Comparison
Settlement Share Across Underinsurance: What a Shortfall in Cover Costs on Every Claim, Not Only a Total Loss
If the sum insured is less than the value at risk, what share of a claim is actually paid — and does the answer depend on how large the loss is?
The short answer
The share of a claim that is paid is the share you are insured for — and it is the same for a small claim as for a total loss. At 80% cover, a loss of 10% of the value pays 8.00% of the value and not 10.00%. The missing 20.00% of every claim is borne by the holder.
This is the part that surprises people. Underinsurance is widely understood as something that bites when everything is destroyed, and the arithmetic does not work that way: the proportion is applied to each claim, whatever its size. At 50% cover the holder carries 50.00% of a scratch and 50.00% of a catastrophe.
- Who it applies to
- Any cover where the sum insured is compared against a value at risk and a proportional condition of average applies — buildings, contents, stock and commercial property are the usual cases. It compares ratios, not insurers, policies or products.
- What this does not tell you
- It establishes what share of a claim a stated shortfall in cover leaves unpaid, and nothing else. It does not establish whether any policy contains such a condition, whether a loss is covered, or how often such a loss occurs.
An exact identity rather than a projection: one multiplication on two ratios, so no calculation-model version, horizon, rate or return assumption enters the answer, and nothing here goes out of date.
What is being compared
Two ratios, held against each other across a ladder. Nothing varies between rows except the share of the value that is insured, nothing varies between columns except the size of the loss, and no amount of money appears anywhere on this page.
| Levels of cover compared | 100% · 90% · 80% · 70% · 60% · 50% |
|---|---|
| Loss sizes compared | 5% · 10% · 25% · 50% · 100% |
| Quantity compared | The share of a claim paid, and the share the holder bears |
| Amounts | None. The identity is scale-free, so no sum and no currency appears |
| Value at risk | Taken as given. How it should be measured is a policy question |
| Excess or deductible | Not modelled. It reduces a settlement separately |
| Premium | Not modelled. Whether cover is worth its price is another question |
| Likelihood of any loss | Not modelled, not estimated and not implied anywhere |
Every level on both ladders is a stated level entered into the identity, chosen so the shape can be read. Neither ladder is a survey of how much anyone is typically underinsured by, no insurer or policy is named, and nothing here reports a market condition.
Where the proportion comes from
A condition of average exists so that a holder who insures for less pays less and recovers proportionately less. Where it is written proportionally, it settles a claim like this:
settlement = loss × (sum insured ÷ value at risk)
Divide both sides by the loss and the loss disappears. What remains is the whole finding:the share of a claim that is paid is the ratio of the sum insured to the value at risk, and nothing about the size of the claim enters it. That is not an approximation that holds for small losses or a rule of thumb that breaks at the extremes. It is an identity, and it is exactly as true at 10% of the value as at 100%.
The reference module refuses a sum insured above the value at risk rather than computing one. Past that point the identity stops describing what a policy pays: a settlement is capped at the loss by the principle of indemnity, so over-insurance pays the loss and not more, and printing a settlement larger than the loss would be reporting the formula rather than the mechanism.
What each level of cover pays, on any claim
Two columns, and neither of them takes a loss size. That is the point of putting them in their own table before the grid: these are properties of the cover, not of the claim.
| Share of the value insured | Share of a claim paid | Share of a claim borne by the holder |
|---|---|---|
| 100% | 100.00% | Nothing — the condition does not bite |
| 90% | 90.00% | 10.00% |
| 80% | 80.00% | 20.00% |
| 70% | 70.00% | 30.00% |
| 60% | 60.00% | 40.00% |
| 50% | 50.00% | 50.00% |
The first row is the anchor and the reason the rest are legible. At 100% there is no shortfall to have, and it is reported as a named state rather than as 0.00% — becausethere is nothing to be short of is a different sentence fromthe shortfall is small, and a table that printed a zero there would invite the reader to look for the size of a thing that does not exist.
The same proportion, at five sizes of loss
Below, the settlement as a share of the whole value at risk. Read across a row and the figures grow with the loss, as they must. Read the proportion between any cell and the loss above it and it never changes at all.
| Share of the value insured | Loss of 5% | Loss of 10% | Loss of 25% | Loss of 50% | Loss of 100% |
|---|---|---|---|---|---|
| 100% | 5.00% | 10.00% | 25.00% | 50.00% | 100.00% |
| 90% | 4.50% | 9.00% | 22.50% | 45.00% | 90.00% |
| 80% | 4.00% | 8.00% | 20.00% | 40.00% | 80.00% |
| 70% | 3.50% | 7.00% | 17.50% | 35.00% | 70.00% |
| 60% | 3.00% | 6.00% | 15.00% | 30.00% | 60.00% |
| 50% | 2.50% | 5.00% | 12.50% | 25.00% | 50.00% |
The top row is the one a reader's intuition already agrees with: at 100% cover the settlement is the loss, every time. Every row beneath it is that row scaled by a single number, and the scaling is applied to the small losses just as firmly as to the large ones.
The 80% row is the worked case. A loss of 10% of the value pays 8.00%, leaving 2.00% of the value uncovered; a total loss pays 80.00%, leaving 20.00%. The uncovered amount is ten times larger in the second case and the uncovered share of the claim is identical.
A proportion is not a policy
Everything above is arithmetic, and arithmetic is the smaller half of this question.This page cannot tell anyone whether their policy contains a condition of average, and it does not claim that every policy does. Some carry one written to bite only where the shortfall passes a stated threshold; some insurers waive it; some cover is written on a basis where the question does not arise. Which of those applies is in the wording, and the wording is not arithmetic.
The figures are also conditional on a loss having happened. Nothing here estimates how often such a loss occurs, and no number on this page can be turned into such an estimate: there is no probability in this identity, no frequency, no history and no horizon.
Two holders with the same ratio can still be in different positions, depending on:
- how the value at risk was arrived at, and when it was last revisited
- whether the figure has kept pace with rebuild costs since the policy was written
- index-linking, where the policy carries it
- whether the loss falls inside the cover at all
- the excess, which reduces a settlement separately from the ratio
- limits and sub-limits that cap particular categories of loss
No level of cover is presented here as adequate or reckless. Insuring for less costs less and recovers less; which of those matters depends on circumstances this page does not know and cannot see.
What this comparison does not determine
The figures on this page are exact arithmetic on two stated ratios. They are not policy terms, quoted settlements, claims data or a forecast of anything. The comparison cannot determine:
- whether any particular policy contains a condition of average at all
- whether a policy that does contain one applies it proportionally, or only past a stated threshold
- whether an insurer would waive it, and on what terms
- how the value at risk should be measured — rebuild cost, market value, or a day-one uplift
- how an excess interacts, or whether it is taken before or after the reduction
- what a claim would cost in time, evidence, loss adjustment or disruption
- whether a loss is covered by the policy in the first place
- whether any level of cover is adequate, affordable or suitable for anyone
- the likelihood of any loss, which is not modelled and cannot be derived from anything here
It also holds the value at risk still. A rebuild cost that rises while a sum insured stays fixed lowers the ratio over time and widens every shortfall in the table, which is how most underinsurance actually happens. That is a path, and this identity is a statement about one moment rather than about a path.
None of this is financial, legal or tax advice, and no product, provider, insurer or policy is recommended, ranked or named anywhere on this page.
Where these figures come from
Every number above is produced at build time from settlement = loss × (sum insured ÷ value at risk), applied to the two ladders in the comparison set. Nothing on this page is typed by hand, and nothing is calculated in the browser.
There is no calculation-model version behind them, and that absence is the point. A settlement share is one multiplication on two stated ratios rather than a projection, so it inherits no horizon, no assumed return and no rate. Its regression suite holds every published figure against a second, independently written implementation of the same identity, and asserts the constancy along each row exactly rather than within a tolerance — two implementations that agree are evidence, one checked against itself is not.
The build refuses a sum insured above the value at risk, and refuses a loss of nothing, rather than reporting a figure for either. Neither is a case the identity has content at, and printing a boundary that is not a quantity would be worse than printing none.