股票市场气候风险的分解（英）-国际货币基金组织.pdf

Decomposing Climate Risks in Stock Markets Yuanchen Yang, Chengyu Huang, Yuchen Zhang WP/23/141 IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. 2023 JUN * “The authors would like to thank Koshy Mathai, Pierpaolo Grippa, Tannous Kass-Hanna, Liam Masterson, Lucy Liu, Augustus Panton, Divya Kirti, Henk Jan Reinders, and participants at the Canada 2022 Article IV Analytical Seminar for comments, and Jerry Chaves for research inputs. © 2023 International Monetary Fund WP/23/141 IMF Working Paper Western Hemisphere Department Decomposing Climate Risks in Stock Markets Prepared by Yuanchen Yang, Chengyu Huang, Yuchen Zhang Authorized for distribution by Koshy Mathai June 2023 IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. ABSTRACT: Climate change poses an unprecedented challenge to the world economy and the global financial system. This paper sets out to understand and quantify the impact of climate mitigation, with a focus on climate-related news, which represents an important information source that investors use to revise their subjective assessments of climate risks. Using full-text data from Financial Times from January 2005 to March 2022, we develop machine learning-based indicators to measure risks from climate mitigation, and the direction of the risk is identified through manual labels. The documented risk premium indicates that climate mitigation news has been partially priced in the Canadian stock market. More specifically, stock prices react positively to market-wide climate-favorable news but they do not react negatively to climate-unfavorable news. The results are robust to different model specifications and across equity markets. RECOMMENDED CITATION: Yang, Yuanchen, Chengyu Huang, and Yuchen Zhang, 2023. Decomposing Climate Risks in Stock Markets. IMF Working Paper No. 23/141. JEL Classification Numbers: G11, G12, Q54 Keywords: Climate Mitigation; Machine Learning; Asset Pricing Authors’ E-Mail Address: yyang6@imf.org; chuang@imf.org; yzhang5@imf.org WORKING PAPERS Decomposing Climate Risks in Stock Markets Prepared by Yuanchen Yang, Chengyu Huang, Yuchen Zhang IMF WORKING PAPERS Decomposing Climate Risks in Stock Markets INTERNATIONAL MONETARY FUND 2 Contents 1. Introduction . 3 2. Literature Review 6 3. Methodology and Data 10 4. Results and Discussion 16 5. Robustness Checks 21 6. Conclusion . 26 References . 27 Appendix 31 FIGURES Figure 1. Word Cloud of Climate Policy Keywords . 11 Figure 2. Research Methodology Flowchart . 15 Figure 3. Intensity of Climate News Coverage 16 Figure 4. Intensity of Climate Favorable and Unfavorable News 17 TABLES Table 1. Pricing of Climate News Factor in Canadian Stock Market . 18 Table 2. Portfolio Sorting Analysis of Canadian Stock Market 19 Table 3. Pricing of Climate News Factor in the US Stock Market . 20 Table 4. Pricing of Climate News Factor in the EU Stock Market . 20 Table 5. Pricing of Climate News Factor Using Alternative Asset Pricing Models 21 Table 6. Pricing of Climate News Factor Using Alternative Climate Risk Measures . 22 Table 7. Pricing of Climate News Factor Using Different Sorting Strategies 22 Table 8. Pricing of Climate News Factor (Canada-specific) in Canadian Stock Market . 23 Table 9. Pricing of Climate News Factor (Without ESG) in Canadian Stock Market 24 Table 10. Pricing of Alternative Climate News Factor in Canadian Stock Market . 25 IMF WORKING PAPERS Decomposing Climate Risks in Stock Markets INTERNATIONAL MONETARY FUND 3 1. Introduction The world’s climate is changing. There is general scientific consensus that increased greenhouse gas concentrations are attributable to human activities (IPCC, 2021). The average global surface temperature could rise by 3-6 degrees Celsius by 2100 without immediate action to slow the pace of global warming (OECD, 2021). With increasing global surface temperature, the likelihood and intensity of natural disasters will also increase. Climate change is an existential threat. While countries broadly agree on reaching net-zero emissions by mid- century to avoid the most adverse climate change scenarios, uncertainties surround the transition toward a low- carbon economy. If governments, firms, and households fail to check temperature rise and mitigate climate change, disorderly adjustments in asset prices would occur, with possible disruption to the proper functioning of the financial system and potential spillovers to other sectors of the economy. This paper sets out to examine how the market views and prices climate mitigation policies, with a focus on the Canadian economy. It is widely recognized that risks of climate changes can be divided into two types: physical risks and transition risks (IMF, 2022). The former refers to the risks stemming from severe climate events, such as floods, droughts, wildfires, etc., whereas the latter results from policy changes, technological advances, and market shifts in the process of adjusting to a low-carbon economy. Of different types of transition risks, policy change is the one that has attracted the most attention. With the net-zero pledge constantly calling for more mitigation actions, it would be important to understand the effectiveness of existing mitigation policies, and in particular, whether current policies provide adequate incentives to steer agents’ behavior towards lower carbon emissions. One of the channels for policies to shape agents’ behavior is through price adjustments in financial markets. Assessing the impact of mitigation policies is particularly relevant to Canada, which is among the world’s top carbon emitters and faces a major transformation as the world moves away from fossil fuels. (Government of Canada, 2020; Environment and Climate Change Canada, 2022) The government is pushing for stronger climate policies at both the federal and provincial levels, with profound implications for its oil and gas sector. Prime Minister Trudeau announced during the United Nations Climate Change Conference in Glasgow that Canada will be the first major oil-producing economy to cap and reduce pollution from the oil and gas sector to net zero by 2050. Facing a hard path to decarbonize, the sector will need to continue to adapt, in order to spur innovation and preserve jobs. However, it is not yet well understood how the sector will respond to market and governmental regulatory pressure for green transition. The key challenge to analyzing market responses to climate mitigation is to develop a time series that adequately captures risks from implementing mitigation policies. One approach is to leverage the fact that policy events that potentially signify shifts in climate policies are often covered by newspapers. In fact, news may even serve as the primary source of information for investors to formulate their subjective probabilities of climate risks. Studying the responsiveness of share prices to news coverage of climate policies offers important advantages over performing event studies around major climate-policy events. (Bhattacharya et al., 2009; Cahan et al., 2009) In particular, an event-study methodology would not allow the unexpected components of policy shocks to be easily disentangled, nor would it allow measurement of the intensity of, or sentiment IMF WORKING PAPERS Decomposing Climate Risks in Stock Markets INTERNATIONAL MONETARY FUND 4 towards, these shocks. On the contrary, news text, which has increasingly been used in financial economics research, captures unexpected new information that is just acquired by investors and foretells changes in future investment opportunities. News articles encompass diverse narratives and are timely and focused on the risks most pertinent to the market, and in the spirit of canonical asset pricing, news coverage may serve as the best of all publicly available signals. (Huang et al., 2019; Barkema et al., 2021) We propose a novel, machine learning-based method to investigate how market-wide climate mitigation risks are priced in Canadian stocks. We use state-of-the-art Natural Language Processing model to perform textual analysis of news on climate mitigation policies that appeared in Financial Times over 2005-2022. Risks from mitigation are identified through a combination of detailed narrative analysis and supervised learning algorithms. A set of asset pricing models are applied to Canadian oil and gas companies listed in the S&P TSX composite index to examine whether model-generated climate risk factors are associated with positive/negative risk premia. Finally, we explore whether Canadian firms are more sensitive to news on climate mitigation by comparing them with representative US and EU firms. Our results provide evidence that stock prices of oil and gas companies incorporate information about climate mitigation policies. To avoid neutralizing positive and negative news labels, we generate two climate news indices, with one signaling positive news for climate mitigation and thus lower transition risk, and the other symbolizing negative news for climate mitigation and thus higher transition risk. We hypothesize that an increase in the positive (negative) climate factor signals stricter (lighter) mitigation policies, and thus should be bad (good) news for oil and gas companies. In response to such a negative (positive) shock, investors would sell (buy) oil and gas stocks, thus decreasing (increasing) their prices and increasing (decreasing) their returns. Consequently, we should observe a statistically significant coefficient on the climate risk factor. The results confirm our hypothesis, indicating that firms have started to incorporate climate factors in their portfolio construction. The responses of stocks in the oil and gas sector to positive and negative news about mitigation policies are asymmetrical. While there is a significantly positive coefficient associated with climate-unfavorable news among the group of oil and gas companies, we find that the sensitivity tied to climate-favorable is negative but statistically insignificant. In other words, an ease in climate mitigation constitutes a favorable shock to oil and gas companies but tighter mitigation policies do not necessarily signal a negative shock. The results are robust to various asset pricing models, including market model, Fama-French three factor model, Fama-French five factor model, Carhart four factor model. Canadian firms are slightly more sensitive to both climate favorable and unfavorable news than US and EU firms. The efficiency and extent of asset pricing could vary across markets. We conduct a cross-country comparison by performing the same set of exercises on US and EU companies. In the baseline model, we find that the coefficients on the climate risk factor of US and EU companies are slightly smaller in magnitude, relative to that of Canadian companies. Overall, US and EU oil and gas companies also respond to climate policy news, but with different degrees of sensitivity from those based in Canada. Our study makes an original contribution to the literature. First, we exploit news data to establish a novel indicator of market-wide climate risks. Previously, Engle et al. (2020) relied on news data to measure climate risks and design hedging strategies. Building on their work, Faccini et al. (2021) further disentangled different dimensions of climate risks using a narrative method. Our paper combines the merits of both studies. We IMF WORKING PAPERS Decomposing Climate Risks in Stock Markets INTERNATIONAL MONETARY FUND 5 present evidence on what types of climate risks are priced, enabled by a novel method that combines manual labeling and machine learning algorithms. Second, we document to what extent oil and gas sector stocks are exposed to climate risks, demonstrating how traditional asset pricing approaches can inform climate finance. In addition, we perform cross-country comparison on climate risk pricing, which offers important insights into which markets are more exposed to climate policy risks. Taken together, using machine learning-based news measures, we show that news on climate mitigation has been partially priced in the stocks of Canadian oil and gas companies. An ease in mitigation policies represents a positive shock to their stock prices but stronger mitigation policies does not necessarily lead to negative price effects. These findings suggest that climate mitigation risks captured by news coverage have limited, asymmetric impact on stock valuations. The rest of the paper is organized as follows. Section 2 reviews the literature on climate risk and asset pricing. Section 3 introduces our methodology and data. Section 4 presents and discusses our results. Section 5 provides robustness checks, and Section 6 concludes. IMF WORKING PAPERS Decomposing Climate Risks in Stock Markets INTERNATIONAL MONETARY FUND 6 2. Literature Review 2.1 Classification of Climate Risks It is common to classify climate risks into physical risk and transition risk. The former is defined as risks arising from physical impact of climate change, whereas the latter refers to risks resulting from policy, technology, legal, and market changes that occur during the move to a low-carbon economy. (IMF, 2022) The impacts of climate risk drivers, whether physical risk or transition risk, on economies and financial markets may vary widely depending on geolocations and become increasingly hard to predict. Physical risk drivers can be further classified into acute risks associated with extreme weather events (Network , 2019), and chronic risks related to gradual climate shifts (McKinsey Global Institute, 2020). It has been estimated that natural disasters, most of which climatological, led to roughly $5.2 trillion losses from 1980s to 2018 (Munich Re, 2020). Transition risk drivers are global, although the specific nature of each risk driver may vary by economy. Various stress test results indicate that the losses for financial institutions in the event of a disorderly energy transition could be sizeable. While banks and the non-bank financial sector have been affected by, and have therefore closely monitored, these forms of changes linked to climate transition, the synchronous nature of transition- related changes have the potential to result in a scale of impact that is much greater than previously anticipated. (Netherlands Bank, 2018) In this study, we focus on climate transition risk, particularly the risk from mitigation policies, hoping to shed light on the effect of ch