As I noted last week, Jerry A. Hausman and J. Gregory Sidak recently published Google and the Proper Antitrust Scrutiny of Orphan Books. Their basic model is that Google Books will be a “new good” coming on the market. New goods benefit both the firms who sell them and the consumers who buy them. The prospect of those profits is an incentive to introduce new goods. Thus, the government needs to be careful to avoid stepping on the profits that pull new goods into existence. (If this reminds you at all of the arguments for intellectual property, it’s not a coincidence; Hausman and Sidak mention the role of patents as a source of incentive-producing exclusivity.)
According to Hausman and Sidak, if antitrust investigations have the effect of regulating a previously unregulated market, it hurts incentives by allowing other firms to grab shares of the incentivizing profits, or even to compete them away. Antitrust can also hurt slowing down the introduction of a valuable new product. Thus, antitrust actions should be saved for situations in which the first entrant will have harmful market power. In this case, however, Google won’t. Competitors would face no legal barriers to entering the market as Google did: by scanning books themselves. Thus, there are substantial costs to antitrust intervention, and no offsetting benefits, so the Antitrust Division should stay out and let the settlement go through.
I see three important assumptions in here:
- Competitors won’t face legal barriers to entry.
- Google Books will be a new good.
- The relevant market is currently unregulated.
All three are false, and for the same reason: copyright. The orphan works are existing goods created by copyright law; the market for them is pervasively regulated by copyright law; competitors will be unable to enter without violating copyright law.
In the rest of this post, I’m going to analyze these three assumptions to show where I think Hausman and Sidak go wrong. This won’t be a comprehensive reply to their paper, although there there are other issues they also get wrong, as well as issues they get right. I want to focus on what I see as their fundamental problem: a failure to work through the copyright issues that are intertwined with the antitrust ones.
Do Competitors Face Legal Barriers to Entry?
Hausman and Sidak believe that entry by competitors is unproblematic. I disagree. Competitors face a legal barrier in the form of copyright law; it’s an infringement for anyone to sell copies of a work without a license. Ordinarily, only the copyright owner is permitted to grant licenses, which means that orphan works effectively cannot legally be used at all. Class-action law permits an exception to this principle, and Google now seeks to take advantage of this exception. The question antitrust law asks—under the heading of “barriers to entry”—is whether the exception is available to others. For reasons I’ve discussed at length, I think the answer is “not without substantial qualifications.”
The other major antitrust anti-critics—David Balto, Mark Lemley, and Einer Elhauge—all give reasons to believe that a class action settlement on similar terms would be readily available to subsequent entrants. These reasons may or may not be convincing (I’m unconvinced by Balto and Lemley and am working my way carefully through Elhauge), but the others recognize the problem and attempt to deal with it appropriately. Hausman and Sidak do not. I count four ways in which they misunderstand the legal barriers facing competitors:
First, they assume that individual negotiations could supply the necessary rights:
Given the non-exclusivity of the settlement terms, subsequent entrants will not face legal barriers to entry. (427)
This claim may be true for actively managed works, but it’s false for orphans. By definition, it’s impossible to negotiate permission from orphan copyright owners. Without the possibility of negotiated permission, competitors will be violating copyright law—civil and criminal—if they try to make books available as Google plans to. I’d call that a “legal barrier to entry.”
Second, they believe that the Registry will have the power to license competitors on comparable terms:
Professor Samuelson claims that no other firm can get an equivalent license to Google’s without scanning books and hoping for a similar class action lawsuit, which might be too risky. She also claims that Google will have a monopoly position. We fail to understand these claims because the settlement allows for subsequent licenses so long as the MFN clause is satisfied. Moreover, Professor Samuelson acknowledges that the Registry can license other firms, which seems inconsistent with her claims about monopoly and the implausibility of subsequent entry. (429–30)
This is a misreading of the settlement agreement. It doesn’t vest the Registry with power to grant licenses for orphan works to competitors. The Registry’s main jobs are to pass money to copyright owners and instructions to Google. It has an open-ended power to negotiate with Google over the details of New Revenue Models (though not with competitors), but the actual copyright authorizations for such uses are granted by the settlement directly to Google. And while the Registry has the power to act as copyright owners’ agent in granting licenses if they authorize it to, orphan owners aren’t around to give that authorization. Since the Registry holds nothing, it can’t offer licenses to orphan works.
This passage is also a misreading of Samuelson. Her slides actually say, “BRR can license others (e.g. Amazon) to scan and make books available as to authors and publishers that have registered with it … .” (emphasis added) She’s not talking about the orphan works here, but Hausman and Sidak are. If they “fail to understand” her claims, it’s not her fault.
Third, they argue that competitors could obtain their own licenses through direct negotiation with the “plaintiffs.” They appear to mean the named Plaintiffs and they appear to have future class-action settlements in mind (although neither of these things is explicit in the paper). They write:
As we now discuss, the MFN does not deter entry, and the plaintiffs will have an economic incentive to license future entrants, contrary to the commentators’ concerns. (430)
It is not at all clear that Google will be able to exercise market power in the future. Suppose that the plaintiffs in the future did sign another license with a subsequent entrant, and that the license gave the plaintiffs a share of revenues greater than the 63 percent that they will receive from Google. (431)
The short answer here is that the rights to orphan works are not the plaintiffs’ to license. The longer answer is that in order to decide whether future entrants would be able to obtain licenses, you need to make supportable claims about the class-action process, and the named plaintiffs’ incentives are only one small piece of that puzzle. I’ve tried, and my conclusions were that future settlements are plausible, but risky, and are probably legally required not to resemble the Google Books settlement too closely. On the question of incentives, I’d note that when I asked the named plaintiffs’ representatives at the recent NYPL panel whether they’d agree to strike a similar settlement with others, the reply was cagey, at best.
Fourth, they believe that competitors could offer comparable services through fair use:
It bears emphasis that other companies can also embark on Google’s road of scanning orphan works and providing access to them through fair use. (422)
The “access” that other companies could provide through fair use would be limited to scanning and indexing. It would not include selling full-text access, as Google now proposes to do. No one that I know of has seriously argued that full-text access would be fair use under current copyright law.
In brief, all of Hausman and Sidak’s reasons for believing that competitors could enter this market are based on mistaken assumptions about copyright, class actions, and the settlement.
Is Google Books a New Good?
The next troublesome assumption is Hausman and Sidak’s characterization of Google Books as a “new good.” The analysis is fine if you think of the Google Book corpus as something created ex nihilo. It’s not. The central plaintiffs’ argument in the lawsuit was that Google was trying to assemble its corpus by violating copyright. The central feature of the settlement is that a small group of settling plaintiffs are trying to grant Google a license to the copyrights of millions of others who aren’t before the court. This, therefore, is a “new” good assembled from a great many existing goods.
Hausman and Sidak systematically minimize the importance of orphan works and their copyright owners. They treat orphan works as valueless goods for which there is currently no demand:
Their current economic value, before GBS, is near zero, as evidenced by the fact that insufficient demand exists for them to remain in print. (428)
Currently, there is almost no market demand for orphan books. (437)
This is a roughly accurate description of why many books go out of print, but it’s not an accurate description of the orphan works problem. Imagine that there were vastly increased demand for an orphan book: thousands of people with a high willingness to pay trying to obtain a copy, bidding up the prices of used copies to astronomical levels. Some orphans would find their parents because of all the attention and the incentive to track down the rights to print a new edition. Many others, though, would stay resolutely orphaned, as the relevant knowledge about rights ownership has simply been lost. The actual demand for any given orphan book after the settlement is likely to be much lower than this caricature—and yet Google will find buyers. The natural conclusion is that the problem the settlement solves is not a lack of demand.
Thus, it’s inaccurate to say, as they do:
Google will create the market demand by making a risky investment of hundreds of millions of dollars in scanning these books into a digital library. (437)
This is almost exactly backwards. The demand for orphan works already exists; Google is creating supply. I would rather say that Google will meet demand rather than create it. The problem this market faces is that the legal barriers of copyright law and the technological barriers of high-fixed-cost print runs made it uneconomical to meet that demand.
By saying that the “value” of the works is low, one is led to focus on Google’s role in offering the products, rather than on the rights of the copyright owners that must be judicially licensed to Google to make the products possible. That perspective allows them to treat Google Books as having its own independent demand. That’s why I think they err when they attempt to extract an antitrust conclusion from their discussion of demand and value:
When a firm creates demand for a new product, we do not see how it exercises market power by charging “too high” a price, because demand did not previously exist for the product. (437)
There are two possibilities here. Either we still care about the copyrights of orphan works owners, or we think that the works are truly and irrevocably orphaned. If the former, then the authors were the ones who “created” the demand for their books by creating the books in the first place. Even if it’s not possible to say that the price of a book is “too high” in this sense, it is possible to say that Google’s prices couldd be “too high” given the existing demand for these existing books. (Keep in mind that ordinarily, books are partial substitutes for each other, but that under the settlement Google would become the sole supplier of the whole orphan works corpus.)
If, on the other hand, you believe that orphans are orphans for keeps, then any price greater than zero is “too high” a price for the copyright licenses to them. The most efficient solution is to place true orphans in the public domain, as any other price inefficiently restricts consumption of these information goods. Google, of course, is perfectly entitled to scan and sell copies of public-domain works, but note that it has voluntarily chosen to price them at zero by offering free PDF downloads.
Either way, this is not a story about Google “creating” demand for orphan works. It’s a story about Google finding a new way to sell into an existing market. And either way, we can assess Google’s prices against a coherent baseline: either the prices the orphan owners would charge if they all showed up, or zero if we truly believe they’ll never show up.
Is the Book Market Unregulated?
The third point on which I think Hausman and Sidak miss the importance of copyright law is in their description of the market for orphan books:
In a nominally unregulated market, antitrust intervention (and the delays associated with the threat of such litigation) can become a de facto form of regulation. Relative to the consumer-welfare losses from the delay in the introduction of cell phones and voice messaging, we would expect similarly large losses in consumer welfare from the misguided or strategic use of antitrust intervention to delay the introduction of a new information-based product like consumers’ enhanced access to orphan books through GBS. (418)
The book market is not an “unregulated market.” It’s already regulated … by copyright law. And this is not the sort of “regulation” that balances speedy introduction of new products with external concerns like consumer safety. No, if you believe in the incentives theory of copyright (which Hausman and Sidak presumably do, given how much it resembles their theory of new goods), this is regulation that creates the market in the first place, by setting up the right incentives for authors.
We need, therefore, to be careful about the effects of the settlement (including any antitrust interventions) on the machinery of rights and incentives that copyright law employs. Here’s how Hausman and Sidak frame the issue:
Nonetheless, some academic commentators claim that competitive problems may arise from the GBS settlement. Their primary concern is that Google will use the orphan books settlement so that GBS is able to gain a monopoly position over all book search services. (437)
Others can speak for themselves, but my primary concern is that Google will use the settlement to gain a monopoly position over orphan books. Everything else is derivative of that ur-worry. We should look very closely at the bona fides of effectively transferring some of copyright owners’ bundle of rights to Google. This transfer affects the rights of absent class members (a class-action issue), tinkers with the economic flows associated with orphan works (a copyright issue), and concentrates a great many previously scattered rights in Google’s hands (an antitrust issue).
On the copyright question of incentives, Hausman and Sidak say:
Google will invest in scanning orphan books and compiling them into a digital library, which will significantly increase their use. Thus, GBS will create economic value for the orphan books. No change in the production of future books will occur. (428)
This claim is untrue, because there are almost certainly some books that currently appear to be orphaned but which would have become unorphaned in the normal course of events without the settlement. For these books, Google and the Registry will sell at prices not set by the copyright owner and keep a commission the copyright owner might not have agreed to. After five years, even that revenue will no longer be held for the owner. The settlement will thus satisfy some demand for these books in a way that diverts revenue that the owner would have realized, but for the settlement.
Thus, to the extent that the legal precedent set by the settlement creates expectations on the part of potential authors that some of their future revenue might be similarly diverted, the effect will be to depress the production of future books. Perhaps this class of books is small—but remember that Google’s fear that orphan work owners would reemerge was in part responsible for the settlement in the first place.
Similarly, they say:
Any division of the revenues arising from the use of orphan books between Google and rights holders is a division of rents with no effect on economic efficiency. We expect that both sides bargained in the settlement negotiations, but antitrust law is not intended to, and should not, be used to divide rents. (428)
The same point applies. These revenues are part of the incentive system copyright law sets up. When orphan-work revenues are diverted to other copyright owners, that’s not division of rents; that’s an incentive-distorting transfer.
On their second-last page, Hausman and Sidak make a claim, intended by them as hypothetical, that I think perfectly captures the spirit of their analysis:
Thus, only if future competitors of GBS were unable to negotiate an agreement to use the orphan books would a problem arise. (437)