EViews.com

Hello,

I am trying to insert a time trend into my model estimating the impact of HIV on GDP. I have included capital investment, gov spending HIV prevalence and Life Expectancy as independent variables. I took logs of all variables in the estimation equation so my equation is: log(gdp) c log(govspend) log(capital) log(hiv) log(life). I also estimated this equation with cross-section and panel fixed effects. However, whenever i try and insert @trend to include a time trend within my model to test for serial autocorrelation, I get near singular matrix error. I have no idea what this is and no idea on how to fix this. Please Help. I have attached my workfile and used all the adjusted data for the regression and life expectancy.

You can't have both a time trend and period fixed effects. They're redundant.

You can't have both a time trend and period fixed effects. They're redundant.

Thank You. I assume this means that sice i'm running the fixed effects I don't need to run a time trend?

That's right.

Slightly off topic but at what confidence interval does eviews run regressions at by default? 0.95 or 0.90?

Slightly off topic but at what confidence interval does eviews run regressions at by default? 0.95 or 0.90?

I'm afraid the question doesn't make sense. Regressions aren't run at any confidence interval. After a regression is run, one can use the output to compute any desired confidence interval.