Resilience Is Not Redundancy: Why the Supply Chain Is Not a Chain
Global trade is still moving. The mistake is assuming that movement means robustness.
Global trade is still moving. The mistake is assuming that movement means robustness. In 2026, the real question is whether firms have diversified dependencies—or merely duplicated suppliers that fail together.
Keywords: supply chain resilience; logistics; supply networks; supply chain risk; maritime chokepoints; Panama Canal; Suez Canal; Strait of Hormuz; supplier diversification; route diversity; network resilience; redundancy; optionality; supply chain complexity; global trade
A supply chain is a lie told by a diagram.
It is a useful lie, certainly. Boxes are drawn in a row, arrows move obediently from left to right, suppliers supply, factories manufacture, ports port, ships sail, warehouses warehouse, customers consume, and everyone goes home satisfied that the world has been reduced to something that might fit comfortably on a PowerPoint slide.
The physical economy, regrettably, has never shown the slightest respect for PowerPoint.
What we call a supply “chain” is not a chain at all. It is a network of firms, factories, subcontractors, mines, refineries, roads, railways, pipelines, ports, canals, data systems, financiers, insurers, energy sources, governments, standards, labour markets and physical geography. Some of these are visible to the company whose logo appears on the finished product. Many are not. Some can be substituted in a week. Others cannot realistically be substituted in a decade. Some look diversified because several companies are involved while remaining economically identical because every one of those companies depends on the same upstream refinery, the same port, the same canal, the same electricity grid or the same obscure producer of a component nobody noticed until it stopped arriving.
That distinction is the essence of supply chain resilience.
Resilience is routinely confused with redundancy. The two are not synonyms. Owning two umbrellas does not constitute a strategy if both have been left in the same taxi. Having five suppliers does not make a supply chain resilient when all five containers pass through one chokepoint. Moving from one manufacturer to three manufacturers in neighbouring industrial zones may make the sourcing spreadsheet look splendidly diversified while leaving the physical system exposed to the same ports, the same weather, the same electricity system, the same regulatory environment and often the same upstream component manufacturers.
Corporate management has spent decades learning to count suppliers. It has spent considerably less time learning to count shared dependencies. That omission was tolerable when the dominant question in supply chain management was how to remove another cent from unit cost. It becomes rather more expensive when several different systems begin failing at once.
And that is the world of 2026.
Trade Has Not Died. Certainty Has.
The fashionable description of the present economy is that globalisation is collapsing. It makes an excellent headline and a rather poor description of the evidence. International trade has not vanished. The more interesting fact is that trade is proving remarkably persistent while the infrastructure, routes and assumptions beneath it are becoming less predictable.
The World Trade Organization’s Goods Trade Barometer in September 2026 remained above its baseline, with export orders, air freight, agricultural raw materials and electronic components all showing resilience even as container shipping remained softer. UN Trade and Development likewise reported continued expansion in global trade during 2026, but also stressed that higher energy, freight, logistics and production costs were contributing materially to the value of trade. The important point is not that globalisation is “fine.” It is that trade can continue while becoming more expensive, slower and more fragile.
This matters because resilience is easiest to misunderstand when the system has not completely failed. A factory shutdown is obvious. A ship sunk is obvious. An empty shelf is obvious. Much of supply chain fragility, however, appears first as friction: a voyage takes five days longer; insurance costs double; ships reroute; a supplier requires more working capital because transit times have stretched; inventory that once covered twenty days now effectively covers fourteen; a supposedly cheap source is suddenly not cheap after freight, financing and uncertainty are included; production planners compensate for variability until the compensations themselves begin consuming capacity.
The network can therefore remain operational while becoming steadily more brittle. This is precisely why resilience should not be measured merely by asking whether goods eventually arrived. A man who survives three heart attacks has demonstrated persistence, but his cardiologist would be unwise to describe the arrangement as efficient.
The more revealing question is what the system had to sacrifice to continue functioning. Did lead times expand? Did inventories rise? Did logistics costs increase? Did firms abandon preferred routes? Did they consume emergency capacity that will not be available for the next shock? Did working-capital requirements rise? Did customers accept slower service? Did governments intervene? Did firms transform their networks, or simply endure until the disturbance passed? Those are questions about resilience. They are quite different from asking whether trade technically continued.
What Resilience Actually Means
The academic literature has spent more than two decades trying to define supply chain resilience, and one can forgive practitioners for occasionally wondering whether academia intends to make a concept clearer or simply surround it with enough definitions that escape becomes impossible. Yet there is an important evolution in that literature.
Pettit, Fiksel and Croxton’s foundational work treated resilience as the relationship between vulnerabilities and capabilities. Their insight was not that every company should maximise redundancy. It was almost the opposite. A company with vulnerabilities exceeding its relevant capabilities is exposed; a company that purchases capabilities grossly beyond the vulnerabilities it actually faces can destroy profitability. Resilience therefore requires balance rather than the indiscriminate accumulation of buffers (Pettit et al., 2010; Pettit et al., 2019).
This is a much more useful idea than the corporate instinct to purchase a warehouse full of inventory after the latest crisis and call the exercise “resilience.” Inventory can be valuable. Alternative suppliers can be valuable. Spare capacity can be valuable. Faster information can be valuable. None is universally valuable. Their worth depends upon the vulnerability being addressed.
Tomlin’s analytical work makes the point elegantly. The optimal response to disruption risk depends not merely on the probability that a supplier will fail but on the shape of the disruption. Frequent, short interruptions can favour a different response from rare, prolonged failures. Inventory, reliable alternative sourcing, contingent rerouting and even passive acceptance all have circumstances in which they are economically rational (Tomlin, 2006). One cannot therefore purchase resilience in the abstract. One must understand the failure mechanism.
A second development in the literature is even more important. Earlier approaches frequently borrowed the engineering notion of resilience: a system is disturbed and then returns to its original state. The relevant questions become how much performance is lost and how rapidly normality can be restored. That remains useful. Time-to-recovery matters. Time-to-survive matters. A semiconductor fabrication plant does not become philosophically enriched when its ultrapure water supply fails; it needs the water back.
But Wieland and Durach make the larger point that a supply network is not merely a machine. It is a social and economic system populated by adaptive actors. Sometimes the correct objective is not to return to the state that existed before the shock. The old state may have been the reason the shock was so destructive in the first place. They consequently define resilience as the capacity of a supply chain to persist, adapt or transform in the face of change (Wieland & Durach, 2021). That is a far richer conception.
Figure 1. A useful conception of resilience must include survival, recovery, adaptation and, where necessary, transformation. Returning rapidly to yesterday’s configuration is not always success if yesterday’s configuration contained the vulnerability that caused the loss.
The distinction changes management. An engineering view asks how quickly the company can restore the interrupted route. An adaptive view also asks whether the route should remain dominant. Engineering resilience asks how fast the supplier can recover. Transformative resilience asks whether the business should continue depending upon that supplier, that technology or that geography at all. The first question concerns continuity. The second concerns strategy. Both matter. Confusing them produces either overreaction or complacency.
Efficiency Did Not Cause Fragility. Unpriced Correlation Did.
It has become fashionable to blame lean production and just-in-time inventory for every supply shortage since the invention of the pallet. This is convenient because it permits a complicated problem to be solved by declaring that warehouses should simply contain more things. The real issue is subtler.
Efficiency is not the enemy of resilience. A wasteful supply network does not become strategically sophisticated merely because it owns an impressive amount of inventory. The problem occurs when firms optimise efficiency while assuming that dependencies are independent when they are actually correlated.
Imagine a buyer with four suppliers. Conventional procurement metrics celebrate diversification. No supplier provides more than 30 per cent of volume. The purchasing officer may reasonably report that concentration risk has been reduced. Now add one inconvenient fact: all four suppliers export through the same port. The four suppliers are economically one supplier with four letterheads whenever that port closes.
Add another fact: all four purchase their critical precursor from the same second-tier producer. Now their apparent independence becomes theatrical. Add a third: the four use different ports, but every practical route to the customer crosses the same canal. The purchasing spreadsheet still reports four suppliers. Physics reports one chokepoint.
This is correlated redundancy: assets that appear redundant at one layer but share a common failure mode at another. The idea sits naturally within the work of Choi and Krause, who showed that supply base complexity is not simply a matter of how many suppliers exist. It also concerns differentiation and relationships among them. They warned that blindly reducing the number of suppliers can lower transaction costs while increasing supply risk and sacrificing innovation (Choi & Krause, 2006).
Yet the inverse deserves equal attention. Blindly increasing the number of suppliers does not necessarily reduce risk either. Supplier count is not the same thing as dependency diversity.
Craighead and colleagues showed why network design matters by identifying density, complexity and node criticality as determinants of disruption severity. The important node is not necessarily the largest supplier or the most expensive facility. It is the node whose interruption propagates furthest through the network (Craighead et al., 2007). Kim, Chen and Linderman sharpen this distinction further. A disruption at a node or arc does not necessarily constitute a network disruption. Some failures are absorbed locally. Others propagate through the topology and interrupt the wider system. The difference depends substantially on network structure (Kim et al., 2015).
This means that the correct unit of analysis is not merely “supplier risk.” It is the path by which economic function reaches the customer. A resilient organisation therefore needs to know not merely who its suppliers are, but which dependencies its supposedly independent suppliers share. That means mines, refineries, sub-tier manufacturers, ports, canals, border crossings, energy infrastructure, digital systems, cloud providers, shipping alliances, warehouses, finance, insurance and sometimes extraordinarily mundane pieces of physical equipment. The glamorous part of supply chain strategy may involve artificial intelligence and digital twins. The part that shuts the factory can still be a transformer, a bridge, a bearing or insufficient rain in Panama.
Supplier Diversity Is Not Route Diversity
This is one of the most important distinctions in contemporary logistics. Management tends to diversify the entities it contracts with because entities are visible. Routes are less visible because transport is frequently outsourced and because the route is treated as a service purchased after the sourcing decision has already been made. The result is a conceptual separation between procurement and logistics that the physical world does not share.
A procurement department may proudly announce that sourcing has been diversified across Vietnam, Thailand, Malaysia and southern China. The logistics team may then discover that the resulting cargo converges upon the same transshipment hub, the same narrow maritime corridor or the same destination port complex. Geographic diversity upstream has been compressed into route concentration downstream. The organisation has diversified origins while concentrating passage.
The distinction matters because maritime chokepoints produce precisely this effect. The Panama Canal, Suez Canal, Bab el-Mandeb, Strait of Hormuz, Strait of Malacca and several other passages exist because geography has kindly provided enormous economic efficiency in return for concentrated dependency. A chokepoint is efficient because enormous volumes can pass through a narrow place. A chokepoint is dangerous for exactly the same reason. The engineering accomplishment and the vulnerability are two sides of the same fact.
Jung, Kim, Jo and Cho’s 2025 study of the Panama Canal drought provides unusually good empirical evidence because it combines almost 30 million customs bills of lading with tens of millions of vessel-position observations. During the 2022–2023 drought, container voyages using Panama took approximately 133.8 hours longer, including an additional 33.6 hours stopped, while vessels also moved more slowly. Importantly, the route distance itself did not materially change. The system had not necessarily rerouted around the canal; the same corridor simply became slower (Jung et al., 2025).
This is an excellent illustration of why route resilience cannot be reduced to a binary “open versus closed” variable. A corridor can remain open and still lose much of its economic advantage. Indeed, the drought more than erased much of Panama’s normal-time efficiency advantage for affected traffic. That is exactly how real supply chain losses often appear—not as dramatic disappearance, but as the degradation of an assumption upon which inventory policy, production scheduling and customer promises were built.
Figure 2. The number of suppliers tells us remarkably little unless we also know whether those suppliers share upstream inputs and downstream routes. Apparent diversification can disappear the moment a common dependency fails.
The bottom-right quadrant is what firms imagine they have when they announce a multisourcing initiative. Many are actually sitting in the bottom-left. Five suppliers, one canal. Six factories, one transshipment port. Three countries, one critical precursor. Four logistics companies, all booking capacity on the same underlying carrier network. Different contractual counterparties do not automatically create different physical outcomes.
This also explains why nearshoring, friend-shoring and supplier diversification must be evaluated carefully. They can improve resilience enormously, but only when they actually alter correlated exposures. Moving final assembly closer to the customer can shorten lead time while simultaneously concentrating freight through a smaller number of land gateways. Moving suppliers from one political jurisdiction to several allied jurisdictions may reduce political risk while preserving a common dependence on a single refining country. Substituting an Asian supplier with a Mexican supplier may reduce ocean exposure and increase border exposure. Risk is rarely eliminated. It is transformed.
Good resilience strategy asks whether the transformed risk is more observable, more controllable and easier to substitute—not whether it has miraculously ceased to exist.
The Network Lies to Managers Because Managers See Only Part of It
Choi and Krause make an important distinction between the supply base and the wider supply network. The supply base consists of suppliers the focal firm actively manages. The supply network includes companies further upstream that may influence the focal firm despite having no direct contractual relationship with it (Choi & Krause, 2006). That difference sounds academic until a third-tier supplier fails. Then it becomes an invoice.
The focal company naturally knows a great deal about the firms it pays. It may know very little about the firms its suppliers pay. Yet those hidden tiers can contain extraordinarily concentrated activities. Automotive manufacturers learned this repeatedly. Semiconductor customers learned it. Pharmaceutical firms know it from active pharmaceutical ingredients. Energy-transition technologies exhibit the same pattern through concentrated mineral processing and refining.
The paradox is that the deeper one moves into the network, the less managerial visibility often exists precisely where industrial concentration can become greatest. Complexity compounds the problem.
Bode and Wagner empirically examined horizontal complexity, vertical complexity and spatial complexity in upstream supply chains. Their findings were not merely that each form of complexity was associated with more frequent disruption. The dimensions could reinforce one another. Complexity could become superadditive (Bode & Wagner, 2015). That is an important warning against one of the most persistent errors in risk management: addition.
Traditional registers often identify Risk A, Risk B and Risk C, estimate their individual likelihood and impact, and place them neatly into coloured boxes. The organisation then admires its risk matrix, preferably in a meeting where coffee is supplied and no physical cargo is present to contradict it. Networks do not respect the assumption that risks arrive one at a time.
A drought can reduce canal capacity while geopolitical conflict changes alternative routes. A port strike can occur while inventory has already been depleted by a previous delay. An energy-price shock can increase freight and manufacturing costs just as higher interest rates make additional inventory more expensive to finance. A trade-policy change can redirect cargo into ports that were never designed for the resulting volume. A cyberattack can hit a logistics provider during a physical disruption precisely because the organisation is already operating through unfamiliar contingency processes.
The correct question is therefore not merely: “What happens if this supplier fails?” It is: “What else tends to be unavailable when this supplier fails?” Correlation is where resilience strategy begins.
2026 Is Giving Us Three Different Chokepoint Lessons at Once
If one wished to design a lecture demonstrating that supply chain risk has multiple failure mechanisms, the present world would almost be too obliging. Panama, Suez and the Red Sea, and Hormuz illustrate three different classes of physical-network risk: environmental capacity, security-driven route substitution and geopolitical-energy disruption.
Panama: The Infrastructure Is Working; Nature Changed the Capacity
The Panama Canal did not need to be destroyed to become a resilience problem. Its locks depend upon freshwater. That makes rainfall, water management and shipping capacity economically connected. The historical drought analysed by Jung and colleagues showed how a hydrological shock translated into slower voyages and lost logistical advantage. The lesson is not historical. In September 2026, the Panama Canal Authority introduced additional capacity measures after below-expected precipitation, reducing available daily Panamax slots. The canal is not “closed”; its operating envelope has changed.
The point is that an asset can be entirely intact while its service capacity is altered by a variable that many conventional sourcing models once treated as external scenery. Climate is not scenery when infrastructure depends upon water. For the shipper, this transforms environmental uncertainty into booking risk, waiting time, inventory requirements, scheduling uncertainty and ultimately working capital. The physical mechanism begins as rainfall. The financial mechanism ends on the balance sheet.
Suez and the Red Sea: Route Efficiency Depends Upon Security
The Red Sea presents a different lesson. Here the physical canal can be available while the approach to it becomes commercially unattractive because of security risk. Shipping companies responded to earlier attacks by rerouting vessels around the Cape of Good Hope. The result was not a theoretical exercise in geopolitics. It was extra sailing distance, more fuel, more vessel-days, changes in effective global shipping capacity and longer inventory pipelines.
By September 2026, carriers were experimenting with and expanding selective returns through Suez rather than simply restoring the old network in one movement. That is resilience in motion. A route has not simply switched from “closed” to “open.” Carriers are testing, adapting and selectively reallocating services as the risk-reward relationship changes. The interesting variable is optionality. A carrier capable of shifting between Suez and the Cape possesses a valuable operational option. But exercising it is not free. The Cape requires more distance and capacity. Returning through Suez can reduce transit time but increases exposure to the security environment. There is no magic route. There are choices with different conditional costs.
Hormuz: A Chokepoint Can Become an Energy Shock, a Freight Shock and a Financial Shock Simultaneously
Hormuz demonstrates something more severe: physical-route risk interacting with the global energy system. In September 2026, reported traffic through the strait had fallen dramatically relative to normal levels amid intensified regional attacks and security concerns. The significance extends far beyond tankers.
Energy is an input into almost every physical supply chain. A disruption to oil and gas flows changes bunker fuel, electricity, petrochemical feedstocks, fertiliser, aviation, trucking, industrial heat and consumer purchasing power. Insurance changes. Financing changes. Freight rates change. Governments respond. Companies substitute inputs. Consumers alter demand. What begins as a problem at a narrow geographic passage propagates through systems that never send a ship through that passage at all. This is why resilience cannot be managed by logistics departments alone.
Figure 3. “Chokepoint risk” is not one risk. Different chokepoints fail through different mechanisms, and the appropriate resilience response must therefore differ.
The three cases expose the poverty of a single global “supply chain risk” score. Panama’s problem cannot be solved by the same capability as Hormuz. Extra inventory may help absorb Panama delay. It does not create energy. A Cape route may substitute for Suez. There is no equivalent maritime detour around a globally significant energy shortage. Better forecasting may improve bookings but cannot produce rain. Insurance can transfer part of the financial loss but cannot deliver a component to an idle assembly line. Resilience has to be matched to mechanism.
The Economics of Resilience: The Expense You See and the Loss You Do Not
Managers have a perfectly reasonable objection to resilience investment. It costs money. Spare capacity costs money. Inventory costs money. Qualifying an alternative supplier costs money. Maintaining a supplier that receives only a minority allocation costs money. Mapping multiple tiers costs money. Contracting for transport optionality costs money. Holding additional warehouse space costs money. Designing products to accept alternative components costs money.
Efficiency savings are immediate, measurable and flattering. Avoided disasters are hypothetical. This creates an asymmetry in corporate decision-making. A procurement manager who removes $10 million from annual purchasing cost can point to $10 million. A resilience manager who spends $2 million maintaining an option that prevents a $100 million loss five years later will spend four years appearing less efficient than the colleague who cancelled it.
Accounting records the cost of the lifeboat every year. It records the value of the lifeboat rather dramatically and only on the day the ship sinks.
This is one reason Hendricks and Singhal remain so important. Their examination of 827 supply chain disruption announcements found average abnormal stock returns approaching negative 40 per cent over the period beginning one year before the announcement and extending two years afterward. They also found that equity risk increased materially after disruption and that firms did not simply bounce back immediately (Hendricks & Singhal, 2005).
The magnitude should dispel the notion that supply disruption is merely an operational inconvenience to be delegated somewhere below the board. A severe disruption changes expected cash flows, risk and therefore enterprise value. It can reduce sales, increase costs, create inventory distortions, damage customer relationships, alter financing conditions and absorb management attention. A resilience decision can consequently be a capital-allocation decision even when accountants initially classify it as “extra inventory” or “dual-source cost.”
But here one must resist the opposite error. If resilience is valuable, it does not follow that more resilience expenditure is always better. Pettit, Croxton and Fiksel explicitly make this point. Excessive vulnerability relative to capability creates exposure. Excessive capability relative to vulnerability can erode profit. The task is balance (Pettit et al., 2019).
No company should duplicate every factory, triple every inventory buffer and reserve alternative transport on every conceivable route. A system designed never to experience a shortage could bankrupt itself preparing for one. Resilience should instead be concentrated where the expected economic value of optionality is greatest: critical products, critical nodes, highly correlated dependencies, long recovery times and exposures where substitution after the event would be difficult or impossible. In other words, resilience requires discrimination. The firm needs to know which failures matter.
Do Not Ask “How Many Suppliers?” Ask “How Many Independent Ways Can We Still Serve the Customer?”
This is the question I would put at the centre of resilience planning. Suppose a critical component is supplied by three vendors. At first glance, diversification appears respectable. Now ask whether they use different manufacturing regions. Then ask whether they use different upstream component suppliers. Ask whether they obtain electricity from the same regional grid. Ask which ports they use. Ask whether their freight eventually converges at the same transshipment hub. Ask which shipping alliances provide the capacity. Ask whether every commercially viable route crosses the same chokepoint. Ask whether the alternative supplier has spare capacity precisely when every competitor wants it. Ask how long technical qualification would take. Ask whether intellectual-property restrictions prevent rapid substitution. Ask whether tooling exists. Ask whether the replacement product can be accepted without redesign. Ask whether the alternative supplier itself relies on the failed supplier.
Then count suppliers again. The number may have changed without any company disappearing.
This is why resilience analysis should be built around independent fulfilment paths, not merely entities. An independent fulfilment path is a practicable combination of source, transformation, transport and supporting infrastructure capable of continuing to deliver the required economic function when another path is unavailable. The word “practicable” matters. A spreadsheet can invent alternatives instantly. Physical capacity is less imaginative.
A supplier capable of producing 5 per cent of normal demand is not a full substitute for one producing 70 per cent. A port technically capable of receiving a vessel may lack cranes, storage, rail capacity or customs throughput for the redirected volume. An alternative manufacturer that requires nine months of qualification is not a contingency response to a two-month crisis. A ship route around Africa exists, but the ships required to maintain the same weekly capacity do not appear by enchantment merely because the route is visible on a map. Resilience is therefore constrained by capacity, speed of activation and substitutability. These properties should be measured.
Visibility Is Useful. Visibility Without the Ability to Act Is Surveillance.
Supply chain technology is currently filled with promises of end-to-end visibility, control towers, AI prediction, digital twins and real-time alerts. Many are genuinely valuable. Information can extend warning time. Craighead and colleagues explicitly identify warning capability alongside recovery capability as a moderator of disruption severity. Knowing earlier permits firms to reposition inventory, reserve capacity, communicate with customers and execute alternatives before everyone else attempts the same thing (Craighead et al., 2007).
But visibility has become another term that deserves interrogation. Seeing a disruption is not the same as having an alternative. If a company learns three weeks early that a unique supplier will shut down for six months, the information is valuable only to the extent that three weeks changes the feasible response. If qualification of the sole alternative takes nine months, the control tower has provided a beautifully rendered view of an unavoidable shortage. That is surveillance, not resilience.
The real value of information is therefore conditional upon response capability. Warning time multiplied by optionality is valuable. Warning time multiplied by zero alternatives remains approximately zero, however pleasing the dashboard.
This also changes how firms should think about AI in supply chains. The most interesting application is not another model predicting that a shipment might be late. It is linking prediction to explicit decisions: reroute, reallocate, reserve, substitute, expedite, ration or communicate. Prediction without an action rule is weather commentary. Operational resilience requires decision architecture.
Inventory Is Neither Hero nor Villain
Inventory deserves rescue from two opposing religions. The first regards inventory as sin. The second discovered during the pandemic that sin can be extremely useful and consequently wishes to build warehouses until shortage becomes metaphysically impossible. Both positions are too simple.
Inventory is time stored in physical form. It allows the downstream system to continue operating while upstream supply is interrupted. Its resilience value therefore depends upon how much time is required to activate an alternative or recover the failed source. If a disruption usually lasts two days, nine months of safety stock is an expensive monument to anxiety. If a sole-source component requires twelve months to requalify elsewhere, two days of safety stock is optimism expressed through inventory policy.
Tomlin’s work demonstrates precisely why the frequency and duration of disruptions matter to the optimal choice among inventory, sourcing mitigation and contingent rerouting (Tomlin, 2006). The same logic applies beyond inventory. Capacity is valuable when it can be activated fast enough. Alternative suppliers are valuable when they can actually produce enough. Alternative routes are valuable when infrastructure at both ends can absorb the flow. Contracts are valuable when counterparties can honour them during a system-wide shortage. Resilience is not the presence of an option on paper. It is the ability to exercise it under the conditions in which it is needed.
Complexity Must Be Managed, Not Worshipped or Eliminated
There is a seductive managerial response to complexity: simplify everything. Sometimes this is excellent advice. Reducing unnecessary product variants, standardising components, eliminating redundant handling and reducing gratuitous supplier fragmentation can improve performance and make disruptions easier to manage. But complexity has an awkward relationship with resilience. Some complexity is unnecessary burden. Some complexity represents diversity.
Eliminate the wrong kind and the network becomes beautifully simple because it now has one way to work. It also has one way to fail.
Choi and Krause highlighted precisely this trade-off. Fewer suppliers can reduce transaction costs and improve responsiveness while increasing supply risk and potentially reducing innovation. Bode and Wagner show that greater structural complexity can increase the frequency of disruption. These findings do not contradict each other. They tell us that the problem cannot be solved through a slogan such as “more suppliers” or “fewer suppliers.”
The objective is useful heterogeneity without gratuitous complexity. Standardise where standardisation creates substitutability. Diversify where concentration creates correlated failure. Reduce nodes that add neither capability nor optionality. Preserve nodes that provide genuinely independent capacity. That is not as satisfying as issuing a universal procurement rule. It has the minor advantage of being correct.
Resilience Is a Design Problem Before It Is an Emergency Problem
The worst time to discover supply chain architecture is during a disruption. Yet many firms effectively do exactly this. The crisis begins. Someone asks which second-tier suppliers are affected. Someone else attempts to locate contracts. Logistics asks whether another port is feasible. Engineering discovers that an alternative part was never qualified. Procurement contacts another manufacturer and learns its capacity was purchased six hours earlier by a competitor whose resilience plan included a telephone number.
Emergency response matters, but by then the feasible choice set has already been determined by years of design decisions. Product architecture determines component substitutability. Sourcing policy determines supplier optionality. Factory location determines transport dependence. Inventory policy determines time-to-survive. Contracts determine access to contingent capacity. Data architecture determines visibility. Qualification rules determine switching time. Financial policy determines how much inventory and disruption cost the company can absorb. Network design determines propagation.
Resilience is therefore largely pre-committed before the crisis arrives. This is one of the reasons Craighead et al.’s focus on design characteristics remains so useful. Disruption severity is not just a property of the hazard. It emerges from the interaction between the hazard and the network it encounters.
A storm is meteorology. A storm shutting five factories because they all depend upon one flooded substation is network design. The distinction matters because companies cannot manage the weather. They can manage some of the dependencies through which weather becomes financial loss.
The Objective Should Be Selective Redundancy and Maximum Reconfigurability
If resilience is not synonymous with redundancy, what should firms seek? I would place reconfigurability above raw duplication. A rigid network with enormous spare capacity can survive certain shocks but remain vulnerable to shocks outside the scenarios for which the redundancy was purchased. A reconfigurable network can alter production allocation, transport, components and fulfilment rules as conditions change.
This is the difference between owning backups and possessing options. Options have value because the future state is uncertain. A qualified second supplier is an option. A product designed to accept multiple standard components is an option. A port contract allowing diversion is an option. A logistics network capable of switching modes is an option. Reserved emergency capacity is an option. Postponement in manufacturing is an option. Strategic inventory is an option on time. Information is an option only when receiving it changes a decision.
The aim is not to exercise every option. The entire point is that most options will remain unused most of the time. This is precisely why traditional efficiency metrics dislike them. Unused optionality looks wasteful until the state of the world changes. The managerial task is to price that optionality against the correlated exposure it protects.
Do Not Optimise the Average Day
Many logistics systems are extraordinarily good on average. Average vessel transit time is excellent. Average supplier quality is excellent. Average port performance is excellent. Average production utilisation is excellent. The shareholder does not live in the average. Neither does the customer. Averages are seductive because most days are ordinary. Catastrophic loss is created in the tails.
This does not mean organisations should optimise solely for catastrophe. That would be equally foolish. It means the objective function should include the economic cost of tail states rather than quietly assuming that the world will remain close to its historical mean.
Pettit et al.’s vulnerability-capability balance provides one framework. Time-to-survive and time-to-recover provide another useful operational framework. Network analysis adds node criticality and propagation. Tomlin adds the economics of mitigation versus contingency. Together, these approaches suggest a more disciplined model of resilience than “keep more stock.”
For every critical flow, management should understand how long the business can survive an interruption, how long restoration is likely to take, what alternatives can be activated, how much those alternatives can actually supply, what shared dependencies exist between them, and how the economics change if several supposedly independent disruptions occur together.
The difference between time-to-survive and time-to-recover is particularly stark. If the business can survive 30 days without a node and recovery requires ten days, there is room. If the business can survive ten days and recovery requires 90, there is not a resilience problem to discuss later. There is a design problem already waiting for a date.
Resilience Must Reach the Board Because It Already Reaches the Share Price
Supply chain resilience is too often filed beneath “operations.” That classification may reflect corporate organisation. It does not reflect economic causation. Hendricks and Singhal demonstrated that serious disruptions affect long-term shareholder value and equity risk. Higher inventory alters working capital. Additional plants require capital expenditure. Supplier failure creates credit exposure. Freight disruption affects margins. Energy disruptions change costs throughout the enterprise. Geopolitics alters investment. Trade restrictions can strand assets.
These are board-level issues wearing operational clothing. A sensible resilience programme should therefore connect supply network design with corporate finance. Management should be able to compare the annual cost of maintaining an alternative with the expected economic exposure that alternative mitigates. It should be able to identify critical nodes whose failure would materially affect enterprise value. It should know which “savings” arise from genuine productivity and which arise from quietly consuming resilience.
This last distinction deserves attention. A company can improve reported efficiency by reducing inventories, eliminating suppliers, raising utilisation, centralising production and concentrating logistics. Each decision can be individually rational. Together they can remove the system’s degrees of freedom. The company has not necessarily become more productive. It may simply have converted visible cost into invisible exposure. That exposure remains invisible until the state of the world changes. Then accounting catches up very quickly.
The World Is Not Deglobalising. It Is Repricing Dependency.
The current period is frequently described as the end of globalisation. I think that is too crude. Trade remains enormous. Firms continue sourcing internationally. Comparative advantage has not resigned because a politician held a press conference. Ships continue sailing because moving goods thousands of kilometres remains economically sensible in an astonishing number of industries.
What is changing is the price attached to dependency. Transport time has a price. Policy uncertainty has a price. Energy security has a price. Concentration has a price. Carbon has a price. Working capital has a price. Political alignment has a price. Reliance on a maritime chokepoint has a price. Reliance on a single technology has a price. Reliance on a supplier whose own supplier nobody has identified has a price.
For years some of those prices were effectively treated as zero because they did not appear on the purchase order. They were never zero. They were contingent. This is the great resilience lesson of the past decade. The cheapest source is not necessarily the source with the lowest expected total economic cost. The shortest route is not necessarily the route with the highest expected service performance. The supplier portfolio with the largest number of names is not necessarily the portfolio with the greatest diversity. The network with the smallest inventory is not necessarily the most productive once disruption costs are included. The accounting boundary and the economic boundary are different things.
From Bouncing Back to Becoming Better
Wieland’s work on transformative supply chain management takes resilience beyond restoration. He argues that supply chains should be understood as social-ecological systems embedded within political, economic and planetary systems rather than isolated machines susceptible to complete optimisation (Wieland, 2021). This becomes increasingly convincing as one watches contemporary trade.
Rainfall influences canal slots. Military conflict influences vessel routing. Energy infrastructure influences manufacturing costs half a world away. Trade rules influence factory location. Technology influences the commodity composition of global trade. Environmental constraints influence insurance, infrastructure and capital allocation. There is no clean boundary around the supply chain.
The idea that management can identify every risk, estimate each probability and engineer a permanently optimal network belongs to a world that exists chiefly in introductory diagrams. A better ambition is to build a network capable of changing without losing its economic purpose. That means accepting that the “optimal” configuration is conditional.
A route is optimal under one security environment. A supplier is optimal at one freight rate. A factory is optimal under one tariff regime. An inventory policy is optimal for one distribution of lead times. A concentrated supplier relationship is optimal while the probability and duration of interruption remain within certain bounds. Change the environment and yesterday’s optimisation becomes today’s constraint.
This is why transformation belongs within the definition of resilience. Sometimes resilience means restarting the old system. Sometimes it means refusing to.
What a Serious Resilience Programme Should Measure
The practical implication is that resilience should become measurable without becoming simplistic. I would begin with critical flows rather than every item. Identify products and inputs whose interruption would create material economic damage. Map their fulfilment paths through multiple tiers where feasible. Identify common dependencies among those paths. Estimate time-to-survive under different outage conditions. Estimate time-to-recover for critical nodes. Measure the capacity—not merely the existence—of alternatives. Measure activation time. Measure route concentration. Identify which nodes or arcs create disproportionate network effects. Stress the system with compound rather than isolated shocks.
Then compare capabilities with vulnerabilities. A business exposed to a short port closure may require additional inventory. A business exposed to a unique production technology may require design standardisation or an alternative qualified producer. A business dependent upon one maritime corridor may require route options or geographically separated inventory. A business exposed to a critical raw material may need long-term contracts, recycling, substitution, strategic stock or vertical integration. A business dependent upon fast warning and rerouting needs data and decision rules.
The response should follow the mechanism. This sounds obvious. It is violated constantly because companies prefer standardised programmes. Standardisation makes management easy. Reality remains inconsiderate.
The Most Dangerous Supply Chain Is the One That Works Perfectly
A badly performing supply chain attracts attention. A perfectly performing one attracts optimisation. Every spare minute is removed. Every buffer is challenged. Every secondary supplier is asked to justify itself. Every warehouse is measured against utilisation. Every route is consolidated. Every factory is specialised. Every cost not exercised during normal operations is described as waste.
Eventually, the system becomes magnificent. On Tuesday. The problem arrives on Wednesday.
Resilience is the discipline of remembering Wednesday while operating efficiently on Tuesday.
That does not require abandoning lean systems, global trade or optimisation. It requires recognising that optimisation is meaningful only relative to a model of the world. If the model excludes correlated disruption, chokepoint dependency, recovery time and adaptive capacity, the optimisation has not removed risk. It has merely removed it from the equation.
Conclusion: The Supply Chain Is Not a Chain, and Resilience Is Not a Warehouse
The next era of supply chain management should move beyond a debate between “efficiency” and “resilience.” It is a false opposition. Waste is not resilient. Efficiency is not inherently fragile. Redundancy is not automatically resilience. Diversification is not automatically diversity. Visibility is not automatically control. Recovery is not always transformation.
The real issue is whether the network has been designed with enough independent options to preserve economic function when its operating environment changes.
The evidence accumulated over two decades tells us that severe disruptions impose enduring financial costs. Network structure determines whether local failures remain local or propagate. Complexity can amplify disruption frequency. Supplier portfolios involve genuine trade-offs between transaction cost, responsiveness, innovation and risk. Inventory, dual sourcing and contingent capacity have different values under different disruption patterns. Resilience therefore cannot be reduced to a single practice.
The events of 2026 merely make the theory difficult to ignore. Panama reminds us that physical infrastructure can become constrained without being damaged. The Red Sea reminds us that an efficient route is valuable only while it is safely usable. Hormuz reminds us that a narrow passage can transmit a logistics shock into energy, manufacturing, prices and finance. And global trade itself reminds us of something equally important: systems adapt.
Trade continues. Routes move. Firms substitute. Capacity is reallocated. New intermediaries emerge. Costs are repriced. Networks transform. That is resilience—but not necessarily efficient resilience, and certainly not free resilience. The managerial challenge is to make more of that adaptation deliberate rather than desperate.
Count suppliers, certainly. Then count ports. Count routes. Count upstream dependencies. Count qualification times. Count alternative capacity. Count the days the business can survive. Most importantly, count how many genuinely independent ways remain to serve the customer when the obvious one disappears.
Because a company does not possess five alternatives merely because five names appear in a database. It possesses as many alternatives as remain independent when something important goes wrong. That number is usually smaller.
Finding out how much smaller before the disruption is called resilience. Finding out afterward is called news.
References
Bode, C., & Wagner, S. M. (2015). Structural drivers of upstream supply chain complexity and the frequency of supply chain disruptions. Journal of Operations Management, 36, 215–228. https://doi.org/10.1016/j.jom.2014.12.004
Choi, T. Y., & Krause, D. R. (2006). The supply base and its complexity: Implications for transaction costs, risks, responsiveness, and innovation. Journal of Operations Management, 24(5), 637–652. https://doi.org/10.1016/j.jom.2005.07.002
Craighead, C. W., Blackhurst, J., Rungtusanatham, M. J., & Handfield, R. B. (2007). The severity of supply chain disruptions: Design characteristics and mitigation capabilities. Decision Sciences, 38(1), 131–156. https://doi.org/10.1111/j.1540-5915.2007.00151.x
Hendricks, K. B., & Singhal, V. R. (2005). An empirical analysis of the effect of supply chain disruptions on long-run stock price performance and equity risk of the firm. Production and Operations Management, 14(1), 35–52. https://doi.org/10.1111/j.1937-5956.2005.tb00008.x
Jung, P. H., Kim, K., Jo, A.-H., & Cho, J. S.-H. (2025). Panama Canal drought and supply chain disruptions in Asia–United States trade: Evidence from micro-level trade shipments and vessel trajectory data (ADB Economics Working Paper Series No. 822). Asian Development Bank. https://doi.org/10.22617/WPS250475-2
Kim, Y., Chen, Y.-S., & Linderman, K. (2015). Supply network disruption and resilience: A network structural perspective. Journal of Operations Management, 33–34, 43–59. https://doi.org/10.1016/j.jom.2014.10.006
Pettit, T. J., Croxton, K. L., & Fiksel, J. (2019). The evolution of resilience in supply chain management: A retrospective on ensuring supply chain resilience. Journal of Business Logistics, 40(1), 56–65. https://doi.org/10.1111/jbl.12202
Pettit, T. J., Fiksel, J., & Croxton, K. L. (2010). Ensuring supply chain resilience: Development of a conceptual framework. Journal of Business Logistics, 31(1), 1–21. https://doi.org/10.1002/j.2158-1592.2010.tb00125.x
Tomlin, B. (2006). On the value of mitigation and contingency strategies for managing supply chain disruption risks. Management Science, 52(5), 639–657. https://doi.org/10.1287/mnsc.1060.0515
Tukamuhabwa, B. R., Stevenson, M., Busby, J., & Zorzini, M. (2015). Supply chain resilience: Definition, review and theoretical foundations for further study. International Journal of Production Research, 53(18), 5592–5623. https://doi.org/10.1080/00207543.2015.1037934
Wieland, A. (2021). Dancing the supply chain: Toward transformative supply chain management. Journal of Supply Chain Management, 57(1), 58–73. https://doi.org/10.1111/jscm.12248
Wieland, A., & Durach, C. F. (2021). Two perspectives on supply chain resilience. Journal of Business Logistics, 42(3), 315–322. https://doi.org/10.1111/jbl.12271
Current 2026 Sources
World Trade Organization. (2026, September 9). Goods Trade Barometer. https://www.wto.org/english/news_e/news26_e/wtoi_09sep26_481_e.htm
UN Trade and Development. (2026). Global Trade Update, July/August 2026. https://unctad.org/publication/global-trade-update-julyaugust-2026-global-trade-continues-expand-amid-rising-price
Panama Canal Authority. (2026, September). Panama Canal adopts additional measures to address reduced precipitation in the Canal watershed. https://pancanal.com/en/panama-canal-adopts-additional-measures-to-address-reduced-precipitation-in-the-canal-watershed/
Maersk. (2026, September). Europe market update. https://www.maersk.com/news/articles/2026/09/09/europe-market-update-september
Reuters. (2026, September 15). Hormuz traffic dwindles after Middle East attacks intensify. https://www.reuters.com/world/middle-east/hormuz-traffic-dwindles-after-middle-east-attacks-intensify-2026-09-15/