William M. Landes and Richard A. Posner,*

An Economic Analysis of Copyright Law

18 J. Leg. Stud. 325, 325-33, 344-53 (1989)

Intellectual property is a natural field for economic analysis of law, n2 and copyright is an important form of intellectual property. Yet while there are good introductions to the economics of copyright law, and a number of excellent articles on the economics of copying (as distinct from copyright law), n3 no article examines the field of copyright as a whole, discussing the evolution and major doctrines in the law from an economic standpoint. This article, which is in the spirit of our recent articles on the economics of trademark law, tries to fill this gap—although the field is so vast that our analysis cannot be exhaustive. As in most of our work, we are particularly interested in positive analysis, and specifically in the question to what extent copyright law can be explained as a means for promoting efficient allocation of resources.

A distinguishing characteristic of intellectual property is its "public good" aspect. While the cost of creating a work subject to copyright protection—for example, a book, movie, song, ballet, lithograph, map, business directory, or computer software program—is often high, the cost of reproducing the work, whether by the creator or by those to whom he has made it available, is often low. And once copies are available to others, it is often inexpensive for these users to make additional copies. If the copies made by the creator of the work are priced at or close to marginal cost, others may be discouraged from making copies, but the creator’s total revenues may not be sufficient to cover the cost of creating the work. Copyright protection—the right of the copyright’s owner to prevent others from making copies—trades off the costs of limiting access to a work against the benefits of providing incentives to create the work in the first place. Striking the correct balance between access and incentives is the central problem in copyright law. For copyright law to promote economic efficiency, its principal legal doctrines must, at least approximately, maximize the benefits from creating additional works minus both the losses from limiting access and the costs of administering copyright protection.

Section I develops the basic economic model of copyright protection, including an analysis of the optimal degree of that protection. Section II applies the model to the principal doctrines of copyright law. It considers such questions as the originality requirement for copyright protection, the distinction between ideas and expression, the absence of copyright protection for utilitarian works, the protection of derivative works, and issues of fair use.


I. The Basic Economics of Copyright

We begin with the factors—including, of course, copyright protection—that determine the number of works created. Then we examine the exploitation of the created work—the number of copies and the price per copy. Last, we incorporate the important features of our discussion into a model of the optimal degree of copyright protection.

A. Number of Works as a Function of Copyright and Other Factors

1. General Considerations

The cost of producing a book or other copyrightable work (we start by talking just about books and later branch out to other forms of expression) has two components. The first is the cost of creating the work. We assume that it does not vary with the number of copies produced or sold, since it consists primarily of the author’s time and effort plus the cost to the publisher of soliciting and editing the manuscript and setting it in type. Consistent with copyright usage we call the sum of these costs the ‘‘cost of expression."

To simplify the analysis, we ignore any distinction between costs incurred by authors and by publishers, and therefore use the term "author" (or "creator") to mean both author and publisher. In doing this we elide a number of interesting questions involving the relation between author and publisher. For example, do such principles as droit moral, entitling authors to reclaim copyright from assignees after a fixed period of years or entitling artists to royalties on resales of their art by initial (or subsequent) purchasers, n4 increase or reduce the incentive to create new works? The answer suggested by economic analysis is that, contrary to intuition, such principles reduce the incentive to create by preventing the author or artist from shifting risk to the publisher or dealer. A publisher (say) who must share any future speculative gains with the author will pay the author less for the work, so the risky component of the author’s expected remuneration will increase relative to the certain component. If the author is risk averse, he will be worse off as a result. However, we do not explore such matters in this article.

The second component of the cost of producing a work increases with the number of copies produced, for it is the cost of printing, binding, and distributing individual copies. The cost of expression does not enter into the making of copies because, once the work is created, the author’s efforts can be incorporated into another copy virtually without cost.

For a new work to be created, the expected return—typically, and we shall assume exclusively, from the sale of copies—must exceed the expected cost. The demand curve for copies of a given book is, we assume, negatively sloped because there are good but not perfect substitutes for a given book. n5 The creator will make copies up to the point where the marginal cost of one more copy equals its expected marginal revenue. The resulting difference between price and marginal cost, summed over the number of copies sold, will generate revenues to offset the cost of expression. Since the decision to create the work must be made before the demand for copies is known, the work will be created only if the difference between expected revenues and the cost of making copies equals or exceeds the cost of expression. If we assume that the cost of creating (equivalent) works differs among authors, the number of works created will increase until the returns from the last work created just covers the (increasing) cost of expression.

Two qualifications should be noted. First, for many types of intellectual property some price discrimination may be possible because individual works are not perfect substitutes and arbitrage is preventable. Thus, a book publisher will commonly charge higher prices for hardcover editions and later reduce the price for persons willing to wait for the paperback edition to appear. Similarly, the prices charged by exhibitors for first-run movies will generally be higher than the prices in the aftermarket (cable viewing, video cassettes, and network television). Price discrimination increases revenue and thus the number of works produced, though it may not increase the number of copies of each work. n6 Second, the demand for copies of a given work depends not only on the number of copies but on the number of (competing) works as well. The greater the number of such works (past and present), the lower the demand for any given work. Thus, the number of works and the number of copies per work will be determined simultaneously, and the net effect of this interaction will be to reduce the number of works created.

This description of the market for copies and the number of works created assumes the existence of copyright protection. In its absence anyone can buy a copy of the book when it first appears and make and sell copies of it. The market price of the book will eventually be bid down to the marginal cost of copying, with the unfortunate result that the book probably will not be produced in the first place, because the author and publisher will not be able to recover their costs of creating the work. The problem is magnified by the fact that the author’s cost of creating the work, and many publishing costs (for example, editing costs), are incurred before it is known what the demand for the work will be. Uncertainty about demand is a particularly serious problem with respect to artistic works, such as books, plays, movies, and recordings. Even with copyright protection, sales may be insufficient to cover the cost of expression and may not even cover the variable cost of making copies. Thus, the difference between the price and marginal cost of the successful work must not only cover the cost of expression but also compensate for the risk of failure. If a copier can defer making copies until he knows whether the work is a success, the potential gains from free riding on expression will be even greater, because the difference between the price and marginal cost of the original work will rise to compensate for the uncertainty of demand, thus creating a bigger profit potential for copies. So uncertainty generates an additional disincentive to create works in the absence of copyright protection.

Practical obstacles limit copying the original works of others even in the absence of any copyright protection. But these obstacles, while serious in some cases, can easily be exaggerated. When fully analyzed, they do not make a persuasive case for eliminating copyright protection.

1. The copy may be of inferior quality, and hence not a perfect substitute for the original. In the case of books and other printed matter, the copier may not be able to match the quality of paper or binding of the original or the crispness of the printing, and there may be errors in transcription. None of these is an important impediment to good copies any longer, but in the case of works of art—such as a painting by a famous artist—a copy, however accurate, may be such a poor substitute in the market that it will have no negative effects on the price of the artist’s work. Indeed, the copy may have a positive effect on that price, by serving as advertising for his works. On the other hand, it may also deprive him of income from selling derivative works—the copies of his paintings—himself. (More on derivative works shortly.) To generalize, when either the cost of making equivalent copies is higher for the copier than for the creator or the copier’s product is a poor substitute for the original, the originator will be able to charge a price greater than his marginal cost, even without legal protection. And, obviously, the greater the difference in the costs of making copies and in the quality of copies between creator and copier (assuming the latter’s cost is higher or quality lower), the less need there is for copyright protection.

2. Copying may itself involve some original expression—as when the copy is not a literal copy but involves paraphrasing, deletions, marginal notes, and so on—and so a positive cost of expression. The copier may incur fixed costs as well, for example costs of rekeying the words from the copy he bought or of photographing them. Still, we would expect the copier’s average cost to be lower than the creator’s because it will not include the author’s time or the cost of soliciting and editing the original manuscript. Nevertheless, when the copier cannot take a completely free ride on the creator’s investment in expression and his other fixed costs, the need for copyright protection is reduced.

Between the literal copier and the author who makes no use whatever of previous works, three additional types of producer can be distinguished. One is the author who makes at least some, but perhaps modest, use of previous works; most authors are of this type. Next is the author of a derivative work, that is, a work that draws very heavily on previous works, though the derivative work involves some original elements. Third is the unoriginal copier who nevertheless tries to complicate the author’s task of proving infringement by differentiating the copied work from the original in minor ways. Derivative works and infringement are discussed in Section II; the author who makes some use of previous works figures prominently in our formal analysis, along with the literal copier.

3. Copying takes time, so there will be an interval during which the original publisher will not face competition. This point, which is related to the first because generally the cost of production is inverse to time, n7 has two implications for the analysis of copyright law. First, because modern technology has reduced the time it takes to make copies as well as enabled more perfect copies to be made at low cost, the need for copyright protection has increased over time. Second, for works that are faddish—where demand is initially strong but falls sharply after a brief period—copyright protection may not be as necessary in order to give the creator of the work a fully compensatory return.

4. There are contractual alternatives to copyright protection for limiting copying. One is licensing the original work on condition that the licensee not make copies of it or disclose it to others in a way that would enable them to make copies. But contractual prohibitions on copying may, like trade secrets, be costly to enforce and feasible only if there are few licensees. Where widespread distribution is necessary to generate an adequate return to the author or where the work is resold or publicly performed, contractual prohibitions may not prevent widespread copying. Thus, the greater the potential market for a work, the greater the need for copyright protection. The development of radio, television, and the phonograph has expanded the market for copies and thereby increased the value of copyright protection.

5. Since a copier normally must have access to a copy in order to make copies, the creator may be able to capture some of the value of copies made by others by charging a higher price for the copies he makes and sells. For example, a publishers of academic journals may be able to capture part of the value that individuals obtain from copying articles by charging a higher price because of home taping. n8 Although this possibility limits the need for copyright protection, it does not eliminate it. If one can make many copies of the first copy, and many copies of subsequent copies, the price of copies will be driven down to marginal cost and the creator will not be able to charge a sufficiently higher price for his copy to capture its value in allowing others to make more copies; no one (except the first copier and the most impatient readers) will buy from him rather than from a copier.

6. Many authors derive substantial benefits from publication that are over and beyond any royalties. This is true not only in terms of prestige and other nonpecuniary income but also pecuniary income in such forms as a higher salary for a professor who publishes than for one who does not, n9 or greater consulting income. Publishing is an effective method of self-advertisement and self-promotion. The norms against plagiarism (that is, against copying without giving the author credit) reinforce the conferral of prestige by publishing; to the extent that those norms are effective, they ensure that the author will obtain recognition, if not always royalties, from the works he publishes.

Such points have convinced some students of copyright law that there is no need for copyright protection. n10 Legal rights are costly to enforce—rights in intangibles especially so—and the costs may outweigh the social gains in particular settings. Perhaps copyright in books is one of them. After all, the first copyright law in England dates from 1710 (and gave much less protection than modern copyright law), yet publishing had flourished for hundreds of years in England despite censorship and wide-spread illiteracy. n11 The point is a little misleading, however. In the old days, the costs of making copies were a higher fraction of total cost than they are today, so the problem of appropriability was less acute. Also, there were alternative institutions for internalizing the benefits of expression. n12 And before freedom of expression became generally applauded, publishing was often believed to impose negative externalities—so there was less, sometimes no, desire to encourage it. Finally, while it may be difficult to determine whether, on balance, copyright is a good thing, it is easy to note particular distortions that a copyright law corrects. Without copyright protection, authors, publishers, and copiers would have inefficient incentives with regard to the timing of various decisions. Publishers, to lengthen their head start, would have a disincentive to engage in prepublication advertising and even to announce publication dates in advance, and copiers would have an incentive to install excessively speedy production lines. There would be increased incentives to create faddish, ephemeral, and otherwise transitory works because the gains from being first in the market for such works would be likely to exceed the losses from absence of copyright protection. There would be a shift toward the production of works that are difficult to copy; authors would be more likely to circulate their works privately rather than widely, to lessen the risk of copying; and contractual restrictions on copying would multiply

A neglected consideration—one that shows not that copyright protection may be unnecessary but that beyond some level copyright protection may actually be counterproductive by raising the cost of expression—will play an important role both in our model and in our efforts to explain the salient features of copyright law. Creating a new work typically involves borrowing or building on material from a prior body of works, as well as adding original expression to it. A new work of fiction, for example, will contain the author’s expressive contribution but also characters, situations, plot details, and so on, invented by previous authors. Similarly, a new work of music may borrow tempo changes and chord progressions from earlier works. The less extensive copyright protection is, the more an author, composer, or other creator can borrow from previous works without infringing copyright and the lower, therefore, the costs of creating a new work. Of course, even if copyright protection effectively prevented all unauthorized copying from a copyrighted work, authors would still copy. But they would copy works whose copyright protection had run out, or they would disguise their copying, engage in costly searches to avoid copying protected works, or incur licensing and other transaction costs to obtain permission to copy such works. The effect would be to raise the cost of creating new works—the cost of expression, broadly defined—and thus, paradoxically, perhaps lower the number of works created. n13

Copyright holders might, therefore, find it in their self-interest, ex ante, to limit copyright protection. To the extent that a later author is free to borrow material from an earlier one, the later author’s cost of expression is reduced; and, from an ex ante viewpoint, every author is both an earlier author from whom a later author might want to borrow material and the later author himself. In the former role, he desires maximum copyright protection for works he creates; in the latter, he prefers minimum protection for works created earlier by others. In principle, there is a level of copyright protection that balances these two competing interests optimally—although notice that the first generation of authors, having no one to borrow from, will have less incentive to strike the optimal balance than later ones. n14 We shall see in Section II that various doctrines of copyright law, such as the distinction between idea and expression and the fair use doctrine, can be understood as attempts to promote economic efficiency by balancing the effect of greater copyright protection—in encouraging the creation of new works by reducing copying—against the effect of less protection—in encouraging the creation of new works by reducing the cost of creating them.

[Presentation of a formal model is omitted.]

II. Applications


A. The Nature of Copyright Protection

Our task now is to use our model to explain the principal features of copyright law. We begin with the nature of the protection that a copyright gives its owner. In contrast to a patent, a copyright merely gives protection against copying; independent (that is, accidental) duplication of the copyrighted work is not actionable as such. In speaking of "independent [accidental, inadvertent] duplication" we are addressing only the problem of an independent recreation of the original copyrighted work. The accidental use of someone else’s work might be thought of as duplication, but in that context liability for infringement is strict, much as it is for the trespass on a neighbor’s land made by a person who thinks that he owns it.

The more difficult question is to explain why duplication in the sense of independent recreation is not actionable. Our analysis suggests two possible explanations. The first is the added cost to the author of checking countless numbers of copyrighted works to avoid inadvertent duplication. The costs (if actually incurred—a qualification whose significance will become apparent shortly) would increase e(z) and lower social welfare because both net welfare per work and the number of works created would fall. True, the author’s gross revenues might rise if the reduction in the amount of accidental duplication raised the demand for copies or made that demand less elastic. n29 But since accidental duplication of copyrighted works is rare (except in the area of popular music, discussed below), the net effect of making it unlawful would be to lower social welfare.

In contrast to copyright, accidental infringements of patents are actionable, and the difference makes economic sense. A patent is issued only after a search by the applicant and by the Patent Office of prior patented inventions. This procedure is feasible because it is possible to describe an invention compactly and to establish relatively small classes of related inventions beyond which the searchers need not go. The procedure makes it relatively easy for an inventor to avoid accidentally duplicating an existing patent.

No effort is made by the Copyright Office to search copyrighted works before issuing a copyright, so copyright is not issued but is simply asserted by the author or publisher. There are billions of pages of copyrighted material, any one page of which contain a sentence or paragraph that a later writer might, by pure coincidence, duplicate so closely that he would be considered an infringer if he had actually copied the words in question or if copying were not required for liability. What is infeasible for the Copyright Office is also infeasible for the author. He cannot read all the copyrighted literature in existence (in all languages, and including unpublished works!) in order to make sure that he has not accidentally duplicated some copyrighted material.

The cost of preventing accidental duplication would be so great, and the benefits in terms of higher revenues (and so the amount of damages if such duplication were actionable) so slight because such duplication is rare, that even if it were actionable no writer or publisher would make much effort to avoid accidental duplication, so the increase in the cost of expression would probably be slight. But social welfare would be reduced somewhat. At best we would have a system of strict liability that had no significant allocative effective; and as explained in the literature on negligence and strict liability in tort law, the costs of enforcing such a regime are socially wasted because their only product is an occasional redistribution of wealth (here that would be from the accidental "infringer" to the first author or publisher of the material duplicated). n30

The second reason we expect accidental duplication not to be made unlawful derives from the economic rationale for copyright protection, which is to prevent free-riding on the author’s expression. Accidental duplication does not involve free-riding. Since the second work is independently created, its author incurs the full cost of expression. If the works are completely identical—a remote possibility, to say the least n31—competition between the two works could drive the price of copies down to marginal cost and prevent either author from recovering his cost of creating the work. It is more likely that significant differences between the two works will remain, so that both authors may be able to earn enough to cover their respective costs of expression—particularly if neither author is the marginal author, whose gross revenues would just cover the cost of expression in the absence of accidental duplication.

Although for simplicity our analysis focuses on copyright protection for literature and other written works, it is applicable, mutatis mutandis, to other forms of expression as well. A significant difference between literary and musical copyright is that courts hold that accidental duplication may infringe a songwriter’s copyright if his song has been widely performed. n32 Since most popular songs have simple melodies and the number of melodic variations is limited, the possibility of accidental duplication of several bars is significant. Widespread playing of these songs on the radio makes it likely that the second composer will have had access to the original work, which both increases the likelihood of accidental duplication and reduces the cost of avoiding it. If proof of intentional duplication were required for infringement, composers of popular songs would have little copyright protection and social welfare would fall.

This result may appear to show that musical copyright follows the pattern of patent law rather than of literary copyright, but the appearance is deceptive. Two forms of "accidental’’ duplication of a copyrighted work must be distinguished. The first is independent creation: the duplicator makes no use, direct or indirect, of the copyrighted work. Such a duplicator is never an infringer. The second is unconscious borrowing. Unlike the independent creator, the unconscious borrower is a free rider, and is therefore properly deemed an infringer. Musical copyright is special only in that unconscious borrowing is more likely in the musical than in the literary realm.

To distinguish between copying and independent duplication, the courts use an implicit indifference-curve analysis in which access is traded off against similarity. Where there is a strong showing of similarity, it is more likely that the original work was copied and less likely that it was independently created (particularly for complicated works, in contrast to, say, a few bars of a popular melody). In such a case, the copyright owner can prevail even if he presents only weak evidence that the defendant had access to the original work. On the other hand, where the differences between the original work and its "copy" are substantial, the copyright owner will normally have to provide strong evidence of access to rebut the defense of independent creation.

B. The Scope of Protection

The most difficult economic questions about copyright law have to do with the scope of legal protection. We shall discuss these under the headings of (1) idea versus expression, (2) derivative works, and (3) fair use.

1. Idea versus Expression

i) The economic rationale for not protecting ideas. A copyright protects expression but not ideas. n33 Postponing to the next section the problem of distinguishing between the two, we offer several reasons for the difference in treatment.

a) Suppose that the N works in our model express the same idea differently; for example, each work might be a different novel about a romance between young people who belong to different social classes or religious faiths and whose parents are feuding. If copyright protected the first author’s idea, the cost of expression to each of the remaining N - 1 authors would increase, because each would have to invest time and effort in coming up with an original idea for his work, or to substitute additional expression for the part of his idea that overlapped the first author’s, or to incur licensing and other transaction costs to obtain the right to use the first author’s idea. n34 The net effect of protection would be to reduce the number of works created—in the limit, to one, although the Coase Theorem makes this outcome unlikely—so that social welfare in equation (16) would fall. Granted, the copiers’ cost would also rise if a copyright protected ideas, because copiers use not only the author’s expression but also his ideas. However, this offset is likely to be small. Copiers are copying expression either unlawfully, in which case the marginal deterrence from protecting ideas is likely to be small, or lawfully, for example, because their copying is deemed a fair use. In either case, copyright protection for ideas would have a negligible effect on the copier’s cost of copying. Even if protection did increase the gross profits of the N - 1 authors and thus offset partially the reduction in N, social welfare would fall because both N and welfare per work would fall.

The traditional explanation for protecting only expression emphasizes the welfare losses from monopoly of an idea. We emphasize the increase in the cost of creating works and the reduction in the number of works rather than the higher price (per copy) that is normally associated with monopoly. In a more complicated model the demand for copies would depend not only on the number of copies (our model) but also on the number of competing works. In that case, protecting ideas would raise the price of copies.

b) The author is both a user of ideas developed by others and a creator of new ideas. Suppose our N authors did not know which one would be the first to come up with an idea that the other N - 1 authors would use. Since (as we shall see in the next section) the costs involved in coming up with the kind of new idea normally embodied in an expressive work usually are low relative to the costs in time and effort of expressing the idea, and since the originator of the idea will probably obtain a normal return in one form or another from being first in the market even without receiving copyright protection, the N authors, behind a veil of ignorance, probably would agree unanimously (or nearly so) to a rule that protected expression but not ideas. If so, then that rule would be Pareto optimal.

c) Another objection to copyright protection of ideas is that it would encourage rent seeking. Since the costs of developing a new idea are likely to be low in most cases relative to the potential reward from licensing the idea to others, there would be a mad rush to develop and copyright ideas. Resources would be sucked into developing ideas with minimal expression, and the ideas thus developed would be banked in the hope that a later author would pay for their use. Although the development of new ideas would be accelerated, the dissemination of ideas might not be.

d) A final concern is with the administrative costs involved in defining rights in ideas. Courts would have to define each idea, set its boundaries, determine its overlap with other ideas, and, most difficult of all, identify the idea in the work of the alleged infringer. Yet the total administrative and enforcement costs of operating a copyright system might actually shrink if ideas were protected, because of the significant reduction in the number of works created. Since the optimal level of copyright protection for expression takes account of the costs of enforcement, our first point— the decline in social welfare from fewer works—incorporates the savings in administrative and enforcement costs from few works. The present point, therefore, concerns the increase in these costs if the number of works is held constant.

ii) Distinguishing ideas from expression. Although the line between expression and idea is often hazy, there are clear cases on both sides of it. If an author of spy novels copies a portion of an Ian Fleming novel about James Bond, he is an infringer. If, inspired by Fleming, he decides to write a novel about a British secret agent who is a bon vivant, he is not an infringer. If an economist reprints Professor Coase’s article on social cost without permission, he is an infringer; but if he expounds the Coase Theorem in his own words, he is not.

In both of these cases the original work (novel or article) is the joint output of two types of input, only one of which is protected by copyright law. In the case of the novel, the reason for the limited protection is easily seen. n35 The novelist creates the novel by combining stock characters and situations (many of which go back to the earliest writings that have survived from antiquity) with his particular choice of words, incidents, and dramatis personae. He does not create the stock characters and situations, or buy them. Unlike the ideas for which patents can be obtained, they are not new and the novelist acquires them at zero cost, either from observation of the world around him or from works long in the public domain. Most works of fiction that anyone would want to copy are intended for a mass audience--which means they must use stock characters and situations in order to be understood. To give the author of a work embodying such characters or situations ("idea" in copyright lingo) copyright protection beyond the form in which he molds them into a particular novel or story ("expression") would increase the cost of expression of later authors without generating offsetting benefits.

This discussion helps to show why, in the first part of our discussion of ideas versus expression, we disparaged the welfare benefits from creating property rights to ideas. What passes for "ideas" in the case of most expressive writing normally is not comparable to the sort of inventions, costly to develop, that receive protection under the patent laws. An alternative to distinguishing between ideas and expression, therefore, would be to confine copyright protection to original works, or to a work insofar as it was original, much as in the case of patent law. The problem is that originality or novelty is an even more debatable quality of fiction than of mechanical and other technical processes. If copyright protection depended on originality, authors and publishers would find it hard to know in advance of litigation whether they actually had a property right.

In the case of Professor Coase, the reason for confining copyright protection to the form in which he expressed the Coase Theorem, and not extending it to the theorem itself, is less obvious. The theorem was not obtained at zero cost but reflected extensive education and thought. Also, it is clearly original. But precisely because the theorem is a powerful analytical construct, copyright protection would yield the inventor a very large income over and above the considerable nonpecuniary (as well as indirect pecuniary) income that accrues to a major theoretician. The total income would, in all likelihood, exceed the cost of inventing the theorem, thus creating a problem of rent seeking. Moreover, the cost of copyright enforcement would be much greater than in the case where the article itself was copied. It would often be hard to tell whether an article in economics was really using the Coase Theorem; the author (if he did not want to pay a royalty) would make every effort to explain his results in different terms. Furthermore, mathematical and scientific (including social scientific) ideas often are discovered simultaneously, or nearly so; this would make it difficult to determine whether an alleged infringer was a copier or an independent inventor.

iii) The merger of ideas and expression. By this locution we refer to cases where there are only a few ways of expressing an idea, so that protecting expression fully would, as a practical matter, prevent anyone but the author from using the idea. In such cases, copyright protection is construed narrowly, as we would predict. The classic case is Baker v. Selden. n36 Selden had published a book describing a bookkeeping system that he had invented, and he illustrated the book with blank bookkeeping forms. Baker copied the forms, rearranging columns and using different headings, and sold them to people who wanted to use Baker’s system. This was held not to be infringement, since otherwise Selden would have had a monopoly over his bookkeeping system that he could have exploited by insisting that anyone wanting to use the system buy the forms necessary for using it from him.

It might seem that without copyright protection Selden could not have prevented Baker even from copying Selden’s book, but this is not correct. If Baker had published a book in which he copied verbatim (or by close paraphrase) the expository portions of Selden’s book, he would have been guilty of infringement. If he wanted to sell the forms together with explanatory material, he had to write that material himself. The expressive part of Selden’s book was protected.

Is this the optimal result? On the one hand, denying Selden the right to copyright his forms may have prevented him from recouping the expense in time and effort of inventing a new bookkeeping system. On the other hand, there are other ways he could have cashed in on such a notable commercial innovation; and granting him such copyright protection might have overcompensated him (and thus created a danger of rent seeking)--as well as have created deadweight losses in the market for account forms, by raising the price of those forms above their cost by the amount of royalty that he charged for permission to use his copyright.

The scope of the rule of Baker v. Selden is an issue in the current controversy over whether copyright protection for computer software extends to the visual "desktop" on which the computer operator views "icons" representing documents, files, programs, and so on, and to the organization and sequence by which the operator is led through a program. Because a given desktop display or program sequence can be generated by a variety of different programs, the program copyright itself does not cover these visual aspects of computer use. The question is whether they are separately copyrightable by analogy to a painting—a visual display that could be generated by a variety of different processes too. The argument against copyrightability is that the visual aspects in question have become so standardized in the computer market that copyright protection would enable the copyright holder as a practical matter to exclude competing manufacturers of computers. The counterargument is that the "idea" is the display of documents or other data, or sequences of stops, on a screen, and the "expression" the particular visual symbology. The mere fact that a particular set of symbols has become the industry standard is a tribute to the expressive skills of the particular manufacturer and should not be deemed to convert expression into idea.

We hope the debate will be resolved not by the semantics of the words "idea" and "expression" but by the economics of the problem and, specifically, by comparing the deadweight costs of allowing a firm to appropriate what has become an industry standard with the disincentive effects on originators if such appropriation is forbidden. These disincentive effects may be small. The probability that a particular display format will become the industry standard is small; presumably there are significant returns over and above copyright to a firm that achieves such a position; and the narrowly expressive aspects of the display are protected, thus limiting the effect of free riding.

Another contemporary application of Baker v. Selden concerns copy right of architects’ plans. The plans themselves can be copyrighted, and if they are then copied without authorization the copier is an infringer. But what if someone copies not the blueprints but the building built from them? Is he an infringer? The law’s answer is "no." n37 The economic explanation is that a building is functional as well as formal or decorative, so that if the architect could prevent the copying of the design elements visible in the building he would have much fuller property-right protection than copyright law (with its very long term) envisages. The analogy to the principle that denies trademark protection to "functional" trademarks (for example, trademarks in features, such as shape, that may be essential to the operation as well as appearance of the trademarked product) should be evident. n38 Notice the close analogy to the computer-software problem. The building built without the aid of the architect’s plans corresponds to the desktop display generated without access to the originator’s software. In both cases, the originator’s property right is measured by the savings in cost of duplication that the copier could obtain by taking out a license from the originator.

Returning once more to the Coase example, notice that while in Baker v. Selden expression was incidental to idea, that is not true of Coase’s article, in which, besides announcing the theorem, Coase gave illustrations of it, compared it with previous approaches, defended it, and discussed its implications for law, economics, and public policy.

A difficult problem in distinguishing ideas from expression is posed by technique, which we view as intermediate between idea and expression. The main illustrations are in literature and the arts, and include the sonnet form, stream-of-consciousness writing, the five-act play, perspective in painting, the pointed arch, and serial composition in music. n39 The law assimilates techniques to idea and so denies copyright protection. Another reason besides the monopoly profits that such protection would yield in many cases is that technique is harder to copy well than a work embodying the technique. It is easy to copy someone else’s sonnet--but try writing one! With copies likely to be costly, slow to appear, and imperfect, the originator of a technique will be able to recoup some and perhaps all of his fixed costs even if he has no property right in the technique, as opposed to a right in his own works embodying the technique.

In between works of fiction on the one hand, and works in which expression is largely or entirely merged into ideas (Baker v. Selden) on the other hand, are works of nonfiction. The facts that form the subject matter of such a work may or may not dictate the expression; if they do it is a case of merger, and copyright protection should be denied. The classic example is narrative history. The first author of a history of the United States should not be allowed to copyright the sequence of events narrated, since that would preclude any subsequent author from writing a narrative history of the United States covering the same period as the first author. The argument against copyright protection in this case is even stronger than in a case like Baker v. Selden. The historian has not, of course, invented the facts of history that constitute the subject matter of his work--although it would not be correct to say that he acquired them at zero cost; he may have made a substantial investment in research and study.


* Landes is Clifton R. Musser Professor of Economics at the University of Chicago Law School. Posner is a Judge of the United States Court of Appeals tor the Seventh Circuit and a Senior Lecturer at the University of Chicago Law School. We thank Steven Shavell, George Stigler, and participants in the law and economics workshops at Columbia Law School and Harvard Law School for many helpful comments on a previous draft.

2 See, for example. our articles Trademark Law: An Economic Perspective, 30 J. Law & Econ. 265 (1987), and The Economics of Trademark Law, 78 Trademark Rptr. 267 (1988): and Edmund W. Kitch, Graham v. John Deere Co.: New Standards for Patents, 1966 Sup. Ct. Rev. 293.

3 See, for example, Stanley M. Besen & Sheila Nataraj Kirby, Private Copying, Appropriability and Optimal Copying Royalties, J. Law & Econ. (in press): William R. Johnson, The Economics of Copying, 93 J. Pol. Econ. 158 (1985): S. J. Liebowtiz, Copying and Indirect Appropriability: Photocopying Journals, 93 J. Pol. Econ. 945 (1985); S. J. Leibowitz, Copyright Law, Photography, and Price Discrimination, J. Res. Law & Econ. 181 (1986), S. J. Leibowitz, Some Puzzling Behavior by Owners of Intellectual Products, 5 Contemp. Policy Iss. 44 (1987); Ian E. Novos & Michael Waldman, The Effects of Increased Copyright Protection: An Analytical Approach. 92 J. Pol. Econ. 236 (1984); Michael O’Hare, Copyright: When Is Monopoly Efficient? 4 J. Policy Analysis & Mgmt. 407 (1985). Notable early studies are Arnold Plant, The Economic Aspects of Copyright in Books, 1 Economica (n.s.) 167 (1934): Arnold Plant, The New Commerce in Ideas and Intellectual Property (1953); Robert M. Hurt & Robert N. Schuchman, The Economic Rationale of Copyright, 56 Am. Econ. Rev. Papers & Proc. 42 (1966).

4 See generally Michael E. Horowitz, Note: Artists’ Rights in the United States: Toward Federal Legislation, 25 Harv. J. Legis. 153 (1988).

5 We maintain throughout our analysis the assumption of a downward-sloping demand curve for copies of a given work.

6 The effect of (imperfect) price discrimination on output is normally assumed to be indeterminate. See, for example, F. M. Scherer, Industrial Market Structure and Economic Performance 316, 321 (2d ed. 1980).

7 See Armen A. Alchian, Costs and Outputs, in Readings in Microeconomics 159, 165 (William Breit & Harold M. Hochman ed., 2d ed. 1971).

8 This point is stressed by Liebowitz, Copying, supra note 2.

9 Compare Howard P. Tuckman & Jack Leahey, What Is an Article Worth? 83 J. Pol. Econ. 951 (1975)

10 See Plant, supra note 2, compare Stephen G. Breyer, The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs, 84 Harv. L Rev. 281 (1970). But see Barry W. Tyerman, The Economic Rationale for Copyright Protection of Published Books: A Reply to Professor Breyer, 18 UCLA L. Rev. 1100 (1971).

11 On the history of copyright law—which appears to have emerged first in fifteenth-century Venice—see Bruce W. Bugbee, The Genesis of American Patent and Copyright Law, ch. 2 (1967); Brander Matthews, Books and Play-Books, ch. 1 (1895); Comment, Copyright: History and Development, 28 California L. Rev. 620 (1940); and references in note 11 infra.

12 In England, the Stationers’ Company had long enjoyed a monopoly of printing, and this gave the company the equivalent of copyright protection. With the decline of the company, a need for a copyright law was felt, leading to the passage of the first general copyright law (in England) at the beginning of the eighteenth century, as mentioned. See Benjamin Kaplan, An Unhurried View of Copyright, ch. 1 (1967); Philip Wittenberg, The Protection of Literary Property, ch. 1 (1968).

13 A parallel analysis (independent of ours) of the novelty requirement in patent law is found in Suzanne Scotchmer & Jerry Green, Novelty and Disclosure in Patent Law (Berkeley and Harvard, unpublished manuscript, May 12, 1988). The authors point out that the more stringent the requirement, making it harder to get a patent, the greater the gains from patenting but the less information useful to other inventors will be disclosed (patent applicants must disclose their inventions in the application).

14 Later generations of authors may also differ among themselves on where to set the level of copyright protection. Authors expecting to borrow less than they are borrowed from will prefer more copyright protection than those expecting to be net borrowers. Ex ante, however, before anyone knows whether he is likely to be a net "debtor" or "creditor." authors should be able to agree on the level of copyright protection.

29 If we include in y the copies made by accidental duplication, then expanding the scope of copyright protection to make each "copying" unlawful would shift upward the copiers’ supply curve and raise the residual demand curve of the author.

30 See William M . Landes & Richard A. Posner, The Economic Structure of Tort Law 70, 115-16 (1987).

31 Recall Learned Hand’s remark in Sheldon v. Metro-Goldwyn Pictures, 81 F.2d 49 (2d Cir. 1936), that "if by some magic a man who had never known it were to compose anew Keats’ Ode on a Grecian Urn, he would be an author, and, if he copyrighted it, others might not copy that poem, though they might of course copy Keats." Hand, of course, thought such accidental duplication a remote possibility. The probability of accidental duplication of Keats’ poem word for word is too small to justify courts in treating it as a litigable question, that is, one fairly open to doubt.

32 For example, in ABKO Music, Inc. v. Harrisongs Music, Ltd., 722 F.2d 988, 998-99 (2d Cir. 1983), the court found that George Harrison’s "My Sweet Lord" had infringed "He’s So Fine," recorded by the Chiffons. "He’s So Fine" had been one of the most popular songs in the United States and England during the same year that Harrison (a former member of the Beatles) composed "My Sweet Lord." The court found an infringement even though it also found that Harrison had copied the Chiffons’ song unconsciously rather than deliberately.

34 More formally, the full cost of creating a work would equal , where i is the cost of obtaining the ideas used in the work. Normally, i would not depend on the level of copyright protection. However, if copyright law protected ideas (call this protection to distinguish it from z), then i would rise. In addition, e would probably rise because the author would substitute e for i.

35 See Richard A. Posner, Law and Literature: A Misunderstood Relation, ch. 7 (1988).

36 101 U.S. 99 (1879); see also Morrisey v. Procter & Gamble Co., 379 F.2d 675 (1st cir. 1967), where the rationale for denying copyright protection is explained in terms highly congruent with our analysis.

37 See, for example, Imperial Home Corp. v. Lamont, 458 F.2d 895, 898 (5th Cir. 1972); Demetriades v. Kaufmann, 680 F. Supp. 658 (S.D.N.Y. 1988).

38 See Landes & Posner, Trademark Law: An Enforcement Perspective, supra note 1, at 297-99.

39 See Posner, supra note 34, ch. 7.