Sunday, 8 September 2013

The latest on the price of your broadband

Once upon a time, you were looking at the prospect of cheaper copper-based broadband. And then you had the prospect snatched away. By your own government, no less, though the government says it's in your own best interest to have that nasty cheap copper price taken away.


Never mind: quick recap of how we got where we are, and then, more importantly, the latest twists and turns in the plot (there've been quite a few).

The Commerce Commission was proposing new, lower, prices for your internet service provider (ISP) to have access to the copper telco infrastructure. The price your provider will pay, the Commission said, is going to drop from $44.98 a month to $32.45 a month. That's because the component to pay for hiring the electronic equipment that handles the data traffic ($21.46) is going to drop to $8.93.

Like hell it is, was the government's response (which I wrote about here). Make the copper price at least $37.50 (where it won't be cheaper than the new fibre-based UFB infrastructure we're spending billions on), the government said, or else we'll set it there ourselves.

So here's what's happened in the past few weeks.

First, the Commission came out on August 13 with an update paper on how it might set that equipment price ('UBA' in the jargon). You can read the media release or, for true policy tragics, you can read the full paper and some supplementary expert economic advice that the Commission had got on the topic from Boston University's Ingo Vogelsang.

There were basically two key points.

First, the Commission has revised its proposed UBA price from its original $8.93 a month to $9.91 a month. In both cases, the proposed price had come from averaging the equivalent Danish and Swedish prices for a UBA service (these were the only countries found to have costed the thing the way we would, if we were to do a full cost-modelling exercise). The revision came about for all sorts of reasons - updated local costs at the Danish and Swedish ends, updated exchange rates, fixing a mistake in the Danish data, and pricing an average-data-speed service rather than a low-data-speed one - but the upshot is that the proposed price difference between cheaper copper and dearer fibre has narrowed by nearly a dollar. While this adventitiously defuses some of the government's wrath, it still leaves copper (now $33.43 rather than $32.45) cheaper than fibre ($37.50).

Second, the Commission said it was thinking of picking a higher price than the simple average of the Danish and Swedish prices.


Sweden is arguably more like us than Denmark is, so the Swedish price ($10.92) could well be a closer sighting shot on our likely UBA costs than the Danish price ($8.91).

Plus (as has happened in previous determinations, including ones I sat in on) it may be better from a dynamic efficiency point of view to err on the side of a higher price than a lower one. Yes, there would be a higher cost, and arguably for quite a long time for quite a lot of people, from setting the copper price on the higher side, but on the other hand  "this cost needs to be weighed against the benefits of accelerated migration [onto the fibre network] in bringing forward services dependant on UFB take-up. Thus over time we would expect the value of the additional capabilities of fibre to grow and benefits to end-users to accrue, offsetting the welfare costs of accelerated migration" (para 141 of the paper).

But in leaning towards a higher UBA price for these sorts of reasons (and this is where it gets hairier) the Commission also floated the possibility that its final word on the UBA price might be even higher than the Swedish price.

Come again?

The rationale, as I read it, is that there is (hypothetically) a whole range of UBA cost estimates out there, but we've only been able to corral two of them. And if there's this range, then arguably it extends above the Swedish price. For example (the Commission says), Denmark is $9 (roughly). Sweden is $11 (roughly). These are $2 apart. It's arguable, then, that there could be another price which would also be $2 away from the Swedish price, namely $13. Now you've got a range of $9 to $13. And if you want to pitch your UBA price towards the higher end of that hypothetical range, say at the 75% percentile, it could well come out above the Swedish price.

Well now. You've only got two data points, and yet you're somehow going to arrive not only at a range around and beyond them, but also at a probability distribution for that range. As Captain Jack Sparrow might have put it, that's interesting. That's very interesting.

In any event, the Commission said to everyone involved, get back to us by close of play September 3, and tell us what you think of all of this.

People did. I've read their submissions (they're here, if you 'expand' the documents under 'Submissions on UBA price review update paper'). Here's what I make of them. I've kept to the bigger pricing issues that people raised.

Chorus was somewhat sympathetic to the various tight corners the Commission finds itself in - trying to figure out where the best interests of consumers lie when you have a legacy copper network and a new fibre network coexisting and interacting with each other in complex ways, and having not a lot of guidance from overseas data to help you with the job - and liked that the Commission appeared to be leaning towards making fibre more of a goer quicker (as in that bit I quoted above).

But it also bowled the Commission's idea of trying to create a whole frequency distribution on the basis of two data points.You might well have guessed, intuitively, that the exercise was a non-starter, and you'd have been right. Chorus's statistics wizards - Neil Diamond of Esquant Statistical Consulting and Daniel Young from Competition Economists' Group (CEG) - concluded that "The two observations sourced by the Commission do not provide any basis to inform the shape of the underlying distribution of forward-looking UBA costs.  Nor do we understand there to be any other basis upon which to speculate on the shape of this distribution.  On this basis, we do not believe that a statistically robust estimate of the 75th percentile of this distribution can be estimated".

Chorus restated that its preferred way of getting past the two data points problem was to use more overseas benchmarks, even if they don't meet your comparability criteria straight out of the box, and adjust them so that they do.

Internet NZ, TUANZ (the telecommunication users' association) and Consumer NZ put in a joint submission (assisted by Covec). I'll just cite the Covec paper here. It too skittled the Commission's proposal to bootstrap an entire distribution from two points: it noted (and you might have spotted this, too) that the Commission's imagining a $13 price, $2 up from the Swedish one, could just as equally apply to imagining a price $2 lower than Denmark's, at $7. Covec didn't add, but I'm going to, that there's no end to this: if you've now got possible prices of $7 and $13, $6 apart, you can as easily imagine new prices $6 away from those, too, at $1 and $19. It's getting rather random and arbitrary.

Covec was right (I reckon) to conclude (para 11 of Covec's 'UBA Pricing Issues' submission) that "If the Commission’s proposed methodology becomes established as standard benchmarking practice, it will become very difficult for access seekers, access providers, and infrastructure investors to determine the likely outcome of any future benchmarking exercise" and "would increase regulatory risk and reduce incentives to invest".

Covec weren't impressed by the Commission's argument (quoted above) that a higher UBA price would see us getting the benefits of fibre faster. Like Chorus, they appreciate (para 64) that there are a lot of moving parts here with the copper and fibre (and wireless) networks: "changing the UBA price has very complex effects on outcomes in broadband markets. The immediate effects on consumers and Chorus’s revenues are clear, but the effects on incentives to invest in unbundling, incentives to invest further in UFB, incentives of consumers to migrate to fibre, and so on, are indirect and the magnitude of these effects is difficult to estimate", but for them (para 111) one of the bigger parts is the upfront cost to users of the copper: "probably the largest effect is the fact that at least 70% of end-users will pay the higher copper price and either never receive any benefit [from fibre] or only receive benefits in the distant future (after 2020)".

I'm a fan of letting workably competitive markets do their thing, as you'll have appreciated by now, so I quite liked two other points the Covec paper made.

One was this (para 85): "In summary, competition between copper- fibre- and LTE-based [wireless] broadband services will be lessened by the UBA price. The higher is the UBA price, the less incentive fibre and LTE service providers will have to offer consumers better value, and the weaker will be competition between them".

And the other was this (para 105): "The emergence of a new technology in a competitive market therefore never makes consumers worse off, not even temporarily. In contrast, the Commission is proposing to do exactly that – to make consumers of existing broadband technology worse off in
order to support a new technology that does not yet appear to offer significant benefits to most consumers relative to the old one. That is not how competitive markets work".

Telecom (assisted by NERA) said that you'd have to be very careful about overweighting Sweden on "most like us" grounds. If you gave 100% weighting to Sweden, you've effectively thrown away Denmark completely, and junked one of the only two data points you had (I agree). And like everyone else it didn't warm to the Commission's attempt to conjure up a possible distribution of overseas UBA prices starting with just two observations: "It is not entirely clear to us what the Commission’s justification is for shifting the distribution beyond the range of the benchmark set...In any case, it is not clear how the Commission would determine, in a robust manner, the amount by which to shift the distribution, or indeed what the distribution might look like" (p4 of their submission).

And, like Covec, they weren't convinced by the "fibre faster" benefits: "The Commission and Professor Vogelsang propose that an increment could be added to the UBA price to take advantage of network effects/externalities associated with UFB...This might be correct in concept, although there is a question about how important it is for a small country like New Zealand...But even if there is a positive externality, we question whether increasing the UBA price is the appropriate to way to internalise it. Raising the UBA price would decrease the differential between the UBA and UFB prices, but the more direct and orthodox approach would be to subsidise UFB – which is of course what is already happening.  It is not clear that any additional intervention on top of this is warranted" (p8).

Vodafone (assisted by Network Strategies) also picked up on the Commission's "imagine a distribution" option. "Should the Commission proceed with this option, a price point will be selected that is outside the benchmark range. This approach is entirely contrary to statistical theory – such an action would be purely subjective in nature and made without any recourse to real-life evidence" (p1 of Network Strategies' paper).

And they also noted (p17) the difficulties the Commission has in trying to assess the overall impact of its decision when there are multiple things happening at once and often pulling in different directions: "Professor Vogelsang has quite rightly characterised the issues facing the Commission in this pricing review as a series of trade-offs. As illustrated above, there are a number of potential static and dynamic effects that may occur in the event of over- or underestimation of the UBA price point.
Unfortunately there is no evidence available that would enable us to assign either probabilities to all of the different possible outcomes, nor to estimate the magnitude of the gains / losses". From which they conclude that the Commission can't have a reliable enough base of information to justify overbaking or underbaking the UBA price.

There's a lot of quality analysis in these submissions for the Commission to absorb (and quickly - it's meant to have its final decision out by October 31). It's got its work cut out for it, not only because of the inherent difficulty of the job but also because its proposed Cunning Plan (potentially picking a price higher than the prices actually benchmarked) has got a universal thumbs down.

If, indeed, it gets to make the decision at all.

There was one other submission, from CallPlus, which essentially said that the whole exercise is becoming a waste of time, since either the government will over-rule the Commission's low copper price, or Chorus will exercise its rights to have a full cost-modelling approach instead of all this benchmarking caper. Either way the Commission's proposed prices won't get a look in, so it should get a move on with a full cost modelling process.

It's a realpolitik analysis. But it could be a correct assessment of how things will play out.

No comments:

Post a Comment

Hi - sorry about the Captcha step for real people like yourself commenting, it's to baffle the bots