Tax incidence

Why does the person who writes the tax cheque often not bear its cost?

A legislature writing a tax has to name somebody. Someone must file the return, remit the money, and face the penalty for not doing so. So it seems natural to read the statute as an answer to the question "who pays?" — tax the employer and the employer pays; tax the seller and the seller pays.

But a statute can only bind behaviour it can observe. It can compel a remittance. Can it compel a price? If not, then naming the remitter and deciding who ends up poorer are two separate acts, and only the first is within the law's reach. What decides the second?

Take the simplest case. A tax of a fixed amount is imposed on every unit a shop sells, and the shop must hand it over. What does the shop do the next morning? It raises the price — if it can. That "if" is the entire subject.

Suppose the shop raises the price by the full tax. Some customers now buy elsewhere, buy less, or do without. How many? That depends on how easily they can substitute away. If the good is one they can drop with a shrug, the shop loses so much volume that it prefers to absorb part of the tax in its own margin. If the good is one they will buy nearly regardless — an addictive product, an essential medicine — almost nobody leaves, and the shop can pass through nearly all of it.

Now flip it. Impose the identical tax on the buyer instead, collected at the till. Does anything real change? The buyer's total outlay per unit rises by the tax, so the buyer wants fewer units, so the shop must cut its pre-tax price to keep selling. Work it through and you land in the same place: the same quantity, the same total wedge between what the buyer parts with and what the seller keeps, split the same way. Only the paperwork moved.

That is the result, and it is worth stating flatly: the burden falls on whichever side finds it harder to walk away. Economists call that difficulty inelasticity — a small responsiveness of quantity to price. The side that can readily substitute, wait, or exit shifts the tax onto the side that cannot. Statutory incidence, who remits, and economic incidence, who is made poorer, coincide only by accident. That coincidence is the exception, not the rule.

The cleanest live case is the payroll tax. In the United States it is nominally split down the middle: the employer remits half, the worker sees the other half deducted. But how elastic is labour supply? Most people, faced with a slightly lower wage, still turn up on Monday. Firms, facing a higher cost per worker, have rather more room — to hire fewer, to automate, to hold wages down at the next review. The general finding in the literature is that labour bears most of the total, employer share included, through wages lower than they would otherwise have been. The evidence is real but not unanimous: estimates vary a good deal by country and labour market, and the shifting appears less complete where wage floors or strong collective bargaining block the adjustment.

The limiting case makes the logic visible. Land cannot be manufactured, moved, or withheld from existence. Its supply is fixed, so a tax on land value has nowhere to go: the owner bears all of it, and — because no transaction is discouraged — it destroys no trades. That was Henry George's argument, and it is the mirror image of the cigarette tax, where the buyer's unwillingness to quit is what pins the burden to the buyer.

Three refinements keep this from being applied too loosely.

First, "less elastic bears more" is a statement about the margin, not about wealth or bargaining power. A poor buyer with no substitute is inelastic; so is a rich landowner. The rule tracks options, not resources, which is why incidence and fairness are separate questions that have to be argued separately.

Second, elasticity is not a fixed property of a good. It depends on the time horizon — almost everything is inelastic this week and elastic over a decade — and on the breadth of the tax. A tax on one city's hotels is easy to escape; a tax on all hotels everywhere is not.

Third, and this is where honest disagreement lives: incidence is easy to state and hard to measure. Nobody observes the counterfactual wage. The corporate income tax is the standing example — who ultimately bears it, shareholders or workers, is genuinely contested, with credible estimates spanning a wide range. The principle is settled; the numbers, in most real cases, are not.

• Deadweight loss: why the trades a tax prevents cost more than the revenue it raises.
• Whether a subsidy obeys the same rule in reverse, and who actually captures one.

Every term the collection defines is gathered in the glossary.