On a Fallacy in the CoaseTheorem
and the Theorem of
Steven N.S. Cheung*
Sixtyyears have lapsed since Ronald Coase published “The Problem of Social Cost,” (fn1)and I am delighted to see Steven G. Medema’s piece entitled “The Coase Theoremat Sixty.” (fn2) This paper runs 125pages in fine prints, citing a total of 856 references. A mind-boggling endeavor, and it reminds meof what George Stigler said of Piero Sraffa’s work on David Ricardo: “Here is atask that needs not be performed again.” (fn3)
Medemaaside, however, there are some errors in circulation on the dates of events,and there are some facts distorted by rumors. As myself an old man now, once befriended by Coase and his associates, Ido possess first-hand information on how the wonderful ideas of Coase cameabout. In doing so I should also pointout a major fallacy in the Coase Theorem. Not meant to belittle that theorem, of course. The Coase Theorem will godown in history as certainly as Say’s Law.
The term“Coase Theorem” was coined by George J. Stigler. He told me in person his view that the CoaseTheorem was the single most important idea in 20th centuryeconomics. It would be difficult to arguewith George on this: he was the leading historian of economic thought of thatentire century.
Section I:The Legendary Debate of the Century
First about dates. “The Problem of SocialCost” appeared in the 1960 issue of the Journal of Law and Economics,but that issue did not appear until the early summer of 1961. On the other hand, the Coase Theorem did notappear in that “social cost” paper—in my view, the so-called “InvariantTheorem” attributed to the 1960 “social cost” paper is not a theorem. Rather, the theorem was first enunciated in1959, in a more beautiful paper by Coase on “The Federal CommunicationsCommission.” (fn4) In that 1959 piece there is one sentence that says: “Thedelineation of rights is an essential prelude to market transactions.” This statement is the Coase Theorem in itscomplete form.
Theintroduction of transaction costs in the following 1960 piece, thoughimportant, only supports the working of the theorem stated in 1959. However, as will be seen later in this paper,it is in the introduction of transaction costs that Coase committed a seriouserror.
There wereconsiderable confusions on the exact issue which led to the legendary debate inAaron Director’s home in the spring of 1960. What is known for certain is that when the FCC paper was submitted toAaron Director at the University of Chicago, a galaxy of stars at Chicagoopposed to a key point Coase made. Exactly what that point was, is a question somefuture historians of economic thought would want to know. Coase told me it had something to do with aparking lot. His strongest critic atthat time, Reuben Kessel, also told me it was about a parking lot. Reuben passed away in 1975, and now I findtwo places in the FCC paper where the phrase “parking lot” is mentioned. I put both of them below for the readers toconsider.
On page 14of the FCC paper, Coase wrote:
If one person could use a piece of land for growing a crop,
and then another person could come along and build a house
on the land used for the crop, and then another could come
along, tear down the house, and use the space as a parking lot,
it would no doubt be accurate to describe the resulting situation
as chaos. But it would be wrongto blame this on private enterprise
and the competitive system. Aprivate-enterprise system cannot
function properly unless property rights are created in resources,
and, when this is done, someone wishing to use a resource has to
pay the owner to obtain it. Chaosdisappears; and so does the
government except that a legal system to define property rights
and to arbitrate disputes is, of course, necessary. But there is
certainly no need for the kind of regulation which we now find
in the American radio and television industry.
Then onpage 25:
It is clear that, if signals are transmitted simultaneously on a given
frequency by several people,the signals would interfere with each
other and would make reception of the messages transmitted by
any one person difficult if not impossible. The use of a piece of
land simultaneously for growingwheat and as a parking lot would
produce similar results. As wehave seen in an earlier section, the
way this situation is avoided is to create property rights (rights, that
is, to exclusive use) in land. The creation of similar rights in the use
of frequencies would enable the problem to be solved in the same
way in the radio industry.
We today would not see anything wrong with thearguments in the two paragraphs above, but in 1959 the understanding of thesocial cost issue was
relatively weak even at Chicago: they were still onthe level with Pigou. (fn5) It was on the issue of the radio frequencyinterference and the parking lot interference that Aaron Director urged Coase, who was then at the University ofVirginia, to come to Chicago to clarify his position. Coase responded that he would not give a lecturebut would be glad to discuss the issue with a selected group of economists.
Thediscussion that turned out to be a heated debate is legendary, and no doubtwill be recorded in the future history of economic thought. There now exist a number of different versionsof that debate, including one of over 100
participants. Myversion must be accurate, because among the 10 participants
I was acquainted with eight of them. The ten participants were Martin Bailey,Milton Friedman, Arnold Harberger, Reuben Kessel, Gregg Lewis, John McGee,
Lloyd Mints, George Stigler, Ronald Coase, and AaronDirector.
Afterdinner, in Aaron’s home, Ronald opened with the question: when a factorypollutes the residents nearby, should this factory be held liable for thedamage? Or should the residents pay thefactory owner to reduce the pollution? The proper answer, as we know it today, is that on efficiency grounds itdoes not matter, so long as the right to pollute or not to pollute is clearlydelineated so that the market will work its magic.
McGee toldme that during the debate Harberger was moving furniture around in Director’shouse to form a fence to check the movements of the cattle, yet Al did notrecall himself doing so. There wererumors that the Coase Theorem was first stated by Milton Friedman, who waspresent in that legendary debate, and not by Coase himself. Yet when I pressed Milton on that rumor, hedenied the glory, saying it all belongs to Coase. Coase, however, told me that after more thantwo hours of debate, he was a bit doubtful of his own thinking, although hestill believed he was right: it was the thirty minutes of Friedman summarizingthe main points of his arguments in that evening, with such great clarity, thathe knew he was home free. Stigler’sversion supports Coase’s version. He told me that after more than two hours ofdebate the issue was undecided. Milton stood up and opened fire, and thebullets hit everybody except Coase. Harry Johnson, who was in London at that time, sent a cable to theEconomics Department at Chicago the next day, saying that he heard anEnglishman has discovered a new continent again. Reuben Kessel, who had been the strongestopponent to Coase’s argument, told me in 1972 that we would have to go all theway back to Adam Smith to find an economist of Coase’s level.
Afterthis legendary debate and as the participants were leaving Director’s home,McGee recalled, they were mumbling to each other that they had just witnessedintellectual history. Aaron Director,who was editor of the Journal of Law and Economics at that time, told metwo matters of interest. First, Coase toldme he was rushing to submit the manuscript to Aaron because the deadline ofsubmission was already passed, so he wrote and submitted the 1960 paper sectionby section. But Aaron told me he could notcare less when would Coase complete that paper: his journal could wait for years if needed be. Second, knowing the Journal of Law and Economicspaid authors at that time, I asked Aaron how much his journal paid Coase forthat 1960 piece, he replied that the University of Chicago regulated the feespaid to authors by the page, otherwise he would give Coase all the money at hisdisposal. Knowing Aaron, my guess isthat he would not mind closing his precious journal down after Coase’s 1960paper.
I must takepride to add here that Aaron liked my works! In the spring of 1969 I presented my paper on the choice of contracts inStigler’s workshop. A day later I alonewas having lunch at the Quadrangle Club at Chicago, I saw Aaron walking slowlytowards me. I politely stood up, and Aaronsaid, “The paper you presented yesterday is the best I have read in severalyears.” Then he turned and walkedaway. I was still standing, and tearscame down from my eyes. And when aftersome 12 yeas of research I finally published “The Contractual Nature of theFirm” (fn6) in 1983, Aaron had someone brought me a simple message: After Steve’spaper, the quarrel on what a firm is exists no more.
Section II: The Unfortunate Neglect of Frank Knight
It is unfortunate that in Coase’s 1959 paperon the FCC and his 1960 paper on social cost, no reference is made on amagnificent 1924 paper, entitled “Some Fallacis in the Interpretation of SocialCost,” written by Frank Knight. (fn7) Knight made exactly the same point in 1924as Coase did in 1959 and 1960. Coasewent to the University of Chicago in 1931, from London on a travellingscholarship, and audited Knight’s lectures. He must be familiar with Knight’s classic work.
Knight’s1924 paper rebuts Pigou’s work on social cost. (fn8) He took issue on Pigou’sfamous example of two roads. Two roads,both going from Town A to Town B, one road is narrow but well-paved, and theother is broad but poorly surfaced. Pigou argued that the congestion of cars willoccur on the good road, with the drivers slowing each other down, hence asocial cost problem exists, and this problem could be resolved if a tax isimposed on the use of the good road, while those using the broad but poor road,though with more cars but still with no congestion, are not harmed. Hence a tax imposed by the government for theuse of the good road will reduce the divergence between private and socialcosts.
Thefollowing comment by Knight on this two-road example is profoundly brilliant,which in my view is a Coase Theorem of the 1924 version.
Professor Pigou’slogic in regard to the roads is, as logic,
quiteunexceptionable. It weakness is onefrequently
met with in economictheorizing, namely that the
assumptions divergein essential respects from the facts
of real economicsituations. The most essential featureof
competitive conditions isreversed, the feature namely, of
the privateownership of the factors practically significant
for production. If the roads are assumed to be subject to
private appropriationand exploitation, precisely the ideal
situation whichwould be established by the imaginary tax
will be broughtabout through the operation of ordinary
Yes, thisis the 1924 version of the Coase Theorem, 36 years before what was
enunciated by R.H. Coase.
To myknowledge, Pigou never replied to Knight’s challenge, except his famoustwo-road example was deleted in the later edition of The Economics ofWelfare.
It isimportant--very important-- to note that by “assumptions” Frank Knight meantthe constraints assumed! And this is thespotlight guiding literally all my own research. I was delighted that in an earlier write-upof Frank Knight in Wikipedia, my name was one among five economistsinfluenced by Knight, but then was disappointed that in a later version onKnight, the names were changed.
I wish to note that in December 1968, in Bob Mundell’slavish cocktail party, I had the honor of meeting Knight and told him in personhow much I adored him and learned from his 1924 paper. He looked at me for a long moment, and said,“That was a long time ago!”
Section III: OceanFishery and the Dissipation of Rent
Another unfortunate omission of Knight’spathbreaking work of 1924 is found in yet another important paper, the one onocean fishery as a common property resource. This beautiful piece, written by H. Scott Gordon in 1954, made noreference to Knight, but the geometric diagrams Gordon used were essentially Knight’sdiagrams, as tilted mirror images and relabeling the axes. (fn 9) What Knightdrew to describe Pigou’s two roads now becomes two fishing grounds in Gordon’spaper.
Theimportant--very important--conclusions in Gordon’s insightful work is that theocean rent that may be captured in fishing is absorbed into the cost of fishinglabor and therefore is dissipated under competition, because the ocean fishingground is not privately owned. To myknowledge, the term “dissipation of rent,” which I use often, was first coinedby Gordon. In my view, the frequent useof this concept of rent dissipation is one distinguishing feature of what laterbecame known as the Washington School of Economics. (fn 10)
However,Gordon’s analysis of the dissipation of rent in ocean fishery is flawed. As I pointed out in my 1970 paper on thestructure of a contract, the complete dissipation of rent in ocean fisheryrequires the number of competing fishing boats be approaching infinity. I reached this interesting result byextending Cournot’s duopoly analysis while allowing free entry with homogeneousfishing inputs. That is, even if the ocean is under common ownership so that nocompeting fishing boat has the right to exclude other entrants, some ocean rentwill be captured by each fishing-boat owner so long as the number of fishingboats is in some way restrained. This ismy explanation why unions of various types are so commonly observed in oceanfishery! In fact, my analysis says thatthe more restrictive it is on the number of fishing boats, the more the oceanrent will be captured by each of the boat owners.
Section IV: A Key Error in the Coase Theorem
Let menow turn to what in my view is a serious error in the Coase Theorem. I take issue with a key statement he made atthe beginning of Section IV of his 1960 paper, when he stated his analysis isbased on “the pricing system is assumed to work smoothly (that is,costlessly).” This is the notedassumption of zero transaction cost and the functioning of the market. However, as pointed out in my work WillChina Go Capitalist?, published in1982 (fn 11), I wrote that if transaction costs were truly zero there would beno market:
If all transactioncosts, broadly defined, were truly zero,
it would have to beaccepted that consumer preferences
would be revealed withoutcost. Auctioneers and monitors
would provide freeall the services of gathering and collating
information; workersand other factors of production would
be directed freelyto produce in perfect accord with
consumer preferences;and each consumer would receive
goods and servicesin conformity with his preferences. The
total income received byeach worker (consumer), as
determined costlesslyby an arbitrator, would equal his
marginal productivityplus a share of the rents of all resources
according to any of a numberof criteria costlessly agreed upon.
In other words,production and consumption activities can in
principle be carriedout without a market, to produce the same
result as thoughcostless markets were in operation.
This viewis important, and Kenneth Arrow immediately agreed with me when he read it. Coase also agreed with me a littlelater. However, the full implications ofthis view took me nearly 25 years to obtain.
Section V: TheTheorem of Transaction Costs Substitution
The greatpuzzle is that there are in fact markets in the real world, and by our dailyobservations there are numerous types of transaction costs associated with thesemarkets. If all transaction costs weretruly zero there would be no market, then it makes no sense to say that marketsexist because of the presence of transaction costs. Why do markets exist after all?
Myjourney to solve this major puzzle involved several steps. First, different types of transaction costsoften cannot be logically separated, as a toll collector at the entrance of ahighway performs both the functions of collecting tolls and policing againstintruders. Second, under thisinseparable rule and pushing this rule to the limit, transaction costs mustinclude all those costs that cannot be conceived to exist in a one-person orRobinson Crusoe economy. On this point George Stigler agreed, and it was laterelaborated in my paper entitled “The Transaction Costs Paradigm”. (fn12)
Third,the dissipation of rent is a cost, and because this dissipation can only be theresult of competition, it cannot be conceived to exist in a one-maneconomy. Therefore, rent dissipation istransaction cost. For example, if theprice of a product is restricted by control to below the market price,customers would have to stand in line for, say, half an hour for a purchase,the value of standing time must be added to the price of the product to thebuyer to obtain its true value. Hencethe value of the product reduced is a dissipation of rent.
Fourth—and this the key--of the numerous criteria that may be used todetermine winners or losers under competition in society, only the market priceentails no dissipation of rent. This isbecause in a free market one has to produce something before he can offer toexchange for something else. Theratio of that exchange is the market price, and because one has to producesomething to participate in this competition game, there is no dissipation ofrent.
Hencecomes a beautiful “Theorem of Transaction Costs Substitution”: In order toreduce rent dissipation under competition in a society, all the transaction costsincurred in the market—lawyers, bankers, policemen, middlemen, etc.— are meantto support the use of the one single criterion of determining winners whichentails no rent dissipation, namely, the market price! In other words, the transaction costsincurred in the market are meant to substitute or reduce the dissipation ofrent—another type of transaction costs—which must arise when the market price isnot used.
The ideathat the market price is the only criterion of determining winners andlosers that entails no dissipation of rent was known to me when I was agraduate student, and an elaborate theoretic treatment of the subject is seenin my piece on price control, published in 1974. (fn 13) However, putting otherelements together to obtain the above Theorem of Transaction Costs Substitutiontook a long time. There are numerousother important implications associated with this theorem, because the choiceamong different contracts necessarily entails transaction costs substitution(fn 14). However, if confined to thesubstitution between the costs of using the market price and the dissipation ofrent, the theorem is relatively simple and straight forward. In my view, this latter substitution is atthe core of the theorem of transaction costs substitution which I propoundhere.
In 1979, when China was just talking aboutopening up, I published an article in the Chinese language, bearing the title “OneThousand Rules, Ten Thousand Rules, in Economics There is Only One Rule.” (fn 15)This piece forcefully argues that of the numerous rules that may be used toallocate resources under competition, nearly all entail rent dissipation. All except one—the market price—which entailsno rent dissipation. Some friends toldme that that article was widely circulated in Beijing, leading to complaintsthat the government turned to charge prices for everything.
It was not easyfor me to convince my Beijing friends, however, that private property rightsare essential for the emergence of markets and the use of market prices. It is at this critical point that Coase’sidea of clear delineation of rights works magic. The Chinese culture isallergic to the word “private”, but clear delineation of rights they were eagerto accept. This is the central contributionof Coase’s works on the economic transformation of China.
It isinteresting to note here that in my 1981 pamphlet bearing the title WillChina Go Capitalist?, which correctly predicted that China will reform tobecome a market economy, the underlying elements of an implicit theoreticstructure is essentially the same as the Theorem of Transaction CostsSubstitution which I discuss here. Ittook more than 30 years to put the elements together to form an integratedtheorem.
Section VI: Episodes toRemember
Inclosing, I would like to recall a fond memory in my last meeting with Coase.
It was in December,1990 when Ronald was awarded theNobel Prize. Because that was the 90thanniversary of that prize, all the living Nobel laureates were invited toStockholm for a massive gathering. Mywife and I were also invited to attend this gathering. The reason is that in the evening before thePrize was awarded, there was a dinner party for all Nobel laureates ineconomics, and I as the only non-winner was asked to give a talk at thatdinner, on behalf of Coase, because Ronald had to rest to prepare for theexcitement coming the next day. Two economistsgave talks during that dinner party, Kenneth Arrow and I myself.
Ronald, Marian,Milton, Rose, my wife and myself were together for several days. Two days before the award ceremony, Ronalddelivered his Nobel Lecture. During thatlecture, my wife and I were arranged to sit next to Rose and Milton. It was a huge hall, filled with people, and athunderous standing applause sounded when Ronald was slowly walking down theaisle towards the podium. We all stoodup, and Milton was standing next to me. I whispered to Milton: “Do you think this guy deserve this prize?” He replied:” You mean Ronald? He should havewon it a long time ago!”
Ronald hada deep feeling for China ever since he was a boy, but had never visitedChina. In 2013, a few months beforeRonald passed away, he was to travel to China to see me. Everything was arranged. His expired passport was renewed, and my wifereserved a hotel suite with a nice view for him and his helper. He was to join us in Shanghai on October 1,the beginning of a ten-day holiday. I had arranged a team of doctors to standby just in case medical assistance was needed. I had also alerted several universities in that region for Ronald tovisit. I wanted Ronald to know howunique a hero he was in the eyes of millions of Chinese youths. China owed this man for his ideas, and Iwanted Ronald to see the gratitude with his own eyes.
But It wasnot to be. Ronald passed away onSeptember 2, 2013, at the age of 102.
*Professoremeritus, University of Hong Kong. The ideas contained in this paper are takenfrom Section IV, Chapter 2, Book V of a five-volume treatise entitled EconomicExplanation, written in the Chinese language. In the preparation of this paper I was assisted by Shihan Shen, Yan Zhou, NingWang, and Gary Shiu.
1. Coase, Ronald. 1960. The Problem of Social Costs,” Volume 3, October, Journal of Law andEconomics, pp. 1 – 44.
2. Medema,Steven. Forthcoming, “Coase Theorem atSixty”, Journal of Economic Literature.
3. Stigler, George.1953. “Sraffa’s Ricardo,” Volume 43, September, American Economic Review,pp. 586-599.
4. Coase, Ronald. 1959. “The Federal CommunicationsCommission,” Volume 2, October, Journal of Law and Economics, pp. 1-40.
5. Pigou, Arthur. 1920. TheEconomics of Welfare. London:Macmillan.
6. Journal of Law andEconomics, Vol. 26, April 1983, pp.1-26.
7. Knight, Frank. 1924. “Some Fallacies in the Interpretationof Social Cost,” Volume 38 Number 4, August, Quarterly Journal of Economics,pp. 582-606.
8. See Footnote 5.
9. Gordon, H. Scott,“The Economic Theory of a Common Property Resource: The Fishery,” Journal ofPolitical Economy, (April, 1954).
10. The existence of a unique Washington School ofEconomics is likely first mentioned by Douglas North in his Institutions,Institutional Change and Economic Performance, published in 1990 by theCambridge University Press. On page 27of the book, he wrote in a footnote that “[t]he transaction cost approach isconsistent only in its agreement on the importance of transaction costs; it isfar from unified in other aspects. Theapproach developed here might most appropriately be characterized as theUniversity of Washington approach, originated by Steven Cheung .” Robert Higgs, who was once on the faculty atUniversity of Washington, wrote in 1991 that “I call this proposition, which isa more sophisticated variant of the modernization hypothesis, the theory of theWashington School. Its prime proponentis Douglas North…North draws on theoretical work on measurement and transactioncosts by Steven N.S. Cheung (formerly University of Washington) and YoramBarzel (still there).” See Robert Higgs,1991, “Eighteen Problematic Propositions in the Analysis of the Growth ofGovernment,” Volume 5, Number 1 and 2, The Review of Austrian Economics,pp. 3-40. Commenting on my role in thefounding of the Washington School, Deirdre McCloskey at the University ofIllinois at Chicago said back in 2017 that, “[h]is main legacy was persuadingDouglas North at Washington to take property rights seriously in economichistory. No Cheung, no North.” The quote is from “ Responses to an Inquiryon S. N. S. Cheung’s Economics after a Conference on the Matter in Shenzhen,China, November 2017”. It could accessed via the following link: http://deirdremccloskey.org/docs/pdf/McCloskey_CheungianEconomics.pdf
11. Cheung, Steven.1981. Will China Go Capitalist? AnEconomic Analysis of Property Rights and Institutional Change. London: Institute of Economic Affairs.
12. S. N. S. Cheung, “The Transaction Costs Paradigm,” Economic Inquiry, October 1998. This is the only work of mine which Milton Friedman gavedetailed comments, word by word, line by line, and rendered the verdict that Ialone occupy a position in this paradigm.
13. See S. N. S. Cheung, “A Theory of PriceControl,” Volume 17, April 1974, Journal of Law and Economics, pp.53-71.
14. SeeS.N.S.Cheung, “Transaction Costs, Risk Aversion, and the Choice of ContractualArrangements,” Volume 12, April 1969, Journal of Law and Economics,pp,23-42.
15. Thepiece is written in Chinese, “千規律，萬規律，經濟規律僅一條”
(One Tousand Rules, Ten TousandRules, in Economics There is Only One
Rule), originally published in October1979 in the Hong Kong Economic
Journal Monthly, Volume 3,Number 7.