How does climate change affect firms and the global economy? What are the real effects of net-zero transition? And what does it take to support the transition to a greener economy? A new method could help to answer these and further questions. The method developed by TRR 266 researcher Laurence van Lent (Frankfurt School) together with Zacharias Sautner (University of Zurich and Swiss Finance Institute), Grigory Vilkov (Frankfurt School), and TRR 266 alumnus Ruishen Zhang (Shanghai University) identifies the attention paid by earnings call participants to firms’ climate change exposure and allows predictions about how climate change impacts jobs, innovation, and risk-sharing in capital markets. A considerable number of studies has already built on this data.
In official climate forecasts, researchers calculate how the global climate will change according to different scenarios. A doubling of the CO2 concentration by the end of this century, for example, could increase the average temperature on Earth by 1.5 to 4.5 degrees. The consequences would be devastating: natural disasters, rising sea levels, declining biodiversity. Serious threats that would profoundly affect many areas of our lives. Even the way business is conducted.
While complex models allow those estimations for the global climate, no models can reliably predict how climate change affects jobs, innovation, and risk-sharing in capital markets. Measuring these kinds of effects is difficult since they are multi-faceted – climate change impacts firms in various ways. While some firms face costs from physical climate change, for example, due to rising sea levels, others face costs imposed by regulations implemented to combat global warming (e.g., carbon taxes). Climate change offers opportunities for still others, such as green energy producers.
An important aid for investors, regulators, and policymakers
A method that seeks to measure the impact of climate change on firms reliably must capture these differences across firms. It must also reflect market participants’ assessments of how climate change affects individual firms. Such information is important given the critical role that market participants play in the resource allocation and price discovery process.
Such a method could help investors, regulators, and policymakers quantify firms’ climate change exposure and set them on the right course since it allows them to make important predictions. For example, how firms will respond to net-zero transition and how this affects the global economy. It can also help to predict how to stimulate the creation of green technologies and patents to achieve net-zero by 2050.
New method measures a firm’s climate change exposure based on earnings calls
Therefore, based on previous studies, we developed a new method that measures the climate change exposure of firms or, in other words: how much a firm is affected by opportunities and risks arising from climate change. To measure a firm’s climate change exposure, we use earnings calls – conference calls between the management of a firm and analysts discussing the firm’s financial performance and current and future developments. The advantage of earnings calls? They are less prone to greenwashing since they contain not only the firms’ perspective but also that of analysts. We determine a firm’s exposure by measuring the proportion of the conversation devoted to climate change topics. We do this by using so-called bigrams, two-word combinations indicating climate change talk – such as the bigram “solar energy.”
In other studies, bigrams are often determined with the help of pre-specified training libraries. However, this is difficult in the context of climate change since there is no well-defined dictionary. The terms used by policymakers, journalists, and financial market participants vary widely over time. Creating a new dictionary from scratch is challenging and susceptible to human error. For this reason, we developed an algorithm that uses machine learning to identify word combinations that indicate climate change conversations.
Algorithm uses machine learning to identify climate
The algorithm only requires human input right at the beginning: we developed a short list of initial keywords that most experts would agree are related to climate change. This initial set of bigrams allows the algorithm to identify sentences that unambiguously refer to climate change from the transcripts. The algorithm extracts bigrams beyond the initial set by relying on supervised learning methods. The algorithm extends the rather broadly specified initial bigrams into more specialized word combinations. For example, “solar energy” gives rise to the bigrams “rooftop solar” and “photovoltaic panel.”
To distinguish the different levels at which a firm can be affected by climate change, we constructed four sets of climate change bigrams: The first set captures broadly defined aspects of climate change. The remaining three cover specific climate issues: technical opportunities as well as physical and regulatory shocks. In addition, our method allows determining whether reported events are good or bad news for the firm. We achieved this by constructing “sentiment” measures that count the relative frequency of climate change bigrams in the vicinity of positive and negative tone words.
More opportunities and regulatory risks, more green jobs
Our data is available for over 10,000 firms from 34 countries between 2002 and 2020. Our study shows that this data allows us to predict real outcomes that can guide regulators, policymakers, and investors, especially in the context of the net-zero transition. For example, a firm’s exposure to climate change can predict whether the firm will invest in green technologies and patents, two key drivers of the transition to a low-carbon economy. Firms exposed to more regulatory risks and opportunities due to climate change create significantly more jobs in disruptive green technologies and file more green patents over the subsequent year.
The exposure of firms to climate change also has an impact on equity markets and gets along with increased risk and risk premiums. This is generally true for shares of firms with greater exposure to climate change in general and especially for shares of firms with a higher opportunity exposure. This is plausible as plenty of uncertainty surrounds investments in green technologies or renewable energy. Realizing these opportunities leads to significant gains if successful or large losses if unsuccessful. Thus, investors are willing to pay premiums for downside protection and upside growth potential. This makes the costs for downside crash protection and upside potential relatively more expensive. In addition, our study shows that the costs of upside potential grow faster than the costs of downside crash protection if a stock’s climate change opportunities are higher.
Out-of-sample evidence: a considerable number of studies already cite the new method
Our data have been cited in a considerable number of studies to predict a range of real and financial outcomes of climate change. And the number is constantly growing. This “out-of-sample” evidence is reassuring, indicating that our measures capture meaningful variation across firms. Regulators have also expressed interest in our data. For example, we have presented the study to the educational ESG research department of Vanguard Investment Strategy Group. We are confident that our research can make a difference by contributing to a better understanding of the impact of climate change on firms and the global economy and to effective regulation in the context of the net-zero transition.
To cite this blog:
Van Lent, L. (2023, july 25). How does climate change affect jobs, innovation, and risk-sharing in capital markets?, TRR 266 Accounting for Transparency Blog.