Harberger Triangle

Harberger Triangle is the standard geometric and analytical measure in welfare economics and public finance for the deadweight loss (or excess burden) created when a tax, tariff, price control, or Monopoly markup drives a wedge between the marginal willingness to pay of buyers and the marginal cost of sellers, reducing equilibrium output below the socially optimal competitive level. First conceptualized by French engineer-economist Jules Dupuit in 1844, generalized to multi-market general equilibrium by Harold Hotelling in 1938, and popularized empirically by Arnold C. Harberger in landmark 1954 and 1964 papers measuring the welfare costs of monopoly and taxation, the concept is a cornerstone of Business & Economics Theory. Its central mathematical result—that the deadweight loss of a price distortion is proportional to the product of the market elasticity and the square of the tax or markup rate (τ²)—underpins optimal tax theory, antitrust enforcement, and trade policy analysis.

In a competitive market without externalities (or where externalities have already been internalized via a Pigouvian Tax), the inverse demand schedule P^d(Q) measures the marginal social benefit of unit Q, and the inverse supply schedule P^s(Q) measures its marginal social cost. Competitive equilibrium (P₀, Q₀) equates P^d(Q₀) = P^s(Q₀) = P₀, maximizing total social surplus (the sum of consumer surplus and producer surplus) and achieving Pareto Efficiency.

Suppose the government levies an ad valorem excise tax at rate τ (or a specific unit tax t = τP₀) on the good. This tax drives a vertical wedge ΔP = P_d − P_s = t between the gross price paid by buyers (P_d = P₀ + ΔP_d) and the net price received by sellers (P_s = P₀ − ΔP_s), contracting equilibrium quantity from Q₀ to Q₁ = Q₀ − ΔQ.

• Tax Revenue Transfer Rectangle: On the Q₁ units still traded after the tax, buyers pay more and sellers receive less, transferring the rectangle T = (P_d − P_s) × Q₁ = t × Q₁ from private agents to the government treasury. Because this revenue can fund public goods or lump-sum rebates, the rectangle T is a redistribution rather than a net loss to society.
• The Harberger Deadweight Loss Triangle: On the marginal units between Q₁ and Q₀ (a quantity contraction of ΔQ = Q₀ − Q₁), buyers valued each unit at P^d(Q) ≥ P^s(Q), yet the tax wedge t prevents these mutually beneficial trades from taking place. The lost consumer surplus (the upper sub-triangle) plus the lost producer surplus (the lower sub-triangle) on the eliminated output ΔQ forms the Harberger Triangle:

The Quadratic Tax-Rate Approximation

Using a first-order Taylor expansion around (P₀, Q₀), let ε_d = −(dQ_d ÷ dP) × (P₀ ÷ Q₀) > 0 denote the price elasticity of demand (defined as a positive number) and ε_s = (dQ_s ÷ dP) × (P₀ ÷ Q₀) > 0 denote the price elasticity of supply. For a horizontal constant-marginal-cost supply curve (ε_s → ∞), the entire tax t = τP₀ falls on consumers (ΔP_d = τP₀), so the quantity contraction is:

Substituting ΔP = τP₀ and ΔQ = ε_dQ₀τ into the triangle area (1 ÷ 2)ΔPΔQ yields the classic Harberger Quadratic Formula:

When both supply and demand have finite elasticities (ε_d < ∞ and ε_s < ∞), the combined elasticity is the harmonic-mean term (ε_d × ε_s) ÷ (ε_d + ε_s):

Policy Implications of the τ² Law

Because deadweight loss scales with the square of the tax rate (τ²), doubling a tax rate from 10% to 20% quadruples the excess burden (from 0.01 to 0.04 in τ²), while raising less than twice the tax revenue because the tax base Q₁ shrinks. Consequently:

1. Tax Smoothing Over Time (Barro, 1979): Keeping a constant, moderate tax rate over the business cycle causes less deadweight loss than alternating between low and high tax rates of the same average level (by Jensen's inequality on the convex function τ²).
2. Broad Tax Bases with Low Marginal Rates: Taxing a wide range of goods at a uniform low rate τ generates far smaller cumulative Harberger Triangles than taxing a narrow base at a punitive rate.
3. Ramsey Inverse-Elasticity Rule (1927): To raise a given revenue target with minimal total Harberger deadweight loss across independent markets, commodity tax rates τ_i should be set inversely proportional to compensated price elasticities (τ_i ∝ 1 ÷ ε_i).

Hicksian Compensated Demand vs. Marshallian Demand

As Arnold Harberger (1964) and Richard Musgrave emphasized via the Slutsky Equation, even a non-distortionary lump-sum tax reduces household disposable income and shifts uncompensated Marshallian demand for normal goods leftward. The true excess burden of a distortionary tax compared with an equal-revenue lump-sum tax depends on the substitution effect along the Hicksian (compensated) demand curve Q^c(P, u), where ε_d^c is the compensated price elasticity.

Harberger's 0.1% of GDP Monopoly Finding

In his famous 1954 American Economic Review study ("Monopoly and Resource Allocation"), Harberger assumed unit price elasticity of demand (ε_d = 1) across 73 U.S. manufacturing industries from 1924 to 1928 and used deviations of industry profit rates from the manufacturing mean to measure the price wedge L. Summing the resulting triangles (1 ÷ 2)P_iQ_iL_i² across all industries yielded an aggregate monopoly welfare loss of less than 0.1% of U.S. national income—prompting James Tobin's remark that it takes a heap of Harberger Triangles to fill an Okun Gap.

The Tullock Rectangle and Rent-Seeking Critique (1967)

Why was Harberger's estimate so small? First, as Keith Cowling and Dennis Mueller (1978) pointed out, a profit-maximizing monopolist operates on the elastic portion of demand (ε_d = 1 ÷ L ≥ 1) rather than ε_d = 1, and advertising expenditures hide part of monopoly rents. Second, and more fundamentally, Gordon Tullock (1967), Anne Krueger (1974), and Richard Posner (1975) demonstrated that the monopoly profit rectangle Π = (P_M − c)Q_M is not a costless transfer if firms expend real productive resources—lawyers, lobbyists, campaign contributions, and defensive barriers—competing to obtain or protect the monopoly privilege. Under competitive rent-seeking, the entire Tullock Rectangle is dissipated into social waste, adding a large rectangle Π to the Harberger Triangle. Furthermore, if monopoly power reduces managerial cost discipline (X-inefficiency, Leibenstein 1966) or enables Price Discrimination, the welfare calculus changes substantially.

Feldstein's (1999) Taxable-Income Elasticity Synthesis

In modern public economics, Martin Feldstein (1999) showed that when an income tax at marginal rate τ distorts not only labor hours (L) but also effort, training, occupational choice, and tax-deductible consumption (such as fringe benefits and mortgage interest), all margins can be collapsed into a single sufficient statistic—the compensated elasticity of taxable income (e_TI) with respect to the net-of-tax share (1 − τ):

Why is deadweight loss a trapezoid rather than a triangle when a tax already exists?

When a market is initially undistorted (τ₀ = 0), the marginal buyer's willingness to pay equals marginal cost (P^d = P^s), so a small tax dτ causes only a second-order triangle loss of order (1 ÷ 2)(dτ)². However, if a pre-existing tax τ₀ > 0 (or monopoly markup) already exists in that market or a related market, there is already a vertical gap τ₀P₀ between marginal social benefit and marginal social cost at Q₁. Any additional tax Δτ eliminates a Harberger Trapezoid of height τ₀P₀ and base ΔQ, generating a large first-order deadweight loss ε_dP₀Q₀τ₀Δτ.

How did Jules Dupuit discover the deadweight loss triangle in 1844?

Working as an inspector of bridges and highways in France, Jules Dupuit (1844) investigated how to measure the social utility of public works such as toll bridges and canals. He defined consumer surplus as the area under the demand curve above the toll price and proved geometrically that raising a bridge toll above zero marginal cost destroys a triangular area of "relative utility" lost to travelers deterred from crossing the bridge.

Does first-degree (perfect) price discrimination eliminate the Harberger Triangle?

Yes. If a monopolist can practice perfect first-degree Price Discrimination by charging each consumer their maximum willingness to pay along the demand curve down to the competitive output Q₀ where P(Q₀) = c, all mutually beneficial trades occur, reducing the Harberger Triangle to zero (though the entire consumer surplus is transferred into the monopolist's profit rectangle).

1. Harberger, A. C. (1954). "Monopoly and Resource Allocation." American Economic Review, 44(2), 77–87. https://www.jstor.org/stable/1818325
2. Harberger, A. C. (1964). "The Measurement of Waste." American Economic Review, 54(3), 58–76. https://www.jstor.org/stable/1818490
3. Tullock, G. (1967). "The Welfare Costs of Tariffs, Monopolies, and Theft." Western Economic Journal, 5(3), 224–232. https://doi.org/10.1111/j.1465-7295.1967.tb01923.x
4. Hines, J. R., Jr. (1999). "Three Sides of Harberger Triangles." Journal of Economic Perspectives, 13(2), 167–188. https://doi.org/10.1257/jep.13.2.167
5. Feldstein, M. (1999). "Tax Avoidance and the Deadweight Loss of the Income Tax." Review of Economics and Statistics, 81(4), 674–680. https://doi.org/10.1162/003465399558391