Friday, 23 August 2013

The price of your broadband (5)

I know I've been banging on a bit about broadband policy, but bear with me: we're talking about one of your bigger household bills, not to mention what fibre-based broadband might or might not mean for you (or the country at large), plus there's the little matter of the government spending $1.35 billion of your money and mine to help pay for this new fibre network.

It's also something of a dramatic business story. The government and its fibre-laying partner companies (mostly Chorus, but also others) have basically taken a giant Field of Dreams style punt: build it, and they will come. Lay the fibre, and people and businesses will flock to it for superfast, high bandwidth usages of all kinds. And the payoffs - while highly uncertain - could be huge. One often quoted study from Alcatel-Lucent found that there would be benefits of close to $33 billion over 20 years to business, education, the health sector, and the dairy industry: the annual benefit would be about 1% of GDP. And there's also a sizeable shorter term boost to GDP from the engineering works to lay the fibre.

I don't have any problem with the decision to give this project the green light. You could call it a big gamble - nobody knows for sure what people or businesses will use it for, or even whether they will give it a go at the price it will cost - but you could say that of many large scale commercial investments that work out just fine in the end. You could even argue that the biggest rewards (for both the investors and their customers) come from precisely this sort of high stakes, visionary leap into the unknown. Personally, I'm inclined to the view that fibre-based broadband is on the way to becoming a necessary part of any developed country's infrastructure, if only to stay competitive with the other countries that have it. But who really knows? I'm no more of a successful futurist guru than anyone else involved in this project.

The big point is that it is a lot of money, riding on a great deal of uncertainty about its potential attraction, take-up, use and payoff.

So, recapping the story thus far: the government held competitive auctions for the right to lay fibre in different parts of the country, and out of that process came a wholesale price for what Internet Service Providers (ISPs) will have to pay to use the new fibre facilities: $37.50 per customer per month to start with, rising by $1 a year to $42.50 a month.

In the meantime, you've still got the copper-based broadband network. And that's where things have got messy. As I posted previously, last December the Commerce Commission provisionally reset the prices that your Internet Service Provider will pay to get access to the copper network. The old copper price had been $44.98, comfortably above the $37.50 fibre price, but the new copper price is $32.45 - some $5 a month cheaper than fibre.

You'd think there would be a chorus of approval from end users, welcoming lower copper-based prices. Well, there's been some of that, but as is always the case, the reaction of the few bodies adversely affected tends to be a good deal louder than the reaction of the many diffuse, small-scale beneficiaries (you and me), who tend not to get involved in these esoteric subjects. It's sort of similar to the fuss that happens with trade liberalisation - the four million people who benefit from cheaper T-shirts don't get heard from, but the 500 people employed in doomed domestic T-shirt factories most certainly do.

In any event, the government (among others) has been fuming.

Its biggest concern is that the price advantage that copper will have at these proposed new prices will mean fibre take-up will be modest and slow, undermining the appeal of fibre and doing a lot of damage to the economics of the project. Things like fibre have network effects, for example, where the thing becomes progressively more attractive the more people use it. That won't happen if people don't move in numbers to fibre to start the ball rolling.

And it's right to be concerned. Local research has shown that both consumers and businesses are quite price sensitive when it comes to broadband. In last year's High speed broadband services demand side study the Commerce Commission found (see p36) that 40% of households would pay $0 to $5 more a month for high speed broadband, and another 37% would pay between $5 and $10 a month more: 77% of the population, in short, have a very modest appetite for paying much more for the likes of fibre . Businesses were the same: the study found (p38) that "the cost of high speed broadband services is likely to be a significant issue for many SMEs".

The Commission said at the time that these results were more indicative than anything, but they do agree with what little overseas research there is on broadband price sensitivity. One study by three Japanese researchers of migration from copper to fibre across the OECD in 2000-08 found (p18) that "the price elasticity of FTTx [i.e. fibre] shows -6.39 (p<0.01) which is elastic, and the cross price elasticity with regard to DSL [i.e. copper] 1.19 (p<0.10) which is also elastic". In other words, consumers are very sensitive to the price of fibre and to the difference between fibre and copper. Granted, this was a few years back, and at an earlier stage of adoption of broadband, and things might have changed since then, and New Zealand might be different, but all the evidence is nonetheless pointing to the fibre price, and the relative copper-fibre price, being a hot button for both families and businesses.

The government's response to all of this has essentially been an ultimatum to the Commission: raise those prices, or we'll do it for you. It's proposed three options, shown in the diagram below.

On the left is the status quo - the prices the Commission has proposed ($23.52 for the copper lines 'UCLL' service, plus $8.93 for the electronic gear or 'UBA' service, giving a total of $32.45).
Next is the government's Option 1 - the Commerce Commission is directed to set the total copper price somewhere in the $37.50 to $42.50 range (in other words, at the fibre price). To make things simple, I've assumed the Commission (if this option eventuates) will pick $37.50. In that case the UBA part of the total cost will stay the same, meaning that the UCLL price (as a residual) goes up from $23.52 to $28.57.

Option 3, on the right, is exactly the same - except that this time the government simply decides the pricing, and cuts the Commission out of the loop. Not a good look for the Commission, but not a good look for the government, either.

Option 2 still has the same total price ($37.50), again set by the government, but this time the UBA price is the residual: the UCLL price is set at the Commission's $23.52, so the residual is a UBA price of $13.98.

Under all scenarios, your ISP has to pay $5.05 a month more than it would have otherwise for its wholesale copper services, and that's the cost it will be passing on to you. Your broadband will not go down in price. Nor will you be getting anything by way of extra service or quality from this above-cost charge to you*.

I can't say I like this very much as a policy response. It's as if the government ran an airline with a fleet of older, slower, noisier, fuel-inefficient but cheap and cheerful planes, and then decides to spend up large on a fleet of new, fast, quiet, efficient but more expensive planes. All well and good, but why should airfares on the still serviceable old planes be raised to the level of fares on the new ones? Other than in the commercial interest of the new fleet owner?

The government says two things. One, if we don't do something like this, the new planes won't arrive for years, if at all. And two, it's had second thoughts about whether the Commission's copper prices are in fact right, and whether they are genuinely lower than the fibre price.

The government has come to think that the whole basis of the Commission's copper price setting is, essentially, pointless. Why, the government asks, is the Commission wasting everyone's time going through these benchmarking exercises (if you haven't followed these posts thus far, then benchmarking is covered here and here), when we actually know what the real costs of running out a network are? And we know this (the government says) because we've just had a competitive bidding process where fibre-laying companies told us what the efficient costs of running out a network must be. Plus who'd bother finding out the cost of running out a copper network these days, when any modern cable-layer will be laying fibre, not copper?

This, I think, is rather disingenuous of the government. For a start, the Commission is applying the pricing principles that the government required. And the government itself says that "The pricing principles (the use of forward looking costs for network replacement) remain sound, and reflect international best practice" (p14 of the government's discussion paper).

Plus there's the little matter of one price being all about copper networks and one price being all about fibre networks. Perhaps the government's right, and they might well be similar: one of the odder things, in this sophisticated 21st century, is how much of the costs of this advanced technology come down to the very mundane matters of digging and trenching, which might well be common to both copper and fibre. On the other hand, I've been a participant in several of the Commission's benchmarking exercises, and I've come away with the view that they do, in fact, get approximately close enough to the 'real' cost to be reliable. At this point, I'm somewhat inclined to endorse how the Commission goes about this exercise, though I'd accept that this time round, it is a somewhat more questionable process than usual. The Commission was able to find only two overseas jurisdictions, Denmark and Sweden, to use as sighting shots for the likely cost here, which is a much smaller amount of data than it typically has.

For me, I'd like the final outcome to be two networks, both offering cost-based prices, each acting as a competitive discipline on the other, and offering greater consumer choice. There are people who'll argue that this is productively inefficient: I'm not sure that it is, or, if it is, that the productive inefficiencies aren't outweighed by the competition benefits.

We'll see how it plays out: I'm not especially hopeful at this point that we will see the best result.
At a personal level, I'm the living embodiment of that research that says, households and businesses are going to be very price sensitive to the new fibre products. I looked up Telecom's fibre offerings: for the full-speed (up to 100Mbps downstream) fibre service (including a home calling package), the price is $125 a month. There is a crippleware version, offering up to 30Mbps, at $95 a month.
I'm not impressed.

And I'm especially not impressed by the data caps. Yes, folks: in this wonderful new world of massive fibre data-carrying capacity, the data caps are still there. The $125 full-speed version has an 80GB data cap. If I only got half the maximum download speed, 50Mbps, at non-stop use I'd run through that monthly data cap in 27.3 minutes.

What a deal.

*Original text amended to make it clear that the Commission's lower prices would not have kicked in till next year


  1. Sorry to be pedantic, but the proposed price wasnt to kick in till Dec 2014 (?) so rather than ISPs putting up the price, it is more a case of not dropping the price (other than maybe Flip who were banking on lower costs and have subsequently put their prices back up).

    On the fibre build as a MEA (modern equivalent assets) proxy for the forward looking cost of copper, the counter argument is that the cost of UFB is higher due to it being a brown fields deployment rather than green fields.

  2. Thanks again - I'll fix that. I appreciate the feedback.
    On the relative roll-out costs, my personal view is that it's hard to take any a priori view on general principles: it's likely a purely empirical engineering issue that could go either way. Some of the cost relationships you might expect to operate in telco deployment eg higher density leads to lower cost) don't seem to show up in the data. My default position is that benchmarking is as practical a way as any to start with. Let's hope no-one has to go the full TSLRIC route: I've already spent some days of my life looking at pictures of mole ploughs, and listening to experts who had devised indices of the degree of difficulty of trenching in different conditions. It was interesting in its own way, but I don't think I'd wish it on anyone else...

  3. Thanks for the series, its really good to see the whole picture laid out so well.