manipulation of an unbalanced panel

[t_a_v_o]

Hi,

I'm quite new to eviews and was wanting to manipulate an unbalanced panel to work out the % change across firms

Okay so I have a large number of firms e.g. 1000+ and yearly data of different variables between a certain period e.g. 1995-2008
Some firms however only have data not for the whole period e.g. just have data between 1995-2000 or 1999-2007 etc

firm year variable_x
1 1995 6897
1 1996 7902
1 1997 1890
1 1998 2755
1 1999 2746
1 2000 7888
2 1995 7000
2 1996 7544
3 1995 8776
3 1996 7655
3 1997 6543
3 1998 5432

So I was just wondering what is the best way to work out the percentage change across the firms because they have varying years of data?

Thank-you

[EViews Gareth]

What does "percentage change across firms" mean?

[t_a_v_o]

Sorry I meant the percentage change across years for each individual firm.

[EViews Gareth]

Can't you just use the @phc function?

[startz]

Can't you just use the @phc function?

@PhC requires approval of your dissertation committee. @pch is easier.

[t_a_v_o]

Thank-you

[t_a_v_o]

Also sorry what would I use to work out just the change across years, not in percentage form?

[EViews Gareth]

You'll have to define what you mean by "change", but I think you just want the D(X) function.

[t_a_v_o]

Yeah that was what I meant, however when calculating these, eviews doesn't take into account the different firms and works out the change across the whole data set, e.g.
firm year variable_x
1 1995 6897
1 1996 7902
1 1997 1890
1 1998 2755
1 1999 2746
1 2000 7888
2 1995 7000
2 1996 7544
3 1995 8776
3 1996 7655
3 1997 6543
3 1998 5432

when calculating the change/difference in variable_x it doesn't take into account the fact that the firm has changed, so to take into account the different firms' different years of data I would have to set restrictions right? But because my firms vary in the amount of years they have there's just not one simple restriction I can apply? hope that kinda make sense?

[EViews Gareth]

As long as you have correctly told EViews the nature of your panel data, it will not mix firms.

[t_a_v_o]

Hmmm I think that's the problem...
Finding this a bit difficult so I'm going to run through what I did can you tell me where I may have gone wrong please?

Firstly I have data that is across countries and also different firms across each country (some countries have a lot of firms and others don't). The time period is 1995 to 2008 (looking at annual data). However each firm has varying years of data due to availability (e.g. some have 1995-2008, some 1997-2004, some 1996-2006 etc). Therefore it's an unbalanced panel.

Ultimately I want to just analyse the firms right now and then later take into account the country...

So I did the File--> Open --> Foreign Data as Workfile --> Excel Read Step 1 of 3 --> Next --> Excel Step 2 of 3 --> Next --> Excel Step 3 of 3 (This is where I'm not too sure) What kind of structure should I use?

I hope that all makes sense
Thank-you

[EViews Gareth]

http://forums.eviews.com/viewtopic.php?f=7&t=74

[t_a_v_o]

Thanks,

This might be a silly question but I've got my file structured as a panel.
And am just looking to get some summary statistics...
Each series is structured by countries like your example.
I was wanting to get summary statistics for each individual country, how would I go about doing this when the each series has all the countries in it?

Hope that makes sense.

[EViews Gareth]

View->Descriptive Statistics->Stats by Classification, then choose which stats you want to see, and enter "country" (or whatever your cross-section ID series is called) as the classifying series.

[t_a_v_o]

Thanks again Gareth!