Marine Cargo Open Cover Pricing with Sparse Loss Data

Twelve claims over five years is a thin sample, and the $4.1m total may be dominated by one or two large events. Without shipment-level exposure we cannot calculate a burning cost per unit of transit value or check whether the record reflects the proposed commodity mix, routes, packing, carrying vessels or deductible structure. The 65% loss ratio is therefore an average of uncertain relevance, not a credible estimate for this account.

I would use several approaches: (i) exposure rating by commodity and route, using market benchmarks for cargo loss costs; (ii) a burning-cost analysis with credibility weighting, giving limited weight to the 12 claims; (iii) a scenario analysis for a major accumulation, such as a vessel fire or a port closure, because the cat model does not cover cargo; and (iv) a review of terms, including deductible, warranties on packing and stowage, transit limits and aggregation. I would load explicitly for parameter uncertainty and for the missing exposure data, rather than relying on the historical loss ratio. The output should be a technical rate and a range, with clear conditions.

On the available numbers, the account is already marginal before the reduction: a 65% loss ratio plus 22% expenses plus 4% facultative reinsurance cost gives 91%, leaving 9% for capital, profit and uncertainty. A 15% rate reduction raises the loss ratio to about 76.5% on the same claims and, with expenses and reinsurance, the combined ratio exceeds 100% before any capital charge. That fails the syndicate's 15% return-on-capital target unless the true loss ratio is materially lower than the historical data suggest, which the thin record cannot support.

I would not accept at the requested rate. I would offer a higher technical price, or a reduced line, or both. Conditions should include a per-shipment deductible, an aggregation limit by vessel or port, warranties on packing and stowage, exclusion of certain commodities or routes, a survey requirement, and a profit commission or experience refund only if the data improve. The relationship argument is not sufficient to underwrite at a loss. If the broker will not move, walking away is the correct decision; the opportunity cost of capital and the tail risk are too high.

If written, the account needs a data schedule before inception: shipment-level values, commodity types, routes, carriers, packing, deductible and transit limits. Without this, exposure accumulation and pricing cannot be monitored. I would require quarterly reporting of transit values and claims, with a right to audit, and a review at 12 months. Key monitoring indicators would include claim frequency, average claim size, large claims above $2m, and any change in commodity mix or route. I would set triggers for repricing or re-underwriting, such as a loss ratio above 70% or two large claims in a quarter. Reinsurance recoveries should be tracked by shipment because facultative cover above $2m may be exhausted or unavailable. The underwriter should also monitor cargo accumulation across the whole portfolio, since the cat model does not capture it. These requirements should be written into the binding authorities and the policy warranties, so that the syndicate can act if the risk deteriorates.