Paul Krugman recently posted about Donald Trump’s proposal for eliminating the death tax, aka the estate tax. But Paul really only scratched the surface. The death tax has a number of perverse effects, none of which Paul discusses.

You might think that I’ve already biased the issue by using the term “death tax,” a term many conservatives have used to describe what is known formally as the estate tax. But actually what I’ve done is, however imperfectly, removed the bias. The estate tax is not a tax on estates. You could have a $100 million estate and pay zero taxes on it. What triggers the tax? Your death. So it’s a tax on net worth that kicks in when you die. For that reason, the term “death tax” is just as accurate as the term “estate tax.” Of course, death tax isn’t quite right either because you don’t pay a tax for dying if you don’t have an estate. So the best term might be, however clunky, “the estate tax on people who die.” What I’ll do below is use the terms “death tax” and “estate tax” interchangeably.

I remember reading about 20 years ago in, I think, an NBER study that the estate tax costs the federal government revenue. You might think that claim must be incorrect. A quick look at the data shows, after all, that in 2014, the federal government’s death/estate tax revenues were $19.3 billion, or 0.6 percent of total federal revenue. So how can I say that the tax costs the federal government revenue?

Answer: the unseen. That is, the tax causes people to make other adjustments, typically well before death, that reduce the government’s tax revenues from other taxes. One standard way is to put assets in your children’s name. But then your children are typically in a lower income tax bracket than you are. So the income from the assets is taxed at a lower rate than if you didn’t transfer the assets and the feds raise less revenue from the income tax.

The estate tax causes the government to lose revenue in other ways too. One way is that it reduces the incentive to accumulate capital and the lower capital stock leads to lower productivity than otherwise. Lower productivity means lower total output and, with lower total output, federal corporate and income tax revenues are lower. And it’s not proportional. If the growth rate falls by, say, 0.1 percentage point, federal corporate and income taxes fall by more than 0.1 percentage point: a good guess would be that they would fall by about 0.15 percentage point. Economists refer to this relationship as the elasticity of the taxes with respect to output. Why this more than proportional fall? Because the higher income that would have been taxed would have been taxed at people’s marginal tax rates–and marginal tax rates substantially exceed average tax rates. A given fall percentage drop in this income, therefore, leads to a higher percentage fall in tax revenues.

This somewhat dated piece by David Block and Scott Drenkard of the Tax Foundation, “The Estate Tax: Even Worse Than Republicans Say,” gives other reasons that the death tax could cost the government revenues, as well as showing other perverse results of the tax.