Earnings Projections

There's been a lot of ink spilled about the absurdity of price targets on stocks.  An analyst calling for Google $2000 prompted the latest round of uproar.  There is, however, comparatively less interest in the other half of the analyst's job, coming up with an earnings estimate.  Sure somtimes it's perfectly reasonable to plug a few growth assumptions into a model, pull some levers, and see what the silicon oracle spits out, hence: ''We're upping our FY 2007 estimate for GE from $2.18 to $2.25/share".  Other times, however, this sytem makes no sense.  Today, I took a brief look at regional carrier ExpressJet (NYSE: XJT).  Based on some reasonable assumptions, which I seem to share with Wall St.'s Brightest, the company may make around $1.70-$1.80/share over the next 12 months.  This is really weird in that it gives the company a forward PE of around 4.  This reveals the fact that my forward estimate means nothing.  Investors obviously think there is a reasonable liklihood that their main partner, Continental Airlines (NYSE: CAL) will face major trouble in '06, and that ExpressJet will get slammed.  On the other hand, take a look at their quarterly report; if energy prices abate in '06, maybe the company will make over $2.00/share.  Fact is, no analyst can know what  ExpressJet will earn this year, and even though $1.70 may be the most likely, the range of possibilities is so wide, that it seems silly to just use one number.  It would be better to come up with a range of possible scenarios, and attempt to assign a probability to each one.  However, it won't play very well on Squawk Box, at 7:00 AM, to say "We're assigning an 18% probability to a loss of $.45/share next year, a 35% probability of them making $.75, and 45% probability of $1.70, and a 2% probability to earnings of $2.50".  I'm afraid Mark Haines & Co. would have dozed off by the time the word We're was uttered.

Comments

[Andrew Schmitt]

First link in post is bad

[David]

As to the question, why not use a probability distribution, I think the snooze factor answers it for why the sell side doesn't do it, but I've also been struck by how rare this sort of probabilistic approach is on the buy side. I've read hundreds of investment write-ups, and they almost never try to calculate expected values. Presumably these people are more interested in getting it right than in entertaining others, so why wouldn't they just grit their teeth and crank out a distribution? Of course, they want to determine a company's value, not simply estimate one year's earnings, and it might be that the valuation problem is just too complicated (high-dimensional) to tackle this way. But the right approach being too hard doesn't make an unjustified approach (such as using the "most likely" scenario as the expected value) any less wrong.

[hristop]

They are all so very boring though. To be sure all my custom fonts were seen by everyone with any browser, I made images with the custom font.

[casino bonukset]

Peut-être qu'ils ne poster des articles plus intéressants et des sujets ici.

[Mulberry Alexa]

Thanks, I'm going to have nightmares tonight.