Table 5.10: CAPM, FF3 and FF5 (2016-2020)
Source: Own production
Comparing the descriptive statistics and testing results for the two sub samples, major differences can be observed. The results in the early sub sample gave a clear negative picture of ESG-investing. In the early period, the A portfolios were performing worse than the F portfolios both in terms of return and risk. Moreover, the F portfolios were generating significant positive alphas. However, the later sub sample shows a substantial positive development. According to the descriptive statistics, the A portfolios show superior performance in both the ESGC and S scores. Additionally, both the ESGC, E, and S scores have significant positive alphas in the A portfolios. Moreover, the downside risk difference between the A and F portfolios have also undergone a development across the two sub samples, similar to what was illustrated in figure 5.3. The A portfolios seem to show higher downside risk in the former, whereas the opposite is true for the latter. This development indicates that the ESG focus may have changed over the course of the last 10 years.
“best-in-class” investment strategy. For this part of the analysis, it is not found relevant to include the descriptive statistics. However, the performance benchmark models will be performed both on the full 10-year period and for the 5-year sub samples. Similar to previously, the results will only be reported for the A and F portfolios.
5.4.1 Results: CAPM, Fama-French Three- and Five-Factor Models
In table 5.11 the results are listed across all scores, models, and the two relevant portfolios, using the complete 10-year period of analysis. Starting from the top, the ESGC score have significant positive alphas in portfolio A across all models, but with the highest level of significance in the FF3 and FF5 models. The alphas signal monthly excess returns at 0.2% and 0.3% respectively. These results are similar, but slightly better, than the results obtained in section 5.2 where the significance level were lower and only present in the Fama-French models. In the value-weighted portfolios, the ESG score had significant positive alphas in portfolio F across all three models. Here, it is only present in the CAPM model and disappears when controlling for additional risk factors. For the E score, both portfolio A and F provide significant positive alphas. However, portfolio F is the superior one as it returns a higher nominal alpha. This is the same conclusion as previously drawn. The S score holds similar results to the E score, where both portfolio A and F have significant positive alphas, but portfolio F show higher monthly excess return. These results differ from previously, where the only significant positive alpha was found in the CAPM F portfolio. The same difference is applicable for the G score.
Turning to the risk factors, some similarities and differences are observed when comparing the industry-weighted portfolios and the value-industry-weighted portfolios. Previously, the market factor indicated a somewhat clear tendency, that the A portfolios exhibited lower market-related risk than the F portfolios. This is no longer the case, as the only score exhibiting this trend is ESGC score. The remaining risk factors display the same trends as the previously obtained results, and it is therefore not found relevant to elaborate further.
Table 5.11: Industry-Weighted: CAPM, FF3 and FF5
Source: Own production
5.4.2 Results: Sub samples
As the results from the 5-year sub samples in section 5.3 showed some interesting developments, it is also found important to test the industry-weighted portfolios in the two period splits. In the following section, the testing results will be presented for both the early (2011-2015) and late (2016-2020) sub sample.
The testing results from the early sub sample are listed in table 5.12, which will be compared to the results obtained in section 5.3.1. In the previous analysis, it became clear that portfolio F performed significantly better than portfolio A across all scores. This conclusion is similar for the scores ESG, E, S, and G. For these scores, portfolio F has higher positive alphas than their counter portfolio A, and several of these are significant.
However, the ESGC score show different results. In contradiction to the other scores, the ESGC score show significant positive alphas in portfolio A at a 5% level with monthly excess returns at 0.3%. The nominal alpha is equal to the one obtained for the full period of analysis. Thereby, the results are slightly better for the sector-neutral portfolios compared to the obtained results in section 5.3.1. However, the negative transformation between the results from the full data period and the early sub sample is still visible, similar to the
value-Score Model Portfolio Alpha MKT SMB HML RMW CMA Adj R^2
ESG comb CAPM A 0.002* 0.996*** 0.949
F 0.002 1.011*** 0.891
FF3 A 0.003*** 0.957*** -0.091 0.146*** 0.957
F 0.002 0.992*** 0.340*** 0.101* 0.913
FF5 A 0.003*** 0.946*** -0.102* 0.227*** 0.103 -0.082 0.957
F 0.001 0.976*** 0.344*** 0.360*** 0.484*** -0.059 0.922
ESG CAPM A 0.000 1.064*** 0.941
F 0.004** 0.904*** 0.858
FF3 A 0.001 0.991*** -0.101* 0.275*** 0.965
F 0.002 0.948*** 0.467*** -0.130** 0.902
FF5 A 0.001 0.968*** -0.133** 0.388*** 0.074 -0.200* 0.966
F 0.001 0.922*** 0.447*** 0.111 0.354** -0.178 0.908
E CAPM A 0.001 1.076*** 0.944
F 0.006*** 0.847*** 0.846
FF3 A 0.002** 1.009*** -0.101* 0.253*** 0.963
F 0.004*** 0.896*** 0.437*** -0.153** 0.892
FF5 A 0.002** 0.989*** -0.131** 0.333*** 0.025 -0.175 0.964
F 0.004** 0.851*** 0.378*** 0.095 0.210 -0.377** 0.900
S CAPM A 0.000 1.038*** 0.946
F 0.005*** 0.949*** 0.878
FF3 A 0.002* 0.974*** -0.147*** 0.235*** 0.966
F 0.003** 0.990*** 0.433*** -0.122** 0.913
FF5 A 0.002* 0.961*** -0.169*** 0.281*** -0.005 -0.128 0.966
F 0.003* 0.960*** 0.412*** 0.170 0.444*** -0.196 0.922
G CAPM A 0.000 1.017*** 0.933
F 0.003** 0.900*** 0.895
FF3 A 0.001 0.968*** -0.055 0.185*** 0.943
F 0.003* 0.920*** 0.191** -0.059 0.901
FF5 A 0.000 0.950*** -0.072 0.338*** 0.207* -0.135 0.945
F 0.003* 0.879*** 0.129 0.103 0.043 -0.369** 0.907
* p<0.05, ** p<0.01, *** p<0.001
weighted portfolio analysis. This can be seen through the higher performance gab between the two portfolios, favouring portfolio F.
Table 5.12: Industry-Weighted: CAPM, FF3 and FF5 (2011-2015)
Source: Own production
In table 5.13 the testing results have been listed for the late sub sample, which will be compared to the ones obtained in section 5.3.2. The ESGC score show significant positive alphas in both the FF3 and FF5 models, similar to previously, but no longer in the CAPM model. An important difference is that the ESG score also show positive significant alphas in the FF3 and FF5 models. Compared to the late sample in the value-weighted portfolio analysis, where the only significant alpha in the ESG score was present in the CAPM F portfolio.
The E score results also differ from the ones previously found, as both portfolio A and F have significant alphas in FF3 and FF5. Moreover, portfolio F has higher nominal alphas which indicate that it is superior. This differs significantly from the previous results, where only portfolio A returned significant alphas. The S and G scores provide the exact same conclusion as previously; the S score having significant excess returns in portfolio A, and vice versa for the G score. Overall, the results are similar for the ESGC, S, and G scores, but different for the ESG and E scores. Similar to the value-weighted portfolio analysis, the late sub sample provide a more
Score Model Portfolio Alpha MKT SMB HML RMW CMA Adj R^2
ESG comb CAPM A 0.002 0.974*** 0.937
F 0.003 0.844*** 0.882
FF3 A 0.003* 0.916*** -0.071 0.209*** 0.950
F 0.002 0.923*** 0.358*** -0.041 0.905
FF5 A 0.003* 0.908*** -0.079 0.258* 0.020 -0.120 0.949
F 0.002 0.902*** 0.334*** 0.056 0.020 -0.275 0.907
ESG CAPM A -0.001 1.033*** 0.929
F 0.005* 0.769*** 0.811
FF3 A 0.001 0.950*** -0.086 0.314*** 0.956
F 0.003 0.876*** 0.423*** -0.114 0.848
FF5 A 0.001 0.924*** -0.127 0.326** -0.112 -0.238 0.958
F 0.003 0.867*** 0.433*** 0.122 0.246 -0.316 0.854
E CAPM A -0.001 1.064*** 0.929
F 0.005** 0.798*** 0.804
FF3 A 0.001 0.975*** -0.085 0.342*** 0.961
F 0.004* 0.912*** 0.485*** -0.088 0.848
FF5 A 0.002 0.954*** -0.119 0.343** -0.102 -0.185 0.961
F 0.004 0.886*** 0.466*** 0.136 0.153 -0.445* 0.859
S CAPM A -0.001 1.023*** 0.938
F 0.007*** 0.816*** 0.848
FF3 A 0.001 0.943*** -0.105 0.283*** 0.961
F 0.005** 0.923*** 0.369*** -0.165* 0.879
FF5 A 0.001 0.927*** -0.134 0.241* -0.126 -0.090 0.961
F 0.005** 0.903*** 0.358*** 0.039 0.157 -0.376* 0.887
G CAPM A -0.001 0.981*** 0.939
F 0.001 0.878*** 0.866
FF3 A 0.000 0.936*** -0.022 0.195*** 0.951
F 0.001 0.950*** 0.274* -0.086 0.877
FF5 A 0.000 0.907*** -0.057 0.298** -0.010 -0.349** 0.956
F 0.001 0.916*** 0.238* 0.090 0.056 -0.464* 0.887
* p<0.05, ** p<0.01, *** p<0.001
positive outlook for ESG-investing, compared to the results obtained from both the full data period and the early sub sample.
Table 5.13: Industry-Weighted: CAPM, FF3 and FF5 (2016-2020)
Source: Own production
5.4.3 Summary Industry-Weighted Portfolios
The factor models CAPM, FF3, and FF5, performed across the entire period of analysis, generally provide the same conclusions as the value-weighted portfolio analysis. The ESGC score show significant positive alphas in portfolio A across all performed models, indicating outperformance by the “best-in-class” investment strategy. However, the remaining scores indicate outperformance in portfolio F, although only present in CAPM for the ESG score. According to the market-related risk, portfolio A only exhibits lower risk than portfolio F in the ESGC score, which is reversed for the remaining scores. Similar to previous analysis, the results are also studied for the two sub samples. The early sub sample give somewhat the same conclusions as the full period, with outperformance in portfolio A for the ESGC score and outperformance in portfolio F for the remaining scores. However, the results are still worse from the pro-ESG perspective, similar to section 5.3.1, as the outperformance in the F portfolios are significantly higher in the early period. For the late sub
Score Model Portfolio Alpha MKT SMB HML RMW CMA Adj R^2
ESG comb CAPM A 0.002 1.014*** 0.959
F 0.001 1.153*** 0.926
FF3 A 0.003** 1.012*** -0.195** 0.089* 0.966
F 0.002 1.079*** 0.170 0.178** 0.935
FF5 A 0.003** 0.998*** -0.197** 0.173 0.180 -0.012 0.967
F 0.002 1.048*** 0.189 0.391** 0.546** 0.066 0.944
ESG CAPM A 0.000 1.087*** 0.951
F 0.002 1.020*** 0.907
FF3 A 0.002* 1.042*** -0.203** 0.237*** 0.974
F 0.001 1.017*** 0.392*** -0.152** 0.933
FF5 A 0.002* 1.012*** -0.236*** 0.387*** 0.192 -0.153 0.975
F 0.001 0.994*** 0.393*** -0.006 0.326 -0.007 0.936
E CAPM A 0.002 1.082*** 0.956
F 0.006** 0.887*** 0.879
FF3 A 0.003** 1.051*** -0.188** 0.182*** 0.970
F 0.004** 0.898*** 0.403*** -0.202*** 0.924
FF5 A 0.003** 1.025*** -0.213** 0.318*** 0.191 -0.120 0.971
F 0.004** 0.832*** 0.335*** 0.119 0.418** -0.322 0.933
S CAPM A 0.001 1.048*** 0.953
F 0.003 1.067*** 0.918
FF3 A 0.003* 1.026*** -0.268*** 0.187*** 0.974
F 0.001 1.049*** 0.395*** -0.100 0.938
FF5 A 0.003* 0.999*** -0.306*** 0.306*** 0.104 -0.172 0.975
F 0.001 1.013*** 0.388*** 0.113 0.441** -0.044 0.944
G CAPM A 0.000 1.046*** 0.928
F 0.005*** 0.914*** 0.921
FF3 A 0.001 1.017*** -0.168 0.164** 0.939
F 0.005** 0.912*** 0.125 -0.045 0.922
FF5 A 0.002 1.006*** -0.140 0.265* 0.337* 0.114 0.943
F 0.005** 0.869*** 0.059 0.143 0.136 -0.302 0.923
* p<0.05, ** p<0.01, *** p<0.001
sample, a positive development is observed similar to section 5.3.2. Here, the higher-scoring portfolios outperform the lower-scoring portfolios in both the ESGC, ESG, and S scores. The F portfolios remain superior according to the E and G scores.