Thursday, October 25, 2012

Buy on Earnings Guidance

The standard finding is that a stock’s cumulative abnormal returns drifts in the direction of an earnings surprise for several weeks following an earnings announcement. Often this was presented as a way to make easy money shorting the negative surprises, going long the positive surprises. Now it seems one leg might have the wrong sign. This highlights one problem with trying to be consistent with all the facts: they aren't all true.

There's a newly published paper by Das, Kim and Patro On the Anomalous Stock Price Response to Management Earnings Forecasts. From the abstract:
In the post-announcement period, we find a significant upward price drift for both good news forecasts and bad news forecasts.
Here's the result in a nutshell, x=0 is time of earnings announcement,  the blue line negative surprises, red positive, going out 30 days

A free early version is here.


Anonymous said...

The real surprise is the rampant insider trading and/or leaking of information in the 10 days leading up the the announcement.

Anonymous said...

it is amazing how much time academics spend on stupid stuff like this. I would hate to see what the humanities professors are up to. The only knowledge content in this is that it might be a pattern that can be gamed for a profit - until too many people catch on. There is no benefit outside of profit so why put it in a journal and pretend it is signification in an intellectual sense?

Anonymous said...

The standard result is with respect to reported earnings, this paper is about earnings forecasts. While a reason for different price responses to earnings vs. earnings forecasts may not be obvious, this paper doesn't invalidate the earlier findings.

Anonymous said...

You put it in a journal in the hopes that it inspires AQR or some other quant manager to hire you to do the same historical regressions/backtesting but pay you 4x as much?

Dope Threat said...

While this indeed is another test of post-announcement price drift, this is a test of the drift in the very short window after a company pre-announces its quarter ahead of the expected earnings announcement date, rather than the longer term window over subsequent months.

I don’t believe the findings here regarding negative surprises necessarily contradict previous observations. We should expect behavioural overreaction in the short term (0 to 22 trading days), and underreaction in the intermediate term.

Empirically, there is overreaction to short-term, vivid, recent and negative newsflow, and this tends to correct itself to some degree. This paper provides further evidence. The post earnings announcement drift (PEAD) to which you refer is a driver of longer term momentum is a more intermediate term phenomenon. “Bad News Travels Slowly” by Hong Lim and Stein, and a host of other work, documents this longer term drift as an “underreaction” and it helps to explain long term momentum. But that isn’t what they are testing here. The findings here, at least as they relate to negative surprises, are not inconsistent with earlier work on overreaction.

In fact, I wouldn’t be surprised if that by extending the horizon of this data set out a few more months, we would start to see the PEAD on the negative surprises that you would expect.

However, the behaviour of positive surprises in the short term is somewhat unexpected. In the very short term (within one month of trading days) in the window where overreaction dominates, we shouldn’t expect to see short term price upward price drift after the initial reaction. Yes, we should expect it in later periods as the company reports further evidence that things are getting better, thus causing the analysts to continue updating and revising their anchored, long-term beliefs about the company – but for there to be no overreaction to good news and subsequent correction in the short term is indeed a surprise.

Moreover, as a side note, I do wonder if there is something wrong with the dates used as starting points. The pre-announcement price drift in bad news cases is markedly stronger than you would expect to see – especially as this is a study of surprise profit warnings. Sure, there may be rumblings ahead of a profit warning that puts a little pressure on things, but not to the degree that the chart on page 33 indicates, and certainly not the day before the warning. Thus, the starting point for the PEAD could be too low – and conclusions about its upward drift thereafter biased. This timing issue could be creating a problem for the “good news” cases as well.