score of 0,122 shows that there is a higher frequency of negative returns in residuals with the kurtosis value increasing its combined value.
To test the hypothesis of whether the created regressor of ESG Leaders minus Laggers has an effect on the ex-cess return of the Nordic portfolio, we test it with the two-sided test. The null hypothesis is that it is equal to zero, meaning that it doesn’t affect the excess return against the alternative hypothesis that it is different from zero, meaning that it does have an effect on the excess return. Taking the values from the regression to derive the 𝑡 to the student t distribution, we get the following formula.
𝐻0: 𝛽𝐸𝑆𝐺 = 0 𝐻1: 𝛽𝐸𝑆𝐺≠ 0
𝑡 =−0,134 − 0
0,049 = −2,706
By knowing that for a two-sided t-test, the critical value at the 5% significant level is ±1,982 and the critical value at the 1% significant level is ±2,623 with 107 degrees of freedom. With this information, it can be con-cluded that 𝑡 value is greater than both the critical values and can therefore reject our null hypothesis saying that ESG Leaders minus Laggers doesn’t affect the excess return in favour of the alternative hypothesis.
Table 9: Yearly performance and representation in the Danish Portfolio
(Authors’ own creation, 2021).
In 2016 the ESG Leaders had an incredibly bad year. This was due to two out of three of the companies were the worst-performing stocks in the danish stock market this year. However, this their bad performance wasn’t due to ESG related performances, but these two companies were disappointed investors with their products high prices and achieved financial performance compared to their desired financial performance that year (Berlingske, 2016b).
Table 10: Movement of ESG Score and representation in the Danish Portfolio
(Authors’ own creation, 2021).
Regarding the ESG scores within danish companies, the overall score for companies in Denmark has risen as the 10th, and 90th percentile score has been increasing throughout the years 2011 to 2019. However, it is worth noticing that as in the Nordic portfolio, the highest increase of total companies is from 2016 to 2017 and has a big decreasing impact on the average percentile score for each portfolio. This might lead to the same suspicion that given it is a high percentage of the total companies first ESG score and hasn’t internally been focusing on the area until then. This means that the best comparison for the trend in the average ESG score across danish companies can be seen from 2017 to 2019, which is slightly increasing, and where Laggers has been slightly increasing as well. ESG Leaders, from 2018 to 2019, however, have decreasing ESG scores, indicating that top ESG companies in Denmark have lacked in their attempt to withhold their previous scores.
Year Average Return Number of companies Average Return Number of companies Average Return Number of companies 2011 -27,8% 24 -22,3% 3 -29,0% 3 2012 29,8% 24 24,5% 3 14,3% 3 2013 41,7% 24 78,0% 3 34,3% 3 2014 8,0% 25 9,5% 3 11,8% 3 2015 31,5% 26 39,9% 3 17,8% 3 2016 6,8% 28 -10,5% 3 -1,5% 3 2017 22,2% 38 23,3% 4 19,9% 4 2018 -5,8% 42 -1,5% 5 -9,5% 5 2019 20,1% 42 14,2% 5 15,0% 5 Average 14,0% 30 17,2% 4 8,1% 4
Portfolio ESG Leaders ESG Laggers
Year Number of Companies Average ESG Score 90th Percentile Score Number of Companies Average ESG Score 10th Percentile Score
2011 3 73,3 69,8 3 10,1 16,1 2012 3 72,2 69,1 3 13,3 24,8 2013 3 70,0 67,6 3 16,5 25,7 2014 3 71,0 67,9 3 23,2 30,8 2015 3 74,1 71,1 3 27,1 34,4 2016 3 76,3 71,8 3 28,7 36,2 2017 4 75,9 73,8 4 18,8 28,3 2018 5 79,2 75,5 5 22,2 31,4 2019 5 77,9 75,4 5 22,2 31,4 Average 4 74,4 71,3 4 20,2 28,8
ESG Leaders ESG Laggers
Figure 8: Visual Presentation of Danish Portfolio Performance
(Authors’ own creation, 2021).
Looking at the graph of the overall performance for the danish portfolio and the danish ESG portfolio, it is clear to see that, despite the year 2013, they are behaving in the same patterns as the Nordic ESG portfolio. It is worth noticing that it is performing well when the economies are in bad times and in good times following the trend of the market. In comparison, the combined danish portfolio supports the theory of that size and value matters regarding the optimal portfolio variance as it outperforms the market by far.
The Danish portfolio yield an adjusted 𝑅2 of 0,599 and 𝑅2 of 0,614 representing a strong linear relationship among the regressors in the model. Thus, showing that 61,4% of the variance can be explained by the created portfolio regressor. In the case of Denmark, by looking at the P-value of the Danish ESG portfolio excess re-turns, we see that the P-values yield statistically insignificant results. However, the other three other portfolios, the markets excess return, HML and SMB, yield results that imply them to be statistically significant on 1%
and 5% level respectively. Thus, showing evidence against the hypothesis that there is no relationship between the individual portfolio and excess return regarding the selected Danish companies. However, this cannot be said regarding ESG leaders and Laggers.
Table 11: Danish Portfolio Regression
(Authors’ own creation, 2021).
Delving into the coefficients of our Danish portfolio, first off, we see that the coefficient of the market pre-mium is 𝛽𝑀𝑘𝑡 = 0,853, representing the fact that the portfolio is less volatile than the market, decreasing the risk of the portfolio. The two following portfolios in the graph show coefficient of 𝛽𝑆𝑀𝐵 = 0,311 and 𝛽𝐻𝑀𝐿= 0,088 supporting the theory of smaller companies outperforming larger ones. The negative coefficient for 𝛽𝐸𝑆𝐺= −0,001, again, represents the excess returns produced by the portfolio. Again, the statistics signal that a higher ESG score will reduce risk and reduce the expected cost of equity.
Table 12: VIF Test of Danish Portfolio regression
(Authors’ own creation, 2021).
When checking for multicollinearity through the VIF-test and comparing it to the rule of thumb, see that the VIF values show close to no correlations among the regressors. Furthermore, there is an indication from the mean VIF that the variance can be up to 8% compared to a situation with no multicollinearity at all. Thus, these results support the assumption of homoscedastic in the regression on the Nordic portfolio.
Table 13: Test for normal distribution in Danish Portfolio regression
(Authors’ own creation, 2021).
Continuing to look at the validity of our regression, we move on to look at if the data follows a normal distri-bution. This is important as linear and non-linear regressions should both follow a normal distridistri-bution. By looking at the values derived from the Skewness and kurtosis test, as seen in the figure above, the two values, 0,289 and 0,260, show that the regression is not normally distributed as it does not exceed the 0,500 thresh-olds. The combined score of 0,294 shows that there is a higher frequency of negative returns in residuals and indicating a higher frequency of negative excess returns creating a longer right tail of the distribution.
Lastly, to test our hypothesis of whether the created regressors of ESG Leaders minus Laggers have has an ef-fect on the excess return of the Nordic portfolio, we test it with the two-sided test. Having the same null hy-pothesis as in the earlier sections where there is no effect on the excess returns while the alternative hyhy-pothesis represents a situation where the ESG affects the excess returns. Using the values from our regression
𝐻0: 𝛽𝐸𝑆𝐺 = 0 𝐻1: 𝛽𝐸𝑆𝐺≠ 0
𝑡 =−0,001 − 0
0,055 = −0,02
By knowing that for a two-sided t-test, the critical value at the 5% significant level is ±1,982 and the critical value at the 1% significant level is ±2,623 with 107 degrees of freedom, we cannot reject our null hypothesis of ESG Leader minus laggers not affecting the excess returns.