Number of Analysts from IBES
The premier research database for historical numerical analyst forecast data is IBES (or I/B/E/S ) from Thomson Reuters.
If you want the number of analysts who are covering a company the best estimate is the number of analysts who provide an earnings per share estimate (EPS) for the next (to be announced) financial year (FY1). This is the variable estimated by most analysts.
At the University of Manchester we have access to IBES through WRDS. For the number of analysts we can use the IBES summary data rather than the detailed data at the individual analyst level.
Selecting IBES – Summary History – Summary Statistics
Select dates and the company codes
Measures – EPS (Earnings Per Share)
Forecast Period Indicator – Fiscal Year 1
Identifying Information – <as required>
Other Variables – Number of Estimates
The results will be similar to these.
INTEL CP (ticker INTC, cusip 45814010) in January 2010 had 43 analysts giving an estimate for EPS of next forecast period (fpi 1). The next forecast period was the company fiscal year ending 31 December 2010 (fpedats 20101231) and the number of analysts was calculated on 14 January 20101 (statpers 20100114).
Note that the number of analysts (numest) varies monthly. IBES considers that an analysts estimate is only valid for a certain period of time. If analysts do not update or confirm their estimate within this period then it is removed from the number of estimates.
The IBES summary data is also available through Datastream.
The Datastream variable EPS1NE – Earnings Per Share Total Number of Estimates in the Mean FY1 – is the same variable that was obtained using the WRDS IBES Summary Statistics query described above.
Datastream makes it easy to retrieve EPS1NET – EPS Total Number of Estimates (including those excluded from the Mean) FY1 – in January 2010 this was 44, rather than 43 for EPS1NE.
This expands on the July 2010 post No of Analysts covering a company
Hello there. It appears that F1NE and EPS1NE lead to the same results (At least when testing with Apple’s stock). I also tried accessing the same information through Bloomberg using TOT_ANALYST_REC. The data comes fast and for a large number of companies. Unfortunately it does not correspond to the results reached to using Datastream.
I also compared my data to the Number of Brokers from Yahoo: http://finance.yahoo.com/q/ao?s=AAPL+Analyst+Opinion.
It does not correspond to the Datastream values either (nor to those from Bloomberg).
Do you have any comments on these differences.?
The I/B/E/S database has a long history (which contributes to its good research reputation). I believe that originally EPS (earnings per share) was the only recorded estimate. In recent years there was a significant expansion in the number of estimates collected and as a result there was a revision in the datatypes to make then more consistent. F1NE is an older datatype and EPS1NE is the new recommended version. However both are still legal and should return the same results.
See I/B/E/S Estimates Enhancement – Removal Of Single Character (Dec 2013) on http://extranet.datastream.com/data/IBESConsensus/Index.htm
Comments on getting different values from IBES, Bloomberg and Yahoo Finance are in reply to your other comment on this.
Hi there. I tried using TOT_ANALYST_REC in Bloomberg with the aim of finding that same number, however the results from Bloomberg using the Apple stock, differ from those coming from Datastream using EPS1NE. I also compared to the Number of Brokers for that stock by visiting Yahoo finance, the number appearing there is different from all the others…
Do you have any idea as of what could be the reason for these differences? Thank you
I think that the problem is that analyst coverage is essentially private data – analysts do not have to provide their estimates to Thomson Reuters for inclusion in I/B/E/S or Bloomberg. The IBES number of analysts figure is therefore:
EPS1NE – Number of analysts providing estimate (EPS for FY1) of those analysts who choose to be included in the IBES database
similarly for Bloomberg.
TOT_ANALYST_REC – Number of analysts providing estimate of those analysts who choose to be included in Bloomberg
There is no reason that this figure should be exactly the same. We would expect it to be comparable in that in both sources companies that are studied more such as Apple have a far higher figure.
It will be similar for Yahoo finance – my best guess that this is the number of analysts who choose to make their estimates freely available online and would therefore be lower than the figure from either Bloomberg or IBES.
Dear Mr. Greenwood
Thank you so much for your comment! I think it makes very good sense considering the numbers I find.
BR
Boris
Hi there, I appreciate your comments. Just a simple question. Using Datastream IBES, is there any information on how many analysts in total are covering firms in an specific market?
Cheers.
Felipe
Dear XiaHong. Thanks for your answer. The thing is that probably there are analysts covering more than one company. Therefore, taking the individual company estimation, there is no way to know exactly the amount of analyst covering a specific market.
Cheers!
Felipe
You can use IBES via WRDS. This can give you the codes for analysts who cover individual companies. Those codes are 6 digits unique codes for analysts. For a specific market, you might need to get a few key players in that market, obtain the codes for all the analysts who make estimates for those companies, you could then get an estimated number of analysts for that market. If you don’t have access to IBES via WRDS, you could still get total numbers of estimates for a few key players in that market (on the assumption that major players always get most attention from analysts), the highest number can be used as a proxy for the number of analysts who cover the specific market if that company is representative enough.
Hi guys…is there some way to know the date when the forecast was made by the analyst? I was looking the datastream (IBES) but I can’t find a code with this characteristics. I would like to analyse the difference between the date when the forecast was made and the date when the EPS was announced by the company. I appreciate your help!. Many thanks.
Kind Regards.
Felipe
IBES on Datastream has summary data so there is the date that the mean estimate is calculated (monthly usually between 14th – 20th)
For the date when a forecast was made you need the detailed data, which is at the level of individual analysts. At University of Manchester this is available through the WRDS platform. (On the WRDS IBES demo you can see Estimate Date in the other variables section.)
Hi Mark, thanks for your message, I really appreciate that. The thing is that I want to determine the length of time between the forecasting date and the earnings announcement date. I know that through the code EPSANCDT (IBES) I can get the EPS Actual Announce Date made by the company. Now, my concern is related to the code in IBES to get the date when the forecast was made by analysts.
Hi Felipe, The thing to remember is that the IBES data is collected by Thomson Reuters from the analysts who agree to be included. Therefore there is an “activation date” rather than a “forecast date”. This is the date that the forecast is recorded by Thomson Reuters.
IBES is made available by Thomson Reuters through a number of channels so you need to check your documentation to get the correct code.
(Several years ago I worked with a PhD student using the IBES data that was delivered by FTP. The data came with documentation describing the file format but the actual codes were defined by the SAS code we wrote to read those files.)
Many thanks Mark for your help. I’ll try to find it through IBES. I think that there is something in Thomson One, specifically in the tab detail estimates. If I have something new, I’ll let you know.
Cheers.
Felipe
Hi Mark,
I am using IBES for the first time to find number of analysts following the firms in Indian market. You post made it very simple for me. Thanks a ton!
The coverage of IBES for Indian firms prior to 2011 seems to be very limited. Would you know of any other resource where I can find basic analyst coverage details for international firms?
Thanks
Pranav
Hi Pranav,
I am no longer at the University of Manchester and my knowledge of financial databases is slowly becoming outdated. I do not know of any other resource that has greater historical international analyst coverage than IBES. Target Prices – different sources mentions Bloomberg as a possible alternative – worth investigating if you have access.
The only other thing I can suggest is to look for the sources mentioned in recent papers on analysts’ estimates.
Best of luck
Mark
Hi folks,
i know it has been a while since something has been posted here but maybe someone can still help me out. I am really struggling with finding the quarterly reported eps of a company, not the reported eps for a fiscal year. I know it sounds trivial but I cannot find any variable in Datastream that gives the same values as for example yahoo finance. Really appreciate any help whatsoever.
Kind Regards,
Moritz
Hi Moritz, you may wish to try the following datatypes in Datastream, on a yearly frequency:
Hi Phil,
Thanks for the quick response. I really appreciate it. As a matter of fact, I already found these an hour ago but thanks anyway.
Best regards
Moritz
Hi folks,
I have another question.
If I want to have the reported quarterly earnings of a company and the quarterly earnings consensus forecasts for a specific, which variables in Datastream should I choose?
Thanks
Moritz