This paper proposes a theoretical explanation of the positive consumption multipliers of government spending often found in the data. The explanation requires two ingredients. First, labor demand expands (e.g., prices are sticky). Second, general nonseparable preferences over consumption and leisure should be such that the two goods are substitutes; that is, Frisch labor supply elasticity is lower than the constant-consumption elasticity; this implies that constant-consumption labor supply shifts left. Existing empirical evidence on the relative magnitudes of the two elasticities supports this hypothesis. The parametric conditions under which the result occurs are consistent with restrictions of concavity and noninferiority of consumption and leisure.