James Boyle: A natural experiment

Thomas Hazlett: Changes merit symmetric scrutiny

James Boyle: Symmetry and scepticism

Richard Epstein: Why rule out copyright protection for all databases?

James Boyle: Where do you want your property rights?

Imagine a process of reviewing prescription drugs which goes like this: representatives from the drug company come to the regulators and argue that their drug works well and should be approved. They have no evidence of this beyond a few anecdotes about people who want to take it and perhaps some very simple models of how the drug might affect the human body. The drug is approved. No trials, no empirical evidence of any kind, no follow-up. Or imagine a process of making environmental regulations in which there were no data, and no attempts to gather data, about the effects of the particular pollutants being studied. Even the harshest critics of drug regulation or environmental regulation would admit we generally do better than this. But this is often the way we make intellectual property policy.

So how do we decide the ground-rules of the information age? Representatives of interested industries come to regulators and ask for another heaping slice of monopoly rent in the form of an intellectual property right. They have doom-laden predictions, they have anecdotes, carefully selected to pluck the heartstrings of legislators, they have celebrities who testify - often incoherently, but with palpable charisma - and they have very, very simple economic models. The basic economic model here is “If you give me a larger right, I will have a larger incentive to innovate. Thus the bigger the rights, the more innovation we will get. Right?” Well, not exactly. Even without data, the models are obviously flawed - copyrighting the alphabet will not produce more books, patenting E=MC2 will not yield more scientific innovation. Intellectual property creates barriers to, as well as incentives towards, innovation. Clearly the “more is better” argument has limits. Extensions of rights can help or hurt, but without economic evidence beforehand and review afterwards, we will never know. In the absence of evidence on either side, the presumption should obviously still be against creating a new legalised monopoly, but still the empirical emptiness of the debates is frustrating.

This makes the occasion where there actually is some evidence a time for celebration. What we really need is a test case where one country adopts the proposed new intellectual property right and another does not, and we can assess how they are both doing after a number of years.

There is such a case. It is the “database right.” Europe adopted a Database Directive in 1996 which both gave a high level of copyright protection to databases, and conferred a new “sui generis” database right even on unoriginal compilations of facts. In the United States, by contrast, in a 1991 case called Feist, the Supreme Court made it clear that unoriginal compilations of facts are not copyrightable. (The case is not as revolutionary as it is claimed to be. Most of the appeals courts in the United States had long held this to be the case. In fact, a tenet of the US intellectual property system is that neither facts nor ideas can be owned.) Since 1991 the U.S. Congress has managed to resist frenzied attempts by a few database companies to create a special database right over facts. Interestingly, apart from academics, scientists and civil libertarians, many database companies, and even those well-known communist property-haters, the U.S. Chamber of Commerce, oppose the creation of such a right. They believe that database providers can adequately protect themselves with contracts, technical means such as passwords, can rely on providing tied services and so on. Moreover, they argue that strong database protection may make it harder to generate databases in the first place; the facts you need may be locked up. The pressure to create a new right continues, however, aided by the cries that US must “harmonise” with Europe. So here we have our natural experiment. Presumably the government economists are hard at work both in the US and the EU, seeing if the right actually worked? Umm…. No.

Despite the fact that the European Commission has a legal obligation to review the Database Directive for its effects on competition (they are three years late in issuing their report) no attention appears to be being paid to the actual evidence of whether the Directive helps or hurts in the EU, or whether the database industry in the US has collapsed or flourished. That is a shame, because the evidence is there, and it is fairly shocking.

Intellectual property rights are a form of state-created monopoly and “the general tendency of monopolies,” as Macaulay pointed out, is to “make things dear, to make them scarce, and to make them bad.” Monopolies are an evil, but they must sometimes be accepted when they are necessary to the production of some good, some particular social goal. In this case, the “evil” is obviously going to be an increase in price of databases, and the legal ability to exclude competitors from their use – that, after all is the point of granting the new right. The “good” is that we are supposed to get lots of new databases, databases that we would not have had but for the existence of the database right.

If the database right were working, we would expect positive answers to three crucial questions. First, has the European database industry’s rate of growth increased since 1996, while the US database industry has languished? (The drop off in the US database industry ought to be particularly severe after 1991 if the proponents of database protection are correct; they argued the Feist case was a change in current law and a great surprise to the industry.)

Second, are the principal beneficiaries of the database right in Europe producing databases they would not have produced otherwise? Obviously if a society is handing over a database right for a database that would have been created anyway, it is overpaying - needlessly increasing prices for consumers and burdens for competitors. This goes to the design of the right - has it been crafted too broadly, so that it is not being targeted to those areas where it is needed to encourage innovation?

Third, and this one is harder to judge, is the right promoting innovation and competition rather than stifling it? For example, if the existence of the right allowed a one-time surge of newcomers to the market who then to use their rights to discourage new entrants, or if we promoted some increase in databases but made scientific aggregation of large amounts of data harder overall, then the database right might actually be stifling the innovation it is designed to foment.

Those are the three questions that any review of the Database Directive must answer. But we have preliminary answers to those three questions and they are either strongly negative or extremely doubtful.

Are database rights necessary for a thriving database industry? The answer is a clear “no.” In the United States, the database industry has grown more than 25-fold since 1979 and - contrary to those who paint the Feist case as a revolution - for that entire period, in most of the United States, it was clear that unoriginal databases were not covered by copyright. The figures are even more interesting in the legal database market. The two major proponents of database protection in the United States are Reed Elsevier, the owner of Lexis, and Thomson Publishing, the owner of Westlaw. Fascinatingly, both companies made their key acquisitions in the US legal database market after the Feist decision, at which point no one could have thought unoriginal databases were copyrightable. This seems to be some evidence that they believe they could make money even without a database right. How? In the old-fashioned way: competing on features, accuracy, tied services, making users pay for entry to the database and so on.

If those companies believed there were profits to be made, they were right. Jason Gelman, one of our students, points out in a recent paper that Thomson’s Legal Regulatory division had a profit margin of over 26% for the first quarter of 2004. Reed Elsevier’s 2003 profit margin for LexisNexis was 22.8%. Both profit margins were significantly higher than the company average and both are earned primarily in the $6 billion US legal database market, a market which is thriving without strong intellectual property protection over databases. (First rule of thumb for regulators: when someone with a profit margin over 20% asks you for additional monopoly protection, pause before agreeing.)

What about Europe? There is some good news for the proponents of database protection. As Hugenholtz, Maurer, and Onsrud point out in a nice article in Science Magazine, there was a sharp, one-time spike in numbers of companies entering the European database market immediately following the implementation of the Directive in member states. Yet their work, and “Across Two Worlds,” a fascinating study by Maurer, suggests that the rate of entry then falls back to levels similar to those before the Directive. Maurer’s analysis shows that the attrition rate is also very high in some European markets in the period following the passage of the Directive - even with the new right, many companies drop out.

At the end of the day, the British database industry - the strongest performer in Europe - adds about 200 databases in the three years immediately after the implementation of the Directive. In France there is little net change in the number of databases and the number of providers falls sharply. In Germany, the industry added nearly 300 databases immediately following the Directive - a remarkable surge - about 200 of which rapidly disappeared. During the same period the US industry adds about 900 databases. Bottom line? Europe’s industry did get a one-time boost, and some of those firms have stayed in the market; that is a benefit, though a costly one. But database growth rates have gone back to pre-Directive levels, while the anti-competitive costs of database protection are now a permanent fixture of the European landscape. The US, by contrast, gets a nice steady growth rate in databases without paying the monopoly cost. (Second rule of thumb for regulators: Do no harm! Do not create rights without strong evidence that the incentive effect is worth the anti-competitive cost.)

Now the second question. Is the Database Directive encouraging the production of databases we would not have got otherwise? Here the evidence is clear and disturbing. Again, Hugenholtz et al, point out that the majority of cases brought under the Directive have been about databases that would have been created anyway - telephone numbers, television schedules, concert times. A review of more recent cases reveals the same pattern. These databases are inevitably generated by the operation of the business in question and cannot be independently compiled by a competitor. The database right simply serves to limit competition in the provision of the information. Last week, the European Court of Justice implicitly underscored this point in a series of cases concerning football scores, horse-racing results and so on. Rejecting a stunningly protectionist and one-sided opinion from its Advocate General, the court ruled that the mere running of a business which generates data does not count as “substantial investment” enough to trigger the database right. It would be nice to think that this is the beginning of some scepticism about the reach of the Directive, scepticism that might even penetrate the Commission’s review of the Directive’s anti-competitive effects. Yet the Court provides little discussion for the economic reasons behind its interpretation; the analysis is merely semantic and definitional, a sharp contrast to its competition decisions.

So what kinds of databases are being generated by this bold new right? The answer is somewhere between bathos and pathos. Here are some of the wonderful “databases” that people found it worthwhile litigating over: A website, consisting of a collection of 259 hyper-links to “parenting resources,” a collection of poems, an assortment of advertisements, headings referring to local news, charts of popular music. The sad list goes on and on. The European Commission might ask itself whether these are really the kind of “databases” which we need a legal monopoly to encourage, and that we want to tie up judicial resources protecting. The point that many more such factual resources can be found online in the United States without such protection, also seems worthy of note. At very least, the evidence indicates that the right is drawn much too broadly and triggered too easily in ways that are profoundly anti-competitive.

Finally, is the database right encouraging scientific innovation or hurting it? Here the evidence is merely suggestive. Scientists have claimed that the European database right, together with the perverse failure of European governments to take advantage of the limited scientific research exceptions allowed by the Directive, have made it much harder to aggregate data, to replicate studies, and to judge published articles. In fact, academic scientific bodies have been among the strongest critics of database protection. But negative evidence, by its nature, is hard to produce; “show me the science that did not get done!” Certainly, both US science and commerce have benefited extraordinarily from the openness of US data policy. This is an issue I will deal with in a later column.

I was not always opposed to intellectual property rights over data. Indeed, in a book written before the enactment of the Database Directive, I said that there was a respectable economic argument that such protection might be warranted and that we needed research on the issue. Unfortunately, Europe got the right without the research. The facts are now in. If the European Database Directive were a drug, the government would be pulling it from the market until its efficacy and harmfulness could be reassessed. At the very least, the Commission needs a detailed empirical review of the Directive’s effects, and needs to adjust the Directive’s definitions and to fine-tune its limitations. But there is a second lesson. There is more discussion of the empirical economic effects of the Database Directive in this 2000 word column than there is in the 600 page review of the effects of the Directive that the European Commission paid a private company to conduct. That is a scandal. And it is a scandal that is altogether typical of the way we make intellectual property policy. President Bush is not the only one to make “faith-based” decisions.

The writer is William Neal Reynolds Professor of Law at Duke Law School, a board member of Creative Commons and the co-founder of the Center for the Study of the Public Domain

…………………………………………………………………………………………………………………………..

Thomas Hazlett: Changes merit symmetric scrutiny

Professor James Boyle usefully brings to our attention just the sort of “natural experiment” that should form our analytical database in discerning where intellectual property lines should be drawn. His paean to empirical scrutiny is well-placed, as is his scorn for policy that ignores available evidence in deference to political expediency.

So let’s get ready to rumble.

First, the policy question. The relevant margin is whether intellectual property rights should be, in given instances, strengthened or loosened. Macaulay’s view of monopoly is picture perfect - for exactly half the picture. Joseph Schumpeter saw the whole. We welcome “monopoly” for the competitive process it triggers. Over time, this jockeying for profit produces innovation and growth - two ingredients missing from the textbook view of “perfect competition.”

Errors in either policy direction are problematic. One great database, or miracle medicine, destroyed by lack of investment incentives sacrifices human satisfaction just as does an overpriced, accidental monopoly. Changes merit symmetric scrutiny: loosening European rules or tightening American law.

Second, the financial evidence on database innovation. The policy dichotomy between US and EU law is fertile ground, but the fruits of this investigation must be harvested carefully. Operating profits of database suppliers offer potential evidence, but only in much broader context. Annual or quarterly earnings are often highest when firms are disinvesting - exactly the perverse effect that insufficient IP protection is feared to deliver. By way of reverse example, the CEO of the largest cable TV operator in America in the early 1990s famously pledged that TCI would never show a profit - cash flow was being rapidly reinvested and sky-high capital gains were the payoff. While tax law makes this “profitless” stock appreciation strategy particularly potent in the U.S., the essential point is that operating profits are very loosely correlated with the investment activity that IP laws (arguably) bolster.

Third, a consumer metric. Investigation of the rival policy regimes should focus on the value of databases, not their number. The argument for stronger IP rules (EU) implies not that more databases survive, but that those that do will be more valuable to users. That is because the incentive to invest in accumulating such assets will be higher, and customers will gravitate to those databases that prove relatively useful. The higher failure rate observed in Europe is entirely consistent with the success of the EU IP directive: more entry and exit should be observed when investors respond to the prospect of more lucrative rewards. This, as I read James’ article, is what has transpired. This seems to cut in favour of the strong IP regime, as it suggests an intensification of the dynamic process that produces innovative products for consumers.

Having said this, I suspect that the evidence - evaluated fully in proper economic context - will tend to favor the American rule. But it is, as Prof. Boyle makes clear, well worth finding out.

The writer is a senior fellow at the Manhattan Institute for Policy Research

Back to the top

…………………………………………………………………………………………………………………………..

James Boyle: Symmetry and scepticism

I agree with much of what Thomas Hazlett says. Luckily, there is still some disagreement. We agree that intellectual property policy should be made on the basis of evidence; currently it almost never is. I find that scandalous. We agree that the fact that firms or products leave the market provides no evidence of market malfunction. My point was that the figures on net entrances and exits show the US added many more databases without a database right than the EU did with one. We agree that the data I gave (raw numbers of databases, rates of growth and profit margin) are insufficient - though better than nothing.

What evidence did the EU rely on? According to the articles I quoted earlier the evidence comes down to one EU official’s informal visits to websites and German and Dutch department stores to see how many CD-ROM databases there were on sale. That was the extent of empirical inquiry. This seems too pathetic to be true. In any event, this “study” of the database market has never been released. (Imagine a process of reviewing drugs in which the evidence was confidential!) It is worth remembering that the entire European Commission that adopted the Directive had to resign amidst accusations of fraud and mismanagement. Given the danger the Directive poses to scientific research, it is ironic to note that Edith Cresson, the former Commissioner responsible for Science, Research and Development was criticised in the official report for hiring her dentist, a close friend, as a supposedly distinguished “science advisor.” So some skepticism about the quality of the decision-making process may be warranted.

I like the other metrics Hazlett proposes. I would add one question. How many databases are now created and maintained entirely “free” and thus escape commercial directories altogether? There are obviously many, both in the scientific and the consumer realm. One can no more omit these from consideration, than one can omit free software from the software market.

Now for our disagreements. I must defend my countryman Macaulay. Like Adam Smith, or Jefferson, Macaulay was not against creating monopolies as a spur to innovation. He simply demanded proof they were necessary, and wished to make them as narrow and as short as possible while still achieving the desired result. This is advice the Commission would do well to heed as it reviews the Directive. I disagree, too, about the burden of proof. What if the state had created a telecommunications monopoly to spur innovation and investment in expensive telephone equipment or - going back a few hundred years - had created a monopoly over trade to India to encourage the large capital investment necessary? My guess is that Hazlett would say the burden of proof is on the state to defend the continuing need for, and the specific breadth of, the monopoly. I think we should adopt the same burden of proof here. Indeed, that is the burden that the Commission should apply when it reviews the Directive’s effects on competition, something it is legally required to do. If they used actual economic evidence, I would be even happier.

The writer is William Neal Reynolds Professor of Law at Duke Law School, a board member of Creative Commons and the co-founder of the Center for the Study of the Public Domain

Back to the top

…………………………………………………………………………………………………………………………..

Richard Epstein: Why rule out copyright protection for databases?

I have read James Boyle’s energetic denunciation of copyright protection for databases, and think that the subject is surely one on which sensible minds can differ, as seems to be the case with every disputed issue on matters of intellectual property. So in a race in which I have no horse at all, let me indicate my uneasiness with his willingness to rule out copyright protection, or something analogous, for databases.

The first point is to ask, what is this dispute about at all? Here I take it that there can be no serious argument that the traditional dilemma found with other sources of intellectual property rears its head with databases as well. Any effort to spur the creation of the database will necessarily reduce access to the base once it is created, for the only way the original creator can recover its initial costs is by raising price above marginal cost. Yet with various forms of original works that are protected under copyright, we are generally, as a society, prepared to pay that cost, and to hedge about the monopoly position with limits on duration (which have as of late become something of a joke with the inevitable extensions of pre-existing rights) and on such matters as fair use, which, for better or worse, seem to go much beyond the ability to quote brief passages from a work in order to critique it.

The question is why do databases fall outside that general principle, when the costs of compilation are in many cases substantial for the initial party and trivial for anyone who receives judicial blessing to copy the base? In answering this question, it will not do to say, as the Supreme Court said in the well known decision in Feist Publications v. Rural Telephone Service, (1991) that these compilations are not “original” in the sense that it requires no thought to check the spelling of the entries and to put them all in alphabetical order. But that obvious point should be met with an equally obvious rejoinder. If it requires no thought or intelligence to put the information together, then why not ask the second entrant into the market to go through the same drudge work as the first. After all, just this is done with trade secrets, where each party has to roll its own, and will often do so even if they do not know that the other fellow has the secret that they are imitating. (Just the knowledge that X has a secret to Y process counts a partial loss of the secret, because it allows competitors to know that certain lines of inquiry that might have been dismissed as hopeless actually may be bear fruit.)

Now in Feist there is a solid answer to the question that I just posed on duplicated costs. Rural was a regulated public utility that acquired the information from subscribers at low cost as part of its business. It was, in those days at least, a regulated monopolist with a consistent advantage over all other comers. Given that advantage, we might want to introduce some kind of a sharing regime, but even if that were the case, it seems that the right response would not be the free ride offered by Feist. Rather, it would require the next (and all subsequent) entrants to reimburse the original publisher for some fraction of its actual costs, in order to equalise the burden across competitors. It would be interesting to know exactly what James Boyle thinks about the interaction of these two related forms of monopoly (copyright and the telephone franchise), for the ideal solution is far from clear.

What then is the reason to cut out copyright protection? One that is given by Boyle strikes me as both sensible and odd. He notes that skilled creators of databases are able to get the exclusivity needed for incentives through contract. They can control their list of subscribers; give them each passwords; charge them based on the amount of the information that is used, or some other agreed-upon formula; and require them not to sell or otherwise transfer the information to third parties without the consent of the data base owner. Yet cutting through the details, what Boyle seems to argue is that a regime of trade secrets, expanded by contracts, is sufficient to deal with this problem. And I have on doubt that for many kinds of scientific databases, he’s right. The contractual solution is surely preferable, because general publication will allow for use by others that may not offend the copyright law, but which will block the possibility of payment for the costly information that is supplied.

Even when publication is needed, some countermeasures are still available. In some cases, large database suppliers can slow down the reproduction of copying by inserting in their databases new information that does meet the originality requirements for copyright: head notes, keys, cross references, or whatever. But this is not a complete answer for this nervy solution will not work for phone books, where the listings are the launching pad for the advertisements that are included in the volume. The whole point of classified pages is for people to read them so that they will buy the goods and services that are advertised in them. Wide dissemination is what makes that market work. That option goes out the secret if it is treated as a private database guarded with the tightness of the genome. So while I am quite happy to believe that copyright protection for databases can go too far, I am puzzled by the outright condemnation of the idea. We accept the view that for inventions, trade secrets and patents are both needed alternatives. Why is that impermissible for trade secrets and copyrights? So long as Boyle will allow the trade secrets their protection, then why the objection to copyright as well? Raising the bar on “originality” doesn’t seem to provide use with any clues to the answer.

The writer is the James Parker Hall Distinguished Service professor of law at the University of Chicago and Peter and Kirsten Bedford Senior Fellow at the Hoover Institution

Back to the top

…………………………………………………………………………………………………………………………..

James Boyle: Where do you want your property rights?

Professor Richard Epstein mistakes my argument. Indeed he echoes the point I made 8 years ago when I first wrote about these issues. I observed then that our focus on “originality” as a requirement for intellectual property rights might lead us to overprotect certain kind of information products and underprotect others. I gave unoriginal compilations of fact as an example where we might have too few intellectual property rights. Why did I decide otherwise? Because the evidence suggests there is no need for a new database right. The database industry in the United States is flourishing. The US policy is to concentrate intellectual property rights on expression (copyright) and invention (patent), leaving facts and ideas outside of the property system, the public roads of our information landscape. This policy seems to be working very well indeed. If it were not, I would be happy to propose a narrow database right.

But why is it working? The first thing to realise is that Epstein provides only half the story. We build intellectual property out of other intellectual property. That is the reason why many database producers are against the creation of a new right. They make their living compiling facts into databases. If every one of the producers close to that data could claim a property right in it, database companies fear the resulting “thicket of rights” would be impossible to negotiate. The blockages would be even worse when dealing with data that is thrown off by the operation of the business itself, such as telephone numbers. So the policy of not protecting facts may be a good one, if there are adequate other ways for private producers to recover their investment, or if public provision makes the data available at marginal cost - something I will deal with in my next article.

Contrary to Epstein’s suggestion, trade secret law is of little use to most database owners. But they can rely on technical means, on contracts, on tied services from which non subscribers can be excluded, on advertising, and so on. These methods seem to be working. The point is not that it is impossible to create intellectual property rights in unoriginal compilations of fact. (Though admittedly there are constitutional obstacles to doing so in the United States.) The point is that it is undesirable to do so when those rights are not needed. You get fewer databases and higher costs.

It is theoretically possible, of course, that a narrow, time-limited intellectual property right might actually encourage greater eventual disclosure of data. Think of patents. They make inventors disclose their inventions when they might otherwise have kept them secret. So if there were a problem in generating and then getting access to raw data (and there does not seem to be) I might support a narrow database right that required public access to the dataset once the right expired. Is that the kind of right we have in Europe? On the contrary. We have extraordinarily broad and effectively perpetual coverage under both copyright and a sui generis scheme, coverage which reaches lots of databases that would be produced anyway, and which has almost no exceptions or compulsory licenses for research or for competitors. This is the worst of both worlds.

The writer is William Neal Reynolds Professor of Law at Duke Law School, a board member of Creative Commons and the co-founder of the Center for the Study of the Public Domain

Back to the top