torstaina, tammikuuta 12, 2006

On Levine and Boldrin's Against Intellectual Monopoly

As a kind of armchair economist and someone highly critical of IP, I read with keen interest David K. Levine and Michele Boldrin's online draft of Against Intellectual Monopoly. Amongst other things, I found its historical examples highly salient and compelling. It's wonderful when respectable economists come to put inside single covers a broad argument you happen to agree with. However, in reading the text I also got the sudden urge to to take issue with some of the reasoning presented.

In chapter 3, the writers liken competitive production of ideas to that of shoes. It is easy to agree with the argument upto the subsection titled Fixed Costs and Competition, but from there on, I think the treatment slightly muddles the relevant issue. The way they put it is that the reason why social efficiency might not come about in idea markets is indivisibility in productive capacity. In a sense this much is true, but the writers fail to precisely indicate why it is so. For example, TV series provide an excellent counter-argument which shows that even sizable pieces of tightly knit fiction need not be indivisible; or, from another perspective, markets can be innovative enough to make indivisibility a nonissue or even an asset.

As I see it, the real reason for the problem of indivisibility lies not in the indivisibility of inventions as such, but in the fact that in the presence of effortless copying, even arbitrarily high divisibilities face the higher divisibility of the operations of a pure pirate. When the printing press and, more recently, DVD burners, exist, the process of copying can be completely several from the process of manufacturing the original copy, and obey a radically different model of costs. This is not the case in ordinary markets; no matter how high the fixed cost of designing a car, it is always to a significant degree offset by the shared, fixed cost of automobile production lines. No market entrant can avoid the latter cost, and so everybody on the market will have to amortize some substantial fixed cost over the copies sold. The latter makes for an opportunity cost the original designer can compete against in the medium run.

But in the case of disembodied information, like all-digital movies, copying is a separate process with a different, far higher divisibility. You buy a DVD burner and you're off copying a movie it took substantial physical hardware and a crew of hundreds to produce. No need for a studio. Even the first mover advantage is largely nullified because the equipment used to copy is generic, while movie studios and truly efficient automobile plants display significant asset specificity. DVD burners burn whatever can be cast in bits, within an hour of the user obtaining a source copy and quite without advance notice of the precise kind of investment efficient production might require. (This complements the original argument, which tends to deal with embodied information and/or information used as a factor of further production; pirates do not need to reverse engineer films, nor does inventing a storyline easily enable another quality film to be made.)

What makes this logically equivalent to the treatment of the book is that the machinery used to copy the film is also available to the original producer. However low the marginal cost of copying, the original producer can always at least rival it, and given current optical disc technology plus the legal right to close down pirate operations, also systematically beat it. However, only the original producer faces the fixed cost of making the film. The pirate does not. Neglecting marginal cost of copying (both because it's comparable across actors and because Moore's law is forcing it asymptotically towards zero), the long run marginal opportunity cost of obtaining a copy equals the short run fixed cost of copying. This is already almost negligible compared to the indivisibilities of production, and on its way down as well. In theory the market equilibribrium can be driven arbitrarily close to a corner point where everything that is ever produced, in however small increments, faces a market where the total social cost of copying the increment universally undercuts the unavoidable fixed cost of production. Since the copying machinery is available to the producer as well, we get the result that any invention always forces infinite production capability, but additionally the argument now proceeds at an added level of detail, it no longer places a lower bound on how divisible the inventions need to be in order to be marketable, and it does not depend to the same degree on market size because the divisibility goalpost now moves along. It shows the nice efficiency result to be applicable only to the degree that there are unavoidable short run fixed costs of copying, which there are on ordinary markets where copying includes specialized physical production capability of some kind, but which there aren't on the idea ones.

Even if all this makes some sense?and I'm not certain it does e.g. because industrial automation is continually making conventional production more like copying a DVD?it still doesn't follow that IP should exist. We just have to argue against it differently. The first attack that comes to my mind is to rigorously apply the second welfare theorem. In order to attain efficiency in production, the possibilities of copying also have to be tapped out. In the presence of both ever-decreasing long run marginal costs of copying and intellectual monopoly, they most certainly aren't; the price of movies ought to go to zero real fast over time, yet it doesn't. Instead there are high, persistent monopoly rents, which bring about rent-seeking, overinvestment in production (and especially in highly indivisible means of production), and so significant dissipation of resources. Often it is argued that big budget movies and the like couldn't be produced without IP, but then they can be largely dismissed as a baseline of comparison because much of what makes them what they are in fact results from the described kinds of inefficiency. It is no mistake that spare-no-expense is the norm in big Hollywood productions: rents tend to have their seekers and rent-seeking quickly becomes institutionalized inside uncontested monopolies.

Beyond that, nobody has actually shown that the indivisibilities (and so, long run marginal costs) of producing, say, movies are bounded from below. If we presume the 100% CGI generated sets and actors of tomorrow, there is no a priori lower limit to how low the fixed costs can go. (In this context, you can charge by the extra minute as you generate the movie; only the plot is indivisible to a degree, but in no case to a degree much higher than that of your average TV series. Even cost disease arguments are largely void in this context.) Similarly, TV series and the possibility of physically walling off live performances already show that nothing much follows from indivisibility, per se. Transaction costs in product search and the like plague the pirate as much as they do the original producer, so innovation in distribution still carries an entrepreneurial premium, piracy or no piracy. And so on.

All such veins of argument are economic courses of action nobody has the proper incentive to develop in the presence of intellectual monopoly, so from my point of view the proper argument against IP becomes that intellectual monopoly would in the long run discourage such societally beneficial alternative courses of technological development. The argument for IP from divisibility is not only empirically unsound, but also invalid because circular (the presumed fixed cost is largely composed of dissipation brought on by underlying monopoly right) and based on an unfounded assumption (the divisibility of production cannot keep up with the increasing divisibility of copying).

Ei kommentteja: