Detecting spatial heterogeneity in the determinants of intercity migration in China

Spatial variations exist in migration patterns and the processes causing the observed patterns. Yet, far less attention has been given to the latter. Even if a few studies have engaged in spatial heterogeneity in the process of migration, the scale multiplicity is not emphasised. This research aims to address these gaps with evidence from Chinese cities. To facilitate international comparisons, we use data on intercity migration over a 5-year interval instead of the floating population. Following an elaboration of highly imbalanced migration patterns at the city scale, we detect spatial heterogeneity in the processes underlying these patterns based on 17 potential determinants in four domains (namely labour market conditions, site-specific amenities, agglomerative effects, and institutional dividends) by multiscale geographically weighted regression. The results indicate that cities' attractiveness to migrants depends on a broad spectrum of factors whose influences are location-dependent and exhibit specific patterns, including east–west, south–north, and southwest–northeast gradient patterns, as well as discrete clustered and concentric patterns. The diagnostic analysis has confirmed that the provenance of spatial-varying effects is irrelevant to the nonlinearity. These effects are also scale-sensitive, showing that the influencing scales of the labour market status and agglomeration factors tend to be larger (i.e., their effects are spatially stationary) than those relevant to amenities and policies, which vary considerably (i.e., the impacts of some factors change significantly over space). These findings regarding spatial heterogeneity deepen our understanding of migration in China and highlight the fact that there is no ‘one size fits all’ approach for government policies designed to attract migrants.