File sharing and marginal cost pricing

Last fall, I discussed the future of digital media, especially music, and argued for a system in which digital media is free to users, but artists and producers are paid for it with public funds distributed by observing use. The basic idea, articulated in Terry Fisher’s and Lawrence Lessig’s recent books is that a royalty fund is put aside from taxes (about $10 billion a year, or $100 per family, would do it for the US music market). Playing music files is observed (but not who plays what) by the copyright office through a piece of software running on any device that can see the internet. The Copyright Office then divvies up the fund each year in proportion to plays.

Two developments have brought this development much closer. The first is the demonstration by 411-song that a recording can be identified by software with only fifteen seconds of listening to it over a cell phone, so robust reporting of use is a realistic prospect. The second is the passage by [admittedly, a rump sitting of 58 members of] the French Chamber of Deputies of an amendment to a bill that was supposed to increase filesharing penalties, using digital rights management technology for enforcement. To the astonishment of the Minister of Culture, who offered the bill, M. Alain Suguenot’s amendment goes in exactly the opposite direction, making free file sharing of music and movies legal. The bill includes a tax on internet access to be divided as royalties in an unspecified fashion, so the French are approaching an admirable revolution in intellectual property rights, even if this episode is a piece of political theater for the moment.

(Here’s an AP story in English.) The amendment “freeing culture” in France is here. The operative mechanism

“La publication d’une Ĺ“uvre ou d’une interprétation fixée sur phonogramme, vidéogramme ou tout autre support emporte cession du droit de mise à la disposition du public sur des services de communication en ligne, pour les seuls actes effectués par des particuliers à des fins non commerciales, à une société régie par le titre II du livre III et agréée à cet effet par le ministre chargé de la culture.”

is to transfer non-commercial virtual (en ligne) reproduction rights to a new national quasi-governmental organization as a condition of publication.

The “arm-wrestling”, as Figaro describes the debate, has pitted the country’s largest consumer organization against a variety of publishers and some artists’ groups. Further debate has been put off until after the holidays; this story bears watching.

Future of Music (and movies, and books…)

Earlier this week I attended the Future of Music annual “Policy Summit“, a conference of artists, recording industry execs, intellectual property lawyers, and academics. They gather to predict, view with alarm, recommend, and debate large questions of public and private rights and patrimony such as Lawrence Lessig discusses (for my view on his work, look here), and also the specifics of legal and business arrangements that might assure that artists are properly compensated for their work and also that we have lots of good stuff to listen to. The background context is of course the wheels coming off the traditional model, in which music was sold embodied in something physical like a CD and not copiable without a large investment in expensive equipment. The key image of the current crisis, I think, is the completely implausible business model of the amazingly successful iPod: (i) you buy a player that holds 10,000 songs for a couple of hundred dollars and then (ii) fill it with music from iTunes or legal, purchased CDs (iii) at a dollar a song, for about $10,000. At least part of this scenario has never actually occurred. 20 million-odd iPods are sold every year; guess which part.

The business players are, approximately, a few large record companies and the manufacturers of reproduction equipment (hi-fis, iPods, computers, etc.); innovative, hungry, lively companies offering different versions of web and stream music distribution, CD merchants large and small; and a lot of artists, including soi-disant artists and hopeful artists. Some in the latter two categories are really good and deserve a large audience; many more, no doubt, believe themselves to be. As usual, the “interest group” not at the table in any offical way, but in whose name everyone claims to speak, is the audience and the latent audience for music, meaning pretty much everyone.

The most important issues of the current debate seem to me to be the following:

(1) Marginal cost pricing. The idea that everything should be sold for what it costs to make the last one of it (that is, what it denies you for me to have it) is a cornerstone of what economics has to teach us. In the context of digital music, meaning (now) any recorded music, this principle means that the correct consumer price for listening to a recording is zero, because doing so leaves no less of it for anyone else. (Contrast a seat at a live concert, which does not have this “non-rival” quality). I was astonished, in two days of panels and plenaries, with a fair number of economists around and about, to have never heard this principle stated either in jargon or in plain English.

(2) Price signals, or something like it, for artists and creators, about what kind of music to make in order to create the greatest net social value. In my arts policy course, class discussion usually settles on “more better art consumed by more people” as the appropriate mission statement for the art system. Creative people need to know what work is creating more value for society and what isn’t, so they can adjust their output accordingly.

In the departing system, these signals are extremely noisy. Everyone has CDs they play again and again, and others they put on the shelf after one try, but both sent the same price signal to the creators and publishers in the form of royalties and sales. If we could get away from a system of charging for possession of a file (iTunes or a physical CD) and start observing plays instead, we would all be better off. But obviously the key to cutting the Gordian knot of filesharing lies in being able to pay creators appropriately even if listeners are not paying directly for music.

(3) Compensation for artists. Artists of all kinds labor in a vineyard with two curses. The first is that they are trying to do for a living what other people do for fun, so there is constant labor oversupply of a type that does not afflict, say, bus drivers. The second is a “winner-take-all” marketplace of the type described by Frank and Cook with a few overpaid stars and a lot of also-rans, a situation greatly aggravated by recording technology. Despite these hurdles, it’s obviously essential for a society to assure its artists decent incomes so the good ones (not just the few “best”) can be laying down tracks for us to hear rather than waiting on tables or teaching (the correlation between artistic talent and a talent to maximize the competence of student artists is quite modest). Paying artists a living wage is conceptually distinct from the signaling function in (2).

(4) Search and selection. In a world in which the latent supply of music is enormous, it’s not a simple matter to decide how to commit your next three minutes of attention. Publishers of sheet music and record companies used to make this task manageable by allocating capital and marketing resources to a very few candidates. But in a world of free digital distribution (what we have now de facto) it’s not clear what enterprises (reviews? Amazon-type “people like you liked these CDs” algorithms? Informal networks?) can make the search and selection process tractable, and all the candidates have important downsides and deficiencies.

(5) Goose and gander sauce customization. I was amazed to see how completely the classical music niche has fallen beneath the radar of these music industry players. Classical music creation, distribution, and marketing have important differences from what works for popular music, but all the talk at this meeting was about rock, jazz, and popular/commercial forms.

As a scholar of art policy, I’m familiar with the odd tendency of arts advocates to seek (and, sadly, sometimes attain) policy goals that are against their real interests, like resale royalty rights for painters. But the discussions in the arena of music strike me as especially off-target. The recording industry is simply desperate, completely unable to think outside the box of suppressing file-sharing and copying by lawsuits, digital rights management schemes that can’t work, pained and self-righteous assertions about property rights, and similar rear-guard failing tactics. Artists believe themselves to be consistently cheated under current rules (very few CD releases ever generate any actual royalties after marketing and promotion costs are deducted) and are, perhaps understandably, afraid to think seriously about any alternative scheme. The listening community has no organized voice and just keeps voting with its mice. And though academics have made really promising policy proposals (my favorite, though I think it can be further improved, is Terry Fisher’s enforced license/fund distribution scheme), the kind of theoretically coherent analysis this industry desperately needs seems to be unable to penetrate the cacophony of recriminations and paranoia.

Michael O’Hare on paying for music

Michael O’Hare offers what seems to me one convincing answer to the question of how to pay for recorded music, now that file-sharing is driving the existing system into the ground. (This responds, among other things, to Eugene Volokh’s objection to Paul Boutin’s suggestion of a tax on data transfer capacity and payments to the producers of music on a per-download basis.)

O’Hare suggests that music should be valued, and paid for, not by how often it is accquired (by purchase or download) but by how often it is listened to, and proposes a technically feasible and non-intrusive system to measure that, not perfectly but “well enough.”

Having everyone have access to whatever music he wants for a dollar per person per week sounds like an enormous bargain to me. But O’Hare’s essay is indespensible, even if you don’t agree with his proposed solution, for its clear-headed exposition of the values at stake.

The total now paid for the production and distribution of recorded music (about $10 billion a year) is small enough that O’Hare wants to pay for it out of general revenues rather than a dedicated tax. That makes substantive sense, but might be less politically workable than a self-financed program paid for by taxing data transfer, especially if that could be linked with an effective anti-spam measure. I’d never notice an email tax of a hundredth of a cent per kilobyte, but it would make the peddlers of unclaimed Nigerian bank accounts think twice before sending out their next million messages.

O’Hare doesn’t address the problem of shelling out tax dollars for offensive content, which would be something of a problem for music and a huge problem for video. Does anyone have a suggestion? I’d be for such a system even in the face of those issues, but it would be a hard sell in Congress.

As impottant as music and film are culturally, in economic terms they’re the rounding error in the revenues of the pharmaceutical industry. In that case, the problem of estimating value in the absence of a market is even harder than it is for cultural works.

But the fact that there will be valid objections to any proposal doesn’t mean that a new system wouldn’t be an improvement over the current system of patent and copyright protection. The forgone consumers’ surpluses from charging high prices for goods whose marginal cost of production is near zero are extremely high.

Since such goods will make up an increasing share of economic activity as far into the future as one can see, getting a non-terrible set of solutions is a policy problem of the highest importance. I hope O’Hare’s essay sparks the debate it deserves. Note that it’s still in draft, which means that it shouldn’t be cited but that it may be malleable in the face of convincing criticism or suggestions for improvement.

[Adam Liptak in today’s New York Times reports on the related question of “designer seeds.” [*].]

[My previous thoughts on this topic are here (responding to Brad DeLong [*] and here responding to William Safire.]

[William Fisher of Harvard Law School has a long, scholarly essay here, part of a book-in-process titled Promises to Keep.]

Paying for pharmaceutical develpmentT

Brad DeLong wonders, reasonably, why people hate drug companies so much. He cites a Wall Street Journal/NBC News poll showing the industry with a favorable/unfavorable ratio of 21%/54%, which probably puts it in a tie with Gary Condit. After all, despite the companies’ marketing shenanigans and apparently excessive rates of return, their products do in fact prevent many early deaths (mine, for example, when I had Hodgkins lymphoma a couple of years ago) and improve the quality of many lives.

The root of the problem is, I submit, pricing. Pharmaceuticals are priced in ways that bear no relationship to their marginal cost: the actual expense of making one more dose, once the drug has been developed and approved. The public is aware of that, aware of the way in which the pharmaceutical industry tries to twist the political process in support of above-marginal-cost pricing, and not conversant with, or not convinced by, the economic analysis of nonrival-consumption-goods pricing..

How to pay for creating things that are, once created, very cheap to make in large numbers is a widespread problem. Given its growing importance, it probably deserves more attention than the standard intro micro text gives it. If the goods are priced way above marginal cost to allow a recapture of the initial investment, the result is that some consumers will choose to, or be forced to, forgo them. That forgone benefit represents sheer deadweight loss, or what non-economists call “waste.” That’s as true of a song or a software program or a microprocessor as it is of an AIDS drug, but the problem gets more exciting when the forgone benefit is staying alive.

Owners of intellectual property rights like to say that unauthorized use of their property is theft, just like shoplifting a steak from a grocery store. But of course that isn’t exactly right; the stolen steak can’t be sold to another customer, and costs the store something to replace. That’s “rival consumption” in micro-speak. But listening to a song is “non-rival”: I can listen as often as I like and leave no less song for you to hear. Thus the “theft” of a song by file transfer costs the record company only the opportunity to extract money from that consumer at that moment, and if the consumer is one who wouldn’t have paid the asking price then the company doesn’t actually lose anything: it’s no worse off than if Napster had never existed, while the pirate consumer is better off. [Glenn Reynolds offers Brian Brigg’s funny analogy-by-parody.] You don’t have to be able to define the term “Pareto principle” to have a hard time understanding what’s wrong with an act that makes someone better off and no one worse off.

Some of the practices Big Pharma engages in to protect its revenue stream make it look especially bad: trying to buy special legislation to extend individual drug patents, as in the notorious instance of Claritin; fighting to slow down FDA approval of generics; opposing the cross-border trade that would defeat the “market segmentation” strategy that loads most of the drug-development cost on American consumers. (Jagdish Bhagwati makes a convincing argument that, in this instance, what hurts US consumers helps developing-country consumers. But imports from Canada are another matter.)

[Footnote: All right-thinking people are horrified by the very large proportion of US Gross Domestic Product that goes for health care. But much of that expenditure goes to pay for the costs of the very rapid rate of biomedical progress, not just in pharmaceuticals, but in medical devices, imaging technology, and health-care-delivery techniques as well. That progress benefits both future generations and the rest of the world. So a careful accounting would chalk only part of that expenditure to current consumption; some of it is investment, and some of it a kind of foreign aid. Arguably, there’s nothing wrong with the richest country in the world providing, and paying for, much if not most of the world’s biomedical progress, but there’s no obvious reason the costs involved ought to be borne by sick people and employers who provide health insurance, rather than by taxpayers generally. Oh, sorry, I forgot: Taxes BAD. Profits GOOD. Right.]

There’s no very good solution to the problem of how to pay for the production of non-rival-consumption goods. The one we’ve chosen in the pharmaceutical instance — private development under patent protection — creates great production incentives but generates enormous deadweight loss, and real inequity for those unlucky enough to be both expensively sick and poorly insured. It also contributes to the health insurance death spiral, as rising premiums drive more and more people to be uninsured, which means, in effect, subsidized by the shrinking pool of people who are insured, thus driving premiums yet higher. Price controls on drugs, or reductions in patent protection, or facilitation of “arbitrage” from the countries where the drugs are sold cheap to the U.S., where they are sold dear, would reduce the deadweight loss at any given moment, but at some cost in reduced incentive for drug development.

[ANOTHER FOOTNOTE: If you’re curious about why cannabis has never been developed into a pharmaceutical, the prejudice against it, both as an abusable drug and as a crude plant material rather than a single chemical, constitutes only part of the explanation; the fact that no drug company could get patent protection for it, and therefore no drug company has any incentive to absorb the expense of sending it through the FDA wringer, is the other, and perhaps larger, part.]

One alternative to patent protection is public provision. The National Institutes of Health already pays for a considerable amount of R&D, which the drug companies get to appropriate in the drug-development process. There’s no reason in principle why NIH shouldn’t do its own drug development, hire its own consultants and lawyers to lobby its drugs through the FDA approval process, and then either produce them in-house or (more plausibly) grant non-exclusive licenses to manufacturers to produce and distribute the drugs, trusting that competition will drive prices down close to production costs.

Another approach, discussed in this Cato publicaion by Glennerster and Kremer, is to offer cash prizes for the development of (or even for particular milestones in the development of) drugs for specific uses: set out a set of criteria for, say, an HIV vaccine, and a process for deciding when the criteria have been met, and let institutions of all kinds compete to meet them. [Michael Gluck reviews a range of alternatives, not including prizes, here.]

The most famous use of a prize was by the British Admirality, which offered in 1713 (and finally paid in 1773) £20,000 for the development of method for determining a ship’s longitude. (That doesn’t sound like a huge sum, but in purchasing-power terms it would be a few million dollars in today’s money, and that was back when real incomes, as well as nominal prices, were a lot lower; £20,000 would probably have been close to the highest income earned by any individual in Britain in the year when the prize was offered.) It’s not hard to think of problems with such a scheme, starting with the fact that the reward would be for being first, putting too much emphasis on the time factor and not enough on other characteristics of the drug invented. The underling difficulty is that neither the criterion selection process nor the prize-giving process would be a perfect substitute for evidence of actual medical utility. (Maybe the prize should be stated in terms of a subsidy payment per dose prescribed, though it’s easy to imagine problems with that, too.)

But the fact that there’s no perfect system doesn’t mean that the one we now have in place is the least bad that could be developed. And of course there’s no reason to have only one system running; private development under patent protection, public development with nonexlusive licensure, and prize competitions could exist side-by-side. But as long as drug companies charge thousands, or even tens of thousands, of dollars for drugs that cost tens or hundreds of dollars to actually manufacture, people are going to stay mad at them.


This note generated two very generous compliments, one from Matthew Yglesias, who liked the substance, and one from Demosthenes, who especially liked the concept of finding the least bad alternative.

There were also two substantive comments from experts (as I perhaps should have made clear, I claim no more than advanced amateur status on health care policy):

John Donahue of Stanford writes:

First, my understanding is that Canada uses its monopsony power as the single payer for drugs to drive down the prices of Canadian drugs to only a fraction of the American cost. Of course, this means that if Canada is paying less than its fair share of the fixed costs of drug development, then someone else is paying more than their fair share. If there is deadweight loss here perhaps the US and Canada and other countries can get together and eliminate it (or the US can start exercising monopsony power of its own). Second, if a really important drug comes along the US could always exercise the right of eminent domain to take the patent while paying fair compensation and then distribute it at marginal cost.

[For those who don’t speak Economese, “monopsony power” is the ability of a single buyer to stick it to the sellers: the opposite of monopoly. Canada is a monopsonist because it has a national health system, so there’s only one buyer for drugs in the entire country. The US has less power, because there are many payers, but the Federal government, and federally-subsidized state programs, account for a large enough share of pharmaceutical demand to have considerable market power as well. (Technically, that’s “oligopsony.”)]

UPDATE A reader corrects this as regards Canada, which does not have a single-payer for drugs.

Those are both important points, but I think John neglects what might politely be called the “political economy” of the problem. Yes, the US government could do either of those things; whether those would be good things to do is a matter for debate, and probably depends on lots of other issues and policies.

[The government could also try to acquire patent rights by negotiation, rather than by eminent domain; in situations where there are several nearly equivalent drugs, that could turn into some real high-stakes poker, with each company worried that one of its competitors will take the offer. But the fact that this would be a repeated Prisoner’s Dilemma, with a relatively narrow cast of players, suggests that the companies would probably find ways of colluding to reject the offers without actually exposing their executives to criminal anti-trust charges.]

But the government is hugely unlikely to do either of the things John suggests, or any of the things I suggested, for reasons of politics rather than policy. Big Pharma is a huge campaign contributor, on both sides of the aisle. It’s an election year, and everyone hates the pharmaceutical companies, and yet the House Republicans have bottled up the McCain-Schumer generics bill, which the CBO estimates would save consumers about $6 billion per year, and which passed the Senate 78-21.

(Note that no one who has to deal with real policy issues takes at all seriously the hot debate among campaign finance “reform” advocates and opponents over whether corporate and corporate-influenced contributions actually buy results: of course they do, unless there’s equally big money on the other side or intense popular interest over some specific issue.) .

David Boyum, who also knows a lot more about health care policy than I do, points out that things are more complicated than my post suggested:

With pharmaceuticals, you can’t identify market failure by merely observing a gap between marginal cost and price. Why? 1) health insurance reduces the marginal cost of drugs to many consumers; and 2) from a medical and/or welfare perspective, many drugs are overconsumed, despite prices that exceed marginal cost. (This is especially obvious with antibiotics, where the externality of drug resistance knocks your nonrival argument flat on its face. Do you really think it would be better if Zithromax were sold at marginal cost?)

Bottom line: you cannot conduct your analysis on purely theoretical grounds; you need some serious cost-benefit data.

I’m generally more skeptical than you are regarding government efforts to correct market failures, so the ideas of government prizes for drug development or making NIH a pharmaceutical company strike me as foolhardy at best.

My first-order analysis wouldn’t focus on the alleged deadweight loss indicated by the gap between price and marginal cost. Instead, I’d think about getting the right level of investment and consumption across different classes of drugs. Do we currently overinvest in various classes of drugs, or underinvest? Do we overconsume or underconsume? Once those questions are answered, then you can think about how to play around with government-funded research, patents, FDA regulation (including import policy), concentrated purchasing power by government and private insurance, and so on.

All of which seems right, though of course things are even MORE complex than that: some consumers face reduced prices due to insurance, some don’t, and some insured consumers find that particular drugs aren’t covered, or are covered only after several layers of medical review, because the insurance companies are trying to avoid overprescribing or are in tugs-of-war with the drug companies over pricing. And whether the prices of antibiotics are really a barrier to their overprescription has to be an open question; the place to address that problem may be in the medical schools rather than the co-payment schedules.

That the current system has overconsumption as well as underconsumption complicates the problem, but it’s not as if the two somehow balance out: the fact that primary-care docs routinely prescribe antibiotics for viral infections doesn’t do a thing for the cancer patient whose insurer won’t pay (thousands of dollars) for the blood growth factors that speed recovery from the side-effects of chemotherapy.

All that said, David (as usual) poses the right question, which is always half the battle: what would the socially optimal pattern of drug investment and drug consumption look like, and what set of policies would, dynamically, bring us closest to that optimum?

Wouldn’t it be nice if we had a political system capable of working through the answers to such questions, instead of one where money talks and analysis walks?