EViews.com

Hi,

I am currently a student attempting to conduct a volatility test on the bond prices of five different countries. Consequently, I am trying to conduct a panel GARCH analysis on the prices which unfortunately, is not yet available in EViews8.

As a result, I attempted to write a GARCH program hoping it will work when I use it with panel data. However, as I attempted to run it, it prompted an error message stating I have "missing values in @logl series at current coefficients at observation 8/08/2011 in "DO_LL0.ML(SHOWOPTS, M=1000, C=1E-5)" ".

Could anyone take a look and help me identify the problem?

Up until now I have only been using the arch_t1, garch1, egarch1, bv_garch sample programs provided by EViews and the program written by jma42 in a previous post as reference.
(jma42's post: http://forums.eviews.com/viewtopic.php?f=5&t=4935).
I have no previous experience using Eviews before March this year, and learning how to program has been rather challenging.

Attached is the program and workfile.

Any help is greatly appreciated. Thank you!

Your likelihood depends upon past values of epsilon and H2. Therefore in order to estimate, you need to set the sample to not include the first observation (date), and you need to set pre-sample values for H2 and epsilon.

Thank you so much, EViews Gareth. I am able to run the program now after correcting the samples. However, the output has a "warning: singular covariance - coefficients are not unique" message. Is this problem due to programming or is it associated to the data?

Data or model, probably

Alright, I'll work on it. Once again, many thanks, Gareth. :D