EViews.com

You are correct that it does not allow for solution of the model without prior specification of the underlying observed states and in that respect it is limiting. In a sense it is akin to embedding an equation object in a model without providing exogenous series. In the event that one does generate ...

Try the "propnas" option which produces NAs for partially observed periods.

A fix has been posted.

Sorry I missed seeing this question.

The CSD test is available on a series in a panel workfile. Simply open the series, select View/Cross-section Dependence Test.

If you forecast without the "i" option, you'll get the predicted probabilities.

White period allows for intra cross-section (between period) correlation. White cross-section allows for intra period (between cross-section) correlation.

White period is cross section cluster robust.

We will look into this.

[Edit: in the process of testing a fix.]

Perhaps there is, but to be honest I'm still not clear as to what you want to compute.

It helps a bit, but we are still unclear as to what it is you are trying to do.

What do you mean by accounting for the autocorrelation?

Do you wish to model it and compute the correlations of the residuals? Perhaps compute a long run covariance?

What do you wish to do with the results?

The null is that there is no contemporaneous correlation in the residuals between the disturbances in different cross-section units. The alternative is that there is contemporaneous correlation.

http://www.eviews.com/help/helpintro.ht ... 23ww191025

I've looked in depth at the EViews code and believe that there are no issues (but can be convinced otherwise). I haven't gone through the paper you reference carefully, but the asymptotic analytic standard errors for the Generalized impulses follow directly from the results for the Cholesky standard...

As is well known, the Hausman test statistic is not guaranteed to have a computed covariance matrix that is positive semi-definite. Notably, it is computed as the difference of two matrices which is often negative definite, especially for small sample sizes. The NAs are an indication that the corres...

Hi, From your description, there are a number of things that you may wish to do and with additional information, we can discuss the various ways of doing them. Briefly, however, if all you want to do is to plot the derivatives that you computed against the fitted values, you should use Proc/Forecast...

The R2 are different because the OLS and FMOLS coefficient estimates are presumably different, as are the residuals.

You may use them, but in your case since the FMOLS objective is very different from the OLS estimates, I'm not certain what one would conclude from the differences.