This paper considers what it might mean to describe the VAT as a "money machine," tests whether it is one, and asks if it might consequently be wise not to adopt it. We find broadly persuasive evidence, using panel data for the OECD, for a "weak form" of the money–machine hypothesis: that countries with a VAT raise more revenue than those without. But the effect may not be large. The evidence also supports a "strong form" of the hypothesis: that this association reflects not increased demand for government, but rather the greater effectiveness of the VAT in raising revenue. Models in which citizens/voters are likely to lose by entrusting politicians with a "money machine" rely on quite extreme views of their preferences and/or the effectiveness of electoral discipline.