Setting up panel data

[tommys]

Hello,

I want to run my regression with standard errors clustered by year of observation, where year is not a unique identifier and my data is currently in Excel. What is the best format to import my data in, and how do I make sure my data is set up correctly - currently I seem unable to import it in panel format.

Thanks

[EViews Gareth]

We'll need more information on the structure of your data

[tommys]

basically I have a unique ID for each observation, a non unique year and then my regression variables (Dependent, categorical, continuous). Currently the data is stacked by year. I want to be able to run panel regressions with standard errors clustered by year. I have uploaded my data file for reference. Is there anything else needed?

Thanks

[EViews Gareth]

You don't seem to have any information regarding cross-sections?

[tommys]

How do I set up cross sections in my data? I can seem to find the appropriate section in the user guide.

[EViews Gareth]

It would just be a series (or a column if in Excel still) that contains a cross-section identifier.

[tommys]

As I said before I want to have the cross sections of Year and Group, where neither is a unique identifier. Does that help?

[EViews Gareth]

No. You need to have a way to uniquely identify each observation. A panel is a data set whereby each observation can be uniquely identified by time and cross-section. You do not have that structure. There are multiple observations that have "DND" as their "GROUP", and "2000" as their "YEAR". Thus they are not uniquely identified.

Until you have a way to uniquely identify your observations, you cannot form a panel. This is not an EViews issue, rather it is a basic panel data issue.