Externalities and market failure

Name which failure you are looking at, and what the matching intervention gets wrong.

A market is very good at one thing: finding the price at which everything worth making gets made. It does this without anyone in charge, and the previous lesson showed the arithmetic. This lesson is about the four specific conditions that arithmetic depends on, and what happens to each when the condition fails.

Four failures, and how to tell them apart

Externalities. Someone outside the transaction bears a cost or receives a benefit. The price does not include it, so the quantity is wrong.
Public goods. The thing cannot be withheld from people who do not pay, and one person using it does not use it up. Nobody has a reason to pay for it, so it does not get made even when everybody wants it.
Information asymmetry. One side knows something the other cannot check. Buyers assume the worst, prices fall to match, and the good sellers leave.
Market power. One seller is large enough to set the price rather than take it, and profits by producing less than the crossing would have.

Naming which one you have is most of the work, because each has a different fix and applying the wrong fix makes things worse.

The external cost, drawn

The market below has a cost that neither the buyer nor the seller pays. Add it and watch the true cost line separate from the supply curve.

Buyers keep $625
Sellers keep $625
Harm past what it was worth $50

The market makes 10 more a week than it should, because $10 of the cost of each one lands on somebody who is not buying or selling. A tax of exactly $10 lands on the right quantity.

Cost per unit borne by someone else$10

The red area is the harm done by units whose value to the buyer was less than the damage they caused. It is not the total harm. Most of the pollution comes from units that were genuinely worth making, and the case for stopping those is much weaker.

Worked example

Why the tax equals the harm exactly

Take an external cost of $10 a unit in the market above, which trades 50 units at $35.

The efficient quantity is 40. Past that point, the harm on each extra unit exceeds what the buyer was willing to pay for it.
The market makes 50. Neither party can see the $10, so it does not enter their decision.
The loss is $50. Ten units too many, and the harm on them exceeds their value by an amount that starts at zero and reaches $10 on the last one. Half of ten times ten.
A $10 tax lands exactly on 40. Not approximately. A tax and an external cost are the same wedge between what buyers pay and what sellers keep, so setting one to the size of the other produces the same quantity.

That last point is the whole argument for taxing a harm rather than banning it. A ban takes the quantity to zero and gives up all $1,250 of value to avoid $500 of harm. The tax gives up $50.

It also depends completely on knowing that the harm is $10, and this is where the argument actually happens. Nobody disputes the diagram. People dispute the number.

The figures are exact for the straight-line market drawn above, and the tax result is checked by a test that runs the tax and the externality through separate code paths and asserts they land on the same quantity.

ConventionA tax set to the size of an external cost is called Pigouvian, after Arthur Pigou, who argued for it in 1920. The rival tradition comes from Ronald Coase, who pointed out in 1960 that if the parties can bargain cheaply and somebody clearly owns the right, they will sort it out themselves and the government does not need the number. Both are correct, and which one applies depends on how many people are affected and how easily they can find each other.

The other three, briefly

Public goods fail on a different axis. A lighthouse helps every ship, including the ones that did not pay, so no shipowner has a reason to build one and every one of them wants it built. National defense, basic research, and clean air work the same way. The usual answer is to fund it collectively and argue about the amount, which is what a large part of every budget debate actually is.

Information asymmetry was set out by George Akerlof using used cars. Buyers cannot tell a good one from a bad one, so they will only pay the average price. That price is too low for the owners of good cars, who withdraw, which lowers the average, which pushes out the next tier. The market can unravel entirely. The fixes are all about making the hidden thing visible: warranties, inspections, brands, licensing, and disclosure rules.

Market power is the failure that looks most like success from the inside. A firm with no close competitor makes more money by producing less and charging more, and the units it declines to produce were worth making. The output is lower and the price is higher than the crossing, and the difference is the same triangle as every other failure on this page.

What breaks

Naming a market failure is not the same as showing the fix is better. The government has to estimate the harm, and it faces its own incentives, its own capture by the industry it regulates, and its own delay. A comparison between an imperfect market and a perfect regulator is not a comparison.

Externality arguments are unfalsifiable if you let them be. Almost any transaction affects somebody outside it somehow. The question that keeps the concept useful is whether the effect runs through a price. Your neighbor opening a better restaurant hurts the existing one, and that is competition doing its job rather than an externality. Pollution does not run through a price, which is exactly why it needs one.

The efficient amount of a harm is almost never zero. Getting road deaths to zero means a speed limit of about 5 miles an hour. Everyone already accepts a trade between harm and value in practice while talking as though they do not, and the argument gets much clearer once the trade is stated out loud.

What to do with this

When somebody says a market has failed, ask which of the four they mean. If they cannot say, the claim is usually that they dislike the outcome, which is a legitimate position and a different one.

Then ask what the fix has to know in order to work. A Pigouvian tax needs the size of the harm. An antitrust case needs to know what the price would have been. A disclosure rule needs the disclosure to be readable by the person it is for. The fix that needs the least information to get roughly right is usually the one to prefer, because the information is the part that is missing.

Test yourself

01A factory pollutes a river. Name three fixes and say what each one gets wrong.

Tax the pollution. Gets the quantity right if you know the size of the harm, and you usually do not. Set it too low and too much pollution continues. Set it too high and you have shut down production that was worth more than the damage it caused.

Ban it. Certain, cheap to enforce, and it throws away every unit of output whose value exceeded its harm. A ban is the right answer when the harm is severe and hard to measure, and the wrong one when the harm is moderate and the output is valuable.

Assign the river to somebody. If the people downstream own the water, the factory has to buy the right to use it and the price does the work. It fails when there are ten thousand downstream parties who cannot organize themselves to negotiate, which is usually.

02Why does health insurance need either a mandate or heavy regulation to work, when car insurance mostly does not?

Because the buyer knows more about their own health than the insurer can find out. Healthy people look at the average price, decide it is a bad deal, and leave. The pool gets sicker, the price rises, and the next-healthiest tranche leaves. The market can unravel to nothing even though almost everyone would have preferred to be insured.

Car insurance has less of this because the insurer can observe most of what matters: your age, your record, your car, where you park it. When the seller can see the risk, they can price it, and the pool does not unravel.

03Give an example of a positive externality, and say what the market does wrong there.

Vaccination. Every person vaccinated makes everyone around them slightly safer, and none of that benefit shows up in what they were willing to pay. So the private decision undercounts the value and the market produces too few. Every fix runs the other way from the pollution case: subsidize, mandate, or provide it free. The error is symmetric, which is why the same diagram handles both with the wedge pointing the other way.

A tutor that knows this lesson. It asks before it explains, and it will not tell you what to do with your own money.

Educational material, not investment or policy advice. Figures are cited where they come from a filing or a statistical series, and labelled as illustrative where they do not.