I stumbled upon an interesting blog post on Market Urbanism from a few years ago that talks about Tokyo’s surprising lack of density. I would disagree somewhat with that statement. More below the jump.

Tokyo towards the West as seen from Shinjuku

The author refers back to a post on the “New Geography” blog written by Wendell Cox, quoting him as saying that:

Tokyo does not have intensely dense central areas. The ku area [historic core] has a density of 37,300 per square mile (14,400 per square kilometer). This is well below the densities of Manhattan (69,000 & 27,000) and the ville de Paris (51,000 & 21,000). Only one of the ku (Toshima) exceeds the density of Paris.

On the other side of the spectrum, Matias and Rahul have described the city as low-rise, high-density. They write that:

 Newcomers may be shocked to find that much of residential Tokyo actually resembles the low-rise, high-density habitats one normally associates with cities like Mumbai and Manila.

I don’t think either of the two descriptions is necessarily right. I think a more accurate term for Tokyo’s unique typology would be uniform low-rise, medium-density.

The revelation does not come from comparing Tokyo’s aggregate densities to those of say, Mumbai, but to look at the surprisingly small standard deviation of the various wards’ densities. This means that the low-rise, medium-density typology stretches out across the 23 wards, and even beyond.

Here are some numbers:

Source: Tokyo Metropolitan Government (TMG)

The chart shows the historical development of population densities between the 23 central wards of Tokyo (they did not exist in this form pre WWII, but thankfully, TMG did some backward calculations in one of their Statistical Yearbooks). Comparing the 23 wards to Manhattan misses the mark. They are not Tokyo’s historic core.

In the early period before the war, one has to think of Tokyo as still growing primarily in the East and within the area encircled by the Yamanote Line (much more akin to a “historical core”). The difference in population densities between the various wards is relatively high. What this means is the city had extremely densely populated areas in the shitamachi (downtown) area, while the newer wards east and west of the centre were practically not urbanized yet.

The Allied firebombing during WWII decreased Tokyo’s population, millions of people left the city. The central wards were now almost as sparsely populated as the ones outside the centre. With the end of the war, the city repopulated and reached its pre-war peak in the early 1950s. Prewar patterns establish themselves again, with much of the immediate postwar population growth occurring in the traditionally densely-populated areas. Population density and standard deviation between the wards grow in tandem.

However, the standard deviation between the 23 wards’ densities reaches its postwar peak in 1960, from which point it begins to fall. The growth of postwar miracle Tokyo does not translate into lower density suburban growth but reproduces the same typology (i.e. medium density) of settlement in the new and outlying wards. The most densely populated inner-city wards actually begin to shed density, leveling the playing field ever more. The graph stops in 1970, but standard deviation continues to drop to about 4,000 inhabitants in 2014.

The Market Urbanism blog post thinks this typology is a root cause for several of the city’s ills. Take high housing costs:

Imagine New York City if Midtown and the Upper East and West Sides were still tenement neighborhoods, and everyone living and working above the sixth floor was competing for housing and space in the outer boroughs. Narrow the streets and replace the prewars with postwar buildings, and that’s Tokyo.

Also, people live far from work and have to commute:

High prices also cause people to live very far from work. Many of them still take the train, but the commute is very long, sapping what is becoming an increasingly precious commodity: time. And some commutes are just impossibly long, limiting job opportunities and flexibility.

And aesthetics suffer:

(T)he density caps lead to ugly (not to mention energy inefficient) buildings. Japanese cities have very few historic areas left compared to European and American ones, but redevelopment is limited by the fact that many urban buildings are built to the zoning envelope, and thus tearing them down and building anew will result in higher rents per square foot, but not more square feet.

There is some truth in the high housing costs, although they are caused by many other factors, and blaming them on Tokyo’s “low” density alone misses the mark. Tokyo’s average commute is very long indeed. However, New York’s is longer, at least according to some calculations.

The lack of energy inefficiency is a valid claim. However, despite the apparent waste of energy due to short building lifespans and lack of insulation, Japanese cities have relatively low per capita CO2 emissions, primarily due to the important role of public transport as well as the small apartment size.

Finally, aesthetics and conservation of historical areas are very subjective measures. The frequent comparisons to New York show that the author really struggles to shed his frame of reference.

While there are definitely negatives, the positive sides of Tokyo’s density should also be mentioned: Matias and Rahul have written about the “Tokyo Model”. The low-rise, medium-density typology is historically explained by a lack of zoning as well as the proliferation of live and work arrangements.

Crime is low, as are income differentials within and between the different wards. A lack of public space is partly offset by very livable streets with their private share of greenery. The comments on the Market Urbanism blog post have several other good tidbits about the advantages of Tokyo’s medium density.

Source: Mumbai Human Development Report 2009

Let’s look at the corresponding figures for Mumbai. (I only found ward-level census data stretching back to 2001, but the trend is clear.) On the one hand, we have much higher density in aggregate terms. However, what is striking is the difference between the wards, with a standard deviation of 25,000 inhabitants / sqkm.

There are of course many reasons for this, and density comparisons of political entities are almost always biased. In Mumbai’s case, its northern suburbs particularly have not got very dense settlement patterns as of yet. Nonetheless, the heavily urbanized areas in the south have a high variation in their densities that maps such as this hide because their maximum of 50,000 inhabitants / sqkm naturally hides the extremes.

These differences would only become more acute if we were to divvy up the wards and make a map more in line with the socioeconomic realities of the city, with slums naturally showing the highest density. Dharavi is said to have density as high as 200,000 people per sqkm, although accurate figures are hard to come by.

It is this variety in density patterns that Tokyo lacks, explaining the sea of relatively uniform single- to two-story houses stretching all the way to the horizon, even on clear days. This photo from my balcony is probably more representative than the one at the beginning of the post, which was taken from the Park Hyatt in Shinjuku:

Which leaves me with the last thought of this post: how about the “low-rise” part of the typology description? Is this valid any longer in a city that seems to embrace height more and more? Without going into detail yet (a separate post may follow), high-rise condominium buildings are certainly growing in importance, also for reasons related to real estate economics.

The zoning regime permits those to be built in commercial areas and near train stations. You see them proliferate throughout the city, not just in the central wards, so Tokyo is becoming undoubtedly more high-rise. Due to the relatively uniform distribution of these projects throughout the city, I don’t think that this will dramatically change density patterns as far as variability between the wards goes. Their impact on density as such may also be overrated. I’ll keep looking for some figures to dig more deeply into this part of the equation.