There is a Senate Commerce Committee hearing today on online video, and our friends at Free Press, Consumers Union, Public Knowledge, and New America Foundation argue that it should be used to investigate ISP-imposed data caps.
If data caps had a legitimate economic justification, they might be just a necessary annoyance. But they do not have such a justification. Arbitrary caps and limits are imposed by multichannel video providers that also provide broadband Internet access, because the providers have a strong incentive and ability to protect their legacy, linear video distribution models from emerging online video competition.
As someone who uses an ISP with a data cap and who is a paid subscriber to three different online video services, you might think that I too would concerned about these caps. But to the contrary, I think there are some legitimate economic reasons ISPs might impose data caps, and I don’t see a reason to stop ISPs from setting the price and policies for the services they offer.
The first and most basic reason that ISPs might want to implement a cap is to price discriminate. The term “price discrimination” makes a lot of people uneasy because it contains the word “discrimination.” While that sounds nefarious, price discrimination usually increases economic welfare. If a firm has to charge all consumers $100/month for service, some consumers who can only afford $50/month will be left out. In contrast, if it can charge $50/month to those customers and $150/month to other customers, more customers will use the product. Price discrimination especially benefits those customers at the low end of the consumption spectrum, who would otherwise have to go without.
Comcast and AT&T have 250 GB/month caps on data usage for their residential service. Very few residential customers are likely to bump up against this cap. In fact, residential service is so adequate for most uses that some business customers might be tempted to use residential-class service. By imposing a cap on usage, ISPs are really trying to force these customers to buy their more-expensive business-class service, which does not have a usage cap. This is good for residential customers because it means that businesses are paying a greater share of the fixed costs associated with providing Internet service; business customers are cross-subsidizing residential customers.
Some people argue that price discrimination is bad because it is a sign of market power. However, relatively recent developments in the economics literature do not support automatic linkage between these two elements. Consider for instance that the most notorious price discriminators are airlines, who seem to be continuously going bankrupt! The best paper on the subject is Michael Levine’s “Price Discrimination Without Market Power” (gated published version, ungated working paper). Levine argues that firms in industries with large fixed costs or networks to maintain will be forced by competition to price discriminate to efficiently allocate their fixed costs. From this perspective, it is bans on price discrimination that can be thought of as a restraint of trade.
The second reason that ISPs might legitimately impose data caps is because it is easier for consumers than other superficially more rational approaches to handling congestion. If I were designing a bandwidth pricing scheme for homo economicus, I would impose two-part pricing. Every consumer would pay a fixed fee just to be part of the network, and then a per-bit metered fee so that they bear the costs associated with their own use. Even better, the per-bit fee would be higher when the network was more congested! However, it could be the case that such pricing imposes “mental accounting costs” on ordinary homo sapiens consumers. As Public Knowledge writes in their new paper on Internet data caps and usage-based pricing:
The strongest arguments for flat rates are best explained by the concept of “mental accounting costs.” As the world gets increasingly complicated, people are overwhelmed by the available choices and the need to devote mental efforts to sorting them out, and therefore search for simplicity. They are willing to pay extra for the peace of mind that flat rates offer them.
A 250 GB cap eliminates mental accounting costs for most consumers, relative to the two-part pricing scheme that I proposed above, while eliminating the congestion created by network hogs. My argument is not that a 250 GB cap is optimal, necessarily. ISPs probably have to do some experimentation to find solutions that strike the right tradeoff between congestion management and consumer value. But a 250 GB cap might work relatively well because Grandma doesn’t have to worry about running up her Internet bill. Nor would the vast majority of consumers, for that matter.
A third reason that ISPs might impose data caps is not actually about Internet service at all, but rather about copyright. ISPs quite understandably do not want to be the copyright police. They don’t want to get roped (further) into the ongoing battle between the content-Congressional complex and ordinary consumers. It is in ISPs’ interest, therefore, to find ways to make the content piracy problem go away without alienating the majority of their users. A 250 GB/month data cap probably strikes a nice balance for them on that score. Uncapped service would result in some users running their bit torrent clients at full speed for 23+ hours per day, consuming terabytes of data and distributing lots of copyrighted content. It would also result in further calls by policymakers and content producers for ISPs to do more with respect to copyright. By imposing a cap, ISPs eliminate the most egregious file-sharing practices without overburdening casual file-sharers and without having to monitor users directly.
The bottom line is that off the top of my head, I can think of three legitimate and sensible reasons for ISPs to impose caps. I’m not saying that any one of these is the answer; the real answer might be a combination of these and other reasons. I have no doubt that our friends have the best of intentions, but their claim that data caps have no possible legitimate economic justification suggests to me that either they don’t know much about economics or they’re not trying very hard.