Bruce Bartlett argues in favor of a consumption tax (instead of income taxes) in this op-ed. I don't necessarily agree with all his arguments, but I'm interested in how he handles the point of regressivity:

Liberals also make the mistake of assuming that a consumption-based tax system is regressive -- taking more out of the pockets of the poor than the rich. In fact, over one's lifetime, consumption is roughly proportional to income, because over a lifetime we eventually consume all our income. Thus, a tax on consumption will also be roughly proportional -- taking the same percentage from all taxpayers.

That may be true, although if one considers the diminishing marginal utility of money, taking the same percentage of money from all taxpayers may not really be treating them equally. (I.e., treating people the same is not necessarily treating them equally. Taking 10% of the minimum wage earner's income probably has more effect on that person's life than taking 30% of the $1m-per-year earner's income.)

In any event, the point I'm interested in is the assumption that a consumption tax is necessarily a flat tax. Why couldn't the percentage tax leveled on consumption rise with the dollar amount being spent? In other words, if you buy a $6000 Kia auto, you pay 5% tax, while if you buy a $150,000 Ferrari, you pay a 15% tax. Same for all other consumer goods -- let the percentage sales tax be scaled to the relative dollar amount for the good in question. And since rich people tend to buy more expensive goods and services, they would be taxed at a higher rate, just as under the current system. (On the other hand, if a rich person lives like a pauper and sticks all his money under a mattress, he won't be taxed. To which I say, so what?)

I haven't thought through all the implications of this idea, but it's one I haven't seen anyone else even consider. It would effectively answer, I think, the argument that a consumption tax is necessarily regressive. (For examples of the regressivity argument, see Angry Bear, and a more hysterical version of the argument by David Neiwert).