Abstract
The aim of this paper is to conceptualise the migration duration decision within the expected utility maximisation framework, and from that to derive and estimate an empirical proposition. For this purpose, the conceptual framework in Kotorri (2015) is extended where households decide to return to the home country conditional on their migration duration. In the empirical analysis, the Cox proportional hazards model is employed. This analysis is the first to investigate migration duration based on a random sample stemming from the Kosovo census of population conducted in 2011. The findings suggest rather mixed support for the household approach. The hazard to return decreases with income but not nonlinearly. The results indicate that household return migration behaviour is influenced by demographic characteristics, psychic income, and political factors.