…goes the title of a relatively old paper by Professor David Flath, who  teaches economics at Ritsumeikan University these days. As I study Tokyo’s postwar history and, as part of that, am interested in the density of retail stores, it’s worth summing up the main points and adding a few more thoughts. The high density of retail is a phenomenon that despite several years of de-densification stays roughly intact today.

Evening conbini scene, Nakano-ku

Tokyo’s retail industry has historically been dominated by small and medium sized retail enterprises. In 1954, establishments with 1-9 staff constituted 91.6% of the total number of retail establishments. This stayed roughly the same until the 1980s (1981: 87.7%). A marked change towards larger retail stores took place from then until today (2014: 73.9%). Keep in mind this is number of shops. If we look at employment figures, a much lower proportion than 73.9% of retail employees will work in these small retailers (explained by number of stores times average employees).

For the purposes of my research, just some quick numbers on just how dense Tokyo’s retail network was in the postwar period. I have data for three types of establishments: food retail (your greengrocer or mom and pop grocery store), restaurants, and construction companies. The period under consideration here is 1954-1972, although I also calculated the equivalent values for 1981 and 2014 for context.

Several trends emerge: Food retail shops almost maintain their density amid a rapidly increasing population. While there were 136 Tokyoites per retail establishment in 1954, by 1972 there were 167. This means that the growth in greengrocers etc. turned out slightly lower than overall population growth. Restaurants, on the other hand, grew disproportionately faster in number. Their density almost increased by factor 2. In 1954, there were 279 inhabitants per restaurant, in 1972 only 147. Lastly, another really important feature of the local commercial ecosphere: small construction establishments. Their density also increases considerably, from 490 inhabitants per outlet to 309.

Why is this important? First off, it is remarkable that the density of retail outlets stays roughly stable (food retail) and even increases (restaurants and construction companies) although the city experiences unprecedented, rapid “megacity” growth. As these are very labor-intensive segments of the economy, they absorb a considerable part of the working-age population. My research shows that these retail patterns was also relatively similar across the city, with little variability of these density figures across the 23 wards. Moreover, the ability of the city to regenerate its building stock (with low average building ages) is also made possible (and caused?) by the ubiquitous construction companies.

So why is there such dense retail network? In the 1970s, several explanations were being offered, as Flath shows on p4. These ranged from the low mobility of housewives to low household incomes (and hence lack of consumer durables such as refrigerators). Flath then builds his analytical model for Japan, using data from the 47 prefectures in 1985. He notes a couple of interesting points before going into his data interpretation: higher population density would theoretically favor fewer retail outlets, as the average distance from home to store is low. However, lower average living space per capita suggests high storage costs for non-durable goods. However, not only do consumers face high storage costs, but also retailers.

In a slightly more recent paper, Flath summarizes the model outcomes, concluding that:

• Crowded living space increases the willingness to pay a premium to shop nearby. One of today’s examples is that households rarely store large quantities of drinks at home but buy them incrementally in nearby convenience shops
• The high cost of car ownership renders shopping at more remote malls and large stores expensive
• Japan is a geographically compact country with on average short distances between cities and supply nodes and there is an abundance of trucks and an efficient road network. Taken together, this reduces restocking cost of retailers. They can run small inventories with multiple daily deliveries

Flath shows that accounting for the above, Japan is no regression outlier, i.e. does not behave differently than other countries. Therefore, the density of retail stores is not surprising.

How about regulation? The 1974 Large Store Law (since repealed) restricted the size of new stores (plus they needed MITI approval!), and helped explain why small stores could survive. Today, prefectures are given more leeway to decide on their own. They can turn down applications for the opening of new large stores on environmental grounds.

Also, as GRIDS Vancouver helpfully pointed out on Twitter, zoning regulations explicitly cap retail stores floor area in many districts.

To end this post, herewith a map of convenience store locations in Tokyo courtesy of Tokyo Files / Reurbanist. The concentration around train stations is not surprising but it’s nonetheless a stunning visual representation of something each visitor to Japan feels. There is a conbini anywhere you look!