Studies of neighbourhood effects often attempt to identify causal effects of neighbourhood characteristics on individual outcomes, such as income, education, employment, and health. However, selection looms large in this line of research and it has been repeatedly argued that estimates of neighbourhood effects are biased as people non-randomly select into neighbourhoods based on their preferences, income, and the availability of alternative housing. We propose a two-step framework to help disentangle selection processes in the relationship between neighbourhood deprivation and earnings. We first model neighbourhood selection using a discrete choice framework and derive correction components to adjust parameter estimates in a subsequent neighbourhood effects model for the unequal probability that an individual 'chooses' to live in a particular area. Applying this technique to administrative data from the Netherlands, we find significant interactions between personal and neighbourhood characteristics in the selection model. This confirms individual differences in neighbourhood preferences; individuals non-randomly select into neighbourhoods. The baseline neighbourhood effects model reveals a significant effect of average neighbourhood income on individual income. When we include correction components for the differential sorting of individuals into specific neighbourhoods, the effect of neighbourhood income diminishes, but remains significant. These results suggest that researchers need to be attuned to the role of selection bias when assessing the role of neighbourhood effects on individual outcomes. Perhaps more importantly, the strong, persistent effect of neighbourhood deprivation on subsequent earnings suggests that neighbourhood effects reflect more than the shared characteristics of neighbourhood residents; place of residence partially determines economic well-being.
neighbourhood effects neighbourhood sorting neighbourhood choice income selection bias discrete choice longitudinal data