Chapter 13 — Building an Institution for the Non-Event
The Portfolio That Leaves
A counter-institution can take back part of a rule. It must still pay for whatever will prevent the damage. Prevention then meets a simple difficulty. The expense is certain and immediate. The loss avoided will never appear in a claim file.
The town had flooded twice in four years. The first time, local officials called it an exceptional event. The second time, no one knew what to call it. An exception could not become permanent.
Water had entered through the same streets, cellars and garages. Residents brought out the same pumps, mops and insurance files. The council approved a programme of work. A retention basin would be enlarged, backflow valves installed, riverbanks repaired and several paths rerouted.
For two years, the town lived with machinery, dust, closed roads and slightly higher taxes. Then nothing happened. Five years passed without a major flood. Neighbouring towns that had not completed the same work did not flood either. The question returned. Had it been right to spend so much to prevent an event that might never have occurred?
At an insurance company, an actuary proposed paying for part of the protection needed by the most exposed homes. The programme covered backflow valves, raised boilers, flood barriers and drainage work. The expense would be immediate. Fewer claims would appear gradually. After five years, the expected reduction in losses was meant to cover the programme’s cost.
The pilot was approved. Two years later, almost half the contracts concerned had left the portfolio. Some households had sold their homes. Others had found a slightly cheaper premium elsewhere. They took better protected buildings with them, while another insurer would receive the future savings.
Prevention had reduced the risk. It had also exposed a weakness in the annual contract. Protection remained attached to the building, while the policy and premium moved on.
The problem does not require a choice between prevention and compensation. Even after serious work, some losses will remain too uncertain, too closely linked or too large for one household to bear. Insurance retains its own purpose. It gathers resources before the event and pays the part of the loss that could not reasonably have been avoided.
The imbalance appears when insurance is expected to compensate indefinitely for exposure that planning, standards or poor maintenance continue to increase. It also appears when the absence of collective prevention becomes a reason to withdraw cover from a household that had no power to act. An institution for the non-event must connect two operations without confusing them. It reduces possible loss before the event, then shares what remains.
Giving the Non-Event a Status
Prevention must justify spending through a catastrophe that did not occur. Without the levee, wetland or warning system, the losses would have been greater. This is a counterfactual claim. It compares the world we observe with another possible world in which the intervention did not exist. Archives, comparable territories and models can make the comparison credible. They can never give it the visible force of a collapsed wall.
Old catastrophes are forgotten. People tend to be optimistic and to favour spending whose results appear quickly. These tendencies encourage underinvestment in protection against rare events [1]. Annual budgets, terms of office and short contracts make the problem worse. Present expenditure is certain. Future benefit remains open to argument.
The Sendai Framework asks governments and other institutions to prevent new risks and reduce existing vulnerability rather than concentrate resources on emergency relief and reconstruction [2]. The principle does not provide a budget. It does require inaction to be treated as a decision rather than as the absence of one.
Extreme scenarios serve a related purpose. They do not predict the next event in detail. They test whether warnings, shelters, networks and responsibilities would still work in a situation more severe than those already observed [3]. Their value lies in the decisions they prompt and in their revision when experience contradicts the assumptions.
A counterfactual can also be assessed across a programme. France’s state-owned natural catastrophe reinsurer, Caisse Centrale de Réassurance, estimates that one euro spent on natural hazard prevention avoids about three euros in damage on average.1 The results vary greatly with the hazard and the measure used [4]. This ratio does not prove that every project pays for itself. It does prevent prevention from being described as spending with no measurable return.
Assessment is harder at the level of one building. Institutions have little consistent information about work that was actually completed, what it cost, whether it was maintained and what public support helped finance it [4]. They may be able to estimate the collective value of a programme without following the precise history of one roof, valve or elevation.
The benefits are also dispersed. A strengthened roof protects the owner, insurer, bank and sometimes neighbouring buildings. A retention basin reduces claims, protects property values, lowers emergency expenditure and helps public services remain open. No actor receives the whole benefit. This helps explain why none has a strong incentive to finance the entire project [5].
Giving the non-event a status means identifying the risk, the intervention, its cost, its duration, those who benefit and the scenario against which it is assessed. This visibility must not turn prevention into a test of merit. A map serves prevention when it opens an assessment and practical means of action. It becomes an instrument of selection when it leads only to a surcharge, an unfunded requirement or non-renewal.
Warning Without Losing Trust
Chapter 5 examined trust when the insurer had to honour its promise. Here trust matters before damage. A siren, bulletin or evacuation order works only when people believe that the authority has good reasons to demand immediate action. Yet the alert must be issued before danger is certain. Otherwise it arrives too late.
In 1976, unrest at La Soufrière volcano in Guadeloupe first led people to leave on their own, then to an official evacuation of southern Basse-Terre. The major eruption that had been feared did not occur. Its absence did not erase months of displacement, interrupted work, business failures and permanent departures. It did change how the decision was judged afterwards. If the catastrophe had not occurred, had evacuation been unnecessary? [6]
This is the cry wolf dilemma. A lower warning threshold reduces the chance that disaster arrives without notice. It also creates more evacuations that are not followed by the feared event. A higher threshold reduces false positives, which are warnings not followed by the event. It increases the risk of an accurate warning that comes too late. The threshold chosen today therefore affects how tomorrow’s message will be received [7].
Trust does not always collapse after a false alarm. One study followed communities that had evacuated twice before hurricanes changed direction and struck farther north. It found no general refusal to leave when the next storm approached [8]. Residents did, however, rely less on the official order alone. They looked for other information about the storm’s path, intensity and proximity. The authority did not become inaudible. It lost its monopoly over interpretation.
Trust is better protected by explaining why action may be reasonable before certainty than by promising that every warning will be followed by disaster. Adding a probability can improve understanding and decisions more than simply reducing the number of false alarms [9]. But the same probability can produce different choices when expressed in words or numbers. People also often misunderstand the time window to which it applies [10].
A warning institution should say what it knows, what remains uncertain, how much time action requires and what follows from each possible error. It should also explain afterwards why the decision was reasonable, what was learned and what will change at the next alert. An evacuation without catastrophe is not necessarily a failed warning. It becomes corrosive when no one accounts for the decision.
When the Insurer Also Carries the Long Term
In the early nineteenth century, some New England manufacturers installed fire walls, water reserves and safer processes in their mills. They complained that ordinary insurers continued to price them like less protected businesses. The cost of the work remained with the manufacturer, while the policy barely recognised its value.
In 1835, several manufacturers gathered around the industrialist Zachariah Allen to create a mutual insurer for companies willing to accept inspections and improve their sites. The organisations that became the Factory Mutuals did more than compensate fire losses. They sent engineers, compared processes and tested equipment in order to prevent fires before having to pay claims [11–13].
Collective ownership reduced the problem of the departing portfolio. Savings from fewer fires strengthened reserves, supported lower rates or returned to members. Engineering became a common investment rather than a benefit that a customer could immediately carry to a competitor.
The organisation also produced knowledge that no single mill could have built alone. A fire in one establishment changed recommendations in others. Inspections made buildings, machines and protections comparable. Engineers, archives and a budget gave an institution to the fire that should not happen again.
The model had clear boundaries. A mill that refused the work could pay more or leave the group. Businesses able to invest formed a collective that was easier to protect than poor households or indebted municipalities. The mutual could become a club for good industrial risks.
Its lesson is narrower, but important. The annual horizon of a policy is not inevitable when the people who finance prevention, carry the losses and retain the surplus belong to the same architecture.
Prevention Must Outlast the Contract
Physical protection stays with the property when the policyholder, owner or insurer changes. The duration of the policy must therefore be separated from the duration of the investment. An annual insurance contract can recognise a measure that remains useful for twenty years, provided that the work leaves a credible and portable record.
Portability means that completed work remains recognised when the policy or owner changes. A strengthened roof, building elevation or drainage system can receive a certificate attached to the property. The certificate records the measure, its date, the evidence supporting it, its useful life and its maintenance requirements. A buyer, bank, municipality and new insurer can then rely on the same information.
Recognition requires checks. Roofs age, vegetation grows back and safety procedures are sometimes abandoned. Certification must be capable of renewal, reduction or withdrawal. Such a decision should be explained and open to challenge.
Another mechanism could compensate an insurer that paid for work when the contract ends before the expected savings have been realised. The system would be difficult to design. It would nevertheless make explicit what the annual contract hides. The institution paying for prevention is not always the one that receives the avoided loss [14].
Benefits also move between properties. In places where homes meet forests and dense vegetation, a burning building can project flames and embers towards its neighbours. The resistance of one roof, the maintenance of one plot and the spacing of buildings therefore alter other people’s exposure [15].
Economists call this an externality. It is a cost or benefit that affects people who were not direct parties to the decision. Strengthening one home in a neighbourhood exposed to wildfire may protect nearby homes. The effect also depends on how many other buildings are protected. An isolated improvement may do little for the whole area. Once a neighbourhood approaches a sufficient level of protection, one further intervention may interrupt more chains of fire spread [16].
Prevention must therefore be organised at the scale at which loss spreads. A homeowner can act on a building. They do not control the evacuation road, sewer network or permit issued for the neighbouring plot. A premium reduction can recognise individual effort. It cannot finance common infrastructure on its own.
Making Prevention Visible without Capitalising the Invisible
Making prevention visible also creates a risk that appearance replaces substance. A communication campaign, a new underwriting tool or an ordinary management expense can be presented as risk reduction. Each programme should therefore identify the hazard, intervention, expected mechanism, people or places concerned, time horizon and level of evidence [14].
The evidence will not always take the same form. Some measures rest on tests. Others rely on engineering standards, catastrophe models, actuarial experience, epidemiological research or an audit. Serious reporting should distinguish certified, observed, modelled and experimental interventions. Demanding perfect certainty would prevent investment. Accepting every claim would empty prevention of meaning.
Recognition does not justify treating an avoided loss as an accounting asset. An asset is a resource that can help meet future obligations. An insurer cannot pay claims with a hypothetical saving. Prevention expenditure may remain an expense while being reported separately, with information on amounts, beneficiaries, methods, duration and observed results [14].
The non-event then becomes visible and open to scrutiny without being turned into fictional capital. Reporting makes commitments comparable, shows how benefits are distributed and preserves a record when managers, owners or contracts change.
Markets sometimes recognise prevention when its benefit is durable and attached to the property. Measures that reduce hurricane damage can be partly reflected in real estate values [17]. Recognition is weaker when the work benefits neighbours, depends on continuing behaviour, reduces a rare risk or requires coordination across a territory.
Public action does not need to replace what markets can already recognise. It should concentrate on benefits that markets observe poorly or cannot attribute to one person. It should also prevent future value from going only to households able to pay the initial cost.
Turning the Signal into a Route to Action
Chapter 10 distinguished a signal from a boundary. The difference becomes concrete when information about risk opens a solution before it changes the premium or cover. An assessment without finance leaves the household with an obligation. Assistance without a verifiable standard gives no assurance that the work will reduce vulnerability. A discount without portability may vanish when the insurer changes.
Louisiana’s FORTIFIED programme connects these elements.2 It combines a technical standard for roofs, a public grant, certification and mandatory premium discounts based on the level of reinforcement and the region. By March 2026, more than eleven thousand roofs had been certified. More than four thousand had received support through the grant programme [18]. A model estimates avoided losses, an independent body verifies the work, public funding reduces the initial cost and the premium returns part of the benefit to the household.
The arrangement does not remove inequality. The household must qualify, may need to advance part of the cost and must maintain the protection. It does, however, turn prevention into more than advice addressed to people who can already afford to act. The signal opens finance, portable proof and a reduction attached to the property.
The principle extends beyond housing. Information about health, driving or exposure serves prevention when it opens support and leaves enough time to act. It becomes selection when an organisation demands action that a person cannot finance or obtain permission to complete. A surcharge or requirement affecting an essential need should therefore come with an assessment, time and proportionate means. Without an instrument, a warning becomes an announcement of withdrawal.
Those means should not be limited to people who already hold insurance. A levee also protects tenants, declined households and future residents. A programme confined to each insurer’s portfolio would reproduce the blind spots described in Chapter 7. Prevention must follow the property, territory or exposed community beyond the commercial relationship.
Building Institutions That Can Carry the Long Term
Several principles can guide funding. An actor that created or increased exposure bears some responsibility. An actor that benefits from collective protection may also contribute. Duties should follow the power to act. Municipalities control planning, owners control some work, infrastructure operators control networks and the state sets standards [5, 19]. These principles must be adjusted for ability to pay. Otherwise adaptation will remain reserved for secure property owners.
The role of insurance becomes clearer when three capacities are separated. Insurance produces knowledge about risk, transfers residual loss and can contribute to long-term investment. Residual loss is what remains after reasonable prevention. These capacities support resilience only when they sit within public risk management, infrastructure investment and social protection [20].
Insurance does not replace planning, a hospital, a levee or a sewer network. It finances some consequences, produces signals and may help fund particular works. When the contract becomes the last defence against the failure of a network or housing policy, the premium absorbs costs that the insurer can neither prevent nor control [21].
Individualising risk does not mean that government disappears. It often demands more evidence and surveillance. Each person must document conduct, complete work and answer for the result [22]. An institution for the non-event should reverse this logic. It should not merely ask individuals to behave prudently. It should organise the collective means of prudence.
Elinor Ostrom’s work offers principles of governance rather than a model to copy. Rules should fit local conditions. Monitoring should be accountable, sanctions graduated, appeals available and several levels of decision connected [23, 24]. Protection adds another requirement. Its boundaries cannot exclude those who have already been pushed out of the market.
A meeting without data, a budget or influence over the decision remains consultation [25]. A collective must be able to follow the work, discuss priorities and revise the rules. It may take the form of a prevention fund, territorial governance or national pooling of particular hazards. The form matters less than the powers and resources it brings together.
Procedure must also act before an irreversible consequence has produced all its effects. People should know which data were used, correct errors and have access to an appeal. Aggregate results should be published. Institutions should track non-renewals, increases, completed work, abandoned applications and returns to the ordinary market. Without such information, a sequence of private decisions can transform a territory without ever appearing as a collective decision [26].
An institution for the non-event does not promise to prove every catastrophe it prevented. It organises collective action despite that impossibility. It makes spending visible, distributes its cost, preserves a record of completed work and prevents the institution that paid from systematically losing the benefit when the policy or owner changes.
Returning to the Map
The house in the prologue had not burned. A model had nevertheless changed its social future. Prevention must intervene before catastrophe, at the point when the map becomes a bill or a withdrawal decision.
The map can then open something other than classification. It can trigger an assessment, identify who controls the means of action, preserve evidence of completed work and organise funding. It does not remove danger or the need to share residual loss. It prevents more precise information from being used first to isolate the person who remains exposed.
An avoided catastrophe produces no dramatic photograph and no compensation file. A warning not followed by damage may even create suspicion that the danger was exaggerated. Both still need a budget, a memory and rules. People who receive the map before the water, the score before the care or the price before the loss should not have to wait for destruction before receiving the means to act.
References
Caisse Centrale de Réassurance is a French public reinsurer owned by the state. It provides reinsurance for natural catastrophes with a state guarantee.↩︎
FORTIFIED is a voluntary programme developed by the Insurance Institute for Business & Home Safety. It certifies building improvements that go beyond ordinary construction standards.↩︎