Today’s lecture in my occassional “Economics of Market Power” series comes from the hot policy debate over whether we should let dsl (and cable) providers charge third parties for “premium” speeds to reach their customers. I call this behavior “Whitacre Tiering” (as distinguished from other sorts of tiering traffic or bandwidth) in honor of AT&T CEO Ed Whitacre, a chief proponent of the concept.
Last time, I explained why permitting Whitacre tiering would be a disaster for democracy. This time, I’ll explain why Whitacre tiering produces really, really awful results from an economic perspective. It gives actors all the wrong incentives, adds new layers of uncertainty and inefficiency to the market generally, and discourages investment in bandwidth capacity at every stage of the network (thus aggravating the broadband incentives problem you may have read about recently, rather than solving it, as some defenders of Whitacre tiering maintain).
But hey, don’t blame me, I’m just the messenger! Go do the math yourselves. All you need is a basic knowledge of Econ 101. OTOH, if you have a religious belief, possibly supported by self-interest or fueled by PAC money, that all deregulation is good and all regulation is bad, mmmkay (not that Senator Enisgn is likely to ever read this), I expect you will remain unpersuaded. Rather like passionate believers in Ptolemy’s geocentric model of the cosmos, I expect the true believer neo-cons, the companies whose self-interests are implicated, and their wholly owned subsidiaries in state and Federal legislatures, to devise theoretical models and epicycles to explain away all the nasty empirical problems and assure me I live in the delightful world of competition and frictionless switching to competitors.
It moves, it moves . . . .
Paul Klemperer observed in his priceless paper “Using And Abusing Economic Theory” (available with a bunch of other good stuff here) that while a little knowledge of economics is a good thing for public policy, too much knowledge of economics can be a problem if you allow elegant theory to seduce you away from messy reality. I, happily, am safe from any accusation that I know “too much economics.” As a result, my recent infatuation with the “dismal science” (arising from my frustration with repeating myself) tends to focus on the practical and what Klemperer calls “undergraduate level” economics.
First I will define some terms.
“Customer Tiering”- is when an end user pays more money for more bandwidth. The key point is the customer decides on the size of the pipe, and all bits within the pipe get treated equally (or, possibly, in a manner designated by the customer/end-user).
“Whitacre Tiering” – is when the ISP demands money from a third party for premium access. The ISP makes the decision on how to prioritize packets without regard to the end-user/subscriber preference. Indeed, the end-user may not even be aware that Whitacre Tiering is taking place.
“Provider Provisioning” – is when a third party invests in getting their content or service out as quickly as possible to subscribers. This can be through buying more servers, using technology like Akami or bit-torrent, or whatever. The point is that it is the provider that makes a decision about how much to spend to push out its content.
Customer tiering and provider provisioning rely on an end-to-end architecture to work effectively, since it puts the decision on how much to spend for capacity at the end points. Some technologies may work in tandem. For example, a virtual private network (VPN) would be an example of a form of customer tiering (since the customer is making a decision about how to handle certain packets) and provider provisioning.
Whitacre tiering, definitionally, breaks end-to-end because the network provider must have the freedom to make concrete decisions about treatment of packets without regard to the preferences of the parties at the ends of the networks. Ed Felten has written two good pieces, available here and here, about how this would work from a technology standpoint.
Ed Felten knows a heck of a lot more about this than I do, and I haven’t heard any companies say they can’t prioritize packets (which is a shift from when the debate started in 1999 and the cable companies claimed it was impossible to discriminate based on traffic content or point of origin). So I’ll take as given that any competent broadband access provider can opt to slow down or speed up traffic to the end-user, without the end user being able to do anything about it.
I have previously argued that this is a bad idea from a public policy perspective because it will do serious harm to civic discourse and political speech. Now I want to explore how the economic incentives play out.
Until the Brand X case back in June 2005, and subsequent deregulation of broadband by the FCC last September, we lived in a world that permitted customer tiering and provider provisioning. It did not permit Whitacre tiering. This produced a pretty good internet, over all. And, looking at the economic incentives, this makes sense.
Without Whitacre tiering, content and service providers keep scaling up the bandwidth of content and services until they hit the point that customers can’t easily download them and network congestion becomes an issue. This congestion triggers two positive benefits from a public policy perspective. First, customers will pay more for bandwidth, because they need bigger pipes to get high-bandwidth services. So broadband ISPs have incentive to build bigger pipes and sell them at a higher price, because enough customers will pay more to make this profitable. The bigger pipes, in turn, spur higher-bandwidth intense uses, prompting a continuous cylce of upgrade. This cycle also improves things for the customers that pay for lower speeds, since today’s “high-speed” 5 mbps pipe is tomrrow’s “standard” speed and next week’s “discount” speed.
At the same time, providers have an incentive to engage in provider provisioning. Any provider that can afford it invests in technology that pushes its content faster or improves reliability of its service. Depending on the technology, this does more than improve performance for the specific provisioning provider. It also improves performance for the broader “network of networks” overall. Traffic taken out of the regular backbone and routed in some other way, or that is cached closer to users, is traffic that doesn’t clog the backbone for everyone else. The effect is rather like encouraging people to take public transportation. Fewer cars on the road mean a general reduction in traffic for everyone, not just the folks skipping the traffic jams by taking the bus or subway.
So customer tiering and provider provisioning create a virtuous cycle from a public policy perspective. At every stage, the relevant actors have incentive to build bigger pipes, invest in the network, and innovate with new services. To use a real world example, Ms. Customer has dial up, then buys an iPod. Suddenly, dial up is way too slow, so the customer springs for a 1.5 mbps cable modem. At the same time, Apple has incentive to invest in technologies that push its music downloads to willing customers faster. Apple’s investments don’t just help Apple, they help network congestion as a whole by clearing Apple’s traffic off the more congested portions of the backbone.
A few years later, Apple adds video content to iTunes. For Ms. Customer, a 1.5 mbps pipe has become way too slow. Now she needs more bandwidth. Since she is not alone, the cable guys have been upgrading and now offer a 5 mbps service (at a higher price). Ms Customer switches to 5 mbps, as do a growing number of others, because iTunes isn’t the only service using more bandwidth. Cable can now offer “cheap” broadband of 1.5 mbps at a “bargain” rate to attract more customers, 5 mbps becomes the “standard,” and high-bandwidth users like Ms. Customer will pay a premium for 7 mbps. Meanwhile, Apple and everyone else keep trying to push their own content faster, since there are still plenty of 1.5 mbps customers and even dial up customers they want to reach.
Whitacre tiering screws all this up. As an initial matter, it is important to note that Whitacre tiering doesn’t actually add any value. It is a revenue stream that comes solely as a function of holding a large enough number of subscribers “captive,” so that businesses will pay to get “premium” access rather than move down the slow lane. Or, as Ed Whitacre so succinctly put it the first time he explained the concept, it’s about charging third parties like Google to access “my customers” rather than allowing third parties to “use my pipes for free.”
As these sentiments did not play well, Mr. Whitacre and opthers have spent a great deal of time explaining that no, of course, that’s not what this is all about. But straight forward economics (as opposed to public policy) tells a different story. Companies exist to maximize revenue. The cable and telephone companies have a large customer base, an ability to prioritize packets without customer detection (so customers do not know if switching will help), few facilities based competitors, and significant customer lock-in as a result of switching costs (with no guarantee that switching helps). Of course these companies will try to maximize revenue by charging third parties all the market will bear for access to its customers. Any belief that, absent regulation to the contrary, broadband ISPs will not try to squeeze as much out of third parties as possible is wishful thinking at its worst.
So let us examine what happens in the new world of Whitacre tiering, where only “the market” constrains the ability of companies to charge third parties for “premium” service. Turns out it really screws up incentives for just about everyone.
First, Whitacre tiering removes the incentive to build a bigger pipe for the end-user customer. Worse, it creates an incentive not to build a bigger pipe. Why? Because the value of the “premium” service offered to third parties depends upon it being both scarce and necessary. No party will pay more to deliver its content than it has to. So the ISP has to make Whitacre tiering worthwhile. As a result, as services become congested, the ISP has every incentive to simply sit there and charge a higher price for “fast lane” service to the customer. In doing so, the ISP also saves money, because it does not need to invest in a general upgrade of its systems and wait for the gradual migration of customers to pay for its system upgrade. So not only does the ISP make more money by substituting Whitacre tiering in place of an option to customer tier, it avoides expenses.
Result: instead of an incentive to expand bandwidth availability to the customer, the ISP now has an incentive to supress bandwidth availability as an absolute matter (since it can charge both the customer and a third party for the equivalent of a bandwidth upgrade by using Whitacre tiering).
This has repercussions in the rest of the chain. Providers now have much less incentive to engage in provider provisioning, since ISPs can essentially nullify their investment by slowing down the packets to “non-premium” speeds when they hit the ISPs network. It also discoruages providers from investing in high-bandwidth content, since content that requires “premium delivery” by Whitacre tiering is more expensive than lower bandwidth content. (I recognize that for some people, this kind of “metered pricing” may be a plus. But from a public policy perspective, discouraging third parties from bringing new content or services to market is usually regarded as bad, not good.)
Even without degrading traffic when it hits the end users network, Whitacre tiering still provides a disincentive to invest in provider provisioning, since a provider’s competitive advantage in investing in provider provisioning can be neutralized by Whitacre tiering. In addition, this produces two very bad results from a public policy perspective. First, the virtuous cycle of innovating in delivery mechanisms is reversed. This previous relief of network traffic overall does not occur, since people keep their traffic on the network until it reaches the ISP (where it is “Whitacred” to the customer).
To revert to our traffic model again, Whitacre tiering is like charging a toll for access to HoV lanes. While it works great for those who chose to pay the toll, it reduces the number of lanes available to those who don’t play, so the folks who do not pay the toll experience an increase in congestion.
The second bad result is the introduction of a whole new layer of transaction costs and uncertainties in the broadband content/service business as a whole. If history is a guide, all of these deals will be subject to non-disclosure agreements (NDAs). Rather than the efficient market resolving traffic issues and producing innovation extolled by the defenders of Whitacre tiering, we will see a horrendously inefficient market where no one knows how to price the premium access, there is no basis for comparison, and there are potentially a large number of deals that any content provider will have to make to ensure access to customers. Nor can ISPs efficiently price their Whitacre tiers, since they do not have information on how anyone else is pricing this access. We can therefore expect to see a dramatic rise in transaction costs associated with bringing any new content or service online (which must ultimately get passed on to the customers, further discouraging innovation and uptake of broadband services because of the increase in prices to all parties driven by the increased transaction costs).
The costs to industry increase because of the number of ISPs with which a given provider must deal in order to make to ensure access to the right markets and customers.
Provider provisioning works universally (depending on the nature of technology) and at an understood cost. To achieve the same effect through Whitacre tiering, I need to make a deal with every ISP that serves some threshold number of customers in every market I want to reach.
So to reach New York City customers, for example, it is not enough to cut a deal with Verzion. I will also need to cut separate deals with Cablevision, Time Warner, and any other broadband provider of note in the city. That takes an awful lot of time and effort. And how shall I recoup this cost? Differentiate price to the customer based on their ISP? Average the cost over all customers, so that Verizon customers subsidize Cablevision customer’s “premium access?”
(As an aside, we will also create the potential for a new phenomenom I will call “virtual red lining.” If I have to pay for premium access, damned if I am going to pay extra for “undesirable” customers I don’t want. So an expected market “efficiency” will be for the ISP to offer to sell me premium access based on zip code or other forms of subscriber information that better fit my profile. I shall leave it to the reader to decide if this is a cool new market feature to be encouraged or a social ill to avoid. I’m just making an economic prediction.)
Finally, we will see the same problem emerge in the broadband services/content space that exists in the cable space. Popularity of programming in the cable space is based on what the cable operator chooses to make available on the system. My son used to love PBS Kids network. It was taken off DIRECTV and replaced with Sprout, a network oriented toward much younger kids. But there is no way for me to make my displeasure known because I would have to switch to cable, which is simply not worth it for this level of inconvenience. Similarly, although lots of Comcast customers want to the local baseball team, the Nationals, on TV, Comcast refuses to carry the Nationals. Although customers can switch to either DIRECTV or RCN and see the Nationals, that is way to much effort just for baseball. So Comcast can refuse to carry the Nationals until the Mid-Atlantic Sports Network agrees to fork over a share of equity to reach Comcast’s customers. (Interested folks can follow all this in the FCC’s Comcast/TW/Adelphia transaction docket, 05-192, or read my general summary of how cable market power works here.)
Whitacre tiering replicates the same kind of access issues and incentives found in cable because the ability to charge for Whitacre tiering derives from the same source — control of subscribers as a consequence of lock-in. That’s bad news if we are counting on the internet to continue to be a driver of economic prosperity and job creation. As observed above, scarcity, rather than abundance, is necessary for the broadband providers to maintain control and collect rents from both ends of the transaction.
Ultimately, this is bad for the ISPs themselves, as they will expend resources trying to negotiate the best deals, over price access, under price access, and generally behave inefficiently in an inefficient marketplace. By the time this becomes obvious, however, it’s too late to back out. Every ISP will be chasing the pot of gold at the end of the Whitacre rainbow, and won’t be able to stop chasing it because to do so might yield an advantage to a rival or forgo a revenue stream a rival has found.
This is simlar to what has happened in commercial radio and television. Consolidation has driven down quality of programming costing listeners and viewers, while not providing the promised synergies. In Coasian terms, the firm has grown to a point where the expenses of maintaining the firm no longer outweighs the efficiencies. But the companies remain stuck in the same self-destructive patterns.
Normally, we let businesses make bad decisions and go bankrupt if they so chose. But the internet has become a critical piece of our national infrastructure, as well as an important platform for non-commercial speech. We don’t let drunk drivers pay the “Darwin tax” because they also represent a danger to others. Given what’s at stake, a modicum of caution appears warranted to me rather than allow companies to learn what a stupid idea Whitacre tiering while crippling the development of the broadband internet and associated industries.
The pre-Brand X universe of customer tiering and provider provisioning worked extremely well. There is lots to suggest that Whitacre tiering will screw this up. Others may trust their religious conviction in the inherent superiority of “deregulation” and “the market.” I am not so religiously inclined. I can only hope that Congress will likewise resist the urge to bet against the odds and the basics of economics because of a religious belief that “regulation is bad, mmmkay.”