Nicholas Economides† and Benjamin E. Hermalin have a new paper (pdf), coming out in the Rand Journal of Economics, summarized here, here is the abstract:
Pricing of Internet access has been characterized by two properties: Parties are directly billed only by the Internet service provider (isp) through which they connect to the Internet. Pricing, moreover, is not contingent on the type of content being transmitted. These properties define a regime known as “network neutrality.” In 2005, some large isps proposed that application and content providers directly pay them additional fees for accessing the isps’ residential clients, as well as differential fees for prioritizing certain content. We analyze the private and social incentives to introduce such fees when the network is congested and more traffic implies greater delays. We derive conditions under which network neutrality would be welfare superior to any feasible scheme for prioritizing service. Extending our analysis to encompass isps’ incentives to invest in more bandwidth, we show that the ability to price discriminate increases their incentives to invest. In terms of overall welfare, we show the additional investment may or may not offset any static inefficiency associated with discrimination.
This strikes me as a funny way to put the point. Might the authors also have written?: “A strict welfare calculation is indeterminate in the theory of price discrimination, yet the standard presumption is that price discrimination is welfare-improving. That is all the more likely to be the case when investment in new capacity, and usage, is endogenous”?
For the pointer I thank Richard Harper.
The standard presumption is that price discrimination is welfare-improving, unless the real cost of implementing the price discrimination scheme is sufficiently high. It does not appear to me that investment in new capacity is part of the cost of implementation.
The presumed social welfare cost of price discrimination is that it will be used to establish vertical monopolies.
Holy stacked deck, Batman! Furthermore their model also cheerfully assumes that all content providers are still connected to the same World Wide Web, despite their now-ample incentive to try to vertically integrate in the opposite direction, setting up their own ISPs. Remember the Microsoft Network of 1995? There are also no network effects in this model; no content provider is more valuable because other content providers can be accessed on the same network without additional marginal cost; all content is “consumed” individually like cable TV. This is a rather startling way to consider the value of the Internet.
And after all that, the best they can do is mutter about an indeterminate welfare effect?
The authors are making a reasonable simplifying assumption to seek the welfare consequences which, under even these assumptions, is indeterminate. A result of indeterminance isn’t “muttering.” It’s quite enlightening of the dilemma. The welfare consequences certainly arent going to become more clear when you relax these assumptions.
Undoubtedly the ISPs will supply their own content at potentially anti-competitive transfer rates. But to do so invites the wrath of the FTC on top of the FCC and Congress.
These are not reasonable assumptions, given that these are the issues that one side of the network neutrality dispute emphasizes as fundamental to their case. Solving a policy dispute by assuming away your opponent’s case is not enlightening.
It is generally the case that the ability to price discriminate improves the supplier’s surplus in a commodity exchange market and, intuitively, their incentives to invest in producing those commodities. Indeed what would be unusual if it were not so. It’s fine to publish null results, of course. It’s not so fine when the null result is written to hide the rather unusual assumptions being made.
I remember both MSN and AOL and a couple other vertically integrated ISP/Content Providers. The fact that they are only memories implies to me that vertical integration does not provide significant monopoly power in the Internet marketplace. The AOL/Time Warner merger should provide a good case study for why vertical integration isn’t dominant and why it’s unlikely a Facebook/ATT or Google/Verizon merger wouldn’t likely dominate either.
Content, especially the way it is trending towards ever more diverse creators, just doesn’t gain from integrating access, and vice versa.
Isps have differential charges for speed and volume of their customers now. It is hard to see how charging third parties can produce more price discrimination than already exists. It is easy to see how it extends current monopolies making more investment on their end worthwhile while making less investment by third parties worthwhile.
Re your comment that price discrimination will induce expansion of capacity.
Price discrimination is indeterminate, and dependent on industry structure for its welfare effects. For example, if there barriers to entry, price discrimination can heighten them through various pricing and bundling strategies.
For that reason, price discrimination is indeterminate and dependent on structural issues. Price discrimination does not necessarily mean expansion of output.
By the way, the best site on network economics is Economides webpage which assembles most of the public articles on network economics. Here is the link: http://www.stern.nyu.edu/networks/site.html
“That is all the more likely to be the case when investment in new capacity, and usage, is endogenous”
What does it mean that investment is endogenous?
The decision on whether to invest depends on price and usage, and vice versa.
It means that they can vary based on other conditions in the model. In this case, the ISP gets to decide how much to invest and users get to decide how much to use. If they were exogenous (the opposite of endogenous), the level of investment or usage would be completely static and imposed from outside the model. Endo- refers to something inside and exo- refers to something outside.
Hi Tyler, commentators: can you please translate this to English for us who have only take a mere two semester hours of economics but still have their Macroeconomics textbook by Blanchard? Thanks in advance!
In the garden variety case of price discrimination, an airline charges customers according to their willingness and ability to pay. Unlike markets for other goods, airlines can segment its customers by flight characteristics. They can charge business customers more and non business customers less. The airline would “steal” the consumer surplus of the customers who value the flight highly. The flipside is that low value customers get tickets at low prices. Total surplus might increase, output increases, but there is a transfer of surplus from consumers to producers.
The authors here are looking at a very different type of good, and the welfare effects of the “classroom model” of price discrimination might not hold. Does output increase? Does total producer and consumer surplus increase? Does investment increase?
What makes this different is that the internet has properties of a congestible public good (a shared resource) and natural monopoly (more efficient with fewer providers because of declining long run average costs throughout the relevant range of production)
The upshot is that under several simplifying assumptions, the welfare impact of price discrimination is indeterminate. They don’t know whether total surplus will increase or whether output will increase with fees on content providers. They determine conditions where net neutrality (no price discrimination or prioritization) is welfare improving.
Then they show that fees on content providers can increase investment in new infrastructure because ISPs have an incentive to add bandwidth.
If I got anything wrong, the smart economists in the room will pounce all over me. I’m just a lawyer.
Thanks lawyer, I think your speculations are about as good as mine–my mentioning the Coase Theorem we pretty much are saying the same thing (whether a consumer surplus or producer surplus is immaterial to whether there is a surplus and output increases–the Coase theorem does not care who gets the surplus, as long as there is a surplus). You are correct: the paper seems to be stating assumptions that make Net Neutrality work. As such, it’s a typical speculative econ paper with possibly no real world significance.
No edit button on this blog? But I’ll take a stab at answering my own question, years after taking my two semesters of econ: “>>> Might the authors also have written?: “A strict welfare calculation is indeterminate in the theory of price discrimination, yet the standard presumption is that price discrimination is welfare-improving. That is all the more likely to be the case when investment in new capacity, and usage, is endogenous”?<<<
This says that due to the Coase Theorem assumptions, when transaction costs are minimal it really does not matter who pays. However, a presumption that price discrimination is good derives from the fact that the same good may have different utility to different classes of people, so price discrimination is good (simple example: standing in a queue in a supermarket aisle–if time is money, then some time-starved shoppers should be allowed to cut in front of the queue by paying extra money). This presumption has the assumption that the event is endogenous, meaning arising from within the model. So in the internet example, if the class of users that use and invest in new capacity in the internet are the same players who engage in the price discrimination, namely, if the class comprises only ISPs and consumers, then price discrimination is OK. Left unsaid is whether there is another group of users not a member of this class–maybe 'outside potential internet users or investors' who are not participating in internet usage or investment scheme–if this is the case then the model would fail.
Now think about this: I am not an economist and it's been years since I seriously looked at economics: if any of the above, written by a layperson, is even remotely close to the truth then your industry, Mr. and Ms. Professional Economist, has very little depth and substance. LOL.
When writers use overcomplicated phrases in an artIcle trying to convince me to take a position on a topic, I assume they are trying to trick readers into agreeing because nobody wants to say “I don’t understand what in the blue hell you are talking about”.
“A strict welfare calculation is indeterminate in the theory of price discrimination, yet the standard presumption is that price discrimination is welfare-improving”
Throwing this sentence out there like it is normal dinner table English is absurd. There are few that would understand what this means without having to decifer it using several google searches, and even fewer that would correctly understand what it means in the context of the article.
The writer could simply say, making people pay more for their Internet is good, or making people pay more is not good. Instead they say the equivalent of “people paying more is not necessarily less not good than people paying less than it otherwise could cost”.
I just want to know if being for net neutrality means things stay the same.
The statement is perfectly understandable if you remember Principles of Micro.
Instead of welfare, think of consumer plus producer surplus.
If producers knew what every customer was willing and able to pay, the producer would charge each customer that price. The producer would “steal” every dollar of consumer surplus. However, it would improve TOTAL surplus because output increases. People who couldn’t afford the good before now get as much as they want. The problem is distribution – there is more total surplus, but the producer gets all of it!
Whether or not a particular scheme of price discrimination in the absence of perfect information increases total surplus and output cannot be proven. But our textbook models suggest that it is certainly possible if not likely.
The authors want to know if having net neutrality means the internet will become so congested that it is virtually unusable. It might become so if everyone starts streaming large quantities of data and ISPs have no incentive to invest in greater bandwidth.
So no, it is not likely that net neutrality is a good thing. The paper does not determine whether fees on content users improve overall happyness.
They prove, under their assumptions, that investment will take place with such fees, I.e. more bandwidth.
Willi, The expansion is dependent on industry structure. Say your internet provider is Comcast, the local cable company and internet service provider and now an owner of NBC. It has an interest in preserving its other assets–over the air broadcasting and the cable offerings–what are its incentives to price high bandwith usage of direct streaming through Netflix?
Industry Structure matters when you deal with price discrimination because the revenue can be recovered over many platforms.
What will probably save us all from cable would be the sale of more over the air bandwith and ubiquitous wifi.
True. True.
Fixed infrastructure lends itself to natural monopoly. Flexible infrastructure is more competititve. For example trucks vs. trains, rail vs. road.
Inudstries usually cycle back and forth between fixed and flexible networks because each has its advantages and disadvantages. For example, cable internet is faster and more secure, but its hard to find a cable jack every place you want to be.
I’m not saying that anti-competitive pricing for other content providers isn’t a problem. I’m simply saying the issue of efficiently pricing usage on the fixed I.frastructure is interesting in its own right. Clearly the political aspects of net Neutrality are muddied by market structure.
The net-neutrality fight is essentially between two groups of companies both trying to drag revenues their way. ISP’s, routing gear makers and telecom majors versus Google, Vonage, Amazon, Microsoft etc.
What’s funny is how each side has craftily drafted others to fight their battles. e.g. Google etc. scaring end-users of doomsday scenarios were net-neutrality to be violated.
OTOH, the ISP’s and telecom guys have somehow gotten the libertarians convinced this is a critical freedom issue. (e.g. Cato Institute, the Competitive Enterprise Institute, the Goldwater Institute, Americans for Tax Reform, and the Ayn Rand Institute are apparently all opposing net-neutrality )
Overall, as a consumer I’d rather vote in favor of neutrality because I’m not convinced how a non-neutral regime would make me better off.
You’ll be convinced you are worse off when your internet slows down to the point that it would be faster to send letters via US Postal Service, the only way to get movies would be to go to Red Box or a crowded theater, you will have to pay $2.59 a minute for calls to India on AT&T, you will have to go to brick and mortar stores to buy a book, you must listen to music on CDs, and Tyler would have a new blog post once a month that takes 30 minutes to download.
Net Neutrality: The entire world’s population, information, and entertainment in your living room…all at the same time! Woo, party like it’s 1989!
In economics class, didn’t you people get the part about scarcity? I don’t have a dog in this fight over who gets the steak and who gets the kibble, but I do know that scarce bandwidth has to be rationed, and it is best rationed by price. I know that greater investment in new technology comes from the prospects of higher profits from increased prices and usage.
“No one ever goes to that restaurant. It’s always too crowded.”
It’s already rationed by price. My ISP certainly sends me a bill.
The question is whether your ISP can or should charge the people who provide you with content. If you stream large quantities of data, the amount of that data and its priority becomes an issue. The content provider, e.g. Netflix, makes money on the deal. The provider might be purely advertising based, so you aren’t paying them directly for content.
If lots of people use streaming data, then the network becomes congested. Priority and amount of data must be rationed. If not, internet speeds will grind to a snail’s pace and high value traffic won’t necessarily get priority over low value traffic. Net Neutrality forces the internet to treat all traffic as equal value, equal priority. That is why it is inefficient. If ISPs can’t profit from high priority, high value traffic, then they have no or lessened incentive to expand bandwidth.
At least that’s how I understand it. Some people here seem to know a lot more about networks and technology than I. But people also have a dog in this fight. They want to use Netflix or Bit Torrent without paying for the congestion it causes. The ISPs want to use their monopoly power to charge both ends of the transaction or merely to price discriminate in general. Too many financial interests to have a dispassionate discussion of the positive and normative aspects.
You are in desparate ignorance on this issue. People pay for bandwidth. You want to download movies on netflix? You pay your ISP. Bit Torrent? Same thing. Many ISPs have uncapped sevice, though, and this policy does potentially harm access speed for low volume users. But net neutrality doesn’t prevent your ISP from finding some way off charging the guy constantly streaming video if that is actually a problem. Internet traffic is always paid for at both ends. And when your ISP has to pass of the data to another network, they either trade bandwidth or pay for it (out of the funds you provided). Nobody is getting to push their data on the network for free. All traffic is equal priority because it is all paid for. When your local ISP gets to start slowing down netflix traffic, what that means is that instead of just charging users to provide the bandwidth they are most likely looking to setup their own vertical monopoly in content generation and delivery.
Net neutrality already allows for pricing of internet traffic. This is already quite obvious. Opponents of net neutrality have to establish why content based discrimination will have additional beneficial results when it is so likely to lead to the development of vertical monopolies.
And similarly, Netflix’s ISP/CDNs send them a bill too.
The problem is that your ISP and Netflix’s ISP probably aren’t the same. So Netflix’s ISP uses their revenue to build huge pipes right up to the peering point where they hand off to your ISP, but everybody else’s ISPs just build mediocre pipes from the peering points to your house.
So Netflix’s data moves blazingly fast… about a quarter of the way to your house. The idea is that the other ISPs that Netflix isn’t on can get some revenue from Netflix, not just from their users, and build larger pipes so the data can go blazingly fast all the way.
There are a couple other possibilities that don’t require violations of the net neutrality principal: Netflix could buy internet connections from all the ISPs (or at least all the major ones.) Or non-Netflix ISPs could charge their customers more, or they could cap their customers’ bandwidth usage like they do on cell phones, or they could charge based on bandwidth usage similarly to the old way of charging per minute fees for dialup connections.
Which of these would be the most efficient? I really don’t know, it looks like it might come down to transaction costs. Which would be preferable to end users? I’m thinking the non-net neutral way of charging content providers for preferential treatment would be the most transparent to end users and pit the parties who can most afford to spend time negotiating prices against each other.
Don’t worry the higher speeds and enhanced services will “trickle down” to us all!
\lff
I’m wondering how much of the Internet is already trickling down to us. One of the assumptions this paper appears to make is that Internet traffic is the vast majority of traffic traveling over ISPs backbones. But I doubt that is the case, the same backbone that carries my Netflix stream is also carrying my phone conversation and my employer’s corporate WAN. The revenues and requirements of these will have a major impact on investment decisions, and if they aren’t accounted for we’re just making cute “what if” scenarios.
I only had the time to look briefly at the paper. However, I note that early in the paper the authors state:
========
Over time, the market for connecting websites, particularly commercial sites,
has become relatively competitive. At the same time, however, there is significant
monopoly power in broadband residential Internet access; that is, the connections
households make to their Internet Service Provider (isp). Moreover,
even when isps might compete for residential customers, the fact that these
customers almost always connect through a single isp (single-home) means that
an individual household’s isp has a monopoly on others’ access to it.
==========
And, as I understand their analysis from a quick glance, their analysis is based on the assumption that ISPs are monopoly service providers. If that assumption is incorrect, then much or all of their analysis is flawed. I doubt if their conclusions would hold if users had to pick an ISP at the beginning of each day, even if the ISP were to have a monopoly for the rest of the day.
IIRC, about 20% of U.S. households move each year, and probably most such moves require the household to choose a new ISP after the move.
Moreover, about 25% of U.S. households are single-person households. If I were a single-person household in a location with 4G service, I would certainly experiment with a wireless-only approach to Internet access. Wireless only gives me mobile access and saves me the cost of fixed access at the home. Moreover, I would only need to configure and maintain one connection instead of two.
Given these two factors, I think one must question their assumption that ISPs operate as isolated monopolies. If that assumption is incorrect, then the rest of their analysis is probably unlikely to be correct.
A mental experiment. Suppose Comcast were to block access to Google for a day. Would Verizon despair at the harm to consumers or would they break out the champagne and send sales crews from door to door in neighborhoods in which they provide FiOS service? If you think the latter, then you should have a hard time accepting the results of this paper.
Able
Most broadband markets are a duopoly at best. You can choose between the local cable company for cable modem service or the local telecom for DSL service. It’s highly unlikely that either would be able to make the browsing experience unpleasant enough to make customers switch to satellite. Very few neighborhoods have FiOS service. If you lived in a 4G service area, you might be able to make it work as long as your internet habits included very little media consumption (watching video or downloading a couple games from Steam are easy ways to burn through a 5GB transfer cap).
So if the ISP doesn’t discriminate and prioritize certain content, then popular content will be ‘slower’ for the end user. I wonder what the consumer does if the site he wants is too slow with one ISP but not another….
There is little room for monopolization. Competition is stiff in the communications business. The consumer benefits. (This is not to say that the consumer may rest on his haunches. It is because of his prudence that the ISPs fall in line.)
Not much, because the average US consumer doesn’t have much of a choice with regards to broadband ISPs. If you’re lucky, you get to choose between one cable modem provider and one DSL provider (if you’re REALLY lucky you have the option of FiOS). Satellite is technically an option, but the costs are too high and the drawbacks too great to make it much of an alternative for anyone who has choices other than dial-up.
Satellite is especially bad for latency. VOIP calls might be tricky if not impossible.
This sentence contains an oxymoron. If your clients can no longer access the whole internet, then you have stopped being an Internet Service Provider.
Comments on this entry are closed.