Urbanization in Europe and East Asia

After writing this post about the urban layout of Germany and high-speed rail, I got interested in city size in Europe and Asia more generally. East Asians live in much larger cities than Europeans, and I’m not just talking about Tokyo and Seoul. I was vaguely aware this was the case, but underestimated the size of this effect. By modern definitions of metropolitan areas as those with at least 1 million people, Europe is not majority-urban.


The definitions are hard to harmonize – there are international lists but I don’t fully trust them, many just collating national statistics (like the UN’s) and others making some judgments I am uncertain about, like CityPopulation. I believe the effect I’m discussing is robust to changes in definitions, but its magnitude may be different. I use the following definitions:

  • In Japan, I’m using the major metropolitan area. This is the part I’m least certain about, hence a subsidiary definition based on smaller metropolitan employment areas (MEAs).
  • In South Korea, there are no definitions of a metro area. Seoul is defined as Seoul + Gyeonggi + Incheon, the other cities use city limits.
  • In Taiwan, as in Korea, cities are defined by city limits, except that the capital region consists of Taipei + New Taipei + Taoyuan.
  • In France, a metropolitan area is an aire urbaine.
  • In Germany, a metropolitan area is either the Rhine-Ruhr region, or a Verkehrsverbund.
  • In Italy and Spain, a metropolitan area is cobbled from NUTS-3 regions in 2021 definitions; this can also be used in Germany as a secondary check, but not in France, as French NUTS-3 regions are departments, which aren’t granular enough. Population figures can be found here.



Using the MMA definition, slightly more than half the population lives in the metro areas of Tokyo, Osaka, and Nagoya. Nagoya goes barely over the 50% mark, so the median resident lives either in Nagoya (9.4 million people) or Fukuoka (5.5 million). 56% of the population lives in these four metropolitan areas, 64% lives in metro areas of at least 2 million, 70% lives in metro areas of at least 1 million.

The MEA definition splits polycentric regions like Osaka-Kyoto-Kobe and Fukuoka-Kitakyushu, and also thinks Nagoya is much smaller. On that definition, the median Japanese resident lives in Sapporo (2.4 million), and overall 51% live in metro areas of at least 2 million and 61% in metro area of at least 1 million.

South Korea

The metropolitan area of Seoul has 26.2 million people, which is just a hair over half the population of the country. But it’s so close it’s obligatory to also include the next largest region, which is Busan, population 3.4 million. Together, they have 57% of total population. Daegu adds 2.5 million, and with it 62% of the country’s population lives in cities of at least 2 million. Down to 1 million, the share rises to 70%.


Taipei, New Taipei, and Taoyuan collectively have 8.9 million people, or 38% of total population; Taoyuan is somewhat independent, but Taipei and New Taipei are very closely integrated. The median Taiwanese lives in one of the two second cities, Kaohsiung and Taichung, each with around 2.8 million people; together, 61% live in one of these five municipalities. The next city, Tainan, has 1.8 million people, and including it raises the proportion to 69%.


The Paris region has 20% of the country’s population, and the urbanity of the population craters afterward. Only one more metro area has more than 2 million, Lyon; together, the two regions have 23% of national population. Five more metro areas have between 1 and 2 million people; together, these seven regions have 33% of total population. Even allowing metro areas down to 500,000 only increases the metropolitan share to 44%. The median is either Valenciennes or Le Mans, populations 369,000 and 347,000 respectively.


By any definition, we lump the Rhine-Ruhr into one region of 11,430,361 people, based on the Metropolregion definition (see data here) – usually the German Metropolregionen are way too loose, but in this polycentric region the definition agrees with regional rail extent and with a number of other metro area population lists. In theory this is Germany’s only megacity, but because there are at least 7 independent centers, it doesn’t feel like it’s another London or Paris. Still, it’s 14% of total population.

Using the rather loose Verkehrsverbund limits (still tighter than the Metropolregionen!), we add Frankfurt (6.7 million), Berlin-Brandenburg (6.2 million), Hamburg (3.5 million), Rhine-Neckar (3 million), Munich (2.9 million), Nuremberg (2.8 million), Stuttgart (2.4 million), Bremen (2.2 million), Leipzig-Halle (1.7 million), Karlsruhe (1.4 million), Hanover (1.2 million), Dresden (0.8 million). As a note of caution, some of these regions are drawn very loosely, like Nuremberg, Bremen, and Frankfurt; Hanover and Munich are a lot tighter, Hanover corresponding to a county and Munich to the S-Bahn extent.

On these very loose definitions, the 2+ million regions still only sum up to 49% of the population, and the 1+ million regions to 55%, the median German living in Leipzig-Halle.


Italy has many distinct definitions of metro area, some disagreeing on Milan by factors approaching 2. Eurostat, using provinces as its level of analysis, has metropolitan area for Milan (4.4 million), Rome (4.3 million), Naples (3.1 million), Turin (2.3 million), Brescia (1.3 million), Palermo (1.3 million), Bari (1.3 million), Bergamo (1.1 million), Catania (1.1 million), Bologna (1 million), Florence (1 million), Padua (0.9 million), Verona (0.9 million), Venice (0.9 million), Genoa (0.8 million), Perugia (0.7 million), Messina (0.6 million), Taranto (0.6 million), Reggio Emilia (0.5 million), Parma (0.5 million), Cagliari (0.4 million), Prato (0.3 million).

The four main metro areas comprise 23% of Italy’s population. Going down to 1 million raises the proportion to 37%. These regions combined have only 48% of Italy’s population. This does not mean Italy is majority-rural – the remaining provinces are full of cities in their own right, some larger than Prato or Cagliari – but it’s hard to detect that level of granularity.


In Spain, as in Italy, the division isn’t too granular, at the level of a province. We have metro areas for Madrid (6.7 million), Barcelona (5.6 million), Valencia (2.5 million), Seville (2 million), Alicante (1.8 million), Málaga (1.7 million), Murcia (1.4 million), Cádiz (1.2 million), Biscay (1.1 million), A Coruña (1.1 million), Asturias (1 million), Zaragoza (1 million), Vigo (0.9 million), Tenerife (0.9 million), Granada (0.9 million), Mallorca (0.9 million), Gran Canaria (0.9 million), Córdoba (0.8 million), Gipuzkoa (0.7 million), Navarre (0.6 million), Cantabria (0.6 million), Valladolid (0.5 million), Álava (0.3 million).

The two main metro areas are together 26% of Spain’s population; the next two, down to 2 million people, add up to 35%; the next ones going down to 1 million go up to 55%, the median Spaniard living in a metropolitan province of 1.1 million.


  1. barbarian2000

    Brazil (pop. 211.755.692), going by legal boundaries for each metropolitan area with a population of over 1 million and IBGE population estimates for 2020:

    São Paulo (21.893.842), Rio de Janeiro (13.132.348), Belo Horizonte (6.006.091), Brasília (4.693.763), Porto Alegre (4.363.027), Fortaleza (4.137.561), Recife (4.103.780), Salvador (3.709.555), Curitiba (3.693.891), Campinas (3.304.158), São José dos Campos (2.928.345), Manaus (2.722.014), Goiânia (2.648.226), Belém (2.529.178), Sorocaba (2.166.860), Vitória (2.006.486), Santos (1.881.706), Ribeirão Preto (1.738.000), Natal (1.631.016), São Luís (1.590.852), Piracicaba (1.508.439), Joinville (1.438.561), Maceió (1.346.973), João Pessoa (1.290.223), Teresina (1.232.736), Florianópolis (1.229.335), Londrina (1.194.960), Cuiabá (1.049.312)

    This means that the two megacities comprise 16.5% of the total national population, cities with a population of 2 million comprise 40% of the population and cities with a population of 1 million or more comprise 47.8% of the population. However, like Italy, this doesn’t indicate rurality, and there are a lot of cities that have gone unaccounted for here. It’s also worth noting that most states have a massively dominant state capital, doubly so if that state capital is a large city in its own right: Rio de Janeiro city has 13 million people while Rio de Janeiro state has a total population of 17 million!

  2. michaelrjames

    I don’t have a reference, nor a date, for this data (below) but it confirms the perception that Australia is the most urbanised nation though this measures the concentration rather than urbanisation per se. I don’t think China would be anywhere near the top on any criteria yet, even if on some definitions Shanghai is the world’s largest single city and at 25m is almost the population of Australia. (“Tokyo” as a single city is a construct. It may be the world largest urbanised area.)

    Countries where big cities dominate:
    [share of population living in the five largest cities]

    Australia Fraction of national population: 2019 data
    Biggest city: 20.4%
    Top two cities: 40%
    Top 5 cities: 63%
    Top 10 cities: 70%
    8 state & territory capitals: 66%

    • Alon Levy

      Korea is a lot more than 39%… You have to lump Incheon and Gyeonggi together with Seoul; there are no officially-defined metro areas in Korea, but the Seoul subway extends into Gyeonggi and even meets the Incheon Subway, and the population of Gyeonggi clusters in TOD zones near these subway lines to Seoul.

      Australia is a country of big cities, yes. To some extent so are the US and Canada. This really is an issue of Europe being deeply uncomfortable around big cities – in the French discourse, Parisians aren’t fully French and there’s toleration for rural drivers who come to Paris to throw rocks and graffiti “Juden” and “Rothschild.”

      • Brendan Dawe

        looks like Canada comes in between 47% and 52% depending on which (if any) peripheral metropolitan areas you want to add to the 6 million+ Census Metropolitan Areas, such as Oshawa or Abbotsford

        • seangillis78

          Here’s my numbers for Canada (Population: 37.59 million) using just the main CMA population estimates from Stats Canada.
          Toronto (about 6.4 million) is 17% of the population.
          The ‘Big 3’ of Vancouver, Montreal and Toronto are 36% of the population.
          Metro areas over 1 million people (add in Calgary, Edmonton and Ottawa-Gatineau) are 48% of the population.
          ‘Large’ cities (add in Winnipeg and Quebec, each about 800,000) and you have 52% of the population. These cities are large enough to have a pro hockey team or to be working hard to get one. Seems like a reasonable cut-off for Canada.

          To throw in the mix, Canada’s ‘Big City Mayors’ caucus includes my hometown of Halifax, metro 440,000. So for political purposes a lot of small cities are considered big in Canada. This could be to add perspectives/ clout from different regions, as this is a working group and a political advocacy group on urban issues.

          I think the most interesting question is where does Toronto’s metropolitan area end these days? Along Lake Ontario, the urbanization is continuous from (at least) Oshawa in the east to Hamilton in the west. Hamilton and Oshawa are both separate Census Metropolitan Areas (CMAs) but are plugged into the GO commuter rail network centered on Downtown Toronto. Adding those two would bring ‘metro’ Toronto to about 7.7 million people. I don’t know how this metro area would compare to metro definitions in other countries, which seem to be looser than metropolitan definitions in Canada.

          Then you get into the question of the ‘Greater Golden Horseshoe’: Toronto, Hamilton, Oshawa + Barrie to the north + Guelph, Kitchener, Waterloo, Brampton to the west + St. Catherine’s, Niagara Falls and Welland to the south. Some of those ‘secondary’ cities would be the largest city in a few provinces. Total population: just over 9 million. Some people keep going and include London. Regardless, this is where federal elections are won-lost.

          So we certainly have a large conurbation, that might compare to Rhine-Ruhr, although the distances are bigger and the cities aren’t nearly as interconnected with transit and rail. This is kind of the Canadian experience in a nutshell – we take up a lot of space. Considering the low densities (by global urban standards) it is interesting that we have reasonable transit mode splits in our bigger cities. Density is important, but it isn’t destiny.

          • jcranmer

            The CMAs seem to correlate quite well with the MSAs in the US. Boston, Providence, and Worcester are each their own MSA, but are combined into a single CSA, and that level of distinction feels about right for Hamilton versus Toronto.

            I don’t know enough about international metro area comparisons to know what the right level is compared to other countries, but I believe Alon has generally used MSA comparisons in the past.

          • seangillis78

            Boston-Providence seems to be a good comparison to Toronto-Hamilton. I’m not super familiar with how the US compiles the MSA/ CSA data. I do know that the Stats Can CMA data generally looks/ feels right for where a city and it’s suburbs end. Another interesting product that Stats Can uses is a population centre – essentially the continuous urbanized areas within a CMA. A good product, as the definition is very tight.

            One thing that is interesting to consider is which countries have growing cities and which don’t. With it’s high level of immigration Canada’s cities in general are growing quit quickly, including the big ones. Toronto CMA gained about 7% population in 4 years, and Ottawa grew quicker. Just two examples.

      • jcranmer

        Looking at Wikipedia’s population lists for CSAs and MSAs, if you go by CSA, the median person in the US is in a city of around St. Louis’s size, and by MSA, it’s more like Kansas City. In other words, about 2-3 million people, or 20-30th largest city. I’m actually kind of surprised that the two metrics don’t actually change the result all that much.

        Intriguingly, this means that the median person in the US lives in a city that is about the division point between “no [rail] transit whatsoever” and “farcical attempt at a [rail] transit system”

        • Alon Levy

          Yeah, Americans live in larger metro areas, just with hilariously weak centers. Dallas has about the same CBD job count as Vancouver or Lyon.

        • DMac

          I has a look at the UK data using the OECDs Functional Urban Area with 2001 data (latest I could find on Wiki). By that definition 23% (13 million people) of the UK population lives in the London FUA – which is huge – extending a solid 20/30km out from Greater London. The median Brit lives in the Nottingham/Derby metro area with a population of 1.5 million. Two-thirds of the population live in areas with more than 500k population.

          About 40% of the population live in areas with 2+ million people and 60% in areas with 1+ million people. This makes it pretty similar to Germany I think.

          • Alon Levy

            Yeah, I was trying to avoid either using 2001 data or relying on weird-with-the-UK Eurostat data. The Eurostat data is there, it just has weird definitions of Greater Manchester and such, because as in France and unlike in Germany, Britain’s NUTS-3 areas aren’t granular enough.

      • michaelrjames

        I think that is too much of a generalisation. There were no mega-cities in the world until quite recently, and London & Paris were the biggest cities in the world about a century ago. As a fraction of their nation’s population they are quite high; Paris is ≈20%. Both are still growing. But anyway once a city passes some point, certainly 10 million, I’m unconvinced there is much further difference for anyone to feel “deeply uncomfortable” with. My own theory would be the same as the reasons put for why the modern world evolved in Europe rather than in China (even though China had a head start on many technologies that if they had pursued would have led to industrialisation). This is the theory based on geography elucidated by Graeme Lang and reviewed by Jared Diamond (Nature, Vol 391, Jan. 1998). Diamond writes:

        “Lang sees the autonomous universities on the onion’s outer skin as springing from an underlying layer of European political fragmentation. Mediaeval Europe was still divided into a thousand independent statelets, whereas China was already unified in 221 BC. So it proved impossible to suppress critical thinking for long in Europe: a thinker persecuted in one statelet could (and often did) merely walk into the next. To take just one example, the astronomer Johann Kepler was always able to keep one step ahead of the authorities by moving between Tubingen, Graz, Prague, Linz and Silesia.”

        That’s the proximate cause while geography was the ultimate cause, from Wittfogel’s “hydraulic hypothesis”. I won’t go into the geography but the tribal diversity in Europe persists into the modern era, and in turn this resists agglomeration. Fitting this theory is that the two biggest unified nations, and themselves the biggest empire builders, Britain and France have the two biggest cities. By comparison it is only 1.5 centuries since Germany unified into a nation, and it has a federated structure with each state retaining a major “German” city. The same factor operates in the US and in fact until quite recently the US was not majority urbanised, and remains less urbanised than many other nations.

          • michaelrjames

            That’s a different definition of urban. One really needs a plot of “%total versus town size” to see the trend/inflection point and it wasn’t 1920.

          • adirondacker12800

            The Census Bureau seems to think so. Pity they didn’t consult Your Omniscience in 1921 or whenever they came up with that percentage.

        • barbarian2000

          Geographical determinism, including that of Diamond, has been repeatedly proven wrong again and again. Political censorship in Europe was very common, notably by the Catholic church, to the point that you yourself noted that Kepler had to constantly keep moving around to avoid it. Meanwhile, central authorities in China usually had very limited presence in most of the country, especially in distant and mountainous regions like Yunnan. India also had just as much political fragmentation as Europe until the 16th-17th Century and Indian polities never expanded far overseas. The South Asian Subcontinent is also far more diverse than Europe and has much larger cities, which shows an important counter-example to the “hydraulic hypothesis”

          • michaelrjames

            “repeatedly proven wrong again and again”

            That’s not true because one cannot “prove or disprove” such a ‘hypothesis’. Not a hypothesis if it can’t be disproven, more a speculation. That’s why it will remain an endless unresolved argument.

            I think you are confusing several things. I’m sure geographical determinism is too simple to be all the mechanism but your point about the Catholic church is exactly the point made by Lang et al. The point about China is that after 221BC (at the latest) the Han dominated everywhere. Jcranmer’s point is a non-issue. It may seem to the contrary to realise that China actually has more minority indigenous ethnic groups than almost any other country–about 50–but they are minorities and live in relatively tiny geocaches, mostly mountainous or inaccessible valleys etc, and are careful not to create friction with the dominant Han culture. (Because we can see in Tibet and Xin Jiang what happens if they do.) The East Asians haven’t tolerated diversity within their own cultures, or external influences even less, which we see in the extreme groupism in Japan (who have their own single indigenous minority–the Ainu–who have barely persisted; ditto the same group in Taiwan).

            I agree about South Asia but that just means there are other reasons for the same outcome (ie. failing to modernise despite promising early start). In fact maybe India went too far in the other direction …

            And anyway, my point was about Alon’s thing about city size. India may have big cities (today) but as a fraction of the nation they don’t compare to Paris or London, and India remains largely non-urban (though rapidly urbanising but decades behind China). Clearly one cannot dismiss the unification under the Raj when looking at big Indian cities today.
            Schengen and the EU’s FOM might eventually alter the factors that have limited the growth of really big cities. Paradoxically that is a lot of what Brexit is about, or at least the perception that eastern Europeans are overrunning the joint and taking their jobs etc.

        • jcranmer

          > China was already unified in 221 BC

          And China shattered into the Three Kingdoms in 220 AD, not reunifying until the Sui dynasty in 581. Again shattering at the end of the Tang into the Five Dynasties and Ten Kingdoms in 907, not being reunified until the Mongols conquered them and became the Yuan dynasty in 1279. Again shattering at the end of the Qing dynasty in 1912 and not really reunifying until the end of the Chinese Civil War in 1949. That means there’s several centuries in there where even the Chinese admit there’s several countries vying for domination of the Chinese region.

          • Frederick

            So, in a span of 2200 years (from 221 BC to now), China was unified 70% of the time, according to your math.

            2200 years, and 70%.
            Is there any region in the world which has unified for a longer time?

          • jcranmer

            The Middle East? The Ottomans, the Ilkhanate, the caliphates, various iterations of the Persian Empire, the Babylonian Empire, the Akkadian Empire…

            If you’re going by percentage spent unified, Germany does as well as China, being unified by the Saxon Crusades in the 700s, remaining unified under the Carolingian Empire, its early divisions, reincorporation into the Holy Roman Empire. The Holy Roman Empire breaks down as a unitary state somewhere in the 1500s or 1600s (depending on how you want to count), but the German Empire reunifies in 1871, and some form of Germany lives on until the present day. So 1300 years, 300 of which are disunity.

          • michaelrjames

            You continue to miss the point–which is that in no way did that nominal loss of unification in China promote serious diversification. Everywhere was still dominated by Han Chinese and anything else was suppressed. The power of this was demonstrated by what was a reverse takeover by the Mongols–they totally embraced the Han culture to the point there was no meaningful difference.

          • Alon Levy

            2 things.

            1. No, the Mongols did not embrace Han culture. The Manchus did but that was many centuries later.
            2. “Han Chinese” as an identity is to a large extent “people who acculturated to China.” It’s not genetic, Southern Chinese people look different from Northern Chinese ones. As lowland empires expanded, people on the margins either became Chinese (or Vietnamese, or Thai, or Burmese) or retreated farther up the mountains or across borders. This is also truer in Europe than Europeans would like to admit – France underwent a couple episodes of genocide and ethnic cleansing to get people to be Catholics in good standing and then under the Third Republic engaged in cultural extermination of Occitania.

          • michaelrjames

            Sorry, but must disagree on both points. I don’t know if you are arguing on abstruse technical points but the Yuan dynasty continued almost all significant Chinese traditions. The very fact that Kublai Khan created it and named it a dynasty in Chinese tradition says it all really.
            As for genetics, it is simply not correct as any analysis reveals:

            A 2018 research found that Han Chinese are clearly genetically distinguishable from Yamato Japanese and Koreans, and internally the different Han Chinese subgroups are genetically closer to each other than any of them are to Koreans and Japanese.[210] …. Comparisons between the Y chromosome SNP and MtDNA of modern Northern Han Chinese and 3,000 year old Hengbei ancient samples from China’s Central Plains show they are extremely similar to each other and show continuity between ancient Chinese of Hengbei and current Northern Han Chinese. This showed that already 3,000 years ago the current northern Han Chinese genetic structure was already formed.[212] The reference population for the Chinese used in Geno 2.0 Next Generation is 81% Eastern Asia, 2% Finland and Northern Siberia, 8% Central Asia, and 7% Southeast Asia & Oceania.[213]

            It’s actually quite remarkable for such a large group and spanning such a large geography. Perhaps you are mislead by appearances which can reflect very minor genetic differences, or like comparing blondes versus brunettes within the Nordics for example.

          • Alon Levy

            Yes, the Yuan dynasty continued the old Chinese traditions, but the Mongols viewed themselves as a separate and higher caste and did not personally Sinicize; the Manchus began that way too but Sinicized to a greater extent, to the point that the Manchu language went extinct.

          • michaelrjames

            Well, ok, it’s a matter of degree. But also my focus was on what the Mongol leadership did to China not really those Mongol leaders.

          • Frederick

            1. It is true that the Mongols have never properly Sinicized themselves. This was why their Yuan Dynasty didn’t even last a century (1271-1368). The Mongolian way is, thus, an exception instead of the norm. (On the other hand, Manchus very soon Sinicized after they conquered China.)
            2. It is correct that “Han Chinese” is to a large extent “people who acculturated to China”. What I want to add is that, historically “the culture of China” is to a large extent “the written Chinese language”. Even with modern technology, it is extremely difficult to learn the Chinese language. Hence, when a group of people decided to sinicize, they must spend a lot of time in learning the Chinese language, and as a result they would have much less time practicing archery or other things relevant to their own culture. Many Mongols never bothered to learn Chinese; instead, they asked a Tibetan Monk to invent a script for the Mongolian language.
            3. I don’t know why genetics is brought up here. Haplogroup classification is science and should be science-only. Mixing it up with social constructs like “nation” or “ethnicity” brings us dangerously close to the territory of Adolf.

          • michaelrjames


            I was going to agree to your 1 & 2, but actually your statement “Han Chinese” is to a large extent “people who acculturated to China” isn’t correct. It represents the people–and essentially a single people–who dominated China. Whoever else had occupied the land prior to them (ancestors of some of those 50 ethnic minorities), who were presumably also acculturated to the land, were pushed out to occupy minor niches today. And about point 3, the genetics tells us that that is what happened and cannot be fudged or tampered with by historians with their own agenda. (For example, the Japanese have a lot of trouble sustaining earlier concepts of themselves as the original master race in the face of the genetic tree of East Asia.) The Han did this with very little miscegenation. Over the past two decades (since the Human Genome Project) genetics has clarified a lot of previously murky history, eg. the human movements in the Pacific, or the two separate waves of out-of-Africa, or the miscegenation of H. sapiens and neanderthals, etc.

            It relates to my comment about the diversity argument (whether caused by geography or whatever). I was this morning reflecting on this w.r.t. the just-released documentary White Noise about the US alt-right movement. Because their desire for racial (genetic) purity is more likely to lead to decline rather than all-conquering dominance as they believe. Eugenics has a deserved terrible reputation but that doesn’t mean we should ignore genetics. I’d say that the Han have paid a price for their purity–while producing one of the world’s oldest and most sophisticated societies and dominating east Asia, it also probably stopped them taking the next steps in human evolution (the enlightenment, industrial revolution etc). Just to be clear, I’m not saying it is the genetics which held them back but that it was a proxy marker for a too uniform society and culture. Hybrid vigour in both society and genetics! And to put it into today’s political context, it really does mean that a multicultural, multi-ethnic society like the US will be stronger.

      • michaelrjames

        “in the French discourse, Parisians aren’t fully French”

        I think that that is a slightly odd version of the perceived singularity of Paris within France. One can, and many historians have, make the case for the inverse. After all, the Ile de France doesn’t mean “an area that is an island within France” but almost the opposite. The area was once the only part of the world known as France until much later national coherence, when the emerging nation (until quite recently a collection of different tribes with different languages and cultures) was forced by the centre-based rulers to be known by ‘their’ name France. This has been explicitly set out by historian Marc Bloch in his book “The Ile de France” (my copy is the first eng trs in 1971 of the 1913 book, but coincidence, a new edition was just released on 12 Oct 2020, ISBN-13: 978-1138855946; and no, the 192 page book is not worth the $53. I got mine for a few dollars secondhand).

        My experience, though I’ll admit from the p.o.v. of a Parisian, but also I would say reinforced by cultural normes one sees in media, movies, plays etc, is that all French consider themselves part Parisian. Again, until relatively recently (post-60s) many pregnant women in the provinces journeyed to Paris for their delivery; it’s why so many French actually have “Paris” on their birth certificates (at least boomers and earlier) even though many never actually lived there. Of course the city remains the place many will go for higher education and subsequent career, with many returning to the provinces in midlife. Today there remains a norme that you go to Paris for certain things, and as you have discussed a lot, is reflected in the TGV network radially centred on Paris. France has partly resolved this phenomenon with its la France profonde philosophy: that even when you are a fully integrated Parisian you retain your provincial roots.

        • Alon Levy

          Yeah, elite culture in France is centered on Paris, but still with too much respect to La Mort Aux Juifs or whichever other village they imagine is Real France (that was an actual hamlet, the name only got changed in 2015 after much Jewish uproar). Vernacular culture is different. There’s an entire identity called “periurban,” much like exurban America, i.e. low-density, basically metropolitan but hating the idea of cities and diversity; unlike in Los Angeles, French exurbs are not racially diverse, and their populations descend from the same strain as R-Haine and often directly vote for R-Haine. Moreover, when those people riot, the elite responds with understanding, whereas when an immigrant, a person of color, or a religious minority commits any crime, the state demands law and order; on a day when more than 200 French people died of corona, the interior minister saw fit to say, re a single murder, “the job of the police is to intimidate those who would intimidate us.” But if the intimidation is done by white French natives and involves corona denial or a desire for more pollution, then it’s okay and the entirety of the French government surrendered.

          • seangillis78

            Curious about the comment on ‘exurban’ France and how you would define it physically. From my limited knowledge France does not appear to have exurbs like the North American variety: large areas of un-serviced (sewer and water are on-site via a well and septic system) residential development, with lot sizes near 0.5 ha or greater. A quick look at Google Maps will show them around cities big and small (northern parts of Greenwhich and Fairfield Counties in Connecticut for example). Driving on the roads feels like you might be in the woods or a farming village, but economically they depend entirely on the central city. Sometimes these developments are contiguous with denser, serviced suburbs.

          • Alon Levy

            Peri-urban identity in France is in many ways political, just as “white working-class” is a political rather than socioeconomic identity and in practice excludes people who are white and did not go to university but do not vote for populists or have the cultural habits the elite ascribes to people who vote for populists. To the extent it means anything more than “white people who like driving to the city to throw rocks,” it means living far from a train station, relying on cars for all travel, and probably living at lower density, perhaps in a very small town center or outside one entirely.

      • gc80

        Yes, I suspect that European antisemitism (and Balkan Islamophobia for that matter) is closely tied up with anti-urban sentiment.

  3. Frederick

    One must wonder, whether Asians live in larger cities only because they have more population (or equivalently less land to live in).

    Population density (number of people per square kilometer):
    Taiwan — 652
    South Korea — 517
    Japan — 333
    China — 146

    Netherlands — 422
    UK — 280
    Germany — 233
    Switzerland — 208
    Italy — 200
    France — 123
    Spain — 92.8

    • Alon Levy

      No, not really – England is about as dense as Honshu. The history of why London is smaller than New York or Seoul or Tokyo has to be about centuries of political resistance to dense development, from “gentlemen do not live on shelves” in the Victorian era to the designation of the Green Belt in the middle of the 20th century to present-day NIMBYism creating a massive housing shortage. Every step of the way, romantics viewed the city as less moral than rural living and relied on this moral hierarchy to justify decentralizing development. Some of this mirrored trends on the Continent, like NIMBYism and the Deep England mentality; some was unique to Britain, like the lower residential density of Central London.

      Of course other countries in Europe had their own unique examples of anti-metropolitan romanticism, it wasn’t just Britain. For example, Germany has strong political identities around city limits, whereas the UK is happier setting up metropolitan counties. And France has a combination of Britain-like hostility to growing the capital (though thankfully in the last few years Ile-de-France has become fairly YIMBY) and unique policies choking the growth of secondary centers, so that Lyon, Marseille, Bordeaux, and Toulouse are smaller than Birmingham and Manchester.

      • fjod

        The comparison to Honshu is a bad one – 95% of England could be built on without much trouble, while Honshu is well over half mountains.

        And the Green Belt isn’t inherently a bad thing – in fact, when combined with suburban intensification, it’s net good. The lack of suburban intensification is the problem. Without the green belt (or similar land use planning policies), so much of England is buildable on that you’d just have sprawl and we know that is not conducive for carbon emissions, biodiversity, public transport etc.

        • Nilo

          London’s Greenbelt is trash. Tons of rail stations and Underground stations with easy access to central London have been downzoned for over half a century due to it. Have you ever looked at how many people ride the Metropolitan line on a per/km basis? It’s awful, and it’s due in no small part to the Greenbelt.

          • fjod

            Yes I agree it’s badly implemented. But it is still a) a good high-level idea and b) net good, even considering its flaws, if combined with suburban intensification.

            The point about the Metropolitan line being hobbled by the Green Belt doesn’t fully follow IMO. The Metropolitan line is always going to have worse ridership per km because it is not like other tube lines in that it skips the inner suburbs. And there are like maximum six stations whose ridership is reduced significantly because of the Green Belt, and all but one of these are at the very far end of the line and get much better ridership than comparable stations on mainline routes. However I do agree that the areas around these stations (and even moreso round more egregious mainline examples like Woldingham or Brookmans Park) should be subject to intensification and green belt development.

          • michaelrjames

            I agree.

            I’m sure we’ve gone thru these issues on Alon’s blog several times. London’s Green Belt is 5,139 km2 and could easily fit in a lot of extra housing without significantly impacting its green function:

            In January 2016, the Adam Smith Institute released a paper showing that there was enough space in the green belt around London to build a million homes within ten minutes walk of existing train stations. The report estimated that in the Metropolitan Green Belt, there was around 20,000 hectares [200 sqkm; 3.89%] of land which was within 800m of a railway station, and a further 10,000 hectares of golf course land.

            The problem is that who would have confidence the government or developers or indeed people themselves if by a miracle they had a say, wouldn’t wreck it. The Brits wouldn’t build compact developments but most likely versions of Milton Keynes which, make no mistake, is just another version of exurban sprawl: MK has a mere 230k residents across 89km2 which happens to be exactly the size of inner Paris (105km2 minus the two bois=88km2), or exactly the population of Paris-15 (232,000 on 8.5km2) and MK is neither walkable nor really cyclable (but brownie points for segregated cycle network). Indeed there is a term for it: Green Sprawl. If they took half that proposed 200km2 Green Belt and built at half Parisian density there’s your million residents and it would be truly walkable to those rail stations and everything else.

            Worse, none of anything built would be affordable which is really the only possible justification for building on it at all. Any terrace housing let alone SFH or semi-detached would have astronomical prices, especially being on the tube/rail and in green London! Either that or they (and Alon) would build hi-rise like in east London docklands or southbank London and, in addition to being fugly, that would be equally unaffordable just as all that blingy crap is today; if forced to build some affordable housing it would be the token tower block of crap aesthetics, as usual. Of course proposing to build such stuff in the Green Belt really would light the fires of opposition.
            Then there would be trying to contain development once it started, once the Green Belt was broached.

            The Brits and the Anglosphere in general have shown themselves incapable of any sensible dense urban development between these extremes.

            Only one in 10 green belt homes ‘affordable’, says countryside group
            Campaign for Protection of Rural England says brownfield land offers more opportunities
            Fiona Harvey, 14 Oct 2019
            Only one in 10 homes built on green belt land in England in the last decade has been classed as affordable, undermining the case for the development of supposedly protected areas, according to countryside campaigners.
            The Campaign to Protect Rural England also found that only a third of the 266,000 future new homes currently proposed for green belt land around the country are likely to be affordable. In a report published on Monday, called Space to Breathe, the charity said most green belt developments were inefficient, with only 14 homes for each hectare instead of the 31 for each hectare that is the average on other land.

            One gets articles like the following which can only propose what the developers want: hi-rise stuff that are mostly sold to rich non-residents to bank their rotten money.

            High-density London: past, present and future
            Large numbers of people living on small amounts of land is neither wholly alien to the capital’s traditions nor a sure sign of social deprivation
            Dave Hill, 22 Jan 2017

            There are still fewer Inner Londoners than there were in 1939 and London remains one of the world’s least dense big cities in population terms. Residents of the highest density areas do not seem to regard it as a major factor in their quality of life, according to a 2005 London School of Economics report. This is reassuring given that new housing developments are likely to need to be increasingly compact if housing supply is to come anywhere near meeting demand, thanks to the limited supply of land.
            And the capital already has areas of very high density, some of which may be unexpected. Guided initially by the LSE’s Professor Tony Travers and then by the Office for National Statistics (ONS), I visited the little patches of the city with the highest and second highest numbers of people resident in them per square kilometre. The smallest geographical units from which such data are compiled are called lower layer super output areas – LSOAs – which encompass less space than electoral wards.
            The LSOA with by far the highest population density in London, and indeed all of England and Wales, is just south of Canary Wharf, right next to Millwall Inner Dock. The ONS tells me that in 2015 approximately 3,000 people lived within the LSOA’s 0.03 square kilometres, equating to an exceptional 90,947 persons per square kilometre.
            They reside in the canyons of high rise apartments
            shown in my photograph, all built in recent years, many of them very expensive (and giving planning permission under the now fallen former mayor of Tower Hamlets, Lutfur Rahman by the way). Just around the corner along Millharbour, yet more accommodation of this variety is rising from the ground within easy walking distance of one of London’s principal commercial districts.

            Rather rarely (for a British publication with British commenters, and no I am not User62881!) there was an excellent comment to this article:

            User62881 23 Jan 2017 2:24
            Density is one thing, how it’s achieved is another.
            I think there are stronger arguments against forms of density than density by itself. Discussing a density number can be misleading if the form it takes isn’t discussed in all its long term ramifications.
            It would also be interesting if “town planning” did what it says on the tin: planned towns in which housing plays a great, complex part. Thinking about housing as a number of units, subject to a density co-efficient is a strange, gawky substitute for town planning.
            Tower blocks may be necessary to provide homes quickly. That doesn’t change their awkwardness as urban objects. By this I mean they are objects which aggregate poorly: they do not distinguish between front and back which, in turn, means public and private realms are poorly established.
            British urbanism is not the same as Continental urbanism but lots of towns and cities from here to southern Spain demonstrate the enduring value of plot-based development. Plots are a subdivision of the urban block. From their development the block is made.
            Tower blocks tend to occupy whole blocks. They tend to replace the multitude of building plots with one.
            …the degree to which an urban spatial system allows for self-organization is highly influenced by its particular configuration of accessibility and land division… For example, it has been shown in both social and natural systems that division of land into discrete plots or parcels (in cities) or patches (in nature) can increase both social and biological diversity…
            (Lars Marcus, “Towards an integrated theory of spatial morphology and resilient urban systems”, 2014 )

          • Frederick

            The size of a normal block in London (or around London) is not really suitable for building high rise buildings. For the purpose of upzoning, one should group, say, four blocks into one plot for development, and the streets in between the blocks should be rebuilt as public spaces. Of course, if we do away with the streets then we need to replace it with non-car transportation, so it can only be done near the tube or rail.

          • fjod

            Is the ‘block’ even a useful urban typology for the residential areas of London?

          • Frederick

            Block, defined as “the group of buildings enclosed by a loop of streets”, is most often a useful concept in every city of the world.

            I was basically saying that, when we upzone a residential area, we should also replan the street layout.

          • fjod

            It’s much more useful in some than others. I suggest having a look at residential areas of London and then trying to come up with anything resembling a ‘normal block’. It is not a conceptually useful thing to talk about, like ‘hutong’ is not a useful term when talking about North American urban typology.

            By the way, ‘upzone’ is also not a useful term for a country that does not have zoning regulations.

          • Matthew Hutton

            Ban lifts, set a sensible parking maximum (maybe 1.25 per dwelling), restrict road widths to 4.5 metres and make the footpaths have hard landscaping between them and the road or avoid them altogether, then require a suitable amount of social housing (maybe 50%) split 70:30 between social rent and shared ownership.

            Then also sort out cycling away from the stations to neighbouring towns and villages.

        • michaelrjames

          The only comparison that makes some sense is on arable land.
          Nations: relative densities
          (arable land):
          Nation…….density/km2…..normalised to Aus.
          Australia:………46………………. 1.0

  4. Joseph Eisenberg

    The use of “urban” and “urbanization” in this post is rather confusing, since we are talking only about metropolitan areas with greater than 1 million people. Based on both historic and modern usage, a city can has as few as 100,000 people even in high-density countried like Taiwan, and in areas with few large towns, there can be settlements with all of the functions and characteristics of a city which only have 10,000 or 20k people.

    I would suggest using a term like “metropolitan” instead of urban for these large cities of >1 million people, and “metropolitanization”(?) as the term for the process of formation of a metropolis.

    • Alon Levy

      Yes, I’m sliding in an assumption that a settlement of 20,000 people isn’t really urban. I think this assumption is correct in the 21st century: the extent of economic specialization is such that 20,000 people doesn’t lead to the same kind of economic activity, because so many more general support workers are required to maintain modern infrastructure and society, like teachers and electricians. Certainly, people today who live in municipalities of 20,000 rarely identify as urban, and those places tend to either become suburbs of a larger city, or rural centers. Peoria and Flint, each with around 100,000 people, are barely urban, clustering around just a single plant, which closed in Flint and led to economic devastation.

      • michaelrjames

        @Alon: “Yes, I’m sliding in an assumption that a settlement of 20,000 people isn’t really urban. I think this assumption is correct in the 21st century: ….”

        Exactly, and is what I would have said to adirondacker if I thought he would listen. That 1910 census reduced the definition of urban to a mere 2,500 from the previous 4,000 which was reduced from the previous 10,000 a few years before. Essentially the census bureau decided they wanted the term to include anyone who wasn’t on some isolated rural situation. But try telling Americans in a tiny town of 2,500 that they are urbanites …
        OTOH, the term has lost its meaning.

  5. Yom Sen

    I tried to send this yesterday, but I still don’t see it:

    It’s not that Europe is less “urban” than East Asia or North and South America, it’s more that we are less likely to live in large metro areas but rather in smaller cities
    I would propose 3 possible reasons:
    – Urbanization started earlier in Europe and was slower than in other regions, it happened mostly in a period where transportation was more expensive and slower. Once cities are well established it becomes more difficult to challenge their position.
    – Americas and Oceania were mostly populated from descendants of settlers/immigrants, they would settle in the most populated areas and their descendants would move more easily as less attached to their native region. If you look at Algeria in 1954 for example, 50% of europeans were living in the 4 largest cities (74% of urban population) while muslims were only 7% (43% of urban population) in the 4 cities (http://piketty.pse.ens.fr/files/Despois1956.pdf)
    – East Asia is extremely dense. South Korea and Japan have something like 70% of their territory composed of mountains so real density is actually over 1’000/km2. South Korea has 52M people for 100’000km2, that can be compared to Paris Basin, 21M people for 140’000km2 (but no mountains) with more than half in Paris metro area or Plain of Po, 28M people in 120’000km2 with around a third in Milan metro area

    In Italy, “metropolitan cities” are just an evolution of former provinces with same limits. I don’t see why they can’t be compared to french departments (96 departments in European France, 107 provinces/metropolitan cities in Italy, 59 in Spain, 429 bezirke in Germany). In some cases they can match roughly with metro areas like in Rome (like Alpes maritimes for Nice), but in other cases they don’t match at all, like Milan whose northern suburbs are in 3 different provinces.
    Milan built-up area is at 5M (http://www.demographia.com/db-worldua.pdf), 3rd largest in the EU after Paris and Madrid. Metro area is probably at around 8M (counting the provinces of Monza, Varese, Como, Bergamo, Lodi, Pavia, Lecco, Novara that have all their capitals within 50km of Milan), 3rd largest in Europe as well after Paris and Rhine-Ruhr (and possibly Randstad)
    For smaller cities, using provinces (same for Spain) doesn’t make sense, province of Perugia for example has 6’000km2, almost the same area than Rhine Ruhr and density is half of Italian average. Prato is just 20km from Florence, 20km that are fully built-up…

    • Oreg

      Slight correction: The RMV’s northern reach is actually about right (Limburg, Marburg, Fulda). But it falls short in the the other three directions:
      * East (ends with Mainz; should extend to Bingen, Alzey)
      * West (Dieburg; –> Aschaffenburg, Miltenberg)
      * South (Darmstadt; –> Heppenheim, Worms)

  6. Pingback: Stadtbahn Systems | Pedestrian Observations
  7. Pingback: European Urbanism and High-Speed Rail | Pedestrian Observations
  8. Pingback: The Different National Traditions of Building High-Speed Rail | Pedestrian Observations
  9. Pingback: Meme Weeding: Polycentricity and High-Speed Rail | Pedestrian Observations

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.