Academic journal article Economic Inquiry

Reexamining the Added Worker Effect

Academic journal article Economic Inquiry

Reexamining the Added Worker Effect

Article excerpt

I. INTRODUCTION

The added worker effect refers to a wife entering the labor force when her employed husband becomes unemployed. This labor supply response is a transitory method of intertemporally smoothing family income and consumption, and is of interest to labor economists and to macroeconomists concerned with the labor force participation of married women over the business cycle. The added worker effect is familiar in terms of the income and substitution framework of the labor supply model. When family earnings fall as a result of the husband's unemployment, the income effect influences the wife to increase her labor supply. Similarly, the wife's reservation wage for market work falls when the husband's increased nonemployment time provides a substitute for her home production, and she is more likely to join the labor force as an added worker. The wife's reservation wage is also influenced by the degree of complementarity or substitutability of leisure time between the husband and the wife. Observing the net result of these three factors, the existence and the magnitude of the added worker effect is an empirical question.

This paper expands on the conventional income and substitution explanations in order to emphasize how employment uncertainty and liquidity constraints influence the added worker effect. In the context of a dynamic model of family labor supply, an unanticipated unemployment spell by the husband and the resulting shift in the budget constraint will alter expectations of future unemployment probabilities, which will cause the wife to revise her life cycle profile of current and future labor supplies. The explanation of liquidity constraints as a cause of the added worker effect is discussed in much of the literature: see Mincer [1962; 1966], Killingsworth and Heckman [1986], and Rosen [1992]. As stressed by Lundberg [1985], anticipated unemployment should not produce an added worker effect unless credit constraints limit access to borrowing against future income. Explaining how the timing of labor supply could be affected by variations in family income, Mincer [1962] cites capital market imperfections as a reason why "a transitory increase in labor force participation of the wife may well be an alternative to dissaving, asset decumulation, or increasing debt."

While the added worker effect is easily explained by theoretical models of family labor supply such as those sketched above or that presented by Ashenfelter [1980], empirical work has failed to reach a consensus regarding its magnitude, or even whether it exists.(1) Although Heckman and MaCurdy [1982] and Lundberg [1985] have shown that the added worker effect is small yet significant, Pencavel [1982] does not find an added worker effect once he controls for individual-specific fixed effects, and Maloney [1987; 1991] does not find any evidence of an added worker effect. It is interesting to note the wide variety of data used in this literature. Heckman and MaCurdy use annual panel data, Lundberg uses monthly panel data constructed from quarterly interviews, Pencavel uses annual panel data constructed from quarterly interviews, and Maloney uses annual cross-sectional data.

This paper is motivated both by the conflicting empirical work regarding whether or not the added worker effect exists and by the view that the added worker effect should be analyzed using panel data with a short recall period. Cross-sectional microdata cannot adequately capture the intertemporal decisions of wives joining the labor force when their husbands become unemployed. Annual measures of labor supply are less than ideal for examining the added worker effect since short spells of unemployment by the husband in the preceding year are often forgotten: see Horvath [1982] and Akerlof and Yellen [1985].

My purpose is to document the magnitude of the added worker effect using monthly longitudinal microdata from the Current Population Survey (CPS). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.