Introduction — The Price Assigned to Us
An insurance price distributes a cost among people whose risks differ. Actuarial pricing seeks to measure these differences so that an insurer can estimate the resources it will need to pay claims. It considers how often claims occur, how much they cost and what the contract covers. It may also vary the deductible, which is the part of a covered loss that remains with the policyholder.
From the first mortality tables to contemporary scores, insurance professionals have learned to distinguish increasingly precise groups. They speak of segmentation, adverse selection,1 credibility2 and expected cost [1, 2]. The question is not whether differences should ever matter. It is which differences should change the premium or coverage and which should continue to be carried together.
Vincent had spent seven years in the home insurance pricing team of a large insurer. One Monday morning, he presented the results of a new model. In several municipalities, water-related damage was costing more than expected. The model could now distinguish addresses according to the slope of the land, their proximity to drainage networks, the type of soil and the local history of claims.
The total increase that had to be financed was known. The question was how to distribute it. One option was to raise premiums slightly across the whole portfolio. Another placed most of the increase on the addresses with the highest estimated exposure. A third combined a smaller increase with assistance for installing backflow valves, raising equipment or improving drainage.
Vincent preferred the second option. It seemed more accurate. A European insurance federation had expressed the principle simply. Each policyholder should pay a premium that reflects the risk of loss they bring to the pool [3]. In insurance economics, a premium equal to the average expected cost of the risk is often described as actuarially fair [4].
A colleague asked what the risk belonging to each house meant in this case. Some pipes belonged to the municipality. Planning rules had allowed basements to be built. Some owners could afford protective work and others could not. The increase arose from the same set of losses, but several ways of sharing it remained possible.
The model had measured a difference. It had not decided who should pay for it.
I trained as an actuary and worked in the profession for several years. I know this reasoning from the inside. Part of the job is to identify differences, measure them and calculate the resources an insurer needs to keep its promise. Without that work, prices become arbitrary and the contract may be impossible to finance.
Yet the move from an estimated cost to the price charged is never automatic. An insurer can assign the additional cost to one address, spread it across a wider group, alter the coverage, finance preventive work or withdraw. Calculation informs these choices. It does not make them for us.
A Promise under Strain
Insurance rests on a simple idea. The financial consequences of a fire, illness, death or flood should not fall entirely on the person affected. Contributions are gathered before the event so that the person who suffers the loss can make a claim on collective resources [5, 6].
This promise has never been unconditional. Contracts contain exclusions, limits, deductibles and duties to disclose information. Insurers select risks, investigate claims and sometimes dispute requests for payment. The history of insurance is also a history of conflict over who may enter, which losses deserve cover and what evidence must be produced before payment is made [7, 8].
Policyholders pay before they can fully know the value of the contract. They may never learn how the company would have handled their claim. When a loss occurs, they discover the real reach of the coverage at the moment when they depend on clauses, deadlines, loss adjustment3 and the insurer’s interpretation. Trust is therefore part of the product itself.
Data now allow institutions to distinguish more finely among places, bodies, behaviour, histories and trajectories. People once placed in the same class receive different estimates. Greater precision can correct crude categories and better recognise actions that genuinely reduce risk.
It can also reduce the uncertainty that policyholders carry together. Personalised pricing is then presented as a requirement of fairness. Why should a careful driver pay for a dangerous one? Why should a home with lower exposure contribute to the losses of a threatened area? Why should someone who follows preventive advice pay as much as someone who ignores it?
These questions are legitimate. They are not enough to define a just price. A premium that is too low can conceal danger, preserve exposure and weaken the insurer that will later have to pay. A highly precise premium can instead make protection inaccessible to the person who already carries the risk. The disagreement concerns how the burden should be shared. Which differences should alter the price? Which should remain within the pool? What should be offered to someone who is warned about a danger they can neither reduce nor bear alone?
Can We Really Refuse Insurance?
The significance of a price depends on the real freedom to refuse the contract. In some markets, policyholders can compare offers, reduce their coverage or decide not to buy. That freedom is often imperfect, but going without insurance remains possible.
In other situations, insurance controls access to a good or an activity. Motor insurance is required to drive. A lender will usually require insurance before financing a home. Some professions cannot be practised without liability cover. In the United States, health insurance remains closely tied to employment [9].
The contract remains private in form, but its absence can close access to mobility, credit, work, housing or care. Looking elsewhere is not always an answer. Another insurer may not make an offer. The price may exceed the household’s resources. A scheme of last resort may provide narrower cover at a higher cost.
The state is never absent from these markets. It requires some forms of insurance, defines benefits and duties, supports credit, organises public schemes and absorbs part of the loss when private arrangements no longer suffice [10, 11]. The more a person depends on insurance to lead an ordinary life, the less formal freedom of contract can justify the price, selection and conditions imposed upon them.
From Contribution to Personal Price
French uses two closely related words. Prime suggests the price of a contract. Cotisation recalls membership in a fund, mutual or social scheme. English does not preserve the distinction as sharply, but premium and contribution still direct attention in different ways.
In both cases, the money paid supports a common capacity to meet claims. It must be calculated so that the institution can pay losses, cover operating expenses and retain enough resources for bad years [1]. The two words do not describe entirely separate worlds. They change what comes into view. A premium draws attention to the exchange between a company and a customer. A contribution recalls that many people finance the same promise.
The shift occurs when a common contribution is presented as the exact bill for a profile. Future risk then appears to belong to the person before the event has occurred. Classification no longer serves only to estimate what the group will have to pay. It determines where each part of the cost will be sent.
This changes the meaning of insurance. The institution becomes less concerned with making vulnerability bearable and more concerned with bringing each person closer to the price of their exposure. The solidarity that remains can then appear as a measurement error or an unjustified subsidy rather than a decision about what should continue to be shared [12, 13].
Owning One’s Risk
This shift belongs to a broader transformation of responsibility. Institutions increasingly ask people to construct their own trajectories, anticipate difficulty and answer for outcomes whose causes often extend far beyond their choices [14].
Policyholders are asked to preserve their own access to protection. They should choose the right neighbourhood, avoid certain hours, maintain the home, drive according to the model’s indicators and provide the data that demonstrate prudence. Prevention can then become a series of individual tests that must be passed in order to remain insurable.
The means to act are very unevenly distributed. Moving, strengthening a house, leaving an exposed occupation, replacing a vehicle, correcting a file, understanding a model, paying for an expert assessment or renegotiating a loan all require money, time and alternatives. A premium may reflect the calculated risk and still push a household below a minimum level of material security [15].
Collective protection gives substance to autonomy. It allows people without substantial wealth to depend less completely on the next wage, accident or refusal [16, 17]. Returning risk to individuals in the name of freedom can remove the resources that made choice possible.
A local authority, a bank and an insurer use the same flood map. For a house near a river, the model indicates high exposure.
The local authority uses the map to offer assistance with removable flood barriers and raising the boiler. The bank asks for an assessment before approving a loan for the work. The insurer raises the deductible and announces that it will review the policy at the next renewal.
None of the three institutions disputes the map. They do not give it the same role. The local authority uses it to open access to prevention. The bank treats it as a reason for further verification. The insurer uses it to reduce the protection it provides.
The owner therefore receives three different messages from the same estimate. The risk should be reduced. The proposed work must be documented. A larger share of any future loss will remain with the household.
These decisions were not contained in the map. Each institution made them.
What Numbers Do Not Decide
Insurance protects by drawing boundaries. It sets the conditions of entry, the price, the coverage, the duration of the contract and the events that create a right to payment. It allowed some accidents to be treated as risks to be shared rather than as private faults or misfortunes [5, 18].
This book takes price as its starting point in order to examine how contemporary societies distribute protection. Numbers inspire confidence because they appear to stand outside judgement. Yet data never arrive alone in a file. A claim, an exposure, a behaviour, a neighbourhood and an observation period must first be defined [7, 8].
The expression “raw data” is misleading. Data result from choices about what will be observed, retained and made comparable [19]. This does not make numbers arbitrary. Statistical categories rely on conventions that allow institutions to compare distant situations and coordinate decisions [20, 21].
Calculation does not remove choice. Choice enters the definition of the variable, the collection of information, the construction of the model, the placement of the threshold and the consequence attached to the result. In insurance, that consequence may be a premium, a deductible, a request for protective work, an investigation or a refusal.
Older rating tables already used age, sex, occupation, vehicle use and family status. They confined people within group averages and gave economic force to social categories. Their rules were nevertheless relatively easy to identify. An agent could open a manual, ask a few questions and point to the line that determined the price.
A contemporary score may combine far more attributes, their interactions and their changes over time. It can correct some of the approximations found in older classes. It can also make it harder to know which information mattered, who received comparable treatment and which rule should be challenged [12, 13].
A prediction must therefore be judged by more than its statistical quality. We must also examine the institution’s right to use it, the decision it prepares and the route of appeal left to the person [22, 23]. An accurate score does not by itself justify the consequence attached to it.
Another asymmetry runs through this chain. The institution chooses the model, threshold and action. The person receives an increase, investigation or refusal they did not choose. A future loss may be a risk to manage for the actor making the decision and a danger for the person who bears its effects [24]. Analysis must therefore address power as well as accuracy. Who defines the boundary? Who lives with its effects? Who can ask for it to be corrected or redrawn?
Climate as a Test
Climate disruption makes these questions harder to avoid. Past series describe some possible futures less reliably. Losses can become concentrated in the same place and at the same time. Models then acquire greater authority over premiums, coverage conditions and decisions to withdraw.
In the United States, the conflict appears directly between affordable premiums and the continued availability of homeowners insurance in several markets [25]. France and the United Kingdom distribute more of some costs through national arrangements. No architecture removes the danger. Each decides when it enters the household budget and how the collective takes over what a private contract cannot carry alone.
A map may open access to protective work, assistance or a negotiated relocation. It may also announce a higher premium, non-renewal and a decline in the value of a home. Greater precision does not choose between these uses. Institutions decide what follows from the classification.
Climate therefore reveals a tension already present in all insurance. Calculation must make risk visible without serving only to identify the people who will now have to bear it alone. A society must be able to signal genuine exposure, finance its reduction and share the part of the loss that could not reasonably have been avoided.
The Route through the Book
The book focuses mainly on French, American and British systems. It also draws on European comparisons and several regional arrangements in the Caribbean, Africa and the Pacific. It is not a global history of insurance. It compares different ways of turning knowledge of risk into a price, a right, a procedure or a withdrawal.
The first part, “Insuring, or Carrying the Promise”, begins with the institutions that organise and own the pool. It then asks what makes cooperation durable. It follows the fragmentation of protection, the transformation of private misfortune into a collective claim and the moment of loss, when the value of the contract depends on evidence, delay and the power of the institution expected to pay.
The second part, “Pricing, Seeing, Classifying, Judging”, enters the making of prices. It shows how insurance brings contributions together while distinguishing among policyholders. It examines the blind spots of the portfolio, behavioural data and the conversion of a score into a procedure.
The third part, “When Protection Withdraws”, begins when insurance becomes too expensive, offers only reduced coverage or disappears. Climate brings choices about planning, prevention and adaptation into the contract. The final chapters follow people and places made uninsurable, collectives that bring separate files together and institutions able to finance prevention over time.
No actor possesses the right answer alone. The whole chain must be followed, from information collected to consequence imposed. At each stage, we must ask who decides, who pays, who can act and who can challenge the result.
References
Adverse selection refers to the tendency of people with greater exposure to buy or retain more coverage when the price does not distinguish risk levels well [1].↩︎
In actuarial work, credibility describes how the experience of a particular group is combined with information from a larger population when the group contains too few observations for its own record to be fully reliable [2].↩︎
In claims handling, loss adjustment is the technical assessment of the cause and extent of damage and of the cost of repair or replacement. The insurer appoints its own adjuster. A policyholder may also appoint an independent expert, particularly when the assessment is disputed.↩︎