Those high petrol prices - another view

There's a graph, Figure 3.8 on page 82, in the Commerce Commission's petrol market study that's puzzled me. And not for the first time: it also puzzled me when I first saw an earlier version of it, in MBIE's 2017 go at an inquiry into the petrol industry (where it was Figure 4 on p3). Here is ComCom's one.

It shows the price at the pump of a litre of premium petrol in a wide range of higher income countries, standardised by being converted into US dollars. Eyeballing the graph, you see New Zealand is there at roughly US$1.47. At the exchange rate of the time (March quarter '19) of 68 US cents, the price converts into NZ$2.16, which looks right. All good.

Because the price at the pump is heavily affected by local taxes, for competition policy purposes you need to focus on the price ex taxes, which is shown in blue in the graph. New Zealand does not show to advantage, with the third highest petrol price. Cue song and dance about how bad we are.

But what's been puzzling me is the weirdness of the country rankings. Your first inclination is to go looking for some underlying explanatory patterns - transport costs from major oil fields or refineries? - but it's hard to spot any. The three countries at the top - Mexico, Korea, us - are as odd an assortment as you'll ever see. The three at the bottom - Slovenia, Chile, Finland - don't obviously have much in common, either.

The ordering could of course reflect differences in local competitive intensity. You look at Mexico's top billing, for example, and you wonder about Pemex, a state owned monopoly up to 2013 which still has nearly three quarters of the petrol stations. You wouldn't know about the rest of them without some intensive investigation along our own Commerce Commission's lines.

But I'm also wondering whether the somewhat jumbled pattern mightn't partly reflect the fact that the petrol prices have been converted into US dollars at market exchange rates, rather than at purchasing power parity (PPP) exchange rates.

If this whole 'which exchange rate to use' thing isn't your bag, let's backtrack for a moment. If you're comparing, say, the price of an Apple i-Pad Pro 11" Wi-Fi 256GB, you'll find it's on Amazon at US$799.99 and you'll find it's NZ$1648 at JB Hi-Fi. At today's exchange rate (63.7 cents) the Amazon one costs NZ$1256. Good deal cheaper in the US.

But market exchange rates are fickle beasts and move around a lot. Because an iPad is expensive locally today doesn't mean it mightn't be locally cheap next Tuesday if the Kiwi dollar were to fall sharply against the US dollar over the weekend. You shouldn't be drawing any long-term policy conclusions about iPads - or petrol - on the basis of an exchange rate that might make a fool of you in no time.

Which is why these international comparisons are more normally done on a different basis. Supposing you went out and bought a wide bundle of stuff in the States, and it cost you US$100,000. You do the same in New Zealand, and it costs you NZ$150,000. It would then be fair to say that US$1.00 has the same buying power as NZ$1.50. In that case you wouldn't be in the least bit surprised if a litre of petrol cost US$1 in the States and NZ$1.50 here: that's just what you'd expect, because anything that costs a US dollar in the States is on average likely to cost NZ$1.50 here, as we discovered on our shopping expedition.

Long story short, people making international comparisons tend to use that US$1-equals-NZ$1.50 exchange rate, called the purchasing power parity rate (obvs). And here's what happens when you do that same chart of ex tax prices in blue above, but at that PPP rate instead. I've used the latest (2018) PPP rates as calculated by the OECD (you can find them if you fossick here).

In New Zealand's case, the pre-tax petrol price doesn't change much. It was around 77.5 US cents before (again eyeballing the number from the ComCom graph, as I'm not going to pay the €900 the International Energy Agency wants for the exact data). At the March quarter market exchange rate of the time, 68 cents, that was NZ$1.14. The PPP exchange rate wasn't very different: it was 67.6 cents. So our local price translated into US$ at PPP was 77.1 US cents, rather than the 77.5 US cents price you get at market exchange rates. Same diff.

But other countries' prices move around quite a lot when their PPP rates are used instead of their market exchange rates. And the end result is that our relative position drops quite a bit. We were third highest out of 33 on a market rate basis: on a PPP basis we're 14th out of 33. There's a bit of imprecision here, as I've used eyeball data rather than precise ones, so I wouldn't obsess over whether it's 14th or 13th or 15th*. This PPP ordering also makes a bit more intuitive sense than the market rate one: the bottom three, for example (now Norway, Finland, Iceland) look like a more coherent bunch.

There are still good reasons for having a market study look at the petrol market: those rates of profitability that ComCom found, in particular, need some explaining.  But one conclusion from this exercise is that I wouldn't get carried away by the "we're one of the dearest in the OECD" line of argument.On this, entirely conventional, alternative way of making the comparison, we're a little bit on the expensive side of middle of the pack.

*If you want to see the data I've used, and maybe check I haven't got the wrong end of any sticks, it's here (assuming I've got Dropbox working right).

Did we ask the right question?

Normally I'd take time out early on, to explain some of the telco/regulation jargon, but as I suspect virtually everyone likely to read this post already knows what UCLL, UBA, WACC and TSLRIC are, I'm going to plunge right in.

Unless there is some diehard with a large legal budget to burn, last month's final decision by the Commerce Commission setting the wholesale price of copper-based broadband at $41.19 brings the curtain down on a long, complex and contentious process.

I'm not going to spend much time revisiting the details of the Commission's decision, other than to note that in an exercise like this one, with so many moving parts, there is inevitably going to be room for even reasonable people to disagree.  In their place, I might well have gone a different way on aspects of WACC, for example, or taken a different position on the challenge of digging trenches through New Zealand's allegedly idiosyncratic topography, but so what: overall it was a reasoned, careful piece of work.

The end result looks to have ended up in the right area, and I say “area” advisedly: there is no single “true” point-estimate answer to these sorts of questions. You only have to glance at Figure X1 of the decision, for example, to see how second thoughts on various component bits can have significant impacts on the final price. I was comforted in particular, as I thought it was a realistic result, by the Commission's calculation (in paras X34-5) that its price on a “like for like” basis came out lower, at $31.60, than the entry-level UFB price of $37.50. And no, I don't want comments from conspiracy nutters who think the Commission was steering its decision to within cooee of the UFB price.

But let's step back from the specifics and consider some broader picture issues.

For me, the big one was the task the Commission was constrained to do. By law, it had to set a forward-looking TSLRIC price, and it did. But that task answers the question, “what would be the cost to use a shiny new whizzbang broadband network rolled out efficiently today”, not the question, “is Chorus ripping off my broadband provider, and, at one remove, me?”. Or as the Commission summarised it

E109 … our task is to set a price according to TSLRIC which we and submitters agree is best set relative to the forward-looking efficient costs of the hypothetical efficient operator, rather than the past costs of the regulated entity.

E110 Putting that another way, we do not consider that concepts such as “windfall gains” or “windfall losses” are particularly relevant. A price determined according to TSLRIC does not attempt to regulate the incumbent’s revenues such that it earns a normal return on its actual investments. Rather, it attempts to set a forward-looking price based on modern technology and irrespective of the incumbent’s past investment decisions.

The Commission is completely correct in saying that its hands were tied, and that any financial impact on Chorus was beside the point. And I know that TSLRIC is a good price to use if (for example) your main issue of interest is whether retail broadband providers (or others) are given a fair go at getting into the wholesale broadband business for themselves. And that's an important point if you believe that consumers are best served in the long run by competing suppliers deploying a choice of infrastructures.

But I'm still not convinced that the Commission was asked the right question in the first place. There was a good consumer case for an alternative approach, the one the Commission described in the extract quoted above as an “attempt to regulate the incumbent’s revenues such that it earns a normal return on its actual investments”, and which is standard in many regulatory proceedings.

Consumers, for example, are wearing the cost of a hypothetical brand new network. It's as if the government had decided to regulate the price of cars, and as part of the exercise decreed that only the latest new models in the showroom would be made available. Many of us, however, would prefer to get our cars from Honest John's Japanese import yard. Instead, we're being asked to pay the depreciation costs of a brand new car, and we've lost the option of paying a lower price for an already depreciated second-hand one.

So yes, I'm still concerned that the inherent design of the regulatory regime had the capacity to deliver windfall gains to Chorus (at consumers' expense) or windfall losses (at Chorus's expense). The whizzbang modern technology element probably counts against Chorus if it is still using some earlier generation gear, as does the efficiently deployed element, if it is still carrying inefficiencies from its dominant incumbent Telecom days. But the depreciation aspect is a clear bonus for Chorus: the price is based on a brand new network, with its deployer entitled to a reasonable rate of return on the whole amount invested. Chorus, on the other hand, has already recovered some proportion of its investment, through depreciation, and should only receive a reasonable rate of return on the funds still invested. To that extent Honest John is being paid the new car price.

There's another aspect that also leaves me wondering if the TSLRIC question was the right one to ask.

Earlier, the Commission had provisionally set the broadband price on an international benchmarking basis. It's an approach which I hope survives the current review of telco regulation, as I remain convinced that it can in many circumstances provide a decent enough approximation of what the New Zealand price ought to be. But it didn't work on this occasion: the Commission was (again) constrained, and could look only at the overseas prices in countries that also costed broadband on a TSLRIC basis. And the answer to that quest was – Sweden.

Understandably, various parties weren't happy that New Zealand prices were going to be set on the basis of an overseas sample of one, we moved smartly into the final pricing process that concluded last month, and the initial benchmarking exercise became moot. But it still leaves a big question unanswered: how come there are only two countries in the developed world that are setting broadband prices this way?

I know, there will be times when there are just a few enlightened countries ahead of the pack, and most everyone else is doing it wrong. I'm not sure, though, that this is one of those times. While we're in the mood to have a rethink of our telco regulation regime, why don't we ask ourselves if we can live with mainstream overseas practice – especially if the alternative is fractious, protracted, complicated, and expensive.

There has to be a better way. And there is

I'm convinced there's a better way to get a good fix on some of our more contentious, and important, regulated telecoms prices. Let's deal to some jargon first, and then we'll get properly underway.

If you've got broadband, you get it from an Internet Service Provider (your ISP). And chances are it arrives over the copper wire phone line to your house. You could be on wireless broadband, or you might have signed up for the flashy new fibre network that's being rolled out, but most of us are still on the old copper based system. It's owned by Chorus, and your ISP pays Chorus for the use of that copper line from your house to the nearest telephone exchange. That service is known as the Unbundled Copper Local Loop, or UCLL. ISPs can put their own equipment in the exchange and take the feed from there, or they can rent some gear from Chorus instead of providing their own: that's the Unbundled Bitstream Access service, or UBA.

And finally - and this is where things come closer to your wallet - you've likely noticed that your ISP has said it'll be raising its price to you by $4 a month or so, because the Commerce Commission, which regulates the copper line UCLL price, is in the process of raising it from $23.52 a month (its first stab at the right price to charge) to $28.42 a month (its estimate after going through a full cost modelling exercise).

There's a consultation process going on before the Commission's proposed UCLL goes final (all you can eat here). As part of that process, Spark has come up with this graph, which shows how the Commission's proposed price compares with the price charged for the same service in a range of other developed countries.

You may have seen this already - the Herald's technology columnist, Chris Barton ran with it in a recent article, "Something rotten in our Commerce Commission", where among other things he concluded that "by a curious combination [of] free market ideology and caving to political pressure, it's [i.e. the Commission is] promoting monopoly power and a haughty "let them [the end-users] eat cake"." No doubt the Commissioners sacrifice children to the Great Werewolf, too.

In any event the graph does give you pause for thought about various aspects of how telco prices are set by regulation. My main point: I think there's greater room for using information on overseas prices as a guide to setting our own.

We do it a bit, at the moment: that "first stab" the Commission had at setting the price was required, under our Telecommunications Act, to be set by "benchmarking" against prices overseas. Unfortunately the benchmarking was tightly circumscribed in the Act, and had to be "Benchmarking against prices for similar services in comparable countries that use a forward-looking cost-based pricing method".

You can understand the logic. You wouldn't want prices to be set here solely on the basis of countries that weren't at all like us (eg a highly dense conurbation like Hong Kong), hence the "comparable" test, and you wouldn't want prices to be imported into New Zealand that had all been plucked out of the air on some cockamamie basis. And that's a real risk: regulatory proceedings can easily get captured by one vested interest or another. Money politics can see incumbents' prices set on too-favourable terms; populist politics can set prices that don't cover incumbents' costs. So you can see why the legislation saw fit to use prices only if they were set in a particular way.

Trouble is, you can take intellectual purity too far. After filtering according to the Act,  the latest benchmarking exercise, for the UBA service, ended up with only Denmark and Sweden to look at, which left everyone feeling a bit uneasy. I doubt if even the framers of the Act would have liked a benchmarking process that featured only two smallish Scandinavian countries.

So why don't we take a different tack? Why don't we go the Spark route, and look at the whole range of prices overseas? It would make sense to keep some element of comparability, so we might want to restrict it to say the OECD countries, but even that would leave us with a largeish group of 33. Some prices may well be off, and unfairly tilted towards suppliers or consumers, but on average you'd be inclined to think that the truth will appear somewhere in the middle. You might worry that New Zealand has got some special features that make it impracticable to compare with the average overseas experience: people like to raise the "long and stringy" argument, for example (though you'd think places like Norway and Sweden are much the same). All I can say is that I've seen a lot of folks argue both sides of the "New Zealand is unique" case, and I still don't see a knock-out case for our conditions being completely idiosyncratic.

Regular readers - God bless both of you - will know that I've banged on before (for example here and here) about using benchmarking more extensively in our price regulation, and I'd like to see our revised telco regime, when it eventually materialises, reaching more often for the regulatory equivalent of Number 8 fencing wire. It may be low tech, but it's admirably cheap and serviceable.

KISS

This morning the Commerce Commission released the wholesale price Chorus is allowed to charge to Internet service providers (ISPs), and which therefore is the core component of the retail prices those ISPs charge you for your fixed line broadband.

It's made up of two parts, the first being the bit for the cost of the copper line from your place to a Chorus switch (the 'local loop' or UCLL) and the second ('UBA') being the cost of the fancy electronics that Chorus can (optionally) provide to ISPs to save them having to use their own. The local loop bit will be $28.22 a month and the UBA bit will be $10.17 a month, making a total of $38.39. This compared with the previous price allowed, of $44.98.

These prices are based on explicit, detailed and complex modelling of the costs involved, and are intended to replace the interim hold-the-fort prices that the Commission had previously set, based on the cost of the same services overseas in countries who do things much the same way as we do. This 'benchmarking' exercise had set a local loop price of $23.52 and a UBA price of $10.92, making a total of $34.44.

There are all sorts of issues involved here, big and small, affecting everything from the profitability of  Chorus through to uptake of the country's shiny new ultra fast fibre network. And they directly affect you, too: already some ISPs are saying that the drop in the wholesale price (from $44.98 to $38.39) had already been passed on to you, so you won't be getting any further joy out of it.
In any event, I'd like to pick on one small aspect of the process, even though it's largely moot now, and it's about those interim 'benchmarked' prices.

I think they did a good job of providing a quick, cheap and reasonably accurate initial estimate of the eventual wholesale price. They were pretty much spot-on when it came to the UBA part ($10.92 versus $10.17), which is remarkable given that everyone was agreed that the benchmarking process had only a couple of countries overseas to use as sighting shots. And they weren't far off when it came to the local loop component, either ($23.52 versus $28.22) - especially when you consider that the fully modelled cost estimate involves a whole swathe of judgement calls made by the Commission and its modellers, and is not a glimpse into some eternal truth held in the mind of an omniscient Being.

So I'd take two lessons away from this, both involving the KISS principle.

The first is that over the next couple of years we're going to be taking a close look at the shape of our telco regulatory regime, and I'd like to suggest that we keep the cheap and cheerful benchmarking process. It's relatively fast - a particularly important consideration in fast moving markets like ICT - it's relatively transparent, it's understandable, it's relatively cheap, and it's accurate within some rough-and-ready-justice tolerance. I'd go further, and make it harder for parties to invoke the full cost modelling approach, which introduces layers of cost, delay and complexity, and all for a gain in 'accuracy' that (because of multiple modelling options) may be more illusory than real. And in general I'd like to see the 'good enough' option chosen over the one that keeps consultancies on three continents in business.

The second is that we need to think harder about the increasing complexity and cost of regulation across all sectors, and not just the telco business. I agree with Eric Crampton of the NZ Initiative, when he said on Interest.co.nz that "Too much of New Zealand’s regulatory apparatus would suit a country of forty million rather than the one we have". He's got his own examples: one I came across recently was the Commerce Commission's needing to sign off a $3 million increase in capex spending on a little Transpower project in South Canterbury. The process will take five months from start to finish, and has already spawned a 54 page initial draft decision.

That's a bit of an extreme example, and I should make it clear that it's not the Commerce Commission's fault: it's been lumbered with this ludicrously over-engineered regulatory regime. And I should add that from next April the Commission won't have to get out of bed for anything under $20 million - which is, of course, where the threshold for its involvement should have been in the first place (if not higher again). And I'd have to note that bloodymindedness on the part of Transpower and its customers drew this intrusive regime on their own heads, and a bit of enlightened give and take could have avoided the whole mess.

But it's there now, and it's holding up the sector, and its cousins in other sectors are also increasingly clunky and costly. It's time for more people in the policy analyst community to do what the MD of one company I know used to do: hold up the sign that says, "Does it make the boat go faster?"

Two 5 - 0 defeats

Chorus got bowled like ninepins this morning by the  Court of Appeal, having earlier been skittled by the High Court.

The cases were about the Commerce Commission proposing a big reduction in the price Chorus could charge for UBA, or as it is formally defined, "the additional UBA service component, which allowed access seekers to supply broadband services over Telecom’s copper access lines without investing in their own equipment or software". In other words, the bits and bobs that carry broadband traffic across the gap between the copper line from your place and the start of an ISP's network.

The reduction (roughly halving the price) had been based  on a benchmarking exercise, where the Commission (as required by the Telecommunications Act) looked at the prices overseas for UBA as a quick and dirty proxy for what it might well cost here. There's lots more about the exact details of the benchmarking comparability exercise, but that's the gist of it.

The Court's decision is here as a pdf and there's a shorter media release pdf if you prefer. Chorus's reaction is here: essentially, not surprised by the outcome, but felt they had to make a point about what they see as a regulatory regime mess around pricing of broadband services.

I'd reckoned, as I said some time ago, that (a) Chorus had very little chance of succeeding and (b) in any event the whole exercise was a waste of money, but Chorus went ahead anyway. And this morning, sure enough, they got squashed like a bug. Chorus had raised five issues: the Court said No to each and every one of them, as the High Court had earlier. While Chorus has said it is studying the decision, and I suppose could take it to the Supreme Court, after two successive 5 - 0 defeats you'd think they'd flag the game away.

Chorus didn't get anything helpful about any incoherence in the regulatory regime, either. The reverse, if anything, when the Court of Appeal said at [44], "the mandatory requirement for the Commission to carry out the “benchmarking” exercise...is itself designed to implement the statutory purpose, not to contradict or undermine it". In other words, the Telco Act is internally consistent.

So now on we go to the Commission's final word on the UBA price, which will be determined by modelling the actual costs of an efficient provider in New Zealand (Chorus had exercised its right to object to the benchmark stab at the price and to have local costs estimated explicitly). According to its media statement welcoming the decision, the Commission says it expects to have the first draft of the "real" cost (my words) in December.

Who knows what that price will be, but I wouldn't be in the least bit surprised if it came out within cooee of the original benchmarked stab at it. As I've said before,  my experience across a number of contexts is that  often benchmarking gets you to an approximately correct place, and far quicker and much more cheaply than the model-building route. I didn't start at that position - in fact, I originally thought the exercise would be too flaky to rely on - but what you find, when you get your hands dirty, is that you can say, the cost of this thing is somewhere around X. It might really be 1.1 times X, or 0.9 times X, but it sure isn't twice X or  half X. And that shouldn't be too surprising: in areas like telecoms, for example, companies tend to be delivering much the same sort of thing with the same sort of technology.

It's all moot now, as we're donkey deep in the formal cost modelling, but I'll say it anyway: I have a soft spot for simple, practical-enough regulation options like benchmarking. For all our general preference for light handed regulation and our national propensity to come up with a cheap and cheerful Number 8 fencing wire answer to things, our current regulatory approaches seem to be favouring ever more complicated, ever more expensive, ever slower, ever more intrusive options, as anyone who's had anything to do with the regulation of electricity lines businesses (for example) would agree.

We're currently having a review of the telco regulatory policy regime. It would be nice, as an outcome, if simplicity and speed got more of a look in than they do now.