Thursday, April 17, 2008

Osaka Population

Prodded by Shisaku's post on the Tokyo population rebound, I felt a sudden urge (aided and abetted by a too-long to-do list and gloomy weather) to take a look at some Osaka demographics.

Shisaku shows how Tokyo is growing at a healthy clip of about 0.7% last year. He (she? they? I don't actually know) attributes it at least in part on a reversal of the previous trend of moving out of the city and into surrounding suburbs. If we look at Osaka prefecture, which, like Tokyo, is actually several completely interlinked cities (a major reason, I believe, for large cities being so dominant over medium-sized ones in Japan - but I digress), it produced essentially zero growth the past two years (about 0.07% this year).

But if we break the data down the story is more interesting. The first one is that urban Osaka (at 8,636,197 people) is growing very slightly while rural Osaka (at 191,288 hardy souls) is shrinking. The big surprise here is of course that Osaka even has a rural population, even with the rather relaxed definition of "rural" in this country (where the presence of actual nature seems to be an optional extra).

If we look closer, Osaka city is growing at a fairly healthy clip at 0.31%, which would make it the seventh fastest growing city in Japan; and within Osaka city, Chuo-ku (where we live) is growing the fastest at a very quick 2.8% this year and 3.17% last year. Not hard to believe; looking out the window it seems like new apartment buildings are sprouting around us like mushrooms after an autumn rainstorm.

By contrast, places like Toyono (-2% per year), Tonbabaya and Kawachinagano are losing people (almost 1% per year).In general, the central and western parts of Osaka city - and Osaka prefecture - that face the sea or abuts Kyoto or Kobe, is growing. The east and southeast areas facing the mountains and farmland, and Toyono up in the mountains of the north, are shrinking (so that's where the rural parts are, I guess; should go there sometime).

Osaka Population Change
Osaka prefecture population change over last year.

In general we should expect that rural, sparsely populated areas are losing people, and densely populated centers should gain. So I took a peek at two years worth of population data, and sure enough, there is a positive correlation between density and growth in Osaka, but with a correlation of 0.22 it's pretty weak indeed; nothing I'd want to make anything of. Of course, two years is not a lot of time, and especially for small areas one single event - a factory opening or closing for instance - can make a big difference. Also, the influx of people does depend on other, unrelated factors too, so we'd want to have numerous cities of each size to average out those factors. Many years of data, from a number of population centers, in other words. Not something I really have the time for unfortunately.

But there is another possible factor here as well. Osaka is densely populated - really dense. Southern Swedish city of Malmö has a density of 1700 people per square km, and Stockholm has a density of 4200/km2. Osaka city, by contrast, has a density just shy of 12000/km2. Were Stockholm a city in Osaka prefecture, it would be right on the median, and Malmö would be considered one of those sparse areas out toward the southern border. So perhaps the connection breaks down for very high population densities. Urban enough is urban enough; above a certain point the advantages no longer increase much, while negatives like living costs and crowding does. Take a look at this graph.

Osaka population change and density
Population growth as a function of density, Osaka prefecture.

Above is a graph of population density (on the horizontal axis) and growth (on the vertical axis). The dark green dots are the individual cities and towns in Osaka. The blue line is a polynomial fitted to the data to show the general trend. The orange bars show the average population change for ranges of densities (in the upper X-axis). And true enough: the lowest density areas (0-1499/km2), areas you'd describe as semi-rural or suburban, are losing a significant amount of people (over 1% per year), while urban areas gain people. The densest areas are on average shrinking slightly, but there are so few data points that it's not possible to draw any firm conclusions. It does sort of suggest that there is an optimal density that people will gravitate towards if given the choice, but we'd need a lot more data to tell if it's a real effect.

So, one interpretation is that the same thing is happening locally in Osaka as in Japan at large: Overall population numbers are mostly flat, but the rural areas and small towns are shrinking while the big population centers are growing. And what Shisaku is describing for Tokyo - a return from the suburbs - is possibly exactly the same process; but the vagaries of city borders means it shows up as a net inflow for Tokyo and an static population for Osaka. What will be interesting to see over the next generation or two is what will happen with the three types of area. One possibility is that the superdense areas stay flat while the dense areas slowly approach them and the semi-rural areas empty out. Another possibility - the most likely, I think, is a partial reversal, with the superdense areas shrinking somewhat and the semi-rural areas closest to the urban ones becoming urban areas themselves. A flattening of the density curve as transportation systems improve.

The question mostly on my mind now is of course if or when Shisaku will post a quick analysis of movement patterns in Tokyo, so we have something to compare with.


  1. Herr Morén -

    In order to meet to your challenge I will have to look at the data with some seriousness....which as you know is not quite my idiom.

    As for the patterns of population flow, I am not arguing that the adult outmigrants of the 1970s and 1980s are moving back into Tokyo. The realities of home or condominium ownership (i.e.--the lack of a substantial resale market) would make such a move difficult. The 0.7% annual growth from inward migration more likely represents the moves of the children and grandchildren of outmigrants moving into the core areas of Tokyo, topped off by the long-term inflow from the non-core prefectures.

  2. Nah, I wasn't really serious about asking you for data analysis. Besides, it's fun so if I really wanted it, I'd prefer to do it myself.

    Yes, I meant this the same way you did. With population flows at this level it doesn't normally matter who is moving, only the numbers. If nothing else, saying that "1% move from place A to place B" is really a shorthand for "lots of people in A moved to B, lots of B-people moved to A, and people moved from and to many places from A and B - but the net effect was 1% less in A, with a 1% increase in B"

    At a more detailed level, or when looking for precise causes, then the identities (or rather, the types) of people start to matter. But at this level it does not.

  3. Janne, thank you for a great blog!
    As a human & economic geography student in Gothenburg with an interest in Japan, I'm a little curious about what your research field is. Several of your articles (especially those on Japanese demography) has a certain geographer's touch to them.


  4. Claes, fun to hear from a professional in the field! I know I'm certainly not one.

    My research is not connected to demography or geography in any way; I'm a computer scientist and cognitive scientist by trade, vaguely in the area of doing brain modelling. In my current job I'm trying to create an active early vision and emotional encoding system for our humanoid robot.

    I write about demographics and related subjects as a pure amateur; it is something I know very little about, but which is a lot of fun to consider. Unlike the research fields close to my own, the implications to our lives are quite direct, and the literature is often refreshingly accessible, written for a comparatively wide audience. Also, for some reason other people's research always seem more interesting than your own. :)


Comment away. Be nice. I no longer allow anonymous posts to reduce the spam.