Testing among models of intrahousehold resource allocation

Abstract

Households are the locus for much economic decision making. Households function as producers, investors in human and physical capital, managers of risk, and as consumers. Many resources are allocated within households, and the effects of many policies depend on the dynamics of household decision-making. For example, household decisions about allocating food among individual household members may determine who benefits from programs to improve health, nutrition, or welfare. In many microeconomic studies, the household is treated as an individual for analytical purposes; in other words, it is assumed that a household acts as a single entity with a single set of preferences. Recently, however, a number of economists have recognized that households are sites of conflict as well as cooperation and have begun to use new classes of models to explain how resources are allocated among household members. This literature has especially improved our understanding of households in which income is not completely pooled. In such a setting, standard household models may offer misleading conclusions about the effects of policies on individual household members .I This paper goes beyond recent surveys of intrahousehold issue? by explicitly examining the assumptions, predictions, and empirical implications of different intrahousehold models. In particular, it analyzes empirical tests that distinguish among competing models. By clarifying the similarities and differences among alternative models, this paper also clarifies what has actually been demonstrated (and what has not been demonstrated) by previous empirical studies. In so doing, the paper points to specific topics in need of further research. This paper is organized as follows: Section 2 analyzes the assumptions and predictions of each of the five main classes of intrahousehold resource allocation models. Section 3 describes empirical tests of these assumptions along with an analysis of empirical work that has been generated. Sections 4 and 5 discuss the questions that remain unanswered and directions for future research.