1
Fossil Fuel Divestment and Alternative Energy Investment
Master Thesis MSc FinanceAuthor: B.T. Oegema1 Supervisor: dr. L. Dam University of Groningen Faculty of Economics and Business
Date : 10-01-2019
Abstract
Multiple campaigns have put pressure on institutional investors to divest from fossil fuel companies and/or invest in alternative energy companies to help prevent global warming. To support such moral choices, it is important to understand the financial consequences for fund managers. This study examines whether fossil fuel divestment and alternative energy investment strategies impair performance of well-diversified EMU investment portfolios over the period 1990-2018. In line the empirical literature, this study demonstrates that a fossil fuel constraint does only marginally diminish risk-adjusted returns. Moreover, differences in excess returns are explained by differences in exposure to well-known systematic risk factors. Therefore, fossil fuel divestment and alternative energy investment choices cause no significant out- or underperformance. Findings in this research thus suggest that institutional investors can safely divest from fossil fuel assets and invest in alternative energy assets, although investors should be aware of the changes in vulnerability to certain systematic risk factors like market risk and oil price risk.
Keywords: Socially Responsible Investing, Fossil Fuel Divestment, Alternative Energy Investment, Risk-Adjusted Performance, GARCH
JEL Codes: G11, Q41, Q42
2 1. Introduction
Total fossil fuel emission accounted for approximately 72% of the total greenhouse emission worldwide in 2017 (Olivier et al., 2017). On 12 December 2015, the Paris Agreement was adopted that aims to limit global warming by well below 2 degrees Celsius above pre-industrial levels. Since fossil fuel emission account for such a significant amount of total greenhouse emission, public pressure on funds, endowments, foundations, governments and companies that invest in fossil fuel has grown tremendously. The first campaigns against fossil fuel investments find their origins on US university campuses in 2011 where students put pressure on University endowments to sell their unethical fossil fuel holdings. Such a sell-off for financial, political or in this case ethical purposes is known as divestment (Baron and Fischer, 2015). Since then, almost thousand institutions, including the well-known Norwegian sovereign wealth fund and the Irish national investment fund for example, divested over more than 7 trillion dollars of morally ambiguous fossil fuel assets2. Nevertheless, total energy demand still rises worldwide nowadays (International Energy Agency, 2018). Therefore, the urge for efficient clean energy sources that can substitute conventional high-carbon energy sources is immense (IPCC, 2018; Baron and Fischer, 2015). Institutional investors can pro-actively contribute to climate change with investments in the alternative energy sector3. However, the alternative investment universe is small, the total market capitalization of oil & gas companies is 40 times larger than the market capitalization of wind, solar, geothermal and biomass companies on global exchanges (Ritchie and Dowlatabadi, 2015). Nonetheless, a significantly growing amount of green bonds, alternative energy ETF’s and alternative energy companies arises at the investible horizon. Moreover, a total cumulative amount of 2.2 trillion dollar has already been invested in alternative energy sources worldwide since 2010, with China as current market leader (McCrone et al., 2018). The transition towards a low-carbon economy dominates political agendas nowadays. Various institutional investors already do their moral duty with fossil fuel divestments and/or alternative energy investments. To support such moral choices of fund managers, it is important to understand whether those choices harm financial performance. Fossil fuel divestment campaigns and debates created an ethical incentive for institutional investors to contribute to climate change by decarbonizing their financial assets. The question arises whether this moral incentive undermines the financial incentive of investors and how costly this strategy is. Pension schemes, for example, promise investors a particular amount of retirement income at a certain moment. Such funds primarily aim to fulfil their retirement promises and hence need to be financially incentivized in order to exclude fossil fuel assets from a portfolio. Pension funds and other investors aim to optimize portfolio returns for a given level of risk. From a theoretical perspective, Modern Portfolio Theory (Markowitz, 1952)
2 Source: https://gofossilfree.org/divestment/commitments/ (Data obtained at 29 November, 2018)
3 The alternative energy portfolio in this study includes companies that generate or distribute solar, wind,
3
assumes that exclusion of securities from the investible universe impairs risk-adjusted returns and that a fossil fuel divestment or another exclusion strategy is costly. Although a significant amount of fossil fuel assets already has been divested, limited empirical literature on the financial consequences of such divestment choices exists and results are dissimilar. Several empirical reports find evidence in line with Modern Portfolio Theory and argue that an exclusion strategy reduces diversification benefits (Capelle-Blancard and Monjon, 2012; Cornel, 2015). However, others demonstrate that fossil fuel divestment does not significantly impair financial performance (Trinks et al., 2018; Plantinga and Scholtens, 2016) or even suggest that a fossil fuel constraint enhances risk-adjusted returns (Henriques and Sadorsky, 2018). Although the energy transition dominates political agendas and wind and solar are expected to provide 50% of total energy by 2050 (Henbest et al., 2018), limited literature on the risk characteristics and financial performance of alternative energy sources exists. Since the market capitalization of this sector is expected to grow significantly (Bullard, 2014), the alternative energy sector can develop as an interesting investment substitute for the fossil fuel industry. Therefore, risk characteristics and financial performance of alternative energy sources are interesting in particular for institutional investors who consider substitution of fossil fuel assets with alternative energy assets. This study analyses whether there exists such a financial incentive for institutional investors to divest fossil fuel assets and/or invest in alternative energy assets.
4
This study finds results in line with existing divestment literature (Trinks et al., 2018; Plantinga and Scholtens, 2016)). Exclusion of fossil fuel assets from an investment portfolio does only marginally impair risk-adjusted returns. Moreover, differences in risk-adjusted returns are explained as a compensation for higher exposure to systematic risk factors and no alphas significantly different from zero are observed. Institutional investors can thus safely divest fossil fuel holdings without significantly diminishing risk-adjusted returns, although investors should take differences in exposure to systematic risk factors like market risk and oil price risk into account. Similar to other empirical literature (Ibikunle and Steffen, 2015; Halcoussis and Lowenberg, 2018; Henriques and Sadorsky, 2018),this paper demonstrates that fossil fuel risk-adjusted returns are diminishing and that fossil fuel constrained portfolios are slightly outperforming the benchmark post-2011. Moreover, the alternative energy portfolio outperformed the fossil fuel index post-2011. Although the alternative energy industry is currently small-scaled and has been volatile, this asset class can develop as an interesting fossil fuel substitute for investors.
This study contributes to existing literature and the social debate on fossil fuel divestment and alternative energy investments in three ways. First of all, this research is the first to use an EMU perspective whereas the majority of the literature addressed the financial consequences of fossil fuel divestment of US listed portfolios. This is in particular interesting for institutional investors in Europe with a geographical or monetary home bias4. Second, limited research on the returns and risk characteristics of the alternative energy sector exists. Since European countries are global leaders in terms of alternative power capacity per capita and have been investment pioneers in this asset class (European Environment Agency, 2017), research on the risk-adjusted returns and risk characteristics in this area is interesting in particular. Third, because demand for alternative energy is expected to grow tremendously, it is timely to study the risk-return characteristics of this potential fossil fuel investment substitute.
The remainder of this research paper is structured as follows. Section 2 provides an overview of the existing literature on socially responsible investing, fossil fuel divestment and alternative energy investment. Section 3 presents the research hypotheses and the methodology of this study. The data and descriptive statistics are discussed in section 4. Section 5 provides an overview of the results and discusses the implications of fossil fuel divestment and alternative energy investment. A conclusion is presented in section 6.
2. Literature review
This section provides an overview of relevant literature. First, this study discusses the theoretical implications of exclusion from a Modern Portfolio Theory perspective. Second, this section discusses the theoretical literature on the performance and implications of socially
4 A home bias is the tendency of investors to overinvest heavily in domestic securities, conflicting with Modern
5
responsible investing. Finally, this study presents an overview of the empirical literature of both fossil fuel divestment and alternative energy investments.
2.1 Diversification costs
Modern Portfolio Theory (Markowitz, 1952) states that portfolio managers should combine multiple risky assets to diversify away idiosyncratic risk of a portfolio without shrinking returns. Efficient Markowitz diversification involves combining assets with less-than-perfect positive correlations in order to reduce risk in the portfolio without sacrificing any of the portfolio’s return (Francis & Kim, 2013). In an optimal situation, all rational investors should hold a value-weighted basket of all securities, i.e. the market portfolio. Portfolio returns are a function of the weights and returns of the securities5, whereas a portfolio variance is a product of weights, variances and correlations between stocks6. Since Modern Portfolio Theory assumes the market portfolio is optimal, exclusion of securities leaves investors with a reduced investible universe and increased idiosyncratic risk which has implications for risk and return trade-offs. Divestment of stocks with high risk-adjusted returns or low correlations is costly in particular. Moreover, reduction of the investible universe by excluding industry portfolios increases idiosyncratic in general, since firms in the same industry are more likely to perform bad at the same time compared to firms in dissimilar industries (Markowitz, 1952; Capelle-Blancard and Monjon, 2012). Modern Portfolio Theory thus assumes that exclusion of an industry, like the fossil fuel industry, is costly and impairs financial performance.
2.2 Socially responsible investing
The importance of corporate social responsibility issues exponentially grows in the financial world due to the increased number of specialized institutions, socially responsible mutual funds, academic publications and specialized corporate responsibility reports (Bassen et al., 2016). Broadly speaking, socially responsible investing (SRI) or ethical investing has both a positive and a negative screening component. Negative screening as a SRI strategy involves divestment of undesirable ‘sin stocks’ from a portfolio. Those firms are generally engaged in moral undesirable activities like tobacco, alcohol or gambling (Renneboog et al., 2008). Likewise, positive screening as a strategy incorporates firms with moral desirable characteristics that promote social or environmental sustainability.
Modern Portfolio Theory assumes that screening is a trade-off between ethical objectives and financial performance. The question arises whether this moral incentive to seek ethical objectives undermines the financial incentives of investors. Renneboog et al. (2008) argue social responsible investors are not willing to give up financial performance to seek ethical objectives and exclude sin stocks. As long as SRI screening impairs financial performance,
5 The expected return of a n-security portfolio is derived as: 𝐸(𝑟
𝑝) = ∑𝑛𝑖=1𝑤𝑖𝐸(𝑟𝑖).
6 The variance of a portfolio return equals: 𝜎
6
investors prefer optimal financial performance without screening over suboptimal financial performance with screening. Contradictory, Capelle-Blancard & Monjon (2014) find evidence that some investors’ utility functions have non-financial preferences as well. They demonstrate that a significant amount of investors prefer portfolios consistent to their moral beliefs and give priority to social and environmental sustainability over outperforming risk-adjusted performance of sin stocks.
Although some investor utility functions might have non-financial preferences, it is important to understand the financial consequences of screening or socially responsible investing. Different conclusions on the trade-off between financial performance and social responsible investing are observable. Dam & Scholtens (2015) for example find a strong relationship between financial performance and corporate social responsibility (CSR). El Ghoul et al. (2011) support this and demonstrate that that there exists a strong relationship between CSR, valuation and risk of a company. Companies with high CSR scores tend to have substantially lower cost of equity whereas participation in tobacco or nuclear power industries increases the cost of equity of such companies significantly.
Whereas differences in individual firm performance might be observed due to different financing costs, it is important to study the consequences of CSR screening on a portfolio level. For example, Geczy et al. (2003) find that SRI screening significantly impairs financial risk-adjusted performance. Similarly, Capelle-Blancard and Monjon (2012), argue that financial performance is only impaired as a result of exclusion of an entire section or industry such as the fossil fuel industry. However, the majority of the empirical research on the trade-off between ethical constraints and financial performance contradict with Modern Portfolio Theory. Little to no impairment of portfolio performance is found as a result of CSR screening (Humphrey and Tan, 2014; Viviani and Revelli, 2015; Renneboog et al. 2008; Bello, 2005). Consequently, existing literature suggests that SRI in general does not significantly diminish risk-adjusted returns although exclusion of an entire section might be costly.
2.3 Fossil fuel divestment
7
Trinks et al. (2018) find that, over the period 1927-2015, divestment of US listed fossil fuel companies did not significantly harm financial portfolio performance. Although fossil fuel stocks tend to have high absolute returns, their risk-adjusted returns are not above-market and the industry therefore provides limited diversification benefits. Especially the coal industry stocks tend to underperform the market in terms of risk-adjusted returns. Similarly, Plantinga & Scholtens (2016) examine that due to the higher systematic risk of the fossil fuel industry, a fossil fuel investment constraint does not have a significant impact on the risk-adjusted performance between 1973 and 2015. Moreover, fossil fuel constrained portfolios outperform portfolios including conventional energy stocks in different more recent time frames between 2005 and 2018 (Henriques and Sadorsky, 2018; Halcoussis and Lowenberg, 2018). Investors who are not restricted by short selling even take a short position in the fossil fuel industry optimally. Those three results are in contradiction with the findings of Capelle-Blancard and Monjon (2012), who argue that financial performance is only impaired as a result of exclusion of an entire section or industry and Cornell (2015), who demonstrates that over a time frame of the last 50 years fossil fuel divestment would seriously diminish risk-adjusted returns.
Although various systematic empirical evidence over larger time frames on fossil fuel exists, a strong diminishing trend appears in the returns and performance of fossil fuel companies. Starting with the Industrial Revolution, fossil fuel companies have been the engine of world economy growth for decades. In 1980, the S&P500 top 10 included seven fossil fuel companies, while ExxonMobil is the only one remaining in this top ten in 2018 (Sanzillo et al., 2018). Moreover, from 2013 onwards, the MSCI world ex fossil fuels index significantly underperforms the MSCI world index7 while low-carbon funds significantly outperform conventional funds (Ibikunle and Steffen, 2015; Halcoussis and Lowenberg, 2018; Henriques and Sadorsky, 2018). In general, industries have periods of over- and underperformance with the market. For institutional investors it is crucial to determine whether the fossil fuel might be currently undervalued or not.
First, the development of corporate social responsibility created a financing problem for the fossil fuel industry. Divestment campaigns and political debates raised the cost of debt financing and the required rate of return from investors. (El Ghoul et al., 2018; Ansar et al., 2013). Moreover, several banks already announced that they are not providing loans to the fossil fuel sector anymore in the short term8. A financing problem thus arises for the fossil fuel industry in the near future and investors hence adjust their expectations of future cash flows. Ansar et al. (2013) support this proposition and argue that investors lower their expectations of future cash flows, caused by a growing uncertainty of the fossil fuel industry. They demonstrate that the value of the divested assets is especially affected indirectly by stigmatization in the
7 See: https://www.msci.com/documents/10199/b4b02abd-f3a7-4a4b-b459-e996a672cd8f
8 For example, the World Bank announced that they stop financing the fossil fuel industry by the beginning of
8
long run, whereas the direct effect of divestment creates opportunities for investors to earn above market returns by increasing their holdings in the potentially undervalued fossil fuel stocks in the short run. Morally unacceptable firms are thus under-priced (relatively high book-to-market value) and lie above the security market line, whereas morally acceptable firms are overpriced and lie below the security line. Consequently, positive alphas for morally unacceptable firms are detectible whereas morally acceptable firms generate negative alphas in the short run (Heinkel et al., 2001). Second, some even argue that current fossil fuel companies have overvalued their fossil fuel reserves (Leaton et al., 2013). An oversupply of fossil fuels might decrease fossil fuel stock prices dramatically. Since fossil fuel prices and fossil fuel stocks are positively correlated (Scholtens and Yurtsever, 2012; Sanusi and Ahmad, 2016) the burst of this ‘carbon bubble’ might plunge fossil fuel stock values dramatically with if investors realize that large parts of the reserves cannot be burned (Bütikofer, 2014; Ritchie and Dowlatabadi, 2015). Third, the trend of diminishing returns in the fossil fuel industry seems to be supported by the urge to reduce carbon emissions, commitment of fossil fuel divestment and the future energy market expectations. 195 member countries already signed the Paris agreement that aims to limit global warming by well below 2 degrees Celsius above pre-industrial levels. Moreover, assets committed to divest increased from $52 billion to $6.24 trillion in the past four years (Arabella advisors, 2018) whereas the contribution of burning fossil fuel to total electricity production is expected to decrease to 29%, down from 63% today (Henbest et al., 2018).
Empirical evidence for the consequences of fossil fuel divestment over a long time horizon varies. However, existing literature strongly suggests that fossil fuel returns are diminishing and that fossil-free portfolios are outperforming fossil fuel portfolios in the last decade. Whereas the fossil fuel industry might generate alpha in the short term, investing for the long term is expected to be rather risky due to financing problems, reserves overvaluation, and unburnable carbon risk. Despite the diminishing returns and risky character of the fossil fuel industry, investors should be aware of potential spill-over effects. Oil prices drive costs in other industries. Hence, oil price shocks significantly impact the performance of non-fossil fuel industries (Nandha and Faff, 2008). Moreover, banks have a significant exposure to fossil fuel assets (Battiston et al., 2017). Consequently, interrelatedness between the fossil fuel industry and non-fossil fuel industries might seriously harm portfolio performance.
2.4 Alternative energy investment
9
this asset class only accounts for 0,16% of the total MSCI world index. Therefore, marginal differences in financial performance as a result of investments in the alternative energy sector are expected by Modern Portfolio Theory. Since there exists limited literature on performance of alternative energy stocks in a portfolio, this literature sector particularly discusses the risk-return performance of the alternative energy sector in the 21st century. Moreover, this subsection examines market movements and expectations of this asset class.
Europe has been the pioneer of alternative energy investments. In 2005, European investors funded 46% of all alternative energy investments in alternative energy (European Environment Agency, 2017). China has strongly overtaken Europe as market leader and accounted for 36% of the investments worldwide in 2016, while Europe had a share of 17% and US investors funded 15% of all those investments. Despite the fact that Europe was a pioneer and invested heavily in the early and mid-2000s, Bohl et al. (2015) find evidence for, similar to the dot.com bubble and the bitcoin bubble for example, the presence of at least some irrational exuberance9 among alternative energy investors. They demonstrate that the alternative energy market, especially in Europe, generated significant and positive alphas prior to the financial crisis of 2008. However, during and after the crisis, alphas turned negative and significant and indicate a substantial revaluation of this asset class. Furthermore, Bohl et al. (2015) examine that most discussed alternative energy ETF’s in their paper have negative and significant abnormal returns, high market betas and significant loadings on the size factor between 2004 and 2013. Although they find high market and size systematic risk for the alternative energy market, several empirical evidence demonstrates a positive correlation between oil prices and alternative energy returns (Kumar et al., 2012; Sadorsky, 2012; Gupta, 2017). Hence, an increase in the oil prices increases systematic risk of the clean energy substitute. Higher oil prices generally spur curiosity in developing alternative energy sources whereas interest in developing alternative energy sources drops after an oil price decrease (Sadorsky, 2012). Interestingly, literature suggests (Nandha and Faff, 2008; Klevnäs et al., 2015) that rising oil prices decrease equity returns for non-energy industries as costs of doing business increase. Oil prices tend to have a negative impact on stock prices as oil price increases generally induce inflationary pressures and lower valuations ultimately (Degiannakis et al., 2018). Literature thus suggests that fossil fuel divestment might increase negative exposure to oil prices. Therefore, alternative energy investments might be an interesting asset class to hedge oil price exposure for fossil fuel divesting investors. Moreover, an even more influential empirical relationship between the returns of country-level technology stocks and alternative energy stocks is observable (Kumar et al., 2012; Gupta, 2017). They suggest that the level of technical efficiency and developments of a country has a positive impact on the use and returns of technological driven alternative energy market. Although asset pricing models (Fama and
9 Irrational exuberance refers to overconfidence among investors that drives asset prices above fundamental
10
French, 2015; Carhart, 1997) perform well in modelling systematic risk and risk-adjusted returns, institutional investors who aim to invest in alternative energy should take the sectors exposure to oil prices and the country-specific technology level and returns in mind.
Henriques & Sadorsky (2018) examine the implications of positive and negative environmental screening on a portfolio level. They demonstrate that, between 2005 and 2016, portfolios that included clean energy stocks but excluded fossil fuel stocks performed better than the well-diversified portfolio consisting of conventional energy stocks only. However, a short position on both energy stocks is taken. Consequently, the risk-return differential was driven by fossil fuel exclusion and not by alternative energy inclusion.
Although the current market capitalization of the alternative energy market is small-scaled and the industry yielded negative abnormal returns, the future of the alternative energy market has perspective. With the signed Paris Agreement and the growing amount of divestment campaigns, demand for clean energy sources is expected to grow steadily. Total electricity usage worldwide from fossil is expected to fall to 29% worldwide by 2050, down from 63% today. Moreover, renewable energy is expected to produce 87% of the European electricity by 2050 (Henbest et al., 2018). Despite the fact that the energy market has a bright outlook from the demand side, investments in the alternative energy should still be considered as low-return and rather risky. Exposure to oil prices changes might seriously diminish alternative energy returns when the ‘carbon bubble’ plunges and oil prices decrease significantly (Bütikofer, 2014).
3. Methodology
This section presents the methods used to evaluate financial performance of fossil fuel, alternative energy and the fossil free and alternative energy free portfolios. First, this section discusses the research hypotheses and data requirements of this study. Second, the risk-adjusted performance measures of Sharpe, Sortino and Treynor are explained. Finally, this section explains of the well-known Carhart Four-Factor model and Fama & French five factor multi-factor asset pricing models.
3.1 Research questions
11
characteristics of the industry. This study uses three risk-adjusted performance indicators and two well-known multi-factor models to evaluate the financial performance and the risk characterises of both energy industries.
3.2 Risk-adjusted performance indicators
Investors demand compensation for the risk that is taken, therefore institutional investors use risk-adjusted performance indicators. To measure risk-adjusted portfolio performance, this research uses three well-known measures that are often used by mutual fund managers to rank portfolios. William Sharpe, Jack Treynor and Frank Sortino have developed performance indicators that allow fund managers to consider risk and return at the same time. First, Sharpe’s ex-post reward-to-variability ratio (Sharpe, 1994) measures the excess return per unit of volatility. William Sharpe extended a linear risk-return modelling technique which was developed earlier by James Tobin (1958):
𝑆ℎ𝑎𝑟𝑝𝑒 𝑟𝑎𝑡𝑖𝑜: 𝐸[𝑟𝑝−𝑟𝑓]
𝜎𝑝 (1)
𝐸[𝑟𝑝− 𝑟𝑓] represents the expected excess return or the risk premium of a portfolio, based on a historical average excess return. 𝜎𝑝 presents the standard deviation of the portfolio that captures the total risk of the excess returns. This ratio measures the historical average excess return per unit of total risk and gives the slope of a straight line in a risk and return diagram. Second, Treynor (1965) uses a reward-to-volatility ratio to compare performances of different portfolios. Instead of using the standard deviation as a measure of portfolio risk, Treynor uses a portfolios market beta as a measure of systematic risk.
𝑇𝑟𝑒𝑦𝑛𝑜𝑟 𝑟𝑎𝑡𝑖𝑜: 𝐸[𝑟𝑝−𝑟𝑓]
𝛽𝑝 (2)
Where 𝐸[𝑟𝑝− 𝑟𝑓] represents the expected excess return of the portfolio and 𝛽𝑝 represents the historical responsiveness to market changes with the market. This ratio measures the historical average excess return per unit of market risk. Third, this stduy uses the Sortino ratio (Sortino and Price, 1994) as a complementary risk-adjusted performance indicator. Sortino’s ratio is a modification of Sharpe’s ratio. However, it measures the excess return per unit of downside risk and only takes ‘bad volatility’ and thus the probability of negative excess returns into account:
𝑆𝑜𝑟𝑡𝑖𝑛𝑜 𝑟𝑎𝑡𝑖𝑜: 𝐸[𝑟𝑝−𝑟𝑓]
√1𝑇∑𝑇𝑡=1(𝑀𝑖𝑛 (0, 𝑟𝑝 − 𝑟𝑚𝑎𝑟))2
(3)
12
𝑟𝑚𝑎𝑟 as the minimal acceptable return of a portfolio, for which this paper uses the risk free rate. Consequently, downside risk is derived only from periods with observed negative excess returns and 𝑟𝑝 ≤ 𝑟𝑀𝐴𝑅.
3.3 Multi-factor asset pricing models
The performance indicators of Sharpe, Treynor and Sortino in general report differences in risk-adjusted performance. Asset pricing models however assume that differences in excess returns are explained as compensation for higher exposure to systematic risk factors and that no risk-adjusted returns statistically different from zero are observed. The Carhart (1997) Four-Factor Model and the Fama & French (2015) five-factor asset pricing models capture such systematic risk factors to explain the exposure of portfolio excess returns to those systematic risk factors. Factor premiums can be considered as rewards for investors that endure losses during bad times (Ang, 2014). First this study uses the Carhart Four Factor asset pricing model:
𝑅𝑝,𝑡 − 𝑅𝑓,𝑡 = 𝑎𝑝+ 𝛽𝑝,𝑚(𝑅𝑚,𝑡− 𝑅𝑓,𝑡) + 𝛽𝑝,𝑠𝑚𝑏𝑆𝑀𝐵𝑡+ 𝛽𝑝,ℎ𝑚𝑙𝐻𝑀𝐿𝑡+ 𝛽𝑝,𝑚𝑜𝑚𝑀𝑂𝑀𝑡+ 𝜖𝑡(4) The Carhart Four-Factor model explains the excess returns of a monthly rebalanced portfolio to four systematic risk factors. First, 𝑅𝑝,𝑡 − 𝑅𝑓,𝑡 equals a portfolio monthly return in excess of an one month T-bill. Alpha (𝑎) represents the abnormal return of the industry portfolio. The intercept is expected not to be statistically different from zero since exposure to the risk factors should capture all variation. (𝑅𝑚,𝑡− 𝑅𝑓,𝑡) represents the excess return of the European market portfolio. 𝑆𝑀𝐵𝑡, 𝐻𝑀𝐿𝑡, and 𝑀𝑂𝑀𝑡 are returns on European value-weighted, zero-investment, factor mimicking portfolios for size, book-to-market equity and one-year momentum in stock returns (Carhart 1997). 𝑆𝑀𝐵𝑡, equals the return of a European portfolio long in small stocks (small market capitalization) and short in big stocks. This factor captures the outperformance of small firms relatively to large firms. Likewise, 𝐻𝑀𝐿𝑡 represents the monthly return of a European portfolio long in value stocks (high book-to-market value) and short in growth stocks (low book-to-market value). This risk factor captures the phenomenon that value stocks outperform growth stocks on average. Finally, 𝑀𝑂𝑀𝑡 represents the premium of a portfolio long in stocks that have performed well and the last six months or so (winner stocks), and short in bad-performing stocks (losers) over that period. Winner stocks have the tendency to continue rising for a while, whereas loser stocks generally continue to lose for a while.
Finally, this paper estimates the Fama-French Five-Factor (2015) asset pricing model: 𝑅𝑝,𝑡 − 𝑅𝑓,𝑡 = 𝑎𝑝+ 𝛽𝑝,𝑚(𝑅𝑚,𝑡− 𝑅𝑓,𝑡) + 𝛽𝑝,𝑠𝑚𝑏𝑆𝑀𝐵𝑡+ 𝛽𝑝,ℎ𝑚𝑙𝐻𝑀𝐿𝑡+ 𝛽𝑝,𝑟𝑚𝑤𝑅𝑀𝑊𝑡+
𝛽𝑝,𝑐𝑚𝑎𝐶𝑀𝐴𝑡+ 𝜖𝑡 (5)
13
in excess of the risk free rate, is captured by (𝑅𝑚,𝑡− 𝑅𝑓,𝑡) in the equation. 𝑆𝑀𝐵𝑡 is the difference between returns of a European long-short portfolio with small and big firms. 𝐻𝑀𝐿𝑡 is the risk premium of a portfolio long in high book-to-market ratio stocks and short in low book-to-market stocks. 𝑅𝑀𝑊𝑡 measures the difference in returns between stocks that have robust and weak profitability since firms with higher gross profits ratios have the tendency to perform better. Finally, 𝐶𝑀𝐴𝑡 represents the investment premium of a portfolio long in low investment (conservative) stocks and short in high investment (aggressive) stocks since firms that increase their capital investments tend to achieve lower returns for five subsequent years. 3.4 Heteroscedasticity
Financial data is often subject to heteroscedasticity. Especially during crisis periods, volatility clustering occurs in financial data. With the exception of the basic materials industry portfolio, all industry portfolio returns show heteroscedastic residuals. The residuals test positive on autoregressive conditional heteroscedastic (ARCH) effects. To model volatility clustering, GARCH(1,1) models with conditional variance are used additionally. The ARCH-LM heteroscedasticity tests on remaining ARCH effects of the GARCH(1,1) fails to reject the null hypotheses of no ARCH effect. Appendices B.1-B.2 show that the lagged variance term mostly captures volatility clustering. Both GARCH(1,1) models indicate stationary conditional variances since the sum of the lagged variance term coefficient and the lagged error term coefficient does not exceed one. Moreover, the Jarque-Bera normality test shows an improved normality in the residuals.
The Durban Watson test statistic indicates no autocorrelation in the error terms for all industries in both the Carhart four-factor model and the five factor model. Moreover, the factors in both models indicate no evidence for multicollinearity.
3.5 Oil price exposure
As discussed earlier, existing literature assumes that oil prices are positively correlated with both fossil fuel and alternative energy returns (Kumar et al., 2012; Sadorsky, 2012; Gupta, 2017; Scholtens and Yurtsever, 2012). Moreover, a burst of the ‘carbon bubble’ might seriously impair financial performance of fossil fuel stocks (Bütikofer, 2014) whereas oil price increases generally tend to decrease other equity returns (Nandha and Faff, 2008). Therefore, exposure to oil price deviations might be a significant systematic risk component for investors. To address whether excess returns are explained as compensation for oil price exposure, this study models energy price risk and adjusts the Fama & French 5 factor model (2015):
𝑅𝑝,𝑡 − 𝑅𝑓,𝑡 = 𝑎𝑝+ 𝛽𝑝,𝑚(𝑅𝑚,𝑡− 𝑅𝑓,𝑡) + 𝛽𝑝,𝑠𝑚𝑏𝑆𝑀𝐵𝑡+ 𝛽𝑝,ℎ𝑚𝑙𝐻𝑀𝐿𝑡+ 𝛽𝑝,𝑟𝑚𝑤𝑅𝑀𝑊𝑡+
14
All factors except 𝑂𝐼𝐿𝑡 are identical to equation (5). 𝑂𝐼𝐿𝑡 represents monthly log returns on Brent crude oil barrels. This paper uses log returns on Brent crude oil because European countries commonly use this price as benchmark.
3.6 Sensitivity analyses
To test the robustness of my results, various sensitivity analysis are performed. This study addresses both volatility clustering and oil price exposure as described above. Moreover, this research tests whether divestment performance varies over time. This study uses the Sharpe ratio, Sortino ratio, Treynor ratio and Carhart model to determine whether divestment or investment strategies have periods of significant under- or outperformance over four time frames of seven years. Furthermore, this paper analyses the performance of two zero-investment portfolios. First, zero-investment portfolio long in fossil fuel stocks and short in the fossil fuel constrained portfolio are constructed. Second, this study constructs a zero-investment portfolio long in alternative energy stocks and short in fossil fuel stocks. The zero-investment portfolios test the robustness and changes in vulnerability to certain systematic risk factors of both divestment and divest-invest strategies.
4. Data and descriptive statistics
In this section, this paper first introduces the dataset and explains the construction of the industry portfolios. Afterwards, this section discusses the descriptive statistics of the industry portfolios.
4.1 Data
Datastream provides monthly data of stock returns and market capitalizations of an EMU market index with 1354 constituents from 19 countries10. Returns and market capitalizations are available for the period between April 1965 and August 2018. Monthly stock returns are obtained from Datastream. Datastream also provides monthly market values from all 1354 publicly listed constituents. The market value is measured as the share price multiplied by the number of ordinary shares in issue. All industry portfolios are monthly rebalanced. The weights are determined by the market capitalization of the stocks as a share of the full industry capitalization. The returns in this study do not reflect real returns. Portfolio managers are exposed to rebalancing, monitoring, transaction and other financial costs additionally. Those costs are substantial (Bessembinder, 2016). The portfolio mean returns exhibit the average of all monthly market capitalization weighted returns of the industry portfolios. No extreme outliers are detected. All maximum and minimum values are observed around the dot.com bubble and the financial crisis of 2008.
10 Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania,
15
Testing the four and the five factor model corresponding to equation (4) and (5) requires returns of the European market, size, value, momentum, profitability and investment factor portfolios. Kenneth R. French provides those data. A small adjustment to the factor portfolios of Kenneth R. French is made. This paper uses the excess return of all 1354 EMU stocks as the market proxy. Appendices C.1-C.2 show that the EMU market portfolio generates significant abnormal rates of return with European market factors from Kenneth R. French. The EMU stocks thus tend to outperform the European market portfolio of Kenneth R. French. Therefore, the market factor portfolio of Kenneth R. French is not appropriate to use. Monthly Brent oil prices are obtained from the Federal Reserve Bank of St. Louis. Monthly returns from the European factor portfolios and risk free rates are available for the period November 1990 – Augustus 2018. Therefore, this study analyses the risk-adjusted returns of all stocks that have been active during this time period of 334 months.
4.2 Portfolio construction
The Industry Classification Benchmark (ICB) framework classifies all companies into four levels: the industry code, the supersector code, the sector code and the subsector code. The ICB codes allocate all companies to the subsector that specifically describes the nature of the business as determined by the primary source of revenue from the company. The industry portfolios are derived from the ICB industry codes. To evaluate fossil fuel divestment and alternative energy investment performance, this research divides the ICB industry oil & gas into two new composed industry portfolios: the fossil fuel portfolio and the alternative energy portfolio. Moreover, the subsector coal (ICB code: 1771) from the basic materials is added to fossil fuels, whereas this study adds the subsector alternative electricity to the alternative energy portfolio and subtracts it from the utilities portfolio. Table I shows the constructed EMU portfolios and the current market capitalization of the industries. The dataset consists of eleven industry portfolios, one market portfolio, one fossil fuel free portfolio and one alternative energy free portfolio including 1354 stocks in total.
4.3 Market capitalization
Table I demonstrates that consumer goods has the largest EMU market capitalization in August 2018 with a market cap of 20.3%. The fossil fuel market share of 7.8% indicates that diversification benefits might be seriously harmed as a result of divestment. The alternative energy portfolio only accounts for 0.5% of the full market although alternative energy accounts for only 0,16% of the total MSCI world index. The current EMU energy share of 8.3% is more than 2% higher than the current 6% share of energy stocks as a percentage of the total MSCI world index11. Therefore the EMU has an above worldwide average presence of energy stocks
11 Source: https://www.msci.com/documents/10199/178e6643-6ae6-47b9-82be-e1fc565ededb (Data obtained
16
in the investible universe. Since alternative energy accounts for only 0.5% of the total market, contribution to financial performance of a well-diversified market cap weighted portfolio is limited. Appendix A.1 gives an overview of the 10 largest EMU stocks based on their market capitalization. The fossil fuel stocks Total, Royal Dutch Shell A and B represent 5.36% of the total market capitalization on August 2018. Hence, EMU fossil fuel divestment performance is largely driven by divestment of those three stocks.
Figure I shows the number of stocks and market capitalization of the energy market. The figure shows that, except during the oil crisis of 1998, the share of the alternative energy in the total energy sector grew gradually from 2% to 6%. Moreover, the number of alternative energy stocks grew from two to 22 stocks. This study focusses on financial performance of fossil fuel divestment and alternative energy investment strategies on a portfolio level. Therefore, it is interesting to see that the fossil fuel industry accounted for 9.4% of the full EMU index on average. Currently, the fossil fuel index has a current market capitalization of 7.8%, therefore,
TABLE I: Overview of EMU industry portfolios and construction adjustments
Portfolio # of
stocks % market cap
ICB industries, sectors and subsectors included
Fossil fuel 35 7.8% Sectors: oil & gas producers (0530) and oil equipment, services and distribution (0570)
Subsector: coal (1771)
Alternative energy 22 0.5% Sector: alternative energy (0580) Subsector: alternative electricity (7537)
Basic materials 80 6.2% All sectors from original industry are included except for subsector coal (1771)
Industrials 284 16.1% All sectors from original industry are included Consumer goods 183 20.3% All sectors from original industry are included Healthcare 85 7.2% All sectors from original industry are included Consumer services 175 7.7% All sectors from original industry are included Telecommunication 26 3.1% All sectors from original industry are included
Utilities 45 5.4% All sectors from original industry are included except for subsector alternative electricity (7537)
Financials 325 19.0% All sectors from original industry are included Technology 94 6.8% All sectors from original industry are included All industries 1354 100% All Industries are included
Fossil fuel free 1319 92.2% All Industries are included except for fossil fuels Alternative energy free 1332 99.5% All Industries are included except for alternative energy
17
it seems that institutional investors are either gradually divesting their fossil fuel shares and/or the fossil fuel market is underperforming.
Figure I: Energy market capitalization and number of stocks
This figure shows the market capitalization and the number of stocks of the fossil fuel industry and the alternative energy industry in the energy market. In the EMU dataset, the both energy industries account for on average 9.8% of the full index. The fossil fuel industry has an average share of 9.4% whereas the alternative energy sector accounts for 0.4% of the market on average.
4.4 Descriptive statistics
Table II shows the descriptive statistics of all portfolios. During 1990 – 2018, technology has the highest excess mean return of 2.01% per month and that this industry outperforms all other industries by 0.74% per month in absolute terms. Utilities has the lowest excess mean return over this period with 0.86% per month. Differences in systematic risk exposure generally explains mean return differentials. Except for technology, the differences in mean returns are limited. With a standard deviation of 8.09%, which is significantly larger than all other industries, this industry exposes substantially above average market risk. Appendix A.2 clearly shows the influence of the creation and burst of de dot-com bubble on the above average mean return and standard deviation. With an excessive mean of 3.19% per month over the period 1990-1997 and a standard deviation of 12.34% in the period 1998-2004, the creation and the burst of the dot-com bubble is clearly visible.
Table II demonstrates that over the period 1990-2018, fossil fuels has a higher absolute mean return and a lower total risk (standard deviation) than the alternative energy sector. During this time period, fossil fuels exhibited a mean excess return of 0.98% whereas alternative energy generated a mean excess return of 0.92%. Interestingly, both fossil fuels and alternative energy
0 5 10 15 20 25 30 35 40 % 82 84% 86% 88% 90% % 92 % 94 % 96 % 98 % 100
18
seem to underperform in absolute terms compared to the market. This is in contraction with the result of Plantinga & Scholtens (2016), who demonstrate that US listed fossil fuels stocks were the best performing industry between 1973-2015 and outperformed the market between 2000 and 2010. However, appendix A.2 shows that after 2011, alternative energy outperformed the fossil fuel industry significantly in the EMU. Literature suggest that oil prices movements are positively correlated with fossil fuel and alternative energy returns and negatively with other equity returns (Scholtens and Yurtsever, 2012; Nandha and Faff, 2008). The oil price drop by more than 50% between June 2014 and January 2015 (Klevnäs et al., 2015) might therefore explain the fossil fuel underperformance during this time frame. Moreover, alternative energy outperforms the market post-2011. Although literature suggest that the alternative energy returns are positively correlated with oil prices (Kumar et al., 2012; Sadorsky, 2012; Gupta, 2017), the alternative energy portfolio was the best performing industry in the two years (2014-2015) of the oil price drop in this EMU dataset in terms of excess returns, whereas the fossil fuel industry was the worst performing industry. However, the standard deviation over 2012-2018 shows that the alternative energy industry exposes a substantially higher amount of systematic risk than the fossil fuel industry during that period.
TABLE II: Summary statistics for monthly EMU industry portfolio returns (November 1990 - August 2018, 334 months)
Portfolio Mean Min Max Median SD Beta
Fossil fuels 0.98% -19.32% 16.22% 1.11% 5.14% 0.67 Alternative energy 0.92% -19.60% 23.15% 0.65% 5.92% 0.61 Basic materials 1.16% -26.16% 25.78% 1.42% 5.73% 0.82 Industrials 1.21% -20.45% 26.02% 1.43% 5.55% 0.94 Consumer goods 1.25% -14.51% 17.95% 1.27% 4.79% 0.82 Healthcare 1.27% -14.71% 17.22% 1.50% 4.92% 0.83 Consumer services 1.10% -17.08% 19.58% 1.10% 4.92% 0.92 Telecommunication 0.92% -22.25% 30.17% 0.79% 6.97% 0.66 Utilities 0.86% -13.79% 16.22% 1.01% 4.70% 0.76 Financials 0.95% -24.75% 35.29% 1.20% 6.49% 0.90 Technology 2.01% -22.31% 38.40% 1.91% 8.09% 0.70 All industries 1.16% -15.32% 20.45% 1.50% 4.80% 1.00
Fossil fuel free 1.17% -16.10% 21.70% 1.42% 4.96% 0.99
Alternative energy free 1.16% -15.35% 20.43% 1.52% 4.80% 1.00
This table contains the mean, minimum, maximum, median, standard deviation (SD) and beta with market portfolio of monthly excess returns over a period of 334 months (November 1990 – August 2018). The total number of observations equals 287.957. Monthly returns are obtained from Datastream. The fossil fuel free portfolio represents a monthly rebalanced portfolio of all industry portfolios except the fossil fuel industry portfolio. Similarly, the alternative energy free portfolio represents a portfolio of all industry portfolios except the alternative energy portfolio.
19
market portfolios have the lowest beta with the market. Theoretical literature assumes that divestment of low beta industries seriously impairs diversification benefits (Markowitz, 1952). In line with this theoretical framework, the fossil fuel constrained portfolio exposes 0,16% higher total risk than the unrestricted market portfolio.
Both the mean return and the market beta contradict with existing literature. Trinks et al. (2018) and Plantinga & Scholtens (2016) examine that the fossil fuel has high absolute returns and a high beta. Consequently, existing literature assumes diversification costs are subject to the divestment of high returns, whereas this study demonstrates that exclusion of this low beta industry might impair diversification benefits in particular. Due to the small market capitalization of the alternative energy market, differences between the market portfolio and the alternative energy constrained portfolio are limited.
5. Results
This section discusses the risk-adjusted returns and performance of the portfolios. First, this section uses the performance indicators of Sharpe, Treynor and Sortino to measure risk-adjusted performance. Second, this section uses the Carhart four-factor and the Fama and French five-factor asset pricing models to explain whether exposure to systematic risk five-factors captures differences in excess return. Finally, this section performs several robustness tests.
5.1 Performance indicator results
Table III shows the estimation results of the performance indicators that indicate risk-adjusted performance. Both energy portfolios underperform compared to the market (all industries) in terms of excess returns and exhibited higher risk. The fossil fuel industry has a higher excess return and lower total (standard deviation) and downside risk than the alternative energy industry. Hence, both the Sharpe ratio and the Sortino ratio are higher for the fossil fuel industry. In contradiction with theoretical literature (Markowitz, 1952), but in line with existing empirical literature (Trinks et al., 2018; Plantinga & Scholtens, 2016; Henriques and Sadorsky, 2018), divestment decreases the Sharpe and Sortino ratio only marginally.
20
costs for this EMU divestment portfolio are beta-driven whereas US fossil fuel divestment costs are driven by exclusion of high excess returns. Both EMU energy portfolios are thus interesting asset classes for institutional investors who aim to mitigate market risk.
Table III: Performance indicators: Sharpe, Treynor and Sortino ratio (Nov 1990 – Aug 2018)
Portfolio Mean StDev DR Beta Sharpe Treynor Sortino
Fossil fuels 0.98% 5.14% 3.29% 0.67 0.191 0.015 0.298 Alternative energy 0.92% 5.92% 3.43% 0.61 0.155 0.015 0.267 Basic materials 1.16% 5.73% 4.26% 0.82 0.203 0.014 0.273 Industrials 1.21% 5.55% 3.93% 0.94 0.219 0.013 0.309 Consumer goods 1.25% 4.79% 3.35% 0.82 0.261 0.015 0.373 Healthcare 1.27% 4.92% 3.24% 0.83 0.257 0.015 0.391 Consumer services 1.10% 4.92% 3.23% 0.92 0.224 0.012 0.341 Telecommunication 0.92% 6.97% 4.48% 0.66 0.132 0.014 0.206 Utilities 0.86% 4.70% 3.27% 0.76 0.182 0.011 0.262 Financials 0.95% 6.49% 4.72% 0.90 0.147 0.011 0.202 Technology 2.01% 8.09% 4.90% 0.70 0.248 0.026 0.410 All industries 1.16% 4.80% 3.24% 1.00 0.241 0.012 0.356 Fossil fuel free 1.17% 4.96% 3.34% 0.99 0.235 0.012 0.349 Alternative energy free 1.16% 4.80% 3.25% 1.00 0.241 0.012 0.356
This table shows the mean, standard deviation (StDev), downside risk (DR), beta, Sharpe ratio, Treynor ratio and Sortino ratio of monthly excess returns over a period of 334 months (November 1990 – August 2018). Monthly excess returns of all EMU portfolios are obtained from Datastream, one-month T-bill rates for Europe are obtained from Kenneth Fama’s database. The fossil fuel free portfolio represents a monthly rebalanced portfolio of all industry portfolios except the fossil fuel industry portfolio. Similarly, the alternative energy free portfolio represents a portfolio of all industry portfolios except the alternative energy portfolio. Downside risk measures the industry volatility in times of negative excess returns. The Sharpe ratio equals the mean excess return over the standard deviation. Treynor’s ratio equals the mean excess return over the industries systematic risk (beta). Finally, the Sortino ratio measures the mean excess return per unit of downside risk.
Table IV shows the outcome of the risk-adjusted performance indicators over different time frames. The fossil fuel industry outperformed alternative energy in terms of the Sharpe and Sortino ratio until 2012. In line with empirical literature (Ibikunle and Steffen, 2015; Halcoussis and Lowenberg, 2018), the alternative energy portfolio performs better than the conventional energy portfolio in terms of reward to volatility, downside risk and market risk over the last couple of years. Evidence for this trend might be related to the limited diversification opportunities of the alternative energy portfolio prior to this period or due to the consequences of oil price shocks on equity prices (Scholtens and Yurtsever, 2012; Nandha and Faff, 2008; Klevnäs et al., 2015).
21
constrained portfolio are limited and environmental screening performance is difficult to evaluate. Despite those limitations, the alternative energy seems an interesting asset class for institutional investors. The low beta generates diversification benefits and post-2011 performance looks promising.
5.2 Factor model results
Table V presents the results of the Carhart four-factor model using an ordinary least square estimation. In line with empirical literature (Trinks et al., 2018; Plantinga & Scholtens, 2016), no risk-adjusted returns are significantly different from zero. Although table III demonstrates marginal financial impairment due to fossil fuel divestment, differences in excess returns are explained as compensation for higher exposure to well-known systematic risk factors. The fossil fuel constrained portfolio exposes significantly higher market and lower value risk than the alternative energy free portfolio. Institutional investors are thus exposed to higher market and marginally lower value systematic risk as a result of a divest-invest approach. Whereas, Trinks et al. (2018) argue that US fossil fuel stocks are mostly large cap value stocks, fossil
Table IV: Performance indicators over time: Sharpe, Treynor and Sortino ratio (7 year time frames)
Time frame 1990-1997 1998-2004 2005-2011 2012-2018
Sharpe’s ratio
Fossil fuels 0.327 0.143 0.128 0.184
Alternative energy 0.165 0.085 0.109 0.291
All industries 0.352 0.195 0.125 0.393
Fossil fuel free 0.345 0.187 0.120 0.401
Alternative energy free 0.352 0.195 0.125 0.393
Treynor’s ratio
Fossil fuels 0.019 0.017 0.010 0.012
Alternative energy 0.019 0.011 0.009 0.025
All industries 0.014 0.011 0.007 0.014
Fossil fuel free 0.014 0.012 0.007 0.015
Alternative energy free 0.014 0.011 0.007 0.014
Sortino’s ratio
Fossil fuels 0.819 0.199 0.198 0.309
Alternative energy 0.491 0.184 0.147 0.499
All industries 1.000 0.291 0.187 0.667
Fossil fuel free 0.997 0.285 0.178 0.704
Alternative energy free 1.000 0.291 0.187 0.667
22
fuels only have significant loadings on the EMU market and value risk factors. Moreover, compensation for market, size and value risk largely explains alternative energy excess returns. The adjusted R2 in tables V-VI indicate that a significant fraction of the sample variance is unexplained by the asset pricing models for both individual energy portfolios. Exposure to oil prices and country-level technology stocks for example might capture the unexplained variance of the alternative energy portfolio (Kumar et al., 2012; Sadorsky, 2012; Gupta, 2017). In line with the findings of Trinks et al. (2018), the market factor explains roughly all the variation in the returns of the fossil fuel free portfolio. Since the R2 is nearly one, institutional investors can safely divest fossil fuels, at a marginal tracking error.
Table V: Risk-adjusted return performance of fossil fuel, alternative energy, fossil fuel free and alternative energy free portfolios (Carhart four-factor model, Nov 1990 – Aug 2018, OLS)
Portfolio Fossil Fuel Alternative
energy Fossil fuel free Alternative energy free Alpha 0.0005 (0.0023) -0.0009 (0.0028) -0.0000 (0.0002) -0.0000 (0.0000) MRP 0.6940*** (0.0479) 0.7912*** (0.0580) 1.0289*** (0.0048) 1.0009*** (0.0003) SMB -0.1192 (0.1010) 0.4082*** (0.1223) 0.0092 (0.0102) -0.0020*** (0.0006) HML 0.2764*** (0.0910) 0.2335** (0.1102) -0.0022** (0.0092) -0.0009* (0.0005) WML 0.0553 (0.0582) -0.0076 (0.0705) -0.0157*** (0.0059) 0.0001 (0.0003) Adjusted R2 0.4534 0.3970 0.9940 1.0000
This table provides the results of monthly excess returns of the EMU portfolios regressed on the coefficients of the Carhart (1997) model using an Ordinary Least Squares estimation. Alpha measures the abnormal rate of return. The coefficient of MRP measures the industry exposure to the all industries portfolio. SMB, HML and WML represent the coefficients on K. French’s European size, value and momentum risk factor portfolios. Values in parentheses present standard errors. * p-value <0.10; ** p-value <0.05; *** p-value <0.01. Monthly returns are obtained from Datastream. The fossil fuel free portfolio represents a monthly rebalanced portfolio of all industry portfolios except the fossil fuel industry portfolio. Similarly, the alternative energy free portfolio represents a portfolio of all industry portfolios except the alternative energy portfolio. Adjusted R2 is the goodness of fit statistic.
N=334.
23
exposes lower systematic risk to the profitability risk factor since 7.8% of the portfolio is divested, which has a high and significant exposure to this risk factor.
Table VI: Risk-adjusted return performance of fossil fuel, alternative energy, fossil fuel free and alternative energy free portfolios (Fama-French five-factor model, Nov 1990 – Aug 2018, OLS)
Portfolio Fossil fuel Alternative
energy Fossil fuel free Alternative energy free Alpha -0.0031 (0.0023) -0.0028 (0.0029) 0.0003 (0.0001) 0.0000 (0.0000) MRP 0.8051*** (0.0543) 0.7684*** (0.0671) 1.0210*** (0.0055) 1.0010*** (0.0003) SMB -0.0281 (0.1005) 0.4056*** (0.1240) 0.0002 (0.0102) -0.0002*** (0.0006) HML 0.2087 (0.1269) 0.5125*** (0.1567) -0.0223* (0.0128) -0.0023*** (0.0008) RMW 0.5460*** (0.1652) 0.4284** (0.2039) -0.0722*** (0.0168) -0.0020** (0.0010) CMA 0.4722*** (0.1604) -0.2965 (0.1981) -0.0364** (0.0163) 0.0015 (0.0010) Adjusted R2 0.4825 0.4067 0.9943 1.0000
This table provides the results of monthly excess returns of the EMU portfolios regressed on the coefficients of the Fama-French five-factor (2015) model using an Ordinary Least Squares estimation. Alpha measures the abnormal rate of return. The coefficient of MRP measures the industry exposure to the all industries portfolio. SMB, HML , RMW and CMA represent the coefficients on K. French’s European size, value, profitability and investment risk factor portfolios. Values in parentheses present standard errors. * p-value <0.10; ** p-value <0.05; *** p-value <0.01. Monthly returns are obtained from Datastream. The fossil fuel free portfolio represents a monthly rebalanced portfolio of all industry portfolios except the fossil fuel industry portfolio. Similarly, the alternative energy free portfolio represents a portfolio of all industry portfolios except the alternative energy portfolio. Adjusted R2 is the goodness of fit statistic. N=334.
5.3 GARCH(1,1) estimation results
24
5.4 Periodic out- or underperformance
Both the OLS and the GARCH(1,1) estimations on the asset pricing models indicate no under- or outperformance of the individual industry portfolios or energy constrained portfolios. To assess whether those results are stable over time, this paper evaluates fossil fuel divestment and alternative energy investment through the Carhart four-factor model over four subsamples of seven years. The insignificant alphas in appendix B.3 indicate no significant periodic out- or underperformance. Although chosen time-periods might be biased, results contradict with the findings of Bohl et al. (2015) who demonstrate significant positive and negative alpha’s in the European alternative energy market. Moreover, the insignificant alphas in appendix B.3 during 2012-2018 indicate that divestment campaigns, which started in 2011, do not cause significant periodical underperformance of the fossil fuel industry and that vulnerability to well-known systematic risk factors explain differences in excess returns.
5.5 Oil price exposure
Appendix B.4 shows the estimation results of the adjusted Fama & French model (6) for exposure to oil price movements. In line with existing literature (Scholtens and Yurtsever, 2012; Sanusi and Ahmad, 2016) oil price movements capture a significant amount of fossil fuel return variation. In line with earlier findings, divestment causes significant higher market risk vulnerability and lower profitability exposure. However, appendix B.4 demonstrates that fossil fuel divestment causes a significantly larger and negative exposure to oil price movements. Confirming existing literature (Nandha and Faff, 2008), oil price movement negatively influence returns of non-fossil fuel assets. Hence, total vulnerability to oil price shocks increases as a result of divestment and investors might thus hedge negative oil price exposure by holding fossil fuel assets. Conflicting with other literature (Kumar et al., 2012; Sadorsky, 2012; Gupta, 2017), results indicate no evidence for exposure of alternative energy returns to oil price changes.
5.6 Zero-investment portfolios
25
Above robustness tests support the earlier implications of fossil fuel divestment and alternative energy investment. Robustness tests that incorporate periodic out- or underperformance, oil price exposure, volatility clustering and zero-investment portfolios indicate that divestment and investment strategies do not significantly impair portfolio performance. Hence, differences in excess returns are explained by vulnerability to certain systematic risk factors. Divestment causes lower profitability risk and increases vulnerability to market risk and oil price risk in particular. Moreover, the alternative energy portfolio mainly differs from the fossil fuel portfolio in terms of exposure to size and profitability systematic risk.
6. Conclusion
Various campaigns and political debates have put pressure on fund managers and created a moral incentives to divest fossil fuel assets invest in alternative energy assets. To help prevent an above 2 degrees Celsius global warming above pre-industrial levels by 2050, it is important to study whether this moral incentive undermines the financial incentives of institutional investors. This paper therefore investigates the financial consequences of fossil fuel divestment and alternative energy investment. This research analyses the impact of such decarbonizing choices on a well-diversified EMU portfolio over the period 1990-2018. First, the well-known adjusted performance indicators of Sharpe, Sortino and Treynor are used to evaluate risk-adjusted portfolio performance of fossil fuel and alternative energy investment choices. Second, this research uses two multifactor models to explain whether differences in excess returns are explained by exposure to certain well-known systematic risk factors. Finally, several robustness tests are performed to address the quality of the findings.
26
absolute returns for the fossil fuel industry. Consequently, diversification costs in the EMU over this period are driven by a low beta, whereas existing literature on US stocks assumes those diversification costs are driven by exclusion of high absolute returns. Findings in this research thus suggest that institutional investors can safely divest from fossil fuel assets and invest in alternative energy assets although investors should be aware of changes in exposure to certain systematic risk factors.
The approach of this paper has limitations and drawbacks. First, the fossil fuel portfolio weights of Total and Royal Dutch Shell are substantial. Therefore, divestment performance might be positively or negatively influenced by firm-specific behaviour of one of those companies. Second, divestment transaction costs are not taken into account. Exclusion of fossil fuel stocks and re-investment in the remaining industries market capitalization proportional incurs financial costs that directly affect divestment performance. Third, this study is retrospective and assumes that performance of other industries is not influenced by fossil fuel exclusion. However, oil price shocks induce substantial shocks among different industries for example (Scholtens and Yurtsever, 2012; Nandha and Faff, 2008). Moreover, fossil fuel divestment might also indirectly diminish portfolio performance due to fossil fuel exposure in other sectors like the banking sector (Battiston et al., 2017). Although backward-looking performance assumes fossil fuel divestment and alternative energy investment does not significantly impair financial performance, institutional investors should seriously take future energy market expectations, industry interrelatedness and energy price risk into account.
27 7. References
Ang, A., 2014. Asset Management: A systematic approach to factor investing. Financial Management Association Survey and Synthesis Series. Oxford University press.
Ansar, A., Caldecott, B., Tilbury, J., 2013. Stranded assets and the fossil fuel divestment campaign: what does divestment mean for the valuation of fossil fuel assets? Smith School of Enterprise and the Environment, University of Oxford.
Arabella Advisors, 2018. The global fossil fuel divestment and clean energy investment movement. https://www.arabellaadvisors.com/wp-content/uploads/2018/09/Global-Divestment-Report-2018.pdf
Baron, R., Fischer, D., 2015. Divestment and stranded assets in the low-carbon transition. https://www.oecd.org/sdroundtable/papersandpublications/Divestment%20and%20Stranded% 20Assets%20in%20the%20Low-carbon%20Economy%2032nd%20OECD%20RTSD.pdf Bassen, A., Hölz, H. M., Schlange, J., 2006. The Influence of Corporate Responsibility on the Cost of capital: an Empirical Analysis. Working Paper, University of Hamburg.
Battiston, S., Mandel A., Monasterolo, I., Schütze F., Visentin, G., 2017. A climate stress-test of the financial system. Nature climate change 7, 283-288.
Bello, Z. Y., 2005. Socially responsible investing and portfolio diversification. Journal of Financial Research, 28(1), 41-57.
Bessembinder, H., 2016. Frictional Costs of Fossil Fuel Divestment. Available at SSRN: https://ssrn.com/abstract=2789878
Bullard, N., 2014. Fossil fuel divestment: a $5 trillion challenge. Bloomberg New Energy Finance, white paper. https://data.bloomberglp.com/bnef/sites/4/2014/08/BNEF_DOC_2014-08-25-Fossil-Fuel-Divestment.pdf
Bohl, M. T., Kaufman, P., Siklos, P, L., 2015. What drove the mid-2000s explosiveness in alternative energy stock prices? Evidence from U.S., European and global indices. International Review of Financial Analysis (40), 194-206.
Bütikofer, R., 2014. Fossil fuels as a financial risk: The carbon bubble. https://reinhardbuetikofer.eu/wp-content/uploads/2014/09/PE-Carbon-Bubble-UK-web.pdf Capelle-Blancard, G., Monjon S., 2014. The performance of Socially Responsible Funds: Does the Screening Process Matter. European Financial Management 20, 494-520.
