Do we accurately measure what we consume?

Abstract

Understanding how consumption patterns affect the environment and shape well-being hinges on the rationale that the data collected on what is consumed, who consumes it, and where it is consumed are indeed accurate. To identify these consumption patterns and recommend corresponding policies, researchers and policy makers often rely on national surveys. Studies have explored the accuracy of individual surveys and the level of agreement across surveys of the same type (e.g. household expenditures), but no studies have compared representative national surveys measuring consumption in different ways. This study compares household consumption measured as expenditures and as material consumption (i.e. physical units) to assess how well we currently measure what we consume. We use multiple rigorous, national surveys to estimate meat consumption, household energy use, and private automobile use in the United States, with consumption profiles parsed by affluence, race/ethnicity, and education. Our results indicate that commonly used surveys may not accurately track important aspects of household consumption. For meat consumption, which included 30 consumption profiles detailing the consumption patterns across different demographic characteristics and meat types (e.g. kilograms beef consumed/capita for Caucasians), there is considerable disagreement between data sources for 20 profiles. By contrast, national surveys accurately measure household energy and transport (disagreement for four profiles). Our findings indicate that national surveys more accurately measure consistently tracked, standardized consumables like electricity than irregularly tracked, variable goods such as food. These results cast doubt on studies that use national surveys to draw conclusions about the how the environmental impacts of food, and, potentially, other goods (e.g. manufactured goods) vary across demographic groups. Overcoming this challenge will necessitate new surveys, updating legacy databases, and harnessing breakthroughs in data science.