Studies continue to show that revisions are informative, causing mean announcement-returns of 1% to 2.5% for up revisions and -1% to -2.5% for down revisions. The total value of these announcement-returns points to the release of enormous information value by revisions, given their great frequency.
But they note that many of the revisions occur overnight, and they often accompany company information, such as a new product release or earnings guidance. So, focusing only on analyst revisions that occur intraday, and focusing on a 40 minute window around the forecast revision, generates a measly 0.05% price move.
So, ignore all those talking heads, and glossy binders by your broker? No! It turns out, there is information in analyst forecasts, specifically, shorting your broker. In Long-Term Earnings Growth Forecasts, Limited Attention, and Return Predictability by Da and Warachka, they find that if you take the long-run earnings growth and divide by the recent earnings growth, you get a straightforward metric of optimism: higher is more optimistic. It turns out, the highest decile has a 4% annualized lower risk-adjusted return than the lowest decile; more optimistic stocks have lower returns.
A simple explanation (see here) is that mutual funds "herd" (trade together) into stocks with consensus analyst upgrades and (especially) herd out of stocks with consensus downgrades. Stronger fund herding occurs when analyst recommendation revisions are more unanimous.
However, there is some information that analyst recommendations are more useful at the industry level. Portfolios long in industries about which analysts are optimistic and short in industries about which analysts are pessimistic generate significant abnormal returns (see here).
So, listen to your broker's advice on sectors and stocks, just remember to short the stock picks.
This fits with other research.
One reason boils down to how analysts spend their time. Long-term growth forecasts get very little attention and are late to be revised. Part of the problem is that the "buyers" of the information (insitutional investors) don't demand real analysis and justification for the long-term estimates, focusing greater effort on the near-term results. Consequently, it becomes mostly extrapolation of recent trends.
All this should be a big red flag for lovers of the PEG ratio: the "G" in the formula isn't very good.
so it seems the mean reversion and momentum work at the same time and in the same respect. thank god behavioral finance can explain that... ; )
ps not related to Tom (at least I don't think)
Post a Comment