Transcript of Berkman Report presentation to FCC – Dec 10 2009

[report and more details: here]

UNITED STATES OF AMERICA
FEDERAL COMMUNICATIONS COMMISSION
REVIEW AND DISCUSSION OF BROADBAND DEPLOYMENT RESEARCH
Washington, D.C.
Thursday, December 10, 2009

PARTICIPANTS:
Introduction of Workshop:
SCOTT WALLSTEN
OBI

Panel 1 – Berkman Report: ‘Next Generation Connectivity: A Review of Broadband Internet Transitions and Policy from Around the World’

YOCHAI BENKLER
Berkman Professor of Entrepreneurial Legal Studies, Harvard

Respondents:

HAROLD FELD
Legal Director, Public Knowledge

THOMAS HAZLETT
Professor of Law & Economics, Director of the Information Economy Project, George Mason University School of Law


P R O C E E D I N G S (1:15 p.m.)

MR. WALLSTEN: Thanks, everyone, for coming today to our workshop discussing broadband deployment.

And we’re starting with our first panel, which is on the Berkman Report, “Next Generation Connectivity: A Review of Broadband Internet Transitions and Policy from Around the World.”

I’m going to give very, very short introductions, because people on the panel need no introduction. But in case, for you, they do, read them in the agenda.

We’ll first have a presentation of the report itself for about 20 minutes, and then discussions by two discussants.

So, Professor Yochai Benkler will present the report. And he is the Berman Professor of Entrepreneurial Legal Studies at Harvard, and faculty co-director of the Berkman Center of Internet and Society.

So, Yochai, please.

MR. BENKLER: Thanks so much for the invitation to come and speak and present the report. The report, just to set the context, we were asked by the broadband — by the task force to take a look and see what’s been happening in other countries, and what, if anything, we can learn, either from things that went well or did not go well elsewhere.

What we did — or let me jump right into the key findings, since at least two things were controversial and got comments. So let me identify the two things that were identified after significant research.

The first thing, as I’m sure most of you know, there’s been a good bit of question about what is the condition of the United States. There had been a statement, for example, by the President about where we are and how good a performer we are. And there’s been criticism of that.

So the first thing among our key findings is that we went in and we spent a tremendous amount of time developing new data about the benchmarks, and about how we know what countries have done well and what countries have not done well, using various independent sources, in particular adding a lot on price and speed, not only penetration. And for that, the critical thing is to put facts before interpretation.

First of all, what are the facts? Then let’s talk about what the causal relations may or not be.

The second finding — there are other things, but this is the one that drew the attention so that’s what I’ll talk about now.

And, of course, we can talk about others later — is that open access policies, a bucket of policies, were important in the first generation transition, and that they’re widely regarding by policy-makers as part of the toolbox for their next generation transition. Critical things that we need to keep in mind. This is done in addition to and complementing facilities-based competition, not instead of. It’s not seen as a diametric opposition.

Second, the core idea is to find a policy framework that enables additional entrants.

In particular, we see entrepreneurial entrants alongside one or two large players with their own infrastructure. That’s the core target.

More things that I’ll focus on today are, first, that the background literature is a lot less determinate and a lot more supportive of the positive effects of open access and the absence of negative effects on investment than is widely thought, including by me four or five months ago before we went more deeply into this.

Second, that we spent a lot of time looking specifically at detailed country and firm case studies that support mixed models, not purely intermodal, but with a complex interaction. And, critically, this is not a solved problem. We need to continue study, we need to continue experimentation, we need to do this with an open mind.

So, first of all, the benchmarking.

Three outcomes measures: Quantity, quality and price — basic and important — quantity measured by penetration, quality by speed, and price is price, of course. Important to understand.

OECD data — much maligned data, it’s still the best most comprehensive, with the longest time and the most comparable countries, “comparable” in the sense of “like us,” along relevant dimensions. We also added independent sources, particularly to prices and speeds. We looked also at wireless.

The result is that what we are presenting is a diverse set of sources that are reasonably well correlated with each other, confirming that when you observe through this way and through that way, you find a similar set of findings.

So this is obviously the best-known model — the best-known measure, the penetration per 100, and we all know the United States is 15th. Throughout these, what you’ll see is the United States will be in a red or an orange so that we see its location relative to the other countries.

The critical point here that’s helpful is that this is a measure that has been available for many years. It comes through reporting of number of subscriptions. And essentially, in 2002, we were top quintile. And this basically shows you the top quintile countries and where the U.S. has been, from being in that to falling out of that.

Many others say, no, we should be looking at households rather than at per capita penetration, because that penalizes countries with large households.

What’s important to remember are two things. First of all, by the metrics as supplied by the national statistical agencies that are the formal numbers that are collected by the OECD, the U.S. is 14th rather than 15th. It doesn’t make a difference. In general, the two measures are highly correlated.

Second, it’s important to understand that households are a survey-based method. They get updated less often. They don’t include small business uses that buys consumer-grade broadband.

So they’re, in some senses, exactly the target we want. But on the other hand, they’re updated less regularly, not always at the same time with all of the other countries, so they have some costs.

So we prefer to use both, understanding that each one has advantages and limitations.

And, in the case of the United States, that they confirm each other very well. Even if theoretically, at 100 percent penetration there might be problems, these have been good predictors of each other more recently.

Then we move to speed. And here, let me focus on the things that we did that are newer, the actual measurements of which — in the report, we report on one, and since the report we’ve added another. So these are speed test measurements.

They measure at the end-user machine, downloading and running a machine. We looked at 41 million records from the last quarter of 2008, from the 30 OECD countries. We ran filters to try to clean as much data as possible.

Here, we see the United States is 11th in speeds. And, again, there was some pushback that speed-test is not a fantastic source, even though we worked with it quite carefully to try to clean it up. So since the report, since the comments, what we’ve done is also looked at Akamai. Now, it’s important to understand that Akamai measures and speed-test measures are done in completely different places in the network, and they’re done by completely different companies.

And the United States is 11th on Akamai measures, as well. And the two measures are very well correlated. So what we see is that you look from two — you ask two completely different people, using completely different method to observe the same fact, they converge quite closely, we get a higher level of confidence that this is not nonsense, this is probably the truth.

So we have penetration and quantity. We have speed, or a measure of quality. Now let’s look at prices.

In the initial draft we used two independent sources: OECD, which does its own pricing study, and a market analysis for Tele-Geography. Since then, again, people have come back with but I’ve seen a price here, like this, and price here like that. We added yet a third independent market source, Point Topic.

In total, we have 950 unique observations from the top four providers in each of the countries, 115 different companies of actual research done by three separate entities in three separate studies that are very well correlated with each other.

And what’s the picture they show? The picture they show is that on the very low speed — 768 kilobits up to 1.5 megabits, more or less, the U.S. is doing well. We have decent prices at the very low end of the offerings.

Once you start moving higher up, to more what would count today as contemporary acceptable speeds, and then once you move up from there to 8, 10, and above megabits, the U.S. is much less of a good performer. We go to the mid-teens when we’re looking at today’s and just beyond the horizon, tomorrow’s numbers. Once we look at the most predictive of the future, at the very high speeds, above 35 megabits, the picture is a lot less attractive. Essentially what we’re seeing is that in the United States, relative to other countries, very high speed is a luxury good.

So another way of looking at all of these data is to take all of the points, look at where the price is. And essentially what we’re seeing is exactly confirmed when you look at all three data sets together. At the very low speeds, the U.S. is doing fine. Once you move to today’s speeds, prices are expensive but middling of the pack. Once you get to really high speeds, it gets enormously expensive.

Another way of looking at the same data is to look at a subset of the offers — just the high, the next generation type offers as they’re available. And these ones — each one of these is a flag that marks exactly the best offer available in the three data sets from each of these four — from each of these companies.

And what we see — so, on the bottom right-hand corner, what you see is “Expensive and Slow,” on the top left-hand corner what you see is “Cheap and Fast.” That’s easy.

What you begin to see is that countries start to cluster. If we clear out the countries that are in the “Cheap and Fast” cluster, and keep the North American market, what we see is, actually, a nice clustering in several countries of “Cheap and Fast,” and an unattractive — if I may be so bold as to say — description of the price-speed tradeoffs in the U.S.

So that starts — so those are the facts. Now we can start thinking about what might have caused the effect.

So, obviously, there are things that everybody agrees are important. Urban concentration is important. Income is important.

Education is important. Poverty we think is important, and there’s evidence for that. People throw different things at it to ask the questions.

The basic question we need to ask ourselves is: Does that mean that policy is irrelevant? How much of the performance is talent, just your basic endowment? And how much can you do more, with sweat, to get better?

So this happens to be one way of doing it. This includes those factors that we just talked about. Others have done others.

Fairly regularly what you see is that the U.S. is performing as predicted. There are countries that are performing worse than predicted, for sure. But there are countries that are performing better than predicted. And the question is, what can we do, beyond relying on the natural endowment, to say what can actually work?

So let’s just take an example. One way in which we were able to use the speed data is to look at particular cities. So we looked at the top 55 cities. In the OECD we had data for 55 — but top 2 cities, capital and largest city in the OECD 30. We looked at top download speed, at average download and average upload. New York is not on the top 20 of the OECD, in terms of download speeds. It barely makes it to 13.

Washington is in the mid-30s. Here you’re not talking about urban density. Something is going on beyond this.

So what we did was to look at the detailed country-level case analysis of the political economy of regulation and at the level of what firms entered, when they entered, how they entered, how they responded to regulation. And we looked at about half of the OECD countries, at 14 case studies, to do this.

The key findings that are relevant today, I think, are that open-access or unbundling facilitated competitive entry in many countries, including — and this is important, the United States is not unique in this — including where facilities-based alternatives are available. What access entrants do there is they play an important catalytic role.

Facilities-based and access-based competition complement each other at the system level. In particular, we see entrepreneurial competitors that tend to enter through unbundling.

It gives them greater control, greater room for innovation, instead of just re-selling. We see this also, by the way, as an entry across borders from the incumbents of one country to become entrants in another country next door. We see the Nordic incumbents doing that to each other all the time. We see France Telecom doing it to some extent. We certainly see a Telecom Italia Tiscali doing it.

So, again, if we try to do this, here’s how the incumbents look on the same framework.

And this is just incumbent telcos.

Then what we see is that the cable entrants more or less cluster around, with here and there differences, more or less cluster around their incumbents. When we add unbundling entrants, what we see is that, first of all, obviously they’re all playing on price. They’re all up at the much lower prices. But, second, in the higher performing countries, they’re also pushing on speed.

And so you get a complementarity. And in these same countries, at least, in some of them you also get additional facility-based entrants.

Some of it, like in Sweden, is municipal utility type places. Other places it’s electric utilities. But you see this complementarity, instead of this idea of either-or.

So the question is: What’s the theory?

First of all, the best known — and, in some senses, I think, for most people, the most intuitive — is investment deterrence. This the Houseman ’98, “and later rates are going to be too low, you’re not going to get investment.”

On the pro side, there are a variety of theories. The most well known is “investment ladder” — start small, build the market, have cash flow. This we see in Norway. We see it a little bit, at least, in planned deployments in Illyad. And particularly we see it in shifts in various European countries from the lower investment incentive bit-stream to the higher investment intensive local loop and bundling.

Another is the question of, yes, maybe we’ll get delayed investment, but is it worthwhile in terms of long term welfare? Let’s say you get investment five years later, but then for 30 years you have more than 2 competitors, you have 3, 4, 5, instead of 1 or 2. Are the total consumer welfare effects sufficient to justify that?

That’s a question. It needs to be looked at.

There are arguments that greater competition increases uptake, and that increases cash flow, and then investment. There’s a brand new interesting paper from Johannes Bauer, basically coming up with a new Schumpeterian dynamic model. That is to say, large incumbents, badgered by small innovative entrants, that’s the market structure that you want. You want it to be not too concentrated so that you have some competitive discipline, and not too dissipated so there are no rents to invest.

To do that, you need to do a lot of dynamic fine tuning, and that’s the particular role. Which is to say, again, it’s not one policy implemented on a particular year in the U.S., it’s a family of approaches that look at this problem.

Then we have the question: What’s the existing evidence? And the answer here is — this, we were criticized in the comments and we’ve looked at 50 papers, and we’ll hand the task force the memo outlining — the draft memo outlining these. And we’ll include these in the final reports either today or tomorrow. I mean, not the final report, the memo.

We looked at 50 papers, 14 on penetration. We found six of them had open access had positive effects, two had negative effects, six had both positive negatives. Four of the ones that we looked at had either old data or, in one case, the methods were problematic.

There were 21 papers on investment. We used the (inaudible) as the basis. We added another — this is the brand new, most recent literature review that came out in telecoms policy.

Again, we analyzed them. We’ll offer all of this. It’s about a 50-page addition to the report. Two positive, one positive, one negative, two no-findings, one negative. Fifteen are either empirical or conceptual, or they have real flaws.

I’m not stacking the deck. They’re about equally divided between the two sides, but they’re problematic.

Another few things. Why is this happening? It’s happening because there are two many factors — demography and geography, local market conditions, regulatory decisions and strategic behavior.

Effective regulation, not just formal, is what we’re looking for. There are financial market effects. There are regional diversity, there time diffusion effects. Too few observable observations to account.

What is need is much more micro-level analysis, single country — even more locally.

Natural experiments.

There are a few papers in the last couple of years that have been trying to do it at that level. But otherwise, you have a large risk of overstating the result, of missing influential points, of masking anecdotes as evidence. It’s a real problem.

We also looked at 15 qualitative studies, looking at single-country or multi-country. Eight of these were positive, one negative, one positive and negative, and five had “no effect.”

Important to remember: 20 of the 35 quantitative papers were self-published. It doesn’t mean you don’t read them on their merits, but they don’t have that particular layer of constraint.

Sixteen of the 35 quantitative papers were industry sponsored. Two-thirds of the papers on investments were industry sponsored. It doesn’t mean you don’t look at the evidence. Of course you do. It just means you have to read it with a caution.

So what’s the evidence? Qualitative case studies that look in detail at what, in fact, happened in different countries and different markets to different countries, other regulators’ experience that says this was important, it helped us, and an econometric literature that is largely ambiguous.

So the last couple of minutes, what’s the transposition of open access to next-generation connectivity?

First, you see access rules and their transposition, their translation, being important to next- generation networks in other countries, in terms of their planning.

Critical, you see that the high cost of rolling out next-generation is pushing both countries and companies to find ways to share costs so that the price of entry into the market doesn’t have to be being able to have backhoes all the way to the home and the slow moving expensive elements. And muting, to some extent, the emphasis of the benefits of having redundant facilities as a hedge against regulatory failure if you had the monopoly. The tradeoff essentially is between market failure in a necessarily concentrated market versus risks of regulatory failure with a monopoly infrastructure.

There is a range of approaches currently, ranging from very aggressively regulatory to very cooperative that needs to be seen within this toolbox. There’s open-access transposed. People are looking whether there should be incumbent-only duties or symmetric duties, whether you look at the size of the cabinet in fiber-to-the-node implementations. How you do it exactly in transposition, where you require active or only passive and how it’s done, those need to be studied.

Functional separation, surprisingly enough, after the UK implemented it in late 2005, had very powerful results on the price competition side. Now it’s diffused, either in formal or in semi-cooperative form to six other countries in the OECD.

We’re seeing more voluntary and raised-eyebrow voluntary arrangements, like the KPN-Reggefiber arrangement in the Netherlands, like voluntary sharing and deployment, very interesting — SwissCom essentially saying I’ll roll four fibers in my regions, you roll four fibers in your regions, we’ll flip, and then we’ll compete in both of these markets, rather than having the government regulate us. Very interesting. And all of these, I think, need to be brought in.

There’s new openness to government investment. If there’s government money, there has to be open access. There are new models of market-viable public-private partnership that absorb some of the risk, but nonetheless are market-based.

To conclude, there is a lot of experimentation and experience going on around the world. It’s important to learn from it.

U.S. performance, overall, on all of these measures, is not the kind of level that we can say whatever others are doing, we’re clearly doing better. We’re doing okay. We’re not doing better.

What we need to understand, that the transition to next generations is a transformative moment. We’re looking at infrastructure that will be put in place for decades. What we do now will set the basic market structure of next- generation connectivity.

And the basic question we have to ask ourselves is: Is a market where the price of entry is backhoes all the way, trenches and PVC, are we going to get a competitive market? Is two enough? Or do we need to experiment with one of these ways to get the three or four or five — something more than that?

And there does seem to be, as we look at the things carefully, a role for well-designed policy.

Thank you.

MR. WALLSTEN: Thank you very much. So before we go on, just let me say after the discussants, we will take questions. And for people who are watching online, please do submit your questions because we’ll try to ask some of them.

So the first discussant is Tom Hazlett, who is professor of law and economics at George Mason, and is also director of the Information Economy Project there.

Tom.

MR. HAZLETT: I fear, Yochai, if folks did not have 100 megabit per second connection they may have missed some of that. That was a very impressive summary.

And I’ve been asked to comment on the Berkman Center study, and I’ve got just a few minutes. I’m happy to comment. This is not going to be a top-to-bottom review. And I, in fact, have conducted no such top-to-bottom review. But I am happy to share a few things here.

Could we — MR. WALLSTEN: Oh, yes. I’m sorry.

MR. HAZLETT: I guess this goes that way. So, I pick out three areas to discuss, because this is Washington, D.C., and you need three areas. So every talk has to have three.

You know that, that’s the rule.

So there are some economic and econometric issues that are right there in the paper. I did not read it with, I guess, the same conclusion jumping out at me that Yochai did, to put it mildly. And, in fact, I think that the econometric issues are quite severe, and I invite folks to read the record the FCC has opened up and maybe that will keep expanding as comments come through.

But, in fact, there are very large problems with spurious correlation from omitted variables and when, in fact, the data samples that are used are expanded the results vanish or reverse. And this is important. This is robustness checks that are absolutely standard.

And particularly in an important proceeding, with the important conclusions that are being reached, we don’t need a point estimate that ignores alternative methods of analyzing the data. We certainly need confidence intervals, and that includes a lot of commentary from other parties.

Now one thing that’s very important to note is that the paper does — as has been presented, I think correctly — talk about this relationship between broadband penetration and some sociological variables, but does not really take into account demand. There are, I think, income variables considered. But markets are structured such that demand is important. And when we’re looking at how much and how fast the broadband products are in the United States versus other countries, you do want to take into account the fact that, for example, we have a lot of — a relative lot of — a relatively high degree of dial-up service in the United States due to our unmetered local usage. We could eliminate the 8 million dial-up customers by metering local usage like most other countries in the sample and get somewhere close to 8 million new broadband subscribers. And that would, in fact, jump us up so our performance was better. But it turns out that the dial-up is a cheap and easy substitute for some part of the market, declining as it is.

We also have fairly unique circumstances in Canada and the United States who seem to do very poorly in these comparisons, because we have very successfully opened up competition to terrestrial broadcasters. And we have an enormous amount of programming, video programming, that goes to households through cable and satellite.

That, all else equal, would tend to lower demand for high-bandwidth products in the broadband space, one would think. And so that really has to be considered. I didn’t see that it was.

I would just touch on the fact that I think — I certainly agree when Yochai says that we want to have the facts go first and then the interpretation after. It is very distressing that within the econometric model, the data that’s used to run that is, in fact, adjusted such that data that has been long considered part of the broadband regulation landscape and, in fact, has been reported for years by the OECD and other sources used by the Berkman Center study, is just reversed. So you end up with South Korea becoming a country that, in fact, has unbundled since 1997; in fact, that’s been reported as beginning an unbundling regime in 2002. And the United States is counted as a country that is not unbundled, even though the United States as considered to be a very vigorous pursuer — in fact, an international leader — on the unbundling frontier early on.

So those are just a few little things that I think people should consider and get a wide view of the universe here. Because, in fact, the facts do matter and it is good to get those facts right.

Here’s one little experiment that I ran — well, here are couple of other things that I’m going to mention here, then — if this works.

Is this this way or that way? What do you think? This way? That way. It’s sleeping.

Okay. Well, we don’t have to do PowerPoint. I’m in a 12-step program, so I’ll be able to not do PowerPoint soon, but I’m still grasping at the button here.

There it is. There we go. So I wanted to mention just briefly this issue about the horse race. That’s something I do agree with the Berkman Center report on. And then I’m going to talk a little bit about the U.S. experience, and then I’ll be cut off and thrown out of the room.

So — grasping at electronic devices.

Okay, so as an experiment, a very simple experiment, a couple of months ago, I decided to find out where the U.S. ranked in international broadband penetration. And I went online to get the first sources for — the first reputable sources I could find to show me what the answer was.

And for theoretical reasons, I used the per 100 households. It’s certainly theoretically the correct measure to use. I don’t know why the OECD or Berman want to bother with the other.

There are a lot of other problems, as has just been mentioned, that are problematic to the data, but that doesn’t mean you can’t get that right.

Theoretically you want to use households as the measure.

So at any rate, I want to do per 100 households, and I wanted to rank 5 top countries, because the U.S. does have a peer group — countries like the UK, Germany, France, and England — well, I counted England twice — Japan.

And so in that group, how does the U.S. do?

Well, it turns out if you use data from Point Topic for the most recent quarter, and you use some CIA data on households — excuse me, on population, and UN data on households — average size of households, the United States comes out first in its peer group. And this was not adjusted for a multiple — you could accuse me of laziness. I took the first data from Point Topic, which is a reputable source used by Berkman and others. But the fact is that the countries are fairly well clustered and this is the way those data come out. So maybe in some of the correlations run by Berkman you don’t see a big difference. Maybe here it’s more of a difference.

I don’t know.

The point of this is not to say that the U.S. Leads the world, and it’s not to say the U.S. doesn’t have any policy issues that need very much to be adjusted by information gleaned from U.S. or international markets. But it does point out how volatile these rankings are and, in fact, how you have to be very careful about using any one ranking system and jumping to conclusions about that, particularly when you’re cutting out demand factors and other things mentioned.

Anyway, I wanted to just jump to the U.S. Experience here. And the Berkman Center study can say, well, you know, we were supposed to look elsewhere and so we didn’t look much at the U.S.

But, in fact, many markets are described and, indeed, the details are taken from what’s happening in those markets, and the U.S. market as characterized as not having much of an unbundling policy and dismissed. But the interesting thing about that is the fact that we’ve had a very rich mix of policies in the broadband and, in fact, also in narrowband, is an opportunity for social science. You get natural experiments when you change the regulatory structure.

And so here we have some wonderful observations of what happens in terms of market reactions, right here in the U.S. market. And they’re very important to take account of.

So if you start with this cable modem versus DSL competition, which typifies the broadband marketplace, the fact is that cable modem service has never been regulated through open access. But over time we have three different regimes for DSL. And, in fact, up until first quarter of 2003, there was an open-access regime, and there was activity in that market certainly.

And then in first quarter of 2003, the FCC, in a surprise vote, decided to end line-sharing. Now, line-sharing was really the price reduction that made that market attractive in terms of open access for entrants. When that was eliminated the growth in that market for third-party use of the existing phone company structure for DSL service to retail customers collapsed.

So after that, after that you go to this less- regulated environment for a couple of years.

And then on top of that, in the middle of 2005, the FCC decided to declare the DSL networks to be information services, and that put them in parity, in a deregulated environment, with the cable modem services.

So if we go to the videotape, we have three periods to look at. In the first period when, in fact, cable is regulated with open access and DSL is — excuse me, cable is not regulated in any of these periods. DSL is regulated with open access — you get about a 2-to-1 lead in subscribership all through that period. Here in ’99 through 2002 data the cable modem subscribership is in black, DSL in red.

And then, when you eliminate the line-sharing pricing, very favorable pricing structure for entry in the market — by the way, I should say that there was causality asserted by the FCC. It’s important — and this will be in my written comments — but the FCC did say specifically we’re not going to impose open access because we want these networks to get built. So there was causality asserted there in terms of cable doing well because of a lack of regulation.

Now, when you eliminate the line-sharing regulation on DSL, you can see that the red line after the vertical regime shift, the vertical line denoting the regime shift, starts kicking up very significantly from trend. And, indeed, cable modem subscribership continues to grow at about the same place — it actually picks up slightly.

But the real uptick is in DSL growth.

And that’s noticeable after the end of line-sharing.

Now, you go all the way through this period, you actually get, by the end of 2006, a 65 percent increase in trend for the DSL subscriberships. That means in year-end 2006, you had 10 million more — you had about 25 million DSL subscribers from a predicted level of 15 million DSL subscribers predicted by the extrapolation of the pre- deregulation trend. And that includes both deregulation periods. There actually is a bit of an incremental uptick for the second part, but I won’t show the data here. The data here encompass both deregulatory periods.

So this is actually quite interesting, that indeed there was a very positive reaction in terms of subscribership to the end of open access regulation. And it’s extremely important, I think, to take a look at this.

So this is just the same numbers. Now, the last part here is that to get fiber, the really — the next-generation network out there to the home — I don’t think there’s any real doubt that the investment will not, in this structure, in this institutional structure, be made without what the FCC did, very explicitly, in October of 2004, and that is preempt unbundling obligations for fiber networks. That’s when we got the kick-up in fiber to the home.

Now, the causality is also asserted by analysts in the market. This is a projection of fiber optic equipment sales rendered in late 2004 by Gartner Group that looks at the market and actually adjusted the new forecast outwards, explicitly on the grounds that the FCC had preempted regulation through open access rules of the fiber networks.

This is extremely important to take into account. It is data that goes right to the heart of what happens in this institutional environment, and should be included by the FCC, certainly, in its deliberations as to what policies are rational, reasonable and pro-consumer going forward.

MR. WALLSTEN: So, next we’ll have Harold Feld, who is legal director of Public Knowledge.

MR. FELD: I just — since I’m looking at a smaller screen here I just want to check,- that last slide was actually a projection about fiber-to-the-curb, not fiber-to-the-home, right?

“Fiber to the node,” I think that was —

MR. HAZLETT: (inaudible) fiber to the node (inaudible) DSL (inaudible).

MR. FELD: Right. Well, I just wanted to make sure we were comparing things.

The second to the last slide was, I thought, a measurement of fiber to the home projection, and the last slide, which was the investment causality slide was actually projecting a growth of — is that — MR. HAZLETT: The last one was FTTH.

MR. FELD: That one’s FTTH, right?

MR. HAZLETT: Yes.

MR. FELD: And then one after that, that was the Gartner project, says — I think it says, “Fiber to the curb node equipment.”

MR. HAZLETT: No, no. Well, that’s “equipment.”

MR. FELD: Right. Okay. I just — which, to some degree, serves as a segue to what I want to talk about, which is my name is Harold Feld. I’m not an economist. I’m not a technologist. Among lawyers, I try to pass for ones. So I will not try to dispute with either Tom or Yochai on the merits of the analysis. I do, however — would point out, one, Public Knowledge, we actually filed written comments in response to the Public Notice, where we thought, for the reasons stated therein, that this report was an extremely valuable piece of work; that we supported its conclusions; that these conclusions seem to be consistent with other information which we’ve cited; and that we thought the FCC had the legal authority to proceed to restore an unbundling regime under its existing authority without any further act of Congress, although we recognize that the D.C. Circuit may not agree.

That said, I would like to talk instead about how I think the FCC ought to be dealing with this report, and what role it should play in the record and recommendations, and what I would hope that the FCC will do and continue to do.

First, I would hope that we would all agree — whether you agree with the conclusions or not — that this was an enormous piece of work; that it adds tremendously to the available data; that they conducted meaningful comparisons, even if one thinks they are flawed or believes that there are other comparisons that could be drawn that would be more telling.

And, most importantly, it has provided a point of reference in what was previously an intractable problem. We will remember, for years we have argued not just about the OECD rankings, but every other set of rankings that has ever come out, and we were proceeding without any sort of common frame of reference or common understanding of what we were doing.

Love the report, hate the report, you know, think they used the wrong terms — at least we have a significant document which has undertaken to try to provide a framework in which some intelligent discussions around this issue can occur.

Obviously, there will be disagreements.

I think that’s to be expected and it’s very appropriate. No study can be comprehensive. The data are often subject to multiple interpretation.

The question of correlation and causation is oftentimes difficult and needs to be tested through multiple trials. And we should all recognize that.

I don’t think this report claims — and, certainly, Yochai in his summary made clear, he’s not claiming that this report is more than it is.

But at the same time, I think it’s important that we approach this in the spirit of appropriate inquiry, and not as, unfortunately, some — not Tom, but some have — with an attitude of either dismissiveness or, sadly, of ad hominem. I think that it is not at all helpful to cast aspersions on the motivations of researchers any more than I find it helpful, when I am arguing against incumbents, to make an argument solely based on the fact that incumbents obviously file stuff that is going to further their own economic interests.

You still have to respond to the merits.

And I would certainly hope that we are all capable to do that here. Further, that even if there is some aspect of the report in which the FCC conclude the analysis was not what it should be, or is not telling, or whatever, that certainly does not invalidate the rest of the report.

Each of the aspects that are raised here for comparison need to be evaluated on their merits and the evidence presented.

Which brings me to my next point which is, I mentioned earlier, the evidence available, especially at a public level, is imperfect. It is dynamic. It is changing. And, most of all, we are limited by the protection of proprietary data especially in this country. The FCC in particular has been extraordinary solicitous of the need to protect carrier data, and this raises problems for a number of us. But I would urge the FCC, to the extent that objections are raised, to compel data where necessary to test the underlying assumptions here.

You have that authority. This is too important to rely on unverified assertions when actual data can be obtained by the FCC through its legal processes and through appropriate inquiries.

Ideally, that information should be made as public as possible. Obviously that’s not going to be possible for all data. And given the amount of time that is available to test these I would reluctantly concede that erring on the side of, you know, setting up procedures by which proprietary data could be accessed under protective orders may be more necessary here simply to expedite the process.

But again, if we are to test the underlying validity of these things — to test the investment assumptions, to test the expenditures

— then we must have real data. In point of fact,
you must have real data and you should not shy away from using your power to compel, particularly where objections have been raised by carriers who have full possession of the data and who choose to submit data which is supportive of their arguments, but do not necessarily submit data which is less supportive of their arguments.

I would say that in evaluating this study and the evidence presented by all parties, what is important to remember always is that policy is tradeoffs. I think that the report does a good job of highlighting some of these. I think some of the objections that may be raised to the report and its potential impacts are very likely to have some validity, in terms of what impacts there might be on deployment. But it is important for us to keep in mind that there are tradeoffs, just as there have been tradeoffs to the adoption of the exiting regulatory regime.

And here is where I will disagree to some degree with Tom, and say that in evaluating our own experience with clear eyes, it is important to observe that we have expanded investment, certainly. There has been investment.

There has been greater adoption. But it has not been uniform. And it has not been properly measured and validated, which has led to the mandate to create a more accurate national broadband map.

We do have a real problem, in terms of evaluating how well our policy has worked, with the quality of the data that we have. But even with what we have, we can see certain very visible things.

Let me take the deployment of fiber to the home as an example. The single biggest investment in fiber to the home is Verizon’s FiOS.

And certainly Verizon gets credit for doing this.

They did it in the face of punishing stock devaluation by Wall Street analysts who were looking to the next quarter rather than to five years out. It’s only now that their investment is being validated. This has had positive effects in requiring the rollout of Doxis 3, pushing Cablevision to provide both WiFi coverage in its footprint to retain customers, to switch to an all-digital system. All of these are, indeed, positive aspects of competition in the largest, most dense markets in the United States.

Once you move out from the I-95 corridor or, on the West Coast, from the concentrated urban areas, the situation changes dramatically.

Verizon has, in fact, shed 10 million lines in order to avoid providing them with fiber and to ensure that it has an appropriate rate of return.

And it has stated that the last 10 million customers who are not getting FiOS service are not going to get the fiber to the home service, that they will continue to rely on DSL to service those customers.

So our reliance on intermodal rather than intermodal competition has had consequences and costs, particularly for those who live in urban, rural areas. Furthermore, the franchising system for cable and how that works likewise accentuates it, because cable, while they have been required to serve everybody in their franchise area, their franchise areas for the largest cable operators have included the largest and most dense population centers. They have left the provision of less-densely populated area to smaller providers who do not have the same resources, who are struggling. And therefore, in our rural areas, our policy of “let the market sort it out,” don’t require any sort of competitive requirements, has had a consequence.

Now, would that consequence have been changed by unbundling? I think that’s a very good question.

But the report does not claim that. The report claims — and this is the next most important point — that you use that strategy in synergy with others, and that the companies that have overcome the urban-rural gap and have been most successful about it have used, you know, direct government subsidy and direct government ownership of networks in some cases, and other strategies that have addressed this in a very clear fashion.

The final point I would make is that open access and competition, while very important and, we think, you know, convey a great deal of benefit, do not resolve all issues. There will still be a need for consumer protection regulations, disclosure regulations. We feel network neutrality or some other sort of, you know, common-carrier obligation somewhere in the system that ensures that carriers receive, you know, our — that the critical position that a carrier occupies with regard to its residential customer is noted and protected against as we have in the telephone network, and as we have to some degree with the cable network.

And therefore, in evaluating this report, it is important to recognize that even if, as I hope, the FCC adopts its pro-competitive recommendations with regard to unbundling and some of the other solutions here, it does not lose sight of its consumer protection mandate, either as part of the National Broadband Plan, or as part of its general oversight of the broadband sector.

MR. WALLSTEN: Thanks, Harold. We need to go to questions, but before we do, Yochai, do you want to take a few minutes to respond?

MR. BENKLER: Thank you. So I’ll just actually say a couple of words so that people have a chance to talk. On the question of the econometrics that’s drawn so much heat, I think what’s important to understand is — we said it several times in the report — we were not using the econometrics to prove a point, we were using it as a heuristic to explain some of the systematic problems with these cross-country studies. We focused on influential points, we focused on formal versus effective regulation.

What we do now with the more formal literature review that analyzes all of these problems is try to provide a more systematic so that we don’t confuse this question of are we saying this is the reason to do it as opposed to not.

There’s a real problem with econometrics more generally. There are real values to qualitative study. That’s what we need to focus.

If you look, for example, at the points Tom was making, so that exhibits two of the problems with the data, for example. So you look at the DSL curve, and that looks impressive. But then you note, you think about time- diffusion effects and S-curve of adoption, and you see that DSL gets adopted later than cable, how do you account for S-curve effects in terms of rate of when this is happening?

If you look at the question of strategic withholding of investment in the context of a regulatory negotiation, that’s part of the problem of how you account, right? That’s part of the problem, that’s what makes econometrics in this area so hard. There are so many variables that go into it that are very hard to characterize and harder yet to use with so few points of data. So this is not anything particular, this is something that’s systematic to the genre.

The second thing is the point — and, again, this is important — the point about Gartner.

So what’s the datum? He datum is that a particular firm changed its projection based on its assumptions on what companies would do given their response and investments to an understanding of a regulatory environment and how that would effect. How much of that is genuine straight incentives? How much of that is projection and reflection of a certain set of beliefs? That’s very hard to tell.

And the last point, people keep saying this about all of the massive growth in fiber as somehow being an evidence about the regulation.

It’s important to understand, fiber to the home in the U.S. — and I’m not going, Harold, where you went about Verizon not doing it.

Fiber to the home in the U.S. is Verizon, right? It’s 3.3 million homes as of the last relevant data that we have from Key One 2009.

It’s Verizon. If you’re asking regulatory environment or competitive environment, the question, if you’re genuinely trying to treat this as data is: Why not AT&T? Why not Qwest? Why not Bell Canada? Why not TELUS? Companies that are in the same regulatory environment or in a regulatory environment that is somewhat unbundling but very, very weak, but in the same cable telco competition, who haven’t chosen to make that particular choice.

It’s an interesting case study, it’s an interesting thing to point: What’s interesting about this firm that’s doing something that we care about and think is wonderful? It’s not data about fiber to the home in the U.S. versus other countries. It’s an interesting case study, if it’s not said as said as such, it’s an anecdote.

MR. WALLSTEN: All right. Let’s go to questions. Okay, Dave first, and then we’ll go there second.

DAVE BURSTEIN: (inaudible) the data we’re looking at is so bad and so unrelated to the (inaudible) you want to answer (inaudible).

And I’ve read Yochai’s study. It’s pretty clear that unbundling worked pretty nicely in some places in (inaudible), right. But we’re trying to figure out what to do in 2011.

And in 2011, my gut — and this is what I want to ask you about — is, really, all of these studies are essentially (inaudible) problem that we have (inaudible) — that once you get to a certain size and scale, so now it’s 65 percent (inaudible), it’s particularly unlikely that a new entrant (inaudible) customers (inaudible).

MR. BENKLER: Do you want — DAVE BURSTEIN: Yes, sure.

MR. BENKLER: So I’m going to do my best not to go beyond what I actually have done here rather than talking about other kinds of policies.

I think that one of the interesting entry strategies — assuming — so there are a couple of — first of all, I’m not sure that I buy the assumption that if the costs for an entrant are not including the back hoes, that it’s the same barrier — DAVE BURSTEIN: (inaudible) MR. BENKLER: The primary model that we’ve seen, particularly as there’s been more consolidation of entrants, is incumbents from one country moving into a neighboring country. So if you look — DAVE BURSTEIN: (inaudible) MR. BENKLER: Well, you said how do I project from my — DAVE BURSTEIN: (inaudible) sold out in France and Germany. (inaudible) and there’s no cross-border entry interest in Europe in the last three years.

MR. BENKLER: I would say that Telenor in Denmark and Sweden is very interesting and growing.

DAVE BURSTEIN: (inaudible) that one, I don’t know.

MR. BENKLER: I would say that still interesting and out, not clear what’s happening with FastWeb in Italy, but that’s hard, more generally, because of the regionalization to identify. They’ve already done a lot in the areas that they care about.

If you try to abstract one layer up, the basic question then becomes what sort of framework do you set up to allow the existing incumbents that are very regionalized and have the basis and the knowledge to enter each others’ markets through open access.

MR. FELD: (inaudible) I’d like to take a crack at that for a minute. Because it seems to me that you’re — part of the problem with the question, and I think it is a good question, that is to say, but there are two elements to this question. One is a question of facilities-based entry, and the other is a question of if you did functional and structural separation, would you get the resale non- facilities-based competition in which there is — DAVE BURSTEIN: (inaudible) MR. FELD: I’m sorry?

DAVE BURSTEIN: (inaudible) may be there. But what I’m trying to get to the question is: If — SPEAKER: Could you speak into a microphone, please?

DAVE BURSTEIN: (inaudible) — that it’s not likely to have new entries.

MR. FELD: But this is what I’m saying, is I’m saying you — the question is what do you mean by a “new entrant?” And what you — it seemed to be that what you were defining a new entrant as was somebody who comes in and overbuilds.

DAVE BURSTEIN: (inaudible) No. No, no, no.

It’s now more expensive to do customer acquisition than build — MR. FELD: Right. This is the — DAVE BURSTEIN: (inaudible) I need $500 million to start a new entry in the U.S., and everybody on Wall Street tells me nobody would ever fund it.

MR. FELD: Right.

DAVE BURSTEIN: That’s the issue.

MR. FELD: This is, indeed, part of the question. I mean, this is, to some degree, is switching (inaudible). We faced this is in the cable marketplace. To a certain degree we faced it when we did long-distance competition. And we started in a fairly mature market, where everybody subscribed to AT&T long distance. And there was an incredible acquisition cost. And it took many years for AT&T to drop down to below dominance level in long-distance residential.

The experience of these providers, it seems to me, is somewhat multifold and needs to be evaluated.

One is the question of is there value in having presence in the market, even if one company is still going to have dominant market share?

This was one of the economic models that Yochai referred to as one of the emerging ones. This is the neo-Schumpeterian, you know, little incumbents badgering the giant so that — take France, for example, where France Telecom’s ISP service still has a tremendous market share.

The question is, do you do things to try to minimize the switching cost to facilitate switch? Do you say that, well, you know, there aren’t a lot of little guys, but that’s making France Telecom behave in a particular way that is valuable to have the regime in place, even if they’re not, you know, gathering a lot of customers? Or do you give up on it?

I think the answer is that your question is a complicated one that is not answered easily by either the pre-2005 factors or the post-2005 factors.

Finally, I’d have to say that, you know, in this country we still have a significant problem with — what, we’ve got 60 percent adoption at the moment?

DAVE BURSTEIN: (inaudible) it’s somewhere around 65. It’s not going to go up because you have wireless substitution.

MR. FELD: Well, you know, again, there’s a parallel here to what was the cable market like in 1992? And, you know, we can ask the question of whether it was worthwhile to do things to create the possibility of DBS entry into the market, and whether that’s been successful or not.

But much as I think the story in cable is also very complex, you can’t deny that in 1992, we had 60, 65 percent cable penetration in the country and no DBS providers. And today we have, I don’t know what we’re given for the market penetration of total MVPDs, but we have DBS providers who have at least reasonable shares of the MVPD market. So a new entrant survived in that environment.

So at least in other sectors of the economy where there were similar issues at play, at least enough, there is at least something there to give some confidence that it is possible for new entrants to emerge.

DAVE BURSTEIN: (inaudible) MR. WALLSTEN: Wait, let’s get Tom. Tom wants — MR. FELD: That’s part of the question to answer.

MR. WALLSTEN: And then we’ll go to more questions.

MR. HAZLETT: Yes, I did want to give a quick answer.

Just — look, before you get the new entrant it always looks impossible. And I know you’ve done this for awhile.

In 1987, I was sitting at a conference at the University of California at Santa Barbara, and we were talking about the possibility of cable TV competition. And the coup de grace question, the rhetorical question was: And so I suppose you think phone companies could compete?

That was 22 years ago, okay? Today there is facilities-based competition in the United States for fixed- line phone service.

Well, you know, lower your expectations for a minute. Twenty-two years was that entire transition, basically. Where did it come from?

Existing networks went intermodal on each other.

And, of course, we look around — in fact, if you read the Berkman report, you are thinking about things like, you know, apartment ownership in densely populated areas. You know, this has worked pretty well in places like Japan and Korea, so that you can get more contracting and investment perhaps. You know, and the Swedish model, you know, sort of comes in there in a funny way, as well. The power companies have been very, very helpful in some areas with their own fiber networks and so forth.

And so you think about where that can come from, the obvious place it can come from is what you just said: Wireless substitution.

So we really should be, at all times — no matter what 2011’s going to look like or not look like, we should be figuring out the barriers to entry there that are currently in place, particularly in spectrum where it’s so obvious that there are constraints on the market because we control the flow of the input for, you know, radio spectrum rights through the FCC. That stuff should be out in the marketplace in a very abundant way so we can get that competition without the constraints.

And then you’ll be surprised, wherever it comes from. But it will come from somewhere.

MR. WALLSTEN: Okay. So the question there, and then we’ll take a second question before we do answers.

MS. HELLERSTEIN: Judith Hellerstein, Hellerstein Associates.

While I loved your report on the open-access requirements and on regulatory reform, I was left wondering why you only concerned yourself with OECD countries, and why not look at some, like Singapore, who have really been leading in broadband and have integrated broadband access in all levels of the government, from education, medicine, to mobile government to also have gotten a lot of other new ventures who spread out from the government’s involvement in broadband.

And I thought that we could see a lot of new best practices and new ways to adopt their methodology into our studies.

MR. WALLSTEN: And a question right there?

MR. CHAFFEE: I’ve actually been in telecom since 1982. So if you see the gray hairs, that’s where these have come from.

SPEAKER: (inaudible) MR. CHAFFEE: It would seem as though a U.S. Broadband plan could be simplified for our unserved population by opening up fiber that has already been installed by another carrier.

From what you know, would other nations open up such fiber for the sake of the broader population because of the savings in the overall cost and the access that would be provided by doing so?

MR. BENKLER: So, two answers, and I’ll try to be very brief so as to make sure that there’s time for others if they want to, as well as others here on the panel.

We looked at the OECD because these are, broadly speaking, the most comparable countries that have a period of data that’s collected in a serious way that you can look at. So in the country case studies, for example, we didn’t look at Iceland, because Iceland is just too different.

Singapore, in that regard, given the political economy, given the regulatory frame, given the density, I would be very wary about trying to draw conclusions from a country with the political and geographic characteristics of Singapore for a country like the U.S.

And the OECD, there are a good number of countries that are plausible — some closer, some less close — and interesting, and there’s good consistent longer-term, a little bit, data. And good public records, and relatively enough things to look at to get a sense of what’s going on.

So that’s really the reason of trying to do that. That was just a choice to try to get better quality data.

The question of fiber for unserved areas — mostly what you’re getting — so there are several answers. Let me try to be brief. It’s hard. As you can tell, perhaps, from the report or from the memo, I tend to go on. But there’s lots of details to look at, so it’s worth it, often.

First of all, most of the countries that have done serious planning have essentially a two-tier target. They’re looking to 2 megabits per person throughout the population — in the UK, for example — and then they say — and to some large portion of the population, we’d like something that really is tomorrow’s speeds. Some places are more ambitious because they come from places where they can get more. So it might be 10 megabits as opposed to 100. But that’s the first thing, is you can’t think of one policy for everything. You can. People do.

It’s not uncommon to say we want, really, two things. We want all the population to have really good connectivity. And then we want as much as is feasible of the population to have cutting-edge, next-generation, the thing that will drive the best applications and the economy. And in that regard, most of the thought in fiber is in that second bucket rather than immediately going into the first in the current area.

The second is, very interesting, the European Union just came out with new guidelines a couple of months ago — actually, as we were just finishing the report — on state aid. That is to say — and it’s important to remember, the EU, because they’re set up very much as a way of avoiding some countries that have a history of helping their companies against competitors from other countries, are very strict about state aid.

They don’t want countries to invest a lot, because they worry about benefitting; they nonetheless went there.

And they went there for unserved areas.

Very interesting definition, by the way, of “unserved areas.” “Unserved,” for purposes of high-speed, could be if you don’t have two facilities-based competitors offering 28 megabits per second, with real plans for deployment within the next 2, 3, 4 years or something like that. I can’t remember the exact number. That’s a very different view of what we think of “unserved.”

The second thing is, they’ve basically said if you do have state aid for those areas, it’s got to be open access.

And so in that regard, those are the most relevant things for us to learn and take from here, is what’s the conception of “unserved,” in regard to what kind of competitive market counts as unserved. And how widely accepted open access is that it’s considered to be within the state aid.

MR. WALLSTEN: Okay. Well, we’re completely — we’re into negative time, unfortunately, even though we started late.

So please join me in thanking the panelists for speaking, and for the Berkman Center report.

Thank you. (Recess)