The sum insured that drifts: variations, indices and currency

The sum insured on a construction policy is set before the first excavation, on a figure that does not yet exist: the value of the completed works. It is not a prudent estimate of an existing asset, it is a forecast, made when least is known, for a period during which everything that makes it up will move. The contract will see variations, materials will follow indices, part of the equipment will be paid in another currency, and the schedule will stretch. The contract itself carries the first day's figure, and that figure remains the benchmark against which the indemnity will be computed three years later.

What that figure measures must be seen first, because intuition from other classes misleads. On an existing building, the sum insured is the value of the asset at the time of the loss, and underinsurance is measured there. On a construction site, the sum insured is the value of the completed works, and underinsurance is measured against that final value even when the loss happens a third of the way through. A project owner reasoning that the works were only worth sixty million that day, so an eight million loss is well below, is making the wrong comparison: the ratio that decides is not loss to present value, it is declared sum to actual final value.

Three movements open the gap, and they are not corrected in the same way. Variations to the contract first, which add works: they are visible, they are priced, and yet they almost never reach the insurance contract because they are notified to the design supervisor, whose role that is not. Escalation next: price revision follows contractual indices, it applies month by month with nobody deciding anything, and over three years of an inflationary cycle it moves the value by several points with no document carrying the word increase. Currency last, where a major item is paid abroad: the euro amount on the day of the loss is not that of the day of placement, and the gap is nothing like marginal on a line worth tens of millions.

Policies are not without answers, they offer three, and each has a flaw that surfaces later. The escalation clause, providing for an automatic uplift of the sum insured up to a stated percentage, is the best of the three because it demands no vigilance; but its ceiling was set on the inflation expected at placement. Periodic declaration of values, adjusting the sum against declared statements, is accurate when it is done; it assumes somebody does it, on dates nobody wrote into a works schedule. The adjustable premium mechanism last, which reconciles after completion against the final value, is the most dangerous in its apparent elegance: the adjustment happens at the end, therefore after the loss, and discovering then that the actual value exceeded the declared one helps nobody.

The penalty for the gap is the proportional rule of capitals, and it deserves to be distinguished from proportional reduction for non-disclosure, with which it is often confused. The latter punishes a failure to inform and is computed on premiums; the proportional rule of capitals punishes nothing at all, it applies a sharing arithmetic: an insured that declared eighty percent of the value is deemed to have retained twenty percent of the risk, and the indemnity follows that ratio. There is no fault, no good or bad faith, no possible argument about intent. It is the mechanical consequence of a figure, and that is why it cannot be pleaded.

A secondary consequence is often costlier than the first, and it goes unnoticed until settlement. The ancillary heads of a construction policy, debris removal, adjusters' and design fees, expediting expenses, are almost always capped by a percentage of the sum insured rather than by an absolute amount. A sum insured left at the first day's level therefore carries sub-limits left at that same level, while the costs they cover have followed actual inflation. Clearing a damaged structure is paid at the prices of the day of the loss, not those of the day of placement, and the shortfall adds to the proportional reduction rather than replacing it.

What makes the subject hard is therefore not its technique, which is one division, but the fact that it belongs to nobody. The schedule is kept, the cost is kept, the punch list is kept, each by somebody whose job it is. The insured value is kept by nobody after placement: the broker does not see the variations, the design supervisor does not read the policy, the cost controller follows a budget whose structure is not that of the sum insured. The useful question is therefore asked in a single act, at regular intervals and not at the time of the loss: what is today's foreseeable final value, and what does the contract say it is worth. Two figures, one subtraction, and a gap that cannot be closed once the loss has happened.