In his excellent post on taxes and the incidence of taxes, co-blogger Scott Sumner does not mention another important issue in taxation: deadweight loss. The deadweight loss from a tax is the part of the loss to those who bear the tax that does not go to the government. Thus the term "deadweight." (Scott's graph shows a small deadweight loss, but he does not elaborate on this.)
I noticed when checking the Concise Encyclopedia of Economics that the article on taxation, although it mentions incentive effects of taxation, does not introduce the term "deadweight loss." That's my bad as the editor.
There are three important bottom lines on deadweight loss.
1. The easier it is to avoid a tax (that's usually expressed as a higher elasticity of supply or demand), the greater is the DWL per dollar of revenue raised. That's because the tax has distorted a lot. If, for example, the number of cigarettes people buy drops a lot in response to a given increase in the per-pack tax, the tax has distorted the smoking decision a lot. Of course, some people, those who don't want people to smoke, like this distortion.
This is why the capital gains tax is so inefficient, that is, causes a large DWL. It is very easy to avoid the tax by not realizing your capital gain, that is, by not selling your asset whose value has increased.
2. A tax can cause a DWL and raise zero revenue. Here is my favorite example I've used when I've taught this. When I fly into Winnipeg every summer, I don't rent a car at the airport. Instead, I can save almost 20 days of airport taxes on the car rental by paying about $20, including tip, to take a cab to the Avis in downtown Winnipeg. I save close to $200 in taxes by spending an extra 15 minutes plus $20. The latter is the DWL. Notice that the tax raised zero revenue from me. Of course, I'm not claiming that it raised no revenue. But it led to a DWL on my part with zero revenue from me.
3. Last, and possibly most important, the DWL from a tax is proportional, not to the tax rate, but to the square of the tax rate. So doubling a tax rate will quadruple DWL. Cutting a tax rate by half will reduce DWL by 75%. So, imagine that Republicans somewhat succeed in cutting the corporate income tax rate from 35% to 20% and assume, for simplicity, no state tax on corporate income. That's a 43% drop in the tax rate and the new tax rate is 57% of the old tax rate. The new DWL will be (0.57)^2 of the old DWL. That's 0.32. So the DWL falls by 68%!