Net neutrality may be harder to achieve than we thought

For the last few weeks I have been working on a paper out of the notes I posted on weak and strong net neutrality. The paper is now done (at least as a working paper) and it seems appropriate to summarise its main findings as some were unexpected.

The first finding is one that I had already forecast. Net neutrality requires alot of neutrality to be effective. Put simply, in situations, where consumers have a direct pricing relationship with content providers (such as is the case with Netflix), if you prohibited ISPs from charging different access fees to different content providers, ISPs could use consumer charges to undo any real consequences of that. Specifically, while such weak net neutrality may change prices in the Internet ecosystem, consumers will end up doing the same thing that they would without any regulation while ISPs and content providers would earn the same amount of profits. This is because ISPs could build in content-based price discrimination into their charges to consumers and effectively allow them to extract rents from content providers: think, as a hypothetical, Comcast charging consumers more for high speed video downloads and reducing their willingness to pay for Netflix. The consumer ends up paying the same total amount for Netflix but consumer payments have shifted from Netflix to Comcast. The conclusion from all this is that you need strong net neutrality — outlawing all content-based price discrimination to actually have a real impact.

The second finding relates to the type of impact that is expected. The mechanism by which an ISP obtained rents from content providers was by charging consumers more for content from ‘high value’ providers. Strong net neutrality will put all content providers on the same terms and so that mechanism will not be available. Thus, content providers will hang on to the surplus they create in competition with other content providers while ISPs will capture only baseline value common across all content. From the perspective of encouraging content providers to invest in new and innovative products, strong net neutrality gives them a profit from that activity related to their value in the market place and prevents hold-up from ISPs. This is something we would think of as a strong positive.

The third finding then examines the critical question of ISP investment incentives. Their investment increases the value of all services. If that increase is correlated positively or is independent of content value, then net neutrality regulation has no impact on the ISPs ability to capture the marginal value of their investments. So this is not a situation where price discrimination allows ISPs to cover fixed costs they would not ordinarily cover. Instead, when it comes to network quality, net neutrality regulation does not stand in the way of investments that increase quality in a more or less neutral fashion. Thus, the concerns of ISPs that such regulation will undermine investment incentives are not borne out; at least with my model. A similar issue arises with respect to fast lanes that have been proposed. Net neutrality regulation does not change incentives to provide these but, if you want to ensure that content providers are properly incentivised to provide content that is optimised for fast lanes, there is a role for regulation that ensures all lanes are fast. How this might be achieved is anyone’s guess but, it must be said, that fast lanes themselves do not undermine entirely the effects of strong net neutrality regulation. That, at least, is surely some comfort.

One wrinkle comes when the business model of content providers does not involve a direct monetary relationship with customers (e.g., they earn advertising dollars). In this case, even weak net neutrality (on the content provider side) can have real effects but it is also possible that the effect of this is to cause consumers not to see price signals based on the value they create for content provider’s in their advertising or related businesses. In this situations, consumers could choose the ‘wrong’ content where ‘wrong’ content is content that creates lower overall social value but higher value to the consumer. It has hard to tell how real that possibility is as one usually associates more popular consumer content with more advertising revenue but the possibility is there.

The final result concerns competition amongst ISPs. Many commentators — myself included — have suggested that if only we had enough ISP competition there would be no role for net neutrality regulation. What the paper bears out is that is only partly true. The paper demonstrates that even strong net neutrality will not be effective in ensuring that content providers receive profits consistent with competitive outcomes. So net neutrality regulation is, indeed, ineffective in this case. The reason though is because the consumer does, in fact, hold the ultimate monopoly and their chosen ISP in order to get their business will have incentives to create conditions such that any monopoly profits will be transferred to them. What that means is that content providers are stuffed even when there is net neutrality regulation.

The only ray of sunshine from that gloomy finding is that it is somewhat special. One can imagine restrictions on ISPs being able to pay consumers to connect by handing them content provider fees or ISPs that specialise in serving content providers or consumers but not both being able to get around these issues. However, those ideas quickly become more complex and I gave them short shrift in the paper.

In summary, net neutrality regulation is full of the non-obvious. I fear that many people have couched it on the basis that net neutrality regulation may have certain real effects but it is hard to find a model that really gives us that outcome. On the other hand, claims that price discrimination is necessary for appropriate investment incentives seems like an argument based on simple textbook economic notions rather than something a little more complex and actually represents how the Internet operates. Either way, my guess is this debate is in for a long ride.