How to choose a catastrophe modelling platform

This category holds two kinds of product that look similar in a directory listing and are not interchangeable. Free federal programs — the FEMA National Risk Index and the USGS Earthquake Hazards Program — publish relative risk scores and hazard maps at no cost, funded by public budgets. Commercial catastrophe modellers — Verisk Extreme Event Solutions, Karen Clark & Company, JBA Risk Management and Fathom — sell probabilistic loss models that price and reserve for insurance risk. A relative risk score and a probabilistic loss curve answer different questions, and picking the wrong one means redoing the work later with a different vendor.

Decide if you need relative risk or priced loss

If the job is benchmarking, planning or public communication — "how does this county's disaster risk compare to its neighbour's" — a free relative-risk score is enough and there is no reason to pay for more. The FEMA National Risk Index scores every US county and census tract 0-100 across 18 hazards, combining expected annual loss with social vulnerability and community resilience, and its methodology is fully public. If the job is pricing an insurance policy, setting a reinsurance reserve, or structuring an insurance-linked security, you need a probabilistic catastrophe model that produces an exceedance-probability loss curve, not a relative score — that is what Verisk Extreme Event Solutions, Karen Clark & Company, JBA Risk Management and Fathom are built for, and it is the one requirement a free federal tool cannot meet.

Single-peril specialists vs multi-peril generalists

Fathom and JBA Risk Management are both flood specialists with genuinely global coverage, built on high-resolution flood-hazard science rather than a broader catastrophe-modelling franchise. If flood is your dominant exposure — for a bank's mortgage book, an insurer's flood line, or an international-development lender — a specialist's depth on that one peril is usually worth more than a generalist's breadth. Verisk Extreme Event Solutions (formerly AIR Worldwide) and Karen Clark & Company both model hurricane, earthquake, severe convective storm and more across upwards of a hundred countries, aimed at insurers and reinsurers who need one vendor covering their entire multi-peril book rather than stitching together specialists.

Open model vs closed model

This is the distinction Karen Clark & Company built its business around. Most vendor catastrophe models are closed: you get the loss output, not the assumptions behind it. Karen Clark & Company's RiskInsight platform is explicitly open — it lets you plug in your own or third-party event sets and vulnerability curves rather than accepting the vendor's black box, which the firm's founder, who also founded AIR (now part of Verisk), positions as a direct response to that closed-model tradition. If your actuarial team wants to interrogate and adjust the assumptions behind a loss number, that openness is a real differentiator, not a marketing line. If your team wants a vetted, industry-standard model without the overhead of managing your own event sets, a closed model from an established franchise like Verisk is the lower-effort choice.

Infrastructure-dependent risk is a different question again

One Concern does not price catastrophe losses directly. It builds an AI model of infrastructure networks — power, water, transport — to estimate how long a property or region would be without services after an earthquake, flood, wildfire or storm, producing a resilience or downtime metric rather than a loss curve. That is the right tool if your question is business-interruption exposure or supply-chain resilience, not what a policy would pay out. It is the wrong tool if you actually need a priced loss for underwriting — for that, go to a catastrophe modeller instead.

Earthquake needs its own specialist

For US seismic hazard specifically, the USGS Earthquake Hazards Program publishes the National Seismic Hazard Model free, including near-real-time ShakeMap intensity data after an event and the design ground-motion values that feed building codes — authoritative, public data that most commercial catastrophe models for the US ultimately build on or benchmark against. There is rarely a reason to pay for basic US seismic hazard data when this exists and is free; pay for a commercial model when you need a probabilistic loss estimate specific to your exposure, not just a hazard map.

A shortlist by situation

• You need free, comparable relative risk across US counties for planning or benchmarking. FEMA National Risk Index.
• You need free, authoritative US earthquake hazard data. USGS Earthquake Hazards Program.
• Flood is your dominant global exposure and you need specialist depth. Fathom or JBA Risk Management.
• You are an insurer or reinsurer needing a vetted multi-peril model across a large international book. Verisk Extreme Event Solutions.
• Your actuarial team wants to plug in its own assumptions rather than accept a closed model. Karen Clark & Company.
• Your question is downtime and business interruption from infrastructure failure, not insured loss. One Concern.

Questions to ask a vendor

1. Does the output stop at a relative risk score, or does it produce a probabilistic, priceable loss curve?
2. Which perils and which countries are actually covered — not "global" in marketing copy, but the specific territories in your exposure?
3. Can you inspect or adjust the underlying event set and vulnerability curves, or is the model closed?
4. How is the model updated as climate conditions and historical loss experience change, and how often?
5. If your requirement shifts from underwriting to disclosure (or the reverse), does the same data licence cover both uses?

Common mistakes

Using a free relative risk score where a regulator or reinsurer expects a priced, probabilistic loss estimate — the two are not interchangeable even when both produce a number that looks similar on a map. Buying multi-peril global breadth when your actual exposure is concentrated in one peril a specialist covers better. And treating an infrastructure-resilience metric like One Concern's as a loss estimate, when it answers a related but different question about downtime rather than dollars.

For two direct match-ups, read Karen Clark & Company vs Verisk Extreme Event Solutions and Fathom vs JBA Risk Management. If your question is corporate or investment portfolio exposure rather than insurance pricing, see how to choose a climate risk analytics platform instead. See every tool in this category at every tool in this category.