Does the choice of consumption measure matter? An application to the permanent-income hypothesis

Food consumption in the Panel Study of Income Dynamics is used as a proxy for total consumption in many applications in economics, including tests of the permanent-income hypothesis, tests of separability between consumption and leisure, and tests of intergenerational altruism. Food, however, explains only a small fraction of the variation in total consumption. I propose a measure of composite consumption based on predicted wealth and compare it both to food consumption and to Skinner's (1987) measure of predicted consumption in a test of the permanent-income hypothesis. Using a log-linear intertemporal consumption function I find that food does not reject the permanent-income hypothesis but both Skinner's predicted consumption and the composite measure proposed here do reject the hypothesis.