This paper describes a study of participation in telecommuting activities in relation to spatial and non-spatial characteristics. Based on an extensive data set covering 2,097 respondents in the Netherlands, a right truncated zero-inflated Poisson regression model was estimated to study the impact of spatial and non-spatial characteristics on the frequency of telecommuting. The characteristics included in the study involve gender, age, driving license, car availability, availability of public transport pass, availability of travel cost arrangement, handicap, family structure, household income, urban density, and travel time to work. The model proposed in this paper is innovative in at least two respects. Firstly, it adds on some of the existing models in the literature, more specifically with respect to modeling the existence and interpretation of excess zeros in the data and with respect to the right truncated nature of the dependent variable, i.e. number of telecommuting days per month. Secondly, the model treats both the propensity (yes or no) and the frequency/intensity of telecommuting jointly into one model. The results show that most of the variables were significant in explaining the number of days per month a respondent engages in telecommuting activities. Furthermore, although we find evidence in our data that confirms earlier research findings (e.g. with respect to income level, car availability, etc.), we also find some conflicting results (e.g. with respect to gender). (Author/publisher)
Samenvatting