Will Wilkinson has a great post up on the issue:
- Spending by any other name (The Economist)
The whole article is great, but here’s the really good bit:
At bottom I think income-tax rates ought to be based on income level, not on income level plus a passel of other factors, such as how many children one has, or whether one owns or rents. So, to my mind, the rate determined by income alone functions as the baseline. Thus it seems to me that fecund homeowners are getting paid for their fecundity and home-ownership, since they get a considerable break relative to the baseline. But wait! It’s also possible to see the rate for fecund homeowners as the baseline, and to see higher rates for childless renters as a penalty imposed on those foolish enough to choose pitterpatterless apartments. Why isn’t this the baseline? I suppose I would argue that taxes ought to be used primarily to raise revenue, and not to reward and punish behaviour. So here’s a way to reinterpret Ms Stevenson and Mr Wolfers’ proposal. Income taxes ought to be used only to raise revenue, and rates ought to be neutral to all factors other than income level. Policy intended to shape behaviour in ways not related, or only indirectly related, to raising revenue ought to come in the form of explicit subsidies.
Behavioural taxes are based on the idea that if you tax something you’ll get less of it. Subsidies are based on the idea that if you make something cheaper you’ll get more of it. But income tax subsidies are a perverse hybrid: it’s still income being taxed, but at different rates based on the subsidized conditions. If you let people who own a mortgage deduct the interest from their income taxes, you’re not subsidizing mortgages, you’re subsidizing the income of people who own mortgages. Not the same thing! (Yes, money is fungible, but there are temporal effects that plain ol’ fungibility doesn’t account for.)