Virulent Word of Mouse

September 21, 2009

Completing an upgrade to the broadband network

Proponents of a broadband build out should be aware that some subtle economics are missing from their case. The following question needs an answer: What benefit does the spread of broadband to new areas bring to all the other areas in the country that already have broadband?

That type of question does not normally come up in analysis today. For example, I was just reading a very good USDA report about extending broadband to rural areas. It is very thorough and reviews some very interesting analysis (especially in the latter chapters). Broadly construed, the analysis focuses on the benefits broadband brings to areas that did not have such connectivity.

Do not misinterpret me. There is nothing wrong with that type of focus. However, as I have said elsewhere, it leads to one conclusion. Broadband will help a local economy, but its arrival will not – indeed, cannot – perform economic miracles. It is an exaggeration to say otherwise.

More to the point, standard economic analysis does not provide an overwhelming justification for building broadband anywhere in the country, since the expense is enormous. There is an economic kick to installing broadband where there previously had been none, to be sure, but modest gains do not justify large expenses. Limits are limits.

This should give proponents for an expansive policy some pause. Either non-economic issues need to come to the fore or additional economic arguments needs to be formulated. Today I want to consider the latter.

An expansive policy needs to address this challenging question: What difference does it make to a community outside Denver if a community in Appalachians has broadband or not?

Not much, it would seem. But, actually, I am not so sure.

This is actually an open economic question, and that is the point of this post. If the proponents for an expansive policy find a persuasive answer to that question, then they will have a more persuasive case.

Economics of hooking up new users

This topic often leads to confusion. It goes wrong due to misunderstandings about “network effects.”  Network effects arise when the value of the network increases for all users when additional users participate on the network.

Most observers agree that the Internet is shot through with network effects.

Doesn’t the presence of network effects suggest more users – even from rural areas – will generate additional value for existing users? Yes, but not much. By conventional reasoning the additional gains from building a bit of rural broadband are not big.

Sorry, but the numbers do not add up.

Network effects come in two flavors, direct and indirect. In a direct effect the value of being on a network depends purely on the number of participants. Communications networks have this property, and some parts of broadband networks do too – such as instant messaging.

Unfortunately, the numbers are not large, at least on the surface. Think of it this way: with over 70 million households already using broadband in the US, another 7 million will not bring much (Indeed, as I have said elsewhere, most reasonable estimates say a national broadband buildout will yield far less than 7 million additional new households, but let’s try to tackle the best case…).

My point is this: The bulk of the direct network effect has already been realized. The marginal gain from additional users cannot be large.

There are different issues when we discuss indirect effects, which arise when the size of a network generates additional economic activity. Indeed, on the surface, that is not apparent. After, all, the Internet is also full of examples of indirect effects – e.g., once there were enough broadband users, companies like You Tube showed up with applications to take advantage of those users.

Here is the first problem: A lot of this has already happened. Will adding a few additional users bring more?

More to the point, it is hard to argue that we are below some sort of threshold size in the United States, which is the second largest group of Internet users in the world after China….Do you think increasing the size of network by 5% or even 10% will make much additional difference to the entry decision of another major web supplier? As my kids say, I don’t think so.

In addition, as I stressed elsewhere, a lot of what passes for growth from an indirect network effect – more Internet advertising, or more UPS shipping, for example – boils down to modest economic growth for a local economy. It would be nice to have a technology elixir for local economic growth, but most of the time economic growth is hard to achieve. So it goes.

That would seem to end the analysis on a somewhat gloomy note, but it does not. One more thing needs to be introduced. Building rural broadband is about more than just adding more users. This is also an upgrade to an existing network. In other words, bringing broadband to many areas will also lead many dial-up users to convert to broadband.

It does not sound like a big change in framing, but it opens up several new questions about the economic impact of rural broadband. I would sure like to see the proponents for an expensive policy explore further these open questions.

The right kind of network effects

The open questions are easier to understand with the aid of a historical example.

Let’s start with a very old example, from the era of railroads. Back in the 1880s the old iron network received an upgrade to steel. Purely local economics motivated the upgrade at first, because steel had better durability.  Long story short, many localities did an upgrade simply to save on replacement costs.

But something interesting happened after most — namely, more than 95% — of the national rail network converted to steel. Another property of steel became more relevant – it could hold greater weight. That generated invention of larger freight cars and bigger train engines. And that latter technical change initiated an era of enormous productivity advance.

That sounds like an indirect network effect, to be sure, but there is a twist. The effect operated only after most of the network upgraded. The network effect does not kick in at low levels of adoption. It only kicks in after the network has achieved nearly ubiquitous adoption. There is a discontinuity in economic behavior after a large fraction of users convert/upgrade. Call this a upgrade effect.

That is my point. Upgrade effects are an interesting special case. New technologies could take advantage of the entire network, even in the less frequently traversed parts of the network, but it required the entire network to be upgraded. Something like 80% or 90% of the track would not have been enough.

Lest that first example sounds too ancient, other recent examples have a similar economic logic to it. For example, converting the entire US telephone network to digital switches in the early 1980s enabled a whole range of system-wide innovations. Those innovations could not diffuse until virtually the entire network converted.

Perhaps it is less apparent, but a similar economic logic applies to the recent retirement to analog television spectrum. Turning that spectrum into digital applications is just like an upgrade, but those upgrades could not occur until all users retired the analog spectrum. We are waiting to see the benefits, but those should arrive soon.

Those examples frame the question for broadband. By comparison, will the upgrade to broadband generate a range of new innovation that otherwise would not occur without near ubiquitous broadband? To say it another way, was the presence of a significant number of dial-up users holding back invention for the broadband network?

Open questions

Now we can reframe the open question: What further economic gains require near ubiquitous use of broadband or, what is practically the same thing, the nearly complete retirement of dial-up users who use small bandwidth?

I am willing to believe there are sensible technical answers to that question. For example, next generation web technology may work only with broadband bandwidths, but not dial-up. Similarly, certain kinds of interactive video applications also require near ubiquity of broadband. I am even willing to believe something deep inside Internet standard stack, such as IPv6, may not blossom without near ubiquity of broadband in the country. I am just waiting for somebody who knows better to educate me.

That said, this is not straightforward.

I am a bit more skeptical that there is an economic link from ubiquitous broadband and the entry of new businesses. It looks to me as if we are already well above any scale or size levels – in terms of the number of users – necessary to induce entry of new applications. It looks like quite a lot has already happened. And it does not appear that the presence of a significant number of dial-up networks hindered the deployment of some frontier business applications in the recent past, such as massively on-line gaming.

The Internet is an extraordinarily complicated network, so there is room for quite a lot out there, and far more than I know about. Perhaps some medical innovations need ubiquity. Or maybe some video communications standards. Or maybe something related to Internet security.

This is the debate we ought to be having. What more is possible?


  1. I think broadband has a unique property that some of your other networks are missing: it’s a very flexibly routed network (meaning you can go from New York to LA a lot of different ways with relatively low marginal costs). Railroads required 95% conversion in order to make it feasible to build bigger locomotives because odds are that on that network you’d run into an iron track even on a relatively major route unless you had very high conversion rates. Similar would be true for analog to digital telcom switching.

    On the internet, you don’t really have this problem, so adding more customers is really just adding more eyeballs, wallets or citizens (depending on what kind of app you’re trying to sell). So the conversion of low or no speed broadband for the remaining population is probably going to directly drive additional revenue only in proportion to the actual number of end points added.

    But there is a big social cost to having those citizens unconnected to all the potentially great high-speed services that will be coming out over the next 20 years. I’m not sure how to measure that externality (of having a class of citizens left out), but I think it’s a pressing issue and it should be addressed, for reasons of equity in US society if for no other (and I suspect there are other good economic reasons, but as you point out it’s hard to prove them).

    Thanks for a great read – I’ll be curious what you think of these comments if you have a chance to respond.

    Comment by Steve Midgley — September 21, 2009 @ 12:06 pm | Reply

  2. I don’t know If I said it already but …Excellent site, keep up the good work. I read a lot of blogs on a daily basis and for the most part, people lack substance but, I just wanted to make a quick comment to say I’m glad I found your blog. Thanks, 🙂

    A definite great read..Jim Bean

    Comment by JimmyBean — October 1, 2009 @ 3:06 am | Reply

RSS feed for comments on this post.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at

%d bloggers like this: