Comparing taxes across countries sounds very simple when you first think about it. After you spend a few weeks working on it, you start spotting some problems, but you think you can fix them. As the weeks turn into months and the months into years, you eventually realize that the problems can’t be solved, that the people who claim they have solved them are confused, and that Murphy & Nagel were right in saying that the entire project of “tax justice” is a technical and philosophical black hole.

Here are just a handful of the problems you encounter:

  1. Transfer income programs can be implemented as tax credits and, when they are, they are counted as tax reductions instead of welfare spending. Indeed, this fact is driving a lot of the rhetoric around the current reconciliation debate, with people like Seth Hanlon throwing out graphs saying the Democrats’ tax proposal significantly cut taxes on the lower and middle class. But this is only because the Democrats decided to have the IRS administer a child allowance program as a tax credit. If they were to have the SSA administer the same program, which they should, the middle class tax cut would turn into a substantial middle class tax hike even as nothing much changed about the distributive impact of the program.
  2. Taxes can be implemented as non-tax compulsory payments (NTCPs) and, when they are, they are not counted as taxes. Indeed, this is how the Affordable Care Act (ACA) worked. Instead of requiring people to pay taxes to the government so that the government can finance health spending, the ACA instituted an employer and individual mandate that required people to pay premiums to private health insurers so that those insurers could finance health spending.

    The practical difference between these two is so slim that, during the ACA debate, the CBO actually threatened at one point to score all private health premiums as federal tax revenue. Once the ACA’s mandates hit in 2014, the OECD moved about $930 billion of annual US health spending (5.3 percent of 2014 GDP) out of its “voluntary health care payment schemes” column and into its “compulsory private insurance schemes” column. If the CBO had followed through on its threat to count compulsory premiums as taxes, Obamacare would have been the largest and most regressive tax hike in US history. These days, OECD publications group Medicare spending, Medicaid spending, and private insurance premiums together and report the total sum as US health spending financed by “government/compulsory schemes.”
  3. To calculate someone’s tax rate, you need to divide taxes paid by income. But what is counted as income for these purposes? If you use market income for these purposes, then people who live exclusively on transfer income but who still pay consumption taxes when they spend end up with an infinite tax rate (or more accurately an incalculable tax rate due to division by zero). This is not a fringe occurrence as it describes anyone who, for example, lives solely off of modest Social Security checks. So the bottom decile or so of the market income distribution always pays infinitely high taxes. If you try to solve this problem by using market income and transfer income in the denominator, then tax rates wildly vary among people with identical incomes.
  4. Even if using market income did not have the problem in (3), it would have another problem similar to the one identified in (2), which is that certain tax-and-transfer policies can be done through mechanisms that are not considered taxes. A country that taxes high-earners to subsidize the wages of low-earners is doing something very similar to a country that compresses wage differences through centralized union bargaining, but only in the former is the income reduction of the high-earner considered a tax. Indeed, the tax system itself changes the distribution of market income by changing economic behavior even though we then use the resulting market income distribution as the baseline for assessing the distributive effects of the tax system.

I say all this just to establish upfront that I know what I am about to do in the rest of this piece is kind of silly, but only because the entire discourse about which countries have more or less progressive taxes is incoherent on a very basic level.

Nordic and US Labor Taxes

The easiest way to compare Nordic and US taxes that is also in line with conventional and common-usage understandings of taxes is to look at the marginal labor tax rates in the five countries. I do this in the below graph, which shows the marginal all-in labor tax rates for each country for a single person.

The all-in labor tax rate includes federal income taxes, an average of state and local income taxes, employee payroll taxes, and employer payroll taxes. The employer payroll taxes are counted both as taxes paid and as wages for these purposes. The income thresholds for the various tax brackets are expressed as percentages of the country’s average wage rather than as a currency amount.

Here is the same graph but cut off at 350 percent of the average wage so that those details are able to be seen more easily.

The reason there is a dip at 229 percent of the average wage in the US is because that is the point at which the Social Security tax stops being charged. The Social Security tax is a 6.2 percent employer payroll tax and a 6.2 percent employee payroll tax. When those rates go to 0 percent, that cuts the all-in marginal labor tax rate by 11.36 percentage points. The reason the tax rate later goes back up is not because the Social Security tax starts again, but rather because the federal income tax rate bumps up from 24 percent to 32 percent and then later to 35 percent.

The takeaway from this particular representation is meant to be that the Nordics have much higher taxes across the board, especially but not exclusively on the rich.

This particular representation of the tax differences in the country don’t allow you to calculate some kind of “progressivity index” of the sort people like to use because it’s just showing the institutional details of the countries rather than trying to break the population into income deciles and then compute taxes paid or something like that. But it has its advantages because doing the income deciles thing runs into problem (3) above while this approach doesn’t. This representation also uses a single adult and starts at 50 percent of the average wage, which means we get to avoid problem (1) above as caused by the Earned Income Tax Credit and Child Tax Credit.

The Wage Inequality Tax

Many policy people would be impressed with the above analysis. It’s designed to avoid a couple of the pitfalls of tax comparisons. It properly counts employer payroll taxes in wages and taxes. It anchors everything to national average wages rather than trying to put purchasing-power-parity dollars on the horizontal axis, which would have looked like it was making the different currency amounts comparable but wasn’t really.

Yet, despite all these meticulous adjustments, the true policyheads will correctly tell you that this graph is highly misleading. One of the reasons it is highly misleading is that the distribution of wages among wage-earners is vastly different in the Nordic countries and the US.

According to the OECD, the 50th percentile wage in the Nordic countries is, on average, 44 percent higher than the 10th percentile wage. In the US, the 50th percentile wage is 102 percent higher than the 10th percentile wage. In the Nordics, the 90th percentile wage is, on average, 142 percent higher than the 10th percentile wage. In the US, it is 400 percent higher.

The primary reason why the wage scale is so much more compressed in the the Nordic countries is that wages in those countries are generally established through centralized bargaining with very powerful unions that have for decades pursued a “solidaristic wage policy,” meaning that high-wage workers restrain their wage demands so that low-wage workers can be paid more.

I am not here to convince you that solidaristic wage policy is essentially a tax policy that redistributes gross wages from high-earners to middle-earners and low-earners, but….maybe it is? At minimum, let’s indulge the idea for a bit and put some rough numbers to it.

Using the percentile ratios from above, we can imagine that these countries consist of three groups of workers that each make up one-third of the working population. The first group is low wage workers and they all earn the 10th percentile wage. The next is middle wage workers and they all earn the 50th percentile wage. The last group is high wage workers that all earn the 90th percentile wage. This is obviously stylized a bit but these are the numbers I have and doing it this way actually understates the point I am about to make.

In this scenario, we can convert the percentile ratios into wage shares that tell us what percent of the total wage bill goes to each group, as we see in the graph below.

To move from the US wage distribution to the Nordic wage distribution, what you do is take 12.6 percentage points of the gross wage bill away from the high wage workers and then give 4.5 of those points to the middle wage workers and 8.2 of those points to the low wage workers. Continuing with our imagine-this-as-a-tax exercise, what we would say is that solidaristic wage-setting in the Nordic countries is akin to a 20.3 percent (12.6 / 62.3) tax on high-earners to fund wage subsidies to low-earners and middle-earners. The tax rates in the line graphs from the prior section thus represent the taxes applied to high-earners after this 20.3 percent has already been taken out.

In the most extreme case of Sweden, the solidaristic wage policy “taxes” away 23.7 percent of the wages of high earners and then they are hit with a top marginal tax rate of 66 percent for the earnings that remain.

We can also reverse the way we look at this and use the Nordic wage distribution as the baseline and then describe the lower wage shares for the US lower and middle wage workers as if they are taxes applied to them that fund subsidies to high wage workers.

In this way of looking at it, what is happening is that the US is taking 8.2 percentage points of the wage bill away from low earners and 4.5 points away from middle earners and then giving that money to high earners. Expressed as a tax, which we might call the Wage Inequality Tax, this is equivalent to a 39.6 percent tax (8.2 / 20.7) on low wage workers and a 15.2 percent tax (4.5 / 29.6) on middle wage workers. Then, after this Wage Inequality Tax is applied, the surviving wages are subject to the taxes in the line graphs from the prior section.

If you incorporate this 15 percent upfront Wage Inequality Tax to the taxes paid by Americans earning the average wage, the all-in marginal labor tax rate of the average wage-earner in the US goes to 47.6 percent. This is higher than the marginal tax rate at the average wage in Norway and Denmark and only 1 point below the marginal tax rate at the average wage in Sweden. In Finland, the rate is 7.4 points higher.

Health Insurance Premiums

Generally speaking, starting down the road of saying “shouldn’t the money we have to spend out of pocket on things they get for free be considered taxes” is a one way trip to realizing what I said at the very beginning of the piece: this entire project is a conceptual and technical black hole. But, there is a decent case for at minimum indulging this tendency when it comes to health insurance premiums. As mentioned already, under the ACA, employers and individuals are mandated to pay money to private health insurers and the OECD acknowledged this by placing that money into its “government/compulsory scheme” accounting column.

According to Kaiser, the average premium for a single-person employer plan in 2019 was $6,972, with $5,843 contributed by the employer (akin to an employer payroll tax) and $1,489 contributed by the employee (akin to an employee payroll tax).

If you add the employer premium to the gross wage and count the employer premium and employee premium as a tax, then the all-in marginal labor tax for the average wage goes to 47.6 percent. If you combine the health care premiums with the Wage Inequality Tax above, then the number goes to 53 percent, which is higher than every Nordic country except Finland, where the number is 55 percent.

The last thing on earth I want to do is fight someone about whether wage inequality or compulsory payments to private businesses should be considered taxes or not. As I mentioned already, this whole enterprise is a mistake. But practically speaking, if you want to pop off about tax differences across countries, you should always remember this: to raise revenue, you have to go where the money is.