Replicating rockets and feathers

This paper revisits the literature of asymmetric adjustment in gasoline and diesel prices, also known as the ‘rockets and feathers' hypothesis, to consider the issue of the replication of empirical research. We examine the notion of replication versus robustness proposed by Clemens (2017) and add to the literature with a further review of recent and historic work on replication in economics and other disciplines. We then focus on the rockets and feathers literature, finding that the majority of empirical work performs robustness checks rather than replication of earlier papers. We perform two contrasting replication case studies motivated by the ideas of misspecification analysis, dynamic specification, mark-up, pass-through and asymmetric adjustment. In the first case study we find that results are both replicable and robust, even when data specifications are not identical. However, in the second case study we find that the results using the original sample are overturned when reanalysing the problem using an improved model specification. Furthermore, when extending the sample with more recent data and using a more sophisticated method, asymmetry is detected in both petrol and diesel pricing; different to the findings of the original study. Particular care must be taken in future rockets and feathers replications with regard to model specification and methodology.