Long-range versus Short-range Spectrum Licensing

by on September 7, 2006 · 40 comments

I’m pleased to see that Tim Wu took the time to respond (here and here, scroll down to the bottom of the comments) to my recent posts on his paper on Hayek and intellectual property. Here’s what he had to say on the spectrum issue:

The point made by Tim Lee is decent. It is certainly true that the FCC would have to state some kind of standard to make possible permissionless entry into the spectrum market (as it does for the garage band used by 802.11b). In addition, private actors could, if they wanted, similarly allow permissionless use of spectrum. The question is why they would want to.

In general I cannot understand the strength of Jerry’s and others’ objection to the substance of rules that would create permissionless market entry into the spectrum market. In my view, reflected in that paper, permissionless market entry is one of the holy grails of an effective market system.

Perhaps Jerry will jump in with his thoughts, but I think it’s crucial here to distinguish between short-range and long-range spectrum. For short-range transmission, Wu’s argument has a lot of merit because short-range wireless applications are nearly non-rivalrous. Cordless phones and WiFi seem to work quite well in an unlicensed environment.

A big part of the reason for this is that there are only a handful of people who want to transmit short-range signals in any given geographical location. There are only half a dozen WiFi networks within range of my apartment, and I live in a dense urban environment. Because I’m only competing with a handful of people, informal sharing mechanisms work pretty well. In this case, the WiFi protocol can operate on several different “channels,” and access points self-organize by selecting a channel where their signal won’t interfere with others (at least that’s my rather limited understanding of it–geeks please correct me if I’m wrong).

But I don’t see how this model can possibly scale. When people want to transmit 5 miles or 50 miles, rather than 50 yards, then informal solutions are likely to break down. Unless someone is setting rules concerning who can transmit at what frequencies, and what protocols they use, you’re bound to get a tragedy of the commons in which spectrum is over-used, resulting in no one being able to send clear signals.

Hence, for long range signals, you need rules more complicated than “don’t transmit with more than X watts of power.” Choosing the best rules is likely to be extremely important to getting the maximum use out of the available spectrum. Moreover, the best rules are likely to be complicated, and to change over time. If we put the FCC in charge of making the rules, they’re likely to choose badly, and they’re unlikely to change the rules in a timely manner as technology develops. Therefore, it’s important that the rule-making process be decentralized as much as possible.

Property rights are a mechanism for decentralizing decision making over the use of a scarce resource. To the extent that spectrum is scarce (and it clearly is for long-range applications, although it may not be for short-range), someone will have to decide on its allocation. The only alternative to property-based allocation is centralized, bureaucratic allocation.

Wu asks why private actors would allow permissionless use of spectrum. He’s right that they wouldn’t allow truly permissionless use of spectrum, but it’s easy to imagine a private actor could find it to its advantage to engage in a liberal licensing policy that would have virtually the same effects. Imagine an alternate universe in which all spectrum were propertized. It’s easy to imagine someone buying up a block of spectrum and offering it at a low, fixed rate (say $2 per device) to anyone who wanted to use it for low-powered devices. It seems to me that this arrangement addresses Wu’s concerns about decentralized decision-making nearly as well as unlicensed spectrum.

Moreover, there’s every reason to think that such a strategy could be highly profitable. There are hundreds of millions of devices that use unlicensed spectrum. It’s seems probably that in an environment in which all spectrum were propertized. the profits from offering spectrum on such liberal terms would exceed those from using it for a monolithic purpose such as television broadcasting. Moreover, different firms would likely experiment with different spectrum-sharing rules. The ones that found the most efficient sharing rules would be able to sign up the most licensees and make the most money.

I think this points to an important final point: if I’m right that the correct policy is property rights for long-range spectrum and a commons for short-range uses, there remains the question of how spectrum should be allocated between short- and long-range uses. I think the discussion of the previous paragraph suggests that we should err on the side of more propertizing. If we propertize too much, companies will find it profitable to buy up spectrum and re-license it at low rates to companies making short-range devices (who will be willing to pay extra to avoid the congestion of the unlicensed bands). In contrast, if we create too much unlicensed spectrum, I don’t see any mechanism for converting some of it into licensed spectrum.

Comments on this entry are closed.

Previous post:

Next post: