Initial Thoughts on Supply Chain Complexity

For some time, I have been thinking about complexity in supply chains. Sometimes people say that supply chain management is not rocket science. My response is that supply chain management is in fact much more complex than rocket science.

I would like to have a better understanding of what drives complexity in a supply chain, and I have also been reflecting on if and how it could be possible compare two supply chains in terms of their complexity. In this post, I will give my own first shot at categorising drivers of supply chain complexity, and I share some thoughts on what I think a measure of supply chain complexity should satisfy.

Drivers of Complexity in Supply Chains

I think that complexity drivers take many different forms, depending on the type of supply chain and the specifics of the situation. In the following I suggest four general categories, which I think can be applied to classify complexity drivers.

Numerousness

A classic starting point when discussing complexity is the number of “stock keeping units” (SKU’s) or product identifiers i.e., the number of specific materials sold or handled by the supply chain. But there are also other relevant sets than just materials in the supply chain, such as processes, suppliers, locations.

Constraints

Different types of constraints add complexity to managing supply chains. Some standard examples are:

• Capacity
• Supply lead time (for inbound materials)
• Shelf life
• Storage conditions
• Regulatory requirements (e.g., taxes, customs, traceability)

Interactions

In the book “Complexity – A Guided Tour”, Melanie Mitchell describes how, in general complex systems, not only the number of elements drive complex behaviour, but interactions between the elements are perhaps even more important.

Some obvious examples of interactions in supply chains which drive complexity are given below, but it covers only a small fraction of interactions

• Correlation in demand
• Shared resources
• Relationships of materials being used for production of other materials
• Change over on production lines between different products
• Dependencies e.g., between processes in the chain and on key suppliers or customers
• Communication and exchange of information

Such relations mean that events affecting one item often also has derived effects on other items, and these effects can ripple throughout the supply chain in ways, which are difficult to assess and understand.

Variability & Uncertainty

Variability means that things are irregular. Demand varies or process time varies, or capacity varies due to vacation or maintenance. Variability drives complexity, even if it were 100% predictably, but this is usually not the case.

Variability is often coupled to uncertainty, meaning that the variability cannot be predicted precisely. In most supply chains there is uncertainty in demand, and supply processes can also have hard to predict variability, either in lead time or capacity availability. Uncertainty makes planning and managing of the supply chain even more complicated.

Structure and Dynamics

I think it is relevant to distinguish between complexity in structure and complexity driven by dynamics. Roughly, I think it can be said that numerousness and constraints are primarily related to structure, while interactions and variability are primarily related to dynamics. But structure also have significant impact on dynamic effects, so a clear-cut distinction is difficult to make.

Measuring Supply Chain Complexity

The examples of drivers of complexity which I have put forward above is far from comprehensive, and different supply chains will have different complexity drivers and will not be impacted the same from shared drivers. But would it be possible to create a measure, so that we could still quantify and compare supply chain complexity across industries?

Requirements

What would be the requirements for such a supply chain complexity measure?

Ideally, the measure should make possible to compare two supply chains and conclude which one is more complex. And the result should make sense.

The measure should be uniquely defined, meaning that if two persons were to measure the complexity of the same supply chain, they would reach the same result (assuming they had the same information and made no calculation errors).

Also, I think it should be possible to combine the measure for two parts of a supply chain (e.g., two functions within the same company) to get a total measure for the combined supply chain.

At the same time, the measure should be based on numbers, which are easily available, so that it is possible to calculate in practice.

A Possible Structure

The question is, if it is possible to find a measure, which satisfy all of the above? Probably not – a measure satisfying the first properties will probably be so complex that it cannot also satisfy the requirements of availability and ease of use.

A first idea would be a combination of measures related to the categories suggested above, e.g.

   

where N is a measure of numerousness, I is a measure of interaction and constraint complexity, and V is a measure of variability and uncertainty.

It probably requires a lot of work to define N, I, and V so that a measure of this form satisfies all of the requirements listed above, if it is at all feasible. However, it could be possible to find definitions which are simple enough that they could be estimated in practice to create a usable measure. I hope to get back to this in a later post, but any thoughts and ideas would be welcome.