An Application of Extreme Value Theory to U.S. Movie Box Office Returns

In this paper we use extreme value theory to model the U.S. movie box-office returns, using weekly data for the period January 1982 to September 2006. The Peaks over Threshold method is used to fit the Generalized Pareto Distribution to the tails of the distributions of both positive weekly returns, and negative returns. Tail risk measures such as value-at-risk and expected shortfall are computed using likelihood and profile likelihood methods. These measures can be used as indicators for the film distributors in the preparation of movie prints, or as references for actual or potential investors in the movie industry.