Dynamic quantile stochastic frontier models

This paper introduces the concept of dynamic quantile regression to the context of stochastic frontier models. We develop a Dynamic Quantile Stochastic Frontier (DQSF) in a Bayesian framework to take into account possible shifts of production (i.e. outputs) over time. Not only does the model provide inefficiency measures by various quantiles but also controls for endogeneity and treats the quantile as a parameter and derives its marginal posterior distribution. The model also adopts a more general process for the time-varying parameters of the DQSF, where heterogeneity and dynamics are conveniently modeled using a panel vector autoregressive model. We test the model on a sample of US hotels.