A spatial stochastic frontier model at the P2P listing level

Abstract

This paper presents a novel approach to account for spatial effects in the estimation of the efficiency of peer-to-peer (P2P) accommodation units. Specifically, in a stochastic frontier approach, it analyses the correlation effects (spatial dependence of inputs and outputs) on the frontier itself, the noise term (e.g., unobserved but spatially correlated variables) and the inefficiency term (e.g., agglomeration or competition effects). To do so, a spatial efficiency model recently developed in the econometric literature is used. From this model, direct and indirect marginal effects on inefficiency for each listing can be calculated for the inputs and environmental factors. Geographical patterns of the spatial effects of the inputs and determinants among listings can thus be detected, providing researchers and practitioners with granular geographical information on the spatial heterogeneity of efficiency in the sample. The model was applied to the P2P lodging market in the Canary Islands, Spain.