Why make wasteful investments?

I've been planning to wrote this blogpost for quite a while now, but this excellent, although longish article at ars technica on the legal details of streaming video finally nudged me into doing it.

In April, Wired magazine ran an article  on a small startup called Aereo. The company has built an impressive data center containing more than ten thousand antennas for tv reception. Using this, the company sells a DVR-like service, whereby their customers rent a personal antenna in the data center. The antenna is connected to what the company calls a remote DVR. Essentially, customers use the service as any DVR, recording TV-shows to it for later watching.

When a customer records a TV-show using the remote DVR service, the show is captured from the antenna the customer has rented, and stored as a separate file on Aereo's servers. Thus, if 1000 customers record Mad Men, 1000 copies of Mad Men are stored on Aereo's servers, so that each customer has his own copy.

From a technical point of view this of course doesn't make any sense. If Mad Men is broadcast from one sender, and 1000 Aereo customers wants to watch it at some time other than specified by the broadcaster, then it ought to be easiest to store one copy, sourced from one antenna, and then stream from this one copy to any of the 1000 Aereo customers when needed. This ought to be the most cost effective solution, a solution recommended by an economist. Let's call it the first best scenario, the one that minimizes social costs for a given outcome.

In contrast, the current technology is socially wasteful, in that it incurs more costs (the building and running of the one-antenna-per-user system) than the first best alternative, while having no additional benefits.

Why don't we see the first best scenario in practice? The simple answer is copyright law, as outlined in the ars technica and Wired articles. To circumvent copyright law, Aereo must essentially replicate the hardware of a DVR in their data center for each customer, at a cost significantly higher than the first best solution. By doing this, Aereo avoids having to pay the broadcasters for their contents.

Looking at the economics of this issue, we can draw some further conclusions. The long-term strategy of Aereo need not rely on the one-antenna-per-customer technology. As long as they are able to attract a large enough subscriber base using the current technology, they should be able to transfer to the first best solution. Why is that?

The first best solution is currently not a profitable option, because the broadcasters will not license their programs cheaply enough to Aereo. (As a technical detail, not to license at all can be interpreted as setting a price of infinity). By building the one-antenna-per-customer solution, Aereo shows the broadcasters that Aereo doesn't need their license, and this establishes a credible threat-point (or disagreement-point) for negotiations between Aereo and the broadcasters.

Thus, assume that the monthly cost of the one-antenna-per-customer technology is $10 more per user than the first best solution. This then means that the broadcasters should grant Aereo a license for use of their contents at a price of less than $10 per user per month. Any licensing fee higher than this would be rejected by Aereo, since it could avoid the license by using it's current technology. For the broadcasters, this would be profitable, since any revenue from this licensing deal would be incremental (over and above) to what they otherwise receive, for in the absence of the license their contents would still be available to Aereo's customers.

So the the one-antenna-per-customer technology, while socially wasteful, should perhaps really be seen as a bargaining device, rather than something intended to be a long term large scale solution. The twist to the story is of course that in order for the one-antenna-per-customer technology to be a bargaining device, it needs to be a viable long term large scale solution as well.

As long as the technology is viable, then the end result ought to be the first best solution, with Aereo paying a licensing fee less than the incremental cost of the one-antenna-per-customer solution. This is an example of the Coase theorem  in action - The parties will trade to the outcome that is overall (socially) most efficient. There are obvious limitations to the theorem in general, and to its implications in this particular case. These will be the subject of a separate posting.