• A free mini-guide on how to blog a conference in detail, by Ethan Zuckerman and Bruno Giussani.

Search LoIP

  • Web LoIP

Get LoIP per email

  • Enter your email address:


Books by Bruno Giussani

« links for 2007-05-06 | Main | links for 2007-05-07 »

May 07, 2007


What is this paper really saying? They seem just to show that badly-performing companies on average don't get worse nor well-performing ones better, after some fixed point. On the face it looks like a usual regression to the mean performance -- where's the evidence that the magazine articles had anything to do with anything?

Indeed, you read the piece upside down. They don't say (and not even suggest) that the magazine covers had anything to do with anything. The research was not about companies performance, but about magazine covers, and what their data show is that (business) journalists often "get" a story when it's already over.

I'm not sure I follow. I see for example in the case of the positive categories supporting evidence in the data given over the period of days 1-21, but there is no claim made of significance for the data in this time period, whereas those 5-day returns that are significant don't support the conclusion. (I also see that what's said in the Conclusion of the paper is much weaker than the claims in the Abstract.)

On the other hand, for the longer-term statistics in Table 3 (and again in Table 4 for the significant statistics), the returns for the companies get relatively closer in the successive time periods 1-125, 1-250, and 1-500. But this is just what one would expect if the companies were merely regressing to the mean -- if we drew 550 random companies and sorted them by past performances over some 500 days and watched for 500 more days, we'd expect qualitatively the same table. I don't see how the authors can conclude that this data says anything about magazine covers if the same data would come after any event at all.

Either I had a much too well non-watered down evening of celebration, or I really do not get this.
I find it however interesting that there is some statistical evidence to back up an observation of mine about which companies make the front cover and how that then pans out in the real business. Is then the take home lesson that when a company makes it to the cover, that means -statistically speaking- that it finds itself at a turning point?
I am actually left thinking if making it to the front page of these magazines means anything at all. How about the companies that stayed out of the stage lights of magazine covers?

I will happily leave the statistical speculations to statisticians (which I'm not). What the study tells me, as a journalist, is just that too often companies appearing on biz magazine covers in a positive manner tend to be at the end of a good run; in a negative manner, at the end of a bad run. Otherwise said, journalists too often "get" a story too late. (I'm not saying that this happens all the time, of course).

Magazine covers have long been used as contrarian indicators by quote-unquote contrarians. It's simply a statistical enomoly that once something becomes well known and accepted in the market, it's time to change to something different. The most common magazines used are Newsweek, the Economist and Fortune. If you had of bought bank stocks when the cover of many newsmagazines said "Big Banks Too Big to Fail" or the Economist magazine from 2003 that stated "The end of the oil age." You would have grossly underperformed the market.

It's jsut the idea that once something becomes mainstream it's on its way out already. Not a cut to journalist who are reporting the news, other contrarian indicators are Analyst Sentiments and Put-to-call ratios. "Bull markets climb a wall of worry"

The comments to this entry are closed.

Upcoming conferences