Washington, D.C.
October 14-16, 1998
The Impact of Publisher Mergers
Mark J. McCabe, Assistant Professor of Economics
Georgia Institute of Technology
Editor's note: See slides 1 and 2 for a brief outline of this presentation.
Economists, as you know, like interesting data sets, and if you don’t have a belief and a cause, you can certainly believe in interesting data sets. But I think I have both here, and my legal colleagues certainly helped convince me of this. Nevertheless, it is important to objectively evaluate the data, and report what we discovered.
Now, this happens to be a financial sheet from Elsevier’s Web page (see slide 4). It could be most any U.S. company in the last seven or eight years. You see increases over time in profits, sales, margins, cash flow and so forth. This by itself is not a sign of anything amiss. If we’re thinking of computer companies, for example, Dell Computers financial highlights are much bolder than this, growing at exponential rates rather than tailing off. Yet, we attribute that to lower costs and improved market share, not to higher prices. On the other hand, if we see these same facts associated with a company called Microsoft (where choice is more of an issue than it is with Dell where you can buy different PC’s), we would raise questions about those facts, and, of course, the Justice Department is doing that right now. Microsoft is currently being investigated.
Let me also emphasize that economists do things in a very special way, different than other groups. We build simple models, just like physicists build simple models of the atom, and the facts may not all be explained by those models, but they’re a first step.
About 100 years ago, the Sherman Act was passed which dealt with anti-trust questions. During the subsequent century, the process has evolved considerably. In the last 20 years an automatic process has developed for mergers of certain sizes. Typically, a merger transaction above 20 million dollars will require certain documents to be submitted by the companies to the government. Perhaps for every 100 or so transactions, the government looks closely at one.
Once a big enough deal in an industry occurs, usually the agency responsible for that area (either the Federal Trade Commission or the Department of Justice) will continue to look at subsequent deals. So in some sense, the Reed-Elsevier/Wolters-Kluwer proposed transaction has raised the consciousness of the government. This is important even if the government’s investigation was cut short when the companies decided not to merge. It has still raised the consciousness of the regulatory authorities both here and across the Atlantic.
This automatic process owes its origins to legislation named after after three members of Congress (Hart, Scott and Rodino). It generates numerous documents. It’s also the bane of most attorneys on both sides of the production. Economists sit back and look at selected documents, but the lawyers for both sides have to generate and review boxes and boxes of documents. Some transactions generate so many boxes that the FTC and Justice will rent thousands of square feet of warehouse space in Maryland or Virginia.
Reed-Elsevier/Wolters-Kluwer, I think I can say, progressed to this stage in the process. The other way that investigations of this sort can occur, of course, is due to bad acts or alleged bad acts. Microsoft is not a merger case, rather it's about bad acts that the government alleges. These cases are usually done proactively by the Justice Department or the FTC. So even if you haven’t bought another company to become big, you can grow big through many practices, tying the browser to the operating system, for example. Let me also add that Microsoft is not the only example of this type of case. There are plenty of examples, including the recently announced VISA case.
The Hart-Scott-Rudino process is described in the merger guidelines, http://www.usdoj.gov/public/guidelines/horiz_book/hmg1.html. It’s a fairly rigid process and typically proceeds along this path (see slide 3). The courts have asked the government to lay out well-defined principles so that companies know what they’re facing, essentially, the law, the lay of the land. There are four steps: market definition, market shares, competitive effects, and entries and efficiencies (see slide 5).
A good example of market definition can be gleaned from a case recently before the FTC involving Office Depot and Staples. The question in that case was, is the market simply office superstores or is it every store that sells office equipment? And the answer to that question turns on the following principle.
If you had a merger of all the office superstore companies, so that you had one superstore firm and it raised prices five percent, would this action be profitable? Often, economists can look at the data and decide yes or no on that. Furthermore, we speak with customers (with librarians in the case of journals) and we ask, if the single monopoly firm raised its prices five percent, what would you do? And in many cases, the answer is not much.
If the answer to the question in the case of the super stores is, well, the hypothetical monopoly can’t profitably raise prices five percent, then you have to expand the market definition. Of course, parties to a merger hope to define markets as wide as possible, because then their merger won’t control a very large share of of the market. In the case of superstores, the FTC determined that a monopoly superstore firm could profitably raise its prices and designated only superstores as the market.
And that leads me to the second step, market shares. Okay. You’ve defined the market as super stores. How much of a share do the merging companies have of that market? In the case of journals, the anwser to this question will depend on the discipline. With respect to Reed-Elsevier/Wolters-Kluwer deal, the U.S. was interested in STM titles while the Europeans were concerned with other areas, particularly legal publishing in Europe.
Here in the US, I think the answer for journals in some areas was something on the order of 40 to 50 percent. That certainly passes muster for our threshold for market share. But what is this market share threshold and where does it come from? If you read the merger guidelines they have some very simple rules. Generally speaking, the combined share of the merging companies needs to exceed 30 percent of the market for their merger to receive serious attention.
And if a 35 percent company is merging with a 2 percent company, that probably isn’t an issue. But if it’s 35 plus 10, then we’re talking interest from the anti-trust agencies. As I’m going to discuss later, this is our problem with journals. What’s the market? Individual titles, all titles in a discipline or a large portfolio of titles across multiple fields? What market share is important? Is a different frame work needed, a different paradigm?
Third point, competitive effects. Once you have defined the market, and identified market shares, what is the likely impact of the merger?
Generally, if prices are expected to increase at all, the agencies will express serious concerns about the transaction. However, these potential anti-competitive effects need to be balanced against potential efficiencies, which is point four. If costs go down enough, even if prices increase, then on net, the merger might be beneficial. That’s usually a very tough threshold to meet because although it is easy for companies to claim great efficiencies, improvements and so forth, it is hard to prove. Often, many of these claims don’t pan out. Many mergers fall apart precisely because the projected efficiencies aren’t realized.
Turning now to entry and other issues. In most industries when you raise prices, new companies enter. In journals, if you’re the editor of the American Economic Review or Brain Research, you’re not really too concerned about new entry, at least in the short term—where I mean the short term as probably 10 to 15 years. Normally, in Justice Department terms, it’s two years. You’re probably not concerned about a new journal launched because you raised your price on Brain Research.
That’s an explanation of the structure we use—the steps that we go through when we look at mergers. Of course, 'bad acts' investigations don’t involve mergers, and, therefore, are subject to different standards, but the spirit is pretty much the same.
Now, on the four points that I’ve described, market definition, market shares, competitive effects, entry and efficiencies, I’m going to describe briefly the evidence we have about this particular market for journals.
One of the most enjoyable things in an anti-trust investigation—or at least from our point of view—is talking to companies and talking to buyers to find out what the market is. We’re not supposed to reveal who we talked with. Let’s just say we talked to as many people as possible. And what have we learned from that? Well, we posed this question, what is the market?
Again, my initial skepticism. An economist will look at journals based on her own personal use. She will say an article in her field journal can’t be a substitute for an article in a general economics title, e.g. the American Economics Review, much less a journal in another field. So, therefore, the market must be the individual title. That’s the premise. That’s the sort of prior belief an economist would have.
But then you talk to people, and the people buying most of these journals—particularly the more expensive ones—are librarians. They don’t buy journals that way. They do have customers. They act as an agent. But ultimately, they’re making decisions based on the collective, depending on how big that collective is. It can be as wide as all of bio-medicine or as narrow, in some cases, as economics titles. It differs from library to library.
When the Wolters-Kluwer/Reed-Elsevier merger fell apart, there were other deals also pending, involving biomedical titles. [It is rather unusual for a company, in this case Wolters Kluwer, to be involved in several contemporaneous transactions.]
In this case, since the Reed-Elsevier/Wolters-Kluwer deal had fallen apart, the existence of these other cases allowed us to continue investigating. We learned a lot about market definition, and arrived at some tentative conclusions, ones that can be tested empirically: that the market is certainly broader than an individual title but it’s not the entire universe of journals. Libraries tend to simultaneously purchase titles across broadly defined fields, e.g. physical science and/or engineering, etc. In looking at mergers that encompassed medical journals, the medical libraries told us—almost without exception—that the market was everything they bought.
That market was bio-medicine defined very broadly, and it includes more than clinical journals. When you think of the market this way, and using the available sample of libraries, we found that current subscriptions to bio-medical journals were about 20% Reed-Elsevier, 20% Harcourt, 11% Wolters-Kluwer, and 5 or 6% Waverly.
Now, in considering competitive effects, economists build models. They look at data. They test the data against the models. Ideally, you’d like to have a model which generates predictions that you can test. And to some extent, we’ve made some progress along those lines.
Now, earlier when I was talking about market definition, I focused on the customer—how big the market would have to be before you could raise prices five percent, or how small it could be. The subsequent analysis is looking at the actual companies who sell in that market, what the concentration is and how they would behave vis a vis each other. Therefore, my focus will be on the demand side first as well as on the companies.
Before I go to our evidence, let me just talk for a moment about some alternative explanations for the observed price inflation. Depending on who you listen to, price inflation is anywhere from 50 percent to 200 percent in the last 10 years. Cost increases could be generating these price increases. There’s a fair amount of skepticism about that, needless to say.
As far as I can tell, based on what’s publicly available, cost increases don’t begin to account for the price increases. [And I should also add that if mergers result in cost efficiencies, then the prices of the affected titles may actually decline, all else equal. Economic theory tells us that lower costs translates into lower prices, even for a monopoly.]
In the early 1990's, Stanford professor Roger Noll began working with a group which included Nobel Prize winning economist George Stigler. The group began looking at journals with great interest. When George Stigler died, I believe the project went into a sort of hibernation, but what they asserted and what they tried to test was this following notion: That producing a journal involves both fixed and marginal costs. Typically, the first issue costs are quite significant.
Obviously, the marginal cost is two or three dollars to send a print journal, or electronically zero. But you need a certain number of subscribers to cover fixed costs. For any level of fixed costs you can calculate the number of subscribers necessary to recover these costs, given a title’s price. If your circulation goes down because other people are entering the industry in your particular field, then your circulation goes down at that price and you need to raise your price to cover your costs.
That’s the notion Noll had of what could possibly be going on here in terms of prices. That is, proliferation of journals for bad or good reasons could be raising prices. And, of course, what buyers did about the proliferation of journals has a lot to say about what you do about that problem. A lot more work has to be done in that area as well. But the interesting thing here is that even when you try to do that kind of analysis, there’s still a significant price inflation that you observe. Taking into account the possibility that a decline in circulation has contributed to price increases, you still find very large price increases.
So there we were in February saying, okay, what’s an alternative explanation?
[I mentioned that economists play an increasingly important role in anti-trust investigations, not to the diminution of the attorney’s role. It’s simply that both are important now. And 25 years ago, there was no Hart-Scott-Rodino Act, and there were essentially no economists working on these matters. Now computers and the availability of large databases have made a huge change in the way economics is done, also affecting the way anti-trust investigations are done.]
Now we do something called simulation. This is not so much a simulation as a simple model, you put some parameters in the model, and you come up with some results. This is a story, not to be taken literally, but as a first step.
You will see what the simplifying assumptions are (see slide 6), and that there’s plenty of opportunity to build upon this to make it more realistic. And if nothing else, I can say that in my field, which is industrial organization, the realism has increased tremendously in the last 10 years.
Suppose you had five publishers. They all have one title each. We’re going to ignore their costs, because they all have identical costs, and it doesn’t matter in the analysis. The journals these companies own vary in quality, some high, some low.
There are two types of library budgets out there, big and small. I should add that I’ve built models with more than simply big and small budgets, but for the time being, the simple version will serve the purpose. I have also learned that the budgets of the nation’s libraries do vary more than this. There are more libraries with small budgets than with big budgets. That’s probably true statistically.
Purchasing decisions are based on price and usage. This is an assumption, something that I’m going to talk about along with what the holdings of universities over time reveal about that assumption. The library market attempts to minimize the cost per use. A library may have a budget of a million dollars. They rank their journals, cost per use. When they reach the million dollars, they see what the cost per use is, and they cut all the journals above that point. This is something libraries are actually doing at some institutions. And even if they weren’t doing that literally, an economist is happy if people act as if they are.
(Laughter.)
And what I mean by happy is, it’s not that you literally have to have a calculus for every journal based on collection usage, which is, of course, a great way to do it. Knowing your customers here—in the case of libraries, knowing their students and the faculty—is an excellent idea. It helps make more efficient decisions, but there’s a cost to collecting that data. But I think as librarians, and even as economists, we know which journals have the highest value.
Although most faculty don’t know how much the journal costs—they don’t pay attention to that—they do know the value. You are, certainly, cognizant of the price. So you could probably take a group of 100 journals and rank them without looking at the numbers and do a pretty good job of determining which ones you would cut if you had constraints on your budget.
Given this library behavior as the demand side, commercial publishers are interested in maximizing their profits. Non-profits aren’t in this particular model. They may just try to cover cost. They may try to do more than that, but they certainly have different incentives than the commercial publishers.
In the data sample I’m going to show you, of the 1,387 Institute for Scientific Information (ISI) ranked biomedical journals, perhaps 160 are non-profits. I can also tell you that the cost per use is much different for these latter titles, compared to those that are sold commercially. Of course, the variation within the commercials are high as well, as is the variation within the non-profits.
So publishers try to maximize profits given the way they think libraries behave. When a library cuts a journal, of course, they are still asked occasionally, maybe more than occasionally, to find articles from those journals via interlibrary loan. Here, I’m assuming that all libraries engage in this activity and that the cost is pretty much the same across libraries.
Libraries make their price usage decision by looking at their budgets and the cost of inter-library loan. Based on that demand side, the publishers have to choose who they’re going to market to and what price they’re going to set when they fit into that market. They can choose to sell to everyone, but then they have to have a low price. If they choose to sell to a more select group of universities, they set a higher price. And there are further differentiations here. Very high quality journals may be very narrow in their interests, and in some sense divide the market further.
So how do publishers make their marketing and pricing decisions (see slide 7)? The way economists see these models, there’s a sense that there are two periods here. They decide who they’re going to sell to. Then, they set the price. We solve these problems backwards.
Which market do I want to choose to sell to? It’s a strategic question, and we solve it backwards. So you’re sort of like a chess player. What you do in a last move determines what you do today. We solve the last move first, then we solve today’s move.
There are two simple equations here. Clearly, all the journal publishers know that your budget is some value, and the prices they charge can’t be more than the sum of your budget. That’s a simple, intuitive thing. They take that into account in some general way. The second equation, equation 2, talks about how the firms compete with each other. And here I’ve set up a situation where you have a ratio of the price of the first journal divided by the quality of the first journal. From an empirical point of view, we proxy quality by the number of citations.
The firms take these ratios, compare them with each other, and ultimately, in this model, at least, they set them equal to other journals for that given niche in the library budget world.
Economists will ask these questions. Suppose I’m the first firm here with a high quality journal and I raise the price of my journal. If equation one is already satisfied, meaning that the budget is exhausted by the journals in that market, and if I raise the price, I’m going to get canceled. The library can’t afford me. I’m not worth it in terms of cost per use.
If I lower my price on the other hand, I don’t get any more subscription sales, because most libraries buy one copy of a journal. So I don’t gain anything. So I stay at that “price per quality” ratio. The question is, what is that ratio? Is it .4 or .8? And in a world where you have many, many budgets, you have many, many costs per use numbers, and that’s how you generate the varying values of cost per use for journals.
So those are the two things in this model that the publishers do when they set their prices.
The question is, can we find situations where firms will find a library budget niche which they won’t want to defect or cheat on because they don’t want to change their customer bases in the next period. And if we can find such an outcome we have what economists describe as an equilibrium.
In this particular case with five journals, five independent companies, and varying use value I have put together a table of the prices, cost per use, and profits for the companies (see slide 8 and slide 9). Notice that the journal with use value six charges the highest price, but their profits are low relative to the first three quality journals. This is because they’re selling to a smaller market, just the large libraries. In this particular case, the three highest quality journals are selling to every one, the two lowest qualities sell just to the big budget customers. I can demonstrate to you that no one wants to change what they’ve chosen here. They’re happy in terms of profits, given their quality, given the number of libraries and so forth.
The next question in these simulations is what happens when a merger takes place? And this is something we weren’t doing 10 years ago, even 5 years ago very much.
It’s becoming more and more sophisticated. Suppose a merger takes place between the companies that own value 12 and value 8 journals. Can this increase profits? What’s the logic here? Normally, when you merge two products with two companies, you raise prices. Suppose they’re differentiated products? Suppose Burger King buys McDonald’s?
Some people have a strong preference for McDonald’s. Some people have a strong preference for Burger King. After the merger, they raise the price on Burger King products more than they do on McDonald’s products. McDonald’s franchises will benefit to a certain extent because some people who were buying at Burger King get diverted to McDonald’s.
We call this the diversion ratio. In the same context here, if the new company, the merged company, who owns the journal with use value 12 raises the price on the journal with use value 8 more than 12, when 8 gets cut, 12 benefits, because now it’s competing against a lesser group of journals and in turn can raise its price. That’s the intuition. So it turns out that in this particular case with these parameters that a merged company raises the prices on 8 quite a bit.
Again, for realism sake, this is more than a couple hundred percent, but that’s not important here. The journal with the use value of 6 lowers its price and decides to sell to everyone; 8 and 4 now sell just to the largest libraries. If you look at the profits associated with 12 and 8, 12 is plus 57 and 8 has gone down 33. You add the two together, and you have a positive change.
It’s profitable in this model to buy another journal, raise the price on it so that some circulation losses are suffered. But the first journal, the higher quality one, benefits. In some sense, now it’s competing just with 10, but now 8 has been replaced by 6. Six is lower quality and in some sense, higher costs. If you’re competing with a higher cost firm, you can raise your price. So that’s the model. There are, however, different models.
One of my colleagues has just worked on a slightly different model in the same spirit. You can add budget classes and add journals to make the results more realistic.
Now, perhaps the most enjoyable aspect is the empirical part of the data. This table shows you ISI medical journal titles from major commercial publishers (see slide 10).
The National Library of Medicine has a tremendous database called SERHOLD. It’s centralized so that everyone’s serial holdings are in the same format. Format is the key here. We also received tremendous amounts of information from non-medical libraries. Processing multiple formats is a slower process than processing a single format. But since the investigation shifted, we focused on medical titles. I’ve done some analysis in the last month or so since I moved to Atlanta. What I’ve done on this table is indicate to you that from the 200 libraries, we found 82 for the period 1988–1998 that had perfect data sets essentially. Almost all the records were readable in a single format.
And if you look at the libraries we looked at, they were more or less a representative sample of the 200 which, in turn, was a sample that was representative of the larger medical library population. The commercial titles held in those 82 libraries were compared to the titles that are ISI ranked. And the average journal publisher representation was about 79 percent. In other words, on average, 79 percent of the ranked journals published by the major commercial publishers were represented in the actual holdings of the libraries in this table. I’ve underlined the name of four companies in this table who have, in the past year, been involved in mergers with other companies on the list. This list has now gone from 12 to 8. So as we speak, the number of companies is declining.
I also have a note at the bottom indicating to you how many titles we’re looking at; the 932 titles we looked at constitute roughly three-quarters of the total population of ISI-ranked commercial, biomedical titles. You might ask yourself, what about all other titles that aren’t ranked by ISI? That’s a much larger population. We don’t have information about their quality, and it’s important to have that information to conduct a comprehesive analysis of the remaining 10,000 journals out there.
In this next table (see slide 11) we are looking at the shares for commercial publishers in these same 82 medical libraries. Of those 937 journals that are represented in the libraries, we can divide them into groups held, added and canceled. Held means the journal was held the entire period from 1988 to 1998. Canceled means it was canceled some time during that period, added indicates the converse.
And if you look, for example, at all of the subscriptions that were canceled, 28 percent were Elsevier titles. Twenty-four percent of those added were Elsevier titles, and the held subscriptions is about 20 percent. So clearly, they’re losing some share by doing what they do. That’s what a monopolist does, though. They raise their price. They lose sales. That’s consistent.
You see a company like Waverly, which is part of the transaction with Wolters-Kluwer, and their cancel rate is much smaller than their add rate, and they’re growing. Harcourt, for example, has far fewer potential titles to add. If you look on the previous sheet at how many titles exist for people to potentially buy, they’re still growing as fast or faster than Elsevier in absolute numbers because their prices—as you will see in the next table—aren’t quite as out of sync with quality.
This next slide (see slide 12) presents relative prices, relative citation rates on average, and then dollars per citation. Note that the dollars per cite are not calculated by taking the first column and dividing by the second column; rather, dollars per cite are calculated per title and then averaged. The results of the two approaches are not equivalent
The average price per cancelled title is about $1,000 (at the bottom of the table). For Elsevier it is about $1,800 on average for the canceled titles. Their added titles cost $994. The average cost of a title that’s held is $1,334. Look at the variation in quality for Elsevier. Those that are being canceled seemed to be higher in quality but relatively higher in price as well. So the cost per use is higher for those titles. And held titles are lower in dollars per citation and much higher in quality.
But a journal by Waverly, which is a tremendous deal compared to those by Elsevier, on average is equivalent if not higher in quality. The difference is $1,800 versus $263, for example, for the canceled journals.
This table (see slide 13) offers preliminary answers to the question I posed at the beginning, specifically, how do libraries make these decisions? This table tries to summarize in a careful way what libraries did in a given year in relation to journal cancellations and additions. So for the held titles, you see the average price, the citation, and the average cost per use. You would hope that if people were doing things on a cost per use basis, they would cancel titles with the higher cost per use. The next line indicates yes, it’s more than two times as large. These journals are more expensive, and they have lower citation rates. They’re not as valuable to the institution. The added titles have a lower cost per use than the canceled titles, very similar quality and about a 33 percent lower price. This is what’s happening over 10 years for 82 medical libraries, from big to small, small to big. You can see at the bottom, there’s 8,000 titles that we looked at that were held and another 3,000 that were canceled and added.
Economists use regression analysis to explain prices as a function of various independent factors. Thus, the measured effects are independent of each other. We adopted a reduced form or hedonic method of estimation, where prices are regressed on several factors: Firm portfolio size, journal quality (measured by ISI citation level), a Brandon-Hill dummy (the Brandon-Hill journal list indicates essentially whether a journal is a specialized title, and is thus an exogenous measure of circulation “potential”), and a number of variables that capture the effects of residual price inflation and idiosyncratic firm-specific effects. Briefly, our results for journals sold by commercial publishers indicate that prices are indeed positively related to firm portfolio size, and that mergers result in significant price increases.
We can ask, “If portfolio size increases a certain amount, what’s the effect on price controlling for all the other things?” That’s what we are trying to determine.
One last thing about this model. We’ve simulated price increases due to the merger but people are still spending the same amount of money as they were before on journals. What’s changed, the bad effect, something the economist would focus on would be that the average quality of holdings has declined. Everyone’s buying the journal with a use factor of six. They weren’t buying it before. They were buying eight before, and now their quality has gone down. Certainly, you can also have different effects besides that, but that was one important one.
As to the future (see slide 14), anti-trust policies are always in flux just because the economy is always in flux. And the merger guidelines which I referred to earlier that are on the Justice Department Webpage have changed as well over the last 15 or 20 years. In fact, I don’t think merger guidelines existed before the early 1980’s.
So we’ve made progress in a lot of ways. But because it’s in flux it means it can change, and it can change to accommodate different circumstances. The most general guideline can’t take into account every specific case. We have specific guidelines and not just general ones. We have guidelines for health care, and we have guidelines for intellectual property.
It’s not inconceivable that you would be able to develop guidelines for electronic information or publishing. And if not that, you could achieve greater awareness on these issues and how publishing is different. The merger guidelines have two shortcomings. I’ve discussed them already. One is the market definition. What is the market, is it titles or portfolios? I think I’ve demonstrated that journal markets should be defined broadly, and that the precise portfolio definition will depend on the context.
One major obstacle to change is convincing judges who went to what some refer to as judge school. That means they trained in economics for two days when they became judges 25 years ago. And things changed quickly. The other fact about the anti-trust system in this country—it has its strength and its weaknesses—is that a typical federal judge would see one or two anti-trust cases in their lifetime. So they get their first anti-trust and it’s often their last, and it’s hard for them to understand what’s going on.
It’s incumbent on the parties—whether it’s the government or the other side—to educate the judge, and believe me, they do try. The second shortcoming of the guidelines is their presumption that competitive effects will arise only if market shares are substantial. I think we can show, at least at this point, that companies with relatively small market shares have a big impact on prices when they merge, just because the libraries want to buy everything. Small market shares can mean large price increases. What can be done? As I just mentioned, formulation of industry specific guidelines are key. That’s one direction. You can also initiate bad acts or other non-merger investigations.
Immunity. Companies will come to the Justice Department and say, we want to form this joint venture. You can imagine buyers getting together and getting immunity to do certain things. This requires additional support, not just economic, but legal. There are a lot of data questions and data assistance that I could use and my colleagues could use in furthering this agenda. I’ll leave it at that and entertain questions.
(Applause.)
MR. NEAL (Johns Hopkins): I wanted to cite a phenomenon and how it might influence your model. At Project Muse at Johns Hopkins, where we’re making 45 journals available electronically, we’ve noted a very different decision process going on in terms of how libraries choose to subscribe to that database.
Over 90 percent of the libraries subscribing to Project Muse subscribe to all 45 journals through the electronic version. But we’ve analyzed subscribing libraries in terms of how they behaved with the print counterparts of those journals. We have found that nearly 60 percent of them did not subscribe to any of the print titles and none subscribed to all of the print titles.
So we’re seeing decisions being made not on the basis of a title by title analysis, at least at this point, but buying packages of journals. I suspect that we’ll see similar patterns developing in medical journal literature as well. Could you react to that?
DR. McCABE: What’s the pricing scheme you have for the 45 journals?
MR. NEAL: There’s a package pricing scheme, but individuals can also subscribe to individual titles within the electronic version.
DR. McCABE: Is it encouraged by the pricing to buy all 45?
MR. NEAL: It’s not really a significant difference. The total package is just over $3,000 for all 45 titles, unlike some of the titles you were describing here.
DR. McCABE: There could be issues like, for example, libraries have physical space constraints. The pricing is another issue. And also, I don’t know how this fits in with the rest of the library’s holdings.
MR. NEAL: What I’m suggesting is that we may be seeing a different decision making model for purchasing journals.
DR. McCABE: Absolutely. And it’s very interesting to see whether companies have bundled packages for pricing. This is something we couldn’t find much evidence for in the print version of electronic publications. That will demand a different analysis, but I think in spirit it would be the same.
MR. GRAHAM (Syracuse): In your discussion of the model, you talked about the idea of the finite budget of the libraries. Library buying is somewhat different from commercial buying as you described. One of the ways it’s different is in the publisher’s minds, at least. They see that budget as being expandable. And they have their flags out there lobbying the government and universities to increase the budget that’s available specifically for this purpose. Does that make any difference to the model?
DR. McCABE: The next step in the model building is to relax the assumptions and see what it does. I would suspect that it’s a game again. Who acts first and who responds? It’s a little bit of both probably. I don’t have the answer to your question, but it's something I’m thinking about. It’s clearly important.
Budgets are expandable at five to six percent in some years and 10 percent in others. The question is, how much of that is due to anticipated price increases and how much of it is due to interest or willingness to expand the holdings. I think the publishers have definitely exploited the environment to prevent you from doing the latter. So it’s all about expectations and how you build them into the model.
MR. OAKLEY (Georgetown): I was involved in the issues related to the West Thomson merger, and my question is that it seems as though a lot of what is needed is to educate judges. And one of the ways that is done is through the merger guidelines that are in existence at the Department of Justice.
And so, it would seem important for your model’s validity to incorporate some of these new ideas—such as the portfolio rather than the title concept—into the Department of Justice guidelines. Is there any movement to do that now and, if not, is there something we can do to make that happen?
DR. McCABE: The answer to the first question is I don’t know. The answer to the second question is yes, of course. And to elaborate, I was at the Department of Justice long enough to see changes in guidelines and to see the spawning of new guidelines. And it’s generally created by a realization that the existing guidelines don’t do it.
The government is spending too much time trying to explain away—how should I say this—deviations from the existing guidelines. Investigations are getting killed because the specifics of a case does not satisfy the merger guidelines. One objective might be to frequently raise this issue with the government and encourage changes in the guidelines or at least their interpretation.
Even if you don’t change the guidelines, you can get a case, and one case can create a precedent. That is the benefit of going to court, although critics always say going to court is costly and it’s a signal of war. So there are different ways here. One is the merger guidelines. One is to go to court.
MS. KAUFMAN (University of Tennessee): As a community, we’ve been told over the years that there was very little that we could do legally to act as a community of buyers strategically. I’m very intrigued with the idea of anti-trust immunity for our community as buyers. Could you speculate a little bit on what might be allowable and what might not?
DR. McCABE: Well, let me use this example. There are similarities, excuse the analogy, between nuclear power and libraries.
(Laughter.)
DR. McCABE: Nuclear power was an industry and is still an industry with grave difficulties. And based on what I have seen, libraries’ budgets are in grave danger as well. And grave situations demand important solutions, solutions that go outside the bounds. So the Department of Justice gave the nuclear power companies, the General Electric of the world, a blank slate to collude, if you will, with Westinghouse and other vendors of equipment.
If the argument can be made, and I think it can be here, the prospects are reasonable. So it ultimately depends on who your counsel is and who you’re facing. Experienced anti-trust people even know what day of the year to file their cases, because it maximizes the likelihood of getting it through. If you’re not in a hurry, you can accomplish something here.
MS. SLOAN: Thank you very much, Dr. McCabe.
(Applause.)