46th Congress of the European Regional Science Association: "Enlargement, Southern Europe and the Mediterranean", August 30th - September 3rd, 2006, Volos, Greece
An interesting, but yet largely unstudied question concerns migration behaviour in different labour markets from the point of view of duration. How long are residenece spells? Why do some migrants move quite soon after the move, while others stay for a long time? How do personal and family characteristics account for differences in residence spells and repeat migration? Does earlier migration experience or other prior activity or experience explain for this? What is the role of labour market conditions and other region-specific factors? This paper deals with migration behaviour in Finland in 1987-2002. Migration will be analyzed in a duration-model context where movement is seen as terminating an observed residence spell. The data set is based on a Longitudinal Census File and the Longitudinal Employment Statistics File constructed by Statistics Finland. Since 1987, the two basic files are updated annually. These two register-based data sets, together with some other registers, provide panel data on each resident of Finland, from which a 7 per cent random sample bas been taken for this study. The longitudinal data allow us to observe changes of residence and the length of spells remaining in new location. The data set is very rich including hundreds and hundreds of variables for each year. The individual level panel data will be transformed and pooled into a sample of residence spells that began during the period 1988-2001. The maximum of the observed duration of possibly right-censored residence spells if fourteen years. The duration data is interval censored (grouped) – the status of residence spell is only observed at the end of each year. Therefore, discrete time representation for the hazard rate is used. Duration dependence is measured with a set of covariates. Depending on the character of the covariate, the measurement time varies among the variables: some covariates such as sex or parental variables do not change in time, some are measured on year before the residence spell or at the outset of the spell, and the rest such as age, level of education, marital status, family size and regional unemployment are treated as time-varying.