generating a date id and crosssectid series for dynm panels

[helplessbunny]

Dear EViews Community!
Long time admirer; first time poster!

I have massive trouble generating a dateid series and a cross-sectionalid series for my dynamic panel analysis.
My excel file looks as follows (also available as .txt): - see attachment



.) dateid series:
the dates start in the 2nd quarter of 1982 (1.4.1982) with a value of 18 and end in the 4th quarter of 1998 (1.10.1998) with a value of 84. – how can I generate a dateid series with this data? All the examples and explanations I found deal with dates that are already in date form.

.) crossid series: (the yellow column does not exist in the file - i added it for demonstration purposes)
The cross-sections can be identified by the fact that every 40 (!) rows display one individual in the cross section, ergo rows 2-41 are individual 1, rows 42-81 are individual 2, ...

In principle the data are quite fitting for dynamic panels if only it would include a column with the dates in actual date form and a column with codes for the individuals.

PLEASE help!

Regards
Your humble reader

[EViews Gareth]

It isn't particularly clear how your quarters work (if you have quarterly from 1982q2 to 1998q4, you have 67 observations, yet you say you have 40 rows per cross-section?).

I think I'd just create a new, empty, panel with the dates you want, and the number of cross-sections you want, then just do a simple, non-dated, sequential import.

[helplessbunny]

thank you so much for your early reply!

I forgot to mention - not all cross-section participants exist in all periods - f.e. individual 1 exists in periods 18-57 (row 1-40) and individual 2177 exists in periods 24-63 (rows 87041-87080) - (btw I just figured out how to do my cross-section (even if I may not need it) - its quite easy using a mod calculation for the row numbers) - and I have far more than 67 obs. - in fact I have thousands up on thousands of obs.

I will try the approach you suggested and be back with how it worked! Thank you so much!

[helplessbunny]

Hello again!

I tried your approach: open empty panel wf - periods 1982:2 - 1998:4 and ~14000 cross sections - but the problem is, that now for all 14000 cross sections EViews displays the periods from 1982:2 to 1998:4 - which is not really what I need since my cross sections do not exist in all periods but only in 40 (that are different between cross section participants).

Is there a way to generate a dateid series by decoding the number 18-84 into date? So that 18=4/1/1982 and for every +1 in the number EViews counts one quarter into the future? That would be an ideal dateid to generate my panel.

Regards
Your humble reader

[EViews Gareth]

Code: Select all

series mydate = @trend+18