If we fail Ramsey’s RESET test, then the easiest “solution” is probably to transform all of the variables into logarithms. This has the effect of turning a multiplicative model into an additive one. If this still fails, then we really have to admit that the relationship between the dependent variable and the independent variables was probably not linear after all so that we have to either estimate a non-linear model for the data (which is beyond the scope of this course) or we have to go back to the drawing board and run a different regression containing different variables. does have any 3. solution way?
Yilmaz,
If you fail a Ramsey RESET test you are formulating a model with a violation of functional form. However, there could also be an interaction between the unmodelled functional form and the residual variance.
You could supplement your model with a White general specification error test and see whether the cross-products are significant as well as the squares and cubed terms.
You should also test the residuals for ARCH effects as well to be sure that this is not the case. If you are omitting a quadratic trend, you may have to model it with the proper deterministic component.
Robert
if is it possible may I ask something about egarch models?
does it hve any rule about coefficients like garch model also after estimated egarch model does conditinal variance series (make garch series in eviews) has to be stationary ?
The EGARCH model is a very good general model that handles the logged
series well when examining the conditional variance that may have a volatility
smile or smirk due to the asymmetric impact of negative shocks. You need
to examine your Jarque- Bera tests for residual normality to determine whether
you should be fitting a Gaussian model. You may prefer a student-t, an asymmetric student t or a LaPlace distribution for example if your residuals are fat-tailed, skewed, or pointed. You may want to check it with sign-bias tests to determine whetheryou should be using the asymmetric student t distribution. Your Q statistics on the squared standardized residuals may not reveal this. If you want to do the graphical check, plot the volatility smirk to see how pronounced a smirk you get.
Cheers,
Robert
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression