EPUBs are an experimental feature, and may not work in all readers.

Paper presented at the AAUP/ARL Symposium on Electronic Publishing, November, 1993

In this discussion of the economics of electronic publishing I will concentrate upon the equivalent of the typical academic book — what we have come to refer to as the long linear document. In what follows, lacking a good brief term for this, I will speak of `documents'. Although it is the long linear document that is in the forefront of my mind, most of what I will say would apply equally to documents with hypertext and/or multimedia components. And in some parts of the argument those kinds of document display the characteristics about which I am speaking more acutely than do simple documents.

The Reader

We can start from several directions but I want to focus on the most important element in the process — the reader. We must be careful not to allow either the economics or the technology to distract us from the fundamental purpose of the whole endeavor. The point is not words on a screen or type on a page, but ideas in a mind.

So let us focus on the academic reader. The essential characteristics of academic readers are that they are expert and time constrained. As experts they want to make their own choices of what to read, but being time constrained they wish to make their choices of what to read with least time consumed in the selection process, the least time spent reading things that are not valuable, and they want the actual reading process to be efficient and effective.

An academic reader also wants to know about all the important strands of thinking in her discipline, and wants to stay in touch with work in other fields that have relevance to her work. This last consideration is of increasing significance in the fluid world of present day humanities and social sciences. This may be one major factor differentiating the reading patterns of scholars in the humanities and certain social sciences from reading patterns in the sciences.

A century or so ago most of these were not issues. The intelligent lay person could keep abreast of a wide sweep of human knowledge. Now an academic, even with a light teaching load, struggles to keep in touch with all that is written which is of immediate relevance to her work. And no one expects the pace of new writing to slow. Indeed many predictions suggest that the availability of electronic channels for dissemination will have a liberating effect on the productivity of scholars. That is to say, that the easy availability of electronic channels will make more documents available.

Faced with a scarce resource — in this case time to read — an economist argues that the scarce resource has a real value and that misuse of that valuable resource has a real cost. We may not go through a careful process of assigning a dollar value to it, but we should nonetheless aim to use it as parsimoniously as we spend money. In a profession such as the law where the meter is always running, no one has any difficulty in understanding this. When we look inside the academy, this point is much less widely appreciated. No doubt one explanation is that academics usually love their work and thus draw a much looser boundary between work and leisure than do most people. But their days still only have 24 hours like everyone else's — time constrained means time has a value.

So when we think about an academic reader, we are dealing with a resource — her time — with a real economic value. And that fact must guide our exploration of the way a manuscript, via whatever delivery mechanism, should go from its author to its reader.

The Information Intermediary

In a simple and small world, all authors and readers would have direct contact and documents would move directly from creator to reader. Readers would know authors and their work and decide on the basis of that direct knowledge whether a work was worth their time to read. As the size of the population increases, and as it becomes more dispersed and less homogeneous, that direct method breaks down and a need develops for some intermediation between author and reader.

There may well be disciplines, or at least subfields, where the direct communication model still applies. One can sketch certain necessary characteristics of such a field: small size, relatively low output per scholar, general agreement about the scope of the discipline, and general agreement about quality.

While complete fields of this kind may be rare, there are many disciplines where much of the elite communicates informally through working papers — a form of dissemination that is not intermediated. What is concerning there is the exclusionary consequence. Those not connected to the elite networks tend to be left behind and left voiceless. As we consider the need for intermediaries and their role, broadening access both to readers and authors should be a serious consideration. This is of course a challenge as the selectivity necessary for other reasons can conflict with such democracy.

What are the tasks of these intermediaries? I identify four: gathering, selecting, enhancing, and informing. There are of course at present two main kinds of knowledge intermediary: the library and the publisher, and each of the four tasks has a somewhat different meaning for each intermediary. But one important publisher activity fits outside this schema and deserves a brief mention. Book editors, but also some journal editors, persuade scholars to write works which they believe are needed and would be valuable. In many cases also, the publisher works to persuade a scholar that her work is worth synthesizing into a book-length document. I am not going to explore this activity further, but raise it here not only for completeness but also to make clear that the range of material to be gathered in and selected from is not exogenous — it is not purely a given - the intermediaries create and influence what is written in a wide variety of ways.

In the paper world, the four tasks as done by the publisher primarily precede the four tasks performed by the library. This is something of a simplification of the library's full range of work, but it does apply over a wide range of library activities.

Gathering. The publisher gathers in the writings of authors both reactively and proactively. The existence of a large number of publishers permits them to gather writings from a wide range of sources and to provide a quite comprehensive service which saves scholars from having to seek out the works for themselves. This saves scholars' time and of course makes possible a much wider catchment area (literally and metaphorically) than any one individual could monitor. Now it may be that in an electronic publishing world, the task of gathering will be eliminated, although I would not expect the facility for seamless worldwide searching to be practical and available to all for some considerable time. There seems likely to continue to be a need for some kind of knowledge intermediary hunting and gathering across the net.

Selection. I choose the word `selection' rather than `gatekeeping' because `gatekeeping' has a passive connotation: it conjures a vision of a flow of people autonomously arriving at a gate and being considered for entry. St. Peter may be assured of a reliable flow of candidates, the publisher feels a definite need to seek them out.

As a publisher considers a new work or proposal, two sets of criteria are applied so that two types of signals can be sent to readers to guide their later reading choices. The first kind of signal is a pure quality one: this work is worth your time to read. The more discriminating the publisher, the more the scholar can trust the signal. And of course that trust goes back not just to the publisher but to the academics who as advisors and editorial board members steer the editorial decisions of university presses. The second kind of signal is about the kind of work. Most good publishers have an editorial policy: they endeavor to select works of a particular kind even within a discipline. The scholar with certain interests and tastes can therefore know whether what that publisher offers is likely to be appeal. Thus in two ways the publisher's selection process reduces time costs for the reader.

Before I go on, it is time to point out that what I am talking about is value added. I have avoided the term so far, but the time that is saved for the academic reader is a value added service to the reader. Indeed I would posit time saved as one important way of defining and quantifying value added. And of course an entirely media neutral measure.

Enhancement. The third task of the intermediary between author and reader is enhancement. Under this term I include everything that is done to the author's work which changes it before it reaches the reader. This is probably the area where the electronic world will in due course look very different from the present paper world. However amongst the functions that will in some form carry over are editing (revise and resubmit is surely media independent), copyediting, or some components at least of that activity, and design — the formatting of material to make it easy to read, easy to navigate around, and generally to make it an effective device for conveying the message of the document. As we redefine this general task of document enhancement — whether it be traditional style copyediting, the provision of hypertextual links or integration of multi-media, the dominant question will have to be about the relationship between cost incurred and value added.

Informing. Finally there is the task of informing potential readers — what is vulgarly called marketing. Some believe retrieval tools will make this unnecessary. Considering the complex interest matrices in the humanities, I suspect that many scholars will continue to welcome a contribution to the process by active knowledge purveyors and will not want to rely completely on the passive knowledge/active seeker model which is so often implicit in predictions of the future. That the methods and mechanisms employed will change, indeed are changing, is undeniable.

Library Value Added. Turning now to the definition of these tasks for the library, I must be much more cautious given my lack of relevant expertise. So a few brief comments. First, although payment for value added is interestingly absent from most of the library world, clearly we must think of the library tasks within the same framework of benefits, especially time saving benefits, for scholars. While the publisher gathers and then selects from the manuscript universe, the library has traditionally selected from the universe of publications: works that have survived the publisher selection process.

We might ask the rationale for this two-stage process. There are I think two reasons. First the publishers provide a multitude of entry channels and thus ensure there is no monopoly of ideas. But second there are considerable economies of scale in the selection process: one publisher discriminates for many (indeed all) libraries. The economies of scale argument is reinforced by the decisions of most libraries to use vendor systems to reduce to manageable proportions the number of publications from which they have to choose. The vendor rather than the library gathers and does a partial selection.

At the library level, enhancement is very different. It tends to be a matter of multi-document organization rather than single document enhancement — although we might well see some library intervention in the organization and presentation of documents once they are in electronic form and thus transformable.

The function of informing is illuminatingly different. Although libraries do disseminate some information about what has become available to their user community, the special expertise is the reverse of the publisher's: the active-searcher passive-information mode: the skilled searching for what the reader needs. I have increasingly come to feel that this distinction between the direction of activity along the knowledge conduit is important to re-envisioning the roles of publishers and librarians in the future.

Measuring value added

Having identified a range of ways value is added by the intermediators between author and reader, we need to think about the process by which that value added is reflected in revenue received by the body adding value. And here we come upon an interesting difference: the publisher as intermediary has primarily expected that his efforts at adding value will be recognized by readers paying for those efforts. On the other hand, the library has generally expected that the community, be it college, city or club, will pay for its efforts because of some perceived general utility of the value that it adds. No one seems to believe that the library's level and range of services should be defined by market criteria — what it can sell for a sufficient amount to pay its costs. But the contrary presumption is usually made of university presses.

For the moment I will leave that observation to stand but will return to it later. It does though, I hope, raise questions in all your minds as to whether making decisions based purely on market criteria is wise for intellectual and culturally important services and goods. Subsidization of music and theater by both government and private donors certainly suggests a pervasive belief (but sadly not universal) that some things are too central to our culture to be left to the Darwinian struggle of the market place.

Inadequacy of Market Signals

There are in fact good economic arguments for recognizing that market signals will not always be the appropriate ones for determining what is provided. And I want briefly to present three economic reasons why books and even more so electronic documents fall into the category of goods whose provision should not be solely decided by the market.

Cultural concerns. First, imagine that a university library charged a serious amount for use of its holdings; and decided what it would add to those holdings based on the revenue generated from various sections. With substantial research grants, the scientists would probably be little deterred by charges for use. Indeed even large access charges would hardly be significant relative to their equipment costs, for example. On the other hand, humanities scholars, usually without research money, would be strongly deterred from library use. The consequence would be a marked decrease in the purchase of new humanities works in the library and a shift of yet more resources to science and technology holdings. I am sure that none of us would welcome that. I hope it raises questions about the use of a market mechanism of this kind to guide library collection development.

But there are more fundamental reasons why the market should not be relied upon to generate best solutions in our areas of concern.

Public Goods. Books and even more electronic documents possess tow of the characteristics of what economists call public goods. It is a bsic theorem of economics that the market under-provides public goods. That is to say, if the market is relied upon to determine the supply of a public good, the amount produced and sold will be less than the amount that maiximizes society's welfare. Too few resources will go into producing public goods and too mnay into producing conventional private goods — apples, shirts and handguns.

The two characterisitics that imbue documents with publicness are non-exhaustibility and non-excludability.

Non-Exhaustability. The first of these is based on the non-exhaustability of the document in either paper or electroninc form. One person consumes an apple — it is gone. One person reads a book and it is unchanged, ready to be used again. Something of course that even more true of an electronic file. The economic theory that underpins blind faith in market solutions specifically does not apply to goods that are not used up in the act of consumption. Indeed economic theory shows that such goods will be under-provided relative to simple economic goods, and that non-market devices are needed if society is to have the right amount of such goods. Museums, rural roads, public health endeavors all come under this category.

Non-Excludability. The second reason is based on the non-excludability of documents. In a paper world, the existence of photocopying means that some readers obtain a document not from the publisher but by making a copy of the original publication. They thus gain value from the document but contribute nothing to the publisher to recompense for the costs incurred in creating that value added. The result of this is that some documents are not published which would not only generate value added for readers, but whose total value added would be greater than the costs of making the work available. Society is therefore less well-off than it would have been had the publisher not had to assume that his revenue would reflect only a part of the value added. It seems nearly gratuitous to remark that this phenomenon threatens to be much mores serious in an electronic publishing world with the extraordinary ease and cheapness of copying a file.

Declining Costs. There is another quite separate argument that market based solutions lead to under-provision of books (and again a fortiori electronic documents). This is based on the distinctive cost structure of books, a structure that journals share and so, in a more extreme form, do electronic documents. The essential fact is that most publishing costs are incurred in getting to the point where we can make the first copy, or send the first copy over the net. And the costs of making additional copies are small (in electronic publishing - probably very close to zero). Economic theory argues that output of a good should be expanded until the cost of the last copy made is equal to the benefit to the last buyer who buys the last item sold. If we stop before that point, there are buyers who would gain utility greater than the social cost of making that item who do not in fact obtain it.

***********Figure here [sic]

Society's net benefit is maximized if the price of obtaining the document should equal the incremental cost of producing the document. But that means that sales revenue or user fees or whatever are not available to cover the first copy costs. Thus consider an example:

first copy costs $5,000

incremental cost is $3

cost of 1000 copies

$3,000
total expenses $8,000

revenue from 1000 copies sold

at price=incremental cost

$3,000
Publisher's loss $5,000

So there is a clear and substantial conflict between the publisher's need to cover costs and the optimum of having price equal to incremental cost. And with the incremental cost in the electronic environment close to zero, the situation is more acute than the example I have just presented.

There is nothing new about this situtation. Academic publishers have always had to price well above incremental cost. So buyers who would gain benefit from possession of the book that is greater than society's cost of providing the book have been deterred from purchase. Various mechanisms for recovering some of the first copy costs without loading them onto the individual purchaser have developed. Subsidies and help in kind to university presses, title specific subsidies e.g. from the NEH; subsidies e.g. from Getty to support exceptionally high first copy costs. Also higher prices to libraries than to individuals, either through differential subscriptions or though different editions — paper for the individual, cloth for the library, operate to bring the individual purchase price closer to incremental cost. Interestingly the library can be seen contributing to this end in another way: rather than subsidizing the producer to bring price down to marginal cost, society has preferred a mechanism that subsidizes the consumer by permitting free use but not ownership of books.

Cost recovery solutions

Solutions in which optimal pricing and cost recovery are not in harmony are quite common. There are three types of solutions: socialization, special tariffs, mutualization. I will talk briefly about each as each does raise issues of some significance.

Socialization. It is very common to provide such goods through a government agency. Roads are a good example: huge first user costs, trivial or zero costs per user in non-congested situations. They are naturally provided and supported from ppublic funds. So should we just nationalize the provision of academic writings, or set up some organization jointly armed by all universities? I doubt any of us find these options especially attractive. But raising the topic points to a very important issue in any re-design of our existing system — the need for diverse and independent channels.

It is easy when focusing on economic solutions to loose sight of the crucial importance of our topic: at root we are talking of the exchange of ideas — and the free exchange of ideas. We cannot afford to invent solutions that do anything to endanger the freedom of individuals to express ideas conformist and maverick, radical or traditional. The great strength of our present system is the variety of places to which an author can turn to get his or her work published. This we must preserve.

The university presses have evolved into organizations that do provide this freedom. There are a substantial number, over 100 members of the AAUP, each formulating a distinct editorial policy. Their editorial decisions are made in collaboration between press editors and a faculty committee — university administrators rarely have any say in what is published. In my experience faculty boards are very aware if their responsibility to provide a channel for a wide variety of ideas and defend that freedom jealously.

In addition, university presses have been energetic in supporting emergent fields. The fact that sales do matter in their operations means a sensitivity to new trends and developments in the academy. And the number and variety of presses is a good insurance that routes to publication will be available for those fields and those scholars struggling to put forward new perspectives.

Thus in this consideration of institutional designs, I want to reject monolithic solutions to the cost recovery problem. But more — the provision of diverse routes to publication must be a very central criterion in evaluating any proposed changes.

Special tariffs. The second type of solution to the cost recovery problem is special pricing structures. There is an enormous literature and a very technical one on designing pricing schemes for industries with cost characteristics like those of publishing. Much of this literature is driven by the problems of pricing for electric utilities that are faced with very much the same problem as we are.

This problem can be expressed: we need to recover high initial costs while setting a price close to incremental cost so that no one whose benefit from the product is greater than the incremental cost will be deterred from buying.

One common structure for this kind of problem is what has recently been described to me as country club pricing. You pay a membership fee but then pay incremental costs for all the things you do within the country club. There are many other similar applications of this kind of system. Gregory Rawlings proposal is of this kind. We might envision a journal charging a subscription fee which buys you the right not to receive a traditional journal but to have copies of articles you want sent to you at whatever the incremental cost of sending might be.

However all applications have one thing in common. The customer is a repeat customer who is prepared to pay for something akin to membership. This may well make sense for a journal - it is not very plausible for the individual book publisher as an individual may only infrequently want a book from a particular publisher. Some kind of sales consortium might be a solution here.

I am though dubious that scholars in the humanities and social sciences without research money will be very prepared to pay the quite high annual fee which this system would require. So while I believe it is an avenue for further exploration, I am not confident that this is the way to a full solution.

There is one related point that I do want to make. Some of the suggested solutions to the problems of academic publishing have eliminated the price mechanism altogether and suggested that academic publishing should be fully subsidized.


The question is how should they be recovered. There are basically two options. To subsidize the publishing operation or to collect some kind of fee from the user. I would argue for a combination although with the user fee being much the larger part. I will first explain my reason for prefering the user fee before explaining why I see some role for subsidy at source.

I believe it healthy that the producer have to be to a considerable degree responsive to user demand. A user fee ensures this. The responsiveness is of two kinds, one selecting for publication those things that scholars actually want rather than perhaps those things that authors believe should be published, and second taking into account the users views about what does and does not add value. I think an entirely subvened system would rapidly drift away from user needs. We have the ex-Soviet Union as something of a model for what happens when producers do not have to respond to customers.

However there are works that deserve publication even though the demand will be small. It is to permit some of that kind of publishing that I suggest some level of subvention. In order to set the criteria for such works at a high level, the subvention should be small part of the total revenue so that responsiveness to demand is the dominant consideration. However there is a narrow line between responsiveness to demand and decisions being dominated by pursuit of products that have the highest demand. The subvention can help resist that pressure which is a pressure for the demise of scholarly works. What kind of entity should thus be subsidized? It should be a nonprofit with a very clear mission to serve the academic community rather than to pursue primarily economic objectives. If it is based at a unversity, it should not be the exclusive publishing route for that university's faculty. They must have a wide variety of publishing options open to them. It must be able to maintain standards by selecting from amongst local faculty work and as part of diverse and catholic system of schoalrly publishng must be open to accepting work for scholars at other universities.

Mutualization. This is actually what we have already and so rather than envisioning a new basic framework under this heading, I want to advocate rather the full exploration of the potential of present arrangements — a potential that has been sadly neglected. But this is to jump ahead. First I should return to the basic problem: one entity is worried about cost recovery, while another entity is worried about the impact of increasing prices on its budget. In most cases of this general kind, the two entities are distinct and distant, we therefore need a solution that works through a market-type mechanism to a solution that ensures, at least viability for each entity and moves us to a position that minimizes social costs and maximizes social benefits. Amazingly one can in many instances devise solutions that approximate to those objectives.

However, in the particular situation that we are considering and in which we are involved, we can cut through many of those complications: the main participants are already under common ownership. University presses and libraries and the faculty they both serve are all part of the same institution — the university. Yet a model has become established in which presses relate very much at arms length with libraries. The prevailing mindset is a customer-supplier one. In other words we have mutual ownership but seek none of the benefits that mutual ownership should give us.

Now before I move onto ideas for redesigning our institutions, let me make a caution. Institutions tend to evolve for good reason and we should not casually meddle with them. Presses have evolved away from close identity with their home institution because they saw it essential, if they were to have the power to be selective and thus able to apply high standards, that they not be the tied outlet for all the work of local faculty. The press that is the local publisher ends up taking what faculty write and actually ends up not publishing the best of what faculty writes. As August Frug put it so neatly: "In order to succeed at home, a press has to succeed abroad." So in pursuing mutualization, we have to recognize the need for the press to have wider horizons than just its local institutions. Our challenge is to design a system that gains the benefits of mutuality without forcing the presses back to an exclusively local focus.

I am not going to provide a full solution here. It is something that needs more thought and discussion, indeed mutual discussion, to define suitable arrangements but the essential first step is that libraries and presses on individual campuses begin to think about their problems in a system-wide way. Individual pursuit of solutions to problems perceived in the narrow can combine to perverse solutions. Those of you who read The Fifth Discipline by Peter Senge will recognize a point that he makes and makes most persuasively: one must think of the whole system and not separate units of the system.



Colin Day is the Director of the University of Michigan Press. He can be reached by email at colinday@umich.edu


Notes

1. [Gregory Rawlings add ref [sic]]

2. [August Fruge, The Ambigous University Press, Scholarly Publishing Vol.8 No.1 Oct 1976 p.4.]