In addition to the market factor, the SMB, HML, RMW, and CMA factors are necessary in order to perform the Fama-French 3- and 5-Factor models introduced in section 3.3. Similar to earlier studies (Blitz & Fabozzi, 2017; Kempf & Osthoff, 2007), this study chooses to obtain the explanatory variables from the Kenneth French Data Library (French, n.d.). As explained in the delimitations section, the factors have not been computed manually as the primary objective of the study is not to test the correctness of the factor models but to apply them for analytical purposes. Kenneth French’s dataset holds high credibility, and the factor returns are based on a significant amount of data from prominent sources, which is why the factor-data is considered to be reliable.
Finally, the portfolio analysis also needs to consider if the results are subjects to sector bias. Therefore, the companies’ industries have also been extracted from Refinitiv. The industries are classified by a third-party source, being the Dow Jones FTSE’s Industry Classification Benchmark, that returns the eleven industry classes listed in table 4.1, alongside their distribution. Based on the classifications, it is possible to test whether the performance is affected when you ‘sector neutralize’ the portfolio, i.e. apply the same industry weights as the total index.
Table 4.1: Index Industry Weights
Source: Own production
4.2.1 Portfolio Construction
The fundamental part of the analysis is to study the relationship between ESG scores and financial performance, which is tested through the portfolio approach. Companies are allocated to six different portfolios, namely A to F, based on their ESGC, ESG, E, S, and G scores, respectively. This results in a total of 30 portfolios being tested. Portfolio A will contain the ESG leaders, whereas portfolio F will contain the ESG laggards. By performing this division, it becomes possible to test whether higher-scoring companies perform significantly different than lower-scoring companies. This inclusion approach also portraits the positive screening strategy presented in section 2.2.2. The companies are assigned to the portfolios according to the following rules:
𝐴 = 𝐸𝑆𝐺" ≥ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 b5 6e 𝐵 = 𝐴 > 𝐸𝑆𝐺" ≥ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 b4
6e 𝐶 = 𝐵 > 𝐸𝑆𝐺" ≥ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 b3
6e 𝐷 = 𝐶 > 𝐸𝑆𝐺" ≥ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 b2
6e 𝐸 = 𝐷 > 𝐸𝑆𝐺" ≥ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 b1
6e 𝐹 = 𝐸 > 𝐸𝑆𝐺" ≥ 0b1
6e
The rules stated above indicate that companies will be ranked relatively to the other companies. The percentile rules have been chosen, as they ensure that the number of companies will remain approximately the same within each portfolio each year. This also entails that one portfolio will not benefit from an added diversification effect. Table 4.2 illustrates the number of firms that have been assigned to each portfolio, based on the ESGC score, for the entire period of analysis.
Table 4.2: Number of Firms in the Portfolios
Source: Own production
After allocating the firms to their respective portfolios throughout the period of analysis, the firms’ weights should be assigned in order to calculate the portfolio return and risk. There are two types of weighting schemes in portfolio theory: 1) equally weighted, where each stock is given the same weight or importance, and 2) value-weighted, where the stock’s weight depends on their market value. In this thesis, the portfolio return has been calculated using the value-weighted sum of returns. The primary reason for this, is that the factors extracted from the Kenneth French Data Library (French, n.d.) has been constructed based on value-weighted returns. Thus, the weights have been calculated based on the companies’ market value at the time of rebalancing, i.e. the beginning of each year. The market value, as defined by Refinitiv, is the share price multiplied by the number of ordinary shares in issue. Hereby, the equation for the weights is as follows:
𝑤0 = 𝑀𝑉0 𝑀𝑉.
Where:
𝑀𝑉& is the market value for stock 𝑖
𝑀𝑉' is the total market value for the portfolio
In order to detect possible sector displacement within the constructed portfolios, the value-weighted industry weights have been studied. As illustrated in figure 4.2, all industries are represented in the ESGC score portfolios which indicate good diversification, but the composition varies significantly (see remaining scores in Appendix I). Furthermore, it is also observed that the industry weights within the portfolios vary significantly across time (see Appendix II). These findings highlight the importance to construct and test portfolios that are sector-balanced, i.e. portfolios having the same industry weights as the overall index. This approach can be classified as the “best-in-class” strategy, as companies are being ranked according to their scoring among their peers. Thus, companies not scoring in the top of the index can now be assigned to the higher-scoring portfolios, if they have a good ESG score compared to their industry-peers. For each industry, companies have been allocated to the portfolios A to F according to the same rules as previously stated.
Year A B C D E F
2011 72 71 71 71 71 72
2012 73 70 71 71 71 72
2013 72 71 71 71 71 72
2014 72 72 70 71 71 72
2015 72 71 71 71 71 72
2016 73 70 71 71 71 72
2017 72 71 71 71 71 72
2018 72 71 71 71 71 72
2019 72 71 71 71 71 72
2020 72 72 70 71 71 72
Number of firms in the portfolios
Hereafter, the individual stock has been value-weighted according to its industry and constrained by the industry’s weight in the total index. The constraints are determined by the industry weights for the applied index in 2020, as illustrated in table 4.1. The results hereof are sector-balanced portfolios where all portfolios have the same industry weights across time and score. Similar to the overall scoring study, the sector-balanced approach provides 30 additional testing portfolios.
Figure 4.2: Average Industry Weights for the ESGC Score
Source: Refinitiv (n.d.-a), Own production
4.2.2 Portfolio Performance
In order to qualify the acceptance or rejection of the constructed hypotheses, the performance of the high- and low-scoring portfolios will be tested and compared. Portfolio performance can be measured in many ways. As introduced in Chapter 3, one can look at the simple descriptive statistics of the portfolio’s return and risk properties. For the preliminary examination of the portfolios’ performance, the cumulative returns and annualised average monthly returns have been calculated. However, the return itself does not say much without the risk element, which is examined through the standard deviation, market beta, skewness, kurtosis, and maximum drawdown. The calculation of the standard deviation involves the composition of the covariance-matrix, which is a 428x428 matrix that contains the covariances between all the stocks included in the analysis (see subset in Appendix III). Hereafter, the portfolio standard deviation is calculated by using the following vector equation (Munk, 2019):
𝜎. = p𝜋1∑𝜋
Where:
𝜋. is the transposed weight vector
∑ is the 428x428 variance-covariance matrix 𝜋 is the weight vector
0% 20% 40% 60% 80% 100%
A B C D E
F Basic Materials
Consumer Discretionary Consumer Staples Energy
Financials Health Care Industrials Real Estate Technology Telecommunications Utilities
The market beta, on the other hand, is found by performing linear regressions between the portfolio and market returns. For further interpretation of the portfolios’ return distribution and downside risk, the skewness, kurtosis, and maximum drawdown are also studied.
The aforementioned performance characteristics are the standalone risk and return measures. However, as foretold in the theory chapter it is important to study the portfolios’ adjusted performance. The risk-adjusted measures used are the ones introduced in the theory chapter: Sharpe ratio, Treynor ratio, and Jensen’s alpha. The first two can easily be computed from the above estimates, whereas the alpha requires further analysis. Alpha is examined through the factor models: CAPM, Fama-French 3-, and 5-Factor models, using the statistical programming software Stata (see code in Appendix IV). All the models have two common elements: 1) the independent variable, that is the monthly portfolio return minus the risk-free rate, and 2) the market factor, that is the monthly market return minus the risk-free rate. Moreover, the Fama-French models add the SMB, HML, RMW, and CMA factors. For each portfolio, the alpha is reported and tested for significance. If the test returns a significant alpha, it will be indicated by a star (*) and the number of stars will be determined by the significance level, i.e. the probability of incorrectly rejecting the null hypothesis. The significance is reported on three levels, 5% (*), 1% (**), and 0.1% (***), indicating a confidence interval of 95%, 99%, and 99.9%, respectively. Furthermore, the adjusted R-squared is presented for all performed models. The adjusted R-squared is an unbiased form of the simple R-squared estimator that measures the proportion of the variation in the dependent variable explained by the independent variables. The interpretation of the adjusted R-squared is hereby a bit different as it does not directly signal the proportion of explained variation, but it still gives an indication of the models’ explanatory power. Furthermore, the value is more comparable and reliable across different tests.
In order to study the robustness of the results, multiple variations of the tests are performed. For this part, it is only found necessary to report the results for the A and F portfolios, as these are considered most important in relation to the formulated hypotheses. First, the data is split into two sub samples, one containing the early five-year period (2011-2015) and another containing the late five-year period (2016-2020). This division enables the testing of a potential change in the ESG-investing interest. Furthermore, the constructed industry-weighted portfolios are being tested to see whether the obtained results from the value-industry-weighted portfolio analysis is caused by a sector-displacement rather than ESG performance. Lastly, the models will also be performed on a dataset excluding outliers, and the reason hereof is presented in section 4.3.4.