It’s Been Fun!

Hey - It’s been 103 days since my last post. Reflecting on 2024 and writing out the “big swings” I want to take over the next few years, I noticed that “keeping up with the newsletter” didn’t even occur to me. I’m always evaluating the basket of activities I’m doing and want to do, optimizing for things that matter most to me over a given period. I think a lot about Tim Urban’s post quantifying life phases, particularly how about 90% of a kid’s in-person time with their parents happens from birth until age 18, and I have. There are only a few years left with my kids until we hit that 90% number. All of this said, it’s deeply important for me to sharpen my focus on the most fun and useful professional and personal activities. So it is bittersweet to say I’ll be retiring this newsletter, effective today.

I’m very grateful to everyone who read this newsletter and gave me words of appreciation and encouragement in the past several months. The posts were an invaluable activity to me during my sabbatical last year, and I enjoyed the process of crafting each post, but as you can see from my 3+month silence, I’ve been busy with a host of other exciting things that I’m attending to:

1. Continuing the rapid expansion of my advisory firm, Apex Catalytic. After launching late last year, I have a ton of momentum following some incredible projects spanning AI, impact investment strategy, growth-focused R&D, ESG software, creating circular supply chains, and more. If you’re grappling with a novel + tough sustainability problem, please reach out: jon@apexcatalytic.com.

2. Building out new and enhancing my existing software and data products. I created what I think is the world’s biggest circular economy database, which I’ve been using to support my project work with Apex Catalytic. I also built a powerful decision support tool that helps corporates effectively evaluate and properly invest in data, AI, and software to help them move their sustainability actions faster. Again, if you’re interested in learning about this work, reach out at jon@apexcatalytic.com.

3. I am continuing my work to enable 10,000 sustainability professionals by 2030 (I’m about 20% of the way there based on work last year).

4. Advancing other writing projects. I’m putting energy behind a couple new and exciting writing projects, including Apex Briefings (timely and targeted insights on new developments relevant to impact investors and corporate sustainability leaders, which I send directly to personal and professional contacts), and writing a book (I have a couple outlines and am deciding on a publication route).

Thank you again for reading; please get in touch with me on LinkedIn or via email to stay connected. I don’t know the protocol for ending a newsletter, but I thought these lines sum things up at this point:

Quick Announcements

Hi there! Here are a couple of quick announcements before we dig in:

1. I’ll be in <Book of Mormon voice> Orlando on October 2 at the E-Scrap Conference, running a panel with some great leaders and innovators on Digital Product Passports (DPPs) and how they might help to drive sustainable outcomes for electronic devices and much more in the near future (if you’ve never heard of a DPP, I describe it as ‘Like a barcode or UPC, but for sustainability-relevant data’). If you’re reading this in your email inbox, hit reply and let me know if you’re interested in receiving a short DPP primer that I’ve written for that conference (or email us at sustainabilityatthefrontier<at>gmail<dot>com. Also - say Hi if you’re at the conference!

2. I’ll be in New Haven the first week of November and am delighted to be delivering two lectures, one to Prof. Yuan Yao’s Industrial Ecology class on Next-Generation Corporate Sustainability Action, and one to my mentor and doctoral committee advisor Prof. Dan Esty’s Net Zero Pathways course on Emerging Data and AI Frameworks Driving Deep Decarbonization. Reply to this email if you’re in New Haven or NYC and want to get together.

A big Thank You to those of you who voted in the poll in our last newsletter. We had unanimous feedback for a post on GenAI. There’s so much to cover, we’ll try to scratch the surface in this initial post…I expect at least a couple follow-ons in the coming weeks and months. Enjoy!

If You Read Nothing Else in this Post:

After more than 2 years of using various generative AI (genAI) applications as a small part of the flow of my work, I argue that a fairly narrow range of genAI applications can be useful for sustainability professionals. Recognizing that sustainability teams at corporations and investment funds, who are charged with developing right-sized sustainability strategies and then executing projects to achieve the aims written in these strategies, are often under-enabled, under-funded, and short-staffed, I believe it’s in the interest of most sustainability professionals to achieve some minimum level of competence in using genAI in the flow of their work. I found that a combination of an initial on-ramp of about 15 hours of directed, role-specific practice using a leading genAI tool, coupled with tracking heavy genAI users for inspiration and new capability awareness, can result in an increase in efficiency and unlock some compelling uses that can help create more positive impacts than not. This is a “table-setting” post where I lay out some of the issues or challenges around genAI, some perspectives I’ve gleaned as a moderate-to-heavy user over the past couple of years, and tee up some future posts by pointing to use cases that are well-suited to genAI’s current capabilities.

What this Post Is Not (Yet)

Discussions around large language models (LLMs) - the models underpinning popular genAI tools - are fraught with tension, and with good reason. There are serious ethical concerns (e.g., intellectual property scraped from the internet without the creator’s consent to train LLMs), environmental concerns (we will pick apart and provide clarity on what’s known here in a future post, but briefly, the issues surround water and energy use by data centers on which LLMs train and are run), financial concerns (e.g., whether the growth-at-all-costs ethos of large technology companies is economically sustainable), and practical usage concerns (e.g., are genAI tools even good enough to be helpful?).

I’m not an expert ethicist, so I do not think I’m in a great position to comment deeply on that topic within these pages. Admittedly, I go through waves, and the use of genAI tools at times elicits more of an icky feeling than at other times (e.g., the “ick factor” was high recently when LinkedIn announced by default everyone’s data would be used to train AI models unless a user navigates several screens to opt-out). Where I’ve landed on using genAI looks something like this:

Similarly, I won’t dive into the myriad financial concerns here, but for those interested I recommend Ed Zitron’s newsletter, which goes into painful detail about the ways in which technology companies are doing things wrong and may be creating some dire consequences for themselves and the broader economy as they chase growth and returns associated with AI and genAI. I’ll link to his newsletter at the end of this post.

Finally, this post will not touch on the environmental concerns surrounding the current and potentially increased use and training of LLMs. I’ve gathered and pored over a ton of papers that have come out over the past couple of years and will say that (i) the target is moving a lot because of the evolution of various LLMs and (ii) it’s still tricky to pin down a simple “so how bad is it for the environment if I use a genAI tool?” because there’s still a lot of opacity around specific effects that (say) one prompt to generate an image has. I’ll warn you - the story is complicated, but in the spirit of what we try to do in this newsletter, I’ll do my best to demystify and boil down what we know into some valuable chunks of information - in a later post.

Why Should We Trust You on GenAI, Jon?

As a space getting a ton of time, attention, and investment, there’s an accompanying amount of grift surrounding AI and genAI, just like with other supposedly-revolutionary technology platforms or applications. It was only a few years ago that non-fungible tokens (NFTs) were a thing (one of my favorite grift stories around this is linked at the end of this post)! I’ll admit it - it’s tough to separate the wheat from the chaff regarding genAI and its capabilities. I hear this sentiment from sustainability and other professionals I frequently speak with on the topic. Will it materially transform the way we all do work now and forever? Is it an incredibly hollow and useless technology that amounts to nothing more than a parlor trick? Would the money poured into developing and scaling the tech been better off by being set ablaze?

I fall somewhere in the middle of the Useless-Transformational scale of opinion around GenAI. For me and the kind of work that I do, I think my opinion on the topic is valid because I’ve put in the hands-on-keyboard work to see for myself the handful of ways genAI tools are helpful (and the far larger number of ways they are not applicable or useful). I got selected for GPT-3 access sometime in early 2022 before its public release. I’ve probably spent an average of 1-2 hr weekly since then using various genAI tools with deliberate practice. I’ve used each of the major models, and the ones I subscribe to at a given time change as the tools evolve (Over the years, I’ve been a paid subscriber of OpenAI’s models, Anthropic’s Claude, Google’s Gemini, and Perplexity’s tool). Keep in mind that I am not an unbiased messenger on this matter. I’ve had the opportunity to give talks to thousands of leaders on the promise and peril of genAI in recent years. With my new firm, I advise corporates on how they right-size their use and investment in technology, including genAI, and will continue to do so. But those who know me can attest how (perhaps to a fault) I weigh multiple sides of an issue and have a strong willingness to update my thinking when I get new information.

Hopefully, with this backstory, I’ve given you the confidence that you can trust that the remainder of this post - and future related posts - is well-informed based on my experience using these tools and a lot of reading, speaking, and listening to other professionals in the space. I will present a few unique cases where genAI has been helpful for me, along with a brief narrative or example illustrating the concept. I’ll plan a future post focusing more on ‘the how’ and bringing additional examples to life.

GenAI for Sustainability Brief Discussion of Use Cases, Details coming in Future Post(s)

As promised up front, I’m calling this a table-setting post. I’m already at 2,000 words! In this section, I’ll provide example use cases where I’ve found genAI to be uniquely useful and a helpful complement to my normal workflow.

1. Large-document research and review. There are many examples, but a classic one is reviewing new legislation to understand if and how it applies to your company. New regulations can be hundreds of pages long, and there are often large sets of supporting documents that accompany the official regulatory text (e.g., background on the legislation, summaries of meetings with stakeholders, etc). Trying to glean insights from thousands of pages of documents is daunting to even the most skilled researchers. I’ve found that a good prompting framework and articulating your critical questions in advance can materially reduce the time it takes to ‘get up to speed’ on a new topic, particularly details-oriented information like new regulations and legislation. It varies depending on the topic, but my level-of-effort is probably reduced 20-30% in some (but not all) cases when I’m diving into a new sustainability-related topic and need to research core documents to help inform questions to ask or my overall understanding of the topic.

2. Conversation and negotiation scenarios. Using the voice conversation capability (I’ve used it with ChatGPT so far) can provide unique and under-discussed benefits to genAI. Imagine you’re applying for a new corporate sustainability job - you could upload your resume and the position description to ChatGPT, then prompt it in the voice conversation modality to be the hiring manager, and you can have a mock interview at the drop of a hat. Alternatively, say that you are entering any negotiation (it could be with your manager to discuss a raise or promotion, it could be negotiating terms with a client, etc.) - the voice conversation can be given a persona (e.g., “you’re the world’s toughest negotiator”) and you can hold a back-and-forth conversation to help refine your pitch. It’s a compelling way to work out some talking points or to get practice discussing something simply. To me this is largely a “net new” capability unlocked by genAI - most of us don’t have the luxury of dialing up an expert in <name your topic> who will give us undivided attention and play a specific kind of role just to help us brainstorm or practice. There’s an odd feeling, too, when you “beat” the genAI in a negotiation, too. One of my favorite variants of this use case involves preparing for a conference presentation. I assign ChatGPT the role of a formidable graduate school adviser - they then proceed to relentlessly question the accuracy and importance of my topic, requiring me to think on my feet and turn the topic on its head in multiple ways. I’ve had plenty of (human) sparring partners like this in the past, but there’s something super convenient about getting your practice in at any time.

3. Creating topic-specific ‘podcasts’. Google has a new-ish tool called NotebookLM, which along with summarization capabilities and other features common to other leading genAI tools, it has a unique feature allowing you to quickly and automatically create a 'podcast-style audio file on topic(s) of your choice, informed by documents that you upload. Let’s say you wanted to understand more about a start-up's technology, and you had access to a white paper they wrote, the company’s pitch deck, and a couple of media articles on the company. You can upload these documents to NotebookLM and it will instantly create a podcast-style audio track featuring two people speaking back and forth on the contents of the documents you just uploaded. This feature is a little eerie but I can speak to the convenience of execution of this format, and you can really see this being useful for people who are more audio learners. .

A Related Note on Data Analysis and Predictive Analytics

I will probably explore these areas in a separate post, but I want to provide some commentary on the applications of genAI for data analysis and the subtopic of predictive analytics.

On data analysis - in brief, I perform moderate to complex data analysis as a part of projects and advisory that I deliver with my company, and data-driven work has been a key part of my professional “secret sauce” for a long time. At this point, I only trust genAI tools to do a limited set of tasks for me, mainly around getting the ball rolling for a coding task (R is my preferred language). I’ve thrown simple to moderate tasks at genAI and frequently get erroneous, frustrating, or just plain weird responses. Keep in mind I’ve mainly used genAI tools in this way as an individual user of the Pro (i.e., paid) versions of the largest and most sophisticated models, which means I’m mainly using best-in-class models. Back when I worked at Salesforce and had access to Enterprise tools, I can say that the results were not any more trustworthy. Others who are research scientists deep in AI and machine learning have reported similar benefits and cautions of using genAI for data work.

As for predictive analytics - I do not have much faith in an AI or genAI tool to do any predictive analytics better or faster than I could using standard methods and professional judgment. A key conclusion from a recently-published book by a couple of prominent academic figures in AI from Princeton essentially said “Predictive AI doesn’t work, fundamentally, any better than clipping a coin, and we don’t see how it could substantially improve in the future”. Pretty damning. As with other developments around genAI, I’ll continue keeping a “watching and learning” approach and update my thinking based on hype-free evidence and experience.

Related GenAI Links of Possible Interest

Below are links I’ve referenced in this post. The first two are good bookends for people who want to explore the pessimist and optimist/utilitarian views on genAI.

-Ed Zitron’s Where’s Your Ed At? Newsletter (Free) - Longform posts mostly pessimistic about genAI and the tech companies developing and pushing it.

-Professor Ethan Mollick’s Newsletter - A generally optimist view of genAI filled with lots of use cases and experiments he’s run for himself and with his MBA students at UPenn’s Wharton School

-Awesome NFT Post: What the Hell is this Company the 76ers Just Partnered With?

-AI Snake Oil: What Artificial Intelligence Can Do, What it Can’t, and How to Tell the Difference (Book)

Do you know anyone who may enjoy this post? If viewing this in your email client, share this post with a friend by clicking below!

If you’re viewing this online, simply copy this link and email or post it to those who may enjoy the newsletter. Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

Do Extreme Weather Events Influence How Political Parties Behave Regarding Climate Issues?

If You Read Nothing Else in this Post:

Politicians and political parties routinely publish press releases on various topics, including those related to the environment and environmental policy. The focus or attention of a given party or politician on one issue or another can be reflected by the relative frequency with which they publish a press release on a given topic. Recent research explored how extreme weather events (fatal storms, floods, wildfires, and extreme temperatures - all of which are signifiers of a changing climate) influenced the attention paid by European politicians to climate, sustainability, and environmental policies. After analyzing 260,000 press releases published over a 10-year period from 9 EU countries, they found that only explicitly defined “green” parties increased their attention on climate issues following extreme weather events.

Weather and The Year of Elections

One of my favorite downtime activities when traveling abroad is watching international news outlets cover stories happening in the US. It often feels foreign, both because news is happening there and I’m here and because the tone and temperature of the discourse seems so mild (or at least different) compared to what we’re used to in the US. For example, on a recent trip to the UK, I was stopped in my tracks when, in a story on a new energy development project in the area, a government official stated on camera that (paraphrasing) we “…need to make sure any new development takes into account the life-cycle impacts so that we do not continue contributing to climate change.” Not coincidentally, the news outlet presented that story alongside one on a heat wave that was also sweeping the area. Since returning stateside, there’s been a steady beat of extreme weather stories spanning the early, destructive hurricane that landed in Texas, tornadoes in the Midwest, flooding in Canada, and others.

On top of extreme weather coverage, news outlets both in the US and abroad are, of course all over the elections around the world. It turns out that countries representing nearly 50% of the global population will undergo a national election in 2024, totaling 64 countries plus the European Union.

Our focus in Sustainability at the Frontier is to dig into new research and insights that can help us to understand what’s happening at the leading edge of sustainability and how that might apply to our work or otherwise inform our perspectives. So, with the recent exposure to a deluge of extreme weather news and election coverage, I was delighted to see that Tim Wappenhans and co-authors just published a short research study in leading journal Nature Climate Change that asked a related, compelling question: How do politicians act (and react) on matters related to the environment and climate change when extreme weather events happen?

Let’s dig in.

Press Releases as a Proxy for Political Attention

In simple terms, we might think of how individual politicians and political parties care about one issue or another as reflected by some combination of:

1. What they say

2. The bills that they help write and shape

3. How they vote

By sheer volume, we’d expect to have far more evidence on the attention and posture of an individual politician on a given political issue in (1) more than (2) or (3). It turns out that the press release is an increasingly common instrument for politicians to speak directly on some issue, both to articulate a position or relevant news item to constituents and to provide the press with fodder to (for example) build out a story on some news item of the day.

Before reading this research article, I felt that “yeah, that makes sense; comms teams for politicians probably put out press releases every so often to update people on what’s happening.” However I grossly underestimated the number of releases and the range of topic areas. As one example, I visited the website of US Senator Dick Durbin, a long-serving Senator from Illinois, to understand the number and type of press releases his team put out. I was somewhat shocked by what I found!

Here’s a screenshot of his most recent press release as of July 17, 2024:

Political press releases range from the mundane (past press releases for Sen. Durbin highlighted his pending hip surgery, for example) to more politically important issues (e.g., passages of major bills, funding announcements, etc.)

The content of that specific press release wasn’t so shocking - it was the massive number of press releases contained across multiple pages:

US Senator Dick Durbin’s website boasts 420 pages of press releases dating back to when he became a senator in 2006.

Each page had about 20 unique entries, putting the total number of published press releases since he became US Senator in 2006 at roughly 8,000. The pace at which press releases are put out has increased dramatically - his office put out 16 press releases in the past week alone - nearly five times the total number of press releases they put out in all of 2006!

I bring up these facts and figures as one example to illustrate that politicians increasingly put information out about news and the issues they care about in press releases. As a researcher, this can open up all sorts of interesting questions and possibilities, mainly if you analyze the contents of press releases alongside other data or information to explore a relationship. And that’s exactly what the research team did for the article we’re examining here.

Specifically, the research team did the following:

1. Gathered 260,000+ press releases from politicians and political parties from nine European Countries: Austria, Denmark, Germany, Ireland, Netherlands, Poland, Spain, Sweden and the UK, dated between 2010 and 2020.

2. Analyzed data on “extreme weather” events corresponding to fatal storms, floods, wildfires, and high temperatures in the nine countries between 2010 and 2020.

3. Classified all press releases as to whether the release covers the topic of Environment, defined by the authors as a press release discussing climate, sustainability, and environmental policies. The authors classified the presence or absence of discussion around the topic of Environment as Attention.

4. Analyzed whether, on average, the Attention paid by politicians (again, as demonstrated through press releases) to environmental issues changed after an extreme weather event occurred.

The authors used a statistical technique called Difference-in-Differences to gauge how much Attention changed after an extreme weather event. In brief, this means they established an overall “Average Attention” across all the data. They looked at the Attention to Environmental issues before and after a given extreme weather event. Let’s take a look at a graph illustrating their key findings related to Floods:

Results (Published in part in the original paper as Fig 2) show attention up to 6 weeks before (blue) and six weeks after (red) a major flood event. The results show no change in attention paid toward the environment in press releases, even after a major flood event across the analyzed data set.

A lot is going on in the above graph; here’s how to interpret it:

1. Dashed Line: The average attention paid in press releases on climate and the environment in the data set.

2. Blue triangles and lines: The blue triangles represent an estimated “average attention paid to the environment before a flood event.” The blue lines represent the 95% confidence interval of the mean - there’s a 95% chance the true average falls somewhere between the extents of the line.

3. Red circles and lines: The red circles represent the “average attention paid to the environment after a flood event.” The red lines represent the 95% confidence interval, as described above.

4. The numbers -6 to 6: This refers to observations in press releases up to 6 weeks before (

OK, now we understand the data, methods, and techniques this group used to display their results. So, what did they find when analyzing the political press release data across all types of extreme weather events?

The main concluding figure shows the attention paid to environmental issues in press releases across four types of Extreme Weather events. The results show that extreme weather events did not “move the needle” in terms of attention paid by politicians and political parties.

The key results show that attention paid to the Environment (i.e., sustainability and climate change issues) did not increase across the entire data set they analyzed across nine countries and more than 260,000 press releases over ten years. They did a separate analysis evaluating whether there were any political party-specific trends, finding a tiny, short-term uptick in attention to climate issues in those parties identified as green-leaning. This is a staggering result for a few reasons:

• They in part defined Extreme Weather as events that resulted in fatalities, which you’d expect to produce a stronger signal and spur attention (and action) by politicians more than a run-of-the-mill storm.

• The “null result” shows no increased attention held up across multiple categories of environmental issues (save for the small effect seen with green-leaning parties) across numerous political parties, countries, and years.

• European voters are pretty attuned to climate issues (remember my comment above about the politician bringing up life-cycle thinking?). In fact, the European Investment Bank recently polled more than 30,000 adults and found that the UK and EU have a better understanding of the definition and causes of climate change than those surveyed in the US. Hence, it seems reasonable that an informed population would be a receptive audience to politicians talking about climate change, particularly in the wake of extreme weather events. Here’s a figure showing some results of that European Investment Bank poll on the relative grasp on fundamental definitions and causes of climate change:

Final Insights and Takeaways

The authors of this new study showed compelling evidence that in the geographies and periods studied, extreme weather events appear to have very little influence on the average attention paid by various politicians and political parties on climate-related matters. Their use of press releases as a proxy for the attention paid to environmental or climate issues was a clever one, given the sheer volume and velocity of information contained therein. As we discussed, press releases are one measure to display how political parties and politicians think, but that the “null” result held across multiple years, geographies, and (most) political parties across a range of extreme weather events is telling.

Although it’s crucial not to generalize these results beyond their study boundary, it raises the question of “if extreme weather events can’t get politicians to talk about and pay more attention to climate matters, what will?” There’s an interesting disconnect buried within the paper we reviewed here, namely the citation of previous studies (here and here) showing that extreme weather events influence voter opinions, including on climate change. So if extreme weather events sway constituents, but politicians aren’t swayed, that suggests other influential factors might be needed (e.g., displaying economic growth potential from climate tech, etc.) to spur attention and policy to mitigate or adapt to climate change.

Of course, press releases may be an imperfect measure of what politicians are genuinely paying attention to. Even if there was an observed effect shown in this study that increased attention was paid after extreme weather events, it’s not a given that the increased attention leads to action like bills written to address climate change, and the passage and implementation of those hypothetical bills.

It could be an informative exercise to re-create the methods used by this study’s authors to see if extreme weather in the US has a notable effect on how Senators or congressional representatives pay attention to climate. The results could go further by assessing voting records on related issues (e.g., by linking raw vote information (e.g., via the US Senate’s vote tallies) or leveraging aggregated data put together by others (e.g., that compiled by GovTrack or the scorecards and vote tallies on climate-related issues specifically published by groups like the League of Conservation Voters)). Such work could provide important clues for those advocating for investment and pro-climate legislation to understand better the best mechanisms to reach and spur action from lawmakers.

Do you know anyone who may enjoy this post? If viewing this in your email client, share this post with a friend by clicking below!

If you’re viewing this online, simply copy this link and email or post it to those who may enjoy the newsletter. Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

Fracking as a Means to Meet EV Battery Demand?

If You Read Nothing Else in this Post:

Electrifying the transportation sector is a key lever toward reducing global carbon emissions aligned with consensus targets to avoid the worst impacts of climate change. Vehicle electrification at scale requires a substantial increase in the use of critical minerals (such as lithium for electric vehicle batteries) compared to today’s levels, so finding additional sources of lithium to meet demand is essential. New research found that wastewater produced in hydraulic fracturing operations in Pennsylvania contains large quantities of lithium, enough to meet the lithium demand equivalent to nearly 100,000 Tesla Model 3s. Current management practices of hydraulic fracturing wastewater and uncertainties about the recoverability of lithium from the wastewater streams could limit this as a viable source, but US governmental incentives could spur enough innovation to make its recovery viable at an industrially important scale.

Background on Lithium and its Uses 

If you’re anything like me, you probably didn’t think about the word lithium until around 1991 when Nirvana released their hit album Nevermind. “Lithium” was the third single from that album, peaking at 64 on the US Billboard charts.

Contrary to beliefs held by very few, Nirvana’s song “Lithium” is about a man finding solace in religion, not the mineral lithium. This is an image of the Lithium single released in Europe that (according to Reddit) depicts a doll from Kurt Cobain’s doll collection. Also, “titres” is French for “titles.”

Global chatter around “lithium” has dramatically increased in recent years, though, owed mainly to its importance in high-performance batteries used in electric vehicles (EVs). Let’s take a look at how “lithium” and “EV battery” have trended based on US Google searches:

Google searches for lithium spiked (relative to historical trends) as EV production ramped up in the US.

As we’ve discussed before at Sustainability at the Frontier, the transportation sector creates around 20% of global greenhouse gas (GHG) emissions, and electrifying the transportation sector is an important lever to reduce the reliance on gas- and diesel-powered vehicles and, therefore, GHG emissions. Although numerous innovations contributed to the increased deployment of EVs globally, arguably the development of the lithium-ion battery has had the greatest impact. We won’t go into the entire history here (this is a nice writeup), but the short story is that early lithium-ion batteries were developed beginning in the 1970s by a chemist at Exxon-Mobil, and further developments by two other scientists in the 1980s and 1990s more or less got us where we are today (the trio was awarded the 2019 Nobel Prize in Chemistry). The big innovations provided by lithium-ion batteries were their rechargeability and relatively lighter weight compared to other battery chemistries, making them suitable for a range of applications.

In 2024, around 87% of lithium used in production will go into a battery, which is astounding, and far more than any other use of lithium. Data from the United States Geological Survey’s Mineral Commodity Summary for Lithium shows the different areas where lithium is used:

About 87% of mined lithium is used for batteries, far more than any other use. Data Figure by Jon Powell at Sustainability at the Frontier, data source: USGS 2024 Lithium Mineral Commodity Summary.

We’ve now established how the bulk of lithium used in industrial production goes to batteries, but we’re still in the early stages of large-scale electrification of the transportation sector. This begs the question - what will the demand for lithium look like in the future? The International Energy Agency estimates that when compared to lithium demand in the year 2020, 42 times more lithium will be needed by 2040:

Data from the US Geological Survey indicate that most of the 180,000 tons of lithium produced annually globally come from just 22 mineral, mining, or brine operations, none of which are in the United States. Because lithium demand is expected to spike in the future, researchers and other practitioners in the field are working to understand how the demand will be met. As a related aside, the US Inflation Reduction Act, Defense Production Act, Bipartisan Infrastructure Law all in some way support the clean energy transition (including battery technology) and there are massive financial incentives intended to spur more domestic production of lithium, in part to address potential supply risks from other countries that currently dominate global lithium production (here’s some great analysis on the topic by researchers at Columbia University).

With this massive uptick in lithium demand in mind and a new push for more domestic (US) production of lithium, let’s turn to a newly-published study from National Energy Technology Laboratory researchers in the journal Scientific Reports that highlights a somewhat surprising potential domestic (US) source of lithium to help meet demand: produced water from hydraulic fracturing (fracking) operations in Pennsylvania.

New Study - Estimates of Lithium from Fracking Operations in Pennsylvania 

Before digging into the details of the study, let’s examine some basics about how fracking works. Here’s my one-sentence description: A well is drilled into the earth where oil and natural gas exist, after which fluid and sand are injected under pressure to release the oil and gas - the oil and gas is collected. The resulting water (a mixture of the injected water and waters already present in the area of injection called produced water) is brought to the surface. Here’s a short clip depicting this process:

This diagram depicts the basic processes of fracking. In (1), fluid and sand are injected under pressure, after which oil and gas and produced water are brought to the surface (2), followed by management of the produced water (3). Animation adapted from Mother Jones.

In addition to great cheesesteaks, Hershey Park, and the Crayola crayon factory, Pennsylvania is also home to a large swath of the Marcellus Shale formation, which is where about 20% of US natural gas production derives and is expected to be a key source of oil and natural gas supply for the foreseeable future.

Researchers in this new study point out that the Marcellus Shale was formed during the Devonian Period in geologic time - around 400 million years ago - and the formation contains volcanic ash relatively rich in lithium. So when fracking operations occur, the produced water contains many mineral remnants, including lithium. The researchers further acknowledge that, for the most part, there are few beneficial uses for the produced water made during fracking operations - nearly all produced water undergoes some treatment process and then is recycled as injection fluid - so they set out to explore just how much lithium may be there. If the amount of lithium is substantial, it could create a silver lining (a new domestic source of lithium that can be used in EV batteries) from oil and gas operations (which are big targets for reduction as part of decarbonizing the energy sector).

It turns out that Pennsylvania’s regulations require oil and gas operators to track and report various types of data, including (i) the volume of produced water that flows from each fracking well, (ii) analytical data on chemicals found in the produced water, and (iii) other operational, locational, and characteristic data for each well. You can go here and explore the data for yourself. To answer their key research question, the study authors had to figure out:

1. What’s the lithium concentration in fracking wells throughout Pennsylvania, and how does that vary? To answer this question, they examined 595 chemical analysis reports from 515 different fracking wells.

2. How much produced water is made per well, and how does that change over time? They examined data from more than 2,500 wells to answer this question and some statistical modeling to account for the decline in water produced over time.

Here’s a figure showing their specific focus areas:

Key figure (adapted from Figure 1 in the original study) showing the estimated concentration of lithium across the Marcellus Shale. The results showed that lithium levels in Northeast Pennsylvania were a tick larger than those in the Southwest.

Their key result: The average lithium concentration in produced water was 127 and 205 milligrams per liter (mg/L) in Southwest and Northeast Pennsylvania, respectively. If you’re not used to dealing with chemical concentrations, let’s put that result in context: drop a single marble into a 2-liter bottle filled with water. The marble roughly represents the amount of lithium in the produced water analyzed in this study.

The investigators further found that the average annual creation of produced water in the study area was just short of 9 billion liters. Based on the concentration of lithium, coupled with the amount of produced water created, the researchers calculated a likely theoretical recovery amount of lithium of 1,160 metric tons per year. Here’s a figure putting that lithium quantity in context:

This is a pretty staggering and compelling result, but there are a few things to keep in mind:

1. Remember our 2040 lithium demand discussion - we expect demand to increase dramatically over the next 15+ years, so the amount of lithium available recovery would comprise a smaller proportion of total future US demand.

2. This is a theoretical maximum amount. If we put in a process to recover all of the available lithium from all fracking wells in PA, and if recovery efficiency was 100%, we’d yield 1,160 metric tons per year. In reality, we’d expect only a subset of wells to be covered, and recovery efficiencies would likely be less than 100% in those locations.

How Might Lithium Harvesting from Produced Waters Reach a Meaningful Scale?

Regardless of the contextual notes we provided about how the computed lithium available represents a theoretical maximum and that demand for lithium will increase in the future, the finding from this work is compelling and suggests that this pathway may be feasible for lithium extraction in some cases. The researchers point out that the lithium concentration and produced water volumes in some areas are similar to existing production-scale lithium extraction operations (e.g., in Chile). However, they also point out how much of the fracking operations in the Marcellus Shale region have been optimized over time to maximize the reuse of produced water for further fracking - inserting a new process to explicitly extract lithium could create additional cost or logistical burdens.

Current recovery processes for lithium contained in water typically involve evaporating some of the water and then extracting the lithium. Some emerging technologies—yet to be proven at a commercial scale—more directly target lithium recovery, and dozens of projects are demonstrating the technology and creating critical information like operational efficiency, cost, and more. It remains to be seen whether technological development can overcome some existing barriers to recovering lithium from previously underexplored sources like produced water from fracking.

Final Insights and Takeaways

Here are a couple of concluding insights and takeaways to consider:

1. Value of Open Data. The research highlighted here is yet another example of an unexpected positive benefit from open data systems. Do you think Pennsylvania regulators required oil and gas production operations to monitor chemicals in produced water because they thought there was a trove of valuable stuff in produced water? Highly doubtful. However, requiring the analytical data to be open and available to the public allowed this research team to draw some fascinating and industrially important conclusions. There are many other examples (e.g., I’ve published numerous research papers that leveraged expansive data sets made publicly available because of the US Greenhouse Gas Reporting Program). So, an action for readers who are in corporate sustainability or are active researchers could be: what problems you’re trying to solve now might be helped by harvesting existing open data sources today?

2. New Clean Energy Incentives as Drivers for Innovation. The core of the work highlighted here falls under the broad umbrella of “beneficial use,” which is a term often used to describe finding a productive outlet for something that may otherwise be considered a waste product (other examples include using crushed recycled glass or ground-up shingles in asphalt mixtures, using combustion ashes in concrete, and many more). The influx of US government support and incentives for clean energy, coupled with the urgency to decarbonize, is creating new opportunities in unexpected areas and ways. Time will tell if the promise of meeting some of our growing lithium demand will be met via fracking water recovery.

3. Results in Context - Fracking to Get Lithium. I’ve deliberately avoided a question you may be asking: “Wait, is there an implicit suggestion that we should be fracking so that we may recover the lithium needed to meet demand?” It’s anybody’s guess how the energy mix will look in the US in the future, but projections suggest fracking will be a major part of the energy mix for several years, although its proportion has declined. I think one way to look at it is that - if lithium recovery technologies prove successful, are economically competitive, do not create severely negative environmental consequences, etc. - then produced water from fracking may represent a compelling new lithium source. But keep in mind that it’s not the only new source beyond mining. Another encouraging note (again, spurred in part by several of the US federal bills that support the clean energy transition) is that major investment is also happening in battery recycling technologies that can use or otherwise harvest the lithium and other valuable minerals after a battery reaches the end of its initial useful life. A nice outcome may be that lithium extraction from produced water, along with other novel technologies and sources, creates a bridge until we have enough lithium in products that are then effectively recovered and reused through battery recycling technologies.

Do you know anyone who may enjoy this post? If viewing this in your email client, share this post with a friend by clicking below!

If you’re viewing this online, simply copy this link and email or post it to those who may enjoy the newsletter. Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

Gratitude, Preview, Roundup, and Some Resources

We are on vacation this week, so the format of this post is a bit different. We’ll return next week with a fresh new post, but this week we’ll start with some gratitude, a preview of next week’s newsletter, a roundup of our posts so far, and a couple of resources you may find useful. See you next week!

First, Some Gratitude 🙏 

Thank You to the hundreds of folks who have subscribed to this newsletter since we launched last month! I started this newsletter with a simple aim: to write deeply about new sustainability research or developments that are interesting to me. It’s immensely gratifying to see these topics resonate with so many of you.

I’m likewise grateful for the many words of support, encouragement, surprise, criticism, etc. that I’ve received since we launched.

Sneak Preview of Next Week’s Post

I’m so excited to publish next week’s newsletter - a new study in the journal Scientific Reports found that there’s enough lithium (a key ingredient in EV batteries) in so-called produced water (water that comes out alongside gas and oil in hydraulic fracturing operations) from Pennsylvania to satisfy around 40% of the total US annual lithium demand. But could this future actually be realized 🤔?

The authors leveraged a state of Pennsylvania fracking data set with which I’m deeply familiar, having dug into it as part of one of my doctoral dissertation chapters on non-hazardous industrial wastes. The new study raises all kinds of fascinating questions about how to best handle waste products from large-scale extractive operations like oil and gas exploration, addressing challenges with the supply of critical minerals, transforming supply chains toward more domestic production, technology readiness and trade-offs, and a lot more. I can’t wait to share this next one with you!

Sustainability at the Frontier - Post Roundup

For those of you who recently joined us, we’re taking a page out of the TV show playbook by packaging together our previous work to catch you up (and for those of us who have been here since the very beginning, a chance to reminisce…). As a bonus, I’m including a 1-sentence TL;DR:

• 👩‍💻 Post 1: How do ESG Software Ratings Work? Software companies go through a competitive request for proposal-like process by answering lengthy questionnaires and giving a live demo of the software, and rating companies use a scoring rubric to place all companies against one another, usually on a 4-quadrant matrix.

• ♻️ Post 2: Are Plastic Bags Recycled After You Drop Them Off? Yes, for the most part, and life-cycle comparisons of current recycling fates show superior outcomes compared to disposal.

• 🛰️ Post 3: Can the Eye in the Sky Help to Reduce Greenhouse Gas Emissions? Yes, new aerial and satellite tech solves a longstanding challenge of measuring methane levels from waste sites “on the ground” and will likely spur new operational modes and technology advances that reduce global emissions in the waste sector.

• 💰️ Post 4: Do Banking Financial Models Disincentivize Investment in Low-Carbon Projects? Yes <based on data from many EU banks>, but new methods of modeling the risk of new “green” investments, coupled with more historical evidence of the performance of renewable energy and related technologies, should help to reduce the perverse incentives currently in place for financial institutions when assessing high-carbon and low-carbon infrastructure and technology investments.

• 🚗 Post 5: Do Electric Cars Depreciate Faster than Gas-Powered Cars? Yes, but an improved set of features in newer electric vehicles (like a longer range) is already closing the gap and will likely continue to do so in the near future, thus unlocking some of the more promising aspects that EVs can deliver when considering transportation sector decarbonization.

• 🔥 Post 6: How Does Controlled Burning Lessen the Effects of Wildfires? Mainly, in combination with other manual and mechanical methods, by reducing the fuel available in forest lands if/when a wildfire occurs, but there is a point of diminishing returns from a public health perspective whereby too much controlled burning can create smoke that negatively impacts nearby populations.

Thanks to Troy McClure for hosting my equivalent of a TV retrospective clip show <GIF courtesy Giphy>

A Couple of Assorted Links and Resources to Close This Newsletter Out

1. Climate Tech List. Shout-out to Sustainability at the Frontier reader Steven Zhang, who built one of the most helpful job boards for those looking to find their first or next role doing sustainability-related work - there are about 30,000 positions listed from more than 5,000 companies. I usually counsel at least a few people a week on either getting their first job in climate or otherwise advancing their career, and Climate Tech List is something I’ve probably recommended a couple hundred times.

2. Cal Newport’s Podcast, Deep Questions. Cal is a computer science professor at Georgetown, and in his podcast, he puts out a lot of great advice on minimizing digital distractions and cultivating an environment that allows you to do deeper work. I recommend his podcast, it’s one of the few I’m able to put in a regular rotation. Some rearrangements I’ve made to my ways of working in recent months (allowing me to make room for things like this newsletter!) are thanks in part to some of his suggestions.

How Does Controlled Burning Lessen the Effects of Wildfires?

If you read nothing else in this post:

Large wildfires destroy vast amounts of nature and property and negatively impact public health and the environment. Climate change and other factors have increased the frequency and severity of wildfires in recent years, so forest management planners are implementing new or modified fire prevention and management practices to mitigate the effects of future wildfires, which is a shift from historical practices that almost entirely focused on suppressing fires after they started. Controlled burning, where forest managers deliberately introduce small fires to reduce the amount of fuel available to a future wildfire, is becoming an increasingly common practice. But from a public health perspective, there is a question of “How much controlled burning is too much?”. Recent research found a “goldilocks” approach whereby a moderate amount of controlled burning can strike a balance between reducing the size, severity, and duration of large wildfires while not creating a public health burden from exposure to too much smoke is an advisable path forward.

The US Wildfire Problem in Numbers and Context

Anyone who has experienced the effects of wildfires knows their effects can range from devastating (complete loss of your home, your town, your region) to inconvenient (hazy conditions, having to stay inside during periods of smoke). The scientific literature generally finds all sorts of associations between smoke exposure and a series of respiratory issues (e.g., making asthma worse, causing temporary hospitalizations, and even death).

What are the various causes of wildfires? Let’s turn to the US Forest Service, which has 13 unique fire cause codes:

Wildfires have a range of causes. I love the specific call-out of Children as a unique cause, and I can’t help but think about the date April 26, 1992, when I see Cause 12.

What are the biggest causes of wildfires among these 13 possibilities? The answer is a little murky. You may be surprised to know there’s enormous disagreement among multiple US federal services who have some role in understanding, tracking, measuring, and reporting on the seemingly fundamental question of “how many fires get started and what was the cause?”. Want to see a crazy - but seemingly representative example? Look at this figure below, taken from a US Forest Service paper:

This figure summarizes the number of wildfires reported in Texas between 1992 and 2011 across three different sources. Zoom in on 2005 through 2011, and you can see that the number of wildfires reported differs by several thousand in a given year!

Some government agencies suggest that more than 80% of all wildfires are caused by humans in some way - again, as we can see, there is wild disagreement even in how many fires occur, so I’m not sure how much we can take that 80% estimate to the bank. But there does seem to be general agreement that, for the most part, humans do cause most wildfires, which probably led to the investment and creation of this guy:

I initially mistakenly thought “Woodsy” was the character who told us to prevent forest fires, but I quickly realized that it’s Smokey. Woodsy of course famously suggested that we “Give a Hoot/Don’t Pollute” (source: GIPHY)

How big is the wildfire issue in the US? Data from the National Centers for Environmental Information tell us that, on average, around 70,000 fires occur annually nationally, burning 7 million acres of land. The figure below shows recent data from 2012 through 2022.

If you’re not used to thinking in acres, let’s put that 7 million acres burned annually in context. One of my favorite bits of trivia is that Jacksonville, Florida is the largest city by land area in the US at about 560,000 acres. So every year, about 13 Jacksonvilles are burned in the US from wildfire.

As a final point, going beyond the dimensions of estimated damage caused by wildfires, let’s look at the economic losses linked to wildfires. A recent estimate by a US congressional committee called JEC put the annual financial loss from wildfires at $894 billion, and the figure below represents the estimated annual cost across nine main categories. This figure is at the upper end of their estimated range. The absolute magnitude of economic loss should probably be taken with some grains of salt - as we established earlier, there’s wide disagreement in the most basic measures of wildfires, so we’d also expect financial loss estimates to have a similarly large range. Regardless of the specific magnitude, the physical and economic damage from wildfires is pretty substantial.

3 Main Methods to Reduce the Frequency and Severity of Wildfires

Active forest management seeks to reduce the frequency and severity of wildfires. Three methods are mainly used to accomplish this: Hand Thinning, Mechanical Thinning, and Controlled Burning.

Hand Thinning

Hand thinning is mainly done to remove smaller-diameter trees. These smaller trees are considered “ladder fuels,” meaning they can enable fires burning closer to the ground to migrate upward.

Hand thinning removes smaller-diameter trees by hand using equipment like chainsaws. Image courtesy jessicaplance.com.

Mechanical Thinning

Mechanical thinning goes harder than hand thinning, employing equipment like large cranes, bulldozers, wood chippers, and the (incredibly named!) feller buncher which literally translates to “cut down tree then gather it.”

Mechanical forest thinning leverages heavy-duty equipment like a Feller Buncher, which rips and gathers many large trees and vegetation at once. The base equipment photo is courtesy of John Deere.

Controlled Burning

Controlled burning is a process intended to mirror naturally-occurring fires, and the process to plan and implement a controlled burn is fairly involved (here for your reading pleasure is a 115-page controlled burning manual). Did you know one of the jobs involved in carrying out a controlled burn is called a Prescribed Fire Burn Boss?

Each of the three types of interventions has its part in reducing the severity of major wildfires and is often used in tandem (e.g., thinning may be carried out before starting a controlled burn campaign).

Just an incredible photo of the controlled burning process. Photo courtesy of the Nature Conservancy and Jason Houston.

New Study: How Much Controlled Burning is Too Much?

Now that we have established how frequently wildfires occur, tallied up the physical and economic damage they can wreak, and reviewed interventions used to reduce the frequency and severity of wildfires, let’s turn our focus to a new study that asks a question that I’ll sum up as follows: “Wildfires cause lots of physical damage, economic damage, and harm to public health. Controlled burns can reduce the severity of wildfires, but this type of burning also creates smoke emissions, which harms public health. Is there a balancing point where we can effectively reduce the occurrence and damage from wildfires using controlled burns but not doing so much controlled burning that we create a new public health problem from smoke emissions?”

This type of study falls into a particular archetype common in the sciences that I’ll call “Model Development and Application with a Case Study.” There’s a quote attributed to statistician George Box that “All models are wrong; some are useful.” I tend to agree, but the fact that models that scientists use to understand the world are generally wrong does not mean they lack value. Anyone who has taken high school chemistry knows the ideal gas law that describes the relationship between pressure, volume, and temperature for a gas (PV = nRT). It turns out that this model is wrong, and there are several edge cases and other factors not accounted for in this equation - but it gets you pretty close to reality and is relatively easy to use. So it’s a helpful model.

Lots of researchers develop new models in their work as a way of making sense of complex phenomena. Models may manifest as various analytical or numerical equations or simple (or complex) computer models that link relationships across scientific domains. We’ve previously talked about life-cycle analysis as a helpful modeling framework to account for the emissions and environmental impacts of a product or process across its life cycle. Once a scientific model is developed, researchers will commonly apply it to a real-world case study to demonstrate its usefulness (and perhaps accuracy, robustness, etc.)

In this new study, the researchers highlight a historical disconnect between forest managers (who are charged with implementing preventive measures to avoid wildfires and work with coordinated teams to address wildfires when they occur) and public health officials (who are charged with monitoring and implementing guidance to protect public health including during natural disasters like wildfires). Namely, that controlled burning is a commonly-used intervention to prevent or reduce impacts from wildfires, but the effects of controlled burning are underexplored. So the researchers built a new model to bridge this gap and understand how varying the amount of controlled burning reduced wildfires, but then, consequently, how much the additional smoke from controlled burning could affect public health. They then apply their model to a fire-prone area in the Tahoe-Central Sierra Initiative region in California and test its accuracy and results against nearly 20 years of historical data.

What’s clever about this new study is how it incorporates multiple scientific disciplines, techniques, and data sets to make sense of their study region toward answering their central research question. Here’s how they did it:

• Generate Forest Management Scenarios. They worked with forest managers to create realistic scenarios of more and more aggressive wildfire prevention techniques, including a “do nothing” case, a “business as usual” (BAU) case with mostly hand and mechanical thinning, and then four scenarios of more and more controlled burning.

• Model the smoke emissions that occur. For each scenario, the researchers estimated the amount of smoke that would happen.

• Model how much nearby populations are exposed to smoke, based on the smoke emission estimates and modeling atmospheric conditions.

• Model the direct public health effects occurring based on the amount and duration of smoke exposure, along with population demographics.

So what did they find? First, as you might expect, the amount of smoke created is worst in the “do nothing” scenario, which they called Minimal Management, followed by the BAU case. The four controlled burn scenarios resulted in far fewer smoke emissions compared to the minimal management and BAU scenarios. The values in each panel shown below give a nice depiction of how concentrated the smoke gets in the modeled area and how far that smoke emanates.

The authors built a complex, multi-component computer model to understand how various levels of forest management intervention affected the amount of smoke emissions and how far those emissions would travel and applied the model to a region in California. The results show that all four degrees of controlled burning did a better job of reducing smoke compared to the two scenarios where no controlled burning (Image taken from Figure 3a in the original Schollaert et al. paper)

The authors also found in their model that the scenarios with controlled burning tended to reduce the amount of time that wildfire season would last, on the order of about a month’s difference.

To understand the potential public health impacts from the six forest management scenarios described earlier, the researchers examined outcomes in terms of (i) asthma-related hospitalizations and (ii) emergency room (which they called emergency department) visits caused by increased amounts of smoke exposure in the nearby population. Let’s take a look at the hospitalization modeling results:

When modeling the health effects of the six forest management scenarios, the authors found that each of the four controlled burning scenarios resulted in a net reduction of around 30 asthma-related hospitalizations compared to the BAU case. The do-nothing “Minimal Management” case resulted in a roughly additional 60 hospitalizations annually compared to BAU. The difference between each of the four controlled burning scenarios is pretty small. Image adapted from Figure 5 in the original study.

We can see in the above figure that although the four controlled burning scenarios all resulted in a decrease in modeled asthma-related hospitalizations, the magnitude of the difference is pretty small (maybe a difference of around 5 or 10 hospitalizations per year in the study area in California). Still, there’s a notable difference in the “Fire” scenario versus the more aggressive controlled burn scenarios (Fire+ and Fire++), suggesting that a modest amount of controlled burning reflects a “goldilocks” scenario that balances wildfire smoke reduction, wildfire season duration, and public health impacts.

That the magnitude between each scenario is relatively small underscores an important point: more aggressive controlled burning (which takes additional time, financial resources, more fire management personnel, etc) gives diminishing returns from a public health impact, which is an important finding from this work. More broadly, the integrated model developed by the researchers can be generalized and applied to other fire-prone regions of the US, provided the various multidisciplinary data are available at similar levels of accuracy and granularity.

Insights and Implications

Here are some insights and implications of this new study that come to mind:

1. Most corporate sustainability teams that publish their sustainability goals and progress often incorporate so-called physical climate risk models into their planning and reporting, where the effects of things like wildfires and other natural disasters are evaluated, and the company indicates how they plan to address risks from these events to their operations and supply chains. New, integrated models like that discussed here represent an evolution that may bring into sharper focus the magnitude of physical risks from wildfire based on planned wildfire management practices, which, in theory, should improve decision-making processes by corporations when planning risk mitigation measures.

2. The authors’ main goal was to develop and display a new integrated model that could inform public health effects stemming from various realistic wildfire mitigation scenarios and practices. Health officials have already guided the public in minimizing the impacts of wildfire smoke after it happens. Still, with this new work, there is a new tool where public health officials can be brought into the wildfire and smoke prevention process alongside forest managers and explicitly link various wildfire mitigation measures to health outcomes. Naturally, an improved planning process informed by data should result in better health outcomes for the public living near wildfire-prone areas.

Do you know anyone who may enjoy this post? Perhaps a buddy who is a forester, a cousin who’s a public health official, or your long-lost sister who’s an atmospheric modeling fan? If so, use the widget below to share this post with them!

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

If you view this in your email app, simply hit reply to send us your questions, comments, or feedback - we’d love to hear from you.

Do Electric Cars Lose Value Faster than Conventional Cars?

If you read nothing else in this post:

The transportation sector represents around 20% of global greenhouse gas emissions, so decarbonizing transportation is critical to align the world’s emissions to targets that will limit the worst effects of climate change. Best-available research shows that electric vehicles mainly running or entirely on battery technology significantly reduce GHG emissions compared to conventional petroleum-fueled vehicles across the car's life cycle. However, to realize the potential GHG emission savings from a transition to more electric vehicles (EVs) requires (i) more consumers buying new EVs and (ii) a robust used car market for EVs. How well EVs retain their value relative to their original purchase price matters both to buyers of new EVs (who will want to capture value when it’s time to sell their car) and those in the market to purchase a used EV (who want a good deal and trust the car will last for some time). Newly-published research examining details of more than 9 million used vehicle listings found that EVs do not hold their value as well as comparable conventional (gasoline and diesel-consuming) vehicles, but this evidence is skewed in part by the newness of EVs and the gap between EVs and conventional vehicles narrowed in recent years as battery technology improved. These findings, coupled with other evidence suggesting modern EVs should have lifespans comparable to that of conventional vehicles - around 15 years - suggest growing consumer acceptance of EVs and a strong signal that the promise of EVs to decarbonize the transportation sector could be realized.

The Growth in the Electric Vehicle Market

Transportation accounts for about 20% of global greenhouse gas (GHG) emissions (see the figure below). Because of the global push to decarbonize the world in accordance with science-based targets to mitigate the negative effects of climate change, electrifying the transportation sector has been seen as an important movement to reduce reliance on vehicles powered by gasoline or diesel fuel.

One of our favorite graphs at Sustainability at the Frontier is courtesy of Climate Watch and Our World in Data. It shows that Transport comprises about 20% of the total global GHG emissions.

You might be thinking, “Wait, I thought electric cars have been around for more than 100 years—is the introduction of electric vehicles really news?” and you’d be partly right. Although the first vehicles powered by electricity were invented in the early 1830s, serious uptake of electric cars has only occurred in the past 15 or so years, owing mainly to advancements in battery technology and the deployment of charging infrastructure to serve the masses.

The first EVs appeared in the 1830s, and according to Car and Driver, mid-1800s prototypes were destroyed by railroad workers concerned about the effect that EV growth might have on their job security. Image courtesy Getty via Car & Driver.

Before going further, let’s establish some things about electric vehicles. First, here are some definitions that we’ll use in this post:

1. Conventional Vehicle (CV): A gasoline or diesel-powered car.

2. Hybrid Electric Vehicle (HEV): A car with a small battery that nearly completely runs on gasoline or diesel fuel. The battery alone can only power the car for a mile or so.

3. Plug-In Hybrid Electric Vehicle (PHEV): A car with a larger battery (relative to HEV) that runs short (10-50 miles) distances on battery, followed by gasoline or diesel fuel when the battery charge runs out.

4. Battery Electric Vehicle (BEV): A car that only runs on electricity from batteries.

We see a spectrum of options, with batteries providing anywhere from no energy to power a vehicle (CV) to all of it (BEV), with some partial in-between flavors. Automobile manufacturers have begun developing more production capacity for some flavors of EV—the number of EVs on the road in 2023 was six times greater than in 2018, according to the International Energy Agency.

We will probably see more EVs than other high-concept cars like “The Homer: The Car Built for Homer.” However, The Homer’s $82k price point (1991, or $192k in today’s dollars) resembles some higher-end EVs.

EV sales rapidly accelerated around 2018, primarily because of the scale-up in Tesla Model 3 availability. The figure below shows US EV and PHEV sales. We saw a slight bump in PHEV sales around 2017, but absolute sales represented less than half of BEVs in 2018 and 2019.

Comparing Relative Emissions of CVs and EVs

Let’s now explore the question of how EVs compare to CVs on a greenhouse gas (GHG) emissions basis. The figure below compares averaged data of the four primary vehicle archetypes based on an assumed number of miles driven in a given year (around 11,000). The BEV emissions are far less than the others, and the PHEV emissions are pretty good compared to the HEV and CV.

BEVs have far lower carbon dioxide emissions than PHEVs, HEVs, and CVs. The data here assumes an average annual distance traveled of 11,579 miles and the US national average electricity grid mix. Areas with higher-than-average renewables in the grid will show BEVs and PHEVs outperforming others, while electricity grids with a lower proportion of renewables will narrow the gap between electric and gasoline vehicles.

But carbon dioxide emissions during a car’s active life represent only one piece of the GHG puzzle. Going a level deeper, we can consider the emissions of a given type of car by looking at its entire life cycle. So-called life-cycle analysis (LCA) is a simple-to-understand (but often hard-to-execute) methodological tool commonly used in various sustainability-related disciplines. Part of the LCA process involves creating an inventory along the multiple discrete steps of a process or a product (e.g., cars), summing up the GHG emissions and other environmental metrics along the way. Although in most cases the bulk of all GHG emitted from a car comes from the use phase, or when it’s being driven, an LCA lens is more holistic so we can account for everything.

Let’s take a look at the results of a recently published comparison of the GHG emissions profile of various types of cars that considered emissions from vehicle production, use, and end-of-life disposition:

BEV and PHEV have far fewer carbon dioxide emissions when you consider the entire life cycle of a car, from production to use to final disposition. These results are for mid-sized vehicles, but the trend generally holds across multiple vehicle types. As you can imagine, many factors can influence a vehicle’s efficiency (e.g., the car's weight, the available motor power (which can affect how fast one might drive), etc. I’m deeply sorry that my graphing software prevents me from making the “2” a subscript in “CO2”.

OK, even when we account for the full life cycle of various types of vehicles, we can still see that BEVs and PHEVs outperform CVs that run on gasoline or diesel in terms of total GHG emissions. Although data in the above figure is for passenger vehicles (i.e., does not include things like semi trucks, buses, etc.), the trend of life-cycle performance of EVs generally holds across vehicle types, with some exceptions.

New Study: How Quickly Do EVs Depreciate Compared to CVs?

Now that we have established that (i) the transportation sector is one of the most important to decarbonize globally and (ii) most modern BEVs and PHEVs show a drastic decline in life-cycle GHG emissions and therefore, BEV and PHEV adoption is critical to decarbonizing the transportation sector, let’s dig into the topic of broader EV adoption by examining some compelling data from a new study published by Roberson and co-authors in the highly-cited open access journal Environmental Research Letters, who examined resale value trends for millions of used EV and CV vehicle listings over a multi-year period.

Why is examining the resale market of EVs interesting? First, let’s first put the current market penetration of EVs in context. We showed earlier how EV sales have rapidly accelerated in recent years, but today, EVs only comprise around 5 or 6% of vehicles driven in the US. Several EU countries have far greater EV penetration (here’s a good WRI article detailing other country-level estimates). Some estimates suggest that EVs need to comprise a market share of at least 75% by 2030 for the transportation sector to align with science-based GHG emission reduction targets associated with climate change's worst potential effects. Thus, for the potential of greater GHG emissions reduction from EV adoption, there needs to be greater sales of new EVs, and used EVs must have a viable and robust resale market. If EVs are not retaining their value or other issues are preventing EV uptake on the secondary market, then the climate benefits of EVs will not come close to being fully realized. Also, suppose EVs do not hold their value well. That will influence the purchase rates of new EVs because owners will be concerned about their ability to sell the car reasonably whenever they decide to move on to a different vehicle.

A few previously published research articles showed that EVs do not retain value as well as CV counterparts. But past studies were limited by fairly small data sets - for example, previously-published work on the topic looked at anywhere from 72 used car listings to almost 600,000. In this new paper, the authors pulled data from more than 9 million used vehicle listings from more than 60,000 sellers or dealerships to examine how resale value across various car models compares between CV, HEV, PHEV, and BEV.

The figure below shows one of the key takeaways - we can see in the left panel that the vehicle value retention rate - which is the used car sales price divided by its new manufacturer’s suggested retail price (MSRP) when that vehicle was new based on the model, trim, etc. - is nearly identical between CV and hybrid cars, with almost perfect overlap (the dark line in each graph represents the median value of the vehicle value retention rate, and the color-coded shaded boundaries represent the 25th and 75% percentile, respectively.

This figure is from the Roberson et al. paper. We can see that Tesla has a high retention rate if the car sells within the first couple of years of purchase but then steeply declines around year 3. Interestingly, PHEV and non-Tesla BEVs have a similar trend to CV, whereby value is highly retained within the first couple years of purchase, followed by a steep decline and an overall depressed retained value relative to conventional vehicles. Follow a similar trend. The solid line in each panel represents the median retention rate. The retention rate is calculated as the used car sales price ratio divided by its original MSRP.

The middle panel in the figure shows us that newer-vintage PHEVs retain their value better than CVs if the model is less than a couple of years old, but then the value retention drops for vehicles over 2.5 years old. The authors displayed value retention in the right panel for Tesla and non-Tesla BEVs. We can see that the median value retained by Tesla performs well compared to CV for the first few years of the vehicle’s life, followed by a bit of convergence around the CV value retention starting around year 5. However, the non-Tesla BEV value retention at nearly all points fares poorly compared to CV, except for a roughly new BEV.

As you’d expect, the binary question of “Is this an EV or not?” only has a partial bearing on the actual retained value of a given vehicle. The authors looked at a bunch of other dimensions in their analysis to understand the extent to which the following factors seemed to influence the value retention of a given vehicle across multiple model years:

1. Vehicle mileage

2. Time the car is on the market

3. Operating cost

4. Amount of available driving range remaining

5. Whether or not the original EV owner received a federal subsidy

One consistent result is that a larger number of driven miles negatively affected the car’s value retention across all vehicle types—roughly a 5% decline for every 10,000 miles driven. Also, somewhat expectedly, the amount of available driving range for non-Tesla BEVs had a hugely positive effect (+5.6% for every 10 miles of additional available range) on the value retention rate, far greater than that for Teslas (+1.6%).

Another compelling observation was the effect of a US Federal tax subsidy that’s been in place for several years on value retention. In brief, you can claim up to a one-time $7,500 tax credit for purchasing a designated “clean energy vehicle.” The authors found that cars eligible for the $7,500 tax credit tended to have a lower value retention rate, likely in part because of how they computed the retention rate (which didn’t adjust MSRP to reflect tax subsidies) and that a seller may factor into their sale price the fact that they received a large tax credit when they originally purchased the vehicle, which could be for a couple of reasons.

OK, one of this study’s key results is that BEVs and PHEVs do not retain value as well as CVs, at least for the model years examined. Does this spell doom for the potential uptake of EVs going forward? Not so fast.

Another key observation the authors teased out of their massive data set is how the retention value for most non-Tesla BEVs has dramatically improved in recent years. We can see in the figure below that four different popular BEV makes and models experienced a substantial uptick in retention rate from 2014 to 2018, which the study authors largely attribute to the improved total driving range. This trend does not hold for Tesla’s flagship mass-market Model S. The authors attribute the Model S trend in part to a “diminishing returns” idea for driving range and that there’s probably some midway point that the average consumer finds desirable - and in fact if we eyeball the figure below, we can see that the trendlines would probably converge around 200 miles of driving range, which is (probably not coincidentally) seemingly the value that most new EVs ship with these days.

This figure (Figure 3 in the original study) shows that steady improvements in battery efficiency - and, therefore, the total amount of range one can drive before charging the vehicle again - dramatically positively affect the value retention rate for several popular models of BEV. The trend doesn’t hold for the Tesla Model S, though probably.

So, I think the best way we can draw the main conclusion is that while CVs do not lose resale value as quickly as BEVs and PHEVs, the value retention of BEVs and PHEVs is getting better and will likely be unlocked by future models having a “good enough” range for the average consumer. Let’s also consider a related factor likely influencing value retention, and that is the useful life of the vehicle.

The NHTSA estimates that the average useful life for a passenger CV is around 15-20 years. So if we consider a consumer purchasing a new car, they’ll be “locked in” to the higher GHG emissions profile of that CV for its useful life, and that occurs whether the original owner keeps the car for its entire life or whether they sell it to someone else.

Unfortunately, we do not have great longitudinal data indicating how long the average EV might last because EVs are relatively new. However, we have some promising evidence suggesting service lifetime may be similar to CVs. Telematics company Geotab tracked driving data for 21 different EV models spanning more than 6,300 individual cars, finding that the battery life degraded on average about 2.3% annually (see figure below). These battery data may be a valuable proxy to understand the lifespan of EVs, as the battery is considered the most important and likely most expensive piece of the equation for how long an EV will last (or not). We can see the averaged data across all vehicles goes out around 7 years, which is far less than the 15-20 CV life, but this limitation is a reflection for how long EVs could have been tracked. There are tons of Youtube videos showing older-vintage Teslas with hundreds of thousands of miles - time will tell to see how other BEV and PHEVs stand the test of time, but early returns are positive.

The range of EV and PHEV batteries degrades on average by 2.3% annually, according to an analysis by telematics company Geotab based on data from 21 BEV and PHEV models covering more than 6,300 vehicles.

The study’s key takeaway may be interpreted in a couple of ways depending on whether you put yourself in the shoes of the used EV seller or the used EV buyer. For now, lower value retention of EVs may suggest a more significant secondary market for used EVs, which would be a bonus for the buying public but perhaps less attractive to the seller. The authors did not explore the future effect of this dynamic. Still, I sense that the value retention decline will likely slow with improved EV offerings from more manufacturers, so I don’t see the potential adverse effect on EV sellers being a bit limited to larger-scale EV adoption. Further, if we consider previous work finding EVs tend to have lower maintenance costs than CVs, along with declining costs for new EVs that are generally better-performing than earlier models, I think we’ll continue to see a healthy and rapid expansion of not only new EV sales but a resultantly strong market for buyers of used EVs as well.

Insights and Implications

What have we learned today? There are several insights and implications that are relevant not only to consumers in general but also to those in corporate sustainability roles looking at various ways to decarbonize their transportation footprint.

1. The transportation sector is a significant contributor to global GHG emissions. The peer-reviewed scientific literature is clear that the emissions profile (both in the use phase and across the whole life cycle of the vehicle) for BEV and PHEV passenger vehicles significantly outperforms HEV and CV when considering use-phase GHG emissions and life-cycle GHG emissions.

2. A healthy resale market for EVs, which enables EVs to remain in service longer and, therefore, helps avoid the purchase or use of a new or used CV, is critical to meeting global decarbonization goals in the transportation sector. The best available evidence suggests EVs may have a useful life comparable to a CV, which is a vital part of the transition from CVs to EVs.

3. The retained value of EVs in the secondary market is a useful proxy for understanding the health of the resale market. Generally, CVs retain their value better than BEVs and PHEVs, but the gap between EVs and CVs has narrowed recently, mainly because of improvements in battery range. Adding more battery range tends to diminish value retention at around 200 miles.

4. In the near term, the used EV market could offer some compelling buying opportunities given its relatively rapid decline in value compared to CVs, but if current trends hold, this opportunity will decline in the future. Federal subsidies will likely continue to be depressing and affect EV value retention, but that will likewise decline as EV sales grow and subsidies run out.

Do you know anyone who may enjoy this post? If so, use the widget below to share:

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

If you view this in your email app, hit reply to send us your questions, comments, or feedback - we’d love to hear from you.

Do Bank Financial Models Disincentivize Investment in Low-Carbon Projects?

If You Read Nothing Else in this Post: 

The financial sector is key in the global push to decarbonize various industry sectors in line with science-based carbon emission reduction targets. Decarbonization could occur when banks transition investments from high-carbon industries (like oil and gas production) to low-carbon investments (like renewable energy). A new study on lending practices for 59 European Union banks found that this transition to low-carbon assets could cause pain to banks’ bottom line, with researchers computing an average 15% 5-year profit loss across the banks they analyzed. The profit loss results from risk models used by banks that rely on historical data and classical financial metrics that do not account for potential future climate policy shifts (e.g., a price on carbon emissions). So unless fundamental changes in lending risk models incorporate these potential future changes, banks and bankers will be left to decide: invest in high-carbon assets and reap higher profits, or transition investments into low-carbon assets and take a cut on my bottom line. Which would you choose?

What’s the Role of Financial Institutions on Greenhouse Gas Emissions?

We’ve discussed sectoral contributions to global greenhouse gas (GHG) emissions. Here’s a reminder:

Contribution of ten critical sectors to global GHG emissions from 1990 to 2020. We don’t see the financial sector represented explicitly.

This post focuses on the key role that the financial sector plays in the movement to a lower-carbon economy in line with science-based targets established to avoid the worst effects of climate change. But you might be wondering—the financial sector isn’t even displayed in the sector-based GHG emissions shown in the figure above, so why is this sector worth exploring?

It turns out that the emissions funded by the finance sector are far greater than the actual, direct GHG emissions caused by all the world’s financial institutions. One recent estimate by the nonprofit Carbon Disclosure Project (CDP), which receives annual self-reported GHG emissions data from more than 20,000 companies and institutions, says the emissions created from projects and sectors financed by banks and other institutions are more than 700 times larger than their direct emissions. Further, the report suggested that banks and financial institutions are not adequately considering potential climate risks in their lending and investment activities. Finally, the report indicated that three-quarters of financial institutions responding to a survey saw major financial opportunities in making low-carbon investments. This story sounds simple - financial institutions’ investments are fueling many of the world’s GHG emissions. Still, these institutions see significant financial upside in making “greener” low-carbon investments, so all institutions should probably transition to low-carbon investments, right?

The story is a bit more complicated. We will highlight the results of a new study in Nature Climate Change (one of the most highly rated journals in the climate and environmental science world) that digs deeper into some practical realities of how banks specifically incorporate potential lending risk into their financial models when making decisions. Their findings provide an eye-opening perspective on how current bank lending constructs may create perverse incentives to continue investing in high-carbon investments.

A Focus on Incorporating Risks in the Investment Decision-Making Process

The following phrases should be familiar to anyone who has ever invested:

• All investments carry risk

• Past performance is no indicator of future results

Financial institutions use various quantitative and qualitative techniques to gauge the attractiveness and risk of a given investment. Likewise, they can use various techniques to de-risk a given investment, thereby increasing the odds that the investment will not only be paid back in full but also achieve the returns it is aiming for.

As an example, when I co-led circular economy investment funds at Closed Loop Partners, part of my initial scouting work involved evaluating many potential investments against a series of metrics we developed to indicate whether the shape of the investment aligned with our investment goals and strategies, which in part included assessing the risk of a given investment. For example, we’d determine a project’s execution risk by drawing on our expertise in the waste and recycling industry. A project at the safe end of the spectrum might be “lending $2 million to a municipality to fully fund some new recycling collection vehicles that will collect recyclables from newly-populated areas of the municipality”. A project at the other, far riskier end of the spectrum might be “lending $2 million toward an overall $50 million project for a new advanced recycling technology that’s been proven at a pilot scale but never demonstrated at a larger, commercial scale”. You might “price in” a risker project into your loan by giving yourself more favorable terms like a higher interest rate, a better position in the loan repayment stack, etc.

New Study on EU Banks and Lending for Low-Carbon and High-Carbon Investments

With some basics of lending covered, let’s dig more into the new study by Matteo Gasparini and the co-authors we referenced above. The authors explored model-based lending risk regulations that exist in the European Union. These regulations go something like this.

1. Many banks failing would be bad for the economy.

2. Financial regulations can foster stability in the banking system by requiring banks to implement certain requirements when they underwrite a loan.

3. One of these requirements is for banks to estimate potential losses from loans they make and then put aside additional capital based on the risk tied to those expected losses. This is a concept called a Loan Loss Reserve (LLR). In accounting, LLRs are considered liabilities.

In this study, the authors explored the question, “Do financial regulations requiring things like LLR systematically result in investments in high-carbon sectors being considered lower or higher risk than low-carbon sectors?” The authors used a novel approach to answer this question by first classifying a significant number of investments made by 59 European Union banks as “high-carbon” or “low-carbon,” then estimating how banks priced in risk using regulatory-defined risk models, then running sensitivity analyses to check that their answers were robust by considering other influential factors. Let’s take a look at their crucial takeaway through the below figure, which looks at a metric called the provision coverage ratio (PCR) for the 59 banks as a function of the total loan size for each bank, represented by four quartiles (the smallest loan size bucket is 0-25 (or first quartile) and the largest is 75-100 (fourth quartile)):

The study analyzed data from 59 European Banks and found that regardless of the amount of all loans outstanding for a given bank, the amount the banks have in reserve in case of a loan default (called loan loss reserves) is far greater for investments in low-carbon projects compared to high-carbon projects. Banks may have less incentive to invest in low-carbon projects because a greater provision coverage ratio (PCR, the ratio of loan reserve divided by the loan amount) means lower profits for the bank and individual bankers.

What is the PCR? Mathematically, it’s the ratio of the LLR for each loan divided by the total amount of the loan outstanding. A low PCR means the priced-in risk is lower compared to a higher PCR. Put another way, a lower PCR means the bank thinks it’s less likely a loan will default (i.e., not be repaid) compared to a higher PCR. We can see that regardless of the total amount of loans held by the banks analyzed, the PCR is greater for the low-carbon investments compared to the high-carbon investments. In other words, low-carbon investments are considered riskier. Notably, the authors found this result consistent regardless of the loan size (remember, the loan size is the denominator in the PCR calculation).

Traditional lending models may be inadvertently (negatively) impacting low-carbon projects by inflating the calculated risk of these loans defaulting relative to high-carbon investments. GIF by Sustainability at the Frontier via yarn.co.

Why is this result significant? A given bank loan with a high LLR will be less profitable than one with a low LLR. Here is a simple illustration. Let’s say that a bank has a total revenue of $1 billion, and its total expenses (ignoring any loan loss provisions) are $700 million. Their net income would then be $300 million. However, if the bank increased its loan loss reserves - which, as this study points out, is currently the case for low-carbon investments based on baked-in accounting rules and risk models - by $50 million, that extra loan loss provision would increase its expenses by that amount, resulting in a net income of $250 million.

There’s no explicit regulatory definition of a “high carbon” or “low carbon” investment, so you might think that the authors’ results might reflect how they chose to define high- and low-carbon investments. It turns out that isn’t the case - they modified their initial classifications and found that the effect on the magnitude and direction of the results was minimal.

So then the question is: if current regulatory models used to estimate risk suggest low-carbon investments may result in lower profits than high-carbon investments, how much might profitability be negatively affected if banks fully transitioned to low-carbon investments? The authors answered this question by analyzing five years of historical profit data for a subset of EU banks where such data (along with LLR information) were available, and the results are shown here:

Most EU banks would suffer substantial profit losses by divesting from high-carbon projects and reinvesting all funds into low-carbon projects. The profit loss mainly stems from the fact that having to hold more significant loan loss reserves increases a bank’s expenses on its income statement and, therefore, decreases profitability. This figure has my annotations on Figure 3 from the original Gasparini et al. paper.

We can see that most individual banks analyzed would have had a decreased (and, in several cases, substantially decreased) profit had they been invested only in low-carbon assets, with the average cumulative profit loss of 15% across the whole data set.

Why Do Financial Regulatory Risk Models Seem to Consider Low-Carbon Investments as Riskier Than High-Carbon Investments?

Now, we need to dig into why low-carbon investments are coming out as far riskier than high-carbon ones. Put simply, banks' accounting rules and risk models rely on risk estimates based on historical data of classical financial metrics and often do not explicitly account for potential future trends. Current model-based regulatory risk computations for a given investment are based on a blend of a given entity’s profitability (e.g., what are their earnings before taxes divided by their total revenue), solvency (e.g., total debt divided by total assets), and liquidity (e.g., the amount of short-term debt carried divided by their working capital). Investigators computed risk ratios for 228 oil and gas firms and compared calculated risk ratios for 235 renewable energy firms based on 11 years of financial data, finding that several traditional financial risk measures were lower for the high-carbon oil and gas sector compared to the low-carbon renewable energy sector (e.g., a solvency measure related to interest expense showed 16% for oil and gas versus 32% for renewable energy, with the lower value equating to lower risk). The authors also explored more than a decade of data for the dozens of EU banks to compute risk another way by looking at the probability of loan default, finding the average probability of loan default for renewables was more than double that of oil and gas investments.

The computed probability of a loan default using traditional, backward-looking financial risk models showed the probability of a loan default for hundreds of renewable energy investments was more than double that of oil and gas investments.

These results confirm that lending risk models that primarily account for historical data result in banks systematically classifying low-carbon investments as riskier than high-carbon investments, illustrating a tension between intentions to decarbonize and the pragmatic goal of achieving profitability (or more profitability). This begs the question: is there a fix for this situation?

It turns out, despite their long-standing use in the financial world of some of the classical financial ratios described earlier to express risk, these traditional measures are likely too simplified, and risk models may be better served by accounting for other factors that enable a more balanced approach to assessing an investment’s risk. For example, classical financial ratios do not contemplate how events in the future (e.g., a policy change related to GHG emissions) could affect the probability of potential losses for a loan. Keep in mind, this factor is crucial because a policy change (for example) requiring lower carbon emissions would be expected to make the high-carbon investment riskier and the low-carbon investment less risky. The authors illustrated this point by examining what happens to the risk profile of the EU banks analyzed if a carbon tax on GHG emissions was $100 per ton. The results were dramatic, with the solvency measure for high-carbon investments jumping from 16% to 46%, and the solvency measure for low-carbon investments remaining static at 32% (the renewables solvency is unchanged because there would be no direct penalty on carbon emissions because direct emissions are zero).

Insights and Implications

So, what can we learn from this study?

1. Emissions from activities financed by banks and other financial institutions are large, so decarbonizing investments from the financial sector is critical to meet global GHG emission reduction goals that will help us avoid the worst effects of climate change.

2. Traditional financial risk models used in EU banking tend to view low-carbon investments as riskier. Although we can’t generalize this study's results to the whole global financial sector, the results are robust given the large number of banks analyzed (59) and the large geographic coverage (14 of the largest countries in the EU).

3. Banks and bankers, therefore, likely have a perverse incentive (i.e., make greater profits) to invest in higher-carbon projects because of the net effect on profits when traditional financial models are used. This may be at odds with banks' intentions or public commitments to decarbonize.

4. If financial risk models contemplate potential or likely future events (e.g., a carbon tax), the risk equation can be flipped dramatically, with low-carbon investments are seen as less risky than high-carbon investments.

Do you know anyone who may enjoy this post? If so, use the widget below to share:

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

If you view this in your email app, simply hit reply to provide us with your questions, comments, or feedback - we’d love to hear from you.

Can the Eye in the Sky Help to Reduce Greenhouse Gas Emissions?

If You Read Nothing Else in this Post: 

Accurately accounting for the various greenhouse gas (GHG) emission sources on Earth is essential to mitigating their contributions to climate change. Satellite and aerial techniques to measure GHG have become less expensive and more accurate and are being deployed more to detect “hotspots” of high GHG and track emissions over time. GHG emissions from the waste sector, represented chiefly by methane created from decomposing materials in landfills, are one of ten critical global sectors targeted for more aggressive emission reductions, and recent studies have focused on measuring these emissions sources with new satellite and aerial technology mainly because of challenges in estimating and measuring waste emissions on the ground. A new study quantified GHG emissions from 250 landfills using planes and remote sensors equipped with methane detection technology, and although the results did not suggest fundamental errors in existing estimates that landfills already report to the US Environmental Protection Agency, the work highlights how aerial and satellite technology can provide more frequent and (potentially) more focused measurement of “problematic” emissions from specific landfills. These measurement technologies will likely be most beneficial in international contexts where lax regulations or a lack of performance data inhibit knowledge of methane emissions from landfills and, therefore, the greatest opportunities for emission reduction.

Quick Primer on GHG Emissions from Waste 🗑️ 

GHG emissions come from many natural and human-caused sources around the world. Recent years have seen new global collaboration (like the Paris Agreement in 2015) to pinpoint the largest emission sources and work to reduce these emissions in line with the scientific consensus of how much carbon should be in the atmosphere to avoid some of the more dramatic, negative consequences of climate change. So, where are GHG emissions coming from? The figure below from Our World in Data is one of my favorites as it cleanly shows GHG emission trends over time, by sector, so you can see both the absolute and relative magnitude of each sector:

Contribution of ten categories of GHG emissions sources. This post focuses on the Waste sector, the third category at the top of this graph.

We can see that the Waste sector is not the most significant contributor to GHG emissions, but I think it’s among the most interesting because of why it happens. When stuff gets “thrown away”, odds are that no matter where you are in the world, this stuff will go to a landfill, to the tune of around 2 billion tons annually. A paper I published as part of my PhD dissertation found that each person in the US contributes about a ton of waste to landfills yearly. Methane (CH₄) is a significant GHG and starts being produced in landfills about a year after waste is put there. Bacterial colonies then form, consuming and digesting the solid material, producing gas. The amount of methane produced in a landfill is most extensive shortly (about a year) after waste is put into the landfill, then declines over time. This shows the typical relationship of methane production at landfills, taken from my book:

When waste is landfilled, it usually produces a large amount of methane right away, but then the amount produced declines and persists for several years.

Waste gets landfilled for both good reasons (e.g., some discarded stuff is nasty and it’s better off being contained in a landfill than littered into the environment) and not-so-good reasons (lots of discarded stuff has value today in recycling commodity markets, to the tune of around $11 billion annually in the US). There are also fascinating “nexus” issues for some wastes, like food, because when food waste is sent to landfills, it usually degrades quickly, thus making it challenging to capture the methane produced. What’s more, a lot of food discarded to landfills could have been consumed by people or animals, part of the staggering statistic that about one-third of all food produced worldwide is wasted.

Measuring and Reducing GHG Emissions from Waste 

Researchers have produced estimates of GHG emissions for major sectors for years, and these estimates have improved over time for different reasons like more and better direct measurements from various sources, improved computing power, new software programs facilitating data collection and analysis, etc.

Directly measuring emissions on the ground from landfills is surprisingly tricky, mainly because landfills cover a reasonably large area (dozens if not hundreds of acres, or hundreds of thousands to millions of square meters for metric system fans), and technology to measure methane over the large area from ground level historically did not perform well and was expensive and clunky to deploy. Thus, emissions have historically been estimated by leveraging equations that tell us how much methane would be produced (based on the type and amount of waste put into the landfill) and then applying some factor that assumes how much is collected (if the landfill has a system to collect the gas, otherwise nearly all of the gas is thought to be emitted into the atmosphere). In the US, nearly a thousand landfills have been required to estimate their emissions this way since 2010 and report annually under the US Environmental Protection Agency’s (US EPA’s) Greenhouse Gas Reporting Program law.

Now that we’ve covered traditional methods of estimating methane emissions from landfills, we’ll briefly cover how GHG emissions from landfills are reduced. Since the mid-1990s, US landfills have been required to collect and destroy collected methane once they reach a specific size under laws called the New Source Performance Standards and Emission Guidelines. These laws require landfills to have methane collection systems designed and installed throughout much (but not all!) of the landfill, and the laws include allowances to let landfills “phase in” new collection infrastructure to give time for new waste to be put in and to enable waste disposal operations to continue. Methane is usually collected by a network of wells (like the one below (source)) that extract the methane and route it to a “destruction device” like a flare or energy conversion system. We’re not really “destroying” methane, per se, but rather converting it to carbon dioxide, which is 25x less potent of a GHG than methane.

Landfill methane is collected using wells like that shown here, drilled into the waste, and then connected to a network where the gas is extracted under a vacuum. The left image shows how the infrastructure looks in real life, and the right image shows what’s happening both on top of and beneath the surface. Once gas is collected, it is routed to a destruction device that converts the methane to carbon dioxide, substantially reducing its GHG profile because methane is about 25x more potent of a GHG than carbon dioxide.

How are landfill gas collection systems operated? Once a gas collection system is installed, it operates continuously. Each collection well is typically adjusted by a human operator at least monthly (in line with the US EPA gas collection regulations), increasing and decreasing how much gas is extracted at each location and checking for any unusual conditions that may affect performance. A design thumb rule is that you have one collection well per acre of landfill surface, so many landfills will have dozens if not hundreds of individual wells that are operating. Depending on the landfill size, collecting a reading from every gas well can take anywhere from one to several days. Yours truly has designed and operated many gas collection systems. Still, it doesn’t take on-the-ground experience to know that if you only check on performance monthly, there will likely be periods when the system is not operating optimally.

Yours truly collecting a landfill gas sample back in the day.

New Study: Using Planes and Remote Sensors to Measure Methane Emissions from Landfills

An interesting new study published earlier in 2024 in the (excellent, very tough to get published in) scientific journal Science reported on new methods for measuring methane emissions from landfills by flying a plane equipped with methane detection technology about 3 to 5 km above around 250 landfills, pinpointing emissions for each landfill down to a resolution of about 3 to 5 m. Here’s an image I adapted from the study that gives a good visual of what they did:

The new study by Cusworth and co-authors used a plane equipped with methane detection equipment and encircled US landfills to estimate emissions at a high resolution.

Over the 6-year study period, the authors quantified methane emissions at 250 municipal solid waste landfills in 18 US states, aiming to address these two key questions:

1. Do landfills show so-called “point source” emission behavior, meaning are they showing spikes of large amounts of methane being emitted? And if so, are these significant emission events episodic, or do they occur frequently?

2. Do emission estimates using this new method compare favorably to the estimation methods landfills already use when reporting their emissions to the US EPA Greenhouse Gas Reporting Program?

Let's first put aim (1) in context. The authors defined the significant emissions “point source” threshold at 10 kg of methane (CH₄) per hour.  Is that a lot? Let’s put this threshold into terms that may answer that question. One of the things you measure when monitoring landfill gas wells is the flow rate of gas being collected by the well - a typical value you might see may range from 10 cubic feet of methane per minute to many dozens of cubic feet of methane per minute. OK, the study threshold is in kg/hr, and the standard unit of measure is in cubic feet per minute…so let’s convert the low end of what you typically see flowing through one landfill gas well to see how it compares to the study threshold:

OK, so that means a gas well extracting 10 cubic feet per minute (again, this is at the low end of what you’d typically extract in one well) is the same as it collecting 12 kg methane per hour. Thus, one would expect this threshold to get pretty quickly exceeded, as in theory, it would only take one malfunctioning well to lead to enough methane being emitted rather than collected (remember, a typical landfill will have dozens or even more than 100 operating gas collection wells).

The authors reported that of the 250 landfills examined in the study, 52% exhibited “point source” levels of methane emissions at some time.  However, the converse of this conclusion is somewhat telling - another way to couch the results is “48% of the 250 landfills did not have a detected amount of methane greater than a fairly low threshold of 10 kg/hr”.

So, are these results surprising? Not really. Because US federal landfill gas collection system regulations explicitly allow for monthly gaps between well field adjustments, periods of higher emissions should be expected. For example, a landfill with gas collection wells operating within the confines of federal regulations could potentially have up to 30 days where it is operating sub-optimally, so if we assume that an operator made their monthly measurements on the first of the month, a flyover measuring methane emissions anytime between that day and the subsequent measurement could capture instances where landfill gas performance is sub-optimal. You might think the solution is to “just require more frequent monitoring”, but as described above, the monthly monitoring process is already pretty lengthy. It’s probably safe to assume the federal requirement to monitor wells monthly was a compromise that acknowledged the realities of how long it takes to actually make the measurements.

Beyond federal regulation compliance, are landfill operators incentivized to operate methane collection systems more efficiently? In short - yes. Hundreds of landfills in the US and around the world convert collected methane into energy (e.g., burning the methane and using an engine to create electricity, piping the methane to a nearby industrial facility to burn in a boiler, or cleaning up impurities in the gas and using the cleaned-up gas to power vehicles that can run on natural gas). Landfills get revenue from the sale of the converted energy, so any uncollected methane represents lost revenues that can never be recaptured.

The study identified 113 landfills with high (“point source”) methane emission levels. I cross-referenced this list against a database maintained by the US EPA of landfills converting collected methane into energy. I found that 64 of the 113 landfills have an active energy conversion project. Put another way, the other 49 landfills (43%) with high methane emission rates have no energy conversion system and, therefore, less incentive for aggressive methane collection practices.

The authors also rightly point out that several factors contribute to a landfill exhibiting higher-than-expected methane emissions, including the constant placement of new waste and reconfiguring the landfill. Many of the landfills studied here were open and taking in new waste (most landfills are designed and built to take in waste for several decades or more). Recall how federal regulations specify timing requirements for installing new methane collection systems - in brief, the regulations say you don’t need to collect methane from newly placed waste for up to five years. You might wonder, “Why so long?” Well, the regulation reflects some immutable realities about how landfills operate. First, methane won’t be produced until about a year after the waste is placed, but you wouldn’t put in new gas wells in that area because you’re still filling the area with new waste. Landfills are busy places, and you may have hundreds of trucks daily delivering new waste to the site. Again, partly because of federal guidelines for installing methane collection systems, we’d expect methane emissions to be higher at some point in a landfill’s life than others, owing to the deployment of the collection infrastructure relative to the entire site.

To put this idea in numbers, I analyzed the methane collection efficiency (as reported to the US EPA’s Greenhouse Gas Reporting Program) for each of the 113 landfills exhibiting a large methane emission amount. The figure below shows that only a handful of sites have a reported efficiency near 100% - the rest have far lower values. Not because their gas collection systems are not functioning properly, but simply because the infrastructure hasn’t been overextended (or even most, in a few cases) of the landfill. Again, this mostly reflects regulatory timelines, not explicitly poor performance.

The study reported on landfills that showed “point source” emission behavior. I analyzed the reported gas collection efficiency of those landfills to understand the extent to which the landfill actually has gas collection infrastructure. Most of the sites fall beneath 75%, suggesting the larger emissions were not because of malfunctioning gas collection but rather a lack of coverage.

As for the study’s second aim - it turns out that the aerial method they used to measure methane emissions widely disagreed with self-reported methane emissions estimates at each of the studied landfills sent to US EPA as part of annual Greenhouse Gas Reporting Program submissions, even in cases where the research team measured emissions from the same landfill 10+ times. Interestingly, the discrepancy happened roughly equally on both sides of the self-reported estimates, with about half of the aerial measurements exceeding the self-reported emissions estimates and the other half less than the self-reported emissions estimates.

I don’t think it matters much that the aerial measurements and the self-reported estimates did not agree. If this study’s conclusion was “We flew over 250 landfills, and the emissions we measured were greater than all of the self-reported estimates”, that would be a big deal. However, the techniques used in this study signal a shift toward more frequent and ever-improving monitoring of the waste system. The fact that researchers can now take measurements at multiple landfills at any time with functioning detection technology signals even greater scrutiny on the world’s waste management practices. Further, if we consider landfills in a non-US context, aerial and satellite technologies measuring methane means we can pinpoint large emission sources in parts of the world where landfill details and performance data are not available. You may not be surprised to know that the regulations compelling landfills in the US to gather large amounts of data and report on emissions is pretty unique in the world, so you can see how new measurement techniques can help drive emissions reduction from the waste sector globally. I also think the continued attention and reporting of results in the US from aerial and satellite measurement campaigns will likely spur more aggressive methane collection practices, even in the absence of any significant regulatory shift in the US.

Insights and Implications

The study and its results have implications for a variety of initiatives, sectors, and stakeholders:

Landfill Operators Should Expect to See Additional Aerial Measurement Studies in the Future. The study is the latest research article where measurements are taken at landfill sites using methods that do not explicitly require the researchers to access a landfill site or get permission to take measurements. The authors of the study I analyzed here mentioned they plan additional future satellite-based measurement campaigns with an expected continued focus on landfills, including sites in the US and outside of the US, so I hope we will see more studies of this type (perhaps with improved accuracy and more extensive comparison to other operational conditions to understand the “why” behind specific sites emitting more than others beyond that discussed here). I don’t see aerial methods replacing landfills' traditional estimation methods, though. Still, aerial measurements will probably be an essential check on self-reported emission estimates and help better grasp emission amounts internationally, particularly in areas where good records of waste types and quantities are unavailable.

We may see increased deployment of automation at landfills to better monitor and manage methane collection systems. Even when aligned with regulatory operational standards, manual operation of gas collection systems leaves long periods where gas well performance is unchecked, likely resulting in excess methane emissions. Some landfills have begun to experiment with automated landfill gas collection systems, where logic is programmed into each gas collection well and its settings can be continuously adjusted, automatically, without direct human intervention. Landfills incorporating automation capabilities at individual wells should have a better opportunity to reduce emissions simply because of more significant and frequent visibility on individual well conditions and gas collection rates (Disclosure: In the past I have advised sites using automation systems for landfill gas collection along with the technology providers).

Food Waste Diversion from Landfill Should Remain a Priority, but Infrastructure is Lacking. The simple answer for food waste reduction is this: “We need less edible food going to landfill, and food waste that cannot be avoided should probably go somewhere other than landfill so that we avoid the hard-to-collect methane emissions”. Easier said than done! Many great organizations are working on this very problem - as you can imagine, several factors ranging from simple to complex contribute to food from households and commercial businesses being wasted (I’m only mentioning household and commercials here because those are the food waste sources most likely to go to landfill - on-farm and supply chain losses are another story!). There are technological solutions (namely digesters and compost facilities) that can intake food waste with a far better GHG emissions profile compared to landfills. Still, the infrastructure is lacking, with these facilities only handling 4% of the total annual food waste produced in the US (Source: Center for the Circular Economy’s Composting Consortium). Enhanced efforts to reduce the creation of food waste while increasing diversion from landfills should reduce the amount of methane produced from food waste decomposition at landfills.

Do you know anyone who may enjoy this post? If so, use the widget below to share easily:

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time.

If you view this in your email app, simply hit reply to provide us with your questions, comments, or feedback - we’d love to hear from you.

Do You Know Where Recycled Plastic Bags Go After You Drop them Off?

If You Read Nothing Else in this Post: 

For years, many US consumers have gathered and dropped off accumulated plastic bags in dedicated bins at their local grocery store to be recycled. But are these plastics being recycled? In March 2024, a team of researchers published a clever study where electronic trackers (Apple AirTags and Android Tiles) were placed in plastic bags/film, placed in plastic film drop-off bins at grocery stores throughout the US, and tracked the plastic to their final destination. The study concluded there is “…no evidence that…[plastic] packaging is being widely recycled.” However, my analysis paints a very different picture, where most plastic bundles made their way to an established manufacturer with a documented history of using recycled plastic in their products. My conclusion from the study is that while the recycling system isn’t perfect, the results show that dropped-off plastics were effectively sent for recycling. Still, the authors’ work highlights a vital takeaway: we need checks on and transparency in corporate environmental claims on products and packaging to maintain trust and continue pushing toward the best possible environmental outcome.

Why are Plastic Films Collected Differently Than Other Recyclables Like Cardboard, Aluminum Cans, etc.?

The short story is because plastic bags physically differ from other typical "Curbside” recyclables (i.e., those typically picked up at your home) that are primarily rigid materials like glass, metal, and cardboard. Material Recovery Facilities (MRFs) - the term of art in the recycling industry for a place that sorts all incoming recyclables into their component parts - have historically been designed to target the largest quantity and most valuable stuff. Because film plastic is flexible and lightweight, MRFs are not an ideal destination for plastic bags and films, and in fact most MRFs will try to sort out films early in the recycling process to avoid clogging or damaging separation equipment.

If You Put Plastic Film Bags in Your Curbside Recycling Bin, They Will Most Likely Be Sorted Out Up Front and Discarded to Avoid Damaging Equipment

Thus, dedicated drop-off bins for plastic film recycling sprang up at hundreds of locations, such as grocery stores and retailers, to give consumers an outlet to drop off plastic bags and films. This system allows plastic bags and films to be recycled while avoiding aforementioned problems at the MRF.

A New Era of Transparency and Accountability for the Waste and Recycling Industry

OK, so we’ve described a couple common pieces of the recycling system, namely MRFs and the separate collection systems for plastic films. But how well is this system working? Although recycling has been a familiar and conscious activity for many worldwide for decades, the last decade has seen a reckoning and far more attention being paid to the details of the recycling and waste system than ever before. Why? One of the most influential factors stems from a seminal paper published in 2015 by my friend and colleague, Prof. Jenna Jambeck, who along with her co-authors estimated that as much as 13 million tons of plastic enter the world’s waterways every year. In the decade since, I’ve seen unprecedented attention paid by researchers, technologists, investors, and others to understanding the following:

1. How much plastic and other wastes are entering the environment, why, and working to reduce it.

2. The amount of valuable discarded materials burned or disposed of rather than reused or recycled.

3. The fate of materials sent for recycling to ensure they are recycled, or identifying the “highest and best use” for recycled materials.

The rest of this post focuses on a recently released study relating to item (3) above. Recycling systems are complex and require multiple parts of sometimes complex supply chains to align so that the system “works” (i.e., stuff you recycle actually gets recycled). The study asked the question, “What happens to plastic films and bags after you drop them off in a dedicated recycling bin?”

Who Did the Plastic Film Recycling Study, What Did They Do, and What Were Their Conclusions?

Two non-profit organizations, the US PIRG Education Fund and the Environment America Research and Policy Center, designed a study that included the following steps carried out by the research team and volunteers:

1. Gathered plastic packaging, including bubble wrap mailers, plastic bags, and air pillows. [Note: the study focused on plastic packaging associated with Amazon deliveries, but the results are generalizable to other film plastics with confirmed recycling outlets].

2. The plastic was bundled, an Apple AirTag or an Android Tile was affixed to the bundle, and 93 different bundles were dropped off at grocery store film plastic drop-off bins over an 11-month period at different locations throughout the US.

3. Periodically tracking of the location of each plastic bundle and reported on each bundle’s final location and classifying the final destination type. If one of the trackers stopped communicating for one week, the authors marked the location as “final”.

Plastic Film Tracked by the Study Included <Left to Right> Bubble Mailers, Plastic Bags, and Air Pillows. Image Adapted from US PIRG Report.

The authors concluded the following:

However, a closer examination of the results tells a different story. I reviewed the study’s reported details for all 93 plastic bundles, analyzed the final destination, and created simple plots to understand better what happened. Here are some of my findings:

1. Only forty-three of the 93 bundles had a “confirmed destination”. Of these, 57% went to Trex, 31% went to Landfill, 10% went to a MRF, and 5% went to a waste-to-energy facility. For those unfamiliar, Trex manufactures composite decking and related construction material using (in part) recycled plastic as a raw material.  

Only 43 of the 93 Plastic Bundles Tracked in the Study Had a Conclusive Destination. Most Were Delivered to Trex, an Established Recycler of Film Plastic. Data Here Represent Analysis by Sustainability at the Frontier of the Study’s Reported Data, Created with Datawrapper.

2. The researchers classified the remaining 50 plastic bundles with one of the following fates: (i) never left the drop-off location, (ii) delivered to an intermediate destination, or (iii) the tracking tag “died in transit”. Therefore, no conclusions can be made about the fate of more than half of the plastic bundles tracked by the researchers.

So-Called “Composite” Decking and Lumber Manufactured by Trex and Others May have Up to 95% By Weight Made up of Recycled Plastic (Image Credit: Trex)

2. The claim that the plastics are not being recycled is primarily one of semantics. Twenty-four of the 43 plastic bundles with a confirmed fate went to Trex. The authors contend this is not recycling because Trex isn’t making new plastic bags and films from the plastic bags and films it receives. I think this is too narrow of an interpretation of recycling, as the intent of recycling is to (in part) reduce or eliminate the use of natural resources in production processes. Further, Trex commissioned independent analyses characterizing the environmental performance of its materials compared to traditional, virgin resources (i.e., treated lumber - see here, p. 29), which showed favorable results across several environmental dimensions like total greenhouse gas emissions, air pollutant production, and more.

3. The authors suggest Trex does not want to accept dropped-off film plastics and that Trex products are not recyclable when removed from service, but these claims are not supported by Trex’s sustainability disclosures. The report suggests that Trex does not desire dropped-off flexible film, citing, for example, “...it is unclear how much Trex uses the post-consumer plastic film collected from drop-off bins in stores” [p. 10]. However, Trex’s most recent Annual Environmental, Social, and Governance report (2022, p. 19) cites post-consumer film plastic as one of three key material feedstocks for their products. Further, Trex highlighted progress on research and development efforts around reverse logistics and reprocessing of Trex material taken out of service for recycling into new products [p. 21].

4. Some factual errors in the study may need to be clarified for an already-confused public about which recyclables should go where. Page 8 of the study highlights how four plastic bundles were delivered to MRFs and implies those were the only bundles reaching the correct destination. For reasons I stated at the beginning of this post, MRFs are not an ideal destination for film plastic owing to its lightness and tendency to clog MRF equipment mainly designed for larger, heavier, more rigid items.

5. Some of the Plastic Bundles May Have Been Delivered to a Disposal Location Because of Contamination, Including the Presence of the Electronic Tracker Itself. Using trackers to understand the journey of the dropped-off plastics was a clever but imperfect way for the researchers to assess the fate of materials in this study. First, a device with a battery and other electronic components is decidedly not “film plastic,” so it is unclear how many materials that ended up at a landfill or a waste-to-energy facility did so because the tracker was identified and the bundle separated during the journey. Most recycling processes (MRFs and otherwise) have physical and mechanical processes to ensure only desirable materials continue going down the recycling supply chain. Further, other contaminants may have been placed in the drop-off bins before or after the tracked bundles were placed, resulting in the load being sent to disposal rather than the contaminated materials continuing down the recycling supply chain. For example, page 10 of the study states: “...many of our volunteers reported that they saw food scraps in the collection bins they used for this investigation”. It’s not clear how many “many” is, but it appears clear that contaminants likely played a role in some of the bundles not being routed for recycling.  

6. Plastic Bundles Going to Trex Had Longer Journeys Compared to Other Destinations. I converted the reported start and end point for each bundle into a latitude and longitude and analyzed the “as the crow flies” travel distance for each bundle, and the results are shown below. I’m not surprised by this result, given that there are far more waste handling and disposal facilities than recycling facilities. 

The Average Distance Traveled by Plastic Bundles Was Far Greater Than Other Intermediate or Final Destinations Reported in the Study. Credit: Analysis by Sustainability at the Frontier of the US PIRG Study’s Reported Data.

9. There Appears to be Little Relationship Between the Store Where Plastic Bundles Were Dropped Off and the Reported Fate. The messy figure I made below shows the reported fate across the couple dozen grocery retailers where the plastic bundles were dropped off. All but three of the retailers had 5 or fewer plastic bundles that were tracked, making meaningful statistical conclusions a bit out of reach.

This figure shows the number of plastic bundles tracked according to the retailer where the bundle was dropped off. Note that the study dropped bundles at multiple, different stores even of the same chain (e.g., bundles were dropped at multiple Safeway stores, not just one location). No real trend is event from looking at these data, but an extension of this type of work could look at the consistency with which dropped-off plastics reaches the intended final (recycling) destination.

What Can We Conclude from All of This?

My read is that the plastics tracked here with a confirmed final destination essentially went to a place (Trex) where they were likely recycled into new products. Calling composite lumber made from recovered plastics “not recyclable” is a primarily unhelpful game of semantics. Environmental claims should constantly be scrutinized, and the “receipts” as they were should always be made available by those making the claims to ensure credibility and to maintain public trust, not only for recycling claims but other sustainability-related claims. In this case, I think Trex does a good job of reporting on their progress while acknowledging continued investment and analysis in doing better (in this case, reducing the overall environmental burden of their products and trying to create systems to take back and reprocess their manufactured materials when they reach the end of their useful life). Although the study examined here contained some inaccuracies and misleading statements, studies like this can serve as essential checks on corporate environmental claims and may help contribute to greater transparency and, ideally, better environmental outcomes.

Do you know anyone who may enjoy this post? If so, use the widget below to share easily:

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time. If you view this in your email app, reply to provide us with your questions, comments, or feedback - we’d love to hear from you.

How do ESG Software Ratings Work?

If You Read Nothing Else in this Post: 

Organizations of all sizes are being compelled - through drivers like regulation and consumer demand - to finely track and set goals for their environmental impacts (like their carbon footprint), social progress (like employee turnover rate), and governance (like data security). Accurately quantifying these metrics has historically been challenging, owing to a lack of data, transactional friction in obtaining the data, or an absence of standardized methods of computing these metrics. Environmental, Social, and Governance (ESG) platforms recently emerged as a new software category to facilitate creating effective data-driven baselines and ongoing tracking toward goals, and dozens of providers have entered the market, estimated to reach $1.5 billion by 2027. Third-party raters have simultaneously emerged to help the buying market understand the relative capabilities of various ESG software providers. Still, the raters and methods used vary widely, but the variation and information underpinning ratings are often opaque to the buying market. This post unpacks the process ESG software companies undergo as part of the rating process and highlights how the buying market should consider third-party ratings in their decision-making process. 

How Did ESG Platforms Emerge as a New Software Category?

For years, many large companies and organizations have voluntarily reported on select sustainability-related metrics. In essence, companies decided the factors on which they chose to document and put them out into the world, leaving investors and the public to understand whether their environmental, social, and governance (ESG) activities were any good. Thus, the process of stringing together bits of information, mainly manually, from various places just kind of “worked” in the absence of any real drivers mandating accuracy, completeness, or transparency. 

The old “wild west” ways of tracking and reporting on sustainability progress shifted in recent years, owing to multiple factors like global alignment to make tangible progress on climate, new standards to align everyone on the methods used to compute and report their impacts, and the proposal or passage of regulations requiring uniform disclosure of ESG metrics and progress. With the ingredients of (i) standards alignment and (ii) a woefully inadequate status quo around tracking organizational ESG impacts, the opportunity for software companies to improve the accuracy and efficiency of sustainability and ESG program tracking and reporting was born.

In 2024, the ESG software industry counts dozens of providers vying for market share in what’s estimated to be a multi-billion dollar opportunity. Because of the market and regulatory dynamics, more organizations are creating goals and tracking ESG progress for the first time, while others who previously reported voluntarily are stepping up their game to meet more rigorous demands for accuracy and transparency. As in other software categories like data analytics, customer relationship management, and many others, a host of third-party software ratings agencies have begun to focus on the ESG software space to help buyers understand relative capabilities, limitations, and other factors. Getting the software selection right is critical for corporates and other organizations compelled to implement a rigorous ESG program, so third-party raters have an important role to play in helping sort out the software market’s current and future capabilities, which is accomplished by evaluating each software platform, create metrics for comparison, and reporting on the findings. 

The main result from a software rating analysis is a Quadrant like that shown. Software companies want to land “up and to the right.” Each dot represents a unique software provider. Woe is the software company that lands in the bottom left.

How Does the ESG Software Rating Process Work?

I co-founded a global team at Salesforce called the Sustainability Practice - we sat at the nexus of multiple teams at the company (Sales, Innovation, Distribution, Corporate Sustainability, and the Product team for Salesforce’s ESG Software, Net Zero Cloud) and our main job was to deliver expert advice, guidance, and best practices to sustainability leaders at hundreds of the world’s largest companies. Part of my job involved deeply understanding the ESG software landscape and supporting the product team when they engaged with an ESG rating company, mainly to ensure the software’s capabilities were clearly and effectively communicated to achieve the best possible rating. Positive rating outcomes were a powerful tool for software sales teams when speaking with customers, as these outcomes provided independent validation of the software’s capabilities and the likely strength of planned developments. Conversely - a negative rating outcome created a challenge for sales teams who could struggle to explain why the software platform was not rated or carried a less-than-desirable rating.

ESG raters usually go through these steps in their analysis: 

1. Reviewing the landscape review of available software providers and determining (based on a desktop analysis) which providers are “in scope” for their rating exercise. 

2. Asking in-scope providers if they are willing to participate in the ratings exercise.

3. Sending participating software providers a questionnaire about the software company and its product. 

4. Reviewing the questionnaire responses.

5. Participating in a live demonstration of the ESG software. 

6. Collating all results and drafting their ratings.

7. Publishing the ratings either as is or after debriefing with the software provider. 

The process takes around two to four months end-to-end and requires a good deal of work both for the raters and the software providers. Steps 1 and 2 are pretty straightforward - for the remainder of the post, we’ll focus on Steps 3 through 7.

The ESG Software Questionnaire and Product Demonstration

ESG software companies, once confirming their participation in a rating exercise, usually receive a spreadsheet-based questionnaire that asks about a range of topic areas. The questionnaire typically covers areas like:

1. What industries does your software serve?

2. How large is your customer base?

3. What features and functionality does your software currently have?

4. What features and functionality are you planning to build into your software? 

Here’s an example of how a questionnaire may look. 

Questionnaires to which ESG software companies respond during the rating process can be lengthy, often exceeding 100 questions across topics like carbon accounting, ESG metrics tracking, reporting, security, and consulting services to complement the software.

Pretty daunting, huh? Software companies generally have two to four weeks to evaluate, assemble, and submit their initial responses to the rating company, which typically requires multiple stakeholders to review, provide input, and approve responses before submission. 

Once questionnaire responses are submitted, the rating company will analyze responses and schedule a time to view a demonstration of the software. The demonstration helps to make the written questionnaire responses “come to life” while acting as an essential check on claims made by the software company. A typical demonstration will last around two to three hours, including the software team walking through a series of scenarios demonstrating how the software works. The preparation for a software demo typically requires solution engineers to compile existing “demo assets” and develop new ones that respond to the requests of the software ratings company. The demo addresses pre-defined elements and spontaneous back-and-forth requested by the rating company. 

After the Demo

Once the demo is complete, the bulk of the software company’s work is done, but the work of the rating company is just beginning. They work to combine the objective and subjective information gathered from the questionnaire and product demos from all of the different software companies they are evaluating and use quantitative and qualitative techniques to score the capabilities of each platform they review. Evaluation criteria are typically weighted to enable a relative ranking or comparison between each software provider. 

After the ranking and scoring process concludes, the rating company will furnish draft findings (e.g., overall rating, summary of strengths and weaknesses) and schedule a debrief with each software company. Software vendors commonly have an opportunity at this stage to address any aspects of the review that have been egregiously missed or otherwise misunderstood. However, this only sometimes happens, and a rating company may or may not adjust draft findings in response to the debrief. 

After a grueling multi-week, multi-stakeholder process of responding to questionnaires and delivering demos, ESG software companies eagerly anticipate getting “the call” from third-party raters to hear how they scored.

Obvious and Non-Obvious Factors that Influence ESG Software Ratings

The ESG software rating process includes a blend of straightforward, objective measures, along with a good deal of subjectivity. So, what are the factors that most strongly influence a software provider’s ESG rating? 

1. Experience Level of the Rater(s). Interestingly, while the focus of the ESG software rating process is on the software company and its offering, there are no standards for the raters who carry out the software ratings. Ideally, the rating company and the people doing the rating have expertise in the subject matter (in this case, ESG matters that corporates and other large organizations deal with) and enterprise software itself. The combination of subject matter and software expertise is rare, given the novelty of ESG standards and software solutions. So, it’s essential to consider this fact when reviewing results for a given rater.

2. Time and Capacity Available to Respond. There are many rating companies, and, as described in this post, the process for participating in a rating exercise comprises many weeks of effort across many individuals at each software company. Bandwidth is always an issue, particularly at start-ups. Thus, many software companies may only choose to participate in one or a few rating processes in a given year and may decide to opt out of several others. Thus, if one software company appears in one rating report but not another, it may be that the company had to prioritize one over the other because of available capacity. Other factors for a company’s omission from a rating could be the software being seen as “out of scope” when evaluating the software product not meeting the maturity threshold set by the rater. Companies with dedicated personnel to respond to third-party raters are more advantaged because they can build processes to craft effective responses while participating in more ratings than software companies with relatively fewer resources.

3. Capabilities of the Software Platform. This one is fairly obvious—many evaluation metrics are straightforward, so there is a clear separation between software platforms with a given capability and those without. For example, if an evaluation metric is “Uses generative AI within the platform to help create an ESG report”, a company with that capability will of course score higher than one without.

4. “Sizzle” in the Product Demonstration. The software demo has an important role, and aesthetics and delivery can strongly influence how the capability and sophistication of a given software is perceived and, therefore, rated. Here, legacy enterprise software platforms are often at a disadvantage compared to newer purpose-built platforms simply because legacy architecture cannot be substantively changed. Raters are human, and platforms with flashy demos can sometimes get a substantive bump across multiple evaluation criteria compared to legacy enterprise platforms.

5. Prohibitions to Responding to Certain Common Questions. A significant part of ESG software ratings involves metrics around investments in the product, the number of engineers working on the platform, and the size of the customer base. At Salesforce and other publicly traded companies, strict policies and confidentiality requirements prevent specific data from being given on these metrics, even at a high level. So you can imagine the difficult task of a rater comparing one company that says, “We have 163 customers in 24 countries with $50 million in annual revenue and a team of 35 full-time employees building and delivering our product” versus “Our company policy prevents us from disclosing specific customer numbers, installation base, and internal resourcing for our product.” 

Should Corporate Sustainability Teams Use Third-Party Ratings in their ESG Software Evaluation Process, and if So, How?

Yes, third-party ratings of ESG software can be an essential tool for corporate sustainability teams and others in the market to purchase ESG software. Here are some tips and considerations for how sustainability teams can leverage ESG software ratings:

1. Never over-rely on a Single Third-Party Rating in your Decision-Making Process. Several variables play into how a given software company scores, so use reviews from multiple rating companies and pay attention to software providers who consistently rank highly. 

2. Which Ratings Group Should You Pay Attention To? Figuring out which raters to pay attention to is as tricky as figuring out which ESG software to select, as many raters exist. Gartner, Forrester, IDC, and Verdantix appear to have put more resources and expertise behind their ESG teams, so those companies are good places to start. 

3. Source and Use ESG Rating Reports to Inform Your Request for Proposal. When the ESG software review process concludes, the software company participants can license and freely distribute part or all of the raters’ report for a fee. The ESG software raters also publish and sell these reports to the public, usually costing several hundred or several thousand dollars. However, sustainability teams may dig on LinkedIn, Google, or Perplexity (or you could click here, here, or here to download a rating report from one of the ESG software provider’s websites). The reports typically provide a helpful and well-organized narrative describing the capabilities of each participant, along with the quadrant, which sustainability teams can use to help narrow the field of providers they will invite to bid when they are in the market for ESG software. Corporate sustainability teams should leverage the time, effort, and financial resources that software companies and raters put into developing and disseminating these reports to help speed time to action and clarify their decision-making process when purchasing ESG software.

Thank you so much for reading Sustainability at the Frontier. We’ll see you next time - if you’re viewing this post in your email, please hit reply and provide us with any of your constructive feedback or perspectives.