Edyta Łaszkiewicz, Guanpeng Dong and Richard Harris
As is well known, ignoring spatial heterogeneity leads to biased parameter estimates, while omitting the spatial lag of a dependent variable results in biasness and inconsistency (Anselin, 1988). However, the common approach to analysing households’ expenditures is to ignore the potential spatial effects and social dependence. In light of this, the aim of this paper is to examine the consequences of omitting the spatial effects as well as social dependence in households’ expenditures.
We use the Household Budget Survey microdata for the year 2011 from which we took households’ expenditures for fruits and vegetables. The effect of ignoring spatial effects and/or social dependence is analysed using four different models obtained by imposing restrictions on the core parameters of the hierarchical spatial autoregressive model (HSAR). Finally, we estimate the HSAR model to demonstrate the existence of spatial effects and social dependence.
We find the omitted elements of the external environment affect negatively the estimates for other spatial (social) effect parameters. Especially, we notice the overestimation of the random effect variance when the social dependence is omitted and the overestimation of the social interaction effect when the spatial heterogeneity is ignored.