Decision Economics and Yossi Sheffi’s Network Lens
Supply chain used to be an internal concern, the sort of thing only logisticians and operations people talked about. Over the last twenty years it has become a public fascination. Port closures, pandemics, wars, factory fires, container shortages – suddenly everyone has an opinion on how global production and distribution should work. In that public conversation, few academics have done more than Professor Yossi Sheffi at MIT to explain to a broad audience what is actually happening and why it matters. His work on resilience, logistics clusters, and sustainability has shaped how many executives, journalists, and policy makers think about supply chains.
In my book Introduction to Supply Chain, I tried to do something more modest and more technical: to restate supply chain as a practical branch of economics and to show how software can turn that economics into everyday decisions across inventory, production, and pricing. The aim of this essay is not to critique Professor Sheffi, but to clarify my own position by placing it next to his. We look at the same world. We agree on more than we disagree. But we stand in different places and use different lenses.
Two vantage points on the same landscape
When I read The Resilient Enterprise or The Power of Resilience, what comes through first is the wide-angle view. Sheffi’s attention is on networks and shocks: how earthquakes, tsunamis, labor disputes, cyberattacks, and pandemics ripple through global supplier networks and transportation systems; how some companies shatter while others bend and recover. His heroes are firms that invest in anticipation rather than improvisation, that build flexibility, redundancy, and robust culture before disaster strikes.
In Logistics Clusters, the focus widens further. Instead of looking at one company’s network, he looks at entire regions – Memphis, Singapore, Rotterdam, Zaragoza – and explains why logistics activities concentrate in certain places, how this concentration feeds back into economic growth, and what this means for governments and industries. Supply chain is not just trucks and warehouses; it is a spatial and institutional phenomenon.
Finally, in Balancing Green, supply chain becomes a stage on which the tension between profitability, jobs, and environmental impact is played out. Companies must decide when to invest in greener options, when to hold back, and how to navigate the gap between what consumers say in surveys and what they actually pay for at checkout.
My own work starts at a different scale. Instead of looking first at the global network, I start with a single decision: one more unit to buy or not, one more pallet to move or not, one more price change to apply or not. Supply chain, to me, is the economic discipline that decides how every unit of stock, every hour of capacity, and every dollar of capital should be committed, under uncertainty, to maximize the long‑term payoff for the firm.
If Sheffi is standing on a hill drawing a map of the whole territory, I am sitting at a workbench with a microscope, looking at the wiring of the decision circuits inside the firm.
What is uncertainty, really?
The word “risk” appears frequently in both our work, but we use it in slightly different ways.
Sheffi is preoccupied – rightly – with big, discontinuous shocks. He documents how firms faced port closures, terrorist attacks, typhoons, factory explosions, or financial crises, and either broke or bounced back. A theme that appears again and again is that resilience is built in advance: flexible sourcing, alternative sites, thorough business continuity planning, and a culture that knows how to react quickly under stress. Risk, in that frame, is largely about the tail events that could cripple the entire network.
I am more obsessed with the everyday randomness that never makes the news: the forecast that was off by 20 percent, the truck that shows up three hours late, the supplier that quietly slips from a ten‑day lead time to twelve. None of these events is dramatic. Taken one by one, they look like noise. But they compound. They accumulate into stockouts and gluts, into overtime and idle capacity, into millions of euros or dollars evaporating every quarter.
My starting point is simple: the future never arrives as a single scenario. It arrives as a distribution of possibilities. For every product at every location, there is not “a demand number” next week; there is a range of plausible demand levels, each with a probability attached. The same holds for lead times, returns, and many other drivers. The role of analytics in supply chain is not to predict a single future and push everyone to align to it, but to accept the distribution and ask: given this spread, which decision today is likely to create the best economic outcome?
Sheffi’s scenarios and my probability distributions are two ways of wrestling with the same reality. His concern is to ensure that the network does not crumble under extreme stress. My concern is to ensure that, even on quiet days, the firm is making the right bets with its inventory, capacity, and prices.
Decisions, not flows
When executives say “supply chain,” they often mean the physical flow of goods. Trucks, containers, pallets, conveyors, airplanes. Sheffi emphasizes how these flows connect countries, clusters, companies, and consumers. I do not disagree, but I find it more fruitful to think in terms of decisions.
Every flow is the result of a decision: a purchase order, a transfer order, a production launch, a promotion, a price change. Those decisions may be made by people, by systems, or by a tangled mix of both. The question is not only whether trucks are moving, but whether the decisions that trigger those movements are, on average, good ones.
A good decision, in my vocabulary, is one that uses scarce resources – cash, capacity, time, goodwill – in a way that is likely to generate the best stream of future cash flows, once risk is taken into account. We can phrase this in more technical language as a risk‑adjusted rate of return on each marginal commitment, but the intuition is straightforward: if I use one additional unit of cash to buy this item here, rather than that item there, or to lower this price rather than that price, where does it work harder for me over the next months or years, given all the uncertainty?
This way of looking at supply chain has two consequences. First, it blurs the boundary between operations and commercial decisions. Pricing, assortment, and promotions are not separate worlds; they are simply different levers for making the same economic trade‑offs. Second, it makes software central. Human planners can reason about a handful of such trade‑offs; a large supply chain needs millions of them processed daily.
Here I depart most sharply from mainstream practice, and perhaps more than from Sheffi himself. Most enterprise systems and many planning tools are glorified ledgers and dashboards. They record what happened and display it. They rarely take the final step of issuing decisions automatically, at scale, under uncertainty. For me, the heart of modern supply chain is that final step.
Technology: architecture versus capability
Sheffi has written and spoken extensively about the impact of digital technology, automation, and artificial intelligence on supply chains. He describes warehouses filling with robots, trucks and containers tracked in real time, predictive algorithms helping firms anticipate disruptions, and new types of jobs emerging around these systems. He insists, quite properly, that supply chains remain human networks: technology amplifies what people can do rather than replacing them wholesale.1
I share that view of technology as an amplifier, but my focus is narrower. I care a great deal about the internal architecture of the software that suggests or issues supply chain decisions. Is it merely presenting historical data and asking humans to “decide,” or is it actually computing and executing decisions itself? Can it express complex trade‑offs in code, or is it forcing everything into simplistic safety‑stock formulas and frozen “service levels”? Can the team that understands the business also change the logic, or are they at the mercy of vendor roadmaps and configuration wizards?
In other words, I worry about the programmability of the system that sits between data and action. A supply chain where decisions are encoded in a handful of formulas and thousands of spreadsheets is fragile in a very different way than the one Sheffi describes. It is fragile to subtle errors, to silent assumptions, to the slow creep of complexity that nobody fully owns. The cure, in my view, is not to abandon automation but to make it explicit, auditable, and directly tied to the economics of the business.
From Sheffi’s vantage point, the question is more often: which technologies should a firm adopt to become more resilient, more visible, more sustainable? From mine, the question is: given that you will inevitably have software in the loop, what should that software actually do, and how should it be shaped?
Geography and clusters versus abstraction
One of Sheffi’s enduring contributions is to remind us that supply chains are rooted in place. Logistics clusters arise where infrastructure, companies, labor markets, and institutions reinforce each other. They create jobs, shape trade flows, and give regions strategic importance. Governments care about them, and rightly so.
My own writing is comparatively silent on geography. That is not because I believe it is unimportant. Rather, geography enters my work as an input rather than an object of study. Transit times, port capacities, congestion patterns, local regulations – all of these become parameters in the decision models. I do not ask where to build the next port; I ask what to do with the lead times and capacities that the existing network offers.
In practice, the two views are complementary. A firm deciding whether to locate a new warehouse near a logistics cluster is asking Sheffi’s questions. A firm that has already committed to a footprint and needs to decide how to stock each node is asking mine. Long‑term network design and short‑term operational decisions live on different time scales, but they are part of the same story.
Sustainability: morality, markets, and measurement
Sustainability is a domain where our emphases differ but our conclusions largely agree.
In Balancing Green, Sheffi approaches sustainability as a tangled set of competing expectations. Investors want returns. Employees want jobs and stability. Communities want clean air and water. Consumers want to feel virtuous, but only a small fraction are consistently willing to pay higher prices for green products. Companies must navigate this reality, recognizing that much of their environmental footprint lies upstream and downstream in the supply chain rather than inside their own fences.
I tend to express similar ideas in more economic language. Every decision has side effects – pollution, resource depletion, social impacts – that may not be fully priced into today’s invoices. As the world evolves, regulators, courts, and markets progressively put prices on those side effects: carbon taxes, fines, bans, brand damage, talent attraction. Once those effects are priced, they become part of the same calculus as everything else. Buying from a “cheaper” supplier who dumps waste into a river may turn out to be very expensive once those future liabilities are included.
From that perspective, one role of advanced analytics is to surface these hidden costs as early as possible. If you can measure, even approximately, the long‑term financial impact of a certain decision path, you can encode that impact in your decision rules today. Sustainability is then no longer only a moral add‑on or a public relations strategy; it is part of running the business rationally over the long term.
Sheffi foregrounds the ethical and political complexity and then brings in supply chain. I foreground the economic machinery of supply chain and show how, when properly calibrated, it naturally includes sustainability. We meet in the middle.
People, culture, and the division of labor
Sheffi’s books are full of people: executives who champion resilience, plant managers who improvise after disasters, public officials who help or hinder recovery. He devotes serious attention to leadership qualities, organizational culture, and the human relationships between firms, governments, and communities. Resilience, in his telling, is as much about trust, communication, and empowerment as it is about dual sourcing or safety stocks.
In my experience, the human problem is a little different, though related. Most large companies employ armies of planners whose daily work consists of manipulating spreadsheets, reacting to exceptions, and reconciling conflicting lists of numbers. These people are smart, but they spend their energy fighting the tools instead of shaping the logic.
The organizational move I advocate is to invert this. Rather than having hundreds of people making thousands of micro‑decisions manually, I would rather have a small team that understands both the business and the mathematics, and that owns the “recipe” by which decisions are made. Everyone else then works with the outputs of that recipe: handling the truly exceptional cases, feeding back information when assumptions break, and improving the data.
Sheffi wants more distributed, empowered decision‑making in crises and more thoughtful leadership in normal times. I want fewer manual decisions in normal times so that human attention is freed up for the crises and the redesigns. These are not opposites; they are two sides of the same desire to use human judgment where it matters most.
How far does supply chain reach?
There is one final difference worth making explicit: the scope of the function we call “supply chain.”
In much of the management literature, supply chain covers logistics, sourcing, and operations. Sales and marketing set prices and promotions; finance sets capital structure; supply chain responds. Sheffi largely works within that conventional partition, even as he connects it to issues like sustainability and national resilience.
I am less convinced that those boundaries are useful in practice. The decision to set a price is, at its core, very similar to the decision to buy or move inventory. In both cases, the firm is taking a view of demand, committing resources, and trading off risk and reward over time. Separating pricing from supply decisions often leads to inconsistent choices: the marketing team pushes a promotion that the network cannot support; the supply chain team buffers against demand patterns that prices could have smoothed or shifted.
For this reason, I have long argued that pricing, assortment, and inventory should be treated together, using a common economic engine. That does not mean reorganizing the org chart overnight, but it does mean building decision processes and systems that do not artificially separate these levers. In that sense, my conception of “supply chain” is perhaps broader than Sheffi’s. It is the operational expression of the firm’s economics, not just the movement of goods.
Complementary lenses
It would be easy to frame these contrasts as disagreements, but that would be misleading. I have tremendous respect for Professor Sheffi’s work and recommend his books without hesitation to executives and practitioners. His perspective is indispensable if you want to understand how global supply chains interact with geography, politics, and society; how they break under stress; and how they can be designed to recover.
My own work lives closer to the machinery of decisions and software. I worry less about where to build the next logistics cluster and more about what the replenishment logic should do nightly. I talk less about leadership qualities and more about how to encode trade‑offs in code. I treat supply chain as applied economics and focus on teaching software to make good bets under uncertainty.
Seen together, these are not competing stories. They are a zoomed‑out and a zoomed‑in view of the same phenomenon. If Sheffi hands you the map of the territory – its clusters, its fault lines, its political borders – my ambition is to help you design the circuit board that sits in your company’s basement, quietly deciding what to buy, where to stock, and how to price, day after day.
For a profession that is finally, and painfully, recognized as central to modern life, having both views is not a luxury. It is a necessity.
