Impact of Macro Trends on Supply Chains: Decentralization of Production

By Dr. Chris Caplice is Executive Director. MIT CTL, March 13, 2014,

Traditional mass-production manufacturing plants lower the unit cost of production but in many cases are too inflexible for today’s fast-changing markets. Alternative manufacturing models are emerging that are much nimbler; a development that spells big changes for supply chain management.

In my previous blog post Impact of Macro-Trends on Supply Chains: Digitization of Products, I looked at how product digitization might impact distribution networks. This final post in a four-part series on broad trends that could transform supply chains considers the effects of decentralization.

One of the primary reasons manufacturing is often carried out across the globe is to achieve economies of scale: a single massive plant can dramatically reduce per unit production costs.  Economies of scale arise when there is a large fixed cost that can be allocated over a large number of items.  Two obvious examples are specialized production lines and automated distribution systems.  If these costs were purely variable, then there would be little economic incentive to centralize production.

While centralized or mass production does lower the unit cost, it increases the cost of customization.  There is an inherent loss of flexibility.  Being able to leverage the same equipment or investment over and over creates this per-unit savings.  The initial cost of automating a production line can be exceptionally high, but it is worthwhile if it can be used for thousands of items.  However, if the product changes, the system typically needs to be re-tooled – a very costly and time consuming process.  The same is true for automated warehouse systems.  Changing packaging sizes or dimensions can result in significant costs.

What if new manufacturing technologies and production processes could diminish the benefits of economies of scale while simultaneously decreasing the costs of customization?  The result would most likely be a larger number of smaller manufacturing sites located, logically, closer to centers of consumption.

There are several examples of emerging technologies that have this potential.  Here are just three.

The first example is additive manufacturing (AM) that I discussed in my last blog post.  AM allows formanufacturing in batches as small as one.  The set up time and switching costs are exceptionally low.  The per-unit manufacturing costs are not as low as a mass manufactured item – but there is incredible flexibility and capability to customize.  Also, for items with very sparse demand, the cost of producing items as needed can be lower than the sum of the costs associated with manufacturing, holding, transporting, and product shrinkage.

A second example is where technology has lowered the cost of automation and allows for easier scaling.  Kiva Systems manufactures and sells automated warehousing solutions that feature autonomous mobile robots that essentially bring the product to the picker.  Because the system relies on autonomous, yet coordinated, robots to assist in the distribution, it is easy to start small and add units as needed.  This is especially helpful for small to medium sized warehouses.  There is a reason why Amazon recently acquired Kiva as it begins to role out more regional fulfillment centers poised for same day delivery.

A third example is Baxter.  Manufactured by Rethink Robotics, Baxter is a robot that is designed to work side by side with humans and perform a wide range of repetitive manufacturing tasks.  It is exceptionally flexible and can be trained by line workers.  The idea is that Baxter has the capability to learn “on the job” and therefore grows with the operations,  taking on new tasks or modifying its job as needed.  This reduces the fixed cost of automating a manufacturing line and allows a firm to add automation as needed – similar to what Kiva does for warehouse operations.  This again is automation with a lower initial fixed cost.

These are just three examples of technologies that change the economics of manufacturing and distribution.  There are many more out there – and even more that will come out of the labs over the next year or so.  The general trend is to utilize technology to decrease the cost of flexibility while not dramatically increasing the per-unit cost of manufacturing.

So, will decentralized manufacturing disrupt the dominant design of distribution?  The short answer is yes.  It could change a lot.

First, it is important to note that as manufacturing processes move to smaller scale, the input materials become more expensive.  In fact, the input becomes more like a finished product.  A good example of this is personal photography where the printing of photographs has shifted from large-scale film processing machines to individual home printers over the last 10 years.  The input for the typical film processing machines consist of bulk chemicals and ink while for home printers it is individual ink cartridges where the cost per developed photograph is up to 10x higher than traditional methods.  Even with the higher per-item cost, it is preferred since there is less waste.  One does not need to print out the entire roll of film to find that one photograph to keep.

Another example that is close to home is the single cup coffee machines (such as Keurig or Nesspresso).  The machines are very convenient and allow the user to stock a wide variety of different flavors of coffee.  As with the photography example, however, the cost of the input materials is much higher for the smaller scale production.  The cost of coffee for use in a traditional drip coffee maker is between $5 and $7 per pound while the cost for a Keurig K-cup is equivalent to paying between $35 – $50 per pound!

Manufacturing closer to the point of consumption places more emphasis on the last mile delivery – the most expensive and most environmentally harmful transportation leg.  This also places more manufacturing (albeit smaller size) closer to residential areas; adding to congestion and exasperating urban mobility.

This could also translate into less transportation of finished goods over long distances.  To feed the smaller scale and dispersed manufacturing sites there will be more bulk transport of raw materials.  But, as shown in the two examples above, the input to a smaller scale process has more packaging and bears more resemblance to finished goods than raw materials.

In short, decentralization could dramatically impact and change the dominant design of distribution within the United States.  One of the key drivers of decentralization is the digitization of the product.  This enables different manufacturing techniques that are more amenable to smaller scale production.  Having smaller production sites closer to population areas will reduce the level of finished goods shipped across the globe and will increase the length of haul of the input materials.  There will also be an increase in local delivery in congested urban areas.