Dealing with climate risks: the tech stack

Published in

Future Positive

16 min readApr 20, 2023

1. Introduction — why climate risks matter

The average global temperature has risen by 1.1°C since pre-industrial times, causing a severe increase in the frequency and severity of acute hazards (heat waves, floods) and chronic hazards (e.g. drought, rising sea levels). Each year brings a new record of heavy storms, extensive flooding, severe droughts, widespread wildfires, and extreme heat waves.

The impact of these climate-related events is universal, pervasive and expensive. It affects everyone, everywhere, in every industry — though individuals living in developing countries will be hit first and hardest.

Climate-related hazards and disasters have affected a staggering 4 billion people since 2000, resulting in the loss of 1 million lives. Furthermore, each year, the world experiences economic losses equivalent to the GDP of Denmark due to such events.

Our cities, workplaces, and homes are resolutely unprepared to face the effects of our climate crisis.

The urgency associated with adaptation and mitigation will push nearly every industry towards companies that provide relevant solutions. As these emerging opportunities materialize, companies that monitor, forecast, or provide protection against the effects of unexpected seasonal and catastrophic climate disasters will benefit from this global trajectory and a fast-growing addressable market.

Building on our 2019 investment in Cervest, we took an updated look at potential complements at the intersection of data, finance, and climate adaptation.

2. Grand Challenge: climate resilience — and coping with climate-related losses

Climate-related losses

⚠️ The impact of climate-related events is universal, expensive and growing. They affect everyone, everywhere, in every industry.

Since 2000, the UN estimates that 1.2 million people have died and 4.2 billion have been affected by droughts, floods, and wildfires.
Direct economic losses and physical damage amounted to c. 300bn$ in 2021 (cf. below), equivalent to the GDP of Hong Kong or Denmark lost every year.
The economic losses in 2021 were 15% higher than the median of the last decade, 2011–2020 (Aon), while climate-related disasters have tripled in the previous 30 years (Oxfam).
A 2°C warming scenario could lead to 10% GDP losses due to climate events (Swiss Re).

Top 10 climate hazards in 2022. Source: Aon, 2023

A pressing issue across the globe

Distribution of events and losses. Source: Aon, 2023

Economic losses are spread across the globe: the 421 climate events of 2022 were spread across the world, with associated economic losses, particularly heavy in the US.

Notable climate hazards in 2022. Source: Aon, 2023

What are the risks, and who is impacted?

(ranked from lower magnitude/higher frequency to higher volume/lower frequency)

NB: those risks are often interrelated (”cascading effects”), which makes them hard to predict

Winter hazards

Although climate change would in general reduce the duration and temperature drops of winters because the warmer atmosphere holds more moisture, blizzards are more likely to occur and be more severe in places where temperatures are still cold enough for snow.
Cold weather and snow can lead to business interruption and traffic disruption, individuals' sickness and accidents, as well as power outages and frozen water pipes.

Drought and heatwaves

Higher temperatures induced by climate change lead to droughts that have a significant impact on crop shortfalls of agriculture and plant diseases. Forestry, shipping and energy, and tourism are also the most affected. Other consequences are damage to buildings on soils prone to subsidence (ex: clay) and wildfires.
Wildfires have been a growing concern recently: the largest wildfires recorded in California since 1930 have predominantly occurred since 2000, and global losses in 2019–2021 amounted to US$ 25bn (Munich Re).
Populations are affected mainly by heatwaves short term, leading to fatalities as blood pressure drops, putting pressure on the cardiovascular system.

Floods and storms

Scientific studies have shown that warmer air can absorb more moisture, which increases the potential for heavy rainfall as well as storms.
Cyclones are the most significant loss generators. Rare extreme events like Hurricane Ida can generate significant human and economic casualties (>75bn$)
The flooding in Central Europe in July 2021 was the costliest natural catastrophe in modern European history (54 bn$) and the costliest flood event globally of all time.

Earthquakes and tsunamis

The most significant threats to life can cause enormous financial losses: Fukushima alone generated 210 bn$ losses, the most significant climate event catastrophe ever.

3. The flaws with insurance coverage for climate-associated risks

Uninsured losses

⛔ More than 2/3 of losses each year remain uninsured (Swiss Re). Agriculture is a good example of this protection gap: >75% of global agricultural production is not insured, even though weather risk is associated with almost 60% of yield variability.

Why?

Risks are hard to assess, especially given cascading risks — this induces a vicious circle where increased losses due to incorrect risk assessment drive insurers to reduce the supply of insurance and raise prices.

In 2019, insurers dropped 230,000+ policies in California in regions exposed to wildfires. The state extended temporary protection to 2.1 million Californians, but long-term solutions to stabilize the market still need to be discovered.

2. Exclusions — exclusions from insurance policies (e.g. on people who don’t have qualifying assets such as concrete buildings or other consequential losses like loss of income) impact people who may have severe losses that don’t qualify and are therefore excluded, even though these vents may have a more significant impact on livelihoods and economies.

Emerging markets suffer in particular from these limitations and a less developed existing insurance ecosystem. For example, only 5% of flood losses are insured in these markets (Swiss Re)

3. Lengthy underwriting and claim process — the purchase process is laborious and slow, and claims can take months when consumers need payouts immediately. Human experts are involved in loss assessment, which leads to delays, disputes, and occasional instances of fraud and an inability to scale across geographies.

Example with Hurricane Irma (Category 4 storm in 2017): thousands of new claims filed were still being processed three years after the hurricane, with a high incidence of suspected fraudulent claims.

4. Infrastructure is publicly-owned — many losses involve public infrastructure — such as roads, railways, dykes, and bridges — which is usually uninsured.

The insurance value chain

A very integrated value chain: the insurance industry comprises different intermediaries between the customer who need coverage and the ultimate providers of risk capital that work together.

A new entrant will typically fit somewhere between the customer and the ultimate capital provider; it can go anywhere from:

Providing the analytics
Acting as “managing general agent,” i.e. underwriting policies on behalf of an insurer
Going left of the value chain and replacing agents (i.e., customer-facing) or right and replacing insurers (i.e. direct relationship with capital providers)

🏦 The different risk capital providers (on the right)
Reinsurers: have financial strength, knowledge, partnership approach, and client relationships. Examples: Axa, SwissRe, MunichRe, Hannover Re, Renaissance Re, GP Global parametric reinsurers.
Reinsurance platforms: underwriting syndicates like Lloyds (£30bn+ market for risk capital) organize a marketplace of vetted investors.
Capital markets: insurance-linked securities are listed products whose values are driven by insurance loss events. They’re attractive to traditional capital market investors as their return is uncorrelated with other asset classes.

This intricate value chain leaves little incentive to digitize individual participants’ activities unless the entire value chain is digitized.
The current insurance system works well for historical events across large markets but is non-performing for new kinds of events (because models are not accurately forecasting risks) or smaller markets (because the current costs base is too high for the long tail): this creates opportunities for new trends and micro-segments
The market is still predominantly paper-based, sales reporting is on a 30-day basis, and therefore portfolio exposure and treasury functions are not capital efficient.
Insurers need to factor in climate change in their models and are open to outsourcing modeling to third parties, especially for catastrophes. This is particularly hard and valuable to assess risk properly: companies that provide unique proprietary data and modeling are bringing a lot of value here.

4. Opportunities at the intersection of data science & Insurance

Why now?

Besides the rising problem of increasing, impactful, and interdependent weather events, parametric insurance is enabled by the recent developments in the following fields:

🌡️ IoT and Sensors: to gather real-time data as a trigger to payments
🛰️ Satellite: to monitor risks occurrence in real-time
🧮 AI to better model and forecast risks (this could also be applied to traditional insurance)
⛓️ Blockchain: to build smart contracts that automatically trigger the claims payments.

Now, let’s dig into the dynamics and players at each step of the value chain

1. Get and produce the correct data 💽

Value drivers and sources of USP: Access to massive amounts of data to provide a general view of risks of natural hazards over assets from:

Public and private data sources, such as NOAA weather data and satellites (NASA’s MODIS, Copernicus): recent advancements in space-based technology make access to climate-relevant data more widely available.
Proprietary sensors, for example, satellite constellations (Iceye) or proprietary hyperlocal sensors (Climavision, Dot by Understory, HD Rain, Stormsensor)
Crowd-watching data (ex: in agriculture) and decentralized climate data network based on blockchain to increase the transparency of collected data (dclimate)
Beyond data availability, there is enormous value in breaking down the siloed data to understand the interlinked complexity of climate signals and their underlying environmental and biological processes.
Data enrichment and modeling of climate by using of AI / stochastic or physics-based models, i.e. modeling physics laws of nature to forecast the future coupled with access to high-performance computing

Business models: SaaS for data and/or bundled with modeling (see below) and insurance product

Examples: Cervest, Jupiter, Salient Predictions, Climavision, Climate X, Risilience, Corelogic, HD Rain, Understory, dclimate, Iceye, Jua AI, Tomorrow.io.

2. Model and quote risks 📏

Value drivers and sources of USP: Model design and data. Given the dependency on data to run these approaches, usually strongly linked to data-rich locations (previous category).

Business models: build once, sell many times SaaS platform access

Estimating both hazards (likelihood and severity of perils) and actual risk (exposure and losses estimations). Price per asset or km2.
Uses cases are financial disclosures, risk management, suppliers analysis, and investment planning.
Clients are Property & Casualty (P&C) insurance providers, brokers, and capital markets, credit card issuers, or health care providers.

Companies: Verisk, RMS (Moody’s), Cape Analytics, Plover parametrics, Futureproof, Mitiga Solutions, Tractable, Eigenrisk, Futureproof.

3. Leverage parametric insurance models 🧮

📐 What is parametric insurance?
Parametric insurance (aka index-based insurance) is a non-traditional insurance product that offers pre-specified payouts based upon a trigger event, unlike “traditional” indemnity insurance that offers a payout which is the actual loss incurred (with possibly a deductible) as assessed by an expert.

Value drivers and sources of USP: Parametric insurance brings a new way of covering risks:

Reduces underwriting costs — through more efficient administration and no claims adjustments
Releases capital from loss reserves — (as uncertainty is reduced), thereby increasing capital investment returns:

More accurately determining risk: unlike traditional insurance, the loss amount is pre-determined and not dependent on actual losses incurred assessed by an expert
Live risk exposure monitoring (vs. 30–90 days basis today)

3. Reduces cost base — automated digital platforms allow reaching long tail/B2C clients that are unaddressable with traditional insurance (from a few tens of $ / month)

4. Address core issues of traditional insurance

Adverse selection — ex: in agriculture, farmers who are more likely to suffer losses are the only ones who buy insurance
Moral hazard — ex: farmers cut back on effort or compromise yields to receive the insurance payment

→ Parametric insurance overcomes both issues because the index is based on factors that no one can influence.

4. Provide Cover 🪖

Value drivers and sources of USP: Good access to capital allocators and risk aggregators (brokers) to scale distribution. Vertical integration / solid relationship with data and risk modeling.

Metrics: Gross Written Premium and actual revenues (i.e. deducting what is paid to reinsurers), Retention Rate (to test for recurring revenues), Loss ratio (ratio of losses due to claims to gross premium, which indicates underwriting performance), LTV/CAC (to measure customer acquisition performance)

Business models: Managing General Agent model: underwrites on behalf of insurers — an existing model since the 1990s representing c. 15% of the market. Also, white label insurance platform for insurers with open APIs.

Companies: Descartes, Arbol, Celsius Pro, Skyline Partners

5. Specifics by types of risks 🔍

🚜 Agriculture

Value drivers and sources of USP: Distribution model and especially farmers’ onboarding — need for mobile-first and partnership with telco operators in emerging countries

Partnerships — international NGOs and public institutions co-finance programs or policies ensuring economic resilience (ex: CREAD in Dominica, German government premium subsidy in Africa)
Companies: IBISA Network, Pula, Hillridge, OKO

🔋 Energy

Value drivers and sources of USP: Expand from traditional energy-weather products based on temperature to renewable energy (irradiance, wind…). Propose an alternative to weather derivatives to reinsurers who may prefer less volatile financial products (not mark to market).

Companies: Arbol

🔥 Wildfires

Value drivers and sources of USP? Accurate prediction, as wildfires result from several man-made and natural factors: understand the intersection of vegetation, buildings, and infrastructure (esp. electrical grids) management and forecasts.

Companies: Delos, Kettle, Overstory, Cornea

🌪️ Floods, earthquakes, and hurricanes

Value drivers and sources of USP? being able to model and cover very diverse types (niches?) of floods (ex: storm surges, river floods, flash floods, but also groundwater flooding,dam-break flows, backwater floods, debris- or mud flows)

Companies:
Hurricanes: Raincoat, Yokahu
Earthquakes: Jumpstart, Súper, Bounce Insurance
Floods: 7Analytics, StormSensor, Cloud to Street, Fathom, Wetterheld, Birdeyeview

5. Market size & dynamics

📈 Market dynamics

Climate adaptation 🌍 — in general

Bank of America predicts that the climate adaptation market will double to $2 trillion annually within the next five years. The WEF stated that the world needs to invest $5.7 trillion annually in green infrastructure and other adaptation and mitigation efforts. And current financing flows for adaptation efforts could generate an estimated $7.1 trillion return on a theoretical investment of $1.8 trillion globally over ten years.

2. Parametric insurance ☂️ — all sectors

The global parametric insurance market was valued at $11.7 billion in 2021 and is projected to reach $29.3 billion by 2031, growing at a CAGR of 9.9% from 2022 to 2031 (Allied Market).

3. Parametric insurance ☔ - climate risks:

The gross written premiums for parametric insurance policies related to climate events globally were 500m-1bn$ in 2021, and the majority of the company in the space enjoyed 15%+ you growth (Instech survey among 155 actors) e.g. Arbol communicated on its traction, wentfrom 2m$ GWP in 2020, to 170m$ in 2022.

💰 Funding

In H1 2022, parametric insurance-focused companies raised $250m, already more than any other year (Instech). 2021 was already a record with $116 million investment, more than double the $43m of 2020. Pre-2020, $100m was raised approximately. (Instech, Pitchbook)

The largest funding round to date has been Descartes Underwriting, with a $120 million Series B round in January 2022 that valued the firm, post-money, at $714 million. Descartes is gaining impressive traction by focusing on big individual underwritings from corporates and educated investors.

Regarding risk analytics in general, some companies are listed (Verisk, RMS by Moody’s). CoreLogic is another example of a risk analytics company delisted in 2021 and acquired by private equity firms. Those large risk analysis providers are expanding into climate risks: they are at the same time potential competitors for new entrants as well as exit paths (cf. RMS by Moody’s)

On data collection, a notable fundraise is Iceye, with a $136m funding round in February 2022. Iceye develops micro-satellites to capture images from space, and Climavision, who builds a radar network, raised 100m$ from TPG’s Rise Fund in June 2021.

We also see emerging spaces such as the long tail market, where to focus on product and distribution is key (around 3.2bn people today, growing to 5bn by 2030 (OECD), almost entirely un-serviced).

6. The case against tech-based risk coverage: challenges and risks

🤝 Disruption should embark ecosystem: insurtech took off around 2015, with disruptors working in partnership with incumbents more than independent disruption given the very intricate character of the industry (which is less the case in fintech where disruptors could grow a bit more independently)

Our take is that good connections with the insurance ecosystem are essential for new entrants. Incumbents would necessarily take a part of the value.

🏭 Defensibility comes in insurance from economies of scale and network effects. Especially if a new entrant has no disrupting edge, incumbents can underwrite at a loss until the new entrant goes bust. Given the size of incumbents, access to capital is a limiting factor for new entrants.

Our take: access large risk capital amounts is a bottleneck for new entrants wanting to go beyond data analysis: exploring at early-stage JV, partnerships, or hires with an extensive network is a prerequisite

🕑 Natural catastrophes have long feedback loops: a small portfolio will have variability in the first few years, and actual results will be seen in more extended periods (10–15y). Compared with windstorms, earthquake (like a flood) has only attained a low level of insurance penetration globally because risk can appear remote to clients.

Our take is to work on lower amount/frequency risks to reduce feedback loops and build a resilient solution. Awareness of climate risks is necessary.

🧑🏼‍🏫 Evangelization of clients and distribution can be challenging as this is a new way of approaching risk cover.

Our take #1 — big sophisticated B2B — one alley is to target sophisticated stakeholders that understand well the risks: corporate risk managers and brokers that are educated to coverage through similar tools like weather derivatives

Our take #2 — bundle or B2B2C

Personal loans. Ex: short term revenues insurance cover can help prevent customers from defaulting on the whole loan
Operations tools. Ex: covering heat waves impact on milk production integrated in a cattle monitoring system (Skyline Partners)
Other insurance products. Ex: Weather events’ cancellation insurance is embedded in events insurance (Birdseyeview, Wetterheld)
B2B2C. Ex: hospitality industry is heavily affected by nearby recent climate events in zones exposed to climate events, impacting employers (hotels, restaurants..) and employees. For example, tourism accounts for 6.5% of Puerto Rico’s economy: in 2019, the number of tourists visiting the island was still down 36% from 2016 levels after the 2017 storm. A model where employers finance revenues-loss insurance is beneficial for both parties.

Our take #3 — for B2C, leverage technology to deliver the most seamless user experience. Distribution and education of both insurers and insurees

☢️ Are natural catastrophes even insurable by a private entity? The protection gap is especially high for low frequency/high magnitude as covering those risks, even for global reinsurers, is challenging in case of a Fukushima-like event, leaving residual cover to governments. This issue can’t be fixed with technology.

7. Potential exit paths

🛍️ Actors and motivation: The insurance industry would be the natural exit path of these solutions, with strategic (but also PEs) looking for the kind of actors that usually enjoy high margins (Deloitte 2022 report)

🔢 Market data: the insurance M&A market is vast and active, with 25 “mega-deals’’ over $1 billion in 2021. Regent Bidco’s takeover of RSA Insurance Group for $9.2 billion was the year's largest deal.

Examples of exit so far include Jumpstart insurance (acquired by MGA Neptune flood quite early in their development) and RMS (acquired by Moody’s for 2 bn$).

8. Conclusion

We see tremendous potential to leverage recent developments in data science and sensing technologies to adapt and cover against rising climate risks: we see that the companies pioneering the space have been founded in the last 3–5 years in Europe and the US for the most part.

After conversations with many of the companies featured below, we’re very excited by the space. The critical investment rationale for us is about cracking the main bottlenecks of this industry:

In data collection and analysis: Proprietary and superior data access and/or risk modeling, primarily based on physics
In risk assessment and quotation: Efficient distribution model, would it be through software, product bundling, or client’s profile
In risk cover: Unfair access to different sources of risk capital

NB: Given the major cascading effects between various climate risks, we like to see pure players move into diversified assessments and covers of different risks (most of them being in the early phases of their journey).

9. Adjacent opportunities

🌇 Adapting our societies to climate means CAPEX investments in more resilient infrastructure, like burying power lines or developing distributed energy resources and microgrids, but also working on standards and regulations, such as urban planning for flooding or climate-resilient building standards. Governments are increasing public funding for more climate-resilient reconstruction (like the 3bn$ announcement by the US government last March). Companies: Firemaps

🚒 Emergency response. Complementing monitoring and forecasting tools with real-time operational response management and data sharing between different users, agencies, and devices for tracking incidents as a tech-enabled version of the historical 911 system. Other areas of interest are prevention and monitoring of disasters (like fires) or forecast repair costs.

Companies:

Emergency: Tecnosylva, RapidSOS, Pano AI, Serinus, Tractable, Drone Hopper, FloodMapp
Early detection: Orora Technologies, Dryad Networks, Tenevia

Solutions can be easily expanded to non-purely climate solutions

〽️ Business interruption: taking a step back and recognizing that cascading effects of climate-linked risks can not be all anticipated, parametric insurance can be used to cover any business interruption. Companies: Ott Risk
🏘️ Property: help insurance, mortgage and real estate companies better underwrite and rate homeowners and commercial, using geospatial datasets combined with property databases, also connecting all participants in your value chain to create seamless communication and policy status. Companies: CoreLogic, Zesty.AI
🏨 Others: travel and hospitality, maritime