How Taxes Actually Work on the Economy — From Your Paycheck to the Policy Debate

On July 1, 1993, the federal gasoline excise tax rose by 4.3 cents per gallon — part of the Omnibus Budget Reconciliation Act. The tax was levied on fuel distributors. Within weeks, retail pump prices had risen by approximately 3 cents per gallon — less than the full increase. Consumers paid most of the tax, but not all. Refiners and distributors absorbed the remainder through narrower margins.

Congress set the statutory incidence — who writes the check to the IRS. The market set the economic incidence — who actually paid. The two are almost never the same, and the gap between them is one of the most consequential and least-understood ideas in public economics. Understanding it, along with the efficiency costs taxes impose and the behavioral responses they trigger, is essential to reading any policy debate about taxes honestly.

Statutory vs. economic incidence: who really pays?

The statutory incidence of a tax is the legal obligation — who must remit the payment to the government. The economic incidence is the actual burden — whose real income falls as a result of the tax.

The division of economic burden is determined entirely by the relative elasticities of supply and demand on either side of the taxed market, not by legal designation. The IRS excise tax filing rules specify who files Form 720 and remits payment — but that tells you nothing about who ultimately bears the cost.

The intuition: when demand is inelastic — buyers have few alternatives and must keep purchasing regardless of price — sellers can pass most of the tax forward as higher prices. Buyers bear the burden. When supply is inelastic — sellers cannot easily reduce production or shift to other activities — sellers absorb more of the tax as lower net prices received. Sellers bear the burden.

The formal rule is clean: the share of the tax burden falling on buyers equals supply elasticity divided by the sum of supply elasticity and demand elasticity. At the extremes:

• Perfectly inelastic demand (vertical demand curve): buyers bear 100% of any tax.
• Perfectly elastic demand (horizontal demand curve): sellers bear 100% of any tax.

Three real examples:

Cigarettes. Federal and state excise taxes on cigarettes total several dollars per pack. Cigarette demand is inelastic — research puts the short-run price elasticity at around −0.3 to −0.4, meaning a 10% price increase reduces consumption by only 3 to 4 percent. Supply is relatively elastic. The result: retail cigarette prices have risen by roughly the full amount of excise tax increases over decades, with consumers bearing the vast majority of the burden.

Payroll taxes. The Social Security and Medicare payroll taxes are nominally split 50/50 between employer and employee — the Social Security Administration publishes current contribution rates showing the employer and employee each paying 7.65%. But labor supply is relatively inelastic (people need to work) and labor demand is more elastic (firms can substitute capital, offshore work, or reduce hours). Economic research consistently finds that workers bear most of the payroll tax as lower wages than they would receive in the tax's absence — regardless of the 50/50 legal split. Your paycheck reflects economics, not statute.

Corporate income tax. Economists debate who bears the corporate tax: shareholders (through lower returns), workers (through lower wages as capital becomes less productive), or consumers (through higher prices). The answer depends on the openness of the economy and the mobility of capital — in a world where capital flows freely across borders, more of the corporate tax burden may fall on labor than traditional incidence analysis suggested. This remains an active area of research.

Deadweight loss: the efficiency cost that most people ignore

Beyond the question of who pays is a more fundamental question: how much does the tax shrink the total economic pie?

Every tax drives a wedge between the price buyers pay and the price sellers receive. That wedge reduces the quantity transacted below the level that would exist without the tax. Some transactions that would have benefited both parties — consumer surplus plus producer surplus exceeding zero — do not happen because the tax price is too high for the buyer or the after-tax price is too low for the seller.

This lost value is called deadweight loss — also called excess burden — and it is the real efficiency cost of taxation beyond the revenue transfer from taxpayers to government. The revenue transfer is not a social loss; it is a redistribution. The deadweight loss is a pure destruction of value: transactions that would have occurred, and the gains they would have created, simply do not happen.

The Harberger triangle — named for economist Arnold Harberger — visualizes the deadweight loss as the triangle between the supply and demand curves above and below the tax wedge. Its area depends on:

1. The size of the tax rate: deadweight loss grows approximately with the square of the tax rate. A tax rate that doubles generates roughly four times the deadweight loss, not twice.
2. The elasticities of supply and demand: more elastic markets lose more transactions to a given wedge.

This quadratic relationship is the most important practical implication of deadweight loss theory. It means that concentrating taxes on a narrow base at a high rate is dramatically more costly than spreading the same revenue burden over a broad base at a low rate. A 40% tax on a narrow transaction category creates eight times the excess burden per dollar of revenue as a 10% tax on a broad base covering the same underlying activity at four times the rate.

This is why economists broadly favor tax systems with wide bases and low rates. It is also why excise taxes on specific goods — gasoline, cigarettes, alcohol — are evaluated differently from income or consumption taxes: if the taxed good creates negative externalities (pollution, health costs), the deadweight loss may be offset or exceeded by the corrected externality, making the tax welfare-improving even where it reduces quantity.

The elasticity of taxable income: how behavior responds to rates

The efficiency cost of a tax does not stop at the point of sale. For income taxes in particular, behavioral responses ripple through the entire economy: people work different hours, shift compensation from taxable wages to non-taxable benefits, change the timing of income recognition, reclassify income across categories, and in some cases engage in outright tax avoidance.

The elasticity of taxable income (ETI) captures this: it measures the percentage change in reported taxable income in response to a 1 percent change in the net-of-tax rate (1 minus the marginal tax rate). A higher ETI means more revenue leaks through behavioral response as tax rates rise.

The landmark study by economists Jon Gruber and Emmanuel Saez, "The Elasticity of Taxable Income: Evidence and Implications" (NBER Working Paper 7512), estimated an overall ETI of approximately 0.4 for U.S. taxpayers, with substantially higher elasticities for high-income earners — around 0.5 to 0.6 or more for incomes above $100,000. This means a 10% increase in the net-of-tax share (say, a marginal rate cut from 40% to 36%, which raises the net-of-tax share from 60% to 64%, an increase of about 7%) would be expected to raise reported taxable income by roughly 2.8 to 4.2% for high earners.

For policy, the ETI matters in two ways. First, it determines the revenue cost of a tax cut — more behavioral response means less revenue gained from a rate increase (or less revenue lost from a rate cut) than a static calculation would predict. Second, it is connected to deadweight loss: a higher ETI implies a larger welfare cost per dollar of revenue raised from that tax.

The Laffer curve: real, but overused

No discussion of tax rates and revenues would be complete without addressing the Laffer curve — the idea, associated with economist Arthur Laffer and popularized in policy debates since the late 1970s, that there exists a tax rate above which further increases reduce total revenue because behavioral responses shrink the tax base faster than the higher rate expands it.

The Laffer curve is real in a tautological sense: at a 0% rate, revenue is zero; at a 100% rate, revenue is also zero (since no one would work for income the government takes entirely). Therefore, there must be some revenue-maximizing rate in between. This is not controversial.

The policy debate is about where that revenue-maximizing rate is. The honest answer from the empirical literature is: considerably higher than the rates that typically prevail in high-income countries for broad income taxes. Most academic estimates of the revenue-maximizing marginal rate for ordinary labor income fall in a range well above current U.S. top marginal rates. The ETI estimates above imply a revenue-maximizing rate for top earners somewhere above 60% to 70% of income — well above the current 37% federal top rate, and still above the combined federal plus state marginal rates most high earners face.

This does not mean high tax rates have no behavioral costs — they clearly do. But the Laffer curve argument is most defensible as a claim about rates on capital income, on narrow and highly elastic tax bases, or in historical contexts of extremely high rates. As a general argument against raising any tax rate from its current level in the United States, it is not well-supported by evidence.

Why this reaches your paycheck

These abstractions have direct consequences for the numbers on your pay stub and your tax return.

The incidence result means that when your employer pays "their half" of payroll taxes, the economic research suggests much of that cost is factored into your total compensation — your wages would likely be somewhat higher without the employer-side tax. The take-home pay you see already reflects an incidence adjustment baked into labor market equilibrium.

The deadweight loss result means that tax-advantaged accounts — 401(k)s, IRAs, HSAs — are not just government gifts to savers. They are, in part, a response to the efficiency cost of taxing investment returns: by exempting certain savings from tax, policymakers reduce the distortion that discourages saving and investment. The same logic applies to the preferential rate on long-term capital gains versus ordinary income.

The ETI result means that high earners have stronger incentives than most people to structure compensation as deferred income, stock options, carried interest, or other forms that receive preferential treatment — not because they are particularly crafty, but because the behavioral response to high marginal rates is large enough to make it worthwhile. Understanding this does not tell you whether those rates are right or wrong. It tells you what to expect from them.

Taxes are necessary to fund public goods, address externalities, and finance a social insurance system. They are also not free. Every dollar of tax revenue comes with an efficiency cost that varies enormously depending on how the tax is designed. The machine that converts tax rates into revenue, behavioral change, and deadweight loss is not obvious from the statutory code — but it is the thing that actually governs how tax policy works in the economy.