Saturday, September 20, 2014

European, American and Japanese approaches to height and density

Okay, I've talked about TOD quite a bit recently, and I may be approaching rambling level... or I may be at the end of a reflection on certain key concepts and specific observations culminating in a synthesis of all the elements I've learned recently. Again, a reminder that I'm thinking out loud, I don't have years of planning studies behind me, I may take a seemingly authoritative tone, but that's just my way of being neutral and self-effacing. I completely agree that I may be mistaken and welcome corrections.

Anyway, here goes.

We often talk about the need for density to achieve sustainable urban areas, but density can take many forms, and indeed, it does. Many different cities across the world may have similar density overall, but the distribution of density is quite different, and that has significant effects on how cities work. To try to be witty "the density of density matters".

 Here is what I noticed, with three major models in the developed world.

European Model

Classicism, harmony, order, a wide mid-density center

What defines the European city seems to be its classicist approach to city-building: orderly, based on formal rules to guarantee harmony between buildings and uniformity. The Europeans by and large seem to abhor anything that stands out, they have built remarkably few elevated highways or trains and seem to be hostile to tall buildings, strict height limits aren't rare in European cities, apart from a flirtation in the post-WWII era with "towers in a park" modernism (which may be part of the reason why they have turned against towers). Europeans are also strict preservationist, keeping old buildings around seem to be of prime importance to them.

European cities are also very old, so most have pretty wide old dense downtowns. These downtowns were mostly made in earlier era, when building methods and the lack of elevators limited building height to 5 or 6 stories only. The result is that, for the most part, European city downtowns are extremely similar one to another, if not in the detail, in the generic form they take:

London from the air

Paris from the air
Frankfurt, Germany

Frankfurt from the air
Stockholm, Sweden

Stockholm from the air
Amsterdam, Netherlands

Amsterdam, from the air
Prague, Czech Republic
Prague, from the air
As an aside, it's interesting how the traditional European urban bloc is the negative of what a "tower in the park" is. A tower in the park has a green park next to the street, with a tall structure in the middle, the traditional European urban bloc has a series of small or mid-rise structures built wall-to-wall right next to the street, with a green courtyard inside the buildings. 
Traditional European Urban Bloc to the left, a small park surrounded by buildings, Tower in the Park to the right, a tall building surrounded by parks
Though old city centers are very similar in all European cities, new developments in the suburbs vary widely. They still tend to be denser than what is seen in North America, either by having plenty of low-rise multi-family buildings (Sweden, Germany) or by having dense rowhouses (UK, Netherlands), or simply by having smaller houses and lots (France). In Eastern Europe, the communist governments liked to build high-rise apartment blocs in parks in the suburbs.
Percentage of population living in each dwelling type

Europe seems to have much the same problem as North America as in that they seem unable to build new neighborhoods like the old ones. All the really dense areas tend to be pre-existing ones, with the new areas being of significantly lesser density.

This model of mid-density spread about a large urban core seems to work very well with walking, European cities have pretty high walking mode shares (source): London and Amsterdam have 20%, Munich has 28%, Paris has 47%, Bilbao has 60%!

For transit, it's a mixed blessing. The urban core, if wide enough, is quite dense enough to support an underground rapid transit system very well, which has also the advantage of being about the only form of transport that can go fast across such an area, even cars are quite slow amongst the narrow streets and frequent intersections. However, go beyond that core and the effectiveness of transit quickly declines because the newer areas are much less dense, much less mixed, and the European style of development means density is evenly spread out rather than concentrated. Furthermore, as the core has no great concentration of density but instead density spread around evenly (but at a high level), you need a tightly packed grid of lines to offer correct service. The result is that, apart a few exceptions (Paris and Lyon), the performance of European subway systems is quite middling per length of track.Which justifies investing in LRT, S-bahns and tram-trains rather than subways.

Another effect is a tendency for housing prices in the urban core, as it is in demand yet struggles to expand, to be quite high. There is also a lack of housing variety in cities, when all buildings are the same, all housing options are the same too.

North American Model

Skyscrapers surrounded by endless sprawl

In contrast to the Europeans, North Americans by and large seem to embrace tall buildings... but also the idea of use separation. So build that skyscraper if you want, just keep it far away from houses, put it amongst its siblings instead. There are a few exceptions, Washington DC is following the European pattern instead, and older neighborhoods like in Brooklyn may sometimes have an European feel, though their origin is different. Whereas old European cores were built for walking, predating mass transit, most similar places in North America are in fact streetcar suburbs, built around streetcar lines.

Anyway, with all the skyscrapers, American downtowns are probably nearly the densest in the world, at least they would be if it weren't for all these parking lots. But as befits the obsession with use separation, over time the old walkable downtowns have been converted into dense office and commercial zones exclusively, like gigantic industrial parks.

Outside these downtowns, apart from a few exceptions built way back, Americans have built sprawl, really low-density areas, mostly single-family homes on large lots. One of the reasons for this sprawl is that density only works if it comes with proximity. Living in an apartment bloc isolated from everything else is far from ideal, to say the least. So as cities sprawled, the new developments built were on the outskirts of everything. What kind of building do you build on the outskirt of a city? Low-density housing of course, that's what makes sense to build as these areas will be the farthest from everything else. But then you'll need to build another layer, and another, each layer being best fit for low-density housing when they are built. And the zoning and use separation obsession means that residents of the older layers will oppose densification of these layers, no matter how pressing it is.

The result is an almost entire lack of walkable neighborhoods in most cities, limited to a few old streetcar suburbs that were spared odious "urban renewal". Even the downtown has largely been converted to a handful of offices and restaurants. Even in transit, this is terrible, as, yes, there is a great density of jobs downtown, but the people going there are spread all over the area in low-density neighborhoods. Subways are out of the question for most of the area. In recent years, some cities like Dallas, Denver and Houston have gravitated towards LRT to fill the gap, but even they seem overwhelmed by the low density and the spread out nature of places of employment, often only connecting a few chosen neighborhoods to a downtown which now has a relatively small share of regional employment.

The one point in favor of the North American model is the skyline, which is easier to see and more impressive due to being an island of high-rises surrounded by an ocean of low-rise developments.

Dallas Downtown
The Northern part of Dallas' Sprawl, an ocean of single-family homes with industrial parks and malls put her and there
St Paul's downtown
St Paul's ocean of SFH


Japanese model

Rational chaos, density where it counts (especially near subway/train stations)

I've already written about Japanese zoning laws, and these seem to represent the Japanese mentality around city-building well: very lax controls and respect for private property rights, high tolerance for building heights and mixing uses. The Japanese are also much more tolerant, if not outright welcoming, of neighborhood changes, and the dominant mentality says that houses with wooden frame ought to last 20-25 years and housing with concrete frame should last 30-40 years, after which they can, and perhaps ought to, be replaced.

The result is very striking in Japanese cities. Density happens near places where it makes sense, so it's not rare for there to be high-rises next to train or subway station, with low-rise constructions a bit further out. Commercial areas like malls will also tend to attract high-rise developments to them.

Whereas both in Europe and North America density tends to progressively fall as one gets further away from the center of the city, in Japan, though the center does tend to be denser, the density of cities tend to extend in ribbons around rapid transit lines.

One of the clearest examples of that is the area around Shinjuku station in Tokyo:
Around Shinjuku station in Tokyo
There are skyscrapers here, but they're massed around the major arterials and the subway stations, less than a kilometer away, you have low-rise areas. This offers a variety of housing that is rarely seen in Europe or North America. It also means that points of attraction (malls, offices, etc...) concentrate around stations, where they are most easily reachable in transit. That is especially important, because if jobs are only accessible by car, then people will have to own cars to get to work, and once they have cars, they have the best incentive to use them.

Sendai, near a subway and train station

Sendai, next to the subway...

...and 800 meters away
Sapporo, near subway station

Sapporo, near other subway station

Sapporo, 1 km away from subway station
Niigata, a city without subway, mall to the right, mid-rise housing across the street
It's important to point out too that low-rise areas in Japan still have relatively high density.

The Japanese model of development is chaotic and organic. Different types of buildings mix, whereas in Europe and North America, types of housing are often clearly separated, stuck in their respective neighborhoods. But there is a rationality to this chaos, high-rises are grouped at important transit nodes or near services and businesses.


If I could sum-up with a few schema, here is how I would represent building height (and correspondingly, density) in the different models:

Distribution of density in European cities

Distribution of density in Japanese cities

Distribution of density in American cities

One of the impacts this has is on transit use of rapid transit lines. The Japanese system tends to gather density around subway lines, so as subways are built, they tend to attract development to maximize the subway line use. In Europe and North America, this attraction is much less present and tend to be the exception (like the Rosslyn-Ballston corridor in Arlington and the "City Centers" strategy in Toronto of building city centers around subway stations in the suburbs, both of which are planning efforts to do what the Japanese spontaneously do when they have a subway line). Old, dense European cities also offer great potential for rapid transit, but this is piggybacking on ancient density, and so the potential of these networks to expand is relatively weak.

Subway systems by track length and ridership per km

We can notice here that all Japanese subway systems have at least 4 million passengers per km of track, and Tokyo Metro is way up there at 12. European subway systems are all over the map, with those of Paris, Rome and Milan getting exceedingly high ridership per km of track, Rome and Milan despite having very small systems, probably reflective of the very dense old urban cores of these cities. However, many European cities perform relatively poorly, with at least 33% less ridership than equivalent Japanese systems, and that despite the fact that Japanese subway systems are much more expensive for the end-user than the subsidized European systems (Japanese subway systems are profitable, not so for European systems). How much more used would the Japanese systems be if people could buy a 80$-100$ monthly pass that would give them unlimited rides across the network like they do on most European and North American systems? (Most Japanese subway systems have unlimited one-day passes for 8 to 10$, which seems to be the only way of getting unlimited rides across the network, there are commuter passes, but, get this, they only work between two given stations, get one station further out and you have to pay)

North American subways are even worse, and I did not add LRT systems which on average have only 0,5 million passengers per km of track (Calgary and Edmonton top out at about 1,55). In fact, it's interesting to point out that North American subway systems, where they exist, are no less developed than most European systems. The issue with most of them isn't that they are not developed enough, but that land use around the areas they serve leads to sub-optimal use of the lines that exist. Special mentions for Chicago and Atlanta which both have very long networks yet very poor performance, at the level of LRT systems.

European subway systems that leave the old urban core also offer relatively poor performance, London's system being one of the most developed in the world, yet its performance per kilometer is middling at best, being inferior to essentially all Japanese systems and equal or inferior to many American systems (New York, Toronto, Montréal and Boston). 

All in all, it means quite a few American and European cities should be trying to reform land use to make better use of what rapid transit they have before thinking of investing billions into adding new lines or extending them further.

The chaos of Japanese cities also show positive results in terms of car mileage. Even when the Japanese use cars to get around, the proximity of density to jobs and services mean that people can use them less.
Average mileage of cars in select countries
Thus, the average car travels 20 000 kilometers in the United States, around 13 000 miles, but only 10 000 km (6 500 miles) in Japan. Europe is in the middle. Truth be told, Japan has a further advantage that on long-distance trips, it is often better to take trains, that are faster, affordable and connect well the country, whereas in North America and even Europe, traveling long distances is more likely to be done in cars. Still, the fact that Japan can do it is often due to density, and thus locations of interest, being built near train stations.

Another criterion to judge cities on is the active transport mode share, the percentage of all trips made by residents taking place on a bike or on foot.
Percentage of all trips made on bike or on foot in select cities, red for Japanese cities, blue for European cities and green for North American cities
Overall, both Japanese and European cities enable active transport thanks to density and proximity, but there is more variation among European cities, whereas Japanese cities all tend to be between 30 and 40%, some European cities reach a low of around 20%. And of course, it's a complete disaster in the North American cities, where decent levels of active transportation in some central neighborhoods get totally drowned by the deserts of sprawl all around them. Use separation is the crucial aspect that kills the potential for active transport.


So these were my observations on the different ways the Japanese, the North American and the (Western) Europeans build their cities. Both the Japanese and European ways of building cities tend to work well to favor alternative transport modes, despite their evident differences, with Japan having chaotic cities with buildings of very different height and shapes and Europe generally preferring orderly neighborhoods of similar buildings. North America is way behind on every metric.

I'd give the edge to Japan for the ability to build functional cities with self-funded transit and to keep building dense, walkable areas, which Europe seems to struggle to do, with transit being quite subsidized and recent cities faring much less well than their old urban cores. Without the preservationist instinct of Europeans for their old cities, they would likely fare much worse.


  1. Probably the most Japanese-like city in the US is Los Angeles, which has a relatively weak central CBD and secondary clusters of high-rise office towers spread out over a fairly wide area, especially along Wilshire, with plenty of SFR (which is really 2FR, since everyone has an illegal ADU in their backyard) not far away. The best strategy is probably to allow development along the boulevards to intensify to mid-rise densities, let the high-rise clusters keep growing around Metro stations, and keep the SFR (with legalized ADUs) in the remainder.
    On a smaller scale, I feel like Boston is also pursuing the polycentric model, with Kendall Square and the Seaport growing up as secondary CBDs in addition to the existing Downtown/Back Bay ones, with the former built around a heavy rail line and the latter around lots of parking and a bus tunnel. It'll be interesting to see what happens when Kendall is more built out, and whether development keeps moving up the line to Central Square.

    1. It's more than being polycentric, many mid-size Japanese cities are not very polycentric at all. When the traditional CBD is saturated, then you see other centers spring up along rail lines.

      It's mainly about allowing density near centers of employment and of commerce, so that the largest concentration of population is right next to the largest concentration of jobs, or at least a short transit ride away. If the density of residential areas around subway stations or around CBDs is not significantly higher than the density of residential areas further away, it's not really the Japanese model.

    2. Ah, I see. That makes sense. Boston does have that to some extent with high-density residential neighborhoods (Back Bay and the like) around major job clusters, but due to the core city's relatively small size, I think that ends up feeding walking more than transit ridership. I believe Boston has the highest walk-to-work mode share in the US. LA is weird because the transit network just isn't there yet for the most part, and zoning has greatly inhibited the pattern of gradual densification that built most of the inner part of the city.

    3. Los Angeles is still the only major American city that is denser than in 1950 though. Most are significantly (half or less) less dense than in 1950.

      Anyways, I think some cities in Europe have built a fair bit of dense neighbourhoods more recently. If you take a neighbourhood like Delicias in Zaragoza for example, I don't know exactly how it came to be the way it is, but the buildings appear to be all modernist (post 1930). If you check out the outlying neighbourhoods of many Spanish cities, you'll see a lot of densely packed mid rises that appear to have replaced 1-2 storey row houses.,2.162244,3a,75y,301.41h,96.67t/data=!3m4!1e1!3m2!1sDby0JOxjVVYu7TKqpVU71g!2e0

      The more recent (housing boom era) neighbourhoods of Spanish cities are not as dense and more cookie cutter and less organically developed, but are still largely 4-6 storey apartments (with some townhouses). Greek and Italian cities seem to have this too.

      The per km ridership of the Barcelona metro is towards the lower end of the Japanese range, but it seems to have a more developed rapid transit network than most cities its size. All the Japanese cities that do better have much larger networks and presumably larger populations too. Barcelona only has 20% auto mode share which leaves little room for transit growth.

  2. Where are the walk/bike mode share numbers from? I'm asking because the North American numbers seem to be trip-to-work mode shares, whereas the others seem to be trip mode shares. Way less than half of Paris walks to work. The importance of this is that transit users usually take transit to work but walk to other errands, and this means that the transit all-trip share is much lower and the walking/biking share much higher than the trip-to-work mode shares. There's a survey of mode share in Sydney that I can try digging up that cites both numbers, which shows this pattern neatly.

    1. For Europe:
      For Japan:

      For Montréal:
      For Vancouver:

      For Japan, Montréal and Vancouver, the data is trip survey for all purposes. For the US, I posted the numbers I could find, I've struggled to find explicit trip surveys before I posted this article. Since then, I've found a few more, but I've also learned that the typical methodology in the US currently seems to be a bit weird.

      What I mean is, take walking your dog. How do you count that? Is it zero trip (because it's a round trip to go nowhere in particular)? One trip with destination and origin being the same? Or is it two trips?

      If I understand correctly, since 2001, the NHTS in the US splits round trips (so says page 6 of this document: This comparison between mode shares in the US and Germany also had to take American double-counting round trips into account to get similar numbers and pointed out that the American survey prompted people many times to count all their walking trips, unlike the German survey: (FTR, the numbers they used showed Germans walking in twice as many trips as Americans and biking 10 times more, overall having active transport mode share three times that of Americans, in line with the numbers on my graph).

      This is pretty major. Someone who owns a dog (and from certain estimates, 36,8% of American households do) and walks it 15 minutes every day would, on an average day, have maybe two car trips (to work and back) and 2 walking trips (walking the dog, counted twice), so has a 50% walking mode share. If you ask me, that's a pretty stupid methodology that balloons walk trips without being representative of anything. Which may account for, for example, walk mode share between 2000 and 2010-2012 increasing from 8,4 to 16,6% in California according to the DOT study:

      The US methodology also yields much higher number of trips per person than the methodology used in Canada or in Japan (3,6 versus 2,0-2,5), probably largely due to counting round trips twice.

      So I am in a quandary here. The graph is important to show the impact of urban form on active transport as an utilitarian option. However, the comparability of the data is hard to ensure, given the different methodology. I may update it with better data if I can find them, but I am confident that what it shows is quite comparable to reality if not in the particular, at least in the trend, with active modes being 2 to 4 times more frequent in Japan and Europe as in North America, if only because the Canadian data comes from trip surveys that seems to be of similar methodology to that used in Japan and Europe.

  3. Are there any other significantly different approaches to height and density? For example, what about mainland East Asia, or Latin America? IIRC Mexico has an even higher home ownership rate than the USA, in spite of being a considerably poorer country...