Not all index cat bonds are the same: Why an "own" view of risk matters
Index-based catastrophe bonds promise transparency – but models can mislead. For investors, disciplined analysis and an independent view of risk are key to separating signal from noise.
Profily autorov
The insurance-linked securities (ILS) market has matured significantly over the past decade, attracting institutional investors seeking uncorrelated returns and resilience in volatile markets. Yet, as the market deepens, so too does its complexity.
Within catastrophe bonds (cat bonds), one of the largest and most liquid segments of the ILS universe, an important evolution is underway – particularly around index-based structures.
Cat bonds relying on industry loss index triggers – meaning payouts are triggered when specific, externally-defined parameters are met – have become a familiar feature of the market. But their risk characteristics, model dependencies and potential performance divergences are less well understood.
In short and beneath the surface, not all index cat bonds are created equal. Understanding the nuances – and working with managers with the capabilities to take an independent, data-driven “own” view of risk – can therefore result in material differences in performance over time.
The growing role of index-based cat bonds
Cat bonds are structured to transfer extreme natural catastrophe risk from insurers or reinsurers to capital market investors, in return for reliable and collateralised income premiums. The trigger mechanism – the condition that determines when investors’ capital is used to cover losses – is one of the most important determinants of risk and return.
Over the past decade, index-based triggers, where payouts depend on industry-wide loss estimates published by third-party agencies such as PCS or PERILS, have accounted for between 12% and 37% of annual cat bond issuance. As of mid-2025, nearly one in five outstanding cat bonds used this type of trigger.
Such index triggers are often perceived as more transparent and objective than indemnity-based structures, which depend on the specific loss experience of an insurer or reinsurer. They also tend to exhibit less basis risk from an investor’s perspective, as payouts are based on standardised loss reporting rather than an individual sponsor’s claims handling or reserving practices.
Taken together, these factors have contributed to tighter pricing for such bonds in recent years (see chart) – particularly for those covering the dominant “peak perils” of North Atlantic windstorm and global earthquake. However, this perception of standardisation can mask significant differences in structure, exposure and model reliability.
Cat bond return per unit of risk over time
Source: Artemis, Man Group, 17 July 2025.
A market dominated by reinsurers — and model risk
Historically, index-triggered cat bonds have been issued almost exclusively by reinsurers, reflecting the nature of their underlying portfolios.
Reinsurers typically manage thousands of contracts with varying renewal dates, risk models and exposure data, making aggregation complex and subject to model inconsistencies. For such portfolios, index-based triggers are operationally simpler and less dependent on granular, up-to-date portfolio data.
Yet even in this more standardised segment, investors face important variations. Specifically, a significant proportion of index-based cat bonds – roughly two-thirds – provide “aggregate” coverage, meaning protection against multiple qualifying events in a given year.
While this structure can be attractive for reinsurers seeking multi-event protection, it introduces greater sensitivity to model assumptions and event clustering, particularly for so-called “non-peak” perils.
When “non-peak” perils become material
The ILS market was conceived to provide protection against extreme but infrequent events: large earthquakes and major windstorms. However, in recent years, the drivers of insured loss have shifted.
Specifically, perils once considered “secondary” – severe convective storms, winter storms, and wildfires – now account for a majority of annual insured catastrophe losses in the US and Canada.
This evolution has profound implications for cat bond investors. The models used to assess these “non-peak” perils remain less mature and less empirically calibrated than those for major wind or quake events – and many underestimate potential losses and “return periods” (the average modelled time for an event of the same type and with the same or greater scale of losses to occur again). This is particularly true for emerging risks such as wildfire.
It is notable in this respect that for non-peak perils the key sensitivity is to frequency, rather than severity. For instance, in 2024 the US National Weather Service recorded 30,878 severe weather events (1,791 tornadoes, 20,081 wind events and 9,006 hail events), while in California alone the Department of Forestry and Fire Protection recorded 8,110 wildfires in 2024.
The model risk associated with these perils is generally known amongst professional reinsurers offering traditional indemnity coverage. Given the known deficits of the models, it is not uncommon for the lower or so-called “working” layers of a reinsurance programme to be priced on an experience basis, or a combination of modelled perils plus an experience adjustment.
ILS markets, in contrast, have focused on rare and relatively remote events, where the protection buyer’s motivation is principally capital protection. This can be seen in the weighted expected loss of outstanding cat bonds, which has ranged between 1% and 2.5% and as of 1 August 2025 stood at 2.24%. This translates into a weighted return period for the cat bond market of one event every 44.6 years.
As a result, in the ILS market, where investors typically rely on third-party vendor model outputs, the view of risk for index-trigger bonds covering both peak and non-peak perils that are exposed to multiple smaller events occurring with greater frequency, can therefore be misleading.
Case study: When modelled risk diverges from reality
Two index-based catastrophe bonds issued in 2021 — Herbie Re and Claveau Re — illustrate the point. Both offered aggregate coverage across a wide range of perils and geographies, combining peak and non-peak risks under an index trigger. On paper, their modelled expected losses appeared modest and the spreads offered seemed attractive in a tightening market.
Comparison table of bond characteristics
Source: Artemis, Schroders Capital, 31 July 2025.
However, the ensuing years revealed the limits of this model-based view. As multiple moderate events accumulated — including Hurricane Ian, Winter Storm Elliot, the Turkish earthquake, and several US wildfires — both bonds eventually experienced full principal losses.
These were the first-ever industry index-triggered aggregate cat bonds to suffer total losses since the inception of the publicly-traded 144A market in 1998. The outcome underscored how model limitations and peril interactions – particularly the aggregation of smaller, secondary events – can erode protection more quickly than expected.
Why an “own” view of risk is essential
The lessons from these cases go beyond individual transactions. They demonstrate the importance of developing an independent, internally validated view of risk – one that complements and enhances, rather than replicates, vendor model outputs.
When a reinsurance sponsor comes to the market with an index-triggered bond providing protection against both peak and non-peak perils, relying on a vendor modelled expected loss for the pricing, it is justified to question whether the sponsor really believes the model view, or whether their own view of risk substantially differs from the output.
At Schroders Capital, our ILS investment process begins with a fundamental question: does the model truly reflect reality? We assess the freshness of exposure data, the relevance of model versions, and the empirical validity of assumptions for each peril. Where data is outdated or model science lags behind observed loss trends, we make our own adjustments — applying in-house expertise to develop proprietary analyses or supplemental modelling.
For example, our modelling approach considers event clustering, where certain climatic conditions make back-to-back events more likely, such as consecutive North Atlantic hurricanes or European windstorms within the same season. Standard vendor models often overlook such dependencies, leading to underestimation of aggregate loss potential.
We also adjust for economic and inflationary changes in insured values when model data predates current exposures, ensuring our expected loss estimates remain grounded in today’s realities rather than yesterday’s assumptions.
Quantifying the difference
These adjustments can make a substantial difference. When applying our proprietary “own view of risk” (SCILS view), the expected losses for the 2021 Herbie Re and Claveau Re bonds effectively doubled relative to the vendor model estimates. Consequently, the apparent excess spread – the premium investors were expected received over modelled expected loss – narrowed dramatically.
From our perspective, the compensation offered for the additional exposure to secondary perils and regions was inadequate. By refraining from investing, we avoided what ultimately became two of the most significant losses in the index-triggered cat bond space.
Comparison of the vendor model view of risk vs the Schroders Capital ILS view of risk
Source: Schroders Capital, TouchstoneRe, 31 July 2025.
Following the 2025 wildfire losses in California, issuance of cat bonds covering non-peak perils – either alongside peak risks or on a stand-alone basis – has increased. These transactions now offer wider spreads relative to similarly structured peak peril-only bonds, reflecting the market’s recognition that vendor model outputs for non-peak risks require greater scrutiny.
However, the range of spreads offered remains broad, and the market’s confidence in modelled risk remains uneven. This reinforces the point that within the index-triggered cat bond segment, apparent homogeneity conceals significant heterogeneity in true risk-return characteristics.
For investors: Differentiation is key
For investors, the takeaway is clear. Cat bonds using index-based triggers are not commoditised instruments. Their attractiveness depends not only on headline yield but on the quality of the underlying data, the credibility of model assumptions, and the investor’s ability to interpret them independently.
At Schroders Capital, we take an own view of risk to enhance our ability to model these bonds effectively. This ability to construct and continuously refine an internal view of catastrophe risk is a differentiator – enabling us to identify mispriced risk, avoid concentration in structurally weak segments, and build portfolios that can deliver resilient, risk-adjusted returns over time.
In an era of shifting climate patterns, evolving modelling science and rising economic exposure, the need for informed judgement and disciplined analysis has never been greater. Not all index cat bonds are the same — and that’s precisely why independent thinking remains one of the most valuable assets in the ILS market.
Profily autorov
Témy