Vantage Re is a Bermuda-based reinsurer writing global property catastrophe, specialty lines, and a growing book of climate-linked risk. On a Friday afternoon, the cat modelling team sends out the quarterly portfolio update, the same PDF to six different recipients.
By Monday, six different conversations have started. The head of underwriting is asking about a specific treaty in the Gulf of Mexico book. The group actuary is comparing the updated EP curve against the reinsurance programme structure. The CRO has flagged that the 1-in-200 year net loss has crept above the internal capital limit. The retrocession team is renegotiating its own reinsurance cover. An ILS investor who participates in Vantage Re's cat bond programme has emailed a question about their attachment probability. A ratings analyst at a major agency has requested a call.
Same model run. Same numbers. Six conversations, six different questions, six different things each person needs to understand from the output.
A cat model run produces numbers. What those numbers mean, which ones matter, and what decisions they inform depends entirely on who is reading them and why. The same AAL figure is a pricing anchor to the underwriter, a loss ratio input to the actuary, a capital adequacy test to the CRO, and a basis for attachment probability to a cat bond investor. Understanding what each stakeholder needs from the output is not peripheral knowledge, it is what allows a cat modelling analyst to structure a run correctly, present results usefully, and flag the right caveats to the right audience.
This lesson maps the six primary stakeholders who use cat model output, what each one does with it, which metrics they focus on, and what question they are trying to answer. It does not cover how to construct those presentations or conversations, that is a professional skill developed with experience. It covers the foundational picture: who uses what, and why.
The underwriter uses cat model output to make the accept/decline decision and set the price. When Vantage Re's head of underwriting receives the quarterly update, she is checking whether the Gulf of Mexico treaty she wrote in Q2 is performing as expected against her pricing assumptions, whether the modelled AAL is close to what she used to set the premium, and whether the risk's contribution to Vantage Re's Gulf of Mexico accumulation position has changed as the book grew around it. For individual risk assessment, she wants the AAL and the 1-in-100 year OEP, the expected annual cost and the single worst-event estimate at the primary reference return period. The cat model output is the factual input; the underwriting judgement about whether to write the risk and at what price is hers.
The actuary uses the EP curve to validate that Vantage Re's reinsurance programme structure makes sense given the current portfolio. Vantage Re buys retrocession, its own reinsurance protection, to limit net losses above its retention. The actuary needs the full AEP curve at every return period to check that the programme's attachment point is at the right level and that the limit is large enough to cover losses up to the return period Vantage Re has agreed to protect against. She is not pricing individual risks, she is looking at the aggregate portfolio loss distribution and asking whether the reinsurance programme correctly bounds Vantage Re's net exposure. A shift in the EP curve tail between quarters signals a possible need to adjust the programme at the next renewal.
The CRO is concerned with one question above all others: does Vantage Re have enough capital to survive a 1-in-200 year event and remain solvent? The cat model's 1-in-200 year net loss, after all retrocession recoveries, is the primary input into Vantage Re's internal capital model and its solvency capital requirement. When the CRO flags that the 1-in-200 year net loss has crept above the internal capital limit, it means the modelled extreme tail loss now exceeds the capital held against it, a position that requires either additional capital, additional retrocession cover, or a reduction in risk-taking. This is the cat model's most direct connection to corporate governance and regulatory compliance: the number sets the minimum capital Vantage Re must hold.
The retrocession team uses cat model output to negotiate and structure Vantage Re's own reinsurance purchases. Vantage Re's retrocession programme sits above its retention and pays when a single event or an aggregate of events causes Vantage Re's gross loss to exceed a defined threshold. The team needs to know the probability that the gross loss will exceed the retro attachment in any given year, which reads directly from the gross EP curve, and how much cover above that attachment is needed to protect against the full range of plausible large-loss years. When the quarterly model update arrives, the retrocession team is comparing the current gross EP curve against the structure of the existing programme to assess whether coverage remains adequate or whether the growth of the book has pushed exposure above the programme's effective protection.
The ILS investor who participates in Vantage Re's cat bond programme uses model output as their sole analytical basis for pricing. Unlike the underwriter or actuary who bring professional judgement and market context to their reading of the numbers, the ILS investor typically has no independent framework for assessing catastrophe risk, they rely on the modelled expected loss percentage and the attachment probability disclosed in the bond documentation. When the investor emails about their attachment probability, they are asking: given the current portfolio and model, what is the annual probability that losses reach the level at which their note begins to lose principal? That figure comes directly from the gross EP curve at the attachment point. The investor's question is narrow and specific, they are not interested in the full distribution, only in the probability of losing their investment.
The ratings analyst uses cat model output to assess whether Vantage Re's capital base is adequate relative to its catastrophe exposure, and whether its risk management framework is credible. A downgrade in Vantage Re's financial strength rating would increase its cost of capital and make it harder to win business from cedants who require minimum rating thresholds. The analyst is looking at net PML figures as a percentage of available capital, the stress scenario losses against which Vantage Re holds reserves, and whether the retrocession programme actually limits net exposure to within the bounds Vantage Re claims. Cat model output is the evidential basis for those assessments, a reinsurer that cannot provide clear, well-documented cat model results is one the agency cannot rate with confidence.
Cat model output does not travel unchanged from the analyst's desk to the final decision-maker. At each step, the output is interpreted, contextualised, and often transformed before informing the next decision. Understanding this chain is important because it explains why accuracy and clear documentation matter far beyond the immediate recipient of your model run.
Consider the chain for Vantage Re's 1-in-200 year net loss figure. The cat modelling team produces the number, say, USD 480M, from a specific model run with specific settings, specific exposure data, and specific retrocession programme terms. That number reaches the CRO, who compares it to available capital of USD 520M and concludes there is USD 40M of headroom. She presents this to the board as "we hold 8% surplus capital above our 1-in-200 year stress scenario." The board approves the current business plan based on that headroom figure. The ratings agency, reviewing the board's disclosures, accepts that Vantage Re is adequately capitalised at the 1-in-200 year level.
Each step in that chain depends on the accuracy of the original model run. If the exposure data had a systematic geocoding error that understated the Japan earthquake exposure by 20%, the USD 480M figure might in reality be USD 576M, which would flip the "adequate capital" conclusion to a capital shortfall. The error originates in the exposure data, but its consequences reach the board, the regulator, and the ratings agency. This is not an argument for paralysis, it is an argument for understanding why data quality and model documentation matter beyond the immediate model run.
The cat modelling analyst sits at the beginning of all six of these decision chains. The underwriter cannot price without the model run. The actuary cannot structure the programme without the EP curve. The CRO cannot assess capital adequacy without the net loss figure. The retrocession team cannot negotiate without the gross EP curve. The ILS investor cannot price without the expected loss percentage. The ratings agency cannot assess capital adequacy without the stress scenario results.
This position carries a specific responsibility. The analyst is typically the only person in the chain who understands the assumptions behind the number, which model settings were used, what the data quality was, where the exposure is incomplete, and how sensitive the result is to those assumptions. By the time the 1-in-200 year net loss reaches the board, those caveats have often been stripped away. The analyst's job is not just to produce the number but to ensure the right caveats travel with it to the right audience.
Knowing where your output goes changes how you produce it. If you know the CRO will use the 1-in-200 year net loss for a capital adequacy presentation, you need to know whether the retrocession programme in the model exactly matches the current in-force programme, because any mismatch will silently bias the net figure. If you know the ILS investor is relying on the attachment probability, you need to flag any mid-year portfolio changes that could shift that probability. Producing a technically correct model run for the wrong version of the portfolio, or with misconfigured reinsurance terms, is not a minor error, it is an error that travels the full length of the decision chain before anyone catches it.
The underwriter is one of six primary audiences. The actuary, CRO, retrocession team, ILS investors, and ratings agencies each have equally legitimate and consequential uses for the output. Optimising a model run purely for the underwriting conversation while ignoring what the actuary needs for the programme review, or what the CRO needs for capital adequacy, leaves significant gaps in how the output serves the organisation.
The analyst is the only person who knows the assumptions behind the number. Those assumptions, data quality, model settings, exposure completeness, reinsurance configuration, determine whether the downstream decisions are well-founded. The responsibility to communicate what the number does and does not reflect does not end when the report is sent.
They do not. The ILS investor needs an expected loss percentage and an attachment probability, providing them with a full EP curve table and model uncertainty ranges adds noise rather than clarity. The actuary needs the full EP curve at multiple return periods, giving her only the AAL and 1-in-100 loss leaves her unable to structure the programme. Matching the level of detail to the decision being made is a core professional skill.
The model output is the result of many choices: which vendor, which settings, which exposure data, which policy conditions, which reinsurance configuration. Each of those choices introduces the analyst's judgement into the result. The number is not purely objective, it is a product of the data and the methodology chosen to process it. Presenting it as a simple fact without context conceals the choices that produced it.
The reinsurance that a reinsurer buys for itself, protection purchased by Vantage Re to limit its own net losses above its retention. The gross-to-net difference in the cat model reflects the modelled retrocession recoveries.
The annual probability that losses reach the level at which a reinsurance layer or cat bond begins to pay. Read directly from the EP curve at the attachment point. The primary metric ILS investors use to price cat bond risk.
The AAL expressed as a percentage of the notional principal of a cat bond. The primary output metric for ILS pricing, tells investors what fraction of their principal they should expect to lose on average each year if the bond is held to maturity.
The minimum capital a (re)insurer must hold under regulatory frameworks such as Solvency II. Cat model output at the 1-in-200 year net loss level is the primary input. Directly links cat model accuracy to regulatory capital adequacy.
An insurer or reinsurer that transfers (cedes) risk to a reinsurer. Vantage Re's clients are cedants, they pass a portion of their catastrophe exposure to Vantage Re in exchange for a reinsurance premium.
A rating agency's assessment of a (re)insurer's ability to meet its financial obligations. Cat model output, particularly net PML as a percentage of capital, is a key input to the ratings process. A downgrade increases cost of capital and can restrict business access.
Five questions · 4 of 5 correct (80%) to pass