Supply Chain Volatility

Updated: May 17

Dealing with volatility in the age of the pandemic.

The NY Giants, coached by Bill Parcells, won the Super Bowl in 1987. Their defensive coordinator was Bill Belichick. The defense employed a scheme that became known as a “bend-but-not-break” defense. The team gave up a lot of yardage between the twenties, but inside the red zone, where the field narrowed and the player density increased, the defense held strong and did not give up a lot of points. In other words, it did not break. I don’t think it was intended to be that way, but the results were such that the name stuck.

Today’s supply chains, under the unprecedented stress caused by the coronavirus pandemic, are spending a lot of time inside the red zone. The field is narrow and the stakes are high. Supply chains are giving up a lot of yardage, but they are not yet breaking. That said, too much time spent in the red zone will cause any defense to eventually break. Evidence of this is starting to crop up in food supply chains, as manufacturing sites have to shut down to protect workers from infection. My sincere gratitude goes out to the production workers, truckers and shippers, warehouse workers, and retail workers who are operating in the red zone every day.

In this article, I examine what’s changed and what companies need in their supply chain decision processes going forward.

Supply Chains of One are Now Supply Chains of Many

Trucks are going out full and coming back empty, warehouses are reducing their SKU counts, production has been shut down or repurposed, and front-line workers are getting paid premiums to show up. Yesterday’s world – just two months ago – was about making sure you got your Starbucks Triple, Venti, Half-Sweet, Non-Fat, Caramel Macchiato served quickly and at the perfect temperature. Today’s world is about making sure you have enough water, meat, and toilet paper. In a matter of days, supply chains went from focusing on the top of Maslow’s hierarchy of needs to focusing on the base, in other words, basic needs. Twenty years of supply chain tuning focused on higher and higher levels of personalization had to be thrown out the window.

For years I have been talking and writing about finer and finer segments towards supply chains of one. This has all been turned on its ear in the past two months; in fact, in the course of a week, the world’s supply chains had to adapt to a new reality. Demand for discretionary consumer products cratered; demand for consumer stables and basic human needs skyrocketed.

Below are some changes the pandemic has wrought:

  • SKU counts are down

  • Large batches are back

  • Lean is out

  • Redundancy is in

  • Micro segments are out

  • Macro segments are in

  • Algorithms are out

  • Humans are in

  • Efficiency is out

  • Precision is out

These changes are likely to be temporary (a big open question is how long). That said, some learnings from this will last forever. The pandemic casts a bright light on the need for companies to improve their agility (discussed below). It also highlights the need for companies to find the right mix of algorithms and human ingenuity. Self-driving is a great concept, but not when your algorithms have never experienced a cliff. Algorithms trained on data don’t know what to do when demand goes from 100 to zero in one day. They think such demand numbers are not real. Humans know they are very real.

The past twenty years of supply chain management has been about delighting the individual – the supply chain of one – and providing as much choice as possible – the product, the time, the price, the location. Companies were on a long journey to profitably provide this choice. In order to do this, everything had to operate with increasing levels of precision – precise products to serve consumer demand, precise delivery windows, precise loading, putaway, picking and packing, and precise production cycles. This precision has been completely upended in the past two months.

The pandemic has also caused significant dislocations in some supply chains, particularly food, in which there is an excess of supply but no demand channels to absorb the supply. Perishable products, including milk, lettuce, and pork have had to be destroyed. In some cases, demand is there – such as the case of people in need of food but no money to pay for it – but no physical channels by which to supply such demand. There is no economy-level dynamic demand-supply structure to deal with situations like this. Perhaps there should be. Various individuals and organizations have formed coalitions to fill the gap, albeit without the scale to do it nationally. The economy – like a unicycle – does not handle slow well.

This brings us to a discussion of variability and volatility. These terms are widely used, but often misunderstood. Here we attempt to provide some clarity in the context of the pandemic-induced stress in today’s supply chains.

Variability and Volatility

Variability and lead time are two of the most important facts of life in supply chains. Inventory is a result of both of them. Long lead times and high variability result in high inventory. At zero lead time and zero variability, inventory goes to zero.

Variability is inherent in all processes. There may be a process in the world that has zero variability, but I’ve not seen it yet. Understanding and managing variability is based on statistical analysis of process capabilities. Process capabilities also change over time as the result of drift, or as the result of disruption.

And lead time? Well teleportation doesn’t work well even in science fiction movies, so lead time is going to be with us for a while. Globalization over the past thirty years has significantly increased lead times; the geopolitical climate in the past several years has questioned this trend, and may very well lead to a reversal with more sources of supply closer to sources of demand.

Variability has a multiplicative effect on system-wide performance. For example, if two sequential processes both perform at a level of 90%, the system has an overall performance of 81% (90%*90%). Supply chain practitioners and industrial engineers seek to continuously improve process capabilities to drive variability to as close to zero as possible. The difference between actual variability and zero variability must be buffered with some combination of inventory, capacity, and time. The key thing is consistency; if I know a process is always performing within a certain band, I can plan and manage that. If a process fluctuates wildly without any statistical pattern, it is very difficult to manage.

What about volatility? “Volatility” and “variability” are often used interchangeably. However, volatility is a significantly different thing than variability.

Volatility is typically anything that occurs outside of the bounds of how the supply chain system was designed to operate. Variability, on the other hand, is the normal fluctuation that occurs in all processes along the supply chain.

In other words, variability is normal; volatility is not normal. Variability can be planned for. Volatility has to be reacted to. Variability can be buffered; volatility requires intervention.

Supply chains professionals deal with volatility every day: an important customer asks to “drop in” a large order inside of lead time; a critical parts shipment is stuck in port; a supplier experiences an unplanned outage; a hurricane hits the Gulf of Mexico; a celebrity endorses your product on social media.

The above simple diagram illustrates variability and volatility. For discussion purposes, let’s assume it represents demand for a particular product. The chart shows a variability of between plus 15 and minus 15 around an average value of 100. It also shows a volatility event of 40% upside relative to the average value and an extreme variability event where demand doubles, and one where demand drops to zero.

Now, let’s assume that this extreme volatility was not just an event, but it was persistent. In other words, demand for some items rose to more than double their average value and stayed there for weeks, while demand for others dropped to zero and likewise stayed there for weeks. This is the volatility caused by the pandemic: it is unique it its magnitude and its duration. A key question for businesses is when, if ever, demand and supply will return to their normal levels.

Human and algorithm decision making have contributed to the amplitude of the volatility caused by the pandemic. Even Amazon, recognized as supply chain master by Gartner, was ordering early and often, in a real-life example of the beer game. Processes that for years operated within precise bands, were now oscillating back and forth like the Tacoma Narrows Bridge.

So where are we headed now?

The Future Is One of Many Possible Scenarios

Right now, no one knows what life will look like a month from now, much less a year from now. The future is one of many possible scenarios. If you ask ten experts what the future will look like six months from now, you will get ten different answers. Probably one of them will be right. Given this uncertainty, what should companies and their supply chains do?

What if no matter what the future is in six months, you were able to handle it and even improve relative to your competitors? This is called agility. Thirty years ago, when Rick Dove and others from the Agility Forum at Lehigh University were evangelizing agile manufacturing, they defined agility as the ability to thrive in a continuously changing, unpredictable environment.” This is a simple and good definition. But what does agility look like?

I sit on the board of Kinaxis, which makes software that allows companies to thrive in a continuously changing, unpredictable environment. John Sicard, Kinaxis’ CEO, likes to ask the question: “would you rather be infinitely accurate, or infinitely agile?” One of the ways Kinaxis enables agility is to plan for and anticipate a future of many different possibilities. It allows businesses to create a near limitless number of scenarios, to understand their potential impact, and how the business must operate under each. They also couple this capability with a robust collaboration capability, which enables the reduction of decision-making variability.

This capability is such that it is unlikely that when the future arrives it is not something that has already been seen and anticipated by the organization. But what about once-a-century events like the current pandemic? Could this have been anticipated? Hindsight of course is 20-20, but the answer is yes (see the “Black Swan?” below). However, even if an organization did not anticipate the pandemic or its effects on demand and supply, once it arrived, a limitless number of demand and supply scenarios can be very valuable.

One reason for this is that for the past several thousand years, at least since the creation of empires, the future has proven devilishly difficult to predict. At the same time, a principal reason for human success (indeed, any advanced biological system), has been its ability to learn and evolve. In other words, its agility. Part of that learning is a recognition that while it’s important to try to predict the future, it’s even more important to be able to adapt to a wide range of possible futures once one of them arrives.

Scenarios themselves include a prediction. When a company operates against one prediction of the future, the likelihood of that one prediction being right is quite low, particularly in times of high volatility.

When a company operates against many predictions of the future, it is highly likely that one of them will be correct. And, even if an outside-the-domain-of-scenarios event does occur (such as the current pandemic), you are now operating from a different point on the chess board from which you can establish a new set of scenarios and reconnoiter from there. In many ways, this is the difference between a deterministic (single possibility) and stochastic (range of possibilities) view of the world. A deterministic view is very brittle to deviations while a stochastic view is very resilient.

This is not to say that accuracy is not important. As most people in business know, accuracy should always be a goal when predicting the future. That said, accuracy is fool’s gold; it is a fleeting, moment-in-time achievement. It is like Sisyphus rolling the rock up the hill, only to have it roll back down, requiring a never-ending restart; the moment accuracy is achieved, the underlying problem shifts, causing the rock to roll back down. And, unfortunately, in many cases, this accuracy achievement creates a false sense of security that is realized in surprises weeks, months, and even years after the fact.

Another thing that scenarios enable is accumulation of knowledge and the placing of bets against that accumulated knowledge. Last year, I read the book “Thinking in Bets,” by Annie Duke, former world poker champion. She has parlayed her success in poker into a consulting and speaking career based on the idea that all decisions are essentially bets. Cycles of learning allow you to build up a repository of potential scenarios and to place bets at times when the odds start to move in your favor. If there are four Kings in a 52-card deck and you pull 30 straight cards without having seen a King, then you better start planning to see one. Likewise, paradoxically, a long period of low volatility may be the best time to plan for a spike.

Triage Phase is Over

When doctors have to treat patients that are experiencing extreme volatility – for example, multiple critical organs are failing at the same time – they have to make rapid, intelligent decisions regarding where to attack the problem first. Likewise, when their facilities are overtaxed, they have triage protocols for determining the sequence across patients, in an attempt to effect the best overall outcome.

It seems that supply chains are largely through the triage phase of dealing with the effects of the pandemic. Demand priorities, allocations, and lead times are now operating in a new normal. Amazon is a proxy for what is going on in supply chains. Some of Amazon’s 500M+ SKUs are temporarily not available; the ones that are available are given different priorities. Sellers of discretionary items that rely on FBA (fulfillment by Amazon) were temporarily on their own for fulfillment. Prime is no longer one day; it is now five to seven days, which is what it typically was before Prime was introduced in 2005. Demand incentives, promotions, and Prime Day are indefinitely on hold. These moves are all classic demand management moves, albeit the reverse of what we have seen for two decades (and of course done on a massive scale). The pandemic has turned demand incentives into demand disincentives.

Note: Amazon has just announced that Prime will start evolving back to its normal delivery. Getting there required two months and the hiring of thousands workers – a remarkable feat.

Black Swan?

Many people describe the pandemic as a “black swan” event. However, Naseem Taleb, who wrote the book “The Black Swan,” says no. Says he: “It’s a white swan. How can it be a black swan when movies have been made about it?” Good point. Let’s assume for a moment that a pandemic of this magnitude is a once-in-a-century event. If the last one we saw was in 1918, then it was almost a certainty that one was coming soon. As we all know by now, Bill Gates said as much in 2015.

What may be more of a black swan are some of the consequences of the pandemic, as pointed out by Alex Danco. Danco suggests that the virtual shutdown of large parts of the US economy and the resultant unemployment are 30 sigma events, meaning they are, in his words “outrageously unlikely, at universe-scale.” He is effectively saying that even if the pandemic is a white swan, who predicted the economy would be shut down? Of course, one could also say we had a case study right in front of us in the form of large parts of the China economy. Furthermore, once the pandemic door was opened – like a door opening in the Monte Hall Problem – the conditional probability of a number of other doors increased significantly, including record unemployment and the price of oil dropping below zero.

Now that we are in this predicament, it seems virtually no one is predicting that it will go back the way it was as quickly as it came. This brings me back to the need for all companies to enable agility. I will close with a final quote from Alex Danco:

“The future hasn’t happened yet. More things can happen than will; and more things will happen than have happened. The future holds infinite possibility, while the past only offers a finite set of examples to learn from.”

I cannot think of a better way to state the case for infinite agility.