Do Stock Markets Value Climate Risk
Leadership of Financial Institutions?
Bachelor Thesis Economics & Business – Finance & Organisation Track By Damilola E. B. Ojutiku - Student nr: 10658920
Supervised by Professor Jeroen E. Ligterink February 17, 2017
Abstract
Our findings imply that there are potential short-term gains in financial performance to be made from climate risk policy leadership for UK banks, as positive alpha investment strategists raise stock value, conversely, punishing UK banking laggards. As suggested by Stenek et al. (2010), banks will find themselves playing the role of climate change role models for other financial institutions, and backed by our results, could inspire a sector-wide shift towards greater climate risk policy activity.
This document is written by Student Damilola Ojutiku who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Contents
I. Introduction ... 4 I. I. Research Question ... 4 II. Literature ... 6 III. Hypotheses ... 9 II. I. Hypothesis 1... 9II. II. Hypothesis 2 ... 9
II. III. Hypothesis 3 ... 9
II. IV. Hypothesis 4 ... 9
IV. Data ... 10
IV.I. Sample Summary Statistics ... 10
V. Research Methodology ... 11
V. I. CDP Scoring Methodology... 11
V.II. Regression Analysis ... 13
V. II. i. Regression 1 ... 13
V. II. ii. Regression 2 ... 13
IX. Appendix... 22
IX. I. Appendix 1 ... 22
I. Introduction
In December 2015, at the Paris climate conference, policy makers from 195 countries adopted the first ever economy-wide, legally binding global climate change agreement, at COP21. These countries committed themselves to a long-term limit of the increase in the global temperature of 1.5°C above pre-industrial levels, outlining future developments, such as increased accountability infrastructure, and continued target revision (United Nations, 2015). Changing corporate culture and practices play an important role in achieving the targets agreed upon at COP21, as commodity intensive industries, carbon asset-rich corporations and all the lines of business connected to them face risks, both direct and indirect, from changes in the natural world, but also due to revaluation of assets because of policy changes. Now shareholders and investors increasingly recognise the need for
information about corporate practices concerned with mitigating climate, to protect their financial interests, and more pressure is being applied to this effect (Burton, 2010). Corporations are now expected, and in some cases required to provide climate risk policy information, and work with public policy makers to facilitate a cultural change (Schoenberg, 2016).
Pressure on the corporate world to deal with, and publicise their climate risk practices has been increasingly evident, with experts like Paul Fisher warning of the potential for climate change to cause the next global financial crisis (Cadman, 2016), and Barclay’s energy analysts forecasting potential losses in carbon intensive industries in excess of $30 trillion (Ryan, 2016). The financial sector, however, has largely managed to minimise obligations to climate risk disclosure, as the focus remains predominantly on primary and secondary industries.
The Asset Owner’s Disclosure Project (2015) established a decrease in climate risk policy disclosure by institutional investors between since 2014 (Shankleman, 2016). This is despite recognition of the importance of climate risk policy to the reputation of financial institutions, and the financial value that reputation carries (Stenek, Amado, & Connell, 2010). With Stenek et al. (2010), citing public relations costs, talent acquisition difficulties, and potential decreases in access to financial markets for financial institutions, we choose to focus in this thesis, on the effects on climate risk activity on market valuation of financial institutions.
I. I. Research Question - Do stock markets value climate risk policy leadership of financial institutions?
This thesis uses 2016 Carbon Disclosure Project (CDP) climate risk policy disclosure scores, published in the Autumn of 2016, as a proxy for leadership and straggling in the field of climate risk policy. To control for market risk factors, Carhart’s extended four-factor capital asset pricing model is applied to a trading strategy analysis, through regression analysis, of 151 publicly traded British, French, Benelux, German, Swiss, Austrian, Scandinavian,
Spanish, Portuguese, and Italian financial sector corporations which fall under the jurisdiction of the COP21 Paris agreement, over the six quarters (18 months) from July 2015 to
December 2016. We compare short-term financial market performance of climate risk leaders versus laggards, by forming leadership and laggard portfolios, and calculating abnormal returns for the difference between the portfolios. Although this is not an event study, proximity of the data to the agreement date aids the relevance of our results.
Our sample is split up into United Kingdom and European financial institutions, with further separate analysis of banks in both regions, as some UK and European banks already lead the way in climate change risk management (Stenek et al., 2010), and due to the difference in risk factors for these regions. Results of our analysis, prove to be largely inconclusive, however, we find a tentative implication that a trading strategy in which one goes long in stocks of UK leadership portfolio banks, and sells short stocks of UK laggard portfolio banks generates positive abnormal returns over the 18 months ending December 2016. Furthermore, a trading strategy long in stocks of both European leaders and laggards, generates positive abnormal returns over the same period, with the laggard portfolio generating higher abnormal returns than the leaders. This result is replicated in the European bank portfolios.
These findings provide insight into three previously neglected areas of climate risk leadership in financial institutions, namely: the recent nature of financial performance analysis, in correspondence with newly legitimised climate change regulation, a focus on financial sector institutions, and a focus on the UK financial sector, ranked first in the world (Yeandle, 2016). The findings imply that there are potential short-term gains in financial performance to be made from climate risk policy leadership for UK banks, as positive alpha investment strategists raise stock value, conversely, punishing UK banking laggards. As suggested by Stenek et al. (2010), banks will find themselves playing the role of climate change role models for other financial institutions, and backed by our results, could inspire a sector-wide shift towards greater climate risk policy activity. In contrast, our results suggest European
financial institutions have no downside financial performance risk, associated with climate risk policy lagging. This is indicative of a slow response of European investors to climate change legislative changes, and its implications for European financial institutions.
II. Literature
TU’s (2009) Do Stock Markets Price Climate Change Risks? is a rudimentary research paper focusing on US primary and secondary industry corporations, facing direct climate risk. The paper provides an entry point into the subject, using a single factor regression. TU (2009) asserts that a multifactor regression would likely lead to clearer/more reliable results. The analysis uses CDLI scoring, a previous version of the CDP scores used in this thesis, and it is acknowledged that the CDLI scores had shortcomings as they were in 2009, with regards to being an accurate indicator of the climate change risk management itself (TU, 2009). Consequently, the hypotheses of this thesis are distanced from an assessment of climate management, or the risk itself, and focus on perceived leadership in the field. TU (2009) also uses Climate Change Governance Checklist (CCGC) scores provided by a 2006 Ceres report, but recognises these scores are limited to the behaviour of United States corporations. The paper tests the returns of companies from five carbon intensive industries: electricity, oil & gas, construction, aviation, and automobile in the Eurozone market, citing Europe’s leading role in political climate change activism (TU, 2009). This view is somewhat corroborated by the Paris agreement, which has been in discussion by European policy makers for almost a decade. Our focus on European institutions is motivated partially by the perception of Europe as the global leader in climate change activism, but also ensures more homogeneity in our sample. TU (2009) finds varying relationships and statistical significance between CDLI scores and stock performance between 1992 and 2008, with positive
correlations established for electricity, and automobile industries, but not for oil & gas, while construction showed a very weak positive relationship. The main limitations to the research of TU (2009): very small sample size, CDLI scoring as a measure of management
performance, and single factor regression analysis with control for market risk factors are ones which this thesis aims to mitigate, with a large sample size, Carhart’s four-factor CAPM, and reapplication of an already improved scoring system.
Konar & Cohen (2001) elect to ask whether the market values environmental performance, by using Tobin’s q. They acknowledge the pitfalls of small sample sizes and limited objective environmental and financial performance measures, and elect to analyse the financial
performance of 321 S&P 500 companies, against two environmental measures: toxic chemical releases (TRI88) and number of environmental lawsuits (LAW89). By doing this Konar & Cohen (2001) ensure their results represent their own definitions of environmental and financial performance. Here, regression analysis included control variables that were lacking in TU’s (2009) research, specifically market share, industry concentration ratios, growth, advertising intensity, R&D intensity, firm size, and import intensity, relevant to the varied range of US corporations in their sample. Konar & Cohen’s (2001) sample did not include banking and insurance companies, but their methodology highlights the importance of a multifactor analysis.
Identifying growth as a determinant of financial performance, Konar & Cohen (2001) presented stock growth as a significant control variable. In our research, book-to-market ratios are utilised as a proxy for stock growth (Capaul, Rowley, & Sharpe, 1993), used to calculate the market risk factor ‘HML’.
The methodology used in this thesis comes from corporate governance research by Gompers, Ishii, and Metrick (2003), who compare multiple portfolios containing corporations with differing levels of shareholder rights, by buying stocks of corporations with high ‘governance index’ scores and selling stocks of corporations with low index scores. A relevant refinement of Ishii et al.’s (2003) methodology is presented by Ziegler, Busch, & Hoffmann (2011), who study the relationship between climate change risk policy disclosure and stock performance of European and US stocks. This is done by comparing risk-adjusted returns, calculated via the Carhart four-factor CAPM, and the Fama-French single factor CAPM. Ziegler et al. (2011) calculate and evaluate Jensen’s alpha, in the same way as this thesis, treating it as risk-adjusted abnormal return, employing a trading strategy, of going long on stocks of
corporations with climate risk policy disclosures, and shorting stocks of corporations without climate risk policy disclosures. Given their paper’s focus on climate risk policy disclosure between 2001 and 2006, and a large sample size of 499 European firms, and 515 US firms, including financial sector corporations, the results of Ziegler et al.’s (2011) study are most relevant for forming the hypotheses of this thesis.
The main findings show that the relationship between climate risk policy disclosure and stock performance has been significantly positive for energy firms in the US (Ziegler et al., 2011). This result is interesting, because given the scope of the industries studied and number of firms, only one US sector showed a significant positive value of alpha. Nonetheless, Ziegler et al. (2011) attribute the positive relationship in the US energy sector to increased awareness of the need for climate risk policies, and heightened stringency of climate policy and
institutional pressure. They conclude that the positive relationship between stock
performance and climate risk policy disclosure is stronger in places where climate change policies are more ambitious (Ziegler et al., 2011). Similarly, in Europe Ziegler et al. (2011) establish that their trading strategy, consisting of buying stocks on corporations with climate risk policy disclosures and selling stocks of corporations with no policy disclosures has become more worthwhile over time, and expect these investments to become even more attractive in Europe in the future.
As this thesis focuses on data for 2016 and late 2015 in European markets, using risk-adjusted alphas as indicators of abnormal stock performance, and the increased urgency of European climate policy, capped by the Paris agreement, similar results to the study by Ziegler et al. (2011) were expected.
Ziegler et al.’s (2011) findings are somewhat corroborated by the research of Liesen (2015), who uses a five-factor asset pricing model to examine returns over different industries and generate alphas. Although the focus of this study was to assess the level of market efficiency with respect to climate induced systematic risk, Liesen (2015) also finds that going long stocks of companies with emissions reporting, and shorting stocks of companies without emissions reporting generated a statistically positive alpha. Liesen (2015) takes the analysis further, suggesting that emissions disclosing companies with lower systematic risk, generate higher returns.
Finally, Blyth, Bradley, Bunn, Clarke, Wilson, & Yang’s (2007) study of investment risks under uncertain climate change policy, uses real options methodology to try to quantify regulatory risks and explain the effects of climate risk policy uncertainty on investors’ behaviour. Blyth et al. (2007) focus on investment options in specifically coal and gas-fired power plants, and carbon capture and storage technologies, differing from the financial focus of this thesis, however it is their verdict that climate policy uncertainty creates a risk premium that offers some insight into expectations of the relationship between stock performance and
corporate climate risk activity. Most notably, Blyth et al. (2007) find that in close
chronological proximity to climate policy changes, climate risk policy uncertainty becomes a dominant risk factor. Subsequently the required return on projects increases with policy uncertainty. Blyth et al.’s (2007) findings, provide potential explanation for seeming lack of positive abnormal financial performance in this thesis, in the relatively short time-frame of our analysis with regards to the Paris agreement.
III. Hypotheses
Based on a review of the relevant literature, we form the following hypotheses, upon which the results of the analysis in this thesis will be evaluated:
II. I. Hypothesis 1
𝐻0: A trading strategy which consists of buying stocks of European financial sector
corporations with climate risk policy ‘leadership’ status and short selling stocks of European financial sector corporations with climate risk policy ‘laggard’ status will generate no significant abnormal returns.
II. II. Hypothesis 2
𝐻0: A trading strategy which consists of buying stocks of UK financial sector corporations with climate risk policy ‘leadership’ status and short selling stocks of UK financial
corporations with climate risk policy ‘laggard’ status will generate no significant abnormal returns.
II. III. Hypothesis 3
𝐻0: A trading strategy which consists of buying stocks of European banks with climate risk policy ‘leadership’ status and short selling stocks of European banks with climate risk policy ‘laggard’ status will generate no significant abnormal returns.
II. IV. Hypothesis 4
𝐻0: A trading strategy which consists of buying stocks of UK banks with climate risk policy ‘leadership’ status and short selling stocks of UK banks with climate risk policy ‘laggard’ status will generate no significant abnormal returns.
IV. Data
This thesis uses monthly data of a sample of 151 British, French, Benelux, German, Swiss, Austrian, Scandinavian, Spanish, Portuguese, and Italian financial sector corporations, traded on the FTSE 350, Euro Next 100, DAX, XETRA, SMI, ATX, IGBM, FTSEMIB, OMXC20, OMXH, and OMXS30 indices, retrieved via Thomson Reuters DataStream database.
Initially our sample contained all companies reported in the 2016 CDP reports in the UK and Europe, for which 2016 CDP climate change scores were disclosed as of Autumn, 2016. This excluded Irish companies. The data was then cleaned by excluding non-financial sector corporations as categorised by Bloomberg Markets, to include only banks, insurance companies, investment trusts, and other institutions involved in diversified financials. After excluding corporations with scores, B, B-, C, C-, D, and D- (see section V. for CDP scoring methodology), the sample consisted of 168 corporations. Finally, for meaningful regression analysis and risk factor calculation, corporations were required to have monthly data
available for stock returns and market-to-book ratios for 30 months prior to December 2016, and market capitalisation data for at least one month in 2016, narrowing the sample down to 151. The resulting omissions can be seen in Appendix 2, Table X.
IV. I. Sample Summary Statistics
Table I displays summary statistics for the full sample of 151 institutions. This is information is included with the intention of providing further insight for analysis in section VI of this thesis.
Table I - Summary Statistics Full Sample
Observations Mean Minimum Maximum Kurtosis Correlation with CDP Dummy CDP 151 0.24 0 1 2.51 1.00 Market Cap 151 7,651 22 160,376 32.6 0.48 BTMV 151 1.51 0.04 67.61 143.6 -0.04 MER 151 0.23% -0.21% 0.95% 3.6 -0.04
‘CDP’ represents the CDP score dummy variable (see section V for information on the derivation of this variable). ‘Market Cap’ represents the average market capitalisation of a corporation over the 18-month sample period in US Dollars. ‘BTMV’ represents the average book-to-market ratio of a corporation over the 18-month sample period.
‘MER’ represents the average excess market return of a corporation over the 18-month sample period.
Summary statistics for the UK, Europe, UK banks, and European banks can be found in tables V, VI, VII, and VIII respectively, in Appendix 1. These are also provided to aid
analysis in section VI. For this reason, a list of all the banks included in our sample can also be found in Table IX, of Appendix 2.
For the sample period, 1-month interbank overnight rates are retrieved via DataStream, to be used as proxies for risk-free returns in our regression analysis, motivated by the use of these rates for the same purpose by Ziegler et al. (2011).
Regression analysis for the European companies in our sample uses monthly Fama-French, and momentum factors retrieved online via the Kenneth R. French data library. UK monthly factors could be found online at the University of Exeter’s Xfi Centre for Finance and Investment website, however, data for the months of July 2016 to December 2016 is unavailable. Thus, the UK factors for the sample in this thesis are calculated manually in accordance with the methodologies of Fama & French (1993), and Carhart (1997) in section V. II.
UK regression analysis of requires prior-year information, therefore market returns, market-to-book ratios, and market values of 350 FTSE 350 companies is retrieved via DataStream for 30 months prior to December 2016, for construction of Fama-French, and momentum factors. This was cleaned to 333 companies, with complete information for the 30-month period. Further market information for the 18-month sample period was retrieved via
DataStream for the complete Euro Next 100, DAX, XETRA, SMI, ATX, IGBM, FTSEMIB, OMXC20, OMXH, and OMXS30 indices, for construction of market premiums, and to perform a robustness check for a composite European market premium constructed in section V. II. of this thesis.
As the scores are formulate and published for, and on behalf of investors, by the well-established Carbon Disclosure Project, and are widely reported upon annual release, for this thesis it is assumed that climate risk policy status of companies with CDP scores is known to investors and financial markets.
V. Research Methodology
V. I. CDP Scoring MethodologyComprehensive CDP scoring includes thorough surveys on Governance, Strategy, Targets & Initiatives, Communications, Risks, Opportunities, Emissions Methodology, Emissions Data, Energy, Emissions Performance, and Emissions Trading. As well as scope 1, 2, & 3, GHG
emissions information (Both direct and indirect climate risk). This outlines CDP scoring’s suitability as an indicator for climate risk policy disclosure.
One main limitation of CDP scoring is that, scoring methodology was revised as of 2016, making it impossible to directly compare scores of companies in 2016, to scores in prior years. For this reason, analysis in this thesis takes place over six quarters, from 01/07/2015 to 31/12/2016. This includes the dates of publication of 2016 CDP scores, and the COP21 Paris agreement.
CDP’s scoring is as follows, based on responses to questions on the categories, companies are scored on four different categories, out of 100%:
Leadership - meaning a company has taken the steps in implementing the very best practices
in the field of environmental management.
Management – focuses on implementation of policies and strategies companies use to address
environmental issues.
Awareness – Looks at companies’ assessment of environmental risks in relation to its
business.
Disclosure – Measures the completeness of a company’s response.
A score of 0-39% in ‘disclosure’ is awarded a D- ranking, a score of 40-74% in ‘disclosure’ is awarded a D ranking. A 75%+ score in ‘disclosure’ then entitles the company to a potential ranking in ‘awareness’ C-, being upgraded to a C, for a score of 40% or more in this
category.
Companies must achieve at least 75% in each category to be eligible for a score in the next category up. For example, if firm A, scores 83% for disclosure, 85% for awareness, 76% for management, and 68% for leadership, it is awarded a score of A-, and if firm B scores 83%, 86%, 72%, and 67% for each category in the same order, it would be awarded a score of B. This method is depicted for ease of understanding in Table II.
Table II – CDP Ranking Cut-offs
Category Score Ranking
Leadership 75-100% 0-74% A A- Management 40-74% 0-39% B B- Awareness 40-74% 0-39% C C- Disclosure 40-74% 0-39% D D-
An additional ranking of ‘F’ is used for a lack of climate risk policy disclosure.
This thesis focuses solely on companies with extreme scores of A, A-, and F. Categorising A and A- scores as climate risk policy ‘leadership’ status companies, and F scores as climate risk policy ‘laggard’ status. This exclusion of moderate rankings (management, awareness, and disclosure) is done to provide clearer differences between portfolios. For purposes of statistical analysis, this thesis uses a CDP dummy variable, by converting leadership scores A and A- to ‘1’, and laggard score F to ‘0’.
V. II. Regression Analysis
Two regression models are used in this thesis, motivated by the methodologies of Ishii et al. (2003), and Ziegler et al. (2011). These are as follows:
V. II. i. Regression 1
𝑟𝑖𝐿𝐸− 𝑟𝑖𝐿𝐴 = 𝛼𝑖 + 𝛽𝑖1( 𝑟𝑚𝑡− 𝑟𝑓𝑡) + 𝛽𝑖2𝑆𝑀𝐵𝑡+ 𝛽𝑖3𝐻𝑀𝐿𝑡+ 𝛽𝑖4𝑊𝑀𝐿𝑡+ 𝜀𝑖𝑡 V. II. ii. Regression 2
𝑟𝑖𝑡− 𝑟𝑓𝑡 = 𝛼𝑖 + 𝛽𝑖1( 𝑟𝑚𝑡− 𝑟𝑓𝑡) + 𝛽𝑖2𝑆𝑀𝐵𝑡+ 𝛽𝑖3𝐻𝑀𝐿𝑡+ 𝛽𝑖4𝑊𝑀𝐿𝑡+ 𝜀𝑖𝑡
Where 𝑟𝑖𝑡 represents monthly returns on the respective market index of each corporation over the six quarters from 01/07/2015 to 31/12/2016, retrieved from Thomson Reuters
DataStream. Excess returns 𝑟𝑖𝑡− 𝑟𝑓𝑡 are computed in Microsoft Excel using the 1-month interbank lending rates, used as proxies for risk-free returns for each market, where applicable, also retrieved via DataStream, in line with the methodology of Ziegler et al. (2011) in their four-factor regression model. 𝑟𝑖𝐿𝐸 represents the average monthly excess
returns of a leadership portfolio, and 𝑟𝑖𝐿𝐴 represents the average excess monthly returns of a laggard portfolio, calculated in Microsoft Excel using the same method.
The focus of our analysis is on the computed values of alpha (𝛼𝑖) in both versions of our regression, particularly regression 1, with regards to the hypotheses. Given the inclusion of the risk factors ( 𝑟𝑚𝑡− 𝑟𝑓𝑡), 𝑆𝑀𝐵𝑡, 𝐻𝑀𝐿𝑡, & 𝑊𝑀𝐿𝑡, these alphas represent abnormal returns on the market indices for each portfolio. 𝑟𝑖𝐿𝐸− 𝑟𝑖𝐿𝐴, represents the trading strategy of going long in leadership portfolio stocks, and short selling laggard portfolio stocks, in accordance with the methodology of Ishii et al. (2003).
Market premiums 𝑟𝑚𝑡− 𝑟𝑓𝑡 for each market involved are calculated by subtracting the
relevant risk-free rates from the relevant Index returns retrieved via DataStream. Specifically, 𝑟𝐹𝑇𝑆𝐸− 𝑟1𝑀𝑈𝐾𝐿𝐼𝐵𝑂𝑅, 𝑟𝑆𝑀𝐼− 𝑟1𝑀𝐶𝐻𝐹𝐿𝐼𝐵𝑂𝑅, 𝑟𝐷𝐴𝑋 − 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, 𝑟𝑋𝐸𝑇𝑅𝐴− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅,
𝑟𝐸𝑢𝑟𝑜𝑁𝑒𝑥𝑡− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, 𝑟𝐴𝑇𝑋− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, 𝑟𝑂𝑀𝑋𝐶20− 𝑟1𝑀𝑁𝑊𝐼𝐵𝐾, 𝑟𝑂𝑀𝑋𝐻− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, 𝑟𝐹𝑇𝑆𝐸𝑀𝐼𝐵− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, 𝑟𝑂𝑀𝑋𝑆30− 𝑟1𝑀𝑆𝐼𝐵𝑂𝑅 and 𝑟𝐼𝐺𝐵𝑀− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅.
Using these market premiums, an equal weight average market premium was constructed in Microsoft Excel for Europe 𝑟𝐸𝑈𝐴𝑉𝐺− 𝑟1𝑀𝐸𝑈𝑅𝐼𝐵𝑂𝑅, for ease of analysis. Table IV in Appendix 1 shows the correlations between this manually constructed market premium and our
benchmark indices.
Given the very strong positive correlations between the benchmark indices and the composite market premium, with an average correlation of 0.9317, excluding the IGBM index of Spain which still shows a moderate positive correlation of 0.4142, we are confident using this composite European market premium for European regression analyses.
For Europe, data for 𝑆𝑀𝐵𝑡, 𝐻𝑀𝐿𝑡, and 𝑊𝑀𝐿𝑡 from the Kenneth R. French data library for the entirety of our sample period. However, there were no factors available for the UK market, which meant the factors had to be constructed manually, with methodology from Fama & French (1993), and Carhart (1997) cited by Ziegler et al. (2011) as follows: Using the 333 FTSE 350 market firms, six portfolios were constructed using market capitalisation (𝑀𝒆) data and daily book-to-market ratios (𝐵𝒆
𝑀𝒆
⁄ ), calculated in Microsoft Excel, by dividing DataStream market-to-book ratio data points by 1, over the six-quarter sample period. Stocks were ranked by average market capitalisation in Microsoft Excel, and the median was computed as a break-point for firm size, i.e.: small (S) or big (B). Next, the
companies are ranked from high to low in each month based on book-to-market ratios, this time using the 70th percentile and the 30th percentile as breakpoints, with the highest band being high (H), middle band medium (M), and lowest band low (L). Then, six portfolios were constructed using the intersections of S, B, H, M, and L, for each month. Portfolios were S/H, S/M, S/L, B/H, B/M, B/L, where portfolio B/H consists of stocks that have both a market capitalisation above the median and a book-to-market ratio within the 70th percentile in each month.
𝑆𝑀𝐵𝑡, used as a mimic for the risk factor in returns related to firm size (Akgul, 2013), was computed for each day as the difference between average excess returns between the three small and three big portfolios, as shown in equation 1.
Equation 1.: 𝑅𝑆𝑀𝐵 = ( 𝑅𝑆𝐿+𝑅𝑆𝑀+𝑅𝑆𝐻 3 ) − ( 𝑅𝐵𝐿+𝑅𝐵𝑀+𝑅𝑩𝐻 3 ) Equation 2: 𝑅𝑩𝐻 = ∑ 𝑟𝑖 ∙ 𝑀𝒆𝒊
∑ 𝑀𝒆𝒊 , where 𝑟𝑖 is the excess returns of a stock from the B/H portfolio,
and 𝑀𝒆𝒊 is the market capitalisation of that stock. The other five portfolios are computed in the same way.
𝐻𝑀𝐿𝑡 mimics the risk factor in returns related to the book-to-market ratio (Akgul, 2013), and is computed as the daily difference between the sample average of returns between high and low portfolios, as shown in equation 3.
Equation 3: : 𝑅𝐻𝑀𝐿 = (
𝑅𝐵𝐻+𝑅𝑆𝐻
2 ) − (
𝑅𝐵𝐿+𝑅𝑺𝑳
2 )
In accordance with the methodology of Carhart (1997), 𝑊𝑀𝐿𝑡 is computed in Excel using prior-year returns. Specifically, equal-weight average of sample corporations with the highest 30% (winners) 11-month excess returns lagged one month minus the equal-weight average of sample corporations with the lowest 30% (losers) 11-month excess returns lagged one month is calculated for each month of the 18-month sample period. This monthly difference is 𝑊𝑀𝐿𝑡.
Descriptive statistics for excess returns in Stata SE, the sample showed that 3 of 4 the datatypes used to construct the regressions in this thesis are leptokurtic for this sample, as shown in Table I. For this reason, all regressions are computed as robust regressions in Stata SE, to mitigate the effects of the ‘heavy tails’ in the data. Finally, 𝑟𝑖𝐿𝐸− 𝑟𝑖𝐿𝐴 and 𝑟𝑖𝑡 − 𝑟𝑓𝑡 are regressed on the relevant risk factors using the ‘rreg’ command in Stata SE.
VI. Results & Analysis
The results of all regressions computed using Stata SE are shown in Table III.
Table III - 4-Factor Regression Results
Portfolio 𝛼 𝑟𝑚𝑡− 𝑟𝑓𝑡 𝑆𝑀𝐵𝑡 𝐻𝑀𝐿𝑡 𝑀𝑂𝑀𝑡 UKd 0.000 (0.16) 0.005 (0.23) 0.030 (1.51) 0.031 (0.99) 0.007 (0.79) EUd 0.000 (0.56) -0.003 (0.87) -0.003 (0.34) 0.019* (1.79) -0.012 (1.36) UKBd 0.006 (0.61) 0.004 (0.13) 0.018 (0.13) 0.005 (0.05) -0.102 (0.71) EUBd 0.000 (0.92) 0.003 (0.34) -0.002 (0.11) 0.018 (0.81) -0.033* (1.80) UK1 0.000 (0.13) 0.026** (2.71) 0.045 (1.09) 0.060* (2.08) 0.003 (0.08) UK0 0.001 (0.50) 0.018*** (3.37) -0.002 (0.07) 0.014 (0.84) -0.006 (0.25) EU1 0.004*** (5.63) 0.012 (0.82) -0.013 (0.34) 0.035 (0.84) -0.049 (1.40) EU0 0.005*** (16.63) 0.020*** (3.54) -0.055*** (3.30) 0.031* (1.95) 0.013 (0.79) UKB1 -0.001 (0.17) 0.026 (1.06) -0.073 (0.68) 0.009 (0.12) 0.017 (0.15) UKB0 0.002 (0.54) 0.026* (1.86) -0.170** (2.80) -0.042 (1.01) -0.019 (0.30) EUB1 0.004*** (4.51) 0.011 (0.67) -0.012 (0.28) 0.046 (0.98) -0.060 (1.54) EUB0 0.005*** (47.39) 0.014*** (6.70) -0.071*** (9.69) 0.028* (2.13) 0.002 (0.22)
Subscripts *, **, and ***, indicate coefficients are statistically different from zero at a 10% level of significance (p<0.10), 5% level of significance (P<0.05), and a 1% level of significance (p<0.01) respectively. (Absolute values of t-statistics are displayed in parentheses). ‘UKd’, ‘EUd’, ‘UKBd’, and ‘EUBd’ represent the difference in average excess returns between UK leader and laggard portfolios, EU leader and laggard portfolios, UK bank leader and laggard portfolios, and EU bank leader and laggard portfolios respectively.
‘UK1’, ‘UK0’, ‘UKB1’, and ‘UKB0’ represent the average excess returns for the UK leader portfolio, the UK laggard portfolio, the UK bank leaders portfolio, and the UK laggard portfolio respectively. For Europe ‘UK’ is replaced with ‘EU’.
Analyses of the difference between average excess returns of leadership and laggard
portfolios results in alphas of less than 0.000 for hypotheses 1, 2, and 3. With regards to the 4th hypothesis the alpha for the difference in average excess returns for the leadership and laggard portfolios of UK banks is 0.006. Although, none of these results exhibit statistical significant above the 10% level, this result indicates that a trading strategy which consists of buying stocks of UK banks with climate risk policy ‘leadership’ status and short selling stocks of UK banks with climate risk policy ‘laggard’ status generates positive abnormal returns over the 18-month sample period, and a potential positive relationship between UK bank climate risk policy leadership and short-term positive abnormal returns. However, due to lack of statistical significance, we cannot reject any of hypotheses 1, 2, 3, or 4.
From the single portfolio regressions, statistically significant alphas at the 1% level are found for the European leader, European laggard, European bank leader, and European bank
laggard portfolios. European leader portfolios generated alphas of 0.004, while European laggard portfolios generated alphas of 0.005. This suggests that investment strategies long in European financial sector stocks, or bank stocks regards of climate risk policy activity, generates positive abnormal returns over the 18-month sample period. More notably, Across European financial sector institutions, and European banks, long investments in laggard stocks generate higher positive abnormal returns, than long investments in leadership stocks over the 18-month sample period.
VII. Conclusions
Inconclusively, due a lack of statistical significance, our results imply that there are potential short-term gains in financial performance to be made from climate risk policy leadership in the UK banking sector. The lack of statistical significance is likely due to the short time span of our analysis. More conclusive results may be derived from a study of the same sample over many years, of directly comparable CDP scoring, rather than six quarters. Whilst this research sought to provide new information, in light of recent political change, an increase in data would surely lead to the computation of more significant alphas for long-short trading strategy regressions. Stenek et al. (2010), refer to banks in our sample such as HSBC being part of climate risk leadership groups. Given the importance of reputation with regards to financial performance cited in their report, it is reasonable to speculate that longer-term
analysis would show trading advantages or disadvantages more conclusively, factoring in the possibility that markets maybe be slower to respond to enhancements in reputation.
Our results for European financial institutions suggest strictly positive abnormal short-term returns are made via long investments in European financial sector stocks and European bank stocks, of both leaders and laggards. Furthermore, long investments in laggard portfolio stocks generated higher abnormal returns that leadership portfolio stocks over the 18-month sample period. Whilst these results do not contradict the findings of Ziegler et al. (2011), they do not corroborate them either. Considering Ziegler et al.’s (2011) research used European stocks of all sectors in their analysis, over a 5-year period, we can conclude that the same strategy of going long in leadership portfolio stocks and short selling laggard portfolio stocks, cannot be applied exclusively to European financial sector stocks, or bank stocks in the 18-month investment period ending in December 2016. This could be due, again to the slow response of European investors to climate change legislative changes, and its implications for European financial institutions, or reputation which takes time to establish. Furthermore, in line with Blythe et al. (2007), the timing of the Paris agreement itself could be responsible for creating some short-term uncertainty, with regards to the response of corporations to the legislation in both the UK and Europe, potentially being blurring the lines between investment leadership and laggard investment analysed in this thesis.
Further research and improvement of the results of this thesis is straightforward. In the coming years, the financial sectors in the UK and Europe will adjust fully to the new legislation, scoring more directly comparable CDP scores will become available, and reputations will be further established based on climate risk activism. Our methodology can be reapplied over a greater time-span for more conclusive findings, potentially extending the regression models used to include control variables that aid statistical significance.
VIII. References
Akgul, B. (2013). Fama - French three factor regression on European stock markets – before
and after EMU (Master’s Thesis). Retrieved from http://arno.uvt.nl/show cgi? =1285-
20
Burton, C. (2010). An Inconvenient Risk: Climate Change Disclosure and the Burden on Corporations. Administrative Law Review 62(4), 1287-1306.
Blyth, W., Bradley, R., Bunn, D., Clarke, C., Wilson, T., & Yang, M. (2007). Investment Risks Under Uncertain Climate Change Policy. Energy Policy, 35(11), 5766-5773. Retrieved from http://www.sciencedirect.com/science/article/pii/S0301421507002170 Cadman, E. (2016, October 24). Climate Change May Trigger Next Financial Crisis, Fish
Says. Bloomberg. Retreieved from https://www.bloomberg.com/news/articles/2016- 10-24/climate-change-may-trigger-next-financial-crisis-fisher-says
Capaul C., Rowley, I., & Sharpe, W. (1993). International Value and Growth Stock Returns. Financial Analysts Journal, 49(1), 27-36. Retrieved from
http://www.jstor.org/stable/4479610
Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance,
52(1), 57-82. doi: 10.1111/j.1540-6261.1997.tb03808.x
Carbon Disclosure Project Worldwide (2016). CDP climate change ceport 2016, France and
Benelux edition. Retrieved from the CDP website: https://www.cdp.net/en/media
Carbon Disclosure Project Worldwide (2016). CDP climate change report 2016 – DACH
350+ edition. Retrieved from the CDP website: https://www.cdp.net/en/media
Carbon Disclosure Project Worldwide (2016). CDP climate change report 2016, Iberia
edition, a business led path to a low carbon economy in the post COP 21 period.
Retrieved from the CDP website: https://www.cdp.net/en/media
Carbon Disclosure Project Worldwide (2016). CDP climate change report 2016, Italian
edition the road from Paris: How is the COP21 shaping the Italian business landscape? Retrieved from the CDP website: https://www.cdp.net/en/media
Carbon Disclosure Project Worldwide (2016). CDP Climate Change Report 2016 Nordic
Carbon Disclosure Project Worldwide (2016). CDP climate change report 2016 United
Kingdom edition. Retrieved from the CDP website: https://www.cdp.net/en/media
Fama, E. F., & French, K. R. (1993). Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics, 33(1), 3-56. doi: 10.1016/0304-
405X(93)90023-5
Gompers, P., Ishii, J., Metrick, A. (2003). Corporate Governance and Equity Prices. Q J
Econ, 118 (1), 107-156. doi: 10.1162/00335530360535162
IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II,
and III to the Fifth Assessement Report of the Intergovernmental Panel of Climate Change [Core writing team, R. K. Pachauri and L. A. Meyer (eds.)]. IPCC, Geneva,
Switzerland, p. 151, Retrieved from http://ar5-syr.ipcc.ch/
Konar, S., & Cohen, M. A., (2001). Does The Market Value Environmental Performance?
The Review of Economics & Statistics, 83(2), pp. 281-289. Retrieved from
http://www.jstor.org/stable/3211606
Liesen, A. (2015). Climate Change and Financial Market Efficiency. Business & Society,
54(4), pp. 511-539. Retrieved from http://journals.sagepub.com/doi/abs/10.117-
7/0007650314558392
Ryan, J. (2016, July 11). Fossil Fuel Industry Risks Losing $33 Trillion to Climate Change.
Bloomberg. Retrieved from http://www.bloomberg.com/news/articles/2016-07-
11/fossil-fuel-industry-risks-losing-33-trillion-to-climate-change
Schoenberg, M. (2016). Improving the pricing of risk: Aligning the EU financial system and
climate change (Institutional Investors Group on Climate Change publication).
Retrieved from IIGCC website: http://www.iigcc.org/files/publication- files/IIGCC_2016_Financial_Regulation_paper_v15.pdf
Shankleman, J. (2016, May 2). Funds Ignoring Climate Change Risks Rose Last Year Despite BOE Warning. Bloomberg. Retrieved from
https://www.bloomberg.com/news/articles/2016-05-01/funds-ignoring-climate-risks- rose-last-year-despite-boe-warning
Stenek, V., Amado, J. C., & Connell, R. (2010). Climate Risk and Financial Institutions, Challenges and Opportunities. Retrieved from International Finance Corporation,
World Banking Group website: http://www.ifc.org/wps/wcm/connect/topics_ext- _content/ifc_external_corporate_site/cb_home/publications/climaterisk_financial_inst itutions
Stock, J. H., & Watson, M. W. (2015). Introduction to Econometrics. Harlow, England: Pearson Education Limited.
TU, D. (2009). Do Stock Markets Price Climate Change Risks? (Master Thesis, école des Hautes Etudes Commerciales de Paris, Paris, France). Retrieved from
http://www.vernimmen.net/ftp/DoStockMktsPriceClimateChangeRisks_Danni_TU.p- df
United Nations, Framework Convention on Climate Change, Conference of the parties, Adoption of the Paris Agreement, proposal by the President, FCCC/CP/2015/L.9/Rev. 1 (12 December 2015), retrieved from https://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdf
Yeandle, M. (2016). The Global Financial Centres Index 20. Retrieved from http://www.longfinance.net/images/gfci/20/GFCI20_26Sep2016.pdf
Ziegler, A., Busch, T., & Hoffmann, V. H. (2011). Disclosed Corporate Responses to Climate Change and Stock Performance: International Empirical Analysis. Energy Economics,
33, pp. 1283-1294. Retrieved from http://www.sciencedirect.com/science/article-
IX. Appendices
IX. I. Appendix 1Table IV - Market Portfolio Correlations with Composite Europe Market Premium
Switzerland XETRA DAX Austria Euro Next
0.9256 0.9674 0.9569 0.9446 0.9702
Sweden Norway Italy Finland Spain
0.9499 0.8237 0.9188 0.9282 0.4142
Table V - Summary Statistics UK
Observations Mean Minimum Maximum Correlation with CDP Dummy
CDP 66 0.26 0 1 1.00
Market Cap 66 7,466 978 160,376 0.41
BTMV 66 0.92 0.36 2.02 0.25
MER 66 0.03% -0.14% 0.14% -0.16
‘CDP’ represents the CDP score dummy variable (see section V for information on the derivation of this variable). ‘Market Cap’ represents the average market capitalisation of a corporation over the 18-month sample period in US Dollars. ‘BTMV’ represents the average book-to-market ratio of a corporation over the 18-month sample period.
‘MER’ represents the average excess market return of a corporation over the 18-month sample period.
Table VI - Summary Statistics Europe
Observations Mean Minimum Maximum Correlation with CDP Dummy
CDP 85 0.22 0 1 1.00
Market Cap 85 7,795 22 83,059 0.57
BTMV 85 1.96 0.13 67.61 -0.06
MER 85 0.38% -0.21% 0.95% -0.01
‘CDP’ represents the CDP score dummy variable (see section V for information on the derivation of this variable). ‘Market Cap’ represents the average market capitalisation of a corporation over the 18-month sample period in US Dollars. ‘BTMV’ represents the average book-to-market ratio of a corporation over the 18-month sample period.
‘MER’ represents the average excess market return of a corporation over the 18-month sample period.
Table VII - Summary Statistics UK Banks
Observations Mean Minimum Maximum Correlation with CDP Dummy
CDP 7 0.57 0 1 1.00
BTMV 7 1.10 0.46 2.02 0.75
MER 7 0.02% -0.08% 0.14% -0.74
‘CDP’ represents the CDP score dummy variable (see section V for information on the derivation of this variable). ‘Market Cap’ represents the average market capitalisation of a corporation over the 18-month sample period in US Dollars. ‘BTMV’ represents the average book-to-market ratio of a corporation over the 18-month sample period.
‘MER’ represents the average excess market return of a corporation over the 18-month sample period.
Table VIII - Summary Statistics European Banks
Observations Mean Minimum Maximum Correlation with CDP Dummy
CDP 36 0.36 0 1 1.00
Market Cap 36 12,900 87 83,059 0.66
BTMV 36 3.36 0.26 67.61 -0.14
MER 36 0.38% -0.21% 0.88% -0.17
‘CDP’ represents the CDP score dummy variable (see section V for information on the derivation of this variable). ‘Market Cap’ represents the average market capitalisation of a corporation over the 18-month sample period in US Dollars. ‘BTMV’ represents the average book-to-market ratio of a corporation over the 18-month sample period.
‘MER’ represents the average excess market return of a corporation over the 18-month sample period.
IX. II. Appendix 2
Table IX - Sample Banks
Name Code CDP Score Country
Bank of Georgia Holdings BGEO F United Kingdom
HSBC Holdings plc HSBA A United Kingdom
Lloyds Banking Group LLOY A United Kingdom
Onesavings Bank OSB F United Kingdom
Paragon Group of Companies PAG F United Kingdom
Royal Bank of Scotland Group RBS A- United Kingdom
Standard Chartered STAN A- United Kingdom
Bank Coop AG S:CB1 A Switzerland
Bank Linth Llb AG S:LINN F Switzerland
Basler Kantonalbank D:BSLX A Switzerland
Cembra Money Bank AG S:CMB F Switzerland
Comdirect bank AG D:COMX F Germany
Commerzbank AG D:CBKX A− Germany
DVB Bank AG D:DVB F Germany
Graubündner Kantonalbank S:GRKP F Switzerland
Luzerner Kantonalbank S:LUKN F Switzerland
St. Galler Kantonalbank S:SGKN F Switzerland
Thurgauer Kantonalbank S:TKBP F Switzerland
UBS S:UBSG A Switzerland
Walliser Kantonalbank S:WKBN F Switzerland
Zuger Kantonalbank AG S:ZG F Switzerland
Banco BPI SA P:BPI F Portugal
Liberbank SA E:LBK F Spain
CaixaBank E:CABK A Spain
Banco Comercial Português SA P:BCP A- Portugal
Banco Santander E:SCH A- Spain
Bankia E:BKIA A- Spain
Intesa Sanpaolo S.p.A I:ISP A Italy
UniCredit I:UCG A- Italy
Banca Carige I:CRG F Italy
Banca IFIS SpA I:IF F Italy
Banca Monte dei Paschi di Siena Group I:BMPS F Italy
Banca Popolare di Sondrio I:BPSO F Italy
Mediolanum S.p.A. I:BMED F Italy
DNB ASA N:DNB A- Norway
SEB W:SEA A- Sweden
National Bank of Belgium B:BN@B F Belgium
BinckBank N.V. H:BINC F Netherlands
BNP Paribas SA. F:BNP A- France
Dexia SA. B:DEXB F Belgium
Union Fin. Fra. SA. F:UFF F France
Table X - Sample Omissions
Name Code CDP Score Country
Aldermore Group ALD F United Kingdom
Assura AGR F United Kingdom
BH Macro BHMG F United Kingdom
Countrywide CWD F United Kingdom
Esure Group ESUR F United Kingdom
Hastings Group HSTG F United Kingdom
Riverstone Energy RSE F United Kingdom
Shawbrook Group SHAW F United Kingdom
Deautsche Pfandbriefbank AG D:PBBX F Germany
Erste Group Bank S:ERST F Switzerland
Raiffeisen Bank International D:RAWX A Germany
Poste Italiane I:PST F Italy
Compagnie du bois Sauvage SA. B:BSV F Belgium
European Assets Trust H:EURA F Netherlands
Flow Traders H:FLOW F Netherlands