The leading evidence against the unitary household models is that "who gets what" is significantly dependent upon "who earns how much." However, it is difficult to pin down the causal effect of relative earnings on intra-household resource allocation because households jointly decide both labor supply and consumption. I utilize longitudinal data to analyze the spouse's individual budgets – "pocket money." This unique data set allows for the specification of the simultaneous process of household decision-making in a fully stochastic fashion. By doing this, it is possible to differentiate unobserved spousal bargaining power from heterogeneity at the household level. The results imply that the balance of power between spouses is stable over time and robust to transitory changes in relative earnings. Public policies targeting the disadvantaged within households should be designed and implemented on the long-term basis.