One explanation for the comparatively lower quality of movie sequels is selection bias, known in personnel economics as the Peter principle (Lazear, 2004). Only abnormally successful movies are selected for a sequel. Another explanation is a deterministic depreciation in quality due to the decline in the novelty of the sequel's characters and storyline. Both explanations predict that, relative to the original, the sequel's performance will revert towards the mean. We develop a structural model to decompose the two explanations, and estimate its parameters using detailed data on 306 franchise films and 2,823 non-franchise films between 1995 and 2014. Parameter estimates provide evidence of selection bias for action & adventure and horror movies, and evidence of a deterministic decline in quality for comedies.