Results from Europe

We present the European results from our analysis of the decile portfolios in Table 7. When looking at the alpha, or the decile portfolios intercept, we find that the decile portfolios performed quite differently compared to each other. However, only one of the ESG portfolios (PF7) produce a statistically significant alpha on the 5% level in the FF3 model¹⁸.

We observe that the annualized alphas for the FF5 model follow a somewhat downward sloping tendency, with values decreasing from 1.18% in PF1 to -1.44% in PF2 and then increasing to 1.26% in PF5. Furthermore, we observe that PF7, which consist of the seventh decile of stocks with the highest ESG score, is producing the lowest alpha of -2.48%. The results show that ESG-rated portfolios have neither systematically higher nor systematically lower excess returns or risk.

Additionally, we observe that a long-short strategy in PF10-PF1 would generate an alpha of -2.78%, which is of economic relevance but not statistically significant.

In FF3, the alpha is decreasing from -1.13% in PF1 to -1.42% in PF2 before increasing to 1.18%

in PF6. When we introduce the momentum factor (Winner minus loser or WML) in the C4 model we see slightly higher positive and negative alphas but the same overall tendency. The C4 model show a negative alpha for PF1 of -1.28%. The alpha is increasing to 2.19% in PF5 before decreasing to -4.02% in PF7 and then increasing to -1.61% in PF10. The general indication that is formed, is that there are return and risk differences between the portfolios, but that the differences are mainly driven by portfolio specific criteria rather than a homogenous ESG factor. Additionally, we see that some alpha value changes significantly across different factor models. For instance, for PF1, alpha is negative in C4 and FF3, but positive in FF5.

18 Additionally, we have conducted the same cross-sectional analysis on the underlying E, S, and G scores for the European region. Similar to the aggregated ESG score, the results from the FF5 model do not yield any significant alphas, for any of the isolated pillars, across the decile portfolios or in the long-short portfolio (PF10-PF1). The results from the FF5 mode, for all pillars, can be found in Appendix (28)

Page 65 of 146

Table 7: Empirical results for the European region using aggregated ESG scores (Alpha) Table 7 presents the results from the FF3, C4 and FF5 for all decile portfolios, including a long-short portfolio (PF10-PF1) in the European region and their ability to earn alpha, when controlling for risk factors. In this strategy

the stocks have been sorted into decile portfolios based on their ESG scores. The 10% stocks with the lowest (highest) ESG scores are found in PF1 (PF10). All alphas are annualized, and the square brackets present the t-statistics. Significant codes: 0.001 (***), 0.01 (**), 0.05 (*). The sample period is between January 2007 and

December 2020.

Table 8 presents the coefficients for all decile portfolios in the European region from the FF5.

We observe that the regression yields statistically significant market coefficients, otherwise specified as the traditional beta of the portfolio, with coefficients that are significant at the 0.1%

level. This indicates that the returns of our portfolios are heavily correlated with the market excess return and that market excess return explains a lot of the variation in the portfolio excess return.

Furthermore, the majority of these coefficients are close to one, which indicates that our portfolios are almost as sensitive, or volatile, as the market. The market coefficient is increasing from 0.99 in PF1 to 1.16 in PF10, indicating that the highest rated ESG-portfolios are relatively more volatile.

For the long-short strategy in PF10-PF1 we find very little systematic risk with a market coefficient of 0.16, indicating that the portfolio has close to no market exposure.

For the SMB coefficient we observe that PF1 to PF4 have positive SMB coefficients ranging from 0.18 to 0.41, where PF2, PF3 and PF4 are statistically significant at a 5% level, indicating that the average excess returns of the portfolios are positively exposed to the SMB factor. This positive exposure suggest that companies included in the four portfolios are tilted towards stocks with smaller market caps. For PF5 to PF10, the SMB coefficients are negative and statistically significant for PF8, PF9 and PF10 at a 1% (PF8 and PF10 at the 0.01% level) level. The

Page 66 of 146

coefficients range from -0.01 to -0.39, indicating that the companies in these portfolios are tilted towards stocks with larger market caps.

Table 8: Empirical results for the European region using aggregated ESG scores (Factors)

Table 8 presents the results from the coefficients for the FF5 for all decile portfolios in the European region. In this strategy the stocks have been sorted into decile portfolios based on their ESG scores. The 10% stocks with the lowest (highest) ESG scores are found in PF1 (PF10). The square brackets present the t-statistics. Significant codes:

0.001 (***), 0.01 (**), 0.05 (*). The sample period is between January 2007 and December 2020.

Table 8 show that 9 out of 10 HLM coefficients are statistically insignificant, indicating that the hypothesis stating that HML is different from zero cannot be rejected. HML are generally negative for PF1 to PF4, ranging from -0.27 to 0.02, indicating that companies in these portfolios are tilted towards growth-stocks. Conversely, for PF8 to PF10 we find positive HLM coefficients, where the HML coefficients of PF9 is statistically significant at the 1%, suggesting that there is a tilt towards value-stocks in this portfolio.

Additionally, the RMW coefficients show no clear tendency. We observe that 8 out of 10 RMW coefficients are statistically insignificant. Only PF1 and PF10 is statistically significant at respectively the 5% and 1% level, both showing negative coefficients of respectively -0.45 and -0.32. From Table 8 we also observe that 4 out of 10 CMA coefficients are negative and statistically significant at the 5% level (PF1, PF4 and PF5 is on the 1% level), suggesting a negative exposure to this factor. The overall influence of the risk factors in the European region is not very consistent. Some ESG portfolios had statistically significant and positive exposure to a given

Page 67 of 146

factor, while others had a significant but negative exposure to the same factor – and some had insignificant or no noteworthy exposure to that same factor.