Food environment in the United States as a complex economic system

The food environment in the United States is complex. Sixteen socio-economic-demographic variables from various public data sources are studied with a machine learning algorithm to ascertain the causality structure associated with the food environment in the United States. High levels of unemployment and poverty are direct causes of high levels of food insecurity, while low income causes high levels of food insecurity via increased levels of poverty. Unemployment is a common cause for both increased levels of food insecurity and poverty. We find that food insecurity and participation in Supplemental Nutrition Assistance Program (SNAP) are related, yet no direct causality is observed. Contrary to past studies which find that SNAP participation decreased the occurrences of poverty, in contemporaneous time, we find that poverty and SNAP participation are related through several back-door paths, via food insecurity, unemployment, race and food taxes. Obesity and SNAP participation are indirectly related via several back-door paths, namely, race income, poverty and food insecurity and unemployment. Also, food insecurity and obesity are related by several back-door paths. Low income, high food taxes, and race (being Black and non-Hispanic) are direct causes of obesity. The complex causality structure in the US food environment reveals that policy variables cannot be treated independently of their rich causal structure. Government agencies responsible for designing policies for food assistance, poverty alleviation, combating food insecurity and obesity need to consider the interrelationships among these variables. (C) 2016 Elsevier Ltd. All rights reserved.