Food environment and weight outcomes: a stochastic frontier approach

Abstract

Food environment includes supermarkets, restaurants, warehouse clubs and supercentres and other food outlets. As such, it can affect weight outcomes through healthy food-access cost and availability, which in turn shape eating styles and habits. This article evaluates the impact of food environment on weight outcomes through a stochastic frontier approach (SFA) that comprehensively includes various components of the food environment. The model is estimated using body mass indexes (BMIs) as the dependent variable and demographics and food environment data from New England counties as explanatory variables. Empirical results indicate that supercentres and limited service restaurants are positively associated with weight gain, while full service restaurants are negatively associated with BMIs. In metropolitan counties, the presence of full and limited service restaurants has a significant impact on BMIs; in non-metropolitan counties, these establishments do not, except in those adjacent to metropolitan areas. Empirical results also validate SFA is a more appropriate specification than the standard linear model. In addition, SFA allows us to rank counties and identify targets for effective policy interventions. An ordered logit model of weight status is estimated with the same explanatory variables, confirming that the directions of the effects of the food environment components are similar.