What is the difference between targeting and universalism? 

Tax churn.  Or so I will suggest.

There are two basic ways to improve the economic position of disadvantaged Americans using the income tax system.  The first approach, targeting, uses refundable tax credits to put more money in the hands of lower-income households.  Subsidies decrease for households with higher earnings.  The second approach is to use a Universal Basic Income, which gives refundable tax credits to everyone, regardless of income. 

The idea of a UBI is growing in popularity, but there is a good deal of confusion about the differences between targeted tax credits and a UBI.  I think the confusion is at least partly due to the fact that phasing out a tax credit as income rises looks suspiciously like giving a payment to everyone and then funding it through a tax on income. 

In this post I will argue that

1. Universalism = targeting + tax churn
2. Tax churn has political and economic costs

The easiest way to see this is by looking at a very simple example. 

Imagine an economy with two types of jobs, low skill jobs paying $10k, and high skill jobs paying $100k.  Suppose that half the population is high skill, and half is low skill.  

The income tax initially is a flat tax with a marginal tax rate of 20% and a personal exemption of 20k.  This means that low skill workers pay no income tax, and high skill workers pay tax of 16k (20% of the 80k they earn in excess of the personal exemption).  The government collects 8k in revenue per person (it collects 16k from each high income person, which is 8k per person because half of the population is low income).  This money is spent on public goods.

Now assume we want to help low income workers.  Let’s compare a targeted approach to a Universal Basic Income.  

Targeting

Let’s start with the targeted approach.  Suppose we give each low income person a refundable tax credit of 10k.  This tax credit counts as part of their taxable income, so the pre-tax income of low income people is now 20k (10k earned, 10k in a refundable tax credit).  None of this is subject to tax, however, because the exemption amount is 20k.  (If the tax credit pushed low-income workers over the exemption amount, then they would be subject to income taxation, and they might be discouraged (somewhat) from working, but that is not the issue I want to focus on here.)

The government needs to continue to collect revenue of 8k per person to cover existing spending priorities, and it needs to collect additional revenue to pay for the new targeted tax credit.  Clearly, the income tax rate needs to be increased, but how far?  Well, the income tax will need to collect 16k from each high income person to cover existing spending plus 10k to cover the cost of the tax credit.  (Because there are equal numbers of high and low income people, each high income person needs to cover the cost of one tax credit.)  This means the tax rate on income needs to rise to 32.5% (we need to collect 26k from each high income person, on 80k in taxable income: 26/80 = 32.5%).

Univeralism

Now suppose that instead of a tax credit targeted to low income people, we implement a Universal Basic Income.  This means that everyone gets 10k, even high income people.  To pay for this, the taxes of each high income person need to go up by 10k, to 36k.  The tax rate needs to rise to 45% (36/80 = 45%).  

The after-tax position of high and low income people are the same under these two tax programs.  Low income people have 20k (10k earnings plus 10k tax subsidy), high income people have 74k (100k earnings less 26k in tax payments with targeting, 100k income plus 10k tax subsidy less 36k tax with universalism).

The important lesson here is that extending the targeted tax credit to high income people results in pure tax churn.  We give high income people a tax credit and then turn around and increase their income taxes by exactly that amount to keep the government’s budget in balance.  In other words, in this simple model, a UBI is equivalent to a tax credit targeted to low income people coupled with pure tax churn for high income earners.  

The economic costs of tax churn

Tax churn is harmless (and also pointless) in this model, but in the real world giving people money and then raising their taxes results in efficiency losses.  And the efficiency losses from a UBI would be substantial, because the amount of money that would need to be raised to implement a generous UBI (which is needed to ensure a decent standard of living for low income people) would be enormous.  For some rough numbers, see Hoynes and Rothstein here. 

The political cost of tax churn

Universalism is often thought to have political benefits, and this may well be true if we focus only on the benefit side of the equation.  But taking taxes into consideration cuts the other way.  In the models I described above, the net redistribution is the same, but the nominal tax rate is much higher under a UBI.  To the extent that people suffer from “tax illusion”, or “nominal tax aversion”, this suggests that targeting has a potential political benefit. 

There are lots of complications I am glossing over here.  Universalism is simpler to administer than targeting and it avoids clawback problems.  (However, I suggested that these problems with targeting can largely be avoided through better program design.)  Some people really do have special needs, which again argues for some degree of targeting.  A patchwork system of targeting can lead to both low participation rates and high marginal tax rates on the disadvantaged who do participate.     

I was prompted to write this by a few posts on targeting and universalism that I found confusing:  initially this post Scott Alexander, and more recently by Matt Breunig (here, here).  Scott Sumner responds to Alexander, and I think he’s making a point similar to mine, but I’m not sure.  I wrote down this simple example to clarify my own thinking and figured I would share it.