Misguided UBI skepticism: Krugman on Yang

Plenty of progressives are weighing in against Andrew Yang's run for mayor of New York, and I'm sure they have some good reasons. Yang's signature policy proposal is of course a universal basic income (UBI), set at something like $1,000 per month for every adult. I'm a fan of the UBI, but many lefties are not– at least, not when it is Yang pushing the UBI.

Count Paul Krugman among the skeptics. First, he notes, one of Yang's main justifications for the UBI is concern about massive job loss through automation (robo-apocalypse), for which there is so far scant evidence. Fair enough, although we are in early days of AI. In my view there are plenty of other reasons to support a UBI, including basic distributive justice arguments, so I don't find this critique decisive.

Second, Krugman dings the UBI for its high cost, which he considers unsustainable. The seemingly modest $1,000 a month comes with what appears to be a staggering price tag: "... the Yang proposal to pay $12,000 a year would cost... well over $3 trillion a year, in perpetuity."

The math here is pretty simple, but misleading and I think disingenuous. To get $3 trillion, we take $12,000 x 250 million adults. That unconditional payment to every adult really adds up! Why not just give the money to the people who really need it? Krugman: "The point is that for now, at least, the best way to provide an adequate safety net is to make aid conditional." 

This sounds kind of reasonable, but it ignores the fact that in practice a UBI would be paid for by a tax that varies with income, and thus the $12,000 UBI "payment" to a rich person would be more than taxed away. In fact, the net effect of a UBI plus taxes on disposable income is indistinguishable from an equivalent means-tested transfer program that assures every person at least a minimum disposable income– i.e., Krugman's conditional aid. I'll call such a program a basic income guarantee (BIG). It is also known as a negative income tax.

Let's do the math. 

To keep things super simple, I'll assume that the only government program is a transfer system with taxes that just pay for it (i.e., balance the budget). Each person i's pre-tax/transfer income is fixed at yi; I ignore any incentive effects of the tax on incomes. I'm also going to stick to a flat marginal tax rate of t (100t%) on each additional dollar of income. There are N adults. Total income pre-tax is I = Σyi, and the average per-adult income is Y = I/N.

UBI: 

Suppose we wanted to give every adult a UBI that is a fraction u of average income Y. Then we would set UBI to an amount U = uY, and the total spending on the UBI is NuY = uI. In the United States, a $12,000 UBI would be roughly 1/6 of average adult income, requiring u = 1/6. 

To pay for this UBI, the government has to raise on average U = $12,000 per adult in taxes, and using a flat marginal tax rate of t it is trivial to see that a balanced budget requires t = u = 1/6: UBI total cost = uI = uΣyi = tax revenue = Σtyi = tΣyi, thus requiring t = u to balance the budget.

After taxes and transfers, each person i's disposable income Di is:

Di = yi + U - tyi = U + (1-t)yi 

The total "cost" of this program to the government is of course simply N*U = roughly $3 trillion.

BIG: 

It's easy to construct a means-tested basic income guarantee that leaves every adult with exactly the same disposable income as the above UBI. Under such a BIG, low-income individuals receive a means-tested transfer payment Mi, which we make equal to U if their income yi = 0 (hence a guaranteed basic income), and which is reduced (phases out) by a fraction t for every dollar earned, up until it becomes 0: Mi = U - tyi until Mi = 0 at yi = U/t. Overall, the disposable income of the transfer recipients is:

Di = yi + Mi = yi + U - tyi = U + (1-t)yi

For people who make yi > U/t, the transfer Mi is zero, and instead they face a linear tax of Ti = tyi - U. Notice that their tax bill is based on their income but "discounted" by the amount U. Their disposable income is again given by the same formula:

Di = yi - Ti = yi - (tyi - U) = U + (1-t)yi 

Comparing BIG and UBI:

My simplified UBI and BIG programs leave each and every adult with the same disposable income, given their pre-tax income, and thus each person is on net paying or receiving the same amount vis-a-vis the government. Hence if the UBI scheme balances the budget, so does the equivalent BIG.

Despite the programs being identical in their net outcomes for both individuals and the government, the total transfer spending and taxes collected by the government appear to be much greater for the UBI than the equivalent BIG. Under BIG, only people below the income cutoff U/t get any transfer payment, and only those at zero income receive the full amount U. The total amount paid out (and therefore the amount collected in taxes) would depend on the distribution of income below the average, but for a ballpark, suppose half the population received some transfer and the average transfer were half the UBI; then total transfers and taxes under BIG would be a quarter of the UBI scenario, so something like $750 billion annually. Not peanuts, but not the scary $3 trillion number that Krugman throws out either.

This difference is strictly an accounting difference, not a real difference, because there is no difference between these policies in terms of the net amount people receive or pay, and thus no difference in their post-tax financial situation. Furthermore, because either a UBI or a BIG would presumably be administered through the income tax system, the net effects would be pretty obvious to transfer recipients and taxpayers alike. 

Nonetheless, maybe appearances matter, in that people get scared by the big numbers or care about the gross flows. From a political point of view, the $3 trillion price tag of the UBI looks a lot worse than the $750 billion price tag of the BIG. On the other hand, the UBI has the feature of paying everyone the same social dividend, which has always been part of its rhetorical appeal. Anyway, as an economist, Paul Krugman knows that we should look at net fiscal impacts, not gross flows, to determine how these programs really affect both individuals and the government budget. In other words, he knows better. Let's engage the substance of the debate over transfer programs, not the trivia.