A Revealed Preference Approach to Measuring Hunger and Undernutrition

PWP-CCPR-2010-022

Abstract

In most developing countries, caloric intake and minimum calorie thresholds are used to assess hunger and nutrition and to construct poverty lines. However, there are several significant problems with these approaches: there is no consensus on the thresholds needed for subsistence (or whether such thresholds even exist); any such thresholds would vary considerably across individuals (and for the same individual over time) by both observed and unobserved factors; caloric intake cannot be accurately assessed due to imperfect absorption; and empirically, the responsiveness of calories to wealth is both low and non-monotonic, even for households far below recommended thresholds. We propose an economic approach to measuring hunger or undernutrition that overcomes these challenges by using consumption behavior rather than calories. Our approach derives from the fact that when a person is below their nutrition threshold, there is a large utility penalty due to the physical discomfort associated with the body's physiological and biochemical reaction to insufficient nutrition. At this stage, the marginal utility of calories is extremely high, so a utility-maximizing consumer will largely choose foods that are the cheapest available source of calories, typically a staple like cassava, rice or wheat. However, once they have passed subsistence, the marginal utility of calories declines significantly and they will begin to substitute towards foods that are more expensive sources of calories but that have higher levels of non-nutritional attributes such as taste. Thus, though any individual's actual subsistence threshold is unobservable, their choice to switch away from the cheapest source of calories reveals that their marginal utility of calories is low and that they have surpassed subsistence. Accordingly, the percent of calories consumed from the staple food source, or the staple calorie share (SCS), can be used as an indicator for nutritional sufficiency. We also provide an application of this measure for China, showing that there is a well defined SCS subsistence threshold that is consistent with both our model and a variant of the minimum-cost diet problem, and which is also largely "need neutral" with respect to individual characteristics. Finally, we show that SCS captures recent economic gains in China much better than calorie measures.