« A New Workout | Main | "Got Milk?" Not early humans. »

Seeing things at Disney before they got rid of Eisner

February 28, 2007 08:02 AM

I wrote this as an Op Ed to the LA Times back in 2002. They declined to publish it and I can see why. It was not so long after this that Eisner was removed as Chairman. But, the malaise still lingers and, more troubling, financial markets still over react to forecasts of earnings and "earnings disappointments." They ought to fire the forecasters.

As I write this, Disney is downgrading its quarterly earnings forecasts because its animated motion picture Treasure Planet opened to lower than expected revenues. This event reveals the depth of the troubles at Disney and more---revised earnings estimates and tumbling stock prices have become a common occurrence in financial markets. In light of recent revelations of the errors and misdeeds of financial analysts, I have to ask: when did a company like Disney and when did financial markets generally adopt the belief that financial forecasts are infallible?

In science when a model predicts incorrectly you get rid of the model and look for a better one. And, you don't allow yourself to be fooled by randomness. What Disney did was shoot their movie rather than their financial analysts when they got the forecast wrong. It isn't the movie's fault the forecasters got it wrong and every forecast has an error band around it that should be acknowledged. But Disney chose to announce their forecasting error as a ``disappointment'' in Treasure Planet's opening gross, elevating the model to infallibility and ignoring randomness. The irony is that what began as a forecast error may become a self-fulfilling prophesy if Disney's actions make people believe Disney doesn't think the movie is any good.

The syllogism is strange. Let me see if I can get it down right. First, the accountants or forecasters at Disney predict opening revenues and get it wrong. Then, rather than acknowledging that the model was wrong, they say they are ``dissappointed'' in Treasure Planet's opening. This shifts the blame from the model, where it belongs, to the movie, where it doesn't. Then they use this model (that got it wrong at the opening) to forecast revenue far into the future, predicting all of Treasure Planet's theater revenues through its run, compounding the initial error exponentially. This series of errors then becomes the basis for revising Disney's expected earnings downward.


This is troubling because these kinds of mistakes are made at other studios too and in financial markets in general. First, financial models are not very good at forecasting earnings and they become worse the farther into the future they project. Second, financial variables are volatile. This means that the models will often be wrong because of randomness rather than changing fundamentals. Third, financial models fail miserably when it comes to forecasting motion picture revenues because even God Almighty doesn't know what a motion picture will gross.

On these grounds, there is little basis for Disney's announcement that they were revising their earnings estimates downward and less reason for the price of its stock to move. Since the stock price is the discounted present value of a company's future earnings, a random variation or forecasting error for one quarter's earnings should have little effect on the share price. This could only happen revised earnings estimates for one quarter significantly altered expectations over a long future. But, Bayes theorem tells us that a random variation in one quarter's earnings shouldn't carry much weight. This fundamental analysis suggests there is altogether too much reaction in financial markets to ``disappointing news.''

Where Disney went wrong is in having too much faith in their financial analysts. No one knows what a movie will gross. My research (Hollywood Economics: Chaos in the Film Industry) shows (and every movie fan knows) that motion picture revenues are not forecastable; the forecast error is infinite. There is no correlation between opening revenues and total revenues for any movies that are successful. Opening revenue is only a good predictor of total revenue for movies that die in their second or third week. All successful Disney movies (most of the animated ones are) enjoy long runs so that opening revenues become a small portion of their total revenue. And, all successful movies reach a point (about four or five weeks into the run) where they separate rapidly from the pack. This bifurcation point in the mapping is a signature of chaos in the non-linear dynamics of motion picture revenues.

Given how little information is contained in a movie's opening, the weight studios place on it is troubling, particularly at Disney which is a studio where film revenues depend on long runs. Disney movies endure and do not have to open big. The difference between opening big and enduring is like the difference between a one-night stand and a long relationship.

Joseph E. Levine was the master of the one-night stand. Levine promoted his movies heavily so they opened big (pumping up the yokels it was called). By the time the audience found out it was a stinker, the movie was out of town. The movie ``road-showed'' in a series of one-night stands with the usual results: big expectations, short runs, and broken hearts. Disney movies have never been like one-night stands; they bond with the audience in a long-term relationship and do a lot of repeat business.

There is a difference between a movie that opens big and a movie that is good and enduring. If Disney is trying to make movies that open big, then its heritage is at risk. So it is troubling and, perhaps, symbolic of Disney's malaise that some bean counter there is writing off a \$140 million movie in its first week and management is falling for it. When the message is wrong, shoot the messenger not the movie. And, don't be fooled by randomness.

· The Movie Business

Comments

A similar process occurs in my industry of enterprise product development software. Before a new software project is started, the customer will commission an ROI (Return on Investment) analysis. The analysis is used by the project "champion" to justify the project to management. The stated goal is usually to determine the number of months in the future in which the project will have paid for itself. The analysts use static models (think government forecasts of the effect of a tax rate change assuming no change in claimed income) that put a monetary value on the improved product quality, reduced "headcount", and faster time-to-market the software will allow. The craziest part is that this study is frequently done by the software vendor itself!

Some projects succeed, some are canceled midway through, and some fail, but I'm certain there is no routine postmortem review of the initial ROI calculation.

I think this all points to a need in management (and perhaps human nature itself) to have the feeling of being in control, whether that control is possible or not. And maybe a devotion to the normal distribution, in spite of evidence to the contrary?

Posted by: adamsn03 [TypeKey Profile Page] at March 1, 2007 9:04 AM

Post a comment

Thanks for signing in, . Now you can comment. (sign out)

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)


Remember me?