Planet NGFS vs Planet VUCA: Why Standard Climate Scenarios Fail Short-Horizon Decisions

The Network for Greening the Financial System publishes what have become the de facto reference scenarios for climate risk work in banking and insurance. Every major regulator points to them. Most ORSA scenario analysis chapters I review use them — sometimes unmodified, sometimes with a token stress overlay. It is worth stating plainly what these scenarios describe.

NGFS scenarios describe a world without storms, floods, wildfires, droughts or meaningful sea level rise within the horizon. The economic damage from warming grows gradually, and for several decades, modestly. There are no pandemics. No wars. No supply chain shocks. No energy crises. No financial crashes. No unemployment spikes. Markets clear. Carbon prices rise smoothly and fossil capital reallocates in orderly fashion. Negative emissions technologies deploy at scale roughly when the models need them to.

This is the story built into the integrated assessment models underneath NGFS, the IEA net-zero scenarios and most equivalent public pathways. The last five years should have forced a reassessment.

1. What the Models Leave Out

The omissions are not accidental. Integrated assessment models optimise under smooth assumptions because that is what makes them tractable. The modelling choice becomes a communication problem when the scenarios are used in settings where the smoothness is the exact thing being challenged.

The obvious 2020-2026 counter-evidence: COVID-19 and its associated economic dislocation; the war in Ukraine and the energy price shock that followed; supply chain fragility exposed by container shortages and semiconductor bottlenecks; the fastest rate-hiking cycle in 40 years; and an accelerating pattern of climate-attributable extreme weather — from Western Cape drought and KZN floods to European heatwaves and North American wildfires. None of this exists inside the standard scenarios.

A useful distinction: NGFS scenarios are climate scenarios. The world we actually make decisions in is a VUCA world — volatile, uncertain, complex, ambiguous — where climate is one driver interacting with geopolitics, energy markets, monetary policy and extreme weather in ways the integrated assessment models simply do not represent.

2. Why This Matters for Short-Horizon Decisions

For a 30-year systemic risk assessment, NGFS scenarios are probably fine. The smoothness washes out over decades and the narrative direction (orderly, disorderly, hot house) does useful work.

The problem is that most of the operational decisions insurers actually make are not 30-year decisions. They are 1-5 year decisions: underwriting appetite on coastal property, reinsurance programme structure, investment mandate changes, loan-to-value covenants, product pricing, capital planning. Over 1-5 years, the dynamics that NGFS ignores — shocks, discontinuities, feedback loops — dominate. Using NGFS to inform a 3-year underwriting strategy is a category error.

flowchart LR
    A[Climate Question] --> B{Horizon?}
    B -->|30+ yr systemic| C[NGFS-style<br/>Integrated Models]
    B -->|10 yr sector| D[Transition<br/>Planning Scenarios]
    B -->|1-3 yr book| E[Book-specific<br/>VUCA scenarios]

    style A fill:#0A0F1E,stroke:#C9A84C,color:#F0F0F0
    style C fill:#0A0F1E,stroke:#00C2CB,color:#F0F0F0
    style D fill:#0A0F1E,stroke:#00C2CB,color:#F0F0F0
    style E fill:#0A0F1E,stroke:#00C2CB,color:#F0F0F0

3. A Tiered Scenario Stack

The pragmatic answer is not to discard NGFS. It is to sit NGFS on top of a tiered stack, where each tier is matched to the decision it supports.

Tier 1 — Systemic, 30-year horizon. NGFS-like scenarios, used for strategic risk appetite, group-level balance sheet stress, and regulatory disclosure under IFRS S2 resilience analysis. Smooth, narrative-driven, globally calibrated.

Tier 2 — Transition planning, 10-year horizon. Sector-specific transition scenarios tailored to the insurer’s underwriting and investment exposures. A motor insurer’s transition scenario looks very different from a life insurer’s — the first needs EV penetration curves and the economics of stranded ICE assets; the second needs longevity sensitivity to health effects of warming and workforce transition risk. These scenarios need to carry real transition economics, including policy reversal and technology deployment risk.

Tier 3 — Operational, 1-3 year horizon. Book-specific scenarios that combine physical triggers (1-in-50 drought on a crop book; named storm on coastal property; wildfire season on a fire book) with transition shocks (sudden carbon price change; sector-specific subsidy withdrawal; tariff escalation). These are closer to traditional cat modelling than to climate modelling, and deliberately so. They are the scenarios that drive underwriting appetite decisions and reinsurance purchase.

Most insurers have a version of Tier 1 in their ORSA because the regulator asks for it. Many have partial Tier 3 via existing cat models, though rarely labelled as climate. Almost none have a coherent Tier 2 — and Tier 2 is exactly where transition planning decisions live.

4. The Real World Climate Scenarios Initiative

The gap I’m describing is not a private observation. The Real World Climate Scenarios initiative, which has been growing across the UK, Europe and increasingly Africa, exists specifically to build scenarios that embrace the VUCA conditions the integrated assessment models exclude. The early outputs focus on short-horizon, bespoke scenarios for government and corporate use, driven by the interaction of transition risks and extreme weather rather than by smooth IAM outputs.

For actuaries, the initiative’s practical contribution is showing that credible short-horizon scenarios can be built without the machinery of integrated assessment models. The tools are stochastic cat modelling, macro shock overlays, geopolitical risk mapping and structured expert judgment — all tools actuaries already use.

5. Implications for ORSA and IFRS S2

For SAM or Solvency II ORSA, using NGFS alone is defensible today because the regulator has not yet pushed beyond it. That is changing. The Bank of England’s biennial exploratory scenarios have moved in the VUCA direction. The PRA’s expectations on short-term climate stress are tightening. The Prudential Authority in South Africa has been explicit that scenario analysis should be fit for the insurer’s horizon and risk profile, which quietly signals that unmodified NGFS is insufficient for many books.

For IFRS S2, the resilience disclosure requirement is agnostic about which scenarios are used, but it is explicit that the scenarios must support the entity’s actual risk management. An insurer using NGFS-only in S2 disclosure, while making underwriting and pricing decisions on completely different assumptions, creates an audit finding waiting to happen.

6. What to Actually Do Next

Three practical next steps for insurers re-examining their scenario stack:

• Audit current scenarios against the tier they are being asked to support. If Tier 1 scenarios are informing Tier 3 decisions, that is the gap.
• For Tier 2 transition planning, build one sector-specific scenario for the insurer’s largest exposure, with explicit policy, technology and demand components. Do not outsource this to the climate team; keep it with the underwriting or investment function that will use it.
• For Tier 3, connect existing cat modelling to short-horizon climate overlays. The data is often already in-house; it just is not labelled climate.

NGFS is not wrong. It is incomplete. Treating it as the answer to every climate question is what creates the unreality. Treating it as one tier in a stack that also includes the VUCA conditions insurers actually operate in is how scenario analysis becomes a decision tool rather than a disclosure exercise.