Streetwise Professor

November 22, 2017

Obama Turned the Net Into a Neut: But It’s Getting Better!

Filed under: Economics,Politics,Regulation — The Professor @ 9:52 pm

Few subjects generate such intense reactions as net neutrality. It has become freighted with much emotional baggage, largely because it has been framed–artfully–as a matter of free choice and free speech vs. censorship and control of information by malign interests.

The Trump FCC’s announcement of its plans to reverse the Obama FFC’s 2015 net neutrality rule has led brought the issue–and the Manichean rhetoric–back to the fore.

One would hope that applying some basic economics might shed some light, and cool some of the rhetorical heat. I will give it a go.

The basic economic issue is straightforward. It is basically a matter of price discrimination, a subject that economists have analyzed and understood for years. The crucial feature of net neutrality is its ban on ISPs charging different prices for different types or categories of service. So for example, your Internet provider cannot charge higher prices for more intense consumers of bandwidth (e.g., streaming services).

Although the term “price discrimination” has bad connotations–which net neutrality supporters emphasize–it can be good, and it is difficult to identify conditions in which it is unambiguously bad compared to the real world alternative.

One reason that charging different prices for different types of customers can enhance efficiency–and why suppressing the ability to do so can be inefficient–is that the costs of providing a service can differ between customers. Some customers are more expensive to serve, or demand a differentiated service that is costlier to deliver. Providing the price signals that give the incentive to consume, produce, and invest in capacity efficiently requires price discrimination: higher cost customers should pay more than lower cost customers.

This is an issue in providing Internet services. Some services and users that consume more bandwidth, and impose greater risk of congestion on the system. A pricing structure that does not charge such users/services a higher price to reflect these higher costs induces overconsumption of these services, and imposes costs (e.g., poorer quality of service) on those who do not put such demands on the system. Furthermore, preventing ISPs from charging prices that reflect higher cost distorts their incentives to invest in more capacity, or in technologies and congestion management techniques that ease burdens on the system.

Prohibiting charging prices that vary by type of service or customer therefore results in cross-subsidization (low cost customers subsidize high cost ones), which both transfers wealth and undermines efficient allocation of resources.

Price discrimination can also occur as a result of market power. There are different “degrees” of price discrimination. To keep things simple, the most common kind is “third degree price discrimination”, in which a firm with market power who can segment customers based on their demand elasticities: less price sensitive customers pay higher prices than more elastic demanders.

It is plausible that demand elasticities for Internet services differ, and that elasticity may vary by the type of content, e.g., that the demand for streaming services is less price elastic than the demand for email or cat videos. In this case, charging a different price for streaming services vs. more mundane uses of the Internet could well be a form of 3d degree price discrimination.

It has long been known that the welfare effects of 3d degree discrimination are ambiguous: as compared to a single price for all services/customers charged by a firm facing a downward sloping demand curve, welfare (consumer plus producer surplus) or consumer surplus can be higher of lower with price discrimination. Furthermore, if a firm faces substantial economies of scale, the efficient way of covering fixed costs typically involves 3d degree price discrimination (“Ramsey Pricing”).

So one cannot say a priori that even if price discrimination by ISPs reflects market power, suppressing price discrimination improves welfare: the market power remains, and the ISP might exercise it in a way that causes welfare to be lower than if it exercises it by price discriminating.

Moreover, there is reason to doubt that a predicate for inefficient price discrimination–ISP market power–exists, or is more acute in this market than it is in many other markets where price discrimination is common (and believe me, that is pretty much every market). The days of the “last mile” monopoly are over. A very large fraction of Internet users in the US have access to multiple ISPs. Furthermore, wireless service (4G, and perhaps soon 5G services) competes with traditional cable and DSL service. Between wireless and cable, off the top of my head I can think of 8 providers that I can access. Yes, there is some overlap (e.g., ATT provides both types of service), but the number of choices most Americans have for Internet access is greater than they have for many other goods and services. Meaning that it is unlikely that market power problems are so acute in this market as to justify regulations unheard of in other markets where price discrimination is widely practiced.

I should also note that some kinds of price discrimination can unambiguously improve welfare relative to simple monopoly pricing. First degree (rare in practice) or second degree (e.g., quantity discounts, two part pricing) is superior to simple monopoly pricing. I would wager that some ISPs will adopt such efficiency enhancing price policies if freed from net neutrality restrictions.

Thus, if your concern is that ISPs exercise market power by price discrimination, suppressing price discrimination is not the best way to tackle the problem: attack the market power directly by reducing entry barriers or antitrust actions against ISPs. Furthermore, it is not at all clear that price discrimination in this market is driven primarily by market power, given the competitive conditions in the market. Lastly, and perhaps most importantly, attempting to squelch (at best ambiguously inefficient) market power driven price discrimination also precludes efficiency enhancing price discrimination based on differences in service/customer cost. Doing so imposes substantial costs.

For all these reasons, I conclude that net neutrality is (a) a solution to a non-existent problem, and (b) can do positive harm by preventing the development of efficient pricing systems that give appropriate incentives to consume and invest in system capacity, and to optimize its use.

Another aspect of the net neutrality rule is to prevent ISPs from favoring its content (which it either produces, or buys from others) over that supplied by independent parties.  Again, a necessary condition (but definitely not a sufficient one) for this to be a problem is ISP market power, which as noted above is unlikely to be a particularly serious issue. Furthermore, it is typically not in the interest of a downstream firm with market power to restrict its customers’ access to upstream product. Offering suboptimal product variety reduces the demand for the putative monopolist’s services, which reduces its profit. It is typically more profitable for the monopolist to offer the optimal product variety, and profit by charging a higher monopoly price.

There are some rather contrived models in which vertical restrictions (e.g., tying or exclusive dealing) can be used to lever market power from one good to another: the practical applicability of these models is dubious at best, as some of the modelers themselves acknowledge. But there is also an extensive literature (much of it originating in Chicago in the 1960s, including seminal contributions by my thesis advisor Lester Telser) showing that such vertical restrictions are usually efficiency-enhancing responses to some incompleteness in property rights or information problem. Indeed, in US antitrust law, horizontal restrictions (e.g., cartels) receive far more scrutiny that vertical ones, precisely because academic research on the potential efficiency enhancing effects of vertical restrictions, and the difficulty of using them to increase monopoly power, has informed antitrust policy–under administrations of both parties I might add. (As an aside, this makes the Trump DOJ challenge of the Time Warner-ATT deal somewhat strange, and intellectually at odds with the FCC’s move against net neutrality.)

In sum, I favor jettisoning net neutrality. No, I do not believe that the ISP market is perfectly competitive, but that is a red herring. Even acknowledging the possibility of imperfect competition in that market (although I do believe fear thereof is overblown), net neutrality is not the right way to address it, and indeed, might actually mean that market power is exercised in a way that reduces efficiency. In other words, the Obama FCC wanted to fight ISP market power in the worst way–and it did!

So if net neutrality is an inefficient policy, why did it prevail in the US, at least for a while? That is, what is the political economy of net neutrality?

Well, Chicago has a lot to say about this as well. Indeed, the work of another of my former advisors–Sam Peltzman–is directly on point. Sam’s amazing 1976 JLE article “Towards a More General Theory of Regulation” has an important, but widely overlooked prediction: regulation is likely to occur in industries where there are substantial differences in costs of serving different customers, and that regulated price structures suppress these cost differences. That is, regulated price structures cross-subsidize high cost customers. As Sam put it: “cross—subsidization follows a systematic pattern in which high cost customer groups are subsidized by low cost customers.” And: “The important contribution of politics is to suppress economically important distinctions and substitute for these a common element in all prices.”

That is net neutrality in a nutshell. Put simply, the Obama FCC bought political support Google, Facebook, Amazon, Microsoft, Netflix, et al by implementing a policy that cross subsidized their services. They used the political system to push regulation that suppresses economically important distinctions.

This result is less surprising from a political economy/public choice perspective than the Trump administration’s reversal of net neutrality. My first stab at an explanation is that this reflects the fact that a Trump administration can never expect to obtain political support from these companies, and can doesn’t really have to fear additional opposition: they already hate him with the heat of 1000 suns. So why  not stiff them?

Karma is a bitch, eh?

Now for some laughs!

Print Friendly

8 Comments »

  1. We did do the nose. And the hat.

    Comment by Christopher — November 23, 2017 @ 8:29 am

  2. The Canadian and Indian governments have come out and said something different to the US on this.

    Comment by Vanesa Rose — November 29, 2017 @ 12:34 pm

  3. I appreciate the economic analysis and agree with it. I’d like to play Devil’s Advocate and articulate the fears of VCs: how can I fund a startup that will potentially use bandwidth and then won’t be able to afford it? How can I fund the next generation of innovation if only Google, Facebook Netflix etc can afford to pay for bandwidth? Isn’t repealing net neutrality automatically giving them an oligarchy?

    Comment by Jeffrey R. Carter — December 1, 2017 @ 6:09 am

  4. @Jeff–If Google et all thought that ditching NN would cement their oligarchy, they would be supporting Pai intensely. Instead, they are NN’s biggest boosters and propagandists. They are the beneficiaries of the cross-subsidization inherent in NN.

    As to your startup concern. I see things exactly the other way. Paying for the costs imposed on a network won’t occur until a firm gets successful and big, at which time it likely can afford to pay. While it is starting out, it will impose little burden on the network. In a NN world, this wouldn’t matter because it would be the source of cross subsidies, and effectively taxed by paying the average cost of congestion/bandwidth even though it imposes a less than average cost on the network. In a non-NN world, it would pay a lower price than Google et al, commensurate with its lower cost. This would make it easier not harder for startups to get traction. NN is a tax precisely on this type of firm.

    The ProfessorComment by The Professor — December 1, 2017 @ 8:19 am

  5. Hiya,

    Thanks for the analysis. I was wondering where you were getting the data to support this statement: “The days of the “last mile” monopoly are over. A very large fraction of Internet users in the US have access to multiple ISPs.” and if you could be more specific on the fraction of US internet users that have access to multiple broadband ISPs.

    Thanks.

    Comment by Dan — December 12, 2017 @ 12:47 pm

  6. @Dan–This is a good source.

    The ProfessorComment by The Professor — December 13, 2017 @ 10:48 am

  7. The conclusion of this article is totally absurd..the GAFA are gazillionaires, they don’t give a f*** about ISPs which are just insects compared to them (but still annoying, like mosquitoes or head louse) . If the big Content Providers really wanted to control the full chain of value, they could build their own network instrastructure very easilly, and tell them to go f*** themselves..in fact Google already made a step forward in this path, and it’s likely that others will follow…it will the new FTTH era..FaceBook to the Home !

    Comment by clement game — December 15, 2017 @ 3:29 pm

  8. @clement-If NN is so irrelevant to GAFA, why do they exert so much effort and spend so much money on its behalf? Because it’s a helluva lot cheaper than building their own network, perhaps? The absurdity of your comment is demonstrated by its clear inconsistency with the behavior of GAFA on this issue.

    The ProfessorComment by The Professor — December 17, 2017 @ 5:25 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress