Pablo Reyes-Polanco, Sebastián Astroza, David Palma, Juan Antonio Carrasco, Sergio Jara-Díaz
Abstract
Current microeconomic time-use models incorporate committed time as a relevant explanatory variable, which is the total amount of time spent in activities to which an individual must assign a minimum amount of time and cannot or is not willing to assign more. Committed time includes mandatory transport and limits the total time available to be allocated to other activities such as work and leisure, and, therefore, has a direct impact on time assignment and the values of time. Committed activities are often assumed to be known by the modeller and common to all individuals. However, the differences in preferences between individuals make it necessary to consider this aspect as an endogenous decision to understand the whole picture of time allocation. A confirmatory latent classes approach is proposed and used for the endogenous segmentation of individuals, modelling different behaviours on committed time under a probabilistic approach. An application using nationwide data collected in 2015 in two surveys in Chile illustrates the method. Results show that time assigned to meals or sleep may be characterized both as committed and as uncommitted for different (latent) classes of individuals. Latent class models exhibit significantly better model fit and resulting values of leisure and work that differ from those obtained using the models with exogenous imposition of committed time. We conclude that the sets of committed activities for individuals with different socioeconomic characteristics can be obtained as part of the econometric procedure instead of an a-priori imposition of the modeller.
