In recent decades, the correlation between U.S. men's wages and hours worked has reversed: low-wage men used to work the longest hours, whereas today it is men with the highest wages who work the most. This changing correlation accounts for roughly 30 percent of the rise in the variance of male earnings between 1975 and 2015. In this paper, we rationalize these trends in a model of joint household labor supply. Our quantitative model generates similar changes to what is observed in the data as a reaction to shifts in women's education and labor supply, the gender gap, and assortative mating. Our model is consistent with the observations that the changing wage-hours correlation among men is driven by married men, and that there is little change in the wage-hours correlation among employed women and at the household level. The results suggest that taking into account joint household decision making is essential for understanding the dynamics of labor supply.