The Pigouvian Tax: Making Polluters Pay the True Cost

The problem: private cost ≠ social cost

In a competitive market, a firm produces up to the point where its private marginal cost equals the price — the marginal benefit to buyers. That's efficient only if the firm bears every cost of production. It doesn't when it pollutes. A coal plant pays for coal, labour, and capital, but the asthma, crop damage, and climate harm land on third parties who never signed the contract. That uncompensated harm is a negative externality.

Write it as three curves. The marginal private cost (MPC) is what the firm actually pays. The marginal external cost (MEC) is the harm per unit imposed on outsiders. Their sum is the marginal social cost:

MSC(q) = MPC(q) + MEC(q)

The market clears where price = MPC. But the socially efficient quantity is where price (= marginal social benefit) = MSC. Because MSC sits above MPC, the free market produces too much: q_market > q*. Every unit between q* and q_market costs society more than it's worth — that gap, summed up, is the deadweight loss. The market isn't 'broken' by greed; it's optimizing over the wrong cost.

Pigou's fix: set the tax equal to the marginal external cost

Pigou's insight (1920) was elegant: don't ban the activity, reprice it. Impose a per-unit tax τ exactly equal to the marginal external cost evaluated at the efficient quantity:

τ* = MEC(q*)

Now the firm's perceived cost of producing one more unit becomes MPC + τ* = MPC + MEC(q*) ≈ MSC. The supply curve lifts by the tax wedge until it coincides with the true social-cost curve. The firm still maximizes its own profit — nothing coercive — but its private optimum now lands on q\*. The externality has been internalized: the polluter now feels, in its own ledger, the cost it used to push onto others.

Crucially the tax doesn't drive pollution to zero. It equalizes the marginal cost of abating the last unit with the marginal damage of emitting it. Cheap-to-cut pollution gets cut; pollution that is ruinously expensive to eliminate but does little harm survives — which is exactly the efficient allocation. The tax also raises revenue, but that revenue is a transfer, not the point; the welfare gain is the recovered deadweight loss.

A worked example with real numbers

Suppose demand for electricity from a coal plant is P = 100 − Q and private marginal cost is MPC = 20 + Q (P and MPC in \$/MWh, Q in thousands of MWh). Each MWh also inflicts MEC = 30 in external health and climate damage.

• Free market: set P = MPC → 100 − Q = 20 + Q → Q_market = 40, price \$60.
• Social optimum: MSC = MPC + 30 = 50 + Q. Set demand = MSC → 100 − Q = 50 + Q → Q\* = 25, price \$75.
• The overproduction: 40 − 25 = 15 thousand MWh too many.

The Pigouvian tax is τ* = MEC = \$30/MWh. Adding it to supply gives 50 + Q, which clears at exactly Q = 25. The deadweight loss it removes is the triangle between MSC and demand over the excess 15 units: ½ × 15 × (harm gap) = ½ × 15 × 30 = \$225k of net welfare recovered. Tax revenue is \$30 × 25 = \$750k — a transfer to the public purse, on top of the efficiency gain. The tax equals the marginal damage, \$30; note that setting it higher — say \$40 — would push output to Q = 40 − 40/2 = 20, below q* = 25, and create a new deadweight loss on the other side.

Where it shows up in the real world

Almost every 'green tax' is Pigouvian in spirit:

• Carbon taxes. Sweden introduced one in 1991; it now runs around €120/tonne CO₂, among the world's highest, while the economy grew ~80% and emissions fell ~30%. Canada, Switzerland, and the UK's Carbon Price Support all follow the same logic. Estimates of the social cost of carbon — the ideal τ — range from the US EPA's ~\$190/tonne (2023) down to older \$40–51 figures.
• Congestion charges. London's 2003 charge (now £15/day) taxes the externality one driver imposes on all the others (delay, pollution). Singapore's Electronic Road Pricing dynamically varies the toll — a near-textbook Pigouvian instrument.
• Sin and 'internality' taxes. Tobacco, alcohol, and sugary-drink taxes (e.g. Mexico's 2014 soda tax, ~10% price rise, ~6–8% drop in purchases) target external and self-inflicted costs.
• Waste and fuel. The UK Landfill Tax, plastic-bag fees, and fuel duties all price disposal or emission externalities.

The reverse case exists too: a Pigouvian subsidy pays for positive externalities — vaccination, R&D, education, rooftop solar — where private actors under-provide because they can't capture the full social benefit.

The catch: you have to know the damage curve

The theory is exact; the world is not. The tax is only optimal if the regulator knows MEC(q*) — the dollar value of harm per unit — and that is genuinely hard. What is a tonne of CO₂ worth? The answer depends on the discount rate, climate sensitivity, and how you value future and foreign lives; credible estimates span an order of magnitude. Set τ too low and pollution stays excessive; too high and you crush output below q* and manufacture a new deadweight loss.

Three further wrinkles:

• Heterogeneous polluters and places. A tonne of SO₂ over a city harms more than one over open ocean. A single national tax rate ignores this; the true Pigouvian tax is location- and source-specific, which is administratively brutal.
• General equilibrium and the double dividend. Pigou's partial-equilibrium picture ignores existing taxes. Recycling green-tax revenue to cut distortionary income or payroll taxes can yield a 'double dividend' — but the tax-interaction effect can also shrink the gain, a subtlety formalized by Bovenberg & de Mooij (1994).
• Political economy. Concentrated polluters lobby; diffuse victims don't. That's why real carbon prices sit below most social-cost-of-carbon estimates.

The rival view: Coase and cap-and-trade

Pigou is not the only remedy. Ronald Coase (1960) argued that if property rights are clear and bargaining is costless, the parties can negotiate the efficient outcome without any tax — the polluter and the victim will trade their way to q* on their own, and the same quantity results regardless of who holds the right. Coase's real target was the assumption that a tax is always needed; in his view, high transaction costs (many diffuse victims, as with global CO₂) are what justify intervention. So Pigou and Coase are complements: bargaining works for a factory and one downstream fishery; a tax works for seven billion people and the atmosphere.

The other great alternative is cap-and-trade. Instead of fixing the price of pollution (a tax), you fix its quantity (a cap) and let firms trade permits, so the market discovers the price. Weitzman's classic 1974 result 'Prices vs. Quantities' shows which is better depends on curvature: if the marginal-damage curve is steep (a tipping point), fix the quantity (cap); if marginal abatement cost is steep and damage is flat, fix the price (tax). The US SO₂ program (1990) and the EU ETS (2005) chose quantity; British Columbia and Sweden chose price.

Two market-based ways to price pollution: a Pigouvian tax fixes the price, cap-and-trade fixes the quantity.

Feature	Pigouvian Tax	Cap-and-Trade (Permits)
Instrument fixes	Price per unit of pollution	Total quantity of pollution
Who bears price risk	Regulator (emissions uncertain)	Firms (permit price swings)
Revenue	To government (can cut other taxes)	Only if permits are auctioned
Best when	Damage curve flat, abatement cost steep	Damage curve steep, near a tipping point
Real example	Sweden CO₂ tax (1991), UK landfill tax	EU ETS (2005), US SO₂ program (1990)