Migration-prone and migration-averse places. Path dependence in long-term migration to the US

Does past migration beget future migration? Do migrants from different backgrounds, origins and ethnicities, and separated by several generations always settle – in a path dependent way – in the same places? Is there a permanent separation between migration-prone and migration-averse areas? This paper examines whether that is the case by looking at the settlement patterns of two very different migration waves, that of Europeans at the end of the 19th and early 20th centuries and that of Latin Americans between the 1960s and the early 21st century. Using Census data aggregated at county level, we track the settlement pattern of migrants and assess the extent to which the first mass migration wave has determined the later settlement pattern of Latin American migrants to the US. The analysis, conducted using ordinary least squares, instrumental variable and panel data estimation techniques, shows that past US migration patterns create a path dependence that has consistently affected the geography of future migration waves. Recent Latin American migrants have flocked, once other factors are controlled for, to the same migration prone US counties where their European predecessors settled, in spite of the very different nature of both migration waves and a time gap of three to five generations.