Standard deviation over 5 years

[Brandolf]

Hi!

I have a question about the standard deviation function in EViews. I got the GDP of a country over a period of about 30 years. On the left hand side of the regression I would like to have the standard deviation of the GDP. Because its not possible to compute the standard deviation of a single year, I have to compute it for let's say 5 years. So the first value would then be for example the standard deviation from 1990-1995 and the second value 1991-1996 and so on. I know further, that there will be an autoregression-problem if I estimate this equation.

How can I estimate a regression of this type in EViews? And how can I avoid the autoregression-problem? Is there anybody who can help me?

Thanking you in anticipation!

[trubador]

You can define your dependent variable as @movstdev(y,5) and estimate it along with other variables. Afterwards, you'll probably have to build an ARMA model or a GARCH model to overcome autoregression and/or heteroscedasticity problem.

[Brandolf]

Thx for your response! I will try it.

Edit: It worked and it's exactly the solution to my problem. Thx! The autoregression problem is not an EViews problem I guess. I have to think about it (thx for your inputs).