Forecasting Model for Box-Office Revenue of Motion Pictures

The main objective of the paper is to develop an econometric model to forecast box-office revenue of motion pictures. Considering the importance of forecasting demand for new products, marketing researchers have developed various demand forecasting models. However, these models forecast future demands based on either several months of initial sales data after new product introduction or the survey data on customer purchase intention. Different from these models, we forecast future demands of new products based on the analysis of historical sales patterns of similar products. Even though we apply our model to the case of revenue forecasting for motion picture industry, it can easily be applied to forecast future demands for several other industries such as books, music albums, videos and pharmaceutical products. We forecast box-office revenue of a motion picture given its characteristics such as genre, reviews by movie critics, star power of main actors/actresses, directors, ratings, advertising expenditure and so on. Our results show that modeling weekly sales (compared to directly modeling total sales) improves forecasting accuracy significantly. We also show that some movie characteristics are useful to forecast the box-office revenue for the first week while other movie characteristics provide information on forecasting the decaying rate of weekly revenues