GARCH (conditional variance models) for panel data?

I would like to run an ARCH-GARCH model for my unbalanced panel data (100 panels and monthly data from 1950 to 2020).

I have Matlab2018 and the Econometric toolbox. I found examples to use GARCH with time-series, but not with longitudinal (panel) data.

Can this be done with Matlab? May you direct me to a help-file for it? I don't find it...

Thank you so much.

Answer 1

[Tejas]

Hello Marta,

The garch model can be applied to panel data, provided that the data for each panel is considered as an individual time series.

To apply the garch model to each set of panel data, one should loop through each one. Within this loop, include the following steps:

• Define the garch model. Below is an example of how to do this:

    mdl = garch('GARCHLags',1,'ARCHLags',1); 

• Obtain estimations from the model.

    [estMdl, estParamCov, logL, info] = estimate(mdl, panelData); 

For further details on the garch model, consult the following documentation:

https://www.mathworks.com/help/releases/R2018a/econ/garch.html .