Bimodal length of stay in the accommodation sharing economy

Abstract

This paper analyses the determinants of tourist length of stay in sharing accommodation. Methodologically, the paper proposes a new approach to account for the observed bimodality of length of stay. More specifically, the determinants of length of stay are analyzed by modeling both the conditional mean and conditional modal frequencies, which represent shorter and longer tourist stays, unlike previous contributions on length of stay where modeling has been based exclusively on the conditional mean. The empirical analysis examines the length of stay in sharing accommodation lodgings (Airbnb and Homeaway) of tourists visiting the Canary Islands (Spain) before and after the outbreak of the COVID-19 pandemic. The results show a bimodal distribution of length of stay: 1 and 7 nights in the pre-pandemic period, reduced to 1 and 2 nights in the pandemic period. The results also indicate that the model approach followed in this paper is preferable to other previous bimodality models in terms of estimation simplicity and fit. In addition, the model allows an analysis of the determinants of length of stay and differentiation of the influence of each determinant on shorter and longer stays.