Skip to main content

Structural Household Analysis using Revealed Preferences

Final Report Summary - SHARP (Structural Household Analysis using Revealed Preferences)

Household consumption (including labor supply) is usually the outcome of a complex interaction between multiple household members. Understanding this consumption behavior requires a non-unitary modeling of households, which recognizes that households are not unitary decision makers (with a single, household level utility) but consist of interacting individuals (with own utilities). Interestingly, such a non-unitary analysis is no longer restricted to focusing on the aggregate consumption but can also address research questions on consumption patterns within households (e.g. related to intrahousehold resource sharing or the welfare of individual household members).
This project wants to build a powerful toolkit for empirical analysis of household consumption through structural models of individual preferences and within-household interactions. It aims at opening the “intrahousehold black box” while only using the (limited) consumption information that is observed at the aggregate household level.
The distinguishing feature of my project is that I will develop a revealed preference methodology that is intrinsically nonparametric. This method (only) uses preference information that is directly revealed by the observed consumption choices. It avoids confounding the analysis by imposing (nonverifiable) parametric/functional structure on within-household decision processes. My ultimate goal is to establish revealed preference methods as widely used tools for structural analysis of non-unitary household behavior.
To this end, my specific theoretical objective is to integrate marriage market (matching) dynamics into the structural modeling of household consumption. The marriage market defines the “outside options” (or “threat points”) of (adult) household members, which indirectly impacts the intrahousehold consumption allocation (e.g. through individual bargaining positions). Modeling these marriage market effects implies a better description of the intrahousehold decision process, which in turn yields a more powerful analysis of household consumption.
In addition, at the empirical level, my objective is to integrate revealed preference restrictions on non-unitary household consumption with nonparametric estimation of demand systems (subject to non-unitary shape restrictions on household demand). This will enhance the applicability of revealed preference methods to widely available (pooled) cross-section data sets. It also allows for addressing empirical issues such as unobserved heterogeneity across households and limited discriminatory power of revealed preference restrictions.
As for the theoretical objective, we developed a novel framework to analyze the structural implications of the marriage market for household consumption. We defined a revealed preference characterization of efficient household consumption when the marriage is stable. We characterize stable marriage with intrahousehold (consumption) transfers but without assuming transferable utility. Our revealed preference characterization generates testable conditions even with a single observation per household and heterogeneous individual preferences across households. The characterization also allows for identifying the intrahousehold decision structure (including the sharing rule) under the same minimalistic assumptions. Several empirical applications demonstrate the usefulness of our theoretical results.
As for the empirical objective, we proposed a novel method to set identify bounds on the sharing rule for a general collective household consumption model. Unlike the effects of distribution factors, the level of the sharing rule cannot be uniquely identified without strong assumptions on preferences across households. Our new results show that, though not point identified without these assumptions, strong bounds on the sharing rule can be obtained. We get these bounds by applying revealed preference restrictions implied by the collective model to the household's continuous aggregate demand functions. We obtain informative bounds even if nothing is known about whether each good is public, private, or assignable within the household, though having such information tightens the bounds. Empirical applications show that our method obtains narrow bounds that yield useful conclusions regarding the effects of income and wages on intrahousehold resource sharing, and on the prevalence of individual (as opposed to household level) poverty.