When Senator Joe Manchin declared that he would only support a Build Back Better (BBB) bill that was half as large as the one that was originally proposed, Democrats in Congress responded by using a variety of budget gimmicks — such as sunsetting programs after a few years — that made the cost of the bill appear lower. Manchin consistently said he opposed these kinds of gimmicks and torpedoed the BBB bill in part because of them.

Despite his apparent opposition to budget gimmicks, Manchin’s own policy preferences appear to be dominated by one of the worst budget gimmicks of all: phasing out cash benefits by income, which is also known as means-testing.

Manchin’s push for more extensive means-testing has been focused primarily on the Child Tax Credit (CTC), which the Senator apparently wants to phase out starting at $75,000 of income.

The primary appeal of this kind of means-testing is supposed to be that it makes the program cost less: providing benefits to some people is cheaper than providing benefits to all people. But these cost-savings are based on an accounting gimmick, not reality.

In reality, phaseouts function like revenue-raising taxes. They don’t reduce the cost of a program. Instead, they provide tax revenue to help fund it.

To understand my point on this, consider the following four policy options:

  1. $3k CTC
  2. $3k CTC + phaseout of 5 cents for every dollar of income beyond $75k
  3. $3k CTC + 5% surtax on families with children on income beyond $75k
  4. $3k CTC + 2.2% surtax on all families for income beyond $75k

For clarity purposes, let’s stipulate that the phaseout in (2) and the surtaxes in (3) and (4) all raise/save the same amount of money.

The last three programs have the exact same net fiscal cost: ($3,000 * number of children) – money saved/raised by surtax/phaseout. But under the prevailing accounting methods, options (3) and (4) show up as being vastly more costly than option (2). In fact, options (3) and (4) show up as being just as costly as option (1).

The reason for this strange outcome is that the money raised/saved by the phaseout is subtracted from the cost of the CTC, but the money raised/saved by the surtaxes is not. Instead, the surtaxes are scored as revenue that helps pay for a universal CTC.

This accounting convention is clearly out of touch with reality. Options (2), (3), and (4) are fiscally identical while options (2) and (3) are both fiscally identical and distributively identical. Saying option (2) costs way less than (3) and (4) is as silly as saying that sunsetting a program that you intend to extend later saves a lot of money.

Yet playing games with this specific accounting gimmick to drive the official budget score lower is an overwhelming fixation of Manchin and many centrist policy wonks in the US and across the world.

According to Manchin and those who think like him, benefit phaseouts target a fixed pot of money to those most in need. But this is not really true.

The accountants tell us that phaseouts target a fixed pot of spending and that taxes increase the size of the spending pot in order to provide universal, untargeted benefits. But our own eyes tell us that they are actually both doing the same thing. You can call them both targeting devices or call them both revenue-raising devices. But calling one a targeting device and the other a revenue-raising device is just repeating the silly accounting gimmick at the heart of all of this.

Although taxes and phaseouts don’t differ fiscally, they do differ administratively and, at times, distributively. So in the example above, options (2) and (3) are fiscally and distributively identical, but they are administratively different. The phaseout in (2) is administered by reducing benefit payouts for families with incomes over $75,000 while the surtax in (3) is administered by increasing tax withholdings for families with incomes over $75,000. The latter approach is less error-prone and simpler. It is better.

Option (4) is fiscally identical to (2) and (3) but is distributively different. Whereas options (2) and (3) apply the phaseout/surtax only on families with children, option (4) applies its surtax on all families, including those without children. This approach allows policymakers to lower the surtax/phaseout rate from 5 percent to 2.2 percent. It also ensures that all families benefit from income-smoothing when they add a child to their family and increases horizontal equality between families that have the same amount of earnings but different numbers of children. It is better.

A sane policy discourse would have us debating between different ways of “paying for” a child benefit, with the phaseout considered one of many ways to do that. Instead, due to the way the budget accounting works, we are stuck in a special kind of discourse hell where phaseout proponents rely on accounting gimmicks to say that their program simply costs way less than all other alternatives. Insofar as the tax approaches are actually better administratively and distributively than the phaseout approach, these accounting gimmicks are not just creating a skewed policy debate but also a worse society.