United in heart, divided in wallet : Tests of alternative models of family behaviour

Abstract

Over the last decades, a substantial amount of empirical evidence has deemed the unitary model of family behaviour inadequate. A range of alternative models have been developed, but there is no general agreement as to which one is the most suitable. Empirical evidence may even suggest that the appropriate model differs between different parts of the world.

A key feature of non-unitary models is that they are open to the possibility that intra-household income distribution may affect household demand if the husband and the wife have different preferences. In my thesis, I present alternatives from three broad classes of models, all of which have in common that both individual preferences and the decision process matter. One alternative is the non-cooperative model (Ermisch 2003; Ulph 2006) in which it is assumed that family members are economically detached from one another and behave strategically according to their own private agendas. Another alternative is the collective model (Chiappori 1988) which postulates that family members will cooperate, but that the gains from cooperation might be unevenly distributed depending on each family member’s personal bargaining power. Bargaining models (Manser and Brown 1980; McElroy and Horney 1981; Browning and Lechene 2001) take it one step further by specifying a bargaining process directly, usually assuming Nash-bargaining. The unitary model, the non-cooperative model, the collective model and the bargaining model are fundamentally different and in many aspects mutually exclusive.

Using data from the US Consumer Expenditure Survey 2010, I select a sample of 812 quarterly observations of couples with children where both spouses are in full time employment. Based on the individual income records provided in the survey, I estimate how much of family income is formally controlled by the wife. I then use the expenditure data to examine whether the wife’s share of household income has a systematic impact on household spending on several categories of goods. The categories are children’s clothing, household operations, household equipment and health insurance – goods that can reasonably be classified as public to the spouses. I then estimate the impact of the wife’s share of household income on household spending within the context of a variety of model variants that impose different restrictions on the response patterns. Finally, I conduct a series of statistical tests to see if one type of model fits the data better than other models do. My findings indicate that the traditional unitary model of family behaviour has poor explanatory power, and that the collective model (Chiappori 1988) is a suitable candidate for its replacement. In this context, that means that the distribution of income between the spouses has a significant impact on how the household spends its resources.