A grocery store is identical to a mile of cars. Grocery store generally carry most brands. Dealers only a couple. Cluster of different types of grocery stores may work though.

There are differences between real estate and books that make the comparison difficult. Obviously, locating a store has issues around locality, leasing, and significant capital expenses for (perhaps) the building, fixtures, inventory, etc. Those decisions are made with a long view in mind (one hopes) and are expensive to change once they are fixed.

Publishers, however, set their lists around different externalities. Most important, when will the author deliver the finished book? You couldn't have published a book last spring if it wasn't written. Once it is written, the standard publishing contract requires publication within a fixed amount of time. No author is going to let you sit on a book for a year because you'd like to avoid the big push Random House is going to make for Charles Frazier. "Make a bigger one for me," is what you'll hear and it has legal and practical force behind it if you ever want to publisher another book by that author again.

Publishers also have to consider media externalities. The fall of 2008 is going to be a bad time to publish anything that isn't political and election oriented, because there'll be little or no media attention available outside the election. Father's day, holidays, and the seasons all can define when a book is best published. There's a reason most baseball books are published at the beginning of baseball season, and it's a reasonable one. November is a lousy time to publish baseball books.

Unforeseeable externalities also dramatically effect the market for books. Ask a business book author what it was like to publish into the bursting of the bubble in 2000. No publisher is going to risk waiting on a book it thinks could sell a lot of copies (even if it could) when the market seems clear now.

Finally, publishers think of authors like those in the article as having defined audiences, because they usually do, especially for fiction. Established fiction authors tend to have pretty predictable readerships they bring with them, regardless of competition or externalities. Nonfiction has more competitive pressures by nature, the article would have made more sense to look at competiting political books (which are dangerously overpublished right now). But then, the Times wouldn't have had as sexy a competition to talk about as they have with that group of star authors. It's not that there aren't some legitimate points raised in the article, but there isn't as much the publishers can or even want do about the situation as it might appear.

I think it is safe to say that Mr. Konishi and the others are not familiar with the concept of zoning. While there are many reasons for why dealers locate where they do, aside from a few regions (Houston comes to mind), one of the principal reasons for dealership locations is the availability of automotive zoning, which is tightly controlled by local zoning boards, since as you might expect, no one wants an automotive dealership in their backyard. Houston, with no zoning, is like a shotgun whether you are a car dealer or anything else for that matter. California, in particular, which was a leader in the concept of auto malls, and most other mature metro markets have tight controls and limited land on which to locate. Other factors not taken into account is that given the increasing vehicle complexity and lengthened factory warranties, most people now buy (and importantly) service their vehicles at dealerships near where they work or live--further contributing to the clustering effect descibed in the article. A last contributor which I will mention here is that due to the nature of franchised auto retailing, all real estate development is conducted by dealers themselves, who first tie up the land, obtain zoning, obtain the franchise and then build. The dealers often then acquire large parcels of land and are able to achieve the zoning for several dealerships at once, in effect ruling out other locations.

Nice reference to "The Sotweed Factor". I'd like to take a moment here to pump this book. I read it a few months ago and it is now in my top 10 most favorite books. In several places it had me laughing out loud. It sounds like I need to check out Pynchon.

I think you guys are overthinking the book thing: I think it's just ramp-up for Christmas....

Movies also cluster around significant dates, such as Memorial Day and Christmas...

Hi Paul,

This is such an interesting post. Thanks for the links to the extra reading materials (they are comprehensive and loooong). I happens to have read some insightful analysis about the phenomena of clustering (in the context of similar factories clustering in Chinese cities, and other more exotic cases like - prostitutes!) and I will try to dig up some information and share with your readers here.

My stuff will be based on articles written by the retired professor Steven Cheung (you can search for Steven_N._S._Cheung in Wikipeida). He despises Game Theory analysis thinking that they based too little on real world insights. To me, Steven is a very insightful man even though sometimes he comes across as too full-of-himself. (Note: Steven has worked with the likes of Ronald Coase and Milton Friedman and he will and has reminded his readers of this fact oftern. (smile) And Steven is still in contact with these giants of the last century.)

I will come back and post some more after I look up those articles.

Cheers,
Kempton

You can't compare car dealership locations and book publication dates because one is chosen ahead of time and stays that way for a long time and the other just happens to shake out where it does, especially for the less prolific authors that you mention. It would be like comparing the buying process for a house and a pear. An excellent description of how auto dealership zoning works by Mr. Kohler, I bet Konishi wishes he had met him before he wrote his now useless paper. btw, I've never read Pynchon and probably never will but that new book sounds like a weak satire of a western.

btw, I posted this comment once already and was told that it was being held over for approval, probably because I didn't enter an email address the first time. Since I've seen that this approval process just dumps comments in a black hole, I just went ahead and reposted the comment.

The Freakonomics guys were thinking about the same clustering problem a while back (http://qurl.com/434q5) and wondering how it can make sense to have so many water parks in the same corner of Wisconsin.

I tend to the view that it's primarily a marketing costs issue - once you get a few water parks / car dealers clustered together, that area just becomes "the place to go for cars / water slides" which leaves the individual businesses free to concentrate more on improving the competitiveness of their point of sale offering and less on making sure people know where they are. That gets circular too, because if what now matters is competitiveness (a relative quantity) you need to be near the competition to keep an eye on them.

pynchon, can be fun. Gravities rainbow sure covered a lot of ground. The crying of lot 49 is a good starter piece. Also have to like the mystery surrounding Thomas Pynchon. Who is this guy.

Holy crap, Paul -- that has to be the single-largest post I've ever seen you do on any topic. My best guess is about 1,200 words. Is FM paying you by the word now? :-)

Seriously though, some interesting things to think about. I was going to say the same thing Seamus did, about marketing costs. That's as close as I can come to an insight. Oh yes, and Pynchon rules.

Mathew -- I know ... was long. Sadly, FM isn't paying by the word. I just had some things bothering me and I needed to get it all out of my system. It comes over me now and then, but I usually dump it on unsuspecting co-workers and portfolio company CEOs.

it is inverse proportional with how perishable the product sold is: if it has a tendency to stay longer on te shelf then clustering benefits from sharing the same pool of customers. If it is short lived then it is probably commoditized to some extent so that resource pooling does not bring any benefits.

More like this one, please.

Meanwhile, I presume there is a franchise business model that encourages clustering of those who build on their platform.

There is a reputation effect of the host real estate.

Easier to sell a new patient to go to a landmark
Georgia Medical Building than to a generic
820 N. El Camino Real.

Medical and law offices tend to be in the same building.

Food clusters, too: the fruit aisle and vegetable aisle
are in the same market.
Customers buy fruit and vegetables (or beer and diapers)
at the same time and trip.

The auto branding clustering question masks
the endogenous badge engineering phenomenon.
Did you know you could buy a Holden, Daewoo,
or an Opel at a Pontiac dealer ?

Which businesses do not cluster ?
Airports. If JFK was 250 feet from LGA,
that would be convenient.

There is a level of aggregation issue:
should a musician release a single at a time,
or an album ?

Does the publishing industry have anything like
a fashion industry season, where all the
would be big names push their new product at once ?

Few multiplex theatres are neighbors.

In the YVR case of M.E.C vs Coast Mountain Sports
and Taiga, et al, Taiga has followed M.E.C.

M.E.C. has huge street cred, and Taiga free rides
on the foot traffic. These stores are also en
route between UBC and the mountains, destipe
having moved several times.

See also:
Hotelling's law.

Here is part two of my post here. For this post, I will share a few side-comments but the key economic insight will be from the ideas expressed by the economist Steven Cheung (Steven_N._S._Cheung in Wikipeida) in a pair of Chinese articles dated Dec 20, 2002 and Sep 1, 2006 in analyzing the phenomena of clustering which he terms as his Three Laws of Clustering.

Note: I don't do Chinese translation for a living nor am I an economist. So please attributes all insights to Steven and all problems to my hopeless translation.

Steven argues that the reason why similar products (like computers, clothing, etc.) cluster (in Hong Kong and China (Steven always based his analysis and observation on real world examples)) is to reduce the "search cost" (a type of "transaction cost" (see wikipedia)) of the product for the customers who know exactly what they want to buy. Let me gives an example, when someone wants to buy a computer in Hong Kong, they know they can simply go to the Mong Kok district or Sham Shui Po district where they will have a cluster of computer shops to find themselves their computers. I suppose the same applies to Car dealers here. And I suppose the price of the items plays a role here too.

My guess for the lack of a cluster of grocery stores is that the potential savings in search costs is not too significant enough. In a sense, we don't tend to go into a few grocery stores to check for prices first before buying the required items.

Steven also points out other interesting phenomena of clustering in China. Like the world-dominating manufacturing of products in Wenzhou (a city in China) including cigarette lighters (quick google found that there are 500 lighter makers in Wenzhou! Try google "Wenzhou Lighter Makers Await EU Decision"), shoes and Xmas lights. As a retired Economics professor, Steven even suggested these topics as potential ideas for further research.

P.S. If you are easily offended, please skip this "P.S." Thanks. In my first post, I hinted that I might use Steven's prostitutes example to explain things. Well, it turns out that the example is more for "clustering" of similar "quality" of prostitutes (their beauty and looks) in an environment where information cost is very high (the Johns are embarrassed to approach a group of prostitutes in a Chinese hotel lobby to comparison shop for price). That was "first law of clustering" -- clustering of similar quality in an environment of high information cost.

P.P.S. For "completeness", Steven's "third law of clustering" quickly points to shopping malls, where "clustering" happens in a way to gather customers that are "simply browsing or checking things out", "those that want to buy a few things", or "a family shopping together for their diverse needs". And Steven terms this as another type of clustering.