EViews.com

Hi All,

I am regressing the time-series data based on monthly observations.
The regression is heteroskedastic, and not normally distributed.
I would like to test for Autocorrelation by using "Serial correlation LM test" with 12 lags as the data is based on monthly observations.
unfotunately, it is not possible as the error message "positive or non-negative argument to function expected" appears. But when I used maximum up 10 lags it works.

What can be the problem would be kind to help me ,please.

Thanks in advance

Impossible to say without seeing it.

Please see the file attached "test"

I suppose, this because the data is unavailable before the date 2013M02 in "series01"
Because when I use, lags 3, 4, up to 6 it works.

Kindly ask you to check the file.

Thanks in advance for your cooperation!

Yep, that's it.

You have 5 observations in your equation and four regressors (a really bad idea by the way). There's no way to perform the test on more than a few lags.

This is just one of 100 funds. And it is so recent fund (age) that, there is no way to do it differently.
What is the bad idea? could you be kind to explain?
I would really appreciate your explanation, as maybe I am doing smth wrong.

You can't have more coefficients than observations. And having nearly as many coefficients as observations is usually a very bad idea.

Many thanks. I will make a not for it.
One more question in this case.
After checking for heteroskedasiticity with Breusch-Pagan test, I have rejected the null, which means there is a heteroskedasticity.
And as I clearly understood from the literature, I should smooth or correct it by using White's heteroskedasticity-robust standard errors.
Would you be kind to assist me, how to do it in Eviews.

Thanks in advance!

Type "heteroskedasticity" into the help system.

Note that heteroskedasticity tests and corrections require large samples.

Sorry, but there is an info about how to conduct White's but not about the robust standard errors to correct

Go to the manual and on the contents tab drill down to

User’s Guide > Basic Single Equation Analysis > Additional Regression Tools > Robust Standard Errors

Done! Thanks a lot!