In my last post, I showed how expected U.S. storm damages could reach $290 billion per year by 2050 without any additional warming. Today I want to zoom out: how does physical climate risk compare across the world’s major economies, and what happens when you layer warming on top?

What I find is that the climate-driven damage increase hitting Australia is roughly 10× larger than what the U.S. faces. Not because Australia is riskier in absolute terms but because Australia's risk profile is wildfire-dominated, and warming amplifies wildfire far more than it amplifies cyclones.

What Country Is Most Exposed?

The science on where extreme events occur is well established. Hazard maps built from decades of wind, flood, and wildfire data give us a clear picture of which parts of the planet get hit. But how do we translate the science to the economy?

We need to combine where the climate risk is with where the economy is exposed. I’ve been able to do this using Emmi’s global hazard maps product with global gridded GDP data, to measure where economic activity actually intersects with hazard exposure. The result is a set of country-level multipliers, all benchmarked to the USA (1.0×), across three dimensions:

• Magnitude: Expected annual damage where the money is

• Frequency: How often the hazard occurs where the money is

• Severity: How bad the 90th percentile looks where the money is

Wildfire

Australia’s GDP-weighted wildfire magnitude multiplier comes in at 4.85× - nearly five times the U.S. baseline.

Take South-Eastern Australia. The left panel below shows where the GDP is concentrated overwhelmingly in Melbourne and Canberra. The middle panel shows fire hazard (Expected Damage Factor) - highest in the forested ranges to the northeast. The right panel multiplies the two together. The economic exposure lights up exactly where those cities suburban fringe meets the fire-prone bush.

This pattern repeats across Australia’s major cities from Sydney to Perth where they are all coastal cities ringed by high fire-hazard bushland. The interface where suburbs push into eucalyptus forest carries elevated expected damage. GDP weighting captures this because the economic activity is right there at the fire front.

The same dynamic plays out in the U.S. The Southern California map below shows the identical three-panel breakdown - and the January 2025 fires that destroyed thousands of homes and caused an estimated $100+ billion in damages landed exactly where our economic exposure map predicted: the urban-wildland fringe around Los Angeles. The red stars mark the Palisades, Eaton, and Kenneth fires. But note that San Diego’s suburban edge actually shows up as an even larger economic exposure hotspot - a risk that hasn’t made headlines yet.

Globally, Nigeria has the highest relative economic exposure, while Canada is the surprise on the low end: just 0.13×. Despite enormous boreal fire zones, Canadian GDP is overwhelmingly concentrated in southern cities far from the major fire corridors. The land burns, but not where the money is. Russia shows a similar pattern at 0.24×.

Cyclones

By combining the hazard and GDP maps for cyclones, the Philippines economic exposure is 28x the US and it isn’t just about cyclone frequency. Manila and all major population centres sit directly in the typhoon corridor. There’s virtually nowhere in the Philippine economy that escapes cyclone exposure.

Japan reveals something important about fat tails. Its magnitude multiplier is a modest 1.60×, but its severity multiplier is 6.52×, the highest in the dataset. Cyclones making direct hits on the Tokyo-Osaka industrial corridor produce outsized damage. The average understates the risk; the tail tells the real story.

Now Layer on Climate Change

So how much does warming amplify the baseline risk for each country? The answer depends almost entirely on one thing: what type of hazard dominates a country’s risk profile.

Under the NGFS Current Policies scenario - which represents 3°C of warming the relative amplification pattern is clear. The top countries (Italy, Australia, Brazil) are all all wildfire-exposure economies.

Climate warming drives longer fire seasons, drier fuels, and expanded burn zones. These effects compound directly on top of the baseline exposure. Italy and Singapore sit at the top of the amplification ranking, though it’s worth noting both start from a relatively low absolute damage base compared to countries like the U.S. or Philippines. The amplification factor is large, but on a smaller starting baseline number.

Now look at the bottom of the table. The Philippines, Jamaica, and Japan are all cyclone-dominated. Their amplification factors are below 1.0×, meaning warming actually reduces their projected total damage slightly. This is consistent with the cyclone science: warming is projected to reduce overall tropical cyclone frequency (Knutson et al. 2020), and while individual storms get stronger, fewer storms partially or fully offset the intensity gains.

The U.S. at just 1.06× sits in between. Its damage mix is cyclone-heavy, so the projected decline in storm frequency partially offsets growing wildfire and flood risk. Compare the warming-driven increase directly: Australia gets +58%, the U.S. gets +6% which is roughly 10× larger than what the U.S. faces. Since Australia’s risk is dominated by wildfires, it has a much sharper increase in warming-driven risk than the cyclone-dominated US.

Important though: in absolute terms, the U.S. still faces far greater total climate risk than Australia. As I showed in my last post, U.S. storm damages are on track to reach a no-warming P99 of roughly $1 trillion by 2045, driven by economic exposure growth alone. The U.S. baseline is so enormous that even a modest 6% climate amplification translates into tens of billions of additional annual damage. Australia’s 58% warming hit is 10× larger as a proportion, but it’s applied to a much smaller economic base. The 10× gap is the warming signal, not the total risk picture.

Climate risk amplification entirely depends on the dominant type of hazard within a country or region. Under current policies (3°C), wildfire-exposed economies see damages rise 50–80%. Cyclone-dominated ones barely move or even decline.

What Investors Should Take Away

Economics drive the risk. Australia’s wildfire exposure is vastly more than Canada’s despite both having vast fire zones. The difference is where the GDP sits. Price physical risk without accounting for this and you’re mispricing.

Warming amplification is hazard-dependent. Under current policies, Australia’s damage rises 58%, Italy’s nearly 80%, but the U.S. barely moves (+6%) and the Philippines actually falls (−7%).

Severity and magnitude tell different stories. Japan’s cyclone magnitude is a modest 1.6×, but its severity (P90) is 6.5×. If you’re pricing catastrophe bonds or reinsurance, the tail matters more - and warming widens the gap.

The hazard mix drives portfolio risk under warming. Heavy Australian and Southern European exposure compounds under warming. Cyclone-exposed markets like the Philippines face high baseline risk but a stable warming trajectory. The hedge isn’t low-risk countries, it’s knowing which hazard dominates.

The U.S. is deceptively climate stable. The U.S. looks well-positioned, but it’s baseline risk is by far the largest in the world. Its cyclone-heavy mix partially self-hedges as fewer projected cyclones offseting wildfire growth. However this picture changes alot when you drill into region - which I will tackle in a future post.

Until next time,

Ben

Data and Methods

Country risk multipliers calculated using GDP-weighted spatial aggregation of gridded hazard Expected Damage Factor (EDF) maps at 0.1° resolution (via Emmi). GDP weighting uses Kummu et al. (2025) gridded global GDP at 5 arc-minute resolution. All multipliers relative to USA (1.0×). Warming scenario damages derived from 11 NGFS and IPCC pathways spanning 0.8°C to 4.0°C. Coverage: 34 countries across wildfire and tropical cyclone hazards, with flood included in scenario projections. Damages reported as annualized expected values in billions USD.

References

Kummu, M., Taka, M., & Guillaume, J. H. A. (2025). Gridded global datasets for Gross Domestic Product and Human Development Index over 1990–2022. Scientific Data, 12, 61.

Knutson, T.R., et al. (2020). Tropical Cyclones and Climate Change Assessment: Part II. Bull. Amer. Meteor. Soc., 101(3), E303–E322.

Weinkle, J., et al. (2018). Normalized hurricane damage in the continental United States 1900–2017. Nature Sustainability, 1, 808–813.

U.S. politics is rewriting the rules for institutional investors. State pension funds across much of the country are now barred from considering ESG or climate factors, and Congress has recently passed legislation reinforcing the same principle. But irrespective of your views on climate, storm damages in the U.S. are increasing. So it’s important to ask a simple question: what does the future look like if you strip out climate change entirely?

In my last post, I walked through my market-based approach for quantifying how storms hit historical stock returns. It shows that certain weather-sensitive industries likely experience measurable stock price drag in the months following major storm events.

Today I want to tackle the next question for those in the US: how much storm damage should we expect over the coming decades without any additional warming? It’s the no-warming scenario, and the numbers are worth paying attention to.

Thanks for reading Carbon Alpha! Share it to those you think will find it useful.

Share

The US Damage Trend Isn’t About Climate Change

Analysing historical billion-dollar disasters in the US1 (inflation-adjusted to 2024 dollars) the upward trend is clear. Average annual losses ran about $82 billion per year in the 2000s. Over the last five years? Roughly $197 billion per year. A 2.4× increase.

But when you normalize those losses for GDP and population growth, the upward trend disappears. Weinkle et al. (2018) found no trend in hurricane losses normalised for population and wealth from 1900 to 2017. Zhang et al. (2023) found the same for tornadoes. The IPCC Sixth Assessment Report expresses only “low confidence” in attribution of storm trends.

The damage growth is coming from people, wealth, and assets piling up in the places storms like to hit. Simple trend analysis shows damages increasing at approximately $5 billion per year in real terms, projecting to a mean of $290 billion per year by 2050, about triple the historical average.

But Isn’t There a Climate Signal?

There almost certainly is a climate signal in the historical data. But the evidence suggests it’s a fraction of the cause, and one important reason comes down to which hazards dominate the losses.

Cyclones dominate. Tropical cyclones account for 42% of all damages since 2000. Add flooding and you’re at 64%. These categories drive the damage curve and its fat tail.

The science here challenges simple warming narratives. NOAA’s GFDL projects Atlantic tropical cyclone frequency will likely decrease under warming, by roughly 15% for 2°C (Knutson et al. 2020). Storms that do form may be slightly stronger (+4% wind speeds, +10 to 15% rainfall), but these gains are partially or fully offset by fewer storms. As Knutson et al. (2019) concluded: “There is no conclusive evidence that any observed changes in tropical cyclone genesis, tracks, duration and surge flooding exceed the variability expected from natural causes.”

Wildfires are the exception with the clearest warming signal, but they account for roughly 11% of total losses. Even substantial climate-driven wildfire increases get outweighed by what’s happening with cyclones and flooding, where attribution is far weaker.

The hazard with the clearest warming signal contributes a tenth of losses. The hazard that dominates losses shows no detectable frequency trend.

The damage data is primarily telling you a story about where Americans are building and the increasing asset values, not how the climate is changing.

Fat Tails and the Trillion-Dollar No-Warming Scenario

Storm damages don’t follow a normal distribution. The worst year on record (2017: cyclone Harvey, Irma, Maria) reached $410 billion, more than four times the median. Using fixed historical percentile ratios the P99, a 1-in-100 year scenario, crosses the trillion-dollar threshold by 2045 without any climate change attribution. That’s 2.8× the worst historical year, driven entirely by continued exposure growth.

What Investors Should Take Away

First, the no-warming economic damage trend is rising. Even stripping out any climate signal, U.S. storm damages are projected to roughly triple by mid-century on historical trends. This is a demographic and real estate story, not a climate story, but it hits the same balance sheets and insurance markets.

Second, industries exposed to storms face material financial drag without any warming. As I showed in my last post, sectors like electric utilities, insurance, and building materials experience measurable stock price underperformance following major storm events. As the baseline damage trend rises, so does the frequency and severity of those financial impacts across exposed portfolios, irrespective of any warming.

Third, climate amplification sits on top of the no-warming scenario. The projections above represent the floor. In the next post, I’ll show what happens when you layer warming scenarios onto this baseline. A preview: warming doesn’t change the mean as much as you’d expect in the US, but it hits the tails hard. However warming dramatically amplifies the damages across many parts of the world.

Until next time,

Ben

References

Knutson, T.R., et al. (2019). Tropical Cyclones and Climate Change Assessment: Part I. Bull. Amer. Meteor. Soc., 100(10), 1987–2007.

Knutson, T.R., et al. (2020). Tropical Cyclones and Climate Change Assessment: Part II. Bull. Amer. Meteor. Soc., 101(3), E303–E322.

Pielke, R. Jr. (2024). Scientific integrity and U.S. “Billion Dollar Disasters.” npj Natural Hazards, 1, 12.

Weinkle, J., et al. (2018). Normalized hurricane damage in the continental United States 1900–2017. Nature Sustainability, 1, 808–813.

Zhang, J., et al. (2023). Time trends in losses from major tornadoes in the United States. Weather and Climate Extremes, 41, 100579.

I hope you are safe and well.

For the last few months I've been focused on investigating methods to measure physical climate risk for institutional investors, and I want to share an uncomfortable truth:

Institutional investors are paying premium prices for physical climate risk analysis they fundamentally don't need.

In this newsletter I go deeper and share why a fundamental change is needed to better focus on investors, not insurers.

Share

The Current Approach for Physical Climate Risk

Acute physical climate risk assessment in finance today is dominated by methods borrowed directly from the insurance industry. The mathematical framework remains consistent across all major providers: expected cost equals vulnerability multiplied by hazard intensity and exposure value.

This approach works by establishing hazard damage functions that relate physical conditions at specific geo-locations to potential financial losses. Providers assess climate hazards (wildfires, flooding, tropical cyclones) for precise coordinates, then build complex vulnerability models based on sector-specific factors like construction type, building age, and local codes to calculate "mean damage ratios."

The process requires extensive scientific work: downscaling global climate models, statistical corrections, regional modeling, local topographic adjustments, and facility-specific vulnerability assessments. This wasn't designed for investment decisions, it was built for property insurers deciding whether to write a policy on a specific building.

Why This Creates Fundamental Problems for Investors

The insurance-focused approach creates systematic blind spots that affect everyone from pension funds to sovereign wealth funds:

Missing Supply Chain Dependencies: Traditional models focus on owned assets but ignore critical supply chain vulnerabilities. Companies don't operate in isolation—their risks come from suppliers, logistics networks, and customer access points they don't own.

Ignoring Geographic Diversification: Most listed companies are diversified by geography and revenue, creating financial resilience at the company level that geo-location models completely miss. A catastrophic event destroying one facility rarely translates to material company-wide impact.

Overlooking Existing Risk Management: Companies already have insurance, business continuity plans, and operational flexibility. Investors don't need to model replacement costs, they need to understand residual market impacts after companies deploy their risk management strategies.

Requiring Non-Existent Precision: The models demand precise geo-coordinates for every asset and perfect downscaling of climate projections. This level of granularity often doesn't exist in practice and introduces layers of uncertainty without improving investment insight.

Real-World Examples Show the Disconnect

Case 1: Hurricane Harvey's Supply Chain Cascade (2017)

When Hurricane Harvey flooded Houston, it knocked out about one-third of US petrochemical production and 18 oil refineries that compromised 20% of national refining capacity. The disruption cascaded across industries nationwide, affecting companies with no facilities anywhere near the flood zone.

Ford and General Motors had to idle manufacturing plants across multiple states because they couldn't get the plastics and chemicals needed for car production. Newell Brands (maker of Rubbermaid products) lowered their 2017 earnings guidance due to supply chain disruptions, forced to source resins from more expensive alternative suppliers. Electronics manufacturers, construction companies, food packaging firms, and pharmaceutical companies all faced similar disruptions, not because their facilities were damaged, but because the Gulf Coast produces 90% of America's base plastics and key chemical intermediates that flow into virtually every manufacturing supply chain.

Traditional geo-location models would have shown minimal direct risk to Ford's Michigan plants or Newell's facilities in Ohio or tech manufacturers in California. Yet these companies experienced some of their largest climate-related losses not from property damage, but from supply chain disruptions that lasted weeks longer than the storm itself.

The financial impact rippled far beyond the geographic footprint that current risk models assess, affecting companies whose facilities were thousands of miles from the actual climate event.

This is exactly the systematic, enterprise-wide effects that investors need to understand but traditional tools systematically miss.

Case 2: California Wildfires and Kroger's Palisades Supermarket

In January 2025, the catastrophic wildfires in California completely destroyed one of Kroger's supermarket in Palisades, California, a $20+ million facility loss that traditional geo-location models would have likely flagged as high-risk.

The market's response revealed the disconnect: Kroger's stock initially declined 5% but rebounded within three weeks to significantly outperform the wider S&P 500. One destroyed store out of 2,800+ locations represented a rounding error on their financial position.

Even if traditional geo-location risk models correctly predicted 100% facility destruction, it completely missed the investment reality:

Geographic diversification makes individual asset losses (even catastrophic events) financially immaterial for most listed companies.

If Kroger faces material climate risk, it's from systematic impacts like widespread drought affecting food production across large regions, not individual facility losses that models obsess over.

The Premium Price for Immaterial Analysis

Here's the uncomfortable truth: institutional investors are paying premium prices for insurance-focused climate analytics that don't address their core investment needs.

Despite using similar geo-location approaches, research from State Street1 found no correlation in their risk outputs for the same assets amongst ISS, S&P Truecost and MSCI. Yet investors are paying premium prices for this granular analysis built on layers of assumptions.

Meanwhile a pension fund doesn't need the mean damage ratio for a specific Toyota plant located in Thailand. They need to understand Toyota's enterprise-wide resilience to climate-induced supply chain disruptions or demand shifts—insights that current geo-location tools systematically miss.

The Path Forward: Investment-Native Physical Risk

The solution isn't to build better insurance models, it's to recognize that investors need fundamentally different climate risk analytics.

Instead of paying premium prices for granular geo-location analysis that produces immaterial insights, physical climate risk for investors should focus on:

• Market-level impacts across supply chains and consumer demand

• Enterprise-wide effects that persist weeks and months after events

• Systematic risks that affect multiple portfolio holdings simultaneously

This requires a more streamlined analysis compared to the current complexity investors are paying for, but it returns actual investment value rather than insurance precision they don't need. That will fundamentally better support their decision-making.

What do you think?

Do you agree? Are physical climate risk tools misaligned with investment needs in your organization?

Interested to hear your thoughts below.

Thanks for reading.

Ben

The climate finance community erupted in debate this week over a talk by Stuart Kirk, the Head of Responsible Investment at HSBC Asset Management titled ‘Why investors need not worry about climate risk’. The editorial of the Wall St Journal called it “A Financier Tells Some Climate-Change Truths”, so lets dig a little deeper into the arguments.

In the speech, Kirk presents a powerful contrarian view on climate risk in finance. Overall, his core arguments are that there is too much hyperbole on climate, it will only impact markets in the long-term and that we as humans will adapt like everything else. His conclusion: we need not worry about climate risk in finance.

Is this true? And as some asset managers are whispering, is he ‘saying the quiet bit out loud?’ Let’s shed some light on this much talked about speech.

Share

What to agree with on climate risk?

Although Kirk’s overall thesis - that climate risk is nothing to worry about in finance - is wrong (I will get to that), several of the arguments he presented are true:

1. The world will not end because of climate change. Yes, it will impact every living thing on a multi-generational perspective, but it will not cause the end of the earth or humanity.

2. The financial risk of physical climate change tends to be longer-term. While temperatures are rising and human health impacts are already being felt, it won’t impact markets much in the short-term.

3. Humans are amazing at adapting to challenges. You can’t really argue with this one, we are. But, there’s a huge social inequity in the resources needed and as such, the poor will get hit much harder.

4. Humans are innovators and markets will continue to rise. There are big opportunities in the transition.

So where did Kirk’s thesis go wrong?

Well, there are two enormous factors at play that he skimmed over which dismantle the argument that climate risk isn’t important in finance.

But first, some of you might be wondering why a climate scientist like me can agree with him in broadly discounting physical climate risk for finance, at least in the shorter-term.

Glaciers melt too slow for short-term markets

The world is warming, glaciers are melting, sea-level is rising and extreme weather events are likely increasing. This is causing human and social impacts across the world, today. However, the rates of warming, sea-level or extreme events aren’t likely to cause huge systemic changes to markets on a time-frame meaningful for investors. Let me explain.

The carbon we emit today will stay in the atmosphere for hundreds of years, causing multi-generational climate impact. However, there is a large asymmetry in the time-frame of impact on climate versus the time-frame relevant for markets.

Abrupt and irreversible changes to the climate system are known as ‘tipping points.’ Once they happen, there’s no going back. Examples of these include the collapse of the West-Antarctic Ice Sheet, the stagnation of ocean circulation patterns or a permanent shift to global monsoons. These climate tipping points are the ‘Green Swans’ for the financial system. While they grow in likelihood with any degree of warming, they only start becoming a higher risk once we hit 2°C. That level of warming isn’t likely until ~2050 - a time frame not meaningful for the vast majority of investors right now.

So while physical climate impacts will continue to get worse as we approach 2050, they’re not likely to be on the scale to impact financial markets (although there is still a risk). Take sea-level as an example. Ice-sheets and glaciers are melting, but in the short-term (say next 5-10 years), the magnitude is unlikely to cause a major step change in damage to infrastructure globally. Local impacts will occur but from a global perspective, the sea-level risk will be felt much more strongly in the latter half of this century, which give financiers time to invest in adaptation.

The physical impacts of climate change are obviously already happening, and will continue to worsen for centuries. But their magnitude and scale is not systemic enough across the entire economy or on a time-scale that is relevant for short-term financial markets. It’s the classic case of short-termism.

Our response to climate is where the financial risk is

So if physical climate impacts tend to be longer-term, what’s the problem with Kirk’s thesis? Well, he neglected the biggest way markets change - us.

How we as a society respond to climate change is orders of magnitude quicker and more disruptive to markets than the physical impact of climate. I write about this in more detail here, but carbon transition risk is very different to physical climate risk - something that Kirk’s argument missed almost entirely.

COVID shows us governments can act quickly

Governments and authorities respond to challenges that protect the interests of their citizens, whether it’s inflation, COVID or climate change. And these policy changes can be imposed suddenly, causing massive financial readjustments on a timescale that impact markets.

Take COVID-19 as an example. During the early days of the outbreak, only a fraction of the global population were impacted by the virus. Yet virtually every person and business on earth was impacted by how their governments responded to it to protect their citizens against the virus.

Markets didn’t crash because of the mortality rates. They crashed because the world’s government imposed lockdowns and shut businesses to protect its people from the virus. Empty streets and closed shops were in response to the virus. There was an enormous asymmetry between the risk of the virus and the response.

It’s the exact same for climate change - it's not the physical or even direct human impact, but the societal carbon transition to stop climate change that will cause immense market corrections in the shorter-term.

Kirk did mention it in passing, stating ‘the one thing the market could get wrong is a whopping big carbon tax out of the blue, $200 - wham’. Even if it comes gradually over 5 or 10 years, that will cause a huge shift to markets.

Glaciers melt slowly, but humans act quickly. For climate, the response will cause huge changes to markets given the magnitude of change required to limit warming to 1.5-2°C. There’s a reason that Larry Fink, CEO of BlackRock says ‘we are in the midst of a fundamental reshaping of finance’.

Beyond Meat and Tesla show us consumers are shifting

It’s not just governments moving against carbon, it’s consumers. When new products come that do less damage to the environment, markets shift. Solar power is the obvious example, but underlying consumer climate sentiment has undoubtedly allowed Tesla to become a trillion dollar car company leading the transport electrification transition.

Similar shifts are occurring across industries, like those in plant-based meat. Lower carbon is creating and growing entirely new product lines and replacing the old, carbon-intensive traditional products.

Carbon is a risk for consumers. This shift will accelerate over the next 5-10 years as consumers demand lower carbon reliance for products. And of course, financial markets will follow the money. You can see it now how companies are investing hugely into low carbon products because the future demand is in these products.

Climate is a big risk for markets because of carbon

It’s refreshing to see arguments and talks that don’t just follow the consensus. In the end however, it doesn't really matter what one person thinks - the government response and consumer sentiment has already shifted towards transitioning away from carbon. This will only accelerate over the next 5-10 years and on a timeframe very important for markets.

So if you’re an investor, carbon is the real short-term risk and you need to understand that across your investments.

Keen to hear your thoughts.

Ben