Most students can quote the Coase theorem. Fewer can say what it rules out, and the exam questions live in that gap. If you are taking intermediate microeconomics or law and economics at HKU, HKUST or CUHK, this Coase theorem lesson from a Hong Kong microeconomics tutor takes one externality, a club’s live music and the recording studio next door, and works it through both assignments of the right, a tax, and a negotiation that costs money. It is organised around the five errors that lose the most marks. One-on-one sessions run through your own problem sets the same way.
The club, the studio and the number four
A club runs live music for H hours a night. Its marginal benefit from the H-th hour, in dollars a night, is 60 − 10H: worth 60 at the start of the evening and nothing by the end of the sixth hour. The studio next door loses bookings when the music runs, and its marginal damage from the H-th hour is 5H, rising as the evening wears on. Nobody else is affected.
Left alone, the club plays until its own marginal benefit hits zero: six hours. Efficiency asks a different question. Each hour is worth playing only if the club’s benefit from it exceeds the studio’s damage from it, so the efficient number of hours solves 60 − 10H = 5H, which gives H = 4, where both sides equal 20. The club’s total benefit from four hours is the area under its line, a trapezoid with parallel sides 60 and 20 over a width of four: 160. The studio’s total damage over the same hours is a triangle, 40. Net surplus is 120. At six hours the club’s benefit is 180 and the damage is 90, so net surplus is 90, which is 30 less. Those 30 dollars are the deadweight loss of the unregulated outcome, and they are also the whole story of the theorem.
Error one: “the polluter should pay”
Coase’s 1960 paper in the Journal of Law and Economics begins its argument by rejecting exactly this reading. The harm is reciprocal. Stopping the music harms the club as surely as the music harms the studio. The question is not who is at fault but which arrangement wastes less, and the theorem’s first claim is that, if bargaining is free, the answer does not depend on who holds the right.
Give the right to the club. The studio must pay to cut the music from six hours to four. The club’s benefit from hours four to six is a trapezoid with sides 20 and 0 over a width of two: 20 dollars. The studio’s damage over the same hours is a trapezoid with sides 20 and 30: 50 dollars. Any payment between 20 and 50 leaves both better off, and the deal recovers the 30 dollars of deadweight loss. Hours stop at four.
Give the right to the studio instead, so the club must buy permission to play at all. The club’s benefit from hours zero to four is 160; the damage from those hours is 40. Any price between 40 and 160 makes both sides better off, and the club buys exactly four hours, because the fifth would cost it more in compensation than it is worth. Hours stop at four again. Same quantity, both ways round. That is the theorem: with well-defined rights and costless bargaining, the efficient allocation is reached whichever way the right is assigned.
Coase’s own illustration was the 1879 case of a confectioner whose machinery disturbed a doctor’s consulting room next door. The court sided with the doctor, but Coase’s point was that the doctor could equally have been made to pay the confectioner to stop, and an efficient outcome would follow either way.
Error two: “so the assignment does not matter”
It does not matter for hours. It matters a great deal for money. Under the first assignment the studio pays the club between 20 and 50. Under the second the club pays the studio between 40 and 160. The owner of the right is the one who gets paid rather than pays. The theorem is a statement about efficiency and says nothing about fairness; a student who writes “the assignment is irrelevant” has dropped half the result.
The invariance of the quantity also needs an assumption that is easy to miss: the two sides value money the same way whether they hold the right or not. If owning the right made the studio wealthier in a way that changed how much quiet it wanted, the efficient quantity itself could shift with the assignment. That is why textbooks state the theorem for quasi-linear preferences, where income effects on the externality are zero. Two businesses counting dollars, as here, satisfy that by construction; two households may not, and then each assignment reaches a different efficient quantity. There is also experimental evidence that people demand more to give up something than they would pay to acquire it, the endowment effect, which Kahneman, Knetsch and Thaler (1990) tested against the theorem directly. Tokens with assigned cash values traded as the theorem predicts. Mugs did not: about half the predicted trades happened, so who ended up with a mug depended on who started with one.
Error three: “transaction costs are a footnote”
They are the main text. Coase wrote the paper to show how much work the zero-transaction-cost assumption does, not to recommend it. Put a cost on negotiating: lawyers, time, the risk of a dispute, say 40 dollars for either deal to go through.
Under the first assignment the surplus from bargaining is 30, less than 40. No deal is worth striking. The club plays six hours and the 30 dollars of deadweight loss stay lost. Under the second assignment the surplus is 120, more than 40. The deal goes through and hours fall to four. Now the assignment decides the quantity.
Before concluding that the studio should hold the right, count the lawyers. With the right at the club, net surplus is 90 and nobody pays to negotiate. With the right at the studio, the bargain reaches four hours but costs 40 to strike, so net surplus is 120 − 40 = 80. The assignment that forces the costly bargain is the worse one. The principle behind the economic analysis of law follows directly: when bargaining is expensive, set the entitlement so that the efficient outcome needs no bargain at all. Here that means a four-hour limit, or the tax of the next section, either of which delivers 120 with no negotiation. When only an all-or-nothing assignment is available, give the right to the party who would have bought it in a costless world.
Numbers of parties make this bite. One club and one studio can negotiate. A factory and five thousand households cannot, because each household has a reason to hold out for a larger share or to let the others pay. That is why the Pigouvian tax and cap-and-trade exist: they are what you do when transaction costs block the Coasean bargain, and the same free-riding logic is the problem that public goods pose.
Error four: “a tax plus a bargain is belt and braces”
The Pigouvian prescription sets a tax per hour equal to the marginal damage at the efficient quantity. Here that is 5 × 4 = 20 dollars an hour. Facing it, the club plays until 60 − 10H = 20, which is four hours. The tax alone gets the efficient answer with no negotiation.
Now suppose the studio can also sue for its damage, or simply pay the club to cut further, with the tax still in place. The club’s marginal benefit after tax is 40 − 10H. The studio’s marginal damage is still 5H, and at four hours it is 20, while the club’s after-tax benefit from the fourth hour is zero. So the studio has every reason to pay the club to cut further, and the two will settle where 40 − 10H = 5H, at 8/3 hours, about 2 hours 40 minutes. Net surplus there is 106.67 dollars, 13.33 less than at the optimum. Two corrections for one externality over-correct. The damage is now priced twice, once by the tax and once by the studio’s money, whether that money arrives as a damages claim or as a side payment, and the club plays too little.
The exam version of this error is the sentence “the government should impose a Pigouvian tax and the parties should then bargain.” They should do one or the other. Coase’s objections to the tax were different: it ignores the victim’s own cheaper adjustments, such as soundproofing or moving, and setting it requires a damage schedule the authority rarely has.
Error five: “the theorem predicts what will happen”
It is a benchmark, not a forecast. Its content is conditional: if rights are clear and bargaining is free, then the outcome is efficient regardless of assignment. Nearly every real externality fails one of the conditions, and the theorem’s use is to tell you which. A right that is not defined cannot be traded, so a fishery that nobody owns is overfished: there is nobody with the power to exclude, and therefore nobody to negotiate with. A right that is defined but held by thousands cannot be traded cheaply, so a tax or a cap does the work instead. A right that is defined and held by two parties with a lawyer between them will usually be traded, and the law’s choice of assignment will move money rather than output.
Coase’s own conclusion was comparative. Markets, firms and governments are alternative ways of organising the same activity, each with its own transaction costs, and the choice between them should be made by comparing total costs rather than by assuming any one of them works for free. The name “Coase theorem” was given later by George Stigler, and Coase was uneasy about the way the zero-cost case came to stand for his argument.
Three problems, each with a computed answer
Problem 1. A bakery’s ovens vent onto a neighbour’s terrace. The bakery’s marginal benefit from the H-th hour of baking is 48 − 6H dollars; the neighbour’s marginal damage is 2H. Find the efficient hours, the hours the bakery picks alone, the Pigouvian tax, and the bargaining range under each assignment of the right.
Answer. Efficiency: 48 − 6H = 2H, so H = 6, where both equal 12. Alone, the bakery bakes until 48 − 6H = 0: eight hours. The tax is the marginal damage at six hours, 12 an hour. If the bakery holds the right, the neighbour pays to cut hours from eight to six: the bakery loses a trapezoid with sides 12 and 0 over two hours, 12; the neighbour saves a trapezoid with sides 12 and 16, 28; the range is 12 to 28 and the surplus is 16. If the neighbour holds the right, the bakery buys six hours: its benefit is a trapezoid with sides 48 and 12 over six hours, 180; the damage is 36; the range is 36 to 180 and the surplus is 144.
Problem 2. Negotiating costs 20 dollars, paid only if a deal is struck. Under which assignment does the efficient quantity occur, what is the deadweight loss if it does not, and which assignment leaves the two parties with more in total?
Answer. With the right at the bakery the surplus is 16, less than 20, so no deal is struck and the bakery bakes eight hours. The deadweight loss is the surplus that went unrealised, 16 dollars. With the right at the neighbour the surplus is 144, so the deal goes through and hours fall to six. But count the negotiation: net surplus at eight hours with no deal is 192 − 64 = 128, while six hours after a 20-dollar bargain nets 144 − 20 = 124. The assignment that reaches the efficient quantity leaves the parties 4 dollars poorer. A six-hour limit would give them 144 with no bargain.
Problem 3. The city imposes the Pigouvian tax of 12 an hour, and the neighbour can also claim damages of 2H per hour. How many hours does the bakery choose, and how much surplus does the double correction destroy?
Answer. After tax the bakery’s marginal benefit is 36 − 6H. It bakes until that equals the damages it must pay, 2H, so H = 4.5. The loss relative to six hours is the area between the two original lines from 4.5 to 6: the gap 48 − 8H is 12 at 4.5 hours and 0 at 6, a triangle of 9 dollars. Over-correction costs less than under-correction here, but it is a loss, and it is the one a good answer catches.
If externalities, property rights and the Coase theorem are on your syllabus at HKU, HKUST or CUHK, these are the derivations that turn a definition into marks, and one-on-one microeconomics tuition in Hong Kong, in sessions timed for your evenings, can take you through your own version of them. HKU microeconomics problem sets lean on exactly this kind of area calculation, and a PhD tutor can make the trapezoids second nature before the exam.
Microeconomics tuition in Hong Kong for the externalities chapter
Tell us the course and the week you are on, and a PhD tutor in microeconomics will set up online sessions around your timetable, working from the notation your lecturer uses. Message us on WhatsApp from the top of this page, or send the form with the problem set attached.