The behavioral economics of earnings estimates

By Felix Salmon
April 5, 2010
Tom Brakke has a great explanation of how and why earnings estimates move, based on a hypothetical analyst who wants to up his estimate to $1.65 while the consensus ranges from $1.22 to $1.35.


" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Tom Brakke has a great explanation of how and why earnings estimates move, based on a hypothetical analyst who wants to up his estimate to $1.65 while the consensus ranges from $1.22 to $1.35.

Often what is easy to do in a situation like this is to think, “You know, if I use $1.45, I would be well above the high range. I’d still get credit for being right. There’s no sense being too aggressive on this.” And so begins a game of behavioral leapfrog. The other analysts covering the company will immediately notice the new number and think that they too should take another look. They will be asked questions about the estimate you published. Their reviews may discover some of the same improvements that you spotted, and their estimates will move higher, sometimes by a little and sometimes by a lot. Someone who can see that same $1.65 potential as you did will decide to top you with $1.50 or $1.55.

This can go on for quite awhile.

The lesson here is not to pay much attention to forward p/e ratios, since the denominator in such ratios is the consensus earnings estimate, and the consensus earnings estimate is a moving target which pretty much by definition is behind the curve.

And of course this also helps explain why “whisper numbers” are so much more important than official consensus earnings estimates — the estimates don’t necessarily reflect what the analysts really think.

(Via AR)

2 comments

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

1. This isn’t really behavioural economics though is it? It’s a perfectly rational response to the incentives as set up – it’s game theory.

2. Your “lesson” doesn’t follow at all from this description – consensus earnings on this model would be a smoothed version of the “true” average of analysts’ views, but that could just mean it was less volatile and in any case certainly wouldn’t mean it was uninformative.

3. There is, in fact, no sense in being too aggressive. If you put any weight on the views of other analysts covering the same stock (which would be the general case unless you believed them to be totally uninformative), you would always shade your “best guess” back in the direction of consensus – Bayes’ Rule even tells you how much to shade it.

Posted by dsquared | Report as abusive

Large cap earnings are going to rock solid next and will move the stock market to new highs

Posted by Storyburn_com | Report as abusive