An Application of Central Place Theory: Merangin District, Sumatra

Map from here.  Note Jambi Province 

This week I'm down in Bangko, the capital of Merangin district for some research work.  Merangin is just east of Kerinci district, where I live.  It's one of the districts of Jambi province, which you can see from the map to the left.  One of the things I'm interested in is the long- and mid-range planning documents from the regional planning authority (BAPPEDA).  These documents have all sorts of information about the socio-economic aspects of the district, as well as information concerning the policy orientation of the government (i.e. what are the problems facing the district and what are the steps the government needs to take to address the problem).  Knowing these aspects of the district, which has about 20% of its territory covered by Kerinci Seblat National Park, helps me to understand the roots and context of current and potential conflicts between the people and institutions in the district and the park.  And, as a geographer, I like to look at the plans because they have all sorts of cool maps.


One section of the district's long-term (15 year) development plan describes the hierarchy of cities in the district.  As I read through the description, I was reminded of one of the classic models of geography: Walter Christaller's Central Place Theory.

Central Place Theory

Awesome GIF from Wikipedia

Walter Christaller was a German geographer active from the 1920s until around the 1950s(1).  He is most known for his Central Place Theory, which is a staple of introductory geography textbooks.  Central place theory is a model that attempts to describe (and predict) where and why cities of different sizes will be found on the landscape, and how they relate to one another.  The model has some basic assumptions, like most simple geographic models (recall my explanation of Von Thunen's model in a previous post).  It assumes that the world is an isotropic plain and that resources and people are distributed equally.  All people make rational decisions, and they will go to the closest place offering the goods or services they need.  Since the landscape is an isotropic plane, they can take the shortest distance to their destination.  

Christaller divided goods and services into different categories.  Simple or lower order services include things like markets and gas station and can be found in a lot of locations.  More specialized  or higher order services like universities are fewer and farther between.  Directly related to the order of service are two concepts which Christaller pioneered: threshold and range.  Threshold refers to the minimum population required to support a certain type of service.  Think about a minimarket, for instance.  They are everywhere because it doesn't take too many customers to keep a minimarket in business.  The threshold is low.  Range, on the other hand, is the maximum distance people are willing to travel to avail themselves of a service.  Returning to our minimarket example, you're probably not willing to travel too far to find a Circle K.  So the range is low as well.  However, there are some services, like movie theaters, ice skating rinks, Ikeas, etc, that you are willing to travel farther to access.  At the same time, these services need a lot more folks to keep them in business.  So the threshold and range are both high.  Cities or towns that provide only simple services are low order settlements, whereas larger cities that offer a more comprehensive range of services are high order settlements.  

Diagram from Hofstra University's excellent website.

Christaller's basic idea is that in a given region, there will only be a few high-order settlements, but there will be lots of lower-order ones.  What's more, settlements will be spaced equidistant from one another, and higher-order settlements will be surrounding by a number of lower order settlements, which rely on the bigger city for rarer goods and services.  There are several levels in the hierarchy, from huge metropolises on down to regional cities, towns, villages and hamlets.  You can see a diagram of the arrangement of the various classes of settlement to the right.  Each of the different types of settlement has a different market area; large (higher order) cities have bigger market areas.  Remember that people will go to the place closest to them offering the desired good or service; this accounts for the regular spacing (2).  In addition to this market principle (known in the theory as the k-3 principle) Christaller also described transportation (k-4) and administrative (k-7) functions, but we don't need to go into those here.  

Christaller used Central Place Theory in his role as a planner for several evil regimes in Europe.  Since the introduction of his theory, scholars in geography have attempted to apply it to other instances as well in order to prove or disprove its merits.  It has also influenced the discipline of regional planning, as we'll see below.  

Central Place and Merangin District?

The capital of Merangin district is Bangko.  Bangko is surrounded by a number of smaller towns, which in turn are surrounded by villages.  The district is mainly agricultural but there is some industry as well (most of which centers on the processing of agricultural commodities).  As you can imagine, there is a natural ranking of services in the settlements of various size in Merangin.  For instance, every village has a place to buy instant noodles or cell-phone credit.  Larger towns act as transportation hubs, and the capital city functions as an administrative, management, and financial center.  That's all common sense, and in my mind Christaller doesn't deserve much credit for describing the natural way that markets work.  

What is remarkable, though, is that way that the planners have adapted Christaller's ideas in their development planning.  Within Merangin district there are 9 subdistricts (kecamatans).  One of the problems the planners are hoping to solve, though, is an imbalance in the level of development and economic opportunities between subdistricts.  They are also hoping to develop a district-wide system where each subdistrict develops according to the resources located in the subdistrict; for instance, the subdistrict of Jangkat has lots of land with good soil and not many people, so the planners are hoping to increase the amount of palm oil and rubber plantations there.  Likewise, each subdistrict should be part of a larger system with the district capital (Bangko) at the top.  

In order to facilitate this vision, the planners have created a hierarchy much like Christaller's.  In the district system Bangko, the capital, is referred to as an Order 1 (Orde 1) center.  It is a center of transportation, located on the main overland route to the provincial capital at Jambi as well as the roads to other districts (Sarolangun, Bungo, and Kerinci).  It has the most complete economic facilities and is also a center of banking, trade, management, and communication.  In the words of the planning document, it is the Center of Regional Activities (Pusat Kegiatan Wilayah, PKW).  Beneath Bangko in Kota Rantau Panjang, the capital of Tabir subdistrict.  Rantau Panjang is an Order 2 (Orde 2) town in the district framework and is a center of local activities (Pusat Kegiatan Lokal, PKL).  It has good infrastructure and provides subregional services.  Then at Order 3 (Orde 3) we have Sungai Manau and Pamenang.  Both of these towns are about equidistant from Bangko.  In the planning framework they provide lower-order services to their hinterlands and are supposed to spur development in the areas immediately around them.  Lastly we find Muara Siau, a smaller town at Order 4 (Orde 4).  The quality of infrastructure is lower, and it provides still lower-order services to the mainly agricultural hinterlands surrounding it.

The planners of Merangin district use this framework to help plan what sorts of projects need to be developed in various places.  For example, they are working to develop higher-order transport facilities in the district capital, but in the lower-order centers they plan smaller-scale projects aimed at improving specific aspects of the economy in those places.  This helps them to distribute resources, like money and equipment, in a more efficient manner aimed at developing the district as a whole.  You can see the results in the map below, which divides the entire district into development zones, each with its own set of projects and targets.  I've labeled the various centers along with their order.  The spatial arrangement somewhat mirrors that predicted by Christaller, but more important, at least to me, is the function of the various centers.

We can see from this example that geographic models not only help us understand spatial distribution and organization from a theoretical perspective, but they are also important tools in policy making.  Central Place Theory has clearly influenced the planners in Merangin district and has helped them to formulate a development plan that specifically addresses the strengths and weaknesses of each subdistrict.

Notes

(1)  Christaller was not only a NAZI, but a COMMUNIST as well.  I faced something of a moral dilemma when writing this post....I am still somewhat ambivalent about describing a theory that was used primarily for organizing hostile occupations and oppression

(2)  Christaller adopted hexagons rather than circles because hexagons nest together perfectly without any overlap or gaps, unlike circles.

(3)  I am basing this on the 2006 long-range plan; in 2008-9 the authorities experienced a frenzy of new district creation, and now there are 26 kecamatans.

Why Does This Happen? Small Business Clustering in Indonesia

Tire vendors as far as the eye can see! Scenes like the one to the left are common here in Indonesia; when you come here you'll likely encounter agglomerations of merchants offering the same good or service. This include rubber stamp makers, license plate makers, sunglasses salesmen, florists, furniture makers, and others. At first glance it might doubt the logic of 15 nearly identical keymakers setting up shop next to one another; surely there can't be enough demand on a daily basis to support them all. And why would someone set up shop right next to 14 other guys doing the same thing? Doesn't it make sense to start your business someplace else to draw customers from a different location rather than join in the apparently vicious competition? Well, like most things in life, there's more here than meets the eye, and a basic understanding of some simple geographic principles helps us make sense of it all.

Historically there are a lot of examples of like professions being grouped into certain districts. Sometimes this has more to do with regulation than anything else. For instance, in the royal cities of Java professions such as gamelan manufacturers were all located on a particular street. Other places, like New York City's Garment District, gradually became centers because different types of businesses in the same industry (weavers and tailors, for instance) clustered together to minimize transport and other costs. But neither of these explanations seems to help with the keymakers and tire sellers. But there are other possible explanations!

One possibility was predicted by economist Harold Hotelling in 1929 (Hotelling’s Model). Hotelling asks us to imagine a beach filled with thousands of people (which shouldn’t be too hard in Hawaii). An enterprising businessman decides to open an ice cream stand somewhere on the beach. Where do you think his stand is most likely located? If you answered “it the middle”, you’re correct. Now, due to the success of our ice-cream vendor, someone else gets the bright idea to open another ice cream shop to compete. Where do you think our new competitor is most likely to set up shop? Take a look at the diagram below and think for a minute.

According to Hotelling (and according to our common sense), the new vendor will build his stall right next to the existing stall! The reason for this is because he/she can grab half the market from the first competitor (we're assuming, of course, that the buyers only think about how far they have to walk). If an additional competitor were to set up shop, he/she also would tend towards the center (see the diagram below). This might be one way to explain the concentration of tire vendors, key makers, etc. But I suspect that there are other factors at work here. To shed some light on this curious phenomenon, I set out on an adventure of discovery.



My first stop was a long line of 10-15 keymaker’s stalls on Jalan Demangan. Here there wasn’t really any difference between the various products and there didn’t seem to be much pressure to attract customers. After buying a key (rp 5000) to break the ice, I talked with the keymaker.

“Why are there so many keymakers here?” I asked.

“Because this is the key place”, he answered.

“Yeah but why is this the key place?” I prodded.

“Because all the keymakers are here”, he responded.

After this round of circular logic was completed we got down to business. There isn’t a whole lot of cooperation between the keymakers, but apparently there is some. He said he sells around 20 keys a day, which, at 5000 rupiah per key, is somewhere around $13, enough to pay for life’s necessities. He told me all the keymakers are pretty much the same and that he mainly gets customers through repeat business or word of mouth.

Another explanation is that the various vendors are related somehow. It might be that a parent sets his children up in the same business. So next stop was the fruit stand where I do my buah shopping. Although there are several stalls in a row, they are all owned by people from the same family, and so there is a good bit of cooperation here. I saw a variation on this theme on “tire alley”. Some are relations. Others come from villages far away. The way this works is that a “pioneer” opens up a shop, has success, and sends word back to the village, and is soon followed by others who set up their own shops. There are also a couple of different dynamics at work amongst the 20-30 vendors here. The tire salesmen cooperate with one another as well; they even have an association responsible for keeping the area clean and resolving disputes. There was also a high degree of differentiation in terms of products offered as well as equipment used.

The flower vendors told me they had traditionally all been grouped together, but they came to their current location after being displaced from their original spot downtown, which the government turned into a parking area. They were given their new area as compensation. Here again there seemed to be some cooperation as well as family connections among vendors. The furniture guys cooperate the most (based on my short interviews). This is because they are all from the same area in Bantul, a district outside the city limits. I asked them the same question I asked the others: “What if a hotel managers comes and wants to buy 100 cabinets (keys, flower arrangements, etc)….Do you share the work with your neighbors or do you do it all yourself?” Though all the others responded that they would probably fill the order themselves, the furniture guys said they would share the order.

Keith's Rule...

Another reason for clustering is that groups of small businesses together seem, to the rest of the economic “system”, as one large business. This might help them draw customers from further away. Thus their range increases but their profitability threshold remains low (1). Think about how car dealers tend to cluster together. Customers know that they can see a lot of cars without having to go all over town. To illustrate this, I invented a little model. If we imagine the world as a flat isotropic plane (2) with an even distribution of population we can represent the market area as a circle. As we know, the area of a circle is the radius of the circle times the square of Pi. Now for the purposes of my model I've substituted the business threshold for the area. That way I can solve for the radius of the circle, which gives me my hypothetical range. Each additional shop increases the threshold, but it also increases the range. In the table below you can see how it works. The marginal range is simply the additional market radius required to support an additional vendor.

Number of Shops	Threshold	Range	Marginal Range
1	5	1.26	NA
2	10	1.80	.54
3	15	2.19	.39
4	20	2.52	.33
5	25	2.82	.30
6	30	3.09	.27

This is a pretty interesting result. As you can see from the table above, as the number of shops gets bigger, the marginal range gets smaller. That means that the market radius required to support each new shop actually decreases! Eventually this will tend asymptotically towards 0, which means that you can keep adding shops forever without hurting the market (3)! Of course, the real world doesn't work this way...as you get further away from the center of the city the population density decreases, and we can't really assume that the market doesn't get saturated at some point. But this does show us that business clustering isn't necessarily a bad thing.

Anyway, back to the story. When I talked to the sunglasses vendor he told me that there isn't much cooperation between the different vendors, but they all cluster together for the reason I just explained. Thus there's a symbiotic relationship even if there is no cooperation.

This also presents the opportunity to think about the nature of competition. All the key makers were the same; they have pretty much a standard service and inventory that they offer, and there isn't any difference between them. This would be similar (4) to what economic geographers refer to as “perfect competition”. In perfect competition there is no product differentiation. This means that the good or service sold by one vendor is no different from those sold by all the other vendors. In addition, there are no barriers to entry in the market place (anyone can start a business) and there is easy entry and exit from the market. The tire guys at first glance looked like perfect competition as well, but when I started talking to them I realized that there are a lot of differences between them, and they try to find niches by offering different products. There is also significant variability in the tools they use; some have state of the art equipment while others use older equipment. This is closer to monopolistic competition. Look for a future post describing these concepts in deeper detail.

Anyway, I had a great time biking around town talking to folks. They were universally friendly and more than willing to talk. I also wanted to know if the success of the business depends on the location in the line of stalls, but I forgot to ask. But all in all, this was a pretty neat opportunity to apply some theories and ideas from economic geography.

(1) Threshold and range are terms from Walter Christaller’s Central Place Theory. Threshold refers to the minimum market population or income required to support a particular distance. Range is the maximum distance customers are willing to travel to buy something. Can you think of some types of businesses that have short ranges? What about long ranges? What kind of business has a low threshold? A high threshold? Are there any obvious connections between these two concepts? (Diagram courtesy of Dr. Jean-Paul Rodrique, Department of Geography, Hofstra University).

(2) An isotropic plane is pretty much just a flat surface with no variations. Isotropic planes are one of the favorite tricks of geographers in making simple models. Remember that models are merely representations of reality, and they vary in detail. This is a very simple model, much like Von Thunen's market rent model.

(3) My calculus is really rusty, but I think a way to represent this is the following (I hope this is right because it took me 20 minutes to figure out how to use the equation tool in Word:



(4) Instances of perfect competition are rare. A closer example here might be the becaks I described in a previous post. Can you think of an example of perfect competition in Hawai’i? What about examples of monopolistic competition or oligarchy?

Acknowledgments:

Mahalo Nui Loa to Dr. Matt McGranaghan, Department of Geography, University of Hawai'i, and Dr. Gary Fuller, Professor Emeritus, Department of Geography, University of Hawai'i.