Modelling population density over time: how spatial distance matters

Abstract

This study provides an empirical application of the Bayesian approach for modelling the evolution of population density distribution across time. It focuses on the case of Massachusetts by tracking changes in the importance of spatial distance from Boston concerning citizens' choices of residence according to data for 1880-1890 and 1930-2010. By adopting a Bayesian strategy, results show that Boston reinforced its attractiveness until the 1960s, when the city's accessibility no longer represented the unique determinant of population density distribution. Referring to selected historical evidence, a few possible interpretations are presented to endorse these results.

Population distribution over time: modelling local spatial dependence with a CAR process

“…Overall, a general attitude driven by discrimination in location choices has been reinforced over time with a positive dependence across all census tracts in a sort of complementarity effect. As already discussed in Epifani and Nicolini (2017), people with high income tend to settle in selective and exclusive locations where they have access to individual dwellings and, hence, where the population density is lower. As a consequence, the concentration of people with similar income levels in a spatial unit reduces the degree of heterogeneity in terms of average income in that unit and, hence, the Gini index approaches zero and the association with a positive (estimated) direct effect yields low population density.…”

“…Our main assumption is that individuals prefer to settle close to the CBD for several reasons. To commence modelling the spatial structure, we refer to Epifani and Nicolini's (2013) discussion about subjective preferences when presenting the monocentric spatial structure for Massachusetts; the other determinants for location are taken from the literature according to the discussion proposed in Epifani and Nicolini (2017). In the present study, we also aim to introduce the dimension of local spillovers (arising from spatial proximity).…”

“…For features to represent neighbourhood status, we follow Epifani and Nicolini (2017) in using ethnic composition along with education and income as potential quantitative determinants of the population density distribution. For the ethnic composition of the population, we use the proportion of whites over the total population (henceforth Mix).…”

“…Glaeser and Ward (2009) argue that water can be considered a fundamental factor in creating recreational spaces for leisure time. Following Epifani and Nicolini (2017), Z g can be considered to be constant over time. Regarding the spatial surface S g of each county g, Combes and Gobillon (2015) suggest that the land (or surface) area is fundamental for picturing the size of agglomeration effects.…”

“…This type of spatial organization of land stems from the adoption of a specific regulation for the real estate market that prevents free access to all the residential options for everyone. In this respect, Epifani and Nicolini (2017) discuss important features that limited the development of the housing market and free residential access in the US up to 1980s: limited access to credit or racial discrimination drove groups of citizens to concentrate in selected places. Under this perspective, the highly regulated housing and credit markets limited the potentially positive spillovers effects across census tracts.…”