When the Network Fails: How a Disaster Becomes a Payment Shock
This chapter follows the payment shock itself: how a severe weather event moves from a physical threat to a failure of the electronic route used to buy necessities. The focus is the chain connecting power, communications, payment authorization, merchants, and household liquidity. Tracing that chain shows why a disaster can change the feasible set of payments over time, and why stockpiling, cash preservation, substitution, and later drawdown should not be collapsed into one vague idea of “cash use.”
A payment is a chain, not a tap
At the point of sale, a card payment can look almost instantaneous. A customer presents a card or phone, a terminal sends a request, a network routes it, an issuer or processor authorizes it, and the merchant receives confirmation. The visible action is small; the supporting chain is not.
Electricity powers the terminal, router, store systems, and communications equipment. A telecommunications link carries the request. The payment network and data systems must be reachable. The merchant must still be operating, and the customer must be able to get there. A digital wallet or peer-to-peer transfer may use a different interface, but it remains exposed to some combination of power, communications, device, and service availability.
A failure at any one point can change the transaction. If the terminal has no power, there may be no way to enter the payment. If communications are down, a transaction that normally requires an online authorization may not be confirmed. If the merchant's systems are damaged or the store is closed, neither a card nor cash can buy what is not available. These are different failures, but they can arrive together during a hurricane or other severe event.
Cash has a different exposure at the moment of settlement. A note already in the customer's possession does not require a live message to a card network. That gives cash a useful property when the electronic route is unavailable. It does not make cash independent of the wider economy. The customer must have obtained the notes before access was cut off or while a functioning source remains available. The merchant must accept them and be able to make change or otherwise complete the sale. The store must have goods, staff, and a safe way to operate.
This distinction matters because “cash works offline” can be misunderstood as “cash solves a disaster.” It solves a narrower problem: the inability to authorize or settle a transaction through a live electronic channel. It cannot reopen a closed store, restore a damaged road, replenish fuel, or replace goods that never arrived. Payment resilience is one layer of operational resilience, not a synonym for it.
The research on payment methods during extreme weather treats these layers as connected but separable. It combines storm timing with county-level electricity outages, store transactions, card spending, household purchase records, and surveys. The point of using several measures is not to create a longer list of data. It is to distinguish an event that makes people want to prepare from a failure that changes what they can actually do at checkout.
The distinction can be expressed as two questions. First, did the storm change the need or incentive to buy? Second, did an outage change the ability to pay for what was available? A household may answer yes to the first question but no to the second if the local payment infrastructure remains intact. Another household may face both a stronger need to prepare and a narrower set of ways to pay.
The route from storm to payment failure can take several forms. A power outage may disable a terminal, a store's internal systems, or an ATM even when communications networks are still operating. A communications outage may leave the terminal powered but unable to reach the authorization systems it normally depends on. A store or road failure may block the purchase before payment is attempted. These cases have different implications for cash. Cash can help in the second case if the merchant remains open and accepts it, and it may help in the first if the store can operate without powered payment equipment. It cannot help in the third case when the customer cannot reach an open merchant or the merchant has no goods to sell.
The distinction also matters for how risk is shared. An online payment system concentrates convenience in a connected network, but the same connectivity means that a local failure can affect many transactions at once. Cash makes some preparation visible at the household level: someone must already hold it, protect it, and decide how much liquidity to keep available. Merchants, in turn, must decide whether they can safely accept physical currency, make change, record sales, and maintain operations while other systems are down. Resilience is therefore not created by an object alone. It emerges from the interaction between a payment instrument, the people holding it, the merchant willing to accept it, and the infrastructure that lets the transaction take place.
This interaction explains why the research does not treat an outage as a simple switch from “digital” to “cash.” The same county can contain closed stores, operating stores with failed terminals, and operating stores where electronic payments still work. The same household can use a card for one purchase, cash for another, and make no purchase at a third location because the needed item is unavailable. Aggregate spending and payment shares combine these outcomes. Understanding the mechanism means keeping them conceptually separate before looking for their pattern in the data.
The sequence shows why the same customer can use cards before an event and cash during it without changing their underlying preference. The feasible choice set has changed. A payment method can be preferred when all its dependencies work and still be unusable when one dependency fails.
The timing of a payment shock
The timing begins before the storm arrives. When households expect disruption, they may bring purchases forward: food, fuel, water, batteries, and other supplies become more valuable to have at home. The payment data in the resilience study show spending rising before hurricanes, with the increase consistent with stockpiling. That preparation is not the same as panic or irrational overspending. It is a response to the possibility that a store may close, a trip may become difficult, or a later transaction may be harder to complete.
The method used to finance those purchases matters. Before the event, the cash share of store sales falls while card spending, especially credit-card spending, rises. One interpretation is that households use cards to front-load purchases while retaining cash for the period when electronic payments may be impaired. Credit makes it possible to buy now without immediately reducing the physical cash buffer that may be needed later. The evidence supports this timing pattern, but it does not show that every household follows the same plan or that every pre-event card purchase is deliberate cash preservation.
The sequence is best understood as a change in the value of liquidity, not as a claim that cash is inherently better for every purchase. In normal conditions, using a card can be convenient and can leave physical notes untouched. As disruption risk rises, those notes acquire an option value: they preserve a payment path that may remain available after an electronic authorization fails. That option has limits. Cash is finite, can be lost or stolen, may be held in denominations that do not fit the purchase, and must eventually be replenished. Holding it also does not guarantee that a merchant can accept it or that the desired goods will be available. The point of the pre-event pattern is therefore not that households abandon digital payments, but that they may use each instrument for the condition in which it is most useful.
The merchant's position changes at the same time. A store that stays open may be able to take cash while its terminal connection is down, but acceptance still requires staff, pricing and sales records, secure handling, and a workable way to make change. If those conditions fail, cash cannot turn an open door into a functioning market. This is why payment resilience is shared between the household that holds a fallback, the merchant that accepts it, and the infrastructure that keeps the store operating. The observed shift in payment composition is the result of those conditions meeting at one point in time, not a universal preference reversal.
When the storm and outage arrive, the constraints multiply. Some stores close, reducing the number of places where a purchase can occur. Stores that remain open may see fewer customers because roads are unsafe or electricity and local services are disrupted. Even an open store can lose sales if the terminal or communications link cannot complete electronic payments. The resulting decline in spending therefore has both an extensive margin—the number of operating stores—and an intensive margin—the amount spent at stores that remain open.
Cash becomes valuable at the point where the digital route is no longer reliable but a merchant can still operate. The customer may have preferred a debit card or credit card under normal conditions; that preference does not help if the terminal cannot authorize the transaction. A cash balance changes the feasible action immediately. It can pay for fuel, food, or other available necessities, provided the merchant can accept it and the desired goods remain in stock.
This is why a rise in the cash share after a hurricane has to be interpreted with care. In a county with an outage, the rise can indicate substitution away from electronic payment methods that have become unavailable. In a nearby county with no outage, the cash share can also rise after the event, but for a different reason: households may have accumulated cash as a precaution and then spend it once the feared event occurs. The same statistic—more cash as a share of sales—can reflect forced substitution in one place and planned drawdown in another.
The paper calls the latter process “cash burn,” meaning that a household runs down cash previously accumulated for a disruption. The term describes a dynamic inventory decision, not a direct observation of every person's wallet. The available data show timing and payment composition; they do not watch each household's notes being counted out transaction by transaction. That distinction preserves the useful mechanism without turning an interpretation into a literal measurement.
Duration changes the practical importance of every stage. A short interruption may delay a payment and leave little lasting effect if the network returns quickly. The study finds that hurricane-related outage intensity remains elevated for more than a week, with the largest disruption in the first few days. Over that horizon, a fallback cannot be understood as a single emergency purchase. Households may need to complete several transactions while stores reopen unevenly and access to ATMs or card networks remains uncertain. A cash buffer that is adequate for one failed payment may be inadequate for a prolonged local outage.
Recovery is also staged. A store may reopen before its terminal connection is restored, or a digital network may recover while roads, fuel supplies, or inventories remain constrained. As conditions normalize, households can replenish cash or return to cards and other electronic methods. The cash share can then fall without implying that the earlier fallback was unimportant; it has done its job by bridging a period in which the preferred route was not dependable.
This sequence generates predictions that can be tested. If weather alone drives the payment change, nearby places exposed to the same storm should show similar shifts even when their power remains available. If payment-system failure matters, the sharpest expenditure loss and the strongest movement toward cash should occur where outages are realized. The next step is to examine those comparisons, keeping the physical damage of the storm in view while asking what the outage adds.