How Well Can We Measure the Well-Being of the Poor Using Food Expenditure?
August 2006
Thomas DeLeire, Michigan State University and Congressional Budget Office; and Helen Levy, University of Michigan.
Download paper here. 
Abstract
How well does data on expenditure measure consumption? Because time is an important input into consumption that is typically not priced, spending data may not reflect true consumption. This problem may be particularly pronounced when using food spending because households' food consumption is a combination of home production (cooking at home) and market production (eating out). We find using both the Consumer Expenditure Diary Survey and the Consumer Expenditure Interview Survey that food away from home made up an increasingly large share of poor households' food budgets between 1992 and 2000. Total food spending may have increased even if food consumed did not, and analyses of well-being that rely on total food spending may reach erroneous conclusions as a result. We also find that the share of poor households' income that is from work also increased over this period. The incentives that increase work would also be expected to cause a shift from home production to market production. For this reason, the very economic and policy changes that researchers seek to evaluate may introduce bias into the data used to evaluate them. Our results suggest that caution is in order when using total food spending as a summary measure of how the well-being of the poor has changed over time.
Keywords:
Poverty Trends and Measurement