Problems with merging CRSP with Compustat-CRSP merged (Stata)

[Robert Kuijt]

Hello,

For my thesis I would like to merge CRSP with Compustat-CRSP merged. Crsp mainly covers stock market data from exchanges: NYSE, AMEX, NASDAQ. Compustat mainly covers accounting data for public and private companies. Unfortunately I am having trouble by doing so. Here is a small description:

CRSP
2,201,325 obersevations
31 variables
panel data

Compustat-CRSP merged (annual)
164,280 observations
67 variables
panel data

When I try to merge these datasets one-to-one on key variables (permno and date) in Stata I keep getting the message:

variables permno date do not uniquely identify observations in the using data

Has this something to do with fiscal and calender year data?

So far I removed all the duplicates from the two datasets by using:

duplicates list permno date

Can anyone help me with this?

Kind regards,
Robert




Cross-posted here: http://www.talkstats.com/showthread....at-CRSP-merged

[Devra Golbe]

Robert,

duplicates list permno date

lists duplicates; it does not remove them.

More importantly, you should study the extensive documentation for the data files. Compustat firms are uniquely identified by GVKEY. CRSP identifies stock issues by PERMNO. Each GVKEY may be associated with more than one PERMNO. Thanks to various kinds of reorganizations, sometimes one PERMNO links to multiple GVKEYS (though matching by date should eliminate those problems for you.) While Compustat is fiscal-year based and CRSP is calendar-year based, Compustat includes a variable identifying the calendar date for the end of the fiscal year. Here's a link to a support page which gives an introduction to some of these issues.http://www.be.wvu.edu/wrds/CCM_faqs/index.html

Devra

[Robert Kuijt]

Hi Devra,

First thank you for your reply and the link. In the meanwhile I have spoken to the helpdesk of CRSP and she told me this:

The Compustat-CRSP merged (CCM) database itself is a product that already contains all of the CRSP pricing data and all of Compustat’s fundamental data merged into one product. We do the hard work for you of mapping our PERMNO’s to their GVKEY codes. To do your research, you can just open CCM through our CRSPSift tool and get to work – no merging is needed.

With that said, do I still need to merge the datasets? I am using a couple papers as reference for my thesis and they are mentioning a couple of things. For example they talk about share code 10 or 11. In my experience that is a service that only CRSP has.

If this is true a guess from my point of view is that the Compustat-CRSP merged database was not in place when authors like Fama and French (2001) wrote their article.

Regards,
Robert

[David Benson]

I know this is an older thread, but given that merging daily / monthly stock return data from CRSP to annual accounting data in COMPUSTAT, I pulled together some helpful links on merging CRSP to Compustat here https://www.statalist.org/forums/for...rsp-linkytypes