Foreword – Mark Campanale

The foreword argues that pension funds and long-horizon investors are misled by economic models that understate systemic climate risk. Quadratic damage functions ignore tipping points and non-linear escalation, while assumptions of uninterrupted 3% annual GDP growth obscure structural disruption. Under-pricing physical risk leaves diversified portfolios and taxpayers exposed.

Section 1: Mind the gap – The disconnect between climate science and economics

Economic models have historically underestimated damages, with some projecting losses of only 2% of GDP at 3°C. Damage functions in IAMs rely on global mean temperature and smooth relationships.

Recent revisions (Kotz et al., 2024) led the NGFS to increase estimated end-century damages at 3°C from 7–14% to around 30% of global GDP growth. After data corrections, mid-century 1.5°C damages were revised to 17% (range 6–31%).

Reporting damages as GDP-level differences obscures impacts. A 0.8 percentage point annual growth reduction compounds into large long-term losses. Damages affect capital, labour productivity, financial stability and investment, not only output.

Section 2: Scientific understanding – What’s wrong with damage functions?

Expert elicitation of 68 climate scientists found strong consensus that current damage functions underestimate risks.

2.1. Extremes – Not averages – Define the future

Mean temperature masks regional and extreme events. The 2021 Texas winter storm caused over US$195 billion in damages despite limited impact on global averages. Experts stress tail risks and collapse scenarios beyond 2–6°C, reframing risk around catastrophic probabilities rather than median outcomes.

2.2. Like progress, damage is not linear

Quadratic forms fit historical data but diverge under extrapolation. At 3–4°C, expert-informed functions produce damage ranges of 43–88%. Historical disaster data (R² < 0.1) cannot distinguish between functional forms. Structural uncertainty dominates at higher warming levels.

2.3. Damages Are Cascading and long-lasting

Climate shocks propagate through supply chains, credit markets and infrastructure. In the decade to 2025, disasters displaced 250 million people globally. Impacts include migration, conflict, capital destruction and persistent economic scarring.

2.4. The limits to growth

Assumptions of smooth adaptation are rejected. Once thresholds are crossed, systems may fail. Agricultural collapse, although around 4% of global GDP, could trigger disproportionate systemic effects.

2.5. Damage compounds across time, space, and sectors

Regional warming and cross-sector linkages undermine global averages. GDP-centric metrics hide ecosystem loss, health burdens and inequality. Estimates for similar warming levels vary from a few per cent to over 50%, depending on modelling approach.

2.6. Data and modelling issues

Damage estimates extrapolate beyond roughly 1°C of historical variation. Uncertainty at 3°C can vary fourteen-fold and is not reducible through additional data alone.

Section 3: Bridging the gap – Improving damage functions

3.1. How scientists would improve damage functions

Experts prioritise incorporating extremes, tipping points, collapse thresholds and multi-metric approaches beyond GDP. They support cross-disciplinary collaboration and reporting ranges rather than point estimates.

3.2. Direct improvements to damage functions

The report proposes temperature-stratified calibration at 1.5°C, 2°C, 3°C and 4°C, alongside collapse-threshold constraints. An Expert-Calibrated Damage Function (ECDF) integrates multiple functional forms probabilistically, capturing parameter and structural uncertainty. For NGFS scenarios, it recommends reporting damage ranges (e.g. 44–88% at 3°C) and testing robustness across ensembles.

3.3. Indirect improvements to climate modelling and policy

Complementary bottom-up approaches should model capital destruction, sectoral dynamics and extreme-event volatility. A three-track agenda proposes immediate IAM improvements, medium-term interdisciplinary expansion, and improved communication through systemic risk dashboards and scenario narratives.

Section 4: Recommendations

4.1. Key findings: A fundamental disconnect

Economic frameworks are structurally misaligned with scientific understanding. Climate damages are systemic, non-linear and deeply uncertain, exceeding assumptions embedded in current financial modelling.

4.2. Recommendations for researchers, financial regulators & investors

Regulators and central banks should integrate improved damage functions into NGFS scenarios, emphasise tail risks, report ranges not point estimates, and design stress tests for resilience under deep uncertainty. Investors should prioritise systemic exposure, limits to diversification, and mitigation pathways to reduce long-term portfolio risk.