating two equations for the LR statistic.

[mscarlatos]

Thank you. If I do a panel regression on one of my three panels and get the panel regression parameters and other output, does Eviews6 permit stability tests such as Chow or QuandtAndrews, on those coefficients?

[EViews Gareth]

No, stability tests are not available for panel equations.

[mscarlatos]

Would that be because Eviews does not do them or because they are not to be done econometrically? If it is an Eviews characteristic, I guess I can manually generate my relevant Chow F stat by breaking each panel into two (before and after a suspected breakpoint), running a fixed effects regression and simply plugging in the residual sum of squares for the overall panel and the subpanels into the Chpw F formula. Does this seem reasonable?