A multi-inflated hurdle regression model for the total number of overnight stays of Italian tourists in the years of the economic recession

This contribution is concerned with the tourism behavior of Italian residents in the period covering the last economic recession: it investigates whether and how the economic recession has affected the total number of overnight stays in a quarter by modelling it through a hurdle multi-inflated regression model. The assumptions of the hurdle model are consistent with the phenomenon under study, in which firstly a person decides whether to have a vacation trip and then, conditionally to a positive decision, he decides the number of overnight stays. Therefore the binary process concerning the decision to have at least a vacation in a given quarter is modelled through a logit regression model. Then the total number of overnight stays, for those who had at least a vacation, is modelled. Since this variable is naturally concentrated on some specific values (like 2, 6, 7, 14, 20 nights), we use a Multi-Inflated Truncated Negative Binomial regression model in order to control for this peculiar peaks. We analyse data from the quarterly survey on Trips and Holidays in Italy and Abroad carried out by the Italian National Institute of Statistics, in the period 2004 to 2013. The empirical results show that socio-economic characteristics of the individuals and of their families have an important effect on their tourism participation; that these factors, together with some trips-related characteristics, affect the total number of overnight stays; and that the economic recession impacted negatively on both aspects of tourism behavior.