TOP
The Hull market continues to be in a soft cycle in 2026. Capacity remains abundant and rates are under pressure.
At the same time, repair costs and insured values have increased. Labour, steel and spare parts have become more expensive and less predictable. Claim costs per vessel are roughly one-third above pre-pandemic levels.1 The world fleet continues to age, which makes the relationship between repair costs and insured values more sensitive. In a line where margins are typically modest, this increases the pressure on underwriting results.
That is why the topic of indexation continues to come up in renewal discussions. When 10 years of Hull experience are examined, a key question is whether the figures are comparable. A claim from 2016 and a claim from 2025 may carry the same currency sign, but they do not describe the same economic environment. Steel has been through a full cycle since then. Wage development has differed materially between Asia and Europe. Vessel values have changed. And, the premium and terms behind those risks belong to market conditions that have hardened and softened over time.
This is particularly visible in Treaty reinsurance, where historical experience often combines different underwriting years with different claim types, and underlying portfolio developments, such as changes in the underwriting team, strategy, risk appetite, and the corresponding exposure development. In discussions about Hull inflation, the usual reflex is to search for a single market inflation number. The problems start, however, when that single number is expected to answer several questions at once.
“Indexation” is not one task but two, and each asks a different question:
In Hull insurance, no single index does both jobs.
Even within loss trending, not all losses behave the same way:
Take a partial loss first: the repair cost for damage to a vessel was EUR 4 million in 2016, and repair costs have risen by about 40% since then. Trended for cost, the same damage would be closer to EUR 5.6 million today. The damage to the vessel is the same, but the repair costs have increased. That is what a repair index captures well.
A total loss operates differently. A constructive total loss (CTL) does not follow the development of repair costs in the same way. Instead, it is driven by the insured value of the vessel, as this value represents the threshold against which repair costs are measured.
Insured values change for their own reasons, such as the vessel’s age, market cycles, and freight rates. A repair cost index does not reflect these shifts in insured values, but it remains an important consideration. In fact, the likelihood of a total loss increases when insured values rise more slowly than the increase in repair costs.
A primary insurer usually knows, claim by claim, whether a loss was a repair or a total loss. That split is often not available at the Treaty reinsurance level, so an index that reflects a mix of repair-driven and value-driven losses may be needed. The insured-value component is how we pick up total-loss dynamics that we cannot separate manually.
In practice, if it is possible to split the losses by type, a blended index may not be needed at all. Hence, two separate indices may be preferable. A blend is the answer to a data problem, not a statement that mixing is always the right approach.
On the loss side, this framework uses four components as a practical balance between completeness and usability, while avoiding unnecessary overlap between underlying factors. These components are:
Repairs are labour-intensive, and labour is local. Wage development in Asian and European yards has not moved in parallel, and a single global index does not capture these differences. An ocean-going fleet that is predominantly repaired in Asia should have different weights than a local or short-sea portfolio.
Steel prices are cyclical and can move sharply, as seen in the 2021–2022 spike. These movements feed directly into repair costs and can materially affect loss severity. In practice, using the widely available U.S. PPI HRS (Producer Price Index Hot Rolled Steel) is a pragmatic proxy but has its limitations. U.S. and Asian steel prices do not always move in parallel, and tariffs can widen that gap. Where available, regional series should be preferred depending on the regional footprint of a portfolio.
Cost items beyond steel and labour, such as logistics, port services or overhead, lack a dedicated regional index. Broad inflation serves as the residual bucket; not because it measures these items well, but because it is readily available and locally applicable.
The piece a pure repair index leaves out. As mentioned above, vessel values are not repair inflation, but they drive total-loss behaviour because they set the threshold at which a repair is no longer economical.
The weighting of these components is not fixed. It is a design choice, and the decision should reflect the characteristics of the underlying portfolio. The framework is therefore best understood as a toolbox.
So far, we have only looked at losses. The premium side also needs attention. As the name suggests, premium on‑leveling is not applied to claims but to the historical premium, to express what that premium represents in today’s terms.
On‑leveled premium is often used as an exposure measure and as the denominator against which (trended) losses are judged. For this to be meaningful, one unit of premium has to stand for a comparable amount of exposure over the years. A premium in a soft market does not represent the same exposure as the same nominal premium in a hard market.
For the premium index, two drivers do most of the work:
Because the rate-change history is portfolio-specific and not public, on‑leveling cannot be outsourced to a single market index.
A flat premium index does not mean that premium has not grown. It only means that, on a comparable basis, insured values and rate levels have not moved. If premium increases from one year to the next, the increase may come from new business. That premium remains part of the analysis, but it should not be mistaken for an on‑leveling effect. Trying to capture such growth in the on‑level factor would lead to double counting.
This is a fair and common question, but a claims index measures severity. It tells you what a loss costs today, not how many losses occur.
Expected loss is a product of frequency and severity. The framework treats them separately.
Frequency is a separate index, and it measures the expected claim count per unit of on‑leveled exposure. This is where the premium index comes in: because on‑leveling removes the effect of rate changes, on‑leveled premium serves as a proxy for risk volume. Folding frequency into the severity index instead would mix two questions that are better kept apart.
The International Union of Marine Insurance (IUMI)’s Hull inflation index, enhanced again in its 2026 publication, provides a transparent view on repair cost inflation. It is designed to explain the development of Cefor’s average Hull claim costs excluding total losses. That scope (i.e., excluding total losses) is a feature, not a flaw. For repair inflation, it is a very strong starting point.
The benchmark answers the question it was built for. Once a reinsurance history also carries total losses, the analysis needs to go a step further. That is not a criticism of the benchmark, but a reminder to match the tool to the question. The framework described above should be understood as complementary to IUMI’s approach. Where the two align, it builds confidence in the underlying assumptions. Where they diverge, the gap usually reflects a design choice.
Stronger risk management and active claims handling, for instance steering repairs towards more competitive yards, can have a positive impact on individual losses.
On the other hand, headline-making topics such as lithium-ion cargoes, stricter legal regimes and ever-larger vessels with even more challenging fire-fighting scenarios can have a negative impact on loss developments. In softer phases, pricing pressure is also often accompanied by an erosion of terms and conditions.
As noted at the outset, it should also be considered that historical experience often reflects a combination of different underwriting years, varying claim types, and underlying portfolio developments. These include changes in the underwriting team, strategy, risk appetite, and the associated exposure profile.
These factors may not be fully reflected in the underlying data and may be difficult to quantify, but they remain important considerations in the underwriting process.
Especially in Non-proportional Treaty reinsurance, timing introduces an additional dimension. Contracts are typically written for annual or longer periods, while underlying cost levels may shift during the coverage period. As a result, unforeseeable developments such as inflation shocks, as seen during the Persian Gulf conflict in 2026, have a disproportionate impact on excess of loss layers and are not reflected in the pricing of in‑force contracts.
The first question is not which index to use, but what the index is expected to answer. In Treaty reinsurance, loss trending, premium on‑leveling and the treatment of mixed repair and total-loss experience are separate analytical tasks. A single number may be convenient, but without these distinctions it can obscure rather than clarify the analysis.
A robust framework should make these choices transparent and allow underlying assumptions to be tested and challenged. Neither IUMI nor the framework described above aims to provide a definitive answer. This is intentional. Both are designed to help structure the problem rather than resolve it. Their value lies in making assumptions explicit and enabling comparison, discussion, and ongoing refinement.
It is also important to recognize that indexation cannot capture every aspect of the risk. This is not a limitation of the framework but rather a clarification of its intended scope. Indexation and underwriting judgement should be seen as complementary tools, not substitutes for one another.
