Wednesday 11 March 2015

Against capital gains taxes: Sumner edition

Capital gains taxes still do not make economic sense.

Here's Scott Sumner explaining that while wage and consumption taxes can be equivalent, capital gains taxes are effectively a distortionary tax on future consumption relative to present consumption.
The biggest confusion is that people don’t understand why capital income should not be taxed, and why a wage tax is equivalent to a consumption tax. Consider someone with $100,000 in income, who can choose to consume, or invest in a fund that will double in value over 20 years. Suppose we want to raise revenue with a present value of $20,000, from this person. We could have a wage tax of 20%, and raise $20,000 right now. Let’s also assume that this person decided to spend 1/2 of his after-tax income—leading to $40,000 in consumption today, and save the other $40,000, leading to $80,000 in consumption in 20 years. Note that both current and future consumption are reduced by 20% relative to the no tax case.

Alternatively, we could directly tax consumption at the same rate (say with a VAT). Let’s assume the person saved $50,000 and spent $50,000 on consumer goods. After paying VAT they consume $40,000 today, and the government gets the other $10,000. After 20 years the $50,000 saved turns into $100,000, but you must pay $20,000 in VAT, leaving consumption of $80,000. Exactly the same as with a wage tax. The total revenue to the government looks bigger, but is the same in present value terms.

In contrast, an income tax doubles taxes the money saved, once as wages, and again as capital income. So now it’s $40,000 consumption this year, and only $72,000 in 20 years ($80,000 minus 20% tax on the $40,000 in investment income), an effective tax rate of 28% on future consumption. And of course with inflation the effective real tax rate is still higher. Income taxes make no sense at all; if you want progressivity, tax big consumption more than little consumption.
What's a progressive consumption tax? A tax on income minus savings. In that set-up, you'd want to tax returns on investments (capital gains or interest or dividends) that were made in tax-preferred savings vehicles (and effectively then came from before-the-line pre-tax income). As Scott points out, that's not a tax on capital gains or capital income, it's a deferred tax on labour income.

Sumner proposes something pretty close to what New Zealand has. We get progressivity via the income tax rather than via a progressive consumption tax, but we don't tax capital gains, and we doimpose a fringe benefit tax that helps in avoiding nonsense: he proposes that company cars that can be used in off-hours are consumption, not investment; I'm pretty sure those here would attract FBT. Capital income is taxed though, and investment yielding capital income coming out of wage earnings is then double-taxed.

And all of it is highly reminiscent of Seamus's series of posts explaining how capital gains taxes are a bad idea. I get tired of commentators who point to National's reluctance to impose capital gains taxes as evidence that they're somehow bought out by moneyed interests when the economics on capital gains taxation are at best pretty iffy.



  1. Eric: would you consider expanding the argument against CGT beyond the narrow (IMO) economic framing above? CGT is often proposed as a way to reduce entrenched intergenerational inequality. Would it? And if it would, Are the efficiency costs prohibitive? Are there better alternatives? Land tax?

  2. Most of the income of the top 1% in the US comes not through capital bequests anyway: it's almost entirely wages and entrepreneurial income. Dividends, interest and rent make up maybe 15% of those folks' income. See the charts Jim put up here:

    If you're really worried about intergenerational inequality, it's way harder than simple looking things like inheritance taxes, land value taxes, or capital gains. Of the three there, land value taxes are the least bad. But none of them would really get to intergenerational inequality. I rather strongly expect that intergenerational transmission of broad status works not through money bequests but rather through cultural transmission and genetics - Greg Clarke's work here is valuable.

  3. Thanks Eric. Although that link does ignore realised capital gains in the measure of income (by the looks of it). Am I wrong?

    And now for something somewhat different:

    On the face of it, genetic/cultural explanations of inequality provide a strong ethical argument for redistribution, since the wealthy did not get there by hard work alone.

    Now we can debate how that redistribution would be achieved with minimal efficiency loss; ensuring that the wealth creators are still incentivised etc. But the normative side of things is settled if you take Greg Clarke stuff seriously, IMO.

    I'd love to hear a counter argument.

  4. Where dividend income is that low, seems unlikely to matter hugely.

    On your second point, cultural transmission is, in part, work on the part of the parents. So disagree there. Further, while genetics might greatly help in hitting the top reaches, they're hardly going to do it on their own: there needs to be substantial complementary work investment. That can remain pretty elastic to tax rates. And remember the Wilt Chamberlain example: what normative theory would have you toss out Nozick's thought experiment, even if genetics played a large role in determining peoples' willingness to pay to see him play.

  5. Ryan, the genetic/cultural explanations do seem to provide an ethical argument for redistribution, particularly if appealing to a veil-of-ignorance framework. But it seems to diminish the case for taxing wealth transfers across generations, unless it can be shown that these are highly correlated with genetic-cultural transfers and that taxing these transfers involves less efficiency loss than just having regular point-in-time redistributive tax-and-transfer policies. And then it would be another big leap to suggest that a CGT would be an effective, low-cost way of implementing an estate tax. That would be a pretty hard argument to make, I suspect.

  6. John Rawls was against income taxes because they are unjust.

    Rawls lent qualified support to the idea of a flat-rate consumption tax (see A Theory of Justice, pp. 278-79). He said that:

    "A proportional expenditure tax may be part of the best scheme [and that adding such tax] can contain all the usual exemptions."

    The reason why Rawls lent qualified support to the idea of a flat-rate consumption tax was because these taxes:

    "impose a levy according to how much a person takes out of the common store of goods and not according to how much he contributes."

    A simple way to have a progressive consumption tax for Rawls is to exempt all savings from taxation. Taxable consumption is calculated as income minus savings minus a large standard deduction.

  7. Hi Eric:

    Dividend yields are at historic lows. A quick visit to our friendly neighborhood wiki page indicates dividends account for a about a quarter of total returns these days...

    How did they get so low? smaller dividend payouts coupled with rising asset prices.

    Granted these gains are only realized when you sell. But given that capital gains are in the income tax data I would of thought it would have been easy for the other blogger to put together. Anyway, a back of the envelope calculation using those S&P 500 numbers and eye-balling the chart your supplied, it seems reasonable to put the overall share of income from capital somewhere between 15-25% for the top 1%. Perhaps that is small. But it would be great to know the actual figure.

    Your point on tax elasticity is one of the "efficiency loss" details I had in mind.

    RE: influence of genetics vs parental skills: I am unsure the distinction is that useful. The ethical implications go through for any mechanism (genetics or parenting) which implies that the individual cannot choose to be moderately successful. (It would nonetheless be of empirical interest to know whether the parenting or genetics dominate. Know of any studies? It seems that twin or adoption studies could be useful here.)

    The answer to your question (speculative): A normative theory that places a smaller weight on freedom than other concepts/values.

    Was it ever in doubt that Wilt Chamberlain was going to play basketball, regardless of whether he got paid 25c or 10c by everyone? I think Marshawn Lynch would be a better modern example. He apparently gets a lot of pain from playing football.

    Anyway, I imagine you are getting sick of this by now! Thanks for the enlightening chat.