Which identifier to use when computing ratio's and returns (permno or cusip)?

[Julien Maas]

Dear Stata Community,

When dealing with merged data from CRSP and Compusstat.

Is it generally better to compute ratio's and returns by permno, permco or by cusip? And what would you say is the best way to deal with duplicates?

Where each permno is a permanent identifier for each security and each cusip is a identifier for each firm that may change over time.


For the sake of simplicity consider the following data example;

* Example generated by -dataex-. For more info, type help dataex
clear
input long(globalkey lpermno lpermco) int fyear str9 cusip str29 companyname float(totalassets ib)
 1338 12125 5180 1982 "02199Y201" "ALTA ENERGY CORP-OLD"  4.579   .078
 1338 12125 5180 1983 "02199Y201" "ALTA ENERGY CORP-OLD"  6.401   .027
 1338 12125 5180 1984 "02199Y201" "ALTA ENERGY CORP-OLD"  7.185   .145
 1338 12125 5180 1985 "02199Y201" "ALTA ENERGY CORP-OLD"  8.109  -.143
 1338 12125 5180 1986 "02199Y201" "ALTA ENERGY CORP-OLD"  6.691  -.094
 1338 12125 5180 1987 "02199Y201" "ALTA ENERGY CORP-OLD"  7.527  -.046
 1338 12125 5180 1988 "02199Y201" "ALTA ENERGY CORP-OLD"  5.701 -1.564
 1338 12125 5180 1989 "02199Y201" "ALTA ENERGY CORP-OLD"  5.338  -.173
20063 12125 5180 1989 "021270301" "ALTA ENERGY CORP"          .      .
20063 12125 5180 1990 "021270301" "ALTA ENERGY CORP"     19.125  -.078
20063 12125 5180 1991 "021270301" "ALTA ENERGY CORP"     28.702  1.175
20063 12125 5180 1992 "021270301" "ALTA ENERGY CORP"     43.792   .481
20063 12125 5180 1993 "021270301" "ALTA ENERGY CORP"     58.467  -.858
end

Here I have a security who's cusip and globalkey changes yet it's permno and permco remains the same.

So the duplicate is caused by the fact that in the year 1989, I have two different cusips which both report about the same permno.

Thank you in advance for your time and consideration,

Kind regards, Julien

[George Ford]

permno doesn't change like cusip can. so permno I'd think.

[Devra Golbe]

I think the answers to your queries depend on your research question. As you note, Permno is the stock (share class) identifier assigned by CRSP. Permco is the unique company-level identifier assigned by CRSP. I'll bet your variable globalkey is the numeric version of Gvkey, a 6-character string (often with leading zeros which are dropped when it's read as numerical). Gvkey is the unique identifier assigned to each firm in Compustat. In your example, when globalkey changes lpermco changes, too. I don't know if that's always the case.

If you are interested in ratios and returns by stock issue, then use permno as your identifier. If your unit of observation is the firm, then use either gvkey or permco.