By Julian Koelbel and Markus Leippold, University of Zurich; Jordy Rillaert, SFI Swiss Institute, and Qian Wang, UBS
The importance of disclosing climate risks
Efficient climate risk pricing in capital markets requires companies to adequately disclose specific risks. These can be divided into two broad categories: transition and physical risks. Transition risks relate to the low-carbon economic transition, which involves significant policy, regulatory, technological, and market initiatives as we adapt to climate change and mitigate its causes. Physical risks include catastrophic events such as wildfires or longer-term shifts in climate patterns such as more frequent and prolonged droughts.
Both types of risk may have significant financial implications for companies. Without their effective and transparent disclosure, the financial impacts of climate change may not be priced correctly – in the long run, this will lead to significantly higher economic costs, making rapid adaptation with destabilising effects on financial markets more likely.
Our paper asks whether regulatory disclosure provides valuable information on companies’ climate risk exposure and whether this is reflected in market prices, particularly in credit default swaps (CDS).
Why analyse regulatory disclosures?
For our study, we rely on the 10-K filings required by the US Securities and Exchange Commission (SEC). These regulatory disclosures are often criticised for being generic, compliance-oriented, and lacking clear, concise, and insightful information. As a result, it is not apparent whether investors price the climate risk information in them.
However, they are mandatory – failure to disclose climate risks can lead to litigation – and therefore useful. Voluntary disclosures also have disadvantages, namely that companies are prone to cherry-picking non-material climate risk information in their reports. Moreover, since 2006, the SEC requires that the 10-K reports include a specific section (Item 1.A) for firms to identify significant risks to their businesses.
Using BERT to understand climate-risk disclosure
Regardless of the pros and cons, we need a proper tool to extract valuable information from the often repetitive and legal language of corporate disclosures.
Climate change is a complicated topic for natural language processing (NLP) algorithms. As recent research shows, keyword-based approaches struggle considerably in identifying climate-relevant text. Therefore, we adapt and train one of the most powerful NLP methods, a deep neural-network called BERT (Bidirectional Encoder Representations from Transformers).
Unlike traditional word embedding, BERT is a contextual model – it respects word dependencies and sentence structures when analysing words. We use our fine-tuned BERT to assess relevant text in the 10-K filings that deal with transition or physical climate risks.
Why are we considering the CDS market?
Intuitively, an increase in natural disasters and the low-carbon economic transition engender high economic costs. These will partly be borne by companies, impacting their assets, cash flows and ultimately, their creditworthiness.
We can capture this impact on creditworthiness by investigating CDS spreads, which respond quicker to changes in market conditions than bond spreads, as CDS contracts are traded on standardised terms. Moreover, the CDS market is dominated by professional investors that are in the business of hedging risks. In contrast, climate risk premia in equity markets could also be driven by divestment policies that reflect investors’ moral tastes and environmental preferences.
For our analysis, we use data from February 2010 to December 2018. Matching the CDS data with the 10-K filings and firm-specific variables for our regression analysis, we end up with 447 different companies in our sample.
Pricing effects of climate risk disclosure
While the importance of climate risk disclosure for market efficiency is agreed upon, the directional effect on risk premiums is not apparent – it may trigger two opposing effects. When disclosure reveals a novel risk factor, it increases investors’ risk perception, leading to higher risk premiums. However, when disclosure reduces the information asymmetry between investors and companies, it can resolve uncertainty around a risk factor and, consequently, leads to a decrease in risk premiums.
Using our novel measure based on BERT, we find that the CDS market responds distinctively to the disclosure of transition and physical risks. The disclosure of transition risk increases credit spreads, and the effect is stronger after the Paris Climate Accord of 2015, which is consistent with the risk-perception effect. In contrast, the disclosure of physical risks has a negative impact on credit spreads, which is consistent with an uncertainty reduction effect. These effects become significant only when restricting the sample to industries where climate risks are deemed material, as defined by the Sustainability Accounting Standards Board.
In our sample, the effects are not only statistically significant but also economically sizeable. As Figure 1, Panel A, suggests, a one-standard-deviation increase in transition risk leads to an increase of 6.99 basis points in the average five-year CDS spread for the post-Paris period – a 4.3% increase in the average. In contrast, in Panel B, we find that one standard deviation in physical risks results in a decrease of CDS spreads of 7.37 basis points, representing a 4.1% reduction for the average five-year CDS contract in our sample. The effects on shorter-term CDS spreads are even more substantial.
Implications for researchers, regulators, and investors
Our results have several implications for investors and policy makers. For investors, this study adds to emerging evidence that climate risks are, to some extent, priced in financial markets. Given our evidence on the Paris Agreement’s impact on transition risk, the pricing effect becomes particularly significant when these risks reach the political arena. Given that our findings rely on CDS prices – a market where divestment does not play an important role – the paper indicates that the pricing effect is driven by risk considerations, not ethical tastes.
Regulators can learn that the SEC’s approach to climate risk disclosure somewhat fulfills its objective of informing investors about climate risk, although there is still much room for improvement. They should also note that the disclosure of physical risks may have different consequences than the disclosure of transition risk.
As our results show, regulatory disclosure does impact prices. Therefore, it would be interesting for future research to investigate how to transform emerging voluntary standards, such as the Task Force on Climate-related Financial Disclosure recommendations, into an effective, informative, and transparent regulatory framework.
This blog is written by academic guest contributors. Our goal is to contribute to the broader debate around topical issues and to help showcase research in support of our signatories and the wider community.
Please note that although you can expect to find some posts here that broadly accord with the PRI’s official views, the blog authors write in their individual capacity and there is no “house view”. Nor do the views and opinions expressed on this blog constitute financial or other professional advice.
If you have any questions, please contact us at firstname.lastname@example.org
Bidirectional Encoder Representations from Transformers