Circular Migration Measurement

With two counts taken from a migrant's longitudinal history: the number of times they have exited the host country and the cumulative number of years they have spent away across those spells. A migrant with at least one exit and a subsequent return is classed as circular or repeat; zero exits means a permanent stayer, and a single exit without return is a permanent returner. These counts capture both how often a migrant circulates and how much time they spend away, turning circularity into observable, modelable quantities rather than a vague descriptor.

Because the number of exits is a non-negative integer count that is highly skewed and concentrated at zero: most migrants exit rarely while a few circulate many times. Poisson regression is the natural model for such counts, and the negative-binomial extension handles the overdispersion that arises when the variance of exits exceeds the mean, which is the usual case. When there is a large excess of zeros from non-circular migrants, hurdle or zero-inflated versions separate whether someone circulates at all from how often they do. Linear regression would mis-specify the distribution and give biased, inefficient estimates.

Circular migration measurement is about the rhythm of repeated movement — counting how many times someone leaves and returns and how long they stay away — whereas return migration analysis focuses on the nature and success of a return, such as whether it was prepared and resourced. The two are complementary: a return is one event within a possibly circular career. Constant and Zimmermann measure the recurrence, treating permanent return as just one of several mobility types alongside circular and permanent-stay; a return-migration framework drills into the preparedness and reintegration of that return event itself. Studying circularity needs longitudinal crossing data; studying a return needs information on the returnee's resources and readiness.