摘要: Existing quantitative migration studies are mainly at the inter-region or inter-province level for lacking of detailed geo-
referenced migration data. Meanwhile, few of them integrate explorative spatial data analysis and spatial regression model into migration
analysis. Based on aggregated registered floating population data from 2005 to 2008, the phenomena that population floating to Yiwu
City in Zhejiang Province is analyzed at the provincial and county levels. The spatial layout of Yiwu′s pull forces is proved as a V-shaped
pattern excluding Sichuan Province based on map visualization method. Using the migration ratio in 2007 as an explanatory variable, two
models are compared using ordinary least square, spatial error model and spatial lag model methods for county-level data in Jiangxi and
Anhui provinces. The model with migration stock provides an improved fitting over the model without migration stock according to the
model fitting results. The floating population flocking into Yiwu City from Jiangxi is determined mostly by migration stock while the
determinant factors are migration stock and distance to Yiwu City for Anhui. The distance-decay effect is true for migration flow from
Anhui to Yiwu City while the distance rule is not confirmed in Jiangxi with the best fitting model. The correlation between per capita net
income of rural labor forces and migration ratio is not significant in Jiangxi and significant but at the 0.1 level only in Anhui. Further
analysis shows that the distance, income and man-land ratio are important factors to explain population floating at earlier stage.
However, as the dynamic population floating process evolves, the determinant factor would be migration stock.