Search

Calculation of analyst coverage using I/B/E/S summary file
 Posted: 31 March 2013 01:48 AM [ Ignore ]
Jr. Member
Total Posts:  34
Joined  2013-03-31

Dear Joost,

I usually calculate analyst coverage using I/B/E/S detail file.  I calculate the total number of unique analysts issuing earnings forecasts for this firm during the 12-month period before a firm’s fiscal year ending date by using the historical detail file of I/B/E/S.

I find some other papers use the I/B/E/S summary file to calculate analyst coverage. They describe it as “the arithmetic mean of the 12 monthly numbers of earnings forecasts for firm i extracted from the I/B/E/S summary file over fiscal year t”. There are somethings I could not understand.

After the below constraints, I found that for a fiscal year there are 1/6 of obs have more than 12 group of “ticker and fpedats” combined.

DATA SUMMARY;SET IBES.SUMMARY(KEEP=OFTIC CNAME TICKER CUSIP FPEDATS MEASURE FISCALP NUMEST STATPERS FPI);WHERE YEAR(FPEDATS) BETWEEN 1999 AND 2012 AND MEASURE=“EPS”
and FISCALP=“ANN” and FPI=‘1’;RUN;

PROC SORT DATA=SUMMARY;BY TICKER FPEDATS;RUN;

proc sql;create table summary as select *, count(*) as n
from summary
group by ticker, fpedats;
quit;

data test5;set summary;if n>12;run;

I was confused, how do people usually calculate analyst coverage when they are using I/B/E/S summary file? I was wondering which way is better? Using the summary file or the detail file?

I use the following code calculating coverage for a fiscal year using IBES detail file. Is it appropriate?[/color]

data decoverage;set details;if gvkey ne ‘’;run;

proc sort data=decoverage nodupkey;by gvkey ANALYS ANNDATS;quit;

DATA COMP;SET COMP;
FORMAT BDATE YYMMDD10.;
FORMAT EDATE YYMMDD10.;
RUN;

PROC SQL;
CREATE TABLE comb AS SELECT A.*,B.*
FROM comb AS A LEFT JOIN
decoverage AS B
ON A.gvkey=B.gvkey AND A.BDATE<=B.ANNDATS AND A.EDATE>=B.ANNDATS
order BY gvkey,Fyear;
QUIT;

proc sort data=audit3.compfeeib nodupkey;by gvkey ANALYS fyear;quit;

proc sql;create table comb as select *,count(*) as decoverage
from comb
group by gvkey,fyear;quit;

proc sort data=comb nodupkey;by gvkey fyear;run;

Thank you very much,

Anna

 Profile

 Posted: 31 March 2013 01:50 AM [ Ignore ]   [ # 1 ]
Jr. Member
Total Posts:  34
Joined  2013-03-31

I was struggling with the above calculation because I could not get results similar to the prior published paper.

 Profile

 Posted: 31 March 2013 09:42 AM [ Ignore ]   [ # 2 ]
Total Posts:  901
Joined  2011-09-19

hi Luckyanna,

The summary statistics files are derived from the detail files, so, both the number of analysts should be (very) similar.

Maybe I am overlooking something, but shouldn’t there be 12 monthly obs from the summary file? (it says > 12)

Anyway, what I recommend doing is to take a single firm (or a few firms) for a single year, and get the records for these firms from both datasets. That way it is probably easy to see what is going on.

Hope this helps,

Joost

Signature

 Profile

 Posted: 31 March 2013 06:56 PM [ Ignore ]   [ # 3 ]
Jr. Member
Total Posts:  34
Joined  2013-03-31

Hi Joost,

Thank you for your reply. Yes, it can be much more than 12 records.  (After I constrain MEASURE=“EPS” and FISCALP=“ANN” and FPI=‘1’;)

I think I need to further check with wrds.

Best wishes,

Anna

 Profile

 Posted: 01 April 2013 09:46 AM [ Ignore ]   [ # 4 ]
Total Posts:  901
Joined  2011-09-19

hi Anna,

Manual inspection of the data will probably point out what is going on (there should be only 1 observation for ‘eps’, fpi=1/fiscalp=ann per month).

best regards,

Joost

Signature

 Profile

 Posted: 06 April 2013 12:56 AM [ Ignore ]   [ # 5 ]
Jr. Member
Total Posts:  34
Joined  2013-03-31

Hi Joost,

Thank you for your suggestion. I have figured out what is going on by checking the details.

Cheers,

Anna

 Profile

 Posted: 10 December 2013 08:19 PM [ Ignore ]   [ # 6 ]
Newbie
Total Posts:  22
Joined  2013-01-15

Hi,

I have a related question.

for a single company,

I downloaded both detail file and summary file, and want to compute the analyst coverage ,i.e. # of analyst ,

please see attached example, they seem to give different # of analysts, is it 120, or 40 some?

thanks !

File Attachments
 Profile

 Posted: 16 December 2013 02:11 PM [ Ignore ]   [ # 7 ]
Total Posts:  901
Joined  2011-09-19

hi Rochelle,

If you look in column H (ANALYS) of the detailfile, you’ll notice that that value is not unique. In other words, there are revisions in those 120 rows. I count 36 unique values for ANALYS in that spreadsheet (I don’t know what the ‘0’ means for ANALYS).

The summary file does not have this issue (number of analysts is based on unique analysts, excluding revisions).

best regards,

Joost

Signature

 Profile

 Posted: 07 July 2015 07:44 PM [ Ignore ]   [ # 8 ]
Newbie
Total Posts:  1
Joined  2015-07-07
luckygirl2009 - 06 April 2013 12:56 AM

Hi Joost,

Thank you for your suggestion. I have figured out what is going on by checking the details.

Cheers,

Anna

Hi Anna,

Could you please tell me why there are more than 12 group of “ticker and fpedats” combined in summary file? and what should i do with it if i want to get analyst coverage every year? Thanks a lot!

Best regards,
Janice

 Profile