Sheep
October 19, 2011 5 Comments
In economic news of the day:
The credit markets had been going wider, and then some headline appeared on Bloomberg saying that France and Germany were going to agree to a 2-trillion-Euro something-or-other vis-a-vis the EFSF. (I didn’t figure out what the story meant but I’m guessing their plan is just, instead of recapitalizing their banks directly, which would inflate currency and harm their credit ratings, what they were planning to do instead was to launder the exact same liability through the EFSF – or in other words, become swap counterparties to the EFSF, which is a giant CDO – and hope no one noticed the new giant CDO on their balance sheets.) Anyway, it doesn’t really matter what the details are. What matters is that traders saw ’2 trillion’ on their screens and the markets rallied considerably within 10 minutes. Nor did it matter about an hour or two later when cold water was thrown on the whole ’2 trillion’ story which, in the end, doesn’t seem to have really meant anything at all. Markets finished up.
The other thing that happened came at the end of the day, when Apple reported earnings and missed. ‘Missed’, in this context, means that the earnings-per-share number they announced was smaller than some other number (the estimate/forecast/prediction) that traders had seen on their screens. Where did this other number come from? What is it based on? Is it valid? Nobody knows and nobody cares. The estimate was X but the announced EPS was Y<X. So traders said sell and AAPL stock declined.
This is closely related to the weekly and monthly rituals when economic data (e.g., non-farm payrolls) are announced. Say the non-farm payrolls estimate was 260k and it turns out to be 230k. People will groan and the market will go down. Say the estimate was 210k and it turns out to be 230k. People will cheer and the market will go up. Where does this estimate come from? No one knows. It's the estimate. Actual less than estimate = bad.
You know maybe the OWS people have it correct after all, markets just don't work.
Maybe people who were buying at higher prices figured the analysts are probably about right. Then when they realised they weren’t, they became willing to sell the same securities around or below the prices they bought at, causing downward pressure on prices.
Why is that so strange?
There would be nothing strange about it if the ‘figured the analysts are probably about right’ part were based on anything substantial. As things are, you and I both know it’s based on nothing. It’s really easy to ‘figure’ some analysts you’ve never met using methods you haven’t checked to analyze data you have no idea about, are ‘about right’. But in practice that’s the same things as just believing numbers you see on your screen/TV.
You can read my comment to be a complaint that the market doesn’t put enough error-bars around analysts’ estimates. If they did, an under/overshoot of 10% wouldn’t be treated by the market as news, thus wouldn’t affect the price. But it clearly is, because it does.
“As things are, you and I both know it’s based on nothing.”
I think nothing is a bit of an exaggeration… The analysts, fallibility notwithstanding, seem to get remarkably close. Today’s first-claim unemployment report was 403,000 and they were expecting 400,000. Or something like that. If they were off by orders of magnitude all the time then I doubt traders would be figuring those estimates into their prices, but they’re generally close enough that they do. Of course I can’t tell you why the level of confidence they appear to put in the numbers seems to always be so close that any deviation results in buying or selling. It does seem pretty silly sometimes.
Estimates were 400k and actual was 403k. This may be impressive until you realize that prior was 404k. Just taking the prior and rounding to nearest 10k would have been just as good.
If you look at history of these estimates you’lll see two main effects which are (1) a lag and (2) sometimes,an overshoot. In other words these “analysts’ estimates” are basicaly indistinguishable from what you’d get if you took the stats from the past 3months, say, and projected them forward using a spline. Now that’s fine, of course, as long as you interpret the resulting stat with a sufficient error bar. The market doesn’t seem to, though, and that was my point.
Best,
Well, eventually the market figures out the GM is bankrupt. “Eventually” can be – as you well know – a long time.
Same thing for the Euro.
Of course, Keynes had it right: in the long run, we’re all dead.