Capital Gains Taxes

Ignoring Inequality

For decades now, capital gains (and dividends) have been taxed at a lower rather than other income. This has been widely derided by liberals as tax-cuts for the rich. Before, I examine this, I want to stress that you can support a lower capital gains tax rate and still want (in general) to tax the wealthy more.

Let me explain, imagine the US had two tax brackets for regular income. The first was a tax of 20% for all income less than $200,000 and the second was a tax of 30% for all income above this amount. If I thought that lowering the tax brackets on capital gains to 15% and 25%, respectively would boost economic growth to raise utility more than the loss of revenue, that’d be just fine. I’d be supporting an equally progressive tax structure (note: 30%-20% = 10%, and 20%-10% = 15%), I’d just be trying to encourage savings and investment.

Indeed, I could even accept John Rawls’ idea that I should be trying to maximize the utility of only the worst-off member of society "title": "John Rawls", and, given certain beliefs about how economics works, I can still belief that capital gains should be taxed at a lower rate! It all depends on the values of parameters such as the elasticity of savings.

The point I’m trying to make is that (as always), there are two independent dimensions to the question of the capital gains tax:

How strong of an effect does diminishing returns have with regards to income? If Alice has $50,000 and Bob has $100,000, how much less does Bob (on average) value a dollar than Alice? Half as much? A tenth as much? Equally?

If Alice earns $50,000 in capital income and $50,000 in labor income while Bob earns $100,000 in labor income, how much should Alice and Bob pay in taxes?

We’ve previously estimated how important encouraging saving is, and we determined that increasing the savings rate by 1% is (in terms of utility) equivalent to increasing long-term consumption by between 1.9% and 3.6%, so this is the context we’ll be using to examine capital gains v. labor income.

The Wrong Question

First, let me do away with a common liberal misconception. It turns out that the vast majority of academic research has found that the capital gains taxes harm the economy "title": "Supply-Side Economics: An Analytical Review" "title": "Capital Taxation and Accumulation in a Life Cycle Growth Model" "title": "Taxing capital income: A bad idea" "title": "Redistributive taxation in a simple perfect foresight model". Indeed, I only found two papers arguing differently. The first came to no conclusion "title": "Fiscal policy and economic growth: An empirical analysis", and the second found that (under certain assumptions) a capital gains tax could increase growth, but probably not in the long run "title": "Increasing the capital income tax may lead to faster growth". If you don’t believe me, try looking at the research yourself. Sadly, I couldn’t find any meta-analyses.

[As long as I’m myth-busting, I’d like to remind you that an accumulation of capital among the top 1% is not the chief cause of the US’s rising income inequality.]

The question, though, is not whether capital gains taxes are harmful. The question is whether capital gains taxes are more harmful than other forms of government revenue. In particular, the question I’m interested in is whether raising $1 through capital gains taxes reduces utility more or less than raising that same dollar through the income tax.

Effects on Behavior

Theoretical Musings

The rich make the majority of capital gains income, and the rich are more likely to save income than the poor. So, it’s obvious that lowering capital gains will encourage savings more than lowering labor income taxes by the same amount. However, the exact same argument shows that we should lower the top labor-income brackets. As I mentioned before, we’re examining whether we should lower capital gains taxes given a constant level of income.

I’m personally pretty skeptical of the claim that lower capital gains taxes rates will encourage people to save more at any particular income level. Here’s why

  1. If you’re rich, what else are you going to do with your money; if you’re poor you’re probably not saving much anyways.
  2. Lower interest rates are equivalent to higher capital gains taxes, and there’s decent reason to believe that lower interest rates increase savings: the primary reason people save money is to spend it in retirement, and if interest rates are lower, then you need to save more money to maintain the same standard of living.
  3. I’m pretty skeptical that any non-extreme policy will significantly change people’s saving behavior. I think most people save whatever happens to be left over at the end of the week/month/year and that this isn’t a deliberate choice - that’d at least explain American’s 5% savings rate.

Empirical Evidence

This is all nice in theory, but let’s consider practice.

Everyone knows that when capital gains taxes are cut, the next year sees a huge uptick in capital gains and (often) government revenue. However, that’s because investors expect taxes to rise again and are trying to take advantage of temporarily low taxes. The long-term effects of long-term tax policy are more difficult to pin down. This is true for labor income too, but is a much less serious distortion.

Burman and Randolph estimate a long-term elasticity of -0.17 with -6.4 in the short-term "title": "Measuring Permanent Responses to Capital Gains Tax Changes in Panel Data". Auerbach and Siegel estimate a higher elasticity of -0.34 with -4.9 in the short-term "title": "Capital Gains Realizations of the Rich and Sophisticated". Now, obviously, elasticities of around 0.25 are relevant, but they’re a far cry from the short-term effects. Dowd, et al. break it down into more detail by dividing capital income into two parts: those from passthrough businesses and those through mutual funds. I think we can ignore the former [show] and focus on the latter estimate of -0.09 "title": "New Evidence on the Tax Elasticity of Capital Gains: A Joint Working Paper of the Staff of the Joint Committee on Taxation and the Congressional Budget Office".

Of these three studies, the median estimate was -0.17, which is fairly low as far as elasticities go. Coincidentally, this literature review "title": "Chapter 27 - Labor Supply: A Review of Alternative Approaches" also found a median estimate of -0.17 for labor income (though certain demographics are more/less elastic than others "title": "Labour Supply and Taxes"). Admittedly, a different analysis found a long-run elasticity of around 0.26 "title": "The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review", but overall it seems safe to say that the long-term elasticities of capital gains and labor income are fairly similar.

Effects on Capital

However, despite all this, I do think there is one way that reduced taxes on capital gains uncontroversially increase savings even given the same income.

Imagine that Alice and Bob start working and that Alice saves 50% of her income while Bob doesn’t save any of his. If we give $100 of permanent tax breaks to both Alice and Bob, we can expect total savings to increase $50, because Alice will save half her new income. However, if we give the same sized tax break ($200), but give it all to Alice, we can expect savings to increase $100. In this way, we can see that giving tax breaks to people who save more will increase savings - even if no one’s behavior changes.

Here are some realistic assumptions about a typical American:

  1. They save a constant proportion of their labor income, which they invest in a Roth/401k.
  2. All dividends/interest earned by their savings gets immediately reinvested.

If these assumptions are met, we can figure out the long-term reduction in capital that we should expect taxes on labor and capital income to generate. If we suppose a 6% real interest rate, these reductions look like this:

[todo: insert figure]
The x-axis is the tax rate, the y-axis is the amount of capital. The black line represents capital as a function of the labor income tax rate, while the red line represents the same for capital taxes.

It might seem weird that savings continues even with a 100% tax on capital income, but keep in mind that this is empirically true. For most of human history people didn’t get interest from banks when they saved, so they didn’t derive any income from saving - unless, of course, they were very rich and could invest directly ventures. Similarly today, if you want to retire, you have to save - whether or not you earn interest.

After accounting for the fact that capital income is roughly 2/3rds that of labor income, this allows us to “deduce” that capital taxes (in this model) are better only at tax rates above 47% - this threshold is fairly resilient to changes in the assumed real interest rate.


Ultimately, given their similar elasticities and effects on capital, I see little reason for capital gain income to be taxed at a lower rate than labor income.

However, there is one confounding factor that cause naive capital gains taxes to be higher than they appear: inflation.

Imagine you hold a $100 stock that yields 4% in dividends each year and grows 6% in price, while the inflation rate is 2%. On paper, you’ve earned $10 and you will be taxed as such. However, in practice, your real net-worth has only increased by $8, because inflation has reduced it by 2%. Hence, in terms of limiting economic distortions, you are being taxed 25% too much.

The obvious solution is to allow people to deduct inflation from their capital gains, but, alas, we don’t like obvious solutions in the US.

The NYSE Composite index comprehensively represents the performance of all stocks in the New York Stock Exchange, so it seems like as good a place as any to estimate nominal returns. It has made 8.1% average annual returns over the last 40 years "title": "NYSE Composite (DJ)". Over the same time period, inflation averaged 3.5% "title": "Historic inflation United States - CPI inflation". Ultimately, then about 43% of all stock market returns were merely nominal and didn’t represent real income.

Given this, it’d makes sense if capital income tax rates were at roughly 57% the rate of labor income. The current maximum labor income tax rate is 39.6% and 57% of that is 22.5%, which isn’t far from the current top rate of 20%.

So, at the end of the day, I’m pretty satisfied with how capital gains tax rates currently relate to labor income tax rates.

Works Cited [show]