Growth of World Cities (1940-2010)

[nito]

These debates are like birthdays; they keep coming up! I think a big problem in trying to apply North American urban methodologies to London is that (thankfully) it had the Green Belt to prevent substantial sprawl. In turn attempting to apply British methodology to American cities would result in anomalies such as everything west of the Hudson being classified as a completely separate urban area.

One area that I was looking at recently – which had posed some interesting results, but I had yet to finalise – was looking at the number of people that commute into a city from beyond the city boundaries. Naturally that approach has its own issues, but it was an interesting angle to look at nonetheless. These are the figures for New York and London:

New York: 878,954* (2010)
London: 793,400** (2011)

* Commuter adjusted Daytime Population: Places, United States Census Bureau. Note: 608,954 figure is net, 270,000 New Yorkers reverse commute.
** Characteristics of Commuters, July 2015, Census Information Scheme, GLA Intelligence, page 3.

It will be interesting to see what the figures look like in subsequent census data releases due to the regions surrounding London experiencing stronger growth relative to the states surrounding New York City, and the unprecedented improvements to the London commuter network.

And something I’ve been meaning to show off for some time, this pretty cool .gif image of people commuting into the Square Mile from across London and further afield.


Source: Timeout: http://now-here-this.timeout.com/201...ndon-every-day

[Minato Ku]

Anyway if the heart of the urbanised area of London was separated by a large river like Hudson, I think that ONS would have altered their methodology accordingly.

On the other hand, it is not certain that ONS would have considered New York City and Stamford to not be part of the same urban area.
There is no gap of urbanization between New York City and Stamford, yet both cities are not part of the same urban area according to US census.
Seeing this, it is not sure that Hemel Hempstead or St Albans would be part of London urban area if it was calculated according to US definition.

So you should not believe that US methodologies are oversized, stupid or don't take account of the scary sprawl.
If New York City and Stamford are not part of the same urban area, neither the same MSA, this shows rather strict rules.

Remember that despite being almost part of the same contiguous urbanized area, New York City and Philadelphia are two distinct metropolitan area, two distinct CSA.
It is true that New York CSa cover a large surface, most of this land is made of very sparsely populated area. You could divide the area by two and it would not have a big effect on the population.

[nito]

There is nothing inherently oversized or stupid about US methodologies, but it is hard to adopt the same methodologies in another environment which has evolved in a completely different direction. Perhaps in a reality where the UK didn’t pursue the Green Belt strategy (visible in the .gif image posted in my previous post), there would have been sprawl from Cambridge to Brighton, but thankfully that never materialised. Which is why I was curious to look at the actual number of people that commute into New York and London, and that the actual deviation is only 10%.

Interestingly there is a similar situation to the Hudson in the UK: the Mersey separates Liverpool from Birkenhead, and as a consequence the ONS defines each as two detached urban areas, although confusingly both are part of the same ceremonial & metropolitan county.

[Double L]

Houston's statistics look great in this thread!

[TarHeelJ]

Maybe I'm confused as to how you're choosing cities to add to the list...but I can think of several US cities that would easily fit in. DC, San Francisco, Phoenix, Atlanta, Dallas, Seattle, etc. Do these cities not fit your criteria?

[New Brisavoine]

Originally Posted by Crawford

No, the problem in this thread is that you made up arbitrary metro area boundaries, instead of using a consistent metric across cities, where available.

Indeed.

Figures come from the 2011 UK census and can be easily double checked here: https://www.nomisweb.co.uk/query/sel...in+Destination

For London, as for European cities in dense environments like Brussels, Cologne, Amsterdam, Milan, etc., taking 35,883 km² of territory around them just because the NY CSA covers such a territory, or even just 21,482 km² because the NY MSA covers such a territory, makes no sense because these cities are surrounded by other cities strong enough in general to resist their pull and have their own labor markets.

In the case of London, the commuting figures from the 2011 census show that the metro area does not extend very far, whatever the definition we use. If we use the French definition of metro areas (more than 40% of residents in employment commuting to the urban core), the London metro area extends only over 3,976 km² and had a population of 10,326,645 on Jan. 1, 2012.

Using the US MSA definition (more than 25% of residents in employment commuting to the urban core), the following towns and cities would become included:
- Brentwood UA: 39.0% (the figure refers to the percentage of people in the urban area of Brentwood who have a job and commute to the 3,976 km² of the metro area defined at 40% above)
- Sevenoaks UA: 38.3%
- Aveley UA (Thurrock): 38.2%
- South Ockendon UA (Thurrock): 36.6%
- Gerrards Cross UA: 36.4%
- Hertford/Ware UA: 36.2%
- Dorking UA: 34.9%
- Hatfield UA: 33.6%
- Redhill (Reigate) UA: 32.8%
- Grays UA (Thurrock): 32.7%
- Welwyn Garden City UA: 28.0%
- Slough UA: 26.0%

These are the only urban areas beyond the London metro area at 40% which cross the 25% threshold (bar a few other tiny ones that I haven't indicated). Such a metro area defined at 25% commute threshold would have between 11 and 12 million inhabitants (probably closer to 11 than 12 million; I haven't made the exact calculation for that, but could do it), and a land area not much more than 5,000 km².

Using entire administrative districts instead of just urban areas (Built-up Areas, or BUA) would perhaps add a few hundreds of thousands people more, but in any case it looks impossible to get more than 12 million with the US MSA definition applied to London.

Beyond the urban areas mentioned, the labor attraction of London drops rather quickly.

These are the percentages of residents in employment who commute to the metro area of London defined at 40% for a few administrative districts:
- Medway: 20.2% (i.e. 20.2% of the people living in the district of Medway who are in employment work in the metropolitan area of London of 3,976 km²)
- Rushmoor: 19.4%
- Wokingham: 16.4%
- Luton: 15.6%
- Crawley: 10.5%
- Reading: 9.7%

Further away, in zones included by Yuriandrade to reach 34,415 km², it's even worse: barely 6% in Canterbury, less than 5% in Suffolk, etc.

admin note: this post is allowed and cites appropriate data

[Yuri]

Originally Posted by TarHeelJ

Maybe I'm confused as to how you're choosing cities to add to the list...but I can think of several US cities that would easily fit in. DC, San Francisco, Phoenix, Atlanta, Dallas, Seattle, etc. Do these cities not fit your criteria?

I'm out of São Paulo, on vacation. Many of those mentioned by you will be featured on the next rounds as some other areas outside the US.

[Yuri]

Originally Posted by New Brisavoine

admin note: this post is allowed and cites appropriate data

The thing is, you don't care about any data. You just want to prove France is better than anything else, in the most weird obsession I've ever heard of. That's why you are usually not welcomed.

BTW, there are several criteria to establish a metro area, and commute patterns being just one of them. For instance, from Wikipedia:

Australia

The Australian Bureau of Statistics defines 'statistical divisions' as areas under the unifying influence of one or more major towns or a major city.[7] Each of the state and territory capital cities as well as the national capital, Canberra, forms its own statistical division, and the population of the statistical division is the figure most often quoted for that city's population. However, this definition has largely become obsolete with the conurbation of several statistical divisions into a larger metropolitan areas. For example, the City of Brisbane, the capital city of Queensland, has long since become contiguous with the surrounding cities of Ipswich, Logan City, Redland City, Redcliffe, Pine Rivers, Caboolture and the Gold Coast. This conurbation is common to all the major metropolitan areas of Australia, which include all the capital cities and many major regional centres such as the Gold Coast, Newcastle, Wollongong, Geelong, Townsville, Cairns and Toowoomba


Brazil

In Brazil, metropolitan areas are called 'metropolitan regions'. Each State defines its own legislation for the creation, definition and organization of a metropolitan region. The creation of a metropolitan region is not intended for any statistical purpose, although the Brazilian Institute of Geography and Statistics uses them in its reports. Their main purpose is to allow for a better management of public policies of common interest to all cities involved. They don't have political, electoral or jurisdictional power whatsoever, so citizens living in a metropolitan region do not elect representatives for them. There are currently 52 metropolitan regions, distributed in all major regions of the country, the largest of them being the metropolitan area of São Paulo with over 20,900,000 inhabitants, making it the largest metropolitan area in the southern hemisphere, and the 7th largest in the world.
.

[New Brisavoine]

Originally Posted by yuriandrade

That's why you are usually not welcomed.

Not welcomed? I don't think so judging from the messages of sympathy I've received.

Your ad-hominem here is what's not welcomed.

[KevinFromTexas]

Debate the data and critique the sources, but don't use trolling as a way to make your point. This goes for everyone. Don't trust a source? Give the reason why and offer a better one. Don't agree with the outcome even when it's a universally accepted source? Then state your reasons for not adopting that method.

[Minato Ku]

Originally Posted by yuriandrade

BTW, there are several criteria to establish a metro area, and commute patterns being just one of them. For instance, from Wikipedia:

So you are comparing apples with oranges.
This is what several members told you.

A metropolitan area can't be calculated using subjective criterias depending the city or your view.
It is even worse if the main criteria is a view on satellite or based on a surface because you may overestimate size of the metropolitan area in dense regions.
Imagine Semarang with over 30 million inhabitants because there are over 30 million inhabitants on 30,000 km² around Semarang.

Is Semarang the heart of a metropolitan area more populated than Mexico New York, Sao Paulo, Los Angeles? The answer is no but Semarang is located in a densily populated region.

This is why commute rate should be the main criteria, using relatively high threshold.
I think that under 20%, it is a too low.

Using commute threshold, we observe that more a city is located in a dense region, more its metropolitan area covers a small surface.
Why? Because the proximity with other large cities loosens its direct influence over a large area.
In the opposite way, more a city is located in a region with a low density and more its metropolitan area will cover a large surface because the closest large cities are far away.

Note that for a city in a sparsely populated region, the large surface of the metropolitan area does not really matter because it has no significant effect on the population figure.

[Yuri]

Minato, your post is problematic because you are touching subjects that's not even the point of this thread (like the London vs Paris in Brisavoine's attempt to hijack it). London is not even featuring in my list so far.

American forumers are complaining because I downsized US metro areas a little bit, excluding exurb counties which were completely rural back in 1940, 1950, distorting growth rates in those earlier decades as they usually posted negative growth due rural exodus. About international areas, precisely the ones not using commute patterns in their definition, there were no complains. So we cannot talk about São Paulo and New York metro areas, compare them, just because they use distinct definitions (political/statistical)?

Your dense surroundings theory, I find it extremely weak as you refuse the idea metro areas could be polycentric. I mean, I highly doubt 25% of people in Dortmund go to work in Essen, and they form together a metro area. Comparing a random small town in Semarang, with metro areas centered around megacities like Mexico City, São Paulo, New York, Los Angeles also makes no sense at all.

On the lack of uniform definitions internationally, using fix-areas around megacities is much more useful than random commute patterns, completely ignoring polycentric metro areas.

[Crawford]

Originally Posted by yuriandrade

American forumers are complaining because I downsized US metro areas a little bit, excluding exurb counties which were completely rural back in 1940, 1950, distorting growth rates in those earlier decades as they usually posted negative growth due rural exodus.

No, you're completing missing the point.

You're welcome to use any metric you want. Just use a metric. You aren't using any; you're making up fantasy boundaries for each city.

There are many possible metrics out there- MSA, CSA, urban area, etc. Pick one. You've picked none and just decided to create fake population numbers.

[New Brisavoine]

Originally Posted by yuriandrade

I mean, I highly doubt 25% of people in Dortmund go to work in Essen, and they form together a metro area.

Dortmund and Essen are physically linked by continuous urbanization, so the point is moot. As long as they are part of the same urbanized area, they necessarily also are part of the same metropolitan area. That's the very definition of metropolitan areas (an urban core, i.e. an urbanized area, in this case both Essen and Dortmund belong to it, surrounded by towns and villages not physically attached but attracted enough to it by a certain commuter threshold).

Originally Posted by yuriandrade

On the lack of uniform definitions internationally, using fix-areas around megacities is much more useful than random commute patterns, completely ignoring polycentric metro areas.

The only polycentric metro areas are those either whose nucleus is polycentric (the Ruhr for example, or Lille-Roubaix-Tourcoing in France, or Greater Manchester in the UK) or whose nucleus is monocentric but which manage to attract another urban nucleus large enough via the commuter flows (these are very rare... an enormous city like Paris only manages to attract detached urban areas like Meaux, 73,000 inh., and fails to attracts urban areas like Chartres, 89,000 inh., at least with the 40% threshold).

Just because two cities are located close to each other doesn't mean they belong to the same metro area if they are both physically detached (no urban continuity) and without significant commuter flows between them. So you cannot claim that Santos belongs to the São Paulo metro area (the Brazilian statistical office in its recent paper about Brazilian metro areas identified them as two distinct metro areas by the way), or that Reading belongs to the London metro area, or that Philadelphia belongs to the New York metro area just because they happen to be located close to each other.

[Yuri]

Originally Posted by New Brisavoine

Dortmund and Essen are physically linked by continuous urbanization, so the point is moot. As long as they are part of the same urbanized area, they necessarily also are part of the same metropolitan area. That's the very definition of metropolitan areas (an urban core, i.e. an urbanized area, in this case both Essen and Dortmund belong to it, surrounded by towns and villages not physically attached but attracted enough to it by a certain commuter threshold).

The only polycentric metro areas are those either whose nucleus is polycentric (the Ruhr for example, or Lille-Roubaix-Tourcoing in France, or Greater Manchester in the UK) or whose nucleus is monocentric but which manage to attract another urban nucleus large enough via the commuter flows (these are very rare... an enormous city like Paris only manages to attract detached urban areas like Meaux, 73,000 inh., and fails to attracts urban areas like Chartres, 89,000 inh., at least with the 40% threshold).

Just because two cities are located close to each other doesn't mean they belong to the same metro area if they are both physically detached (no urban continuity) and without significant commuter flows between them. So you cannot claim that Santos belongs to the São Paulo metro area (the Brazilian statistical office in its recent paper about Brazilian metro areas identified them as two distinct metro areas by the way), or that Reading belongs to the London metro area, or that Philadelphia belongs to the New York metro area just because they happen to be located close to each other.

About your continuous urbanisation theory, it's quite problematic as well. I decided to look closer on GE, and the gaps between Essen and Dortmund are very similar with the ones between London and Slough or Watford/St. Albans or Leatherhead, areas, if I'm not mistaken, you don't regard as part of London urban area. In fact, even New York and Philadelphia are linked by suburbans ribbons. On the other hand, Bridgeport/New York and Los Angeles/San Bernardino are regarded as separated urban and metro areas by the US Census Bureau.

About São Paulo, some points:

- I'm using the official definition in this thread, not mine I've used elsewhere;

- Secondly, if you are familiar with IBGE's paper, you should know Jundiaí area would be included in São Paulo metro area using the formula they've established. They decided to leave it out just because people tend to see it as an independent city. That would add 700,000 people, not to mention Jundiaí and São Paulo are pretty much linked by continuous urbanisation these days and are linked by CPTM (São Paulo metropolitan train);

- Finally, I must confess I'm not that good with math formulas, so I can't tell whether IBGE's definitions are more strict or liberal than the ones abroad.

Originally Posted by Crawford

No, you're completing missing the point.

You're welcome to use any metric you want. Just use a metric. You aren't using any; you're making up fantasy boundaries for each city.

There are many possible metrics out there- MSA, CSA, urban area, etc. Pick one. You've picked none and just decided to create fake population numbers.

Those metrics mentioned by you will work throughout this seven decade period?

[New Brisavoine]

Originally Posted by yuriandrade

I decided to look closer on GE, and the gaps between Essen and Dortmund are very similar with the ones between London and Slough or Watford/St. Albans or Leatherhead, areas, if I'm not mistaken, you don't regard as part of London urban area.

It's not me who regards or doesn't regard anything as part of the London urban area, it's ONS (the British statistical office). ONS uses the standard international definition recommended by the United Nations to define their built-up areas (BUA), i.e. no gap larger than 200 meters.

Slough is not inside the London BUA, it's a separate BUA, but Watford, St Albans, and Leatherhead are all inside the London BUA, so inside the London urban area, and therefore also inside the London metro area.

Regarding Slough, the gap is between West Drayton and Richings Park, and between Stanwell Moor (which belongs to the London BUA) and the industrial park of Poyle (which belongs to the Slough BUA).

Originally Posted by yuriandrade

On the other hand, Bridgeport/New York and Los Angeles/San Bernardino are regarded as separated urban and metro areas by the US Census Bureau.

The US Census Bureau defines Urbanized Areas which don't follow the international definition of urban areas. This has been criticized, in particular because it artificially treats San Francisco and San Jose as two separate urbanized areas, and Los Angeles and some of its suburbs as two separate urbanized areas. Obviously the UN definition used everywhere else would never consider these as separate urban areas.

[New Brisavoine]

A map posted by vinterriket on SSC (from the UK 2011 census I suppose) which confirms what I already stated in this thread: the pull of London in England is extremely limited, as large enough cities and centers of employment block the extent of its commuter belt very quickly after the borders of Greater London. The map matches my own calculations from the UK 2011 census: the London metro area spreads over a rather limited land area.

[Yuri]

If they annex the blue/green areas to Greater London (and there are talks in that direction), the yellow/orange ones, would probably turn blue/green. In the end, it's just a matter of where to set the official border.

[New Brisavoine]

Originally Posted by yuriandrade

If they annex the blue/green areas to Greater London (and there are talks in that direction), the yellow/orange ones, would probably turn blue/green. In the end, it's just a matter of where to set the official border.

Not true at all, because the blue areas are dormitory communities and not centers of employments, so they attract virtually no one. The color of the further areas would therefore not change much. It's not just something I'm saying by guessing it, but it's something I actually measured in the 2011 census when I made the calculations for the London metro area. Each new iteration added very few new MSOAs to the metro area.

[Yuri]

Originally Posted by New Brisavoine

Not true at all, because the blue areas are dormitory communities and not centers of employments, so they attract virtually no one. The color of the further areas would therefore not change much. It's not just something I'm saying by guessing it, but it's something I actually measured in the 2011 census when I made the calculations for the London metro area. Each new iteration added very few new MSOAs to the metro area.

Firstly, I do think we would see some colour changes all over the board. Also, the blue areas would bring London to 2,000km2 only. They could go much further, up to 5,000km2 for instance, including, therefore, some centres of emplyment. Then, we would see significant changes on the map.

In the end of day, commute rates might be as arbitrary as administrative definitions when it comes to metropolitan areas definition.