Climate change represents, perhaps, the largest risk facing the world today. Financial markets represent, arguably, our most sophisticated tool for pricing risk. But markets haven’t been offering clear signals, incentives, or financial instruments for navigating climate change.
In a new paper, researchers at Yale SOM and NYU use machine learning tools to identify intensity and sentiment of news coverage of the climate crisis and then construct stock portfolios that rise when there is bad news. Their hope is that such a financial tool can help investors direct resources toward companies and activities that ameliorate the negative impacts of climate change.
“Financial markets are just now starting to recognize climate change is a major issue and that we just don’t have many tools that allow us to manage this risk,” says Stefano Giglio, professor of finance at Yale SOM and one of the authors, along with his Yale SOM colleague Bryan Kelly, and Robert Engle, Heebum Lee, and Johannes Stroebel of NYU.
“There’s a lot of work to be done in order to understand how financial markets can help in providing solutions to climate change and the problems that come with it,” Giglio says. “What we’re trying to do in the paper is to show just one example of what you could do.”
In theory, financial markets could be among the most powerful means of combating climate change. “This is the economy’s way of reallocating resources to projects that are, in the long run, mitigating climate risk,” Kelly says. Whether through demand for products, higher value for firms, or lower cost of capital, market signals can facilitate development of promising responses to climate change.
But currently, there’s no financial product aimed at hedging climate change impacts. “Many developing countries are very directly affected by climate risks, but they don’t have a way to buy insurance against it,” Giglio says.
Derivatives are a common tool for managing risk. For example, those concerned about long-term inflation can buy an inflation swap in which a counterparty will pay out if inflation rises above a defined level, for instance. The authors note in the paper that such instruments break down when applied to the climate crisis, since no counterparty can credibly pledge to pay out in the case of a climate disaster decades in the future that might fundamentally upset the economy.
“This is a different problem than most types of risks that financial markets are well-suited to hedge,” Kelly says. “The negative consequences are much further out in the future.” Markets are best able to assess risks that are expected within a few years or over the course of a business cycle. “Now we’re talking about a risk where some of the worst outcomes might be 50 years off or more,” Kelly explains. “You have to worry about issues that you wouldn’t consider with normal stock market risk, like, how do I feel about my descendants’ wellbeing?”
Markets need people with different views to trade against each other—optimists to bet against pessimists. While we certainly have climate optimists and climate pessimists, when you try to focus those positions into a single trade, things get complicated. “We don’t understand the effects of climate change on the economy,” says Giglio. “We don’t understand whether we’re going to be able to adapt or mitigate climate change. There’s so much uncertainty about the underlying process, it’s almost impossible for anyone to figure out what to do. Which in turn is one of the main issues with thinking about constructing a market for climate change risk.”
To construct their proposal, the authors had to make a series of choices. Two of the key choices: What are you going to hedge against? What tool are you going to use?
While the climbing average global temperature is a key part of the climate change discussion, the authors chose against it as the thing to hedge. “A temperature rise isn’t exactly the same thing as climate change, and it’s not the same thing as a negative outcome associated with climate change,” Kelly says. Instead, they chose to focus on the human reaction. “We tried to put together a portfolio that rises in value when people feel especially afraid of what’s going on with the climate.”
They constructed two indices based on text analysis of news coverage: the first measures the frequency of discussion of climate change and related issues in the Wall Street Journal; the second uses sentiment analysis to measure the intensity of negative climate news in more than 1,000 major news sources aggregated by the data analytics service Crimson Hexagon.
The natural language processing analytics was performed by Crimson Hexagon using terms and criteria provided by the authors.
The next choice was the asset class. “When you have a risk that you’re trying to hedge, in a financial markets sense, you want to buy assets that pay off should the bad things occur.” Kelly explains. They chose a portfolio of U.S. equities as an easily trackable and tradable asset. “We decided we wanted it to be as practical and implementable as possible,” Giglio adds. “The objective of all this is to develop a prototype.”
The authors follow standard methods in the asset pricing literature to construct portfolios composed of stocks that would rise when there was bad news related to the climate, as measured by their two indices. They also used third-party firm-level ESG (environmental, social, and governance) ratings to refine which companies are most exposed to climate crisis impacts.
One surprising finding was that it wasn’t successful to simply go long on green energy stock while shorting oil companies. The paper notes, “Our hedge portfolios do not resemble industry bets; rather, they identify, both within and across industries, those firms with the largest exposures to climate change risk, yielding a climate hedge portfolio that is relatively industry-balanced.”
The authors are very much aware of the provisional nature of their proposal. “It’s not going to be the ideal hedging instrument, but it’s a way to start thinking about the problem,” Giglio says.
Markets are powerful because they adapt. While the goal is to create a real product that can be traded, the working paper is just a first step. “This is a very hard question, and so I think any progress is great progress,” Giglio says.
Addressing such a complex problem is going to require iterating on answers, making them better with each new version. “One thing that we like about our approach, is that it’s modular,” Kelly says. “If people have ideas for more useful inputs, they can take the structure that we’ve laid out and plug in the parts that they think are improvements.”
One innovation that may prove fruitful in a variety of research areas is the use of natural language processing and text analytics. By bringing it into their model, Kelly says, “It helps the finance profession recognize that nonstandard data sources can be useful for very practical problems.”
He offers a reminder of the larger goal: “Hopefully by making hedging more accessible, prices will equilibrate more efficiently to reflect the collective demand for less climate risk investments.”