LAUSR

Abstract The treatment of endogenous inputs and outputs in the estimation of multiple–output distance functions is an important issue that has been largely ignored in the hotel and tourism literature. It is well known that directional distance functions with endogeneity are difficult to estimate using maximum likelihood-based methods since the specification of additional identifiable equations for each endogenous variable is highly challenging. In this paper, we propose a limited-information approach to models of this type using a flexible reduced form for the endogenous variables. This approach allows us to easily estimate technical efficiency and does not rely on information about input or output prices, which are typically unavailable. We employ Bayesian methods and propose and use novel posterior measures of weakness and relevance of instruments in an application involving data from major US hotels. We show that controlling for endogeneity has a substantial impact for relative and absolute hotel performance.