Who Pays a Tax? Legal Liability, Economic Burden, and Elasticity

A tax rule identifies who is legally liable for payment or required to collect and remit it. Economists call this statutory incidence. A shop may report a sales tax, an employer may remit a payroll contribution, or an importer may owe a customs duty. This tells you who handles payment under the rule; it does not by itself tell you whose purchasing power, income, or options fall after people and firms adjust.

Economic incidence asks who is worse off after prices, wages, profits, employment, and other choices respond. A seller that remits a sales tax may raise the customer price. A buyer legally charged a tax may negotiate a lower pretax price. An employer-side payroll tax may affect wages, hiring, product prices, or returns to owners. The adjustment can be shared, so “the business pays” and “the customer pays” are not complete incidence estimates.

It helps to separate who is legally responsible, who sends or collects the money, and who bears the change in real economic outcomes. Those roles can belong to the same party, but need not. In the simple market model, economists compare the price buyers pay with the amount sellers keep after tax. For broader policies, workers, capital owners, consumers of related goods, and people entering or leaving the market may also be affected.

Tax revenue is a transfer to government; it is not itself the same as the economic burden. Burden analysis considers changes in buyer and seller surplus and behavior. Public services financed by the revenue can also affect welfare, but that benefit is a separate part of a full distributional analysis.

In a competitive market, buyers care about the price they pay and sellers about the net price left after costs and tax. A per-unit tax of t creates a gap: the buyer price equals the seller net price plus t. The law may assign remittance to either side, but the market outcome has to account for the wedge either way. {source:openStaxElasticityTaxIncidence}

After a tax is introduced, the buyer price usually rises, the seller net price usually falls, and fewer units trade. The difference between those prices is the tax on each remaining unit. The division depends on how readily buyers and sellers can change what they do when prices move.

For a small tax near a competitive-market equilibrium, a useful approximation is:

• Buyer share ≈ supply elasticity ÷ (supply elasticity + absolute demand elasticity).
• Seller share ≈ absolute demand elasticity ÷ (supply elasticity + absolute demand elasticity).

Elasticity measures the percentage response of quantity to a percentage change in price. These are local approximations, not rules for every tax. They assume a competitive market, a small change, a defined product and horizon, and no major simultaneous changes in quality, market power, or policy. The intuition is that the side with fewer practical alternatives has less room to avoid the tax. {source:openStaxElasticityTaxIncidence}

Suppose buyers’ inverse demand is Pᶜ = 100 − Q and sellers’ inverse supply before tax is Pᵖ = 20 + Q. Pᶜ is the buyer price, Pᵖ the seller net price, and Q the units traded. These invented dollar equations make the calculation clear; they do not describe a particular product or country.

Before tax, set 100 − Q equal to 20 + Q. This gives Q = 40 and a price of $60. With the tax, buyers’ willingness to pay must cover the seller net price plus $20: 100 − Q = 20 + Q + 20. The new quantity is 30. Buyers pay $70 and sellers keep $50 after tax. The $20 wedge is split evenly in this example: the buyer price rises by $10 and the seller net price falls by $10.

Revenue is $20 × 30 = $600. That collection is a transfer to the public budget, not the deadweight loss. Quantity falls by 10 units, so mutually beneficial trades that would have occurred without tax do not happen. With the straight-line curves here, lost surplus is a triangle: one-half × $20 × 10 = $100. This distinguishes distribution of the tax payment from the efficiency cost of reduced trade.

Equal sharing comes from these particular slopes, not a general rule. If demand is much less responsive than supply, buyers tend to bear a larger fraction through a higher price. If supply is less responsive, sellers tend to absorb more through a lower net price. The statutory remitter could be the seller in either case.

If buyers have few substitutes over the period studied, demand is relatively inelastic: a price rise causes a comparatively small quantity reduction. Sellers may then pass more of a tax into the buyer price before losing many sales. A narrowly defined essential good might have few short-run substitutes, but the word “essential” alone does not establish elasticity; alternatives and time matter.

Production can also be hard to adjust. Specialized equipment, scarce land, fixed capacity, or long construction times may make supply inelastic. Sellers then have fewer ways to move resources elsewhere and may absorb more through lower net returns. Over time, firms can enter or exit, workers can move, consumers can switch products, and production methods can change. Those responses may alter elasticities and the incidence split.

“Inelastic” does not mean “no response.” It means the percentage quantity response is smaller than the percentage price change in the measured range. If demand elasticity is −0.5 and supply elasticity is 1.5, the local formula implies buyers bear about 75% of a small per-unit tax and sellers about 25%. Reverse those elasticities and the approximate shares reverse. This illustrates the formula; it is not an estimate for a real market.

Elasticity depends on market definition and horizon. A buyer may have few alternatives this week but more next year; a worker may be tied to a location now but move or retrain later; firms may have no spare capacity today while competitors enter later. An incidence claim should identify which people, goods, places, and dates its elasticities describe. {source:openStaxElasticityTaxIncidence}

The basic model shows why the name on a tax form does not determine the final burden, but real markets can depart from its assumptions. With few competing firms, pricing may depend on strategy and markups. Bargaining power, regulated prices, long-term contracts, import competition, or capacity limits can change how adjustment happens. A tax can affect quality, investment, hours, hiring, or market entry instead of only the posted price.

Tax design matters too. A per-unit tax creates a fixed wedge per unit; a percentage tax changes with the transaction price. A tax on all earnings can differ from one that applies only above a threshold. Exemptions, credits, deductions, collection timing, and use of revenue can affect behavior and distribution. Incidence analysis compares a specified tax with a counterfactual; that counterfactual is not always “nothing else changes.”

Timing matters because prices, wages, and contracts do not reset at once. Firms may initially absorb a tax and adjust later, or prices may move in anticipation. In labor markets, wages and hiring can respond on different schedules. The long-run outcome is not automatically a fixed share borne by consumers, workers, or owners; it depends on market adjustment and the rest of the economy.

CBO says shifting an indirect tax to consumers depends on sector conditions, including the number of firms, entry possibilities, and the relative price sensitivity of supply and demand. Its budget analysis also explains that indirect taxes can reduce income available for wages and profits, changing income- and payroll-tax receipts whether consumers or producers ultimately bear the original tax. A budget revenue offset is not an estimate of the consumer-versus-producer incidence split. {source:cboIndirectTaxIncidence2022}

A payroll tax may be legally divided between employers and employees, while its economic effects appear in take-home pay, employer costs, hiring, hours, prices, or owner returns. A model may predict some adjustment through compensation over time, but that does not prove that every specific tax increase is fully passed into wages.

A 2021 CBO working paper says U.S. empirical evidence is limited and does not generally apply directly to federal payroll-tax changes. Its short-run partial-equilibrium model, informed by tax-return and elasticity evidence, estimates that employees bear 58% of the additional burden from a short-run increase in a payroll-tax rate applying to all earnings. The result varies by design: its short-run model estimates 23% for an increase in the Medicare surtax rate and 62% for an increase in the OASDI rate. It estimates a larger employee share of the full burden when a change expands the taxable earnings base through thresholds or the share of compensation subject to tax. These are model-based estimates for specified U.S. changes, not universal observed shares. The paper says longer-run incidence depends on macroeconomic effects and how revenue is used. The paper discusses general-equilibrium effects separately; these percentages are short-run partial-equilibrium estimates, not long-run general-equilibrium predictions. {source:cboPayrollTaxIncidence2021}

A 2024 Census Bureau working paper studies state unemployment-insurance tax increases that unexpectedly affected more-exposed employers. Using matched employer-employee job-spell data, it finds lower employment growth driven by reduced hiring and minimal evidence of pass-through to earnings in that setting. The negative employment effects are strongest for young workers and single-establishment firms. This illustrates an adjustment channel besides wages; it does not show that all payroll taxes behave alike or that earnings cannot adjust later. The tax variation, employers, workers, and observed horizon are specific to the study. This is a Census Center for Economic Studies working paper; it says the views and conclusions are the author's and do not represent those of the Bureau, and no Bureau endorsement should be inferred. {source:censusPayrollTaxIncidenceUi2024}

Together, these papers show why the tax change and measured outcome matter. CBO presents model scenarios for selected federal payroll-tax changes; the Census study examines employer-specific, time-varying state unemployment-insurance taxes and finds hiring effects with little earnings pass-through in its sample. Neither supports “employers always pay” or “workers always pay.” Ask whether a labor-tax claim measures wages, employment, hours, prices, profits, or welfare—and over what period.

Businesses generally remit value-added tax at stages of production, but consumers, producers, and workers may share the economic burden. Consumer-price pass-through is one observable part of incidence: it measures price changes relative to a tax-rate change, not the full effects on wages, profits, quantities, or household welfare.

An IMF working paper studies 1,231 VAT changes across 67 consumption categories in Eurozone countries from 1999 through 2013. In pooled estimates, total pass-through across reform types is about 29% to 32%. The average masks differences: standard-rate changes show anticipation and later adjustment, with a long-run point estimate above full pass-through that is not statistically distinguishable from 100%; reduced-rate changes have an estimated long-run pass-through around 30%; reclassifications show little average price response. These findings depend on the sample and method. The authors caution that narrow product reforms do not necessarily predict a broad standard-rate change. The paper measures consumer-price pass-through: 29%–32% comes from two pooled specifications with controls, while the pooled specification without controls estimates 40%. {source:imfVatPassThrough2015}

An IMF study of Mexico compares two distinct five-percentage-point VAT reforms: in March 1995, the general rate rose from 10% to 15% outside border cities; in January 2014, the border-city rate rose from 11% to 16% while the general rate did not change. For 1995, the authors estimate that each one-percentage-point rise in the rate corresponded to about a 0.4-percentage-point price increase, or roughly 40% pass-through. For the 2014 border-city episode, the estimated overall price response was about 0.14 percentage point for that reform; this is not a per-point estimate. For taxed goods, estimated pass-through was about 50% in 1995 and less than half that in 2014. Because pass-through was incomplete, the study concludes producers and consumers shared the burden. These estimates concern specific episodes and do not forecast a future VAT change elsewhere. {source:imfVatMexicoPriceWelfare2018}

Prices may adjust before implementation, in the month a tax starts, or over following months. The Eurozone study finds timing differences across standard rates, reduced rates, and reclassifications. Competition, tax-base breadth, substitutes, durable versus nondurable goods, inflation, and the ability to reset prices can matter. Consumer-price data also do not reveal the whole distributional outcome: households buy different baskets, and firms may adjust wages, margins, or output. A claim that VAT “passes through to consumers” needs a country, product group, reform type, period, and method. For a guide to how these price measures differ in scope, see CPI vs. PCE vs. the GDP Deflator: How U.S. Inflation Measures Differ.

First identify what a report measures. Legal liability and remittance describe the tax rule and collection process. A change in shelf prices is evidence about consumer pass-through; wages, hiring, and profits capture other channels. A distributional estimate may combine several outcomes and assumptions. These concepts are related but not interchangeable.

Check the counterfactual and time window. Is the comparison with no tax, a prior rate, an indexed base, or another policy path? Does “short run” mean the first month, year, or a model period before firms and workers adjust? Do longer-run estimates include entry, migration, investment, and how revenue is used? A short-run price estimate is not a permanent incidence share.

CBO’s income-and-payroll-tax offset illustrates the distinction between fiscal accounting and incidence. CBO says recent offsets have generally ranged from 21% to 25% of estimated indirect-tax revenue, reflecting lower income- and payroll-tax receipts when the indirect tax reduces income available to workers and firms. It applies the offset whether consumers or producers ultimately bear the original tax. The 21%–25% figure is a budget-revenue adjustment, not the share of tax borne by consumers or employees. This 21%–25% range applies to covered indirect-tax revenue estimates for which CBO uses the offset. {source:cboIndirectTaxIncidence2022}

A careful summary says which group bears an estimated share, through which measured channel, in which study or model, and over what period. Avoid turning that conditional result into “this group pays the tax” without naming the evidence and adjustment mechanism. The answer depends on legal design, the market affected, each side’s ability to adjust, and the time horizon.

For the broader demand effects of tax and spending changes, see Fiscal Multiplier Explained: Spending, Taxes, and GDP and What Are Automatic Stabilizers? Taxes, Benefits, and Recessions for how tax and benefit rules respond during a downturn.