Because of a small direct negative effect on private spending, temporary variations in government purchases as in wartime, would have a strong positive effect on aggregate demand. Intertemporal substitution effects would direct work and production toward these periods where output was valued unusually highly. Defense purchases are divided empirically into "permanent" and "temporary" components by considering the role of (temporary) wars. Shifts in non-defense purchases are mostly permanent. Empirical results verify a strong expansionary effect on output of temporary purchases, but contradict some more specific expectational propositions.