Every answer and its explanation appears here once you have finished the path. Each one then links to the matching glossary entry, where the concept is set out in full with its worked example.
1. A loss occurs midway through the works. Against what value is underinsurance measured?
Against the value of the COMPLETED works: comparing the loss to the present value is the wrong division
Comparing against the present value is what common sense suggests and it reassures wrongly, since the exposed value grows to the end: a construction policy's sum insured targets the completed works, and it is against that the ratio is computed. Taking the initial contract amount means ignoring exactly what widens the gap, the variations. As for the per-event limit, it bounds a payment and enters no proportion: confusing it with the sum insured makes you believe you checked your cover when you read a cap.
Glossary entry · principe-indemnitaire2. Three movements widen the gap between the declared sum and the real value without any document carrying the word increase. Which?
Variations, price indexation and currency movements on equipment paid abroad
None of the three presents itself as an increase in the insured value: variations are notified to the design team, price indexation is automatic and contractual, currency movements are notified to nobody at all. That is what makes the gap invisible where people look, and it is why an indexation clause conditional on periodic declarations is worth no more than no clause at all if the declaration is not made: its date appears on no works programme. The other answers name events costing time or money without changing the value of the completed works, which is the only quantity in the ratio.
Glossary entry · valeur-agreee3. The average rule applies. On what can it be argued?
On nothing of the kind: it is not a sanction and cannot be argued, applying the ratio of the declared sum to the real value with no fault or intent to discuss
It is an arithmetic ratio, and the other three answers share the flaw of importing a debate about conduct where there is none: good faith, the reasonableness of the estimate and the absence of prejudice are arguments against a sanction, and average is not one. Knowing that changes behavior, because you stop preparing a plea and go check a figure, which is the only useful step and is taken before the loss. It is also what distinguishes it from proportional reduction for non-disclosure, which is judged on good faith.
Glossary entry · principe-indemnitaire4. A sum insured frozen for two years freezes something besides the proportion. What?
The annex heads, capped as a percentage of the sum insured: debris removal and professional fees stay frozen while their costs followed real prices
The second effect is quieter than the first and is discovered at settlement: those heads are capped as percentages, so a sum that has not moved carries sub-limits that have not moved either, over works whose cost did follow real prices. You end up clearing a 2026 site with a sub-limit computed on 2024 values. The answer aiming at the premium describes a real effect and one favorable to the insured, which is exactly why nobody worries about it, and it is not where a frozen sum costs.
Glossary entry · tous-risques-chantier-car5. A policy carries an indexation clause conditional on periodic declarations. What is it worth?
It is worth no more than no clause at all if the declaration is not made, and its date appears on no works programme
The clause is in the contract, it reassures on reading, and its trigger is delegated to an administrative step no site tool prompts: that is exactly the configuration producing cover people believe they have. The useful point is therefore not obtaining it but putting its date in the programme, next to the milestones that are actually tracked. Believing in retroactive effect or a shifted burden assumes the insurer would track values it cannot see: it knows of the drift only what it is told.
Glossary entry · assurance-construction