There are some writers in the blogosphere who aim to defend Canadians from the telecom bogeymen. Some of the most popular blog posts get retweeted regularly with endorsements from consumer groups and academics, spreading their influence further.
A few of these writers have created the common misconception that Canada’s high average revenue per user (ARPU) is evidence of high prices. That is not true, no matter how many times the lie is repeated.
Today, we saw an attempt to dispel a claim that Canada’s wireless capital spending per capita is among the world’s highest.
We have previously seen claims that Canadian carriers don’t spend that much on capital [see point 5 of this link], using an argument like “if it was really that expensive to build and maintain those networks, Canadian carriers wouldn’t be sitting near the top of the profit list.”
So it wasn’t surprising that today, we saw a post called “Canadian wireless investment: nothing special“.
In it, the thesis apparently is that if the capital intensity of Canada’s wireless carriers is middle of the road, then “what they’re actually spending is nothing special from an international perspective.” The conclusion is wrong, and not only because the arithmetic calculations underlying the blog post were flawed.
When I first read the piece, my initial reaction expressed surprise that the author was promoting a metric that confirms Canadian wireless revenues are appropriate for the level of capital spending.
After all, capital intensity is measured as capital spending divided by revenues. It is a ratio that aids certain kinds of financial analyses. If two companies have the same capital intensity, but one company has double the revenues per user, then mathematics says that this same company must have double the capital per user. Investors use capital intensity, together with other metrics, to compare past and possible future performance of the companies being analyzed.
Canada’s wireless industry has among the world’s highest levels of revenue per user. It should therefore be expected, that if its capital intensity is comparable to other countries, then Canada’s capital spending per user is among the world’s highest. As it turns out, that is is true.
The author wrote:
I sat down this past weekend and crunched the numbers from 71 of the biggest cellphone companies in the developed world, spanning 23 countries, and here’s what I found: despite the talking point, Canada’s wireless investment is in fact nothing special. It’s actually thoroughly mediocre.
He should have spent more time looking at the numbers he crunched [his spreadsheet – pdf]. The calculations are just plain incorrect. It appears that he may have inverted the ratio, and calculated revenue dollars per capital dollar spent.
The inverted ratio can still be meaningful, but it effectively confirms that Canada’s industry revenues are not out of line when considering the level of capital investment being undertaken.
But that clearly wasn’t what the author set out to do.
He intended to provide a number that could be compared to what the carriers report quarterly. For example, the post says:
For example, according to Telus’s 2012 annual report (links to PDF), wireless capital intensity was 12 per cent, up from 9 per cent the previous two years (the company does indeed typically spend more than its two rivals).
One would think that this quote would have made him pause when looking at the numbers he presented for 2012 in his calculations: TELUS 7%, Bell 8%, Rogers 5%. Wouldn’t you want to reconcile how TELUS reported 12% and your calculations show 7%?
The correct calculations would still show Canada’s capital intensity being middle of the pack. Unfortunately, the writer doesn’t appear to understand that this calculation doesn’t disprove the claim that capital investment is exceptionally high; in fact it proves the point. That is basic mathematics. If the denominator is high but the result is average, it means the numerator must be high.
It was disappointing to see academics and lobbyists re-tweet the “findings”, without pausing to think about what was actually being examined, or the validity of the test of hypothesis. There is a real imbalance in the information being presented. It was perhaps another missed teaching moment.
One of the challenges in engaging with some industry commentators is the need to prepare for a nearly immediate volley of ad hominem attacks. I have been called shill, lobbyist (as though this is a pejorative), sophist, propagandist, consultant. And that is just today’s collection.
I am proud to acknowledge being a consultant, although I am not (nor have I been for a number of years) engaged by any parties involved in the current wireless policy debates. And although I am able to be found in Canada’s registry of lobbyists, I have not yet engaged in lobbying activities. My client is a smaller Canadian wireless services provider, with policy interests often at odds with the so-called Big Three.
I write from the perspective of a Canadian with 33 years invested in launching competitive telecommunications services in Canada and the US. My views are mine.
I welcome your comments and perspectives. Feel free to disagree with my views, but don’t call me names.
As the writer of said post (I’m not sure why you went ahead and dissembled it without actually naming me), I do feel bad about messing up the calculations. I assume that others re-tweeted my initial findings because they have found my previous analyses to be correct, which is why I sought to promptly correct my error.
As you pointed out, the corrected numbers show that the result is the same: Canada’s wireless capital intensity is still middle of the pack when compared to peer countries. The results can be found here http://wordsbynowak.com/2013/08/12/wireless-investment/ or here http://petenowak2000.files.wordpress.com/2013/08/capintensity2.pdf. I don’t agree with your contention that these figures somehow prove exceptionally high investment, but that’s perhaps a matter of perspective.
Oh, and yes, Canada’s high average revenue per user does indeed seem to be the result of high prices, rather than magical voluntary usage or multiple SIM cards. I explain that here http://wordsbynowak.com/2013/06/18/wireless-usage/ and here http://wordsbynowak.com/2013/07/12/wireless-prices/.
Seriously, are there still people still trying to use ARPU as a proxy for comparing cell phone prices around the world?
Let’s say family “A’ spends twice as much money each month on gasoline for their car as family ‘B’ who lives on the opposite side of town. Both families drive the same model of car. Does that mean gas is twice as expensive on the side of town where family ‘A’ lives? Or does it mean family ‘A’ drives their car twice as much as family ‘B’, or does their car need a tune up, is their tire pressure too low, do they carry more or heavier passengers in the car? Until all the possible variables are identified and normalized in the analysis we are just guessing at the reasons for the difference. The same goes for cell phone ARPU.
Fortunately there are simpler and more appropriate ways to compare prices between countries. The OECD and the CRTC/Wall reports both use a “user baskets” approach which attempt to ensure some measure of “likeness” between the cost comparisons, but both reports still have limitations. The OECD report only looks at one carrier per country per basket so they are not showing the “average” prices in a country – that can be a big problem and skew the results if the wrong (i.e. significantly higher or lower than average price) carrier is selected. The CRTC/Wall report does average the prices of several carriers per basket but they don’t compare as large a number of countries.
The links above don’t, in fact, explain the relationship between what is referred to, confusingly, as “ARPU” — unclear whether this refers to average revenue per subscription, though it appears to — and pricing.
The posts also does a weird thing by trying to make sense of multiple SIMs by assuming that they are evenly distributed across the population. In other words, if there are 1.7 subscriptions per pop, they assume that this can be parsed as 1.7 subscriptions for each man, woman, and child, so that ARPU is 1.7 times ARPS.
That doesn’t even make a little bit of sense. Rather than assume that all 3-year-olds have 1.7 cell subscriptions, the more common-sense approach is to try and concentrate them within the proportion of the population that would actually use mobile phones, which is of course smaller than 100%. For instance, there are 98.3% as many mobile subs as humans, but only 30.1% of humans have subs. In Germany, it’s 137.9% and 65%, respectively. And so on. It’s still silly to assume that these subs are distributed evenly across unique users, but if folks are going to do that, they should at least do it in ways that aren’t ludicrous.
That said, if the goal is to make a determination as to the whole market, it doesn’t make that much sense to be starting with anyone’s ARPS figures in the first place. Start with whole-market mobile revenues, and divide by whole-market usage. That’s sort of the point, isn’t it?
I did not understand Mr. Nowak’s straw man, above, regarding a “contention that these figures somehow prove exceptionally high investment”. Surely any such contention is Mr. Nowak’s. He has repeatedly insisted that (a) there is an average ratio between Canada’s prices and its level of capital investment, and (b) Canada has the highest prices in the world. Any remaining argument isn’t a “contention”. It’s simply the corollary of the points on which Mr. Nowak insists.
Don’t forget that Telus and Bell have a network sharing agreement that makes for very efficient use of capital, and has served both the companies and Canadians well. The agreement has allowed rapid deployment of HSPDA, DC-HSDPA and LTE networks without skyrocketing capex intensities. Consider then that despite this agreement the two companies still report average capex intensity and you can assume that without this agreement actual capex intensity for the same services would have been much higher.
Rogers reports lower capex intensity, but this can be seen in their slower rollouts of more advanced network technologies, with a more prolonged reliance on its legacy GSM, for example.