J Popul Econ (1989) 2:237-265 --Journal of Population Economics © Springer-Verlag1989 The marriage premium and compensating wage differentials W. Robert Reed and Kathleen Harford Department of Economics, TexasA&M University, College Station, TX 77843, USA Received September 7, 1988 / Accepted October 11, 1989 Abstract. This paper proposes and tests an alternative explanation of the mar- riage premium that relies upon differences in workers' tastes and compen- sating wage differentials. A key assumption is that marital status proxies for the consumption of family goods - such as children; and that these are costly. Workers whose greater demands for family goods are taste-generated are shown to choose jobs that offer greater wage, and lesser nonpecuniary com- pensation. This creates an observed wage premium that has nothing to do with differences in workers' productivities. Supporting empirical evidence for this hypothesis is presented, including a reevaluation of previous studies. I. Introduction Most studies of labor market earnings report that male married workers earn sig- nificantly more than male single workers. This earnings differential has been termed the "marriage premium" Holding constant human capital variables, male married workers typically earn from 10% to 40% more than male single workers. This makes the unexplained wage differential associated with marital status ap- proximately the same magnitude as that associated with race, sex, and union membership. Table I demonstrates that a marriage premium has been identified for young, old, white, black, and various ethnic and religious groups of workers. 1 Any explanation of the marriage premium must address why some males are married and some are not. Previous research has stressed the role of unobserved differences in workers' productivities: either marriage makes men more produc- tive, with some men being made more productive than others; or more productive 1 A few studies do not find marriage premiums. These include Tomes (1984) for various religious denominations and Duncan and Leigh (1985) for union members. Both studies report marriage premiums for other groups.