Subscribe to our blog: 

Posted on 08 Mar 2016 by Gareth Brown

Statistical Buggery: RBA Urbanisation

Statistical Buggery: RBA Urbanisation


The Australian housing bubble, or lack thereof, has again been dominating headlines in recent weeks, since a segment aired on 60 Minutes. I’m too battle-scarred to see any upside in proffering an opinion on the matter.

But there is one statistical anomaly that keeps rearing its head that I’d like to call ‘bullshit’ on. And that is the extent to which Australia is more ‘urbanised’ than most other countries.

Australia is a highly urbanised country, or perhaps more accurately ‘suburbanised’. More so than most nations. This is offered as a justification for higher house prices, and perhaps not without reason. But there’s an old RBA chart redoing the rounds at the moment that is flat out wrong. It was most recently republished in Fairfax article Property bubble fears are overblown, says CBA.

The chart is at least a few years old, the first reference I can find is Graph 6 in this 2013 speech by Luci Ellis of the RBA, but the data may be older. It certainly appears to have been compiled by the RBA. Here’s a replication of the chart from that 2013 speech.

Statistical Buggery: RBA Urbanisation

Let’s try and recreate the numbers for Australia and a few other nations.

It says the ‘urban population’ of Australia, living in cities of greater than 100,000 people, is a little north of 75%. This is clearly constructed using the entire metropolitan area of each city greater than 100,000. A quick gander at Wikipedia suggests that if we add up the entire metro population of each city from Sydney, population 4.8 million, down to number 16, Toowoomba, the smallest with a population greater than 100,000, about 17.9 million people live in these ‘urban’ areas. That’s about 77% of Australia’s population of 23.13m, according to Wikipedia, and lines up nicely with the claim from the chart.

Now let’s look at the US. Here, according to the data provide by RBA, the population of cities greater than 100,000 people is less than 30% of the nation’s total.

If we take the list of US cities by population from Wikipedia, there’s 297 cities with a population greater than 100,000, and the total comes to 90.5 million, or 28.4% of the total US population. So far, the RBA’s numbers look spot on.

But any demographer worth their salt, or any competent person tasked with putting together a chart on urbanisation, knows that US urban populations are defined very differently than those in Australia.

Those numbers show that New York has a population of 8.5 million, which incorporates the five boroughs—Manhattan, Queens, Brooklyn, the Bronx and Staten Island. It ignores the 15m other people living in the greater New York combined statistical area.

So our US analyst Kevin Rose, who lives in Westchester County, a fairly densely populated suburban area 38 contiguously-settled kilometres from Central Park, is considered ‘non urban’ for the purposes of that RBA chart. Were he to live in less-densely populated Campbelltown or Cranbourne – 61 and 51 kilometres from the Sydney and Melbourne CBDs, respectively – he’d be considered ‘urban’.

Those numbers from which the RBA stats are derived also say that Los Angeles has a population of 3.9 million, making it smaller than either Sydney or Melbourne. Measured the way we measure city size, Los Angeles’ population is actually more like 18 million. Those that live in Santa Monica, LA’s equivalent to Bondi or St Kilda, are considered ‘non urban’ by the RBA’s numbers.

The most accurate like-with-like comparison is to use the combined statistical area of America’s largest cities, it’s the closest equivalent to how we measure city size in Australia. In that case, 239 million people live within the greater metropolitan area of cities greater than 100,000 people. That comes to 75.0% of the US population, very similar to Australia’s 77% and very different to the less than 30% claimed on the RBA chart. Is America also too urbanised to be susceptible to a housing crash?

And it’s not just the US numbers that are wrong. The chart show France’s population at about 15-16% ‘urban’. But more than 20% of France’s population resides in Greater Paris alone. I make the like-with-like number, for direct comparison with Australia’s, closer to 56%.

Italy 22%? Try north of 40%.

Spain at 40%? More like 60%.

South Korea 45%? Ha, more than 50% of the South Korean population live in the Seoul greater metro area alone.

Germany a little over 30%? Please. Almost 52m people, 64% of the German population, live in the greater metro areas of its ten largest cities alone.

Does Australia have a housing bubble? I’ll leave that for others to argue. But if you’re relying on the RBA’s comparison of urbanisation around the world, or the multitude of media reports that have unquestionably replicated the claims, as an underpinning for high house prices, you’re relying on bad data. It’s chiefly a comparison of global urbanisation and Australian suburbanisation.

This chart was put together by clowns, not competent statisticians, and deserves to be appropriately derided.

Statistical Buggery: RBA Urbanisation

15 thoughts on “Statistical Buggery: RBA Urbanisation

  1. Oh “spade” where art thou? Thanks for imparting your knowledge and experience!

    I don’t know if it is the Brisbane city council finally catching up or if house prices really have increased by that much – I received my rates land value notice yesterday and it went up by 27% in one year! I discussed with my neighbour and they had a similar increase. We are 15km south of the city itself. Let’s hope there is a bubble and a crash soon because I am selfish and I want to go bargain hunting.

    • If Brisbane is anything like Sydney, the councils put up their rates on a 3 year cycle, so that’s what you’re seeing.
      The interesting one in Sydney is that Investors will be seeing Land Tax rising substantially over the next 2 years at least because the OSR uses the 3yr average of the Valuer General’s annual valuation… At the same time rents are stagnant… so not a good scenario for investors.

  2. Great analysis and article Gareth.
    Who knows how many times this chart has been used as supporting data in a speech or policy decision. Yet it is clearly wrong.
    I suggest it be forwarded to Glen Stevens at the RBA to ask just who compiled it and perhaps give some sort of explanation. After all shouldn’t we be able to trust the RBA to provide at least accurate charts and data.
    I’ll send it through to the them but don’t want to take away from the credit you should rightly get so I suggest Forager do the same as this is the sort of material that investors require to be accurate.
    The connection with the UN doesn’t surprise me. A nudge here, a wink there and presto they could compile a chart to say anything.

    • I saw you sent it through to the head honcho, good idea Brett. I’ll be chuffed if Mr Stevens reads my blog post. Thanks

  3. Great article.

    My wife and I caught a bus between Sao Paulo and Rio last year and it was one of the most amazing things I’ve ever seen. No joke, there’s bunches of 10 to 12 skyscrapers visible the entire way. We’re talking the distance between Sydney and Wagga Wagga, and I don’t think there was a stretch of road where a tight cluster of skyscrapers couldn’t been seen. We couldn’t believe our eyes. Calling Rio and Sao Paulo separate cities in the same way that Sydney and Melbourne are separate cities doesn’t make sense to me. Which makes the Aus vs Brazil graph pretty pointless. Regardless of where you stand on house prices, the devil is in the detail with international comparisons. It’s so complicated, I’m not even sure that it’s worth the bother.

  4. Great analysis Gareth. A lot of times statistics get recycled (myself included) with authors not taking the time to understand the methodologies used to compile the statistics. A good statistical lesson.

  5. Here another example of unthinking parroting of urbanisation statistics:

    Author claims 40% of Australians live in Sydney or Melbourne, and that is correct if you’re talking about greater metro areas.
    To highlight how high this is, the author then claims, according to IMF, less than 10% of Japanese live in the two largest cities – Tokyo and Yokohama. Well, if you use greater metro areas (as you did for the Sydney and Melbourne stats), the two largest Japanese cities are actually Tokyo and Osaka, and their combined population is over 55m. That’s about 44% of the national population. Japan has even more concentration in its two largest cities than Australia. Where’s the critical thought/fact checking?

  6. In fairness, this sort of statistical sloppiness isn’t just limited to the RBA. It is endemic in just about every aggregation of statistics out there, including [gasp] Bloomberg.

    Most of the packaged statistics that get quoted are either put together by journalists (i.e., entertainers with deadlines) or they are plugged into proprietary databases by anonymous underpaid drones.

    This is why there is simply no substitute for digging up source documents if you want to get a better idea of what is really going on.

    It’s also one of many good reasons why reading newspapers, watching TV news or getting information from any form of mass media is probably a hindrance when it comes to investing.

  7. I’m sure there is much truth in what you say SG but that doesn’t make it right.
    If an organisation like the RBA can’t be trusted to produce accurate charts and data, and it requires one to double check what they produce, it sure makes for a cynical reader. I rather think we should hold the RBA ( as with any media or analysts etc, yes and that even includes company boards and management in their annual reports etc) to produce accurate and transparently correct material.
    I don’t know Glen Stevens personally but have heard him on a number of occasions and I would think he is of the ilk that inaccurate data from the RBA would not please him.
    That’s why I appreciate what these Bristlemouth articles do, they cut through the crap, to reveal what is really going on. Ditto for value investing because it looks at the real business not the PR story that often management and boards like to convey.
    So let’s hold the RBA to a high standard I say.

  8. Gareth. To ask the question (and get you on the topic 🙂 ) why does urbanisation (as defined) lead to higher house prices? I would have thought if Australians all left the cities, whether the average house price went up or down would continue to be a function of interest rates, unemployment rate and LVR ratios as it now?

    • I don’t know if it’s a valid point to begin with. Simply stated, their argument is that more Australians want to live 20kms from the city and close to the coast, and that’s gonna create more competitive markets and a necessity/preparedness to invest a greater % of income on housing than if we were just as happy to live in any available area. But hopefully my data show that it’s fairly invalid assumption to begin with. Australia is barely more suburbanised than the US. It’s also worth pointing our that urbanisation also tends to push up the income level utilised in house price-income and housing affordability ratios.
      Whether urbanisation leads to higher justifiable house prices is one thing, whether it protects against crashes another. Hong Kong is more highly urbanised than just about anywhere, and fell more than 50% top to bottom a few decades ago. Japan is extremely urbanised (more so than on the RBA chart) too, and house prices fell dramatically after their 1980s bubble. I’d take no comfort from urbanisation stats, even if they were compiled properly.

      • Great post, as always Gareth. Why is a population of 100,000 so special anyway? That’s a little like saying a 20c stock is cheaper than a 2$ stock. As you allude in your final comments, surely it’s the density of dwellings that is the more critical factor, as this may correlate wity the supply-demand equation. It would be interesting to see a plot of house prices versus dwelling density (per square metre). What would be interesting would be to see this plot for each country. A comparison of national averages would not suffice, I suspect, as the critical detail would be lost. Then again, perhaps on this measure, Lagos, Nigeria, should have some of the most expensive real estate in the world. Then there’s the the slums of Rio (favelas). Hmmm… back to the drawing board.

  9. I would have thought, using ‘common sense economics’, that one thing urbanisation does (by definition) is increase the competitive pressure on house prices in urbanised areas by having more people competing for a fixed amount of space (land and thus houses on that land). That seems like a recipe for increased prices to me. Might have to dig out my Economics101texts for a more formal explanation but I would think a quick look at real estate prices in say Homebush in Sydney versus say Albury or Bathurst in NSW, and drop on that a chart of population per hectare or square kilometre would corroborate that.

Leave a Reply

Your email address will not be published. Required fields are marked *

Security Test *