Predicting aggregate food consumption for a specific geographic area: an application to southeast Minnesota

Abstract

This thesis develops a framework for estimating food expenditures for a variety of U.S. communities, including regions, states counties and metropolitan areas. The framework is then illustrated by providing estimates of household expenditures for 19 food categories at the national level, in the Twin Cities metropolitan area and in the Southeastern Minnesota area. First household characteristics are related to food expenditures using Consumer Expenditure Survey Data (CEX); then expenditures are aggregated at the community level by applying household demographic profiles from American Community Survey data to the estimations from Consumer Expenditure Survey data. This research is distinctive because (1) it suggests a general and universal model for forecasting food expenditure patterns at almost any regional level; and (2) it provides a good estimation of food expenditure to match with current foodshed analysis. The regression results present a comprehensive relationship between demographic factors and consumer expenditures on 19 food categories. Findings also show that household purchasing patterns are significantly different across regions and that it is not accurate to use average CEX results for the nation to estimate aggregate expenditure by households in a particular locale.