EViews.com

Don't.

You can't take the log of a negative number.

No idea. Sorry.

Don’t take quarterly averages. Use last day of quarter if possible. Taking averages screws up autocorrelations.

standard error = sqrt(var(beta1) + x^2*var(beta2) + 2*x*cov(beta1,beta2))

Check to see if the sector variable has NAs.

Look at the @expand command for controlling for multiple sectors.

Just got
Came after doing crtl-C to copy a frozen table. Happened twice.
Closed frozen table reopened and everything was fine.

Thanks. I'll do that. In what will not surprise you, trying it instantly gave an EViews bug. (I should explain to the casual reader that EViews is programmed to do weird things to me. No one else ever has a problem.)


The results box overlays the search box.

I know this isn't exactly an EViews bug, but it's annoying. Maybe you could goose the FRBStL folks?

If you enter


you get told

Thanks Glenn. I am now getting the same forecasts that you report. Despite having gotten something different 6 times before I posted.

I have a dumb question about dynamic versus static forecasting. I have estimated the following equation from 1953 through 2018 ls spreaddemeaned ar(1) ar(2) ma(1) If I forecast for 2019 from the equation using dynamic forecasting I get all zeros. If I forecast using structural forecasting I get a ze...

You can also replace the coefficient c(i) with exp(c(i)). That guarantees that exp(c(i)) will be positive.

Integrated of order zero means it's not integrated.

If data isn't integrated, it can't very well be cointegrated...

The log likelihood statistics are displayed in the lower panel. Just do the calculation by hand.

The difference in R^2 is because you are changing the variance of the dependent variable. Note that the coefficients, t-statistics, etc., are the same.