Wednesday, January 27, 2010

Passenger Rail, Low Population Density

In the Fall of 2008, I spent three weeks in Japan to learn about the highly acclaimed Japanese rail system. Of special interest to me is how Japanese railways provide service to areas with low population density. You see, people often say the railways of Europe and Japan can't really serve as models for US passenger rail service, because their population is so much more dense than ours. It's very true that most of Japan and Europe are more densily populated than most of the US, and it's also true that passenger rail works best in densely populated areas.

So I made a point of visiting areas where population was relatively sparse: the south and east of the island of Shikoku; the inland areas of the southern island of Kyushu; and the entire northern island of Hokkaido. I've put together a couple of documentary videos with footage from my trip to Hokkaido; what follows is based on parts of the video narrative.

Hokkaido proved the best example, especially for us here in Michigan. It is at exactly the same lattitude as the Lower Peninsula of Michigan (though it gets more snow!). It's just slightly smaller in area than the Lower Peninsula. The population density is slightly less than Michigan's. Yes, you read the right: it's *less* densely populated. We'll get to the figures in a moment.

Hokkaido is Japan's predominant agricultural area. It leads the country in the production of rice and fish, and shares the lead in vegetable farming. The productivity of Hokkaido rice land is 10 percent above the national average. Wheat, barley and potatoes are other staple crops, and dairy farming is a Hokkaido specialty.

Although there is some industry (most notably paper milling, brewing [Sapporo beer!], steel-making, oil refining, building supplies, marine services and food production), most of the population is employed in the service sector. Tourism is an important industry, especially during the cool summertime that attracts campers and hot spring-goers from across Japan. During the winter, skiing and other winter sports continue to bring tourists to Hokkaido. Does that sound like what we might be looking at in Michigan's future?

Let's compare passenger rail systems in Michigan and Hokkaido. What is the actual relationship between population density and rail service? Since Hokkaido, at 173 people per square mile, is actually just a little less dense than Michigan, where we find an average of 179 in each square mile, we might predict that Hokkaido's rail service is similar or a little less that Michigan's.

What we find is radically different. Michigan has no commuter rail, no urban subways, no city street rail, no rail connections to any airports, and ten daily Amtrak round-trips, all starting or ending in Chicago and only serving the southern part of the state. Michigan does have five north-south expressways and three east-west ones.

Hokkaido, with slightly less density, has 4 commuter lines, 3 subway lines, five street rail lines; 72 daily round trips between the Sapporo regional airport and three cities, and about 154 intercity trains, serving every corner of the island. But Hokkaido has only one north-south freeway and one east-west limited access highway, neither of which reach the corners of the island, and one of which is still under construction.

Michigan Hokkaido
sq mi
sq mi
(2008 est.)
(June, 2008)
Population Density
per sq mi
per sq mi
Commuter rail
(so far!)
lines + local services to many towns
Urban subways
(so far!)
lines in 1 city
Street railways
lines in 2 cities
Airport rail connections
72 daily round-trips
Intercity rail
daily round-trips
daily round-trips (approx)
Limited-access highways
North-South (part-way)
East-West (partly under construction)

Why this difference? Not population density, though Hokkaido's rugged topography has tended to concentrate population somewhat in valleys and along the seacoast, which has not happened so much on Michigan's "pleasant peninsulas". Both, however, have areas of concentrated population that could easily be served by rail.

Both Hokkaido and Michigan began serious settlement during the ninteenth and early twentieth centuries, with rugged, pioneering types coming to trade for fur with indigenous people, to fish, to mine, to harvest lumber, and to farm. But while Michigan became a manufacturing powerhouse in the twentieth century, Hokkaido remained primarily agricultural. Both Hokkaido and Michigan were served by extensive rail systems during the late 19th and early 20th centuries, but Michigan's rail network was reduced much further in the late 20th century than was Hokkaido's.

So why the difference? It can be attributed largely to government policies in the post World War II era. Early in the 20th century, all of Japan's trunk railways were nationalized. Naturally, Japanese railways suffered heavy damage during World War II; they were reogranized in 1949 by directive of Gen. Douglas MacArthur's US General Headquarters in Tokyo as Japanese National Railways (JNR), a state-owned corporation.

JNR was operated professionally and was on the forefront of high-speed rail technology, developing the first dedicated high-speed railway in the world, the famed Shinkansen "bullet train", which began service in 1964. Unfortunately, JNR was also a favorite source of "pork" for members of the national legislature. Hundreds of unnecessary lines were built and services legislated by grasping politicians, eager to secure their reelection through favors to their home districts. (Where have we heard this story before?) By 1987, JNR's debt was over $200 billion USD and the company was spending 147% of its income - in spite of Japan's overall density of 888 people per square mile.

To put the railways beyond the grasp of politicians, JNR was privatized in 1987 and divided into seven companies, one of which is the Hokkaido Railway Company (JR Hokkaido). (In case you wondered, the six regional JR passenger operating companies own and maintain their own rails and stations, unlike the British system, in which a nation-wide company owns and maintains the infrastructure, and several competitive passenger and freight companies operate trains.)

Meanwhile, the United States after World War II spent lavishly on the Interstate Highway System. According to the Federal Highway Administration, the total cost of the Interstate system was nearly $1.4 trillion USD between 1958 and 1991, the period during which the system was officially under construction. [] Unfortunately, our road system doesn't pay for itself through user fees, and has required propping up from general funds for the last few years.

To get a handle on the relative spending: the Japanese subsidized its railways by about $1700 USD per person. During approximately the same period, the US spent about $6,000 USD per person, meaning the Japanese citizen got his high-speed and regional rail system for about 29% of what the US citizen paid for his high-speed road system. Of course, the US spent some money on passenger rail during that time - but not much; and Japan spent money on limited-access highways. The US has far greater distances to cover as well.

The trend is clear, though: the US wants its transportation money spent on highways; Japan wants its spent on railways. The effect on population growth patterns is profound. The Japanese rail system subsidized dense urban growth, while the US highway system subsidized difuse suburban growth, popularly known as "sprawl". Nowhere has that been more pronounced than in Michigan, headquarters and devoted fan of the US automobile industry. Only after being afflicted with repeated oil crises has Michigan begun to consider beefing up its urban rail service with commuter and light rail in Detroit, and light rail in Grand Rapids.

Yet in Japan, including sparsely-populated Hokkaido, the railway company earns a profit. JR Hokkaido's bank balance was 68M USD as of March 2008. If you're interested in a more complete business report, a good source is an article in Japan Railway & Transport Review from March 2008, titled "Increasing Efforts to Strengthen Position as Leading Transportation Provider in Hokkaido", written by JR Hokkaido's Management Planning Department. (

Yes, there is government support, particularly for construction of new high-speed lines (now progressing in Kyushu and Hokkaido). But diversification is a strength and a theme of all the railway companies of Japan. Ever since serious abuses by US "rail barons" in late 19th century, anti-trust legislation here has strictly limited what US rail companies can own and what businesses they can engage in. JR Hokkaido has used diversification quite creatively to turn a profit.

A look through JR Hokkaido's Web site (roughly translated by Google) shows the company owns grocery stores, convenience stores, retail kiosks in stations; a medical clinic in Sapporo station; hotels in each of three cities and two in Sapporo; financial services (including stock trading, accounting, and a debit/ticket card); 3 hot springs resorts; 8 quick-cut ten-minute barber shops, bus lines that connect with trains; a travel agency; a toxic and sensitive waste disposal service; a driving school(!); urban development projects; and 10 major station retail and office properties, including a 38-story office tower on Sapporo Station, the tallest building in Hokkaido. One popular JR Hokkaido property is an amusement center under the elevated tracks near Sapporo station, with Internet-enabled "manga cafe", DVD players, TV, billiards, darts, massage chairs, and a large N-guage model railroad layout.

Hokkaido Railway has succeeded in avoiding red ink since 1997 not only by diversification, but by what might be considered "best practices": 1. Invest in infrastructure, maintenance, and cleanliness; 2. Research and develop better ways to provide rail service; 3. Develop a culture of meticulous attention to detail, especially in safety and punctuality; 4. Schedule and dispatch trains with extreme precision

If US passenger services followed these "best practices" and were allowed to diversify, who knows what we might see?


  1. I agree with the concept of the article, but to make your argument more convincing do the math right. Michigan density is only 102 persons per square mile, not 179 as your table shows. And so your premise that the Hokkaido density is less does not hold water. I agree rail is important, but get your facts right.

  2. Anonymous, you're right as far as the figures presented here go: if you divide the population by the number of square miles, you get a density of about 102 for Michigan. I took the 179 number from Wikipedia's article on Michigan. Checking with the Census Bureau (|ST-7S&-_lang=en), we see where the discrepancy arose: the *land* area of Michigan is only 56,803.82 square miles, and since the Census Bureau figures people don't actually live in the water, they divide by land area rather than total area to get population density. That math gives me 176.1, less than Wikipedia's 179, but still more than Hokkaido's 173. Sorry for the confusion!

    BTW - another figure that would be even more relevant would be the density in the Lower Peninsula. I haven't come across that number in any easy format; I thought about adding the population and land area of each county in the Lower Peninsula, but decided the main point would remain unchanged: low population density is *not* a valid excuse for Michigan not to invest in passenger rail.

  3. I stumbled across your blog last night and set to work immediately to add a "comparison" to a project I've been working on for a few years (more on that below).

    First, thank you for the "inspiration" of comparing the land area and population of Michigan's Lower Peninsula with Hokkaido.

    I obtained "land area" statistics for the 15 Upper Peninsula counties from a state document online ("Land Area and Population Density for Michigan, Counties and MCD April 1, 2000"), and 2009 population estimates by county from the Census Bureau Quickfacts.

    The UP includes nearly 30 percent of MI's land area, but houses just 3 percent of its population. Thus, the LP houses 97 percent of the population on 70 percent of the land.

    I have not calculated population densities for "the project." Instead, I calculated "area index" and "population index," e.g. area of Michigan LP / area of Hokkaido.

    The "area index" for Michigan LP is 134, and the population index is 170. That simply means that the LP has 1.34 times the land area - and 1.7 times the population - as Hokkaido.

    One nice thing about these indices (... which might not be obvious ...) is that you can compare population densities. 1.70 / 1.34 = 1.27; in other words, the population density of MI LP is about 25 percent greater than that for Hokkaido.

    I think I'll break this comment down into a few "pieces" rather than post a single very long comment.

    Leroy W. Demery, Jr.

  4. Continuing:

    The land area of Hokkaido, excluding offshore islands, is 77,982 km2.

    The officially-stated "inhabitable area" of Hokkaido (defined as total area, less woodlands, lakes and marshlands) is 21,900.48 km2.

    Japanese sources do not agree on population statistics and estimates (I think the differences arise because of agency, methodology - and time of year). The differences are minor - but they do exist.

    The "best" I can do for the population of Hokkaido island, by taking the official estimate for Hokkaido Prefecture at 2008 and subtracting the offshore islands, is 5,522,000.

    Hokkaido (island) is comparable in land area to Ireland (island), which has slightly more people (6.2 million, at 2008), and to Austria, which has considerably more people (8.3 million, at 2009).

    Here in the U.S., Virginia (7.2 million) and Indiana (6.2 million) are slightly larger in land area, and South Carolina (4.1 million) is slightly smaller.

    I'll stick to my opinion that the most interesting U.S. comparison is with five of the six New England states: Connecticut, Massachusetts, New Hampshire, Rhode Island and Vermont have almost the same (total) land area as Hokkaido, but more than twice as many people – 12.8 million.

    ... but I'll note that your comparison with the Michigan Lower Peninsula is a very strong competitor for "most interesting U.S. comparison."

    In California, the San Francisco Bay Area includes nine counties (Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Solano and Sonoma). These together have about one-fifth the land area but an aggregate population size more than 20 percent greater than Hokkaido.

    Leroy W. Demery, Jr.

  5. Concluding for now:

    The population of Hokkaido is concentrated in a relatively small area. Sapporo housed 1,889,000 at 2007 – more than six times its population at 1950.

    The officially-defined "Sapporo Major Metropolitan Area" includes 10 "associated outlying municipalities." These, together with Sapporo, housed nearly 2.5 million (at 2009) - nearly 45 percent of Hokkaido's population. The land area of this region includes only five percent of the island's area.

    One should resist the temptation to calculate "population density" without adjusting for the large amount of agricultural and mountainous land within this area. The "unadjusted" density of Sapporo city, for example, is highly misleading - on the low side.

    The "Do-o Major Metropolitan Area" is a transportation statistical and survey area. This is defined rather loosely as the area within a 40-km radius of Sapporo station. The boundary does not describe a true circle, but follows political boundaries, and does not include the mountainous area to the west of Sapporo. Total population was nearly 2.5 million. (Do-o is a contraction meaning "central Hokkaidō.")

    Four railway lines extend from Sapporo, of which three are electrified, and electrification extends as far as 137 km distant from Sapporo station. Cities and towns along this portion of Hokkaido's rail network house more than 3.3 million people (at 2009), about 60 percent of the island's population. These municipalities account for only eight percent of Hokkaidō's land area.

    In addition to Sapporo, large cities in this area include Otaru (population 136,000 at 2009), Tomakomai (174,000), Muroran (96,000) and Asahikawa (355,000, second-largest in Hokkaidō by population). For the purpose of "the project," I've labeled this area as "the OSAM towns" because it has no regional name.

    Rural depopulation is a trend characteristic of Japan (among other developed countries) post-1945. At 2000, Japan had 1,171 municipalities with declining populations. Hokkaido had 212 municipalities at 2000, and 152 had declining populations. In other words, 71.7 percent of Hokkaido's municipalities had declining populations. This was the largest such share among Japan's prefectures.

    The "depopulated" municipalities of Hokkaido were not small villages and towns exclusively, but included former coal-mining centers that once housed a significant number of people.

    The population of Hokkaido prefecture increased by about 30 percent from 1950 to 2008 (from nearly 4.3 million to nearly 5.6 million). However, during the same interval, the share of population housed in the three largest cities (Sapporo, Asahikawa and Hakodate) increased from 15 percent to 45 percent. In terms of population, Sapporo grew by a factor of six, Asahikawa grew nearly three-fold, while the population of Hakodate remained relatively static.

    Away from Sapporo, the "OSAM towns" and Hakodate (285,000 at 2009), the largest Hokkaido cities by population are Kushiro (187,000), Obihiro (169,000) and Kitami (126,000). Three fishing ports and railway terminals at the far northern and eastern extremities of the island are quite small: Abashiri (40,000), Nemuro (30,000) and Wakkanai (39,000).

    Fewer than 1.4 million people - about 24 percent of the total population - live throughout the remainder of Hokkaidō, scattered across nearly 85 percent of the island's land area.

    Comparison and contrast with Michigan should be interesting.

    Leroy W. Demery, Jr.

  6. Fascinating comments, Leroy, and helpful statistics. I was particularly interested in the historical changes and the metropolitan area concentration figures. They underscore the point I brought up that "The Japanese rail system subsidized dense urban growth" - a phenomenon often mentioned by urban planners, but clearly shown by these statistics. Of course, there were other factors involved in the depopulation of rural areas and densification of cities, such as the depletion of coal (particularly in the central area east of Chitose) and the ability to transport goods and foodstuffs quickly and cheaply to large cities, enabling metro areas to support larger populations in comfort and plenty.

    I hope you've looked at the Japan Railway and Transport Review article I mentioned ( It focuses on financial and management issues, so it might not prove useful to "the project", but I think you'd find it interesting.

    I'm glad you discovered the blog, and I hope you'll stay in touch!