Copying and Indirect Appropriability:
Photocopying of Journals
S. J. Liebowitz
University of Rochester and University of Chicago
Creators and owners of intellectual properties are alarmed by the growth of technologies that ease the task of copying these properties. This paper, however, shows that the unauthorized copying of intellectual properties need not be harmful and actually may be beneficial. The empirical impact of photocopying on publishers of journals is examined in an attempt to discover if publishers can indirectly appropriate revenues from users who are not original purchasers. The evidence indicates that publishers can indirectly appropriate revenues from users who do not directly purchase journals and that photocopying has not harmed journal publishers.
I would like to thank Gerald Dwyer, Jack Hirshleifer, Steve Margolis, George Stigler, Susan Woodward, and an anonymous referee for their suggestions. Support from the Center for Research in Government Policy and Business at the University of Rochester is gratefully acknowledged.
[Journal of Political Economy1985. vol. 93, no. 5]
1983 by The University of Chicago. All rights reserved. 0022.3808/85/9305.00045$0 1.50
It has become dramatically easier to make copies of printed materials since the introduction of the Xerox 914 copier in 1959. Since that time, users of printed materials have been busily making copies of these materials with this and subsequently introduced copying machines, conveniently and at low cost. The copies are generally not of’ as high a quality as the printed originals, but they are often much more convenient. The producers of copyrighted printed materials see this use of their product as an infringement of their property rights and, more important, as a drain on demand and revenues. But the feasibility of photocopying has two other important effects not generally acknowledged: (1) because the materials can be inexpensively copied, there is an increased demand for them as copyable originals
(i.e., the demand of copiers can be indirectly appropriated by copyright owners), and (2) the total value of the copyrighted good may be dramatically altered. Because of these two effects, photocopying need not have a detrimental impact on the revenues of copyright holders. In this paper I examine photocopying’s influence on journal publishers, with the conclusions being that both of these effects are operative and that photocopying has had a salutary effect on the profitability of publishers of photocopied materials. The debate and litigation between owners and users of copyrighted materials, therefore, may be misplaced.1
I. Copyright Law and Appropriability
Copyright gives authors certain property rights over their intellectual creations, the most important being the sole right to reproduce or publish the work. It is often suggested that without such a right authors and publishers would not receive remuneration sufficient to create their intellectual works. Copyright, however, is rather narrow in scope, protecting merely the expression of intellectual ideas and allowing several authors to copyright similar or identical works independently if they were created independently. The usual term of copyright is the life of the author plus 50 years. There are various exceptions to copyright protection such as performance for charitable causes or copying short passages for use in schools. An important exception, which is most germane to academic and other researchers making photocopies, is known as fair use. Fair use is a defense to a claim of infringement currently provided in Anglo-American copyright law when the copying is done for purposes such as research, teaching, news reporting, or commentary.2 The courts determine whether a particular action constitutes fair use, and no hard and fast demarcation exists.
Although it is often suggested that copyright is required if creators of intellectual products are to be able to appropriate revenues from users of their products, copyright is only one of several possible methods whereby authors or publishers can appropriate revenues from those who use intellectual properties. Plant (1934), for example, claimed that being first in the market allowed authors to capture a large portion of the potential revenue, since he believed that it took considerable time for the inherent monopoly power of the first publisher to erode. Another potential form of appropriation, and the one be to be examined here, concerns the ability of authors to appropriated revenues indirectly from users who do not directly pay authors for the right to use their creation.
II. The Potential Impacts of Photocopying
A copyright holder’s profits are threatened when his ability to appropriate revenues is reduced. The substitution of copying for purchase has generally been viewed as decreasing the potential appropriability of copyright owners. Yet it is certainly not the case that direct payment need be made to sellers of products in order for them to appropriate revenues from users. The discussion of a simple analogy can be used to illustrate this point.
Ford sells new cars to both private individuals and automobile rental companies such as Hertz. These cars are durable goods, lasting for many years. Often for individuals, and almost always for the car rental companies, the cars are resold before their useful lives are finished. The price that Hertz is willing to pay for a car depends not only on the value of a car in the car rental business but also on the resale value of the car when Hertz is finished renting it out. When ‘us Hertz buys new cars it includes the expected discounted value of the resale price of the car in the price it is willing to pay. The purchaser of the used car from Hertz does not pay anything directly to Ford, but Ford received indirect payment when it sold the new car in anticipation of this later resale. Thus direct payment is not necessary in order for Ford to appropriate revenues from future users of its used products over the useful lives of these products.
The same type of analysis can be made for intellectual products, which are capable of being used over and over again regardless of their particular physical manifestation. The copyright owner sells a certain number of authorized copies, from which unauthorized copies are made. The users of unauthorized copies, like the buyers of used cars, may be indirectly paying the copyright owner for their unauthorized copies if the owners of authorized copies take the "resale" value of the authorized copies into account when they purchase them.3 Therefore, the impact of copying on copyright owner revenues is unclear, as are any welfare implications.4
948 JOURNAL OF POLITICAL ECONOMY
The analogy to the used car market is weakened when there exists variability in the number of copies made from each original. It is much more difficult for the publisher to appropriate revenues from copiers when copies are made from some originals but not from others. Since the copying of originals will not have an equal impact on all originals when only some are copied, copying will alter the relative values provided by originals. In order to retain the same degree of appropriability in the face of this copying-induced variation in value, the publisher would need to be able to price discriminate among purchasers of originals, charging a higher price for those originals that would be used to make many copies. If price discrimination were not possible, the publisher could charge a high price, essentially allowing purchase only by those individuals planning to make copies and removing much of their surplus. Or he could charge a lower price, generating a larger quantity sold since both copiers and noncopiers would buy originals, but failing to capture much of the surplus from customers planning to make copies of their originals.5 The inability of any owner of an original or photocopy to appropriate value from those making copies reduces the ability of the copyright owner to appropriate value deriving from his work. In the extreme, if a preponderance of individuals used unauthorized copies and if the chain of appropriability were broken early on, appropriability could be almost completely eliminated.
The existence of cheap photocopying is also likely to alter the relative demands for intellectual properties. For example, cheap copying will alter the work habits of intellectual property users. Those items (books) that are weak complements with photocopying will be replaced by strong complements (journals).6 In fact, the entire network of intellectual communications (or scholarship production function) may change in response to the new reprographic technology. These changes in demands will be referred to as the "exposure" impact.
III. Anatomy of Journal Photocopying
I do not believe that anyone would deny that the number of copies made from each original varies. Most journals sent to individuals are infrequently photocopied, as they are intended primarily for the private use of the subscriber. Journals sent to libraries, on the other hand, are much more frequently photocopied, since their function is to assist the many library patrons interested in research and study. A library’s willingness to pay for journals should increase when photocopying is done on the premises because the availability of photocopying causes a library’s users to value the library’s journal holdings more highly and library funding is (almost certainly) related in some manner to the tastes and values of library users. As long as libraries pay subscription prices related to the valuation of journals by library users, publishers need not be harmed by the photocopying done in libraries; this, of course, implies price discrimination if publishers are also to sell to individuals.
In order for publishers to price discriminate successfully, they must be able to prevent arbitrage. It is fortunate for journal publishers that such discrimination is feasible under current institutional arrangements. Libraries do not usually buy their journals directly from publishers, but instead through middlemen (e.g., Faxon). These agencies can be charged the higher discriminatory price since they are very easy to identify. The gain from using these middlemen helps deter libraries from trying to order journals at the lower price by disguising themselves as individuals. Also, the perceived risk associated with ordering journals at the lower price might loom very large to librarians, who might bear much of the cost but little of the benefit. This risk might appear particularly high since publishers, or their agents, merely have to inspect the holdings of a library to determine if the library were paying the institutional price (some sort of mark could easily be printed in institutional subscriptions to enhance detection).7 For these reasons, therefore, journal publishers are routinely able to charge higher subscription prices to libraries.
The empirical work will be composed of two parts. The first part directly tests for the existence of indirect appropriability and photocopying’s impact on price discrimination. The second portion of the empirical work examines the performance of publishers as photocopying activity has grown.
A. Testing for Indirect Appropriability
The thrust of the empirical work is to investigate whether the value placed on an intellectual property by individuals using copies can be indirectly appropriated by the copyright owner. Since journals in libraries are more heavily photocopied than those owned by individuals, indirect appropriability, if it exists, will raise the demand that libraries have for originals by more than it raises the demand by individuals. If the publisher is able to price discriminate between libraries and individuals, we should find photocopying increasing the price charged to libraries by more than it increases the price charged to individuals. In addition, frequently photocopied items should register larger differentials than items less frequently copied. In order to test these propositions, data for a sample of 80 economics journals were collected;8 these data included institutional and individual subscription prices in both 1959 and 1982, the number of citations received by each journal, the age of the journal, and the type of publisher. Although none of these tests can be considered proof that photocopying is responsible for the price discrimination that currently exists, no other explanations come readily to mind that can explain all these results coherently.
From the discussion above it follows that heavily photocopied journals should have greater price differentials between libraries and individuals than those that are less heavily photocopied. Although data on
PHOTOCOPYING OF JOURNALS 951
the number of photocopies made of various journals are unavailable, the number of citations journals receive is taken as a proxy for popularity and hence photocopying activity.
Several factors besides photocopying intensity may influence the pricing of a particular journal. For example, Fry and White (1976) found that commercial publishers increased their price discrimination during the 1969—73 period to a considerably greater extent than other types of publishers. The same result holds for the economics journals in the sample now under examination: the ratio of institutional price to individual price was considerably greater (though not statistically significant at the 95 percent level) for journals published by commercial publishers (2.08) as opposed to noncommercial publishers (1.50).
It also appears that journals of recent birth are higher and more discriminatorily priced than journals with a longer history. For example, the journals introduced after 1959 had an institutional price per page of $0. 125 in 1983, while the journals in existence prior to 1959 had an average 1983 institutional price per page of $0.065. The average ratio of institutional to individual prices was 1.82 for journals founded after 1959 and 1.50 for journals existing prior to 1959. In large part, the reason that young journals are more aggressively priced seems to be that most new journals are published by commercial publishers (65 percent), whereas most of the older journals (93 percent) are published by noncommercial publishers. These statistics, while dramatically representing the evolutionary superiority of the discriminatory pricing strategy, also indicate possible collinearity problems with these two variables.
In an attempt to examine the pricing policies of publishers with respect to the proxied photocopying of their journals, regressions were run that attempted to control for some of these influences. The dependent variable was the ratio of price charged to institutions (PLIB) over that charged to individuals (PIND). The independent variables included the number of cites per page (cites) received by each of these journals in 1981 to articles written in 1975—79,9 a dummy variable (pub D) that equaled one if the journal had its price determined by a commercial firm,10 and a dummy variable (age D) that equaled one if the journal was in existence prior to 1959. The results are:
PLIB 1.29 + .0065 cites + .650 pub D,
PIND (1.99) (4.14) (1)
N = 80, R2 = .17;
PLIB= 1.38 + .0071 cites + .578pubD - .l60 age D,
PIND(2.14) (3.36) (1.01) (2)
N = 80, R2 = .17
(t-statistics are in parentheses). All variables have the predicted sign. The variable of paramount interest, cites, is significant at the 95 percent level of confidence under both specifications. This significant positive coefficient confirms the existence of indirect appropriability. The more intensive use of a journal in libraries allows publishers to raise the price charged to libraries. This test does not discriminate between various causes of indirect appropriability such as the lending of journals by libraries to readers as opposed to lending journals to copiers. The next two tests will address this particular issue. The coefficient on pub D indicates that the pricing of commercial publishers is significantly more discriminatory than that of noncommercial publishers. Age of journal seems to have little clear effect on the degree of price discrimination.
This test examines the behavior of publishers regarding the price charged to individuals versus institutions. The sample of economics journals provides striking evidence for the contention that photocopying increased the price paid by libraries for journals relative to the price paid by individuals. In 1959, the year the Xerox 914 was introduced, only three out of the 38 journals then in existence (or 8 percent) price discriminated between institutions and individuals. In 1983, 59 of 80 journals (74 percent) charged higher prices to libraries (the 1983 price differential, averaged over all journals in the sample, was about two-thirds of the individual price). The emergence of price discrimination was not limited to economics journals. Fry and White (1976) and Liebowitz (1981) studied journals covering many disciplines, and both found prices for institutions rising more rapidly than prices for individuals during the 1960s and 1970s.
The lack of price discrimination in 1959 might seem surprising considering that libraries have always served many users and had multiple uses (loans) of items. The concept of indirect appropriability, after all, was derived from the analysis of the lending of durable goods (as in the Hertz example), and one would expect the price charged by publishers for the originals used by many persons to be greater than the price charged for originals used by only one person. The reader should note, however, that libraries are currently sold books (which are only infrequently photocopied) at prices no higher than that charged to individuals even though the books are capable of
PHOTOCOPYING OF JOURNALS953
being used many times. There is a possible explanation. Compared with hardcover books sold to individuals, those sold to libraries are likely to give much lower values per reading, since individual purchasers of hardcover books have much higher values of the book than most potential readers. Otherwise, these individuals would buy the much cheaper paperback editions. Therefore, even if many people read the book in the library, the sum of their values might not be much different from the valuation placed on the book by individuals to purchasing a hardcover edition. In addition, since books wear out mate with use, the number of library patrons who can read a book is limited by the durability of the book. Photocopying, on the other hand, allows an unlimited number of copies to be made from a single original and is capable, in principle, of raising the value of copyrighted materials by more than physical lending can do. Similar factors might have been responsible for the lack of price discrimination by journal publishers prior to the era of cheap photocopying.
It appears, therefore, that when the impact of heavy photocopying is excluded, library valuations of printed materials are not sufficiently higher than individual valuations to warrant a higher price to libraries. Since photocopying enhances any differences in valuations between individuals and libraries (and this difference was very small in 1959), the now common price discrimination between individuals and institutions can be attributed mainly to photocopying. The differential pricing by publishers of books and journals conforms nicely to the was differential photocopying practices toward each.
Cheap copying should cause some people to substitute copying for rice the purchase of originals, and indirect appropriability allows publishers to appropriate some of the surplus generated by this copying. Since no such analogous effect should occur for items not frequently photocopied, the valuations of library users should shift toward photocopiable items (journals) and away from those items less often copied (books). This test examines the behavior of libraries with respect to the book-journal trade-off.
Trends in library expenditures for American academic libraries were examined for the time period 1941—81 and are reproduced in table 1. For almost 30 years there was no trend in the ratio of expenditures on books relative to periodicals, but by the late 1960s there appears a very dramatic downward trend in this ratio as shown in table 1. (Similar results are reported in Machlup and Leeson , reproduced on p. 140 of the National Enquiry .) That the shift of did not begin until the late 1960s implies almost a 10-year lag between
954 JOURNAL OF POLITICAL ECONOMY
the introduction of photocopying and its impact on the equilibrium values of journal prices. Of course, this lag is not surprising, since it would have taken years for the stock of photocopiers to permeate the economy and for libraries to respond to the altered behavior of their patrons.11 These data should also dispel the notion that a major shift in library expenditures toward periodicals has been going on for a much longer period of time.
B. The Performance of Journal Publishers
Cheap copying should induce some individuals to switch from personal subscriptions to the use (copying) of journals in libraries. Indirect appropriability allows publishers to generate some revenue from these former subscribers.12 The exposure effect, however, assuming it is positive, tends to increase the number of both subscribers and copiers, thus giving no clear prediction regarding photocopying’s influence on the number of subscriptions. Further clouding the empirical analysis is the impact of new journals, which tends to decrease subscriptions to existing substitute journals. Increases in the population of potential readers should have exactly the opposite effect. It is necessary, therefore, to examine subscriptions per journal, the number and nature of journals, and the number of potential readers when examining the impact of photocopying on publishers of journals.
Consider first the number of subscriptions per journal. Several studies13 show that both titles (estimates ranging from 2 percent to 5 percent) and subscriptions per title (1.5—5 percent) grew during this period at rates that cumulatively appear similar to the growth in potential readership, since annual growth of Ph.D.-level personnel was about 4 percent in the 1960s and 5.5 percent in the 197Os.14 The
RATIO or EXPENDITURES ON BOOKS TO EXPENDITURES ON PERIODICALS
Sources—Data for 1941—53 taker, from "College and University Library Statistics," College and Research Libraries.
March 1943. July 1947, July 1948, April 1950, January 3952, and January 1954. Median expenditures are for type I
(larger) academic libraries. No data are available for 1954—58. Data for 1959—81 taken from the Bowker Annual
(1970—82). Expenditures refer to total expenditures for all academic libraries.
National Enquiry confirms the similarity of subscription growth and growth in potential readership when it states (1979, p. 44): "The number of individual subscriptions per scholar has not changed in recent years." Finally, these studies demonstrate that the size of journals appears to have changed significantly. The COSTC study found a yearly increase in pages per journal of 7 percent, whereas Fry and White found yearly increases of 4 percent. These large magnitudes indicate that even if the subscriptions per capita did not increase, subscribed pages per capita almost certainly did.
Because photocopying increased indirect appropriability and the number of subscriptions per capita did not fall, journal publishers should have found reprography beneficial to profits even if there were no population growth. Population growth should further enhance these profits. Evidence on the profitability of journals can be gleaned from data on their establishment and discontinuance. According to the National Enquiry (1979), the ratio of births to deaths during 1971—77 was six to one, hardly indicative of a bleak market. Fry and White (1976) point out that humanities journals established in the period 1969—73 accounted for 13 percent of all humanities subscriptions in 1973. In addition, the average number of pages per journal in the humanities grew by 16 percent during the period. Fry and White feel that this situation "is particularly interesting in view of the indication that journals in the humanities are in the most consistent financial difficulties of all disciplines" (1976, p. 34).
These data reveal that the growth in reprography has not coincided with a period of decline for journals. Rather, since publishers introduced many new journals, positive profits presumably were expected.
956 JOURNAL OF POLITICAL ECONOMY
V. Conclusions and Policy Considerations
I have endeavored in this paper to demonstrate that the debate between publishers and users regarding photocopying’s influence on publisher revenues has neglected important economic factors such as indirect appropriability, exposure effects, and price discrimination. Cognizance of these factors makes it clear that the determination of photocopying’s impact is a complex problem that cannot be determined without resorting to empirical evidence. The empirical work in this paper indicated that photocopying has not had a detrimental impact on publishers.
Discussions of policy have generally assumed that unauthorized copying must be harmful to copyright owners and have tried to weigh the harm to copyright owners15 from unauthorized photocopying against the gain to users.16 The analysis of this paper alters this view considerably. Also altered is the concept of fair use, since with indirect appropriability fair users may pay for their use of the intellectual product (albeit indirectly) just as other users pay. In addition, recent attempts promoting "clearinghouses" to make the collection of copyright payments more economic may be seen as possibly redundant or unnecessary.17 Finally, the analysis presented here might help us understand the impacts of other reprographic technologies. The copying of other media may or may not have impacts similar to those found for photocopying. Only case-by-case empirical investigations of institutions and markets can discover the impacts of these other forms of copying.
PHOTOCOPYING OF JOURNALS 957
Asser, Paul N. "Some Trends in Journal Subscriptions." Scholarly Publishing 10 (April 1979): 279—86.
Benjamin, Daniel K., and Kormendi, Roger C. "The Interrelationship between Markets for New and Used Durable Goods." J. Law and Econ. 17 (October 1974): 381—401.
Besen, Stanley M. "Private Copying, Reproduction Costs, and the Supply of Intellectual Property." Rand Note N-2207-NSF, Santa Monica, Calif.: Rand Corp., 1984.
Bowker Annual of Library and Book Trade Information. New York: Bowker, various issues.
Committee on Scientific and Technical Communication. Report of the Task Group on the Economics of Primary Publication. Washington: Nat. Acad. Sci., 1970.
Fry, Bernard M., and White, Herbert S. Publishers and Libraries: A Study of Scholarly and Research Journals. Lexington, Mass.: Lexington, 1976.
Johnson, William R. "The Economics of Copying."J.P.E. 93 (February 1985): 158—74.
Landro, Laura. "Movie Studios Put More Emphasis on Home-Video, Pay-TV Markets." Wall Street J., May 1, 1984.
Liebowitz, S. J. The Impact of Reprography on the Copyright System. Ottawa: Dept. Consumer and Corporate Affairs, 1981.
________"Durability, Market Structure, and New-Used Goods Models." A.E.R. 72 (September 1982): 8 16—24.
________ "The Betamax Case." Manuscript. Rochester, N.Y.: Univ. Rochester, 1984.
________."Copyright Law, Photocopying, and Price Discrimination." Res. Law and Econ. (1985), in press.
Liebowitz, S. J., and Palmer, J. P. "Assessing the Relative Impacts of Economics Journals."]. Econ. Literature 22 (March 1984): 77—88.
Machlup, Fritz, and Leeson, Kenneth. Information through the Printed Word:
The Dissemination of Scholarly, Scientific, and Intellectual Knowledge. New York:
Novos, Ian E., and Waldman, Michael. "The Effects of Increased Copyright Protection: An Analytic Approach.,, J.P.E. 92 (April 1984): 236—46.
National Enquiry. Scholarly Communication: The Report of the National Enquiry. Baltimore: Johns Hopkins Univ. Press, 1979.
Plant, Arnold. "The Economic Aspects of Copyright in Books." Economica, n.s., 1 (May 1934): 167—95.
Thatcher, Sanford G. "On Fair Use and Library Photocopying." Scholarly Publishing 9 (July 1978): 3 13—34.