The Effect of IPCC Reports and Regulatory Announcements on the Stock Market

Abstract

This study explores U.S. public companies’ reactions to scientific announcements by the IPCC (Intergovernmental Panel on Climate Change) with respect to updated climate change knowledge and how it affects their stock valuations, given their carbon emission/environmental outlooks. Based on a sample of total daily returns collected for 10 industry indexes from the S&P 500 Index over the period 1990–2014, and using an event study approach, we analyze the connection between IPCC assessment report announcements and firms’ returns to evaluate panel data models. We found that various sectors, regardless of their carbon profiles, react abnormally to IPCC report announcements without remarkable long-run cumulative effects. The implications of these results are that there is no clear violation of the efficient markets hypothesis, yet short-term profits may be gained. Furthermore, the market still reacts to new scientific announcements, even though 24 years have passed since the first IPCC report. In addition, there is a negative relationship for low and medium carbon-intensive industries, especially in the short term.

1. Introduction

The importance of global warming on the economic and political agenda was highlighted by the United Nations Climate Change Conference (COP 21) in 2015, where most nations committed themselves to reduce greenhouse gases (GHG) (namely 189 countries, representing 96% of global GHG emissions). The warming of the climate system is explored by the Intergovernmental Panel on Climate Change with the backing of the UN [1], and the most prominent world scientists continue to contribute to climate science/knowledge, with the publication of IPCC reports once every 5–7 years. This issue is not neglected by the financial sector, as climate change affects financial stability.

The complexity and relevance of this issue are due to the collective nature of the problem, as carbon dioxide, once emitted, becomes a global negative externality by contributing to the common atmosphere with uneven temperature increases, since the formation of the greenhouse effect begins with CO₂ particles being trapped in the atmosphere. This, in turn, melts the snow, thereby increasing the radioactive forcing, as sunlight energy is not reflected back into space. The possible implications of such temperature increases are floods (rising ocean/sea level), melting of the permafrost (contributing to further GHG releases), extreme weather, mass migration, and political unrest. Hirabayashi et al. [2] documented how vulnerable our urban settlements are to, not only rising sea levels, but also to river levels, affecting all human-inhabited continents the longer global warming persists.

This negative global externality of CO₂ emissions has been complex to manage, as international agreements like the Kyoto Protocol and intergovernmental conferences on climate change are governed by voluntary participation and, most notably, the global industrial boom by the majority of nations seeking to boost their production at the expense of environmental pollution. Therefore, since the effects of climate change have become progressively more imminent and higher on the corporate agendas of industrialized nations due to a recent corporate social responsibility (CSR) trend, we intend to measure U.S. public companies’ reactions to scientific announcements by the IPCC in respect to updated climate change knowledge and how this knowledge affects their stock valuation given their carbon emission/environmental outlook.

Furthermore, applicable U.S. federal regulations impacting the carbon footprint of companies will be assessed. This analysis will also be performed as an event-study to measure stock market reactions to U.S. federal environmental policies.

Given the notoriety of this issue, the financial sector (and pension funds especially) are interested in the long-term exposure of their equities to weather changes/carbon policies. Thus, in this study, we explore the stock market’s reactions to the announcements from the IPCC reports and compare them to regulatory announcements on climate and energy policies.

If the results of such climate shock events show a negative impact on carbon-emitting companies, this would reveal the prevalence of stranded assets (e.g., oil reserves on the balance sheets), not only in the fossil fuels sector but also in the non-fossils fuel sector, which is dependent on current carbon-intensive technologies, such as the manufacturing and electricity sectors. Such findings will be of interest to mutual funds with long-term strategic investments in green or decarbonized indexes, which will benefit in the case of new (unexpected) scientific announcements, given that carbon risk is still underpriced by the market.

For regulators of the energy sector in the U.S. (stock data of publicly traded firms in NASDAQ/NYSE are analyzed), the present research will be of high importance. If it turns out that markets react more to IPCC announcements (based on environmental policy) than to actual regulatory changes in the energy sector, this would mean that that the IPCC announcements have a strong market signal and high forward guidance. Such announcements are perceived by markets as having a higher effect (i.e., more exogenous shocks) than actual regulatory changes, for which the price of polluting has already been accounted for. Thus, the study contributes to the existing literature by analyzing the difference between new information releases and policy changes in the agenda and by testing empirically market efficiency in relation to these two types of events on the U.S. market.

The paper is organized as follows. Section 2 contains a review of previous studies of the relationships between new information about climate changes or new regulations and market returns. This review is the basis for our hypothesis development. We test the market reaction of the US market to information about new U.S. federal regulation acts or new IPCC announcements. Section 3 describes the methods of information collection and data sampling, as well as data-processing within the event-study approach. Section 4 shows the results of hypotheses testing and is followed by the discussion and the findings analysis in Section 5.

2. Literature Review and Hypotheses Development

Many researchers explore the relationship between climate change challenges (e.g., carbon emissions, cleaner energy production, energy efficiency) and financial markets development. There is no consensus between academics on the nature of such a relationship. Though there is evidence that financial markets reward greener and more responsible companies [3] with higher returns and a lower cost of capital [4] or better operational performance [5], there are also studies that challenge these ideas. The main problem is that to adjust to new information about climate change and to new regulations, companies should invest more in new, cleaner technologies (or to be ready to be penalized by the states and the markets [6,7]). The outcomes of new technology investments are long and uncertain; moreover, more responsible behavior also leads to a substantial growth of operational costs [8,9]. These costs may be strengthened or softened by government policies and differ from industry to industry [10]. Considering these issues, some researchers conclude that the positive effects of “greening” can be assessed only in the long term [11]. However, for asset pricing and investment strategy elaboration, it is important to know how the market reacts to significant information on climate change that would define the policies for the future.

Beatty and Shimshack [12] assess how the release by a non-profit organization of the firm-level ratings of companies’ plans for measuring, reporting, and reducing greenhouse gas emissions affects capital markets. The results obtained based on the availability of new information show that poorly rated firms’ market valuations fell between 0.6 to 1.6 percent (p. 2), but there was no significant impact on the firms that scored high in their ratings, i.e., no rewards for good performers and no penalties for bad ones. Also, as noted by Portney [13], current studies that attempt to link environmental and firm performance do so (in the U.S.) via “the rate at which firms have reduced their emissions of substances on the EPA’s Toxic Release Inventory (or TRI)” (p. 268). However, the limitation of such studies, when assessing decarbonization and TRI use, is that they do not include the most common air pollutants responsible for premature mortality and only assess readily available data, which ignores a whole range of other possible outcomes related to climate change. Therefore, this study will evaluate multiple events, namely five releases by the IPCC, and regulatory environmental changes and assess whether there is a consistent reward or penalty for industries with varying degrees of CO₂ emissions. Also, unlike TRI assessments (like in [12]), the IPCC reports provide greater detail on not only the current state of the environment but also possible future scenarios. Hence, the exogenous shock generated by the IPCC announcements will quantify the market’s susceptibility to how well current firms are prepared for future environmental challenges.

In a recent study, Hjort [14] assessed related existing literature on the climate risks from global warming and its implications for financial markets; Hjort claims that there is no literature on the “event studies that estimate the market reaction to updated information about the scientific facts on climate change” (p. 54). To deal with this problem, our research intends to assess the impact of the IPCC reports, which have wide implications for policy-making. However, since we analyze the returns of the same companies over a period of several years based on a similar event—the announcement of new knowledge on the climate—we expect the findings to have a high cross-sectional correlation [15] since the event day will be the same for the sample firms. This is due to measuring the same companies’ responses to the updated and new (albeit with similar implications for future decarbonization) information.

Furthermore, one of the possible contributions would be empirically testing how efficient financial markets are with respect to prices reacting instantaneously to newly available information and how such new information is valued, such as whether it has already been anticipated and incorporated or whether has no profound effect in the present. This study, then, will add to the existing literature as a test of market efficiency; if “systematic nonzero abnormal security returns persist” after announcements [16] (p. 4), this would imply inconsistency with market efficiency.

In this sense, there are several studies that assess the increased probability of prospective climate legislation in the Australian market. Ramiah et al. [10] examined the effect of 19 announcements of climate regulation on public companies listed on the Australian Stock Exchange. Their most significant finding is that the Australian stock market is mainly sensitive “to the carbon pollution reduction scheme (CPRS) announcement” (p. 1), while other announcements demonstrate a more sector-by-sector heterogeneity. Also, the different reaction for different sectors of the economy is fixed. The study by Chapple et al. [7] focuses on proposed announcements with respect to the carbon emission trading scheme (ETS), with the greatest number of carbon-intensive firms reacting negatively to the increased chance of ETS inception. The paper by Qian et al. [17] focuses specifically on the stages of carbon regulation in Australia and reveals that policy uncertainty and rapid legislative changes cause the weakening of market responses to carbon policy changes; at the same time, the market becomes more sensitive to the “public call for reward bettering carbon performers” (p. 25). Although these findings are made for the Australian market, they seem global by nature (as inconsistency in policies is quite common), and it is interesting to check their viability for other developed markets, the U.S. market among them. Considering the importance of the U.S. market for global production and consumption, we test the market reaction both to global releases of new information and to federal legislative changes.

One of the possible limitations of an event study approach is cross-sectional correlation. However, risk adjustment is not as prominent in short-term event studies as it is in long-term event studies, as the errors in abnormal returns being influenced by errors in the adjustment of risk are small [16]. Therefore, some simple risk adjustment is required, similar to the type proposed by Brown and Warner [18].

Such cross-sectional dependence may be resolved with the use of a Jensen alpha regression. However, as investigated by Lyon et al. [19], this approach is wrongly specified in non-random samples, such as the sample we intend to use. Moreover, if the event dates are not clustered, which is particularly true for IPCC report announcements (nearly every five years), “ignoring dependence induces little bias in variance estimates” [18] (p. 20), and any adjustment may, in fact, be harmful.

Based on the analyzed literature, we developed several hypotheses. It should be noted that since we test the market reaction, alongside our perception of this reaction, we also foresee null hypotheses claiming the absence of the reaction. To confirm the hypothesis, one should observe a statistically significant difference from the null hypothesis.

Hypothesis 1 (H1).

The effects of the IPCC assessment report announcements have a higher impact on firms’ returns than changes in the regulatory framework.

This statement is driven by the consideration that such assessment reports’ announcements, since they are performed by an intergovernmental panel, will have worldwide implications and set a direction for future mitigation and adaptation, apart from simply providing scientific facts explaining the mechanisms of anthropogenic climate change [20]. Therefore, the IPCC reports’ implications will be unforeseen by the market, as opposed to the regulatory framework, which might have already been incorporated by carbon-emitting companies, i.e., anticipated by the market (into their prices), since they are a result of the outlined actions that were previously published by the IPCC.

Possibly, however, over the last 24 years, the IPCC reports may have reiterated the same message: that the world needs to decarbonize and that the so-called tipping point of the new climate environment is ever more imminent. Thus, the null hypothesis states that the effects of the IPCC assessment report announcements do not have a higher impact on firms’ returns than changes in the regulatory framework, since announcements are only of an informative/scientific nature, unlike regulations, which are binding and have a direct impact.

Hypothesis 2 (H2).

The effects of every successive IPCC assessment report announcement do not lead to lower market reactions than the previous ones.

Pindyck [21] assessed the validity of integrated assessment models (IAMs), which predict how GHG emissions directly impact temperature and thus influence economic well-being. Pindyck’s empirical assessment of IAM models invalidates them, which arguably makes updated climate knowledge of little economic value. Therefore, carbon emitters might already anticipate the new policies outlined in the coming IPCC report or disregard the issue of climate change entirely due to climate change’s very long-term projections and a high degree of error in climate predictions/changes. Moreover, markets may be skeptical in believing that the information provided in the reports is likely to happen, as such information is communicated with degrees of uncertainty ranging from very unlikely to very likely. Budescu et al. [22] claim that the verbal method of probabilities does not match the intended severity offered by the IPCC, even if we disregard differences in confidence about future climate change driven by, for instance, age, gender, and political inclinations. Hence, markets do not treat such reports as external shocks but are instead more “surprised” by the actual unanticipated shock of the regulatory environment. Therefore, the null hypothesis states that the effects of every successive IPCC assessment report announcement lead to lower market reactions than the previous ones.

Hypothesis 3 (H3).

Industries with the highest carbon intensity react moderately to the scientific announcements compared with less intensive sectors.

This study will empirically test the notion introduced by Fisher-Vanden and Thorburn [23] that voluntary environmental reports do not damage the most polluting firms, since they already have a high probability of decarbonization/tougher legislation, but instead affect relatively less polluting companies. Similarly, we will test the S&P 500 industry indices on scientific and legislative announcements; we expect a similar reaction of the market to the most polluting vs. less polluting firms.

However, as Caldecott and McDaniels [24] report, if the IPCC reports/environmental regulation announcements are unanticipated by the market and give rise to a high probability of extreme weather events hampering the extraction process of fossil fuels, the impact on high-carbon emitters will dominate due to stranded assets; therefore, the null hypothesis states that industries with the highest carbon intensity have a greater reaction to scientific announcements compared with less intensive sectors, since it is the most polluting firms that face an immediate market reaction.

3. Materials and Methods

3.1. Data Collection

Although there is overall data scarcity on climate-related subjects, the U.S.’s largest public companies, since 2000, have tended to disclose their GHG-related emissions. Therefore, this study will concentrate on the U.S. market, which is perceived to be much more efficient than the other markets, as empirically tested by Zhou et al. [25]. Hence, less mispricing is expected to be found. We focus on the 10 industries of the S&P 500, as identified by the industry classification benchmark (ICB) [26]. This categorization of companies will allow for a comparison of abnormal returns across industries, in contrast to the carbon intensity of industries provided by the Investor Responsibility Research Center Institute [27], with GHG emissions normalized by revenue (Figure 1). In Figure 1, ICB super sectors are used. We consider S&P 500 to be representative of the global market as a whole due to the market capitalization of traded firms being—as of 2019—30% of total global GDP.

As a first step for this study, we collected the total daily returns for the 10 industry indexes of the S&P 500 Index that are the closest proxies for the 10 industries outlined by the ICB. The necessary criterion for inclusion in the sample was an inclusion in the S&P 500 index with the longest possible time span for every security, thus covering the time frame of the publishing of the IPCC reports between 1990 and 2014. Thus, as a main sample, a so-called “fixed” sample will be used with invariable numbers of firms for the industries. This is done in order to test how incumbents, which had a relatively long history of being exposed to scientific announcements, react to IPCC reports. The limitation here is that the sample size and statistical significance of the results lack a high number of degrees of freedom, and the t-statistic may be invalid for a small sample size. However, were the analysis performed on functioning firms at 17 different announcements at a time, the dataset would have included some firms that have merged or gone bankrupt. These data, then, would not have presented a clear picture of how incumbents (running firms as of 2019) react to scientific news. This selection is based on how current “winners” (i.e., companies that have not reorganized or filed for bankruptcy since the 1990s) continue to stay “winners”. Therefore, by tracking the performance of “winners” only, one can obtain a clear picture of relevant past reactions, which can serve as a proxy for the reactions of successful firms to future related announcements. An improvement of this study could involve an analysis of discarded firms and whether IPCC news played a factor in their demise.

As seen in

Table 1. S&P 500 and ICB industries companies.

No.	ICB Industry	Number of Firms	Carbon Intensity
1	Utilities	63	High
2	Oil & Gas	65	High
3	Basic Materials	66	High
4	Industrials	27	Medium
5	Financial	64	Medium
6	Consumer Services	63	Medium
7	Consumer Goods	73	Medium
8	Telecommunication	20	Low
9	Technology	33	Low
10	Health Care	26	Low

(describing S&P 500 and related ICB industries), the sample contains 500 firms on average per year, with the mean number of firms for the industry being 50. In Figure 1, there are 19 super sectors that have the following characteristics: the highest 5 industries (from Figure 1) have high (H) carbon intensity, the subsequent 8 are medium (M), and the bottom 6 are low (L). We map them into 10 ICB industries, as seen in Table 1.

Our next task is to identify the appropriate event dates (announcement dates) for IPCC report publications/unveilings to be considered new, unexpected information for the stock market that was not already incorporated into the price.

In order to correctly identify the appropriate announcement date of the report unveiling, the procedure identified by Schäfer et al. [28] was followed. In their event study methodology for event identification (national regulatory reforms of the financial sector in the post-financial crisis period), instead of using the signing into law of the reform as an event date, they use prior sub-events by analyzing the front-page headlines of national newspapers. Similar to their study, the LexisNexis database offers news related directly to the announcements from the IPCC panel—preceding, during, or after the plenary sessions. However, since global warming/new scientific discoveries are not very popular topics, in our study, an article in one of the “major world publications” is a criterion for selection as a sub-event. Although these so-called “major world publications” may not be available to traders involved in the NASDAQ or NYSE markets, which are primarily based in the U.S., they act as a good proxy for information exposure; since there are other means for investors to be exposed to relevant information, including, but not limited to, the Internet, news bulletins, and word-of-mouth, “major world publications” act as an indicator that such news could have been accessed/assessed by investors.

Since the IPCC announcements by the panel are made at the end of one of the plenary sessions of their working groups, every IPCC report will be analyzed for the relevant date of the session. Since such meetings usually last for four days, and the findings are finalized with high media coverage, the “leakages” of the main findings are likely to be fresh; thus, the start of the symposium is treated as an event date. This event is then compared to the sub-event date of the “major world publications”, and if this sub-event occurs earlier, only then is it treated as an event day. The main findings of this analysis are presented in

Table A1. IPCC reports, dates, and publications.

IPCC Report	WG Session Start	Source (Major World Publication)	Sub-Event Date	Title of the Publication
1st Report
(a) Working Group I: Scientific Assessment of Climate Change 11	24 May 1990	Guardian, UK	22 May 1990	UN warns of global time bomb: Greenhouse effect set to redraw world map in 60 years
(b) Working Group II: Impacts Assessment of Climate Change 2	5 June 1990	Independent, UK	1 June 1990	Scientists increase pressure for curbs on global warming
(c) Working Group III: The IPCC Response Strategies (merged with the second one (b)	5 June 1990	The New York Times, USA	2 June 1990	Critics Say Draft Report on Climate Shift Ignores Urgency of Problem
2nd Report
(a) Report of Working Group I: The Science of Climate Change (IPCC SAR WG1 1996). 3	27 November 1995	The Daily Mail, UK	27 November 1995	WARMING TO THE IDEA THAT MAN’S A DANGER
(b) Report of Working Group II: Impacts, Adaptations and Mitigation of Climate Change: Scientific-Technical Analyses	no date identified
(c) Report of Working Group III: Economic and Social Dimensions of Climate Change 4	25 July 1995	The Globe and Mail, Canada	29 July 1995	ANOTHER PERSPECTIVE Off base on greenhouse report
(d) The “Full Report” 5	11 December 1995	The Herald Sun, Australia	13 December 1995	GLOBE WARMING HEAT ON MAN
3rd Report
(a) WGI: Scientific aspects of climate (see IPCC TAR WG1 2001). 6	17 January 2001	The Independent on Sunday, UK	21 January 2001	WARMING—IT’S TWICE AS BAD AS WE THOUGHT
(b) WGII: Vulnerability, consequences, and options (see IPCC TAR WG2 2001). 7	13 February 2001	The Ottawa Citizen, Canada	18 February 2001	World’s scientists fear global warming fallout: Mankind facing greatest climate upheaval since Ice Age: report
(c) WGIII: Limitation and mitigation options (see IPCC TAR WG3 2001). 8	28 February 2001	Modern Power System, USA	28 February 2001	COMMENT; REALISM AND RENEWABLES
(d) Task Force: National Greenhouse Gas Inventories Programme. 9	24 September 2001	Western Daily Press, UK	25 September 2001	Climate will “cost billions”
4th Report
(a) Contribution of Working Group I (WGI): Climate Change 2007: The Physical Science Basis. 10	29 January 2007	The Irish Times, Ireland	3 February 2007	A challenge to mankind
(b) Contribution of Working Group II (WGII): Climate Change 2007: Impacts, Adaptation and Vulnerability. 11	6 April 2007	Africa News, Congo	6 April 20072,3	Africa; Worldwide Impact from Climate Change Predicted
(c) Contribution of Working Group III (WGIII): Climate Change 2007: Mitigation of Climate Change. 12	30 April 2007	The Nation, Thailand	5 April 2007	Report “a step backward from Kyoto”
(d) Contribution of Working Groups I, II, and III: The Synthesis Report (SYR). 13	12 November 2007	Christian Science Monitor, USA	15 November 2007	Climate summary fuels worry
5th Report
(a) WG I: The Physical Science Basis. 14	23 September 2013	Canberra Times, Australia	28 September 2013	Global warming: humans to blame
(b) WG II: Impacts, Adaptation and Vulnerability. 15	25 March 2014	The Washington Post, USA	31 March 2014	New call for action on climate
(c) WG III: Mitigation of Climate Change. 16	7 April 2014	Africa News, Congo	13 April 2014	Climate; IPCC Presents Assessment on Measures to Mitigate Climate Change
(d) AR5 Synthesis Report (SYR). 17	2 November 2014 3	UN News Service, New York	2 November 2014	Leaders Must Act’, Urges Ban, As New UN Report Warns Man’s Impact on Climate May Soon Be “Irreversible”

¹ The announcement number and notes are marked in bold. ² The bold-formatted date indicates the announcement date. ³ As the event day falls on a non-trading day, the next trading event is chosen: 6–9 April 2007, 2–3 November 2014.

, which lists the IPCC reports with their respective news announcements if such announcements exist, with highlighted dates treated as an event day for the analysis. Moreover, if the date of publication/session commencement falls on a non-trading day, then the closest successive trading day is treated as an event day.

After identifying the dates at which publication was announced (t = 0), 120 days prior returns to t = 0 and 10 days after the announcement returns are collected, yielding a total return of 121 days for each of the ICB industries. Five reports were given over these years, with multiple working groups being responsible for (in most cases): (a) the physical science basis: (b) impacts, adaptation and vulnerability; (c) mitigation of climate change; and (d) synthesis report; thus, this procedure will be repeated for a total of 17 times (IPCC announcements only).

Moreover, as a contrast with the IPCC announcements, we also conducted the following procedure for the announcements of new regulations on carbon and energy with the same firms. The LexisNexis database was used to find appropriate federal regulation announcements in the U.S. Moreover, because there is a chance that some other major event might have occurred on the announcement day of the new legislation/report, confounding event checks were completed.

In order to contrast the discoveries of scientific information, the following list of federal U.S. regulations dates (implemented/received the partial support of the Congress) will be added with the appropriate event dates:

• American Clean Energy and Security Act of 2009 [29];
• Clean Air Act of 1963 [30].

For the American Clean Energy and Security Act of 2009, the appropriate dates were identified as follows:

• It was offered as a “discussion draft” in the House Committee on Energy and Commerce on 31 March 2009 (Regulation 1).
• The Act was introduced in the House of Representatives as H.R. 2454 by Henry Waxman (D-CA) on 15 May 2009 (Regulation 2).
• On 21 May 2009, the bill passed out of the Energy and Commerce Committee by a vote of 33 vs. 25, largely falling along political party lines (Regulation 4).
• The bill passed the House on 26 June 2009 (219 vs. 212) at 11.21 a.m. (Regulation 3).

Since the bill did not pass through the Senate to later become law, it is natural to look for an event where it was rejected decisively by the Senate, which would have hypothetically increased the stock valuations of carbon-emitters. However, upon careful examination of the LexisNexis database and related news articles after the House of Rep. passage, there has been no definitive event of outright rejection by the Senate, as no vote has yet taken place.

The second federal act closely related to the environmental/climate change agenda is the Clean Air Act of 1963. Although there have been a number of amendments to the bill, namely 1970, 1977, and 1990, with greater federal authority in the field of air quality, the scope of the law does not cover enough industries in the U.S. economy to make an evaluation of announcements for the whole S&P 500 index.

3.2. Event Study Approach

In this research, a standard event study methodology is used. First, the daily returns of the companies were obtained in the sample, with Fama–French three-factor loadings [31]) used for the dates of the IPCC sub-events and four regulation announcements. Daily factor loadings were downloaded from the open resources on the Kenneth R. French website [32] on 28 March 2019.

The daily returns are computed as follows:

$$R_{i,t}=\frac{P_{i,t}-P_{i,t-1}}{P_{i,t}}\ast 100$$

(1)

The return for a stock is explained by the three factors developed by Fama and French [31] and a risk-free rate, as illustrated in (2). These three factors are market return, the high minus low (HML) average of value minus the growth portfolio, the SMB-average return on nine small stocks minus big stock portfolios. The choice of the three-factor model is due to the empirical findings by Fama and French [31] that the market return and the book to market ratio are responsible for most of the cross-section of average stock returns.

$${\mathrm{R}}_{\mathrm{i},\mathrm{t}}={\mathrm{Rf}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_{\mathrm{i}}\ast {\mathrm{r}}_{\mathrm{m},\mathrm{t}}+\mathrm{SMB}\ast {\mathrm{s}}_{\mathrm{i},\mathrm{t}}+\mathrm{HML}\ast {\mathrm{h}}_{\mathrm{i},\mathrm{t}}.$$

(2)

For every event, i.e., for every company, the coefficients${\mathrm{a}}_{\mathrm{i},\mathrm{t}}$,${\mathsf{\beta }}_{\mathrm{i},\mathrm{t}}$,${\mathrm{s}}_{\mathrm{i},\mathrm{t}}$, and${\mathrm{h}}_{\mathrm{i},\mathrm{t}}$ are estimated by regressing (${\mathrm{R}}_{\mathrm{i},\mathrm{t}}-{\mathrm{R}}_{\mathrm{f},\mathrm{t}}$) on (${\mathrm{r}}_{\mathrm{m},\mathrm{t}}$, SMB, HML) in the time period of (−121 to −11) days prior to the announcement. The calculation is as follows: 10 ICB industries with 17 events based on the dates from Table A1 (see Appendix A). Therefore, in total, the number of observations is 10,500 (21 events * 500 companies), and regressions will take place with the residual term (${\mathrm{e}}_{\mathrm{i},\mathrm{t}})$ over 100 days being saved for further analysis (Equation (3)).

$${\mathrm{R}}_{\mathrm{i},\mathrm{t}}-{\mathrm{Rf}}_{\mathrm{i},\mathrm{t}}={\hat{\mathrm{a}}}_{\mathrm{i},\mathrm{t}}+{\hat{\mathsf{\beta }}}_{\mathrm{i},\mathrm{t}}\ast {\mathrm{r}}_{\mathrm{m},\mathrm{t}}+\mathrm{SMB}\ast {\hat{\mathrm{s}}}_{\mathrm{i},\mathrm{t}}+\mathrm{HML}\ast {\hat{\mathrm{h}}}_{\mathrm{i},\mathrm{t}}+{\mathrm{e}}_{\mathrm{i},\mathrm{t}}.$$

(3)

Having obtained ${\hat{\mathrm{a}}}_{\mathrm{i},\mathrm{t}},{\hat{\mathsf{\beta }}}_{\mathrm{i},\mathrm{t}},{\hat{\mathrm{s}}}_{\mathrm{i},\mathrm{t}},{\hat{\mathrm{h}}}_{\mathrm{i},\mathrm{t}}$ for each stock and using Equation (2), the expected returns are subtracted from the actual returns, as demonstrated in Equation (4), to obtain abnormal returns (AR). This procedure was repeated for every stock in the sample for a related announcement date.

$${\mathrm{AR}}_{\mathrm{it}}={\mathrm{R}}_{\mathrm{i},\mathrm{t}}-{\hat{\mathrm{R}}}_{\mathrm{i},\mathrm{t}}.$$

(4)

From Equation (5) onwards, the statistical significance of the excess returns is analyzed using the method developed by Loderer and Mauer [33], who use Z-statistics.

As the first step using Equation (5), the variance of the abnormal return for an individual stock is computed, where ${\mathrm{s}}_{\mathrm{i}}^2$ is the residual return variance obtained by estimating the market model for a period of 100 days prior to the event.${\mathrm{R}}_{\mathrm{m},\mathrm{t}}\mathrm{and}{\overline{\mathrm{R}}}_{\mathrm{m}}$are the returns on the market index and the mean return on the market, respectively.

$$\mathrm{Var}({\mathrm{AR}}_{\mathrm{i},\mathrm{t}})={\mathrm{s}}_{\mathrm{i}}^2\left[1+\frac{1}{100}+\frac{{\left({\mathrm{R}}_{\mathrm{m},\mathrm{t}}-{\overline{\mathrm{R}}}_{\mathrm{m}}\right)}^2}{{{\displaystyle \sum }}_{-121}^{-11}{\left({\mathrm{R}}_{\mathrm{m},\mathrm{t}}-{\overline{\mathrm{R}}}_{\mathrm{m}}\right)}^2}\right].$$

(5)

Having estimated this variance, Equation (6) estimates the so-called standardized excess return on a security by dividing the standard deviation of abnormal returns by the abnormal return for the announcement period:

$${\mathrm{SAR}}_{\mathrm{i},\mathrm{t}}=\frac{{\mathrm{AR}}_{\mathrm{i},\mathrm{t}}}{\sqrt{\mathrm{Var}({\mathrm{AR}}_{\mathrm{i},\mathrm{t}})}}.$$

(6)

We then measure the cumulative performance of all of the securities in the sample over the intervals [${\mathrm{t}}_1,{\mathrm{t}}_2$] ranging from CSAR (0; +1) to CSAR (−10; 10) to determine if there was consistent mispricing that violated the EMH. Thus, Equation (7) calculates the cumulative standard abnormal return for a stock over the interval [${\mathrm{t}}_1,{\mathrm{t}}_2$]. Here, cumulative and standardized abnormal return normality and independence are assumed to have a df = 100.

$${\mathrm{CSAR}}_{\mathrm{i}}=\frac{{\mathrm{SAR}}_{\mathrm{i},\mathrm{t}}}{{{\displaystyle \sum }}_{\mathrm{t}={\mathrm{t}}_1}^{{\mathrm{t}}_2}\sqrt{{\mathrm{t}}_2-{\mathrm{t}}_1+1}}.$$

(7)

The next step is to compute a Z-statistic for each of the 10 ICB industries according to their numbers of firms. From Equation (8), N varies from 20 to 73, with an average of 50 firms per industry. In the results section, appropriate degrees of freedom will be used based on the sample size of the industries.

$$\mathrm{Z}=\sqrt{\mathrm{N}}\left(\mathrm{CSAR}\right),$$

(8)

where $\mathrm{CSAR}=\frac{1}{\mathrm{N}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{N}}{\mathrm{CSAR}}_{\mathrm{i}}$.

4. Results

In this section, we provide the results of empirically testing each developed hypothesis separately and then join them in the discussion of the results.

4.1. First Hypothesis Testing Results

The first hypothesis of our study explored the difference in market reactions to IPCC assessment announcements and announcements on changes in the regulatory framework. As can be seen from

Table 2. Statistically significant abnormal returns to regulation announcements (Day 0).

ICB Industry	Carbon Intensity	Positive Abnormal Returns	Negative Abnormal Returns	Proportion of Negative Abnormal Returns, %
Industrials	medium	0	1	100%
Oil & Gas	high	1	1	50%
Health Care	low	1	1	50%
Consumer Goods	medium	1	1	50%
Financial	medium	1	1	50%
Basic Materials	high	2	1	33%
Technology	low	2	1	33%
Consumer Services	medium	1	0	0%
Utilities	high	0	0	0%
Telecommunication	low	0	0	0%

, most of the industries have a split reaction to the higher probability of the passage of the ETS bill in the U.S., which is hypothesized to have seriously impacted the heaviest of polluters. However, this is not the case, as highly intensive industries (based on carbon imprint) either failed to react or even reacted positively (Basic Materials) to the progress of the ETS scheme. The reason for this result may be market perception/anticipation that the bill would ultimately fail to become a law and, therefore, that there was no reason to price carbon accordingly.

Table 3. Statistically significant abnormal returns by regulation announcements (Day 0).

Regulation	Positive Abnormal Returns	Negative Abnormal Returns	Proportion of Negative Abnormal Returns
Regulation 1	-	7	100%
Regulation 2	3	-	0%
Regulation 3	5	-	0%
Regulation 4	-	-	0%
Total	8	7	Average = 48%

hints at such a possibility. At first, the market is taken by surprise by the initiation of the bill, with only negative reactions observed across industries. However, this situation is reversed when the bill passes the House of Congress (Regulation 3) with all significant ARs being positive. However, theoretically, the bill is closer to passing than in the Regulation 1 event, requiring approval by the Senate and the President only.

These findings, with respect to the first hypothesis (a greater market reaction to the IPCC than to environmental regulations), are difficult to extrapolate and compare due to the regulations’ applicability only in the U.S., whereas the IPCC is a global forum. Moreover, only four regulation events were identified, compared to 17 in the scientific literature. Although there were no statistically significant cumulative abnormal returns (CARs) for regulation announcements (which was not the case for the IPCC), this homogenous reaction—first negative (Regulation 1) and then positive (Regulation 3) abnormal returns on day 0—means that markets may have been surprised by environmental regulations. Hence, the first hypothesis (H1) is rejected.

4.2. Second Hypothesis Testing Results

The second hypothesis (every successive IPCC report does not have lower economic significance) cannot be rejected based on the evidence from Figure 2 and Figure 3. In the first graph (Figure 2), the total number of abnormal returns per report is calculated with a clear upward trend and the highest total ARs in the report of 2014.

The second graph (Figure 3) similarly depicts the tendency of cumulative abnormal returns per announcement to oscillate around a mean value of 4.6 CAR per report, with the highest instance occurring for the IPCC report in 2014. However, limitations in this analysis may result from the small sample size during the first report, which could have resulted in a greater number of CARs and ARs if more industries were included.

Another notable insight concerning the market reaction to IPCC announcements can be seen in Figure 4, which demonstrates statistically significant negative CARs as a percentage of total CARs according to the IPCC reports.

As seen from the clear upward trend from the first until the last IPCC reports, as the proportion (in percent) of negative CARs steadily increases, it provides evidence for the stock market’s prior overpricing of carbon/weather risk, which is a result of greater scientific knowledge, and the imminence is bid down by the market.

4.3. Third Hypothesis Testing Results

Table 4. Statistically significant abnormal returns (IPCC announcements) on Day 0.

Carbon Intensity	ICB Industry	Proportion of Negative Abnormal Returns	Positive Abnormal Returns	Negative Abnormal Returns
high	Basic Materials	86%	1	6
medium	Consumer Services	67%	2	4
low	Telecommunication	67%	1	2
low	Health Care	56%	4	5
low	Technology	55%	5	6
medium	Industrials	53%	7	8
medium	Consumer Goods	50%	3	3
medium	Financial	43%	4	3
high	Oil & Gas	33%	2	1
high	Utilities	0%	0	0

high = high carbon intensity, medium = medium carbon intensity, and low = low intensity.

presents how each sector reacts (with a 10% significance level at least) to the individual 17 IPCC announcements on day 0 when the scientific findings were said to be unveiled. Surprisingly, mispricing via the predictable method of negative ARs is only observed for two out of the five highly intensive industries, with the utilities industry having no reaction on day 0 across any of the 17 events. Secondly, the fact that medium industries (M) are mostly (five out of seven sectors) featured at the top of the table when ranked based on the highest number of negative abnormal returns during day 0 allows us not to reject H3, since industries with the highest carbon intensity have a lower negative abnormal return to the scientific announcements relative to less intensive sectors, especially on day 0.

The second point of the analysis is concerned with investigating whether the abnormal returns are persistent as a result of IPCC announcements, which would violate the EMH if true.

Table 5. Statistically significant cumulative abnormal returns by IPCC reports ¹.

IPCC Report	Industry	Event Window	CSAR/t-Test	AR (%)
1	Consumer Goods	CAR (−2; +2)	1.92 *,2	1.92
1	Consumer Goods	CAR (0; +1)	1.86 *	3.31
1	Consumer Goods	CAR (0; +2)	2.26 **	2.4
1	Consumer Goods	CAR (0; +3)	1.95 *	3.37
1	Health Care	CAR (0; +2)	1.97 *	0.81
2	Health Care	CAR (0; +3)	1.85 *	3.46
2	Utilities	CAR (0; +1)	2.68 *	0.94
2	Technology	CAR (0; +1)	2.48 ***	1.78
2	Technology	CAR (0; +2)	2.2 **	3.72
2	Technology	CAR (0; +3)	2.08 **	3.78
3	Utilities	CAR (0; +1)	−1.76 *	−1.33
3	Industrials	CAR (0; +1)	2.36 **	3.16
3	Consumer Services	CAR (0; +1)	1.9 *	1.14
4	Technology	CAR (−1; +1)	−2.1 **	−1.06
4	Technology	CAR (0; +1)	−2.06 **	−2.37
4	Basic Materials	CAR (0; +1)	−1.79 *	−1.65
5	Health Care	CAR (0; +1)	−2.26 **	−2.12
5	Health Care	CAR (0; +2)	−1.67 *	−2.6
5	Utilities	CAR (−1; +1)	2.03 **	2.26
5	Industrials	CAR (0; +1)	−2.22 **	−0.67
5	Industrials	CAR (0; +2)	−1.78 *	−1.62
5	Consumer Services	CAR (0; +1)	−2.51 **	−2.94
5	Consumer Services	CAR (0; +2)	−1.78 *	−3.4

¹: with df = number of firms in the industry; ²: significance levels: * 10%, ** 5%, *** 1%.

reports significant cumulative abnormal returns—not those for 17 announcements as before, but for five IPCC reports. Table 5 shows evidence of mispricing and opportunities to make short-term profits for investors. Looking at the number of CARs across the five reports from 1990 to 2014, there are ample opportunities to earn short-term profits due to mispricing, as all reports feature at least three instances of CARs and no more than seven, as is the case with the 5th report. Based on these findings, it is possible to reject the third hypothesis, since the effects of every successive IPCC assessment report announcement have led to similar market reactions as the previous ones.

However, although CARs are common, the way in which the IPCC report influences the relevant sectors reversed from the early 1990s to 2014; the reaction was positive during the first 2 reports, but became negative after that. The reason behind this change may be higher uncertainty of the unveiled scientific information in the 1990s, leading to a disregard for the carbon risk inherent in industries such as utilities, with positive CAR because of the second IPCC report. However, later, as the climate risk became more evident and higher on investors’ agendas, investors realized that the companies with high CO₂ exposure were overpriced and that scientific evidence was mounting against them, leading to negative returns for such stocks in the third, fourth, and fifth reports.

Table 6. Statistically significant cumulative abnormal returns (CAR) by IPCC report.

Carbon Intensity	Industry	Total CARs	Positive CARs	Negative CARs	Proportion of Negative CARs
high	Basic Materials	1	0	1	100%
medium	Industrials	3	1	2	67%
medium	Consumer Services	3	1	2	67%
low	Health Care	4	2	2	50%
low	Technologies	5	3	2	40%
high	Utilities	3	2	1	33%
medium	Consumer Goods	4	4	0	0%

H = high carbon intensity, M = medium carbon intensity, and L = low intensity.

provides a summary of Table 5 and demonstrates the occurrences of sectors that had significant CARs. Similar to the ARs on day 0, discussed in the previous section, here, high and medium (in terms of carbon intensity) industries react negatively to the IPCC. Furthermore, the two sectors (utilities and consumer goods) with the most positive occurrences as the share of total significant CARs give further evidence that, in fact, medium carbon-intensive industries are more overpriced than high carbon-intensive firms; this observation supports the third hypothesis (H3).

5. Discussion

The objective of IPCC reports is to raise awareness of forthcoming changes to the climate, with repercussions not only for the general habitat and societies but also for stock market participants. As we have illustrated, various industries, regardless of their carbon profiles, react abnormally to IPCC report announcements, without remarkable long-run cumulative effects. The implications are that there is no clear violation of the EMH, yet short-term profits may be made. Further, the market still reacts to new scientific announcements even after 24 years have passed since the first IPCC report.

Likewise, environmental regulatory legislation is treated as an exogenous event by the stock market. However, due to the lack of an adequate sample size, no concrete conclusions between the two types of announcements may be made. This finding aligns with the conclusions by Qian et al. [17] about the difference in market responses to new policies and new information releases.

The results from this study are noteworthy, in a practical sense, for the implementation of the decarbonized index proposed by Andersson et al. [34], since their main assumptions of underpriced or correctly priced carbon risk (for low-emitting firms) by investors does not seem to be valid for events like the IPCC report announcements. As demonstrated in the previous section, especially in relation to the last two IPCC reports, day 0 ARs and subsequent CARs were negative, especially for low and medium carbon-intensive industries. Therefore, investors who are minimizing carbon risk in their portfolios by the authors’ techniques should be aware that, empirically, their strategy may backfire (yield negative as opposed to expected positive/neutral returns), especially in the short-term period.

Although there are other sources of scientific discoveries related to climate change that enhance the market’s knowledge about pricing carbon, it is very unlikely that there is a direct stock market reaction to, for example, glacier-melting images released by NASA or other one-off meteorological discoveries. This is because IPCC provides verified information from acknowledged scientists around the globe, while not only presenting the knowledge at its face value but also producing an interpretation for policymakers, offering predictions and courses of action to maintain the status quo.

As for limitations and possible areas of further research, there are a number of paths to take. Firstly, this study concentrated only on the U.S.’s exposure to global scientific discoveries, so a relevant option would be to test other countries’ stock market responses to global and local scientific events.

In addition, this study’s events were all hypothesized to be negative in nature, i.e., both regulations (American Clean Energy and Security Act of 2009 [29]) and all 17 IPCC reports were expected to have a negative effect on the stock returns on the studied companies. Thus, it may be of interest to investigate scientific events that are pro-pollution or are expected to result in higher-than-expected returns, as well as deregulation of the current laws related to the environment.

Lastly, GHG emissions profiles or company/industry carbon intensity are undermined by non-disclosure or a lack of resources to accurately calculate Scope 3 (suppliers and consumers) emissions. An alternative classification for carbon, other than simple GHG/revenue, may result in different industry gradations, altering the composition of high-, medium-, and low-carbon intensity groups. Subsequently, greater disclosure by companies, thanks to an awareness of carbon risk by consumers and investors, would match more closely with the true carbon profile, thus eliminating measurement errors and over-/underperformers from certain industries present in this study.