The tax was originally adopted as part of European integration in order to avoid double taxation by national sales taxes. Its key feature is refundability at the border. This requires tax authorities to know exactly how much tax is embedded in the prices of all exports. The VAT provided that paper trail. In its classic form, the tax is paid at each stage of production and distribution, with a credit for taxes paid at earlier stages. Because producers and retailers need to show that they paid the tax to be credited for the taxes included in the prices of the goods they purchased, the system is largely self-enforcing. And the invoice trail allows governments to refund the entire tax on exports at the border.

This suggests there is substantial room for raising broad-based consumption taxes in the U.S. without overburdening the economy. A very broad value-added tax levied on virtually all personal consumption could raise about half a percent of GDP in revenue for each 1% tax rate. But this sort of value-added tax is highly unlikely, though it would be best to treat all consumption equally. In practice, it is unlikely that more than 30% of GDP would be taxed, meaning that a 10% VAT would raise revenues equal to 3% of GDP -- about $350 billion this year. We could raise twice that at a rate no higher than now exists in most European countries.

The great bugaboo of a value-added tax is its regressive feature - - taking more from the poor than the rich. In the short run that's true, because the lower one's income, the higher one's consumption is as a share of income. Over a lifetime, however, consumption is roughly proportional to income, so a VAT would also be proportional - - taking about the same from rich and poor alike in percentage terms.