ELI5 #2: Is everything priced wrong (due to climate risk)? (Part 1)
We don't have good ways of measuring climate risk -- I think that we wildly underestimate it. A fuller understanding of climate risk will require more detailed understandings of our built world.
hey :) I’m Anant and I’m another 20-something conjecturing about technology. My goal is to take complicated ideas, break them down, and develop a perspective. I’m writing to hold myself accountable to learning — hopefully you’ll find it interesting. Maybe you won’t. Thanks for reading either way.
hey everyone ❤️,
Welcome to our new friends :) fun to have you here.
At the time of writing, it’s a balmy Saturday afternoon here in San Francisco, CA. I just came back from a panel on climate activism ahead of SF Climate Week (which had SO much good programming, by the way) — the speakers were inspiring and committed in their own right, but it was hard to heed empty encouragements to “touch grass” or “recycle plastic cups” (I heard a lot of these bromides on Earth Day this year). Calls for climate action stand against the backdrop of potential runaway climate change. God, I’m already such a buzz kill.
And as if I couldn’t be more brutal: I’m going to make the case that we’ve under-estimated the effects of climate change on our daily lives because we don’t have good ways of measuring climate effects. Just stick with me for a sec.
Look, for example, at the below chart of water temperature rise. Which has implications on hurricane severity. Which has implications for weather patterns that influence our lives. Which has implications for when your small business can open or your kid can go to school or how much you pay for groceries. You get the idea. OK — let’s dive in.
In February, the globe passed the internationally agreed upon threshold for irreversible climate effects: 1.5 degree Celsius. What do these effects look like? Well in the short term, they look like hundreds of lives lost in Pakistan and Afghanistan from flooding or like New York skies orange from wildfires or lost biodiversity. And think of all the knock on effects of these disasters: spiraling healthcare costs due to forced displacement, loss in productivity and output from business interruption, school shutdowns due to severe weather. The climate affects our built world, which then affects just about every facet of our lives.
I want to spend some time talking about how we’ve traditionally thought about climate risk and make the case that maybe the effects of climate change are greater than we’ve been able to measure.
What is climate risk?
Climate risk is typically categorized into one of two types: (1) Physical risk, and (2) Transition risk.
Physical risk is the impact of climate change on our built world. Think of this as damage from extreme weather events, like hurricanes, floods, or wildfires, or gradual changes in climate patterns, like rising sea levels or changes in temperature. Physical risks can result in direct costs from repairing damaged facilities and assets, as well as indirect costs from supply chain disruptions or business disruptions. This is, for example, the risk to a business caused by potential flooding from a storm. For the climate homies, mitigation and adaption are both ways of trying to reduce physical risk.
By contrast, transition risk includes the risks associated with the transition to the aspirational low-carbon economy. As societies and economies shift away from fossil fuels and eco-unfriendly activities, there are potential financial risks for companies and investors tied to changes in policy, regulation, technology, consumer preferences. This is, for example, the risk to fossil fuel companies that would come from the government increasing taxes on emissions.
There are, of course, inter-dependencies between these two types of risk, as increasing physical risks drive more rapid policy changes and technology shifts to mitigate climate change, amplifying transition risks. And, in contrast, inadequate transition efforts exacerbate future physical risks.
It’s generally believed that so far, transition risk has been greater than physical risk, or put another way, that the risk to businesses from policy and regulatory changes (e.g., increased taxes) has been greater than the direct impact of climate change on those businesses. But what if we’ve got it all flipped on its head? Let me tell you why that might be.
How do we measure physical risk?
We’ve historically had a narrow definition of physical risk, or at least limited ways of quantifying it across its broad definition. Today, the most straightforward assessments of physical risk look at exposure of physical assets. At the risk of simplifying, think of it like this: an assessor (typically a services provider, like Deloitte, leveraging a climate analytics tool, like Jupiter) asks the question: “How many of Walmart’s stores are in locations of likely flooding?” They’ll pull locations of stores and warehouses, map them against high risk flood zones, and determine that most stores are outside these zones. They may conclude “Walmart is mostly safe! The physical risk exposure is minimal.” Phew! Good work, Walmart (and congratulations, Walmart shareholders)! I’m being a bit flippant, but the point I’m trying to make is that the easiest way to measure climate risk is to look at the direct exposure of our physical assets.
The below chart of insured global losses due to catastrophic events paints a bleak picture of our resilience to climate change. These are mostly losses to our built world, and it doesn’t take a genius to realize they’re increasing.
But, the reality is that actually the second and third order effects of physical damage are far greater than the direct damage itself. For example, 90% of commercial facilities are not in high risk flood zones (according to the National Flood Insurance Program). It feels incomplete to assume that these businesses are therefore not exposed to climate risk. I mean, what about all of the other parts of the built environment that affect the business: What about the delayed ship that can’t deliver merchandise? What about flooding at the electrical transformer that creates a power outage affecting a store miles away? What about transportation blockages that prevent employees and customers from getting to retail locations? Surely these things should also be included in risk estimates (economists call these effects “spillovers”).
It’s estimated that these ripple effects actually account for 95% of the cost from physical risk, or put differently, that 95% of costs are unrelated to effects on a physical building and are not currently included in physical risk calculations! We may be underestimating the effect of climate change by 20x!
How can we measure physical risk more effectively?
Using various methodologies, some (across academia and the private sector) have worked to include assessments of: (1) operations risk, or the risk to the operation of the company (this would, of the aforementioned list, include the risk of energy outages shutting down the store), (2) supply chain risk, or risk to the inputs of the business (think of this as a Walmart supplier not being able to make new merchandise), and (3) market risk, or risk at the point of sale (like blockages that prevent customers from accessing retail locations).
Different companies have taken fundamentally different approaches to assessing climate risk, combining different data sources: (1) Satellite imagery: Planet and Satellogic both leverage satellite imagery to help companies locate assets and identify climate risks (see below picture about their workflow); (2) On-the ground data: Sust Global combines satellite imagery with ground sources to create full-stack geospatial analytics; (3) Company data: Jupiter Intelligence sells through consultants, like PwC, to analyze site data against disaster data and connect results to business losses.
It’s probably no surprise to learn that these things are really hard to measure! In fact, one paper looked at six different climate risk ratings that used different methodologies (some used models; others used news reports, earnings calls, etc.) and found wildly different scores (the paper put it a lot more ahem academically: “a low level of agreement between scores over the sample”).
That’s not so surprising at an individual company level, but what’s more surprising is the divergence even at the market level. Across the six samples, there wasn’t even clear agreement on which industries were most susceptible to climate change. A running average would suggest utilities, energy, and materials, which makes intuitive sense given the risk exposure of physical assets, but you’ll see the variance across scoring methodologies.
Why is measuring climate risk hard?
Ask six experts what they think about your business’ exposure, and you’ll get six different answers. Why is it so hard?
Firstly, different types of businesses are so fundamentally different. The lack of foot traffic in front of a tech company’s headquarters means nothing, but the lack of foot traffic in front of a coffee shop would predictably limit sales. That probably feels obvious, but it means that understanding the risk to a business requires understanding the real drivers of the business, the ins and outs of its operations, the points of vulnerability, and then consequently the likelihood of effects at those points. Getting to the point where you can actually understand all of these interrelated influences requires systems-level thinking across a series of specialists — across climate, AI, urban planning, and business.
Additionally, it requires having a granular understanding of the built world and its interconnected parts. Think of the ripple effects from a catastrophe — you’ve got to know that an outage at Port of Newark (one of the busiest ports in the US) is probably going to have knock-on effects across the greater tri-state area. The more granular you are able to get (e.g., this business is connected to the same power line as this hospital so in the case of a potential outage, they’re likely to be up and running more quickly), the more accurate you can be. This kind of asset, with granular data about the built world, requires lots of up-front investment (not unlike foundational AI models); it’s hard to underwrite!
And finally, time: the effects of physical changes are different on different time horizons. For example, a power outage shutting down a store is immediate, but a week-long delay in parts arriving may not really affect a factory’s operations. Accurately understanding physical risk exposure requires understanding the nuanced effect of time.
So then how do we crack the code?
It’s clear that there are as many approaches to measuring climate risk as there are companies trying to crack this code. But I suspect that the leader, the one that is going to be able to find real signal from all of this noise, will do two things:
They will leverage new and different data sources (not just satellite data), maybe deeper knowledge of our built world, companies’ supply chains, real-time weather sensors. One Concern, for example, is taking the approach of building a digital twin of the Earth’s infrastructure, with the thesis that understanding our built world will help us understand its vulnerabilities. Earth Force is using vegetation data to prevent wildfires. Not quite climate risk, but Windborne is gathering detailed atmospheric data to help predict the weather.
They will go after particular verticals / use cases (not massive datasets applied horizontally for everyone) so as to be able to provide clear and up-front value for customers. For example, Orora is helping predict wildfire risk; ClimateAI is helping with climate risk for agriculture; Near Space Labs is providing data for property and casualty insurance; Arbol and Understory are providing parametric insurance).
Concluding thoughts
I think there are real spoils for the people that are able to figure this out — I’m going to do a whole other post (will link Part 2 here, when it’s done) on the potential use cases and the market size. But let me preview by saying that if the real economy balance sheet is ~$500T, and it’s 1% mispriced (conservative), that’s $5T of mispricing…
It’s worth answering the question originally posed by the title of this piece: Is everything mispriced due to climate risk? You can probably figure out at this point that I think YES. Things are mispriced (publicly traded equities, insurance policies, your lunch), and unfortunately, it’s probably the case that understanding the real cost of physical risk would make you short more stocks than buy to hold, but with the current tools that we have, it’s hard to know by how much. Kind of a cop-out, but have some compelling data on this to share in the follow-up.
And finally: I’ve been chatting with a founder taking an innovative approach to a lot of the above — connecting climate risk to financial risk in ways that will empower us to make smarter decisions about building resilience. He’s brilliant! If you’re thinking about any of the same, I’d love to be in touch.
was JUST talking to a friend about the new waves in climate tech - super impactful and high-growth space. love it!