Misreading the Green: Study of Tiger’s Toll Misses
Attempts to Quantify the Economic Impact of News Events Often Amount to Long Shots; the Traps of Interpreting Market Reaction…
Playing among these analytical sand traps, the Woods study added a few of its own little bogeys in the initial rounds. In its first version, the study said it included stock movements through Dec. 17, though it instead ended the day before. It also included American Express as a sponsor, though the company dropped its Woods relationship in 2007.
 And Gatorade, which has sponsored Mr. Woods, is owned by PepsiCo, which suffered a big drop in share price. But that decline coincided with the company’s downward revision of its forecast for revenue and profit.
 Typically, researchers like to reduce the chances that any numerical changes — in this case, in share price — are due to chance. In general, the threshold is 5%; anything greater is too chancy. But in the Woods study, there was more than a 5% probability that the stock prices moved as they did by chance, and not because of a golfer’s peccadilloes, meaning the initial finding wasn’t statistically significant.
For the record, this blog pointed out #1, and Felix pointed out #2. That article should have mentioned how quickly the financial blogosphere reacted to giving important feedback, because frankly, the financial blogosphere rocks.
Here’s the revised edition of the paper, addressing these critiques.
Also that was me critiquing it as if an analyst brought it to me. If I was critiquing it as if I was referring a finance journal, I would tell them to research the so-called January Effect and comment. It’s bad empirical finance form to look at a handful of stocks for just December, particularly mid-month. A rational agent may sell stocks in December to take advantage of capital-gains losses for tax purposes and buy them back in January, especially for small-cap stocks (which would qualify the other Big Two sponsors besides PepsiCo). This causes a financial ‘anomaly’ where prices go up in January, which markets are aware of, so they do all kinds of weird bidding games on poor performing stocks in the month of December. This has largely disappeared from markets (an argument for “adapative markets” as opposed to “efficient markets”), but this has hardly been an ordinary year for the stock market.
This may bias against them finding the results that they did – small stocks outperforming large stocks, which they can use to bolster their results. But it has been a while since I’ve seen the frontier of that research. Ultimately the question is whether or not these results stick around post-January, which we don’t know yet – and of course this study will now influence the results of that question (performativity!).
They should also comment on the bankruptcy issues of EA, perhaps even looking at it as a case study. EA’s is having bankruptcy issues, and they have a direct financial interest in Tiger from their video game (which is a level more of a operational risk exposure than simply endorsements). They would probably be the ones that could credibly claim a loss – grab a quick Merton Model of Credit Risk and see if a projected decreased in Tiger Woods video games could have push it below a bankruptcy threshold, perhaps?
Is blogging the new first seminar critique?
UPDATE: Mentioned here from the Numbers Guy’s blog. All these arguments make me miss working with empirical equities stuff….