Historical deposition influence in residential location choice: a distance-based GEV model for spatial correlation

Most of the existing literature on residential location choice is cross-sectional and assumes no history effect in people’s residential location preferences. However, human beings are constantly adjusting themselves to the local environment and thus it seems reasonable to assume that preferences are formed over time. We hypothesize a historical deposition effect, which states that people’s location preferences are likely to be a function of the attributes of where they lived before. Furthermore, this historical deposition influence can interact with lifecycle, such that it can become less important or insignificant during a life stage, like parenthood. Our study tests the existence of the historical deposition effect and analyzes its interaction with lifecycle. The study results support the historical deposition influence. Furthermore, we show that the historical deposition effect can be subdued by parenthood. The study also makes a methodological contribution by developing a distance-based GEV (generalized extreme value) model to account for spatial correlation. Spatial correlation is rarely treated in existing studies. The few existing studies typically assume a constant spatial correlation coefficient for adjacent zones. The distance-based GEV model relaxes the constant coefficient assumption and allows a distance-based correlation between nonadjacent zones. The results confirm this distance-based spatial correlation.