Problem with creating for loop

[Katy Red]

Hello Statalist!

I am working with data that looks like this:

analys ticker
72 ATX
72 DMV
81 DMV
81 AZW
96 DMT

I am trying to create a dataset that, as a first step, has an observation for every day of the period (1Jan2007-31Dec2020) corresponding to each of these observations, like this

analys ticker date
72 ATX 1 Jan 2007
72 ATX 2 Jan 2007
.
.
.
72 ATX 31 Dec 2020
72 DMV 1 Jan 2007
72 DMV 2 Jan 2007
.
.
.
72 DMV 31 Dec 2020

and so on. This is only the first step and I will later also have to merge in other data, create running indicator variables, etc. and because the file size will blow up I have resorted to doing this via a for loop. I used the following code:

egen analysbyfirm = concat(ticker analys)


levelsof analysbyfirm, local(tickeranalyslevels)
foreach i of local tickeranalyslevels {
preserve
drop if analysbyfirm != `i'
expand 5114
gen date = mdy(12, 31, 2006) +_n
format date %td
append using experiment.dta
save experiment.dta, replace
restore
}

but at the end of the for loop Stata returns the error:

00BV70 invalid name
r(198);

I am very new to Stata and completely at a loss, any help would be deeply appreciated! Thank you!

[Nick Cox]

This might as simple as

egen long id = group(analys ticker) 
compress id 

tsset id date 
tsfill

but that code could be scuppered by complications, for example if date is not a Stata numeric date variable.

It should certainly not be necessary to loop over all stocks with separate files for each;

Whether this is a good idea is a different question, as it will just imply lots of observations with missing values for weekends, public holidays, and otherwise if any stock was not traded over all of this period.

[Katy Red]

Hi Nick, thank you so much for your response, this really helps! The non-trading days should not be an issue as I'm also merging in a set of "incident dates" which are not restricted to trading days, but this gives me a lot to think about with regards to overall data structure.