How to keep effective dates, after merging panel datasets?

[Edward Wu]

Good day Statalists!

I merged a quarterly dataset with a daily dataset, and did some calculations.

Now there are two date variables, one is crspdate(Stock trading days, skip Saturdays and Sundays), the other is eadate(Earnings announcement date, mostly one date in every quarter, but could be any date during that quarter, and could be on Saturdays and Sundays).

The companies are identified by permno, so basically (permno crspdate) is unique, whereas (permno eadate) is not.

I need to make (permno eadate) unique, which means keep the observations when eadate=crspdate, but the thing is, some eadate is on Saturdays and Sundays that crspdate is not available. In this case, I would like to link the eadate with the next available crspdate, mostly next Monday.

Anyone can help? Great thank!

Here is the sample:

* Example generated by -dataex-. To install: ssc install dataex
clear
input double permno long(crspdate eadate)
10001 15108 15110
10001 15109 15110
10001 15110 15110
10001 15111 15110
10001 15112 15110
10001 15218 15220
10001 15219 15220
10001 15222 15220
10001 15223 15220
10001 15224 15220
end
format %d crspdate
format %d eadate

I need to keep

10001 15110 15110

and

10001 15222 15220

[Clyde Schechter]

This will do it:

by permno eadate, sort: egen next_crspdate = min(cond(crspdate >= eadate, crspdate, .))
keep if crspdate == next_crspdate

[Edward Wu]

It worked! Thanks a lot!