Spousal complementarity in home production

American Journal of Economics and Sociology, The, Oct, 2004 by Michael A. Leeds, Peter von Allmen

I

Introduction

IN THE TRADITIONAL NEOCLASSICAL FRAMEWORK, marriage is an exercise in comparative advantage, with husbands typically specializing in market work and wives in home production (Becker 1976). Research was similarly specialized, with wives generally viewed as secondary workers and husbands' contributions to home production generally ignored. As married women entered the labor force in increasing numbers, economists began to view the "labor market decisions of married couples as joint decisions (e.g., Lundberg et al. 1997: Hersch and Stratton 1997). A similar shift of focus has yet to take place with regard to home production decisions. The prevailing theoretical model of home production remains Gronau's (1976, 1977, 1980) framework, which implicitly assumes that wives are responsible for all the home production in the household. In the popular literature, Hochschild (1990) explicitly denied that traditional sex roles in home production were shifting to accommodate changes in the labor market. Like others before her (e.g., Vanek 1980), Hochschild claimed that, while married women were playing an increasing role in the labor market, housework remained their sole responsibility, forcing them to work a "second shift" at home.

Although a complete system of equations for household production time and labor supply for a married couple can be readily constructed at a theoretical level, the econometrics of estimating such a system are extremely daunting. An empirical model must account for the fact that all household decisions regarding home production and labor supply are likely to be endogenous, as are each spouse's wage and the number of children in the household. Failure to account for these endogeneities would result in simultaneity bias. Second, estimates of a couple's decisions regarding labor supply and the number of children are both subject to selection bias.

Labor economists interested in questions of time allocation are thus left with three options. The first is to ignore the complexities inherent in the problem and to employ a simple linear model. While easily performed and understood, such an approach is subject to the biases noted above and would yield inaccurate results. The second approach is to build a model that completely addresses all the possible sources of bias. A complete model of family decisions that accounts for all possible sources of endogeneity and selection bias, however, is an enormous undertaking that could fill several volumes, let alone a single paper. The final approach, which we advocate, is to build a model that, although

incomplete, answers the most pressing questions and addresses the most significant econometric problems. We believe that such an approach significantly advances the debate by clearly delineating the issues that need to be addressed and subsequently addressing an important subset.

We use data on white households from Wave XXV of the Panel Study of Income Dynamics to examine the interdependence of husbands and wives in home production. Using a system of simultaneous equations; we find that wives who work regard their husbands as substitutes for home production but that wives who do not work do not respond to their husbands' decisions. Husbands respond along a different margin. They regard wives as substitutes when children are present, but do not respond to their wives' decisions when no children are present. This asymmetry in responses could be the source of tensions that frequently accompany childbirth and changes in the labor market status of wives.

In the next section we place our paper in the context of the existing literature. In Section III we provide a theoretical underpinning for the estimation that follows. We present the empirical model and discuss our data set in Section IV and present results in Section V. A conclusion follows.

II

Previous Literature

OUR STUDY FILLS A HOLE in a literature that has applied models of home production only to wives. Such an omission may have been understandable a generation ago, given the traditional neoclassical view of marriage as an exercise in comparative advantage and the prevailing patterns of labor force participation. The lack of attention today, however, is surprising in light of the growing attention paid to spousal interdependence in labor supply.

Studies of family labor supply date back at least to Ashenfelter and Heckman (1974). Their conclusion that husbands supply labor exogenously while wives do so endogenously remains the basis of much recent work (e.g., Triest 1990). Lundberg (1988) reexamines the endogeneity of labor supply by estimating the labor supply functions of husbands and wives as a system of simultaneous equations, allowing the characteristics of one spouse to affect the labor supply of the other. The endogeneity of labor supply decisions reaches its zenith in the growing literature on household bargaining models (for a recent application, see Lundberg et al. 1997).

Previous work, however, typically centers on labor supply decisions. Even recent work by Hersch and Stratton (1997) that explicitly examined housework by both spouses did not look at the determinants of housework. Instead, their study used housework as an explanatory variable in a Mincer-type wage equation.