Internal migration around the world: comparing distance travelled and its frictional effect

This paper examines how internal migration distance and its frictional effect vary between countries. Such comparisons are hampered by differences in the number and configuration of spatial units for which data are available - the modifiable area unit problem (MAUP). We use the flexible aggregation routines embedded in the IMAGE Studio, a bespoke software platform which incorporates a spatial interaction model, to elucidate these scale and pattern effects in a set of countries for which finely grained origin-destination matrices are available. We model the relationship between mean migration distance and mean area size and we show that the frictional effect of distance remains remarkably stable across spatial scale, except where zones have small populations and are poorly connected. This stability allows robust comparisons between countries even though zonal systems differ. We find that mean migration distances vary widely, being highest in large, low-density countries and positively associated with urbanisation, HDI and GDP per capita. This suggests a positive link between development and migration distance, paralleling that between development and migration intensity. We find less variation in the beta parameter that measures distance friction but identify clear spatial divisions between more developed countries, with lower friction in larger, less dense countries undergoing rapid population growth.