The Expected Returns of ESG Excluded Stocks. Shocks to Firms Costs of Capital? Evidence From the World’s Largest Fund (2022)
Erika Berle, Wanwei (Angela) He, Bernt Arne Ødegaard
SSRN Working Paper, URL
After criticizing the subjective character of ESG, this week’s AGNOSTIC Paper examines the ESG exclusions of a popular investor: the Norwegian sovereign wealth fund, also known as the “oil fund”.1More information about the fund here. The fund is one of the largest investment vehicles in the world and, according to the paper, held 9,338 stocks from 65 countries at the end of 2021. The market value of this equity portfolio at this time was about $1,014B.
By and large, the fund is very similar to a passive developed-market equity index. However, already back in 2002, the Norwegian government introduced some ethical guidelines. Over the years, the fund therefore excluded about 190 companies that engage in different types of “bad” practices or products. These exclusions are interesting because they reveal insights about the impact of ESG for one of the largest institutional investors in the world.
Everything that follows is only my summary of the original paper. So unless indicated otherwise, all tables and charts belong to the authors of the paper and I am just quoting them. The authors deserve full credit for creating this material, so please always cite the original source.
Setup and Idea
The paper two weeks ago already shows that the ESG profile of a company has real consequences for it’s fundamentals and stock returns. The authors of this week’s paper build on this insight and present two intuitive explanations from the literature.
The first idea goes like this. Investors are, for whatever reason, willing to pay a permanent premium for companies with a good ESG profile and thereby lower their cost of capital.2For the mechanism, the specific reasons doesn’t really matter. On the contrary, “bad” companies trade at a discount and have a higher cost of capital. All else equal, “bad” companies should therefore offer higher expected returns on their stock. This is exactly the mechanism we would like to achieve for society. More expensive capital and worse access to investors force “bad” companies to change their business models and gravitate towards more sustainable alternatives. The other side of this, of course, is that investors who don’t care about ESG may find some attractive investment opportunities.
The second mechanism how ESG could affect returns is more temporary. If enough investors exclude a particular company, there will be selling pressure on the stock.3This is probably especially relevant when ESG ratings suddenly change. This may push the stock price below it’s theoretical value until some non-ESG investors step in and seize the opportunity. Followers of this explanation argue that the impact of ESG is mostly short term and concentrated around the exclusion trades of large investors. This is quite different (but not necessarily opposing) to the permanent premium of the first idea.
No matter which explanation you like more, both of them suggest that there is some money to be made from stocks with low ESG ratings.4As so often in economics and finance, both mechanisms are probably active at the same time. The authors examine how much by analyzing the performance of the ESG exclusion from the Norwegian oil fund. This is of course just a case study. But given the size and reputation of the oil fund, the authors argue that the results could also be representative for other institutional investors.
Data and Methodology
The most important data source are announcements of ESG exclusions from the ethical council of the Norwegian government and the oil fund itself. The authors collect all 189 ESG exclusions between 2005 and 2021 and match them to stock prices and market capitalization from Refinitiv.5Refinitiv is a high quality data provider, so there shouldn’t be any data issues. The authors also use Yahoo Finance to obtain FX-rates. This is not the most reliable database, but it should still do the job for FX-rates. Of the 189 exclusions, 67 were excluded because of conduct (companies doing bad things like corruption), and 122 because of their products (companies offering bad stuff like tobacco).
Note that compared to the >9,000 stock investment universe of the fund, 190 exclusions are not that much. The excluded companies are therefore a very curated sample with (probably) really bad ESG profiles. Another important aspect is the year 2014 when the oil fund excluded all companies that produce coal or coal-based energy. This led to a considerable spike of exclusions.
The authors use this data to construct exclusion portfolios where stocks enter in the month after the exclusion-announcement. This time lag ensures that there is no look-ahead bias and is best practice in asset pricing research.6Look-ahead bias means that you hypothetically trade on information that wasn’t available at the time of trading. If an exclusion gets revoked (that happens for 26 companies), the stock will leave the exclusion portfolio at the end of the respective month.
Important Results and Takeaways
ESG-excluded stocks generated up to 6.85% alpha per year
The following charts show the cumulative returns of exclusion portfolios between 2005 and 2021 in USD. The authors present both equal- and value weighted portfolios and assume monthly rebalancing. Both exclusion portfolios clearly outperform the global stock market with value weights being substantially better than equal weights. However, the authors also mention that the exclusion portfolios underperformed the market tremendously during the last two major crashes in 2008 and 2020. So there is considerable outperformance but during the worst of all times, the “bad” companies performed even worse.
To analyze the outperformance more thoroughly, the authors next run performance regressions to identify alphas versus the Fama-French 5-factor model. For robustness, they also present alphas versus just the market, the Fama-French 3-factor model, and the Carhart 4-factor model.7The Carhart 4-factor model adds momentum to the Fama-French 3-factor model. Despite robust evidence and popular opposing views, Fama and French still don’t add momentum to their model. The following table summarizes the results.
The results underline the outperformance of the exclusion portfolios. Between 2005 and 2021, equal- and value weighted portfolios generated statistically significant (annualized) alphas of 5.17% and 6.85% versus the 5-factor model. For other specifications, alphas are even higher and reach up to >9% per year. The magnitude of these results is remarkable and suggests that “bad” stocks have indeed characteristics that are not explained by those well-known factors. According to the factor loadings in the table, excluded stocks also tend to have a beta lower than one. The authors argue that is because the exclusion portfolios contain a lot of utility stocks which tend to be more defensive.
The authors also provide some robustness checks to stress-test their results. First, they isolate US-exclusions and find a similar pattern.849 of the 189 exclusions are US companies. Second, they change the rule when stocks enter the portfolio to also capture the month of the exclusion-announcement. Third, they split the sample in two consecutive periods (2005-2015 and 2006-2021) to estimate alphas. And finally, they exclude Walmart (which has a disproportionally large weight) from the value weighted portfolios.
The results are of lower magnitude and less statistical significance for some of those additional tests. But overall, the empirical evidence that ESG-excluded stocks generated meaningful outperformance remains strong.
There seems to be a return premium on “bad” stocks
The exclusion portfolios hold stocks from the month after the exclusion-announcement until the end of the sample period. So with the exception of 26 exclusions that get revoked during the sample period, stocks remain in the portfolio for quite some time. This long term orientation makes it unlikely that outperformance of “bad” companies is only driven by short term corrections of selling pressures. So the first of the two possible explanations seems to be more relevant.
As a consequence, the authors argue in favor of a permanent return premium for “bad” companies. They also suggest to add an ESG-factor to asset pricing models. The good news is that a higher cost of capital forces “bad” companies to work on their ESG issues. In the end, most companies want (and need) to be attractive to more investors than just those who ignore (or can afford to ignore) ESG.9Various regulation forces many institutional investors to consider ESG in their investments. So if a company is on the exclusion list of the Norwegian oil fund, chances are high that other institutions will also not invest.
Conclusions and Further Ideas
This is again a paper which not necessarily meets the quality criteria that I outlined in the introduction to AGNOSTIC papers (at least not yet). It’s an unpublished working paper and the authors just uploaded the second draft on SSRN. Nonetheless, I think it fits quite well to the other papers on ESG and the case study of the Norwegian oil fund is highly practical.
In my opinion, it is most interesting to see that ESG clearly came at a cost. At least on average, the excluded stocks outperformed the market substantially and the fund would have made more money by holding to them. I don’t say it’s better to ignore ESG for higher returns (everybody must decide for herself). But I do think it’s important to have realistic expectations. ESG exclusions narrow the investment universe and unless you only eliminate future losers, a constrained optimum can’t be better than an unconstrained.10If ESG exclusions only avoid future losers, that would be a very attractive investment strategy and we would all become ESG investors, right? I think you see the argument…
From the perspective of a greedy investor who doesn’t care about ESG, the results suggest an interesting investment strategy.11Just to avoid any misunderstandings: this is not necessarily my personal view. Just buy into all exclusions of the Norwegian oil fund and bet on a continuing return premium for those “bad” companies. To do that with more confidence, however, we need some more information about the strategy. How many stocks were in the portfolio at each point in time? Was it reasonably diversified, especially at the beginning of the sample period? What about industry and country exposures? The authors don’t provide much details on such issues. This is unfortunate but okay. Their job is publishing papers, not creating trading strategies.
It would also be interesting to see the return premium on “bad” stocks in the entire cross section of returns. As I mentioned above, the authors call for adding an ESG factor to asset pricing models. But so far, they are just looking at a very curated sample of stocks and a discretionary ESG methodology of one investor. This is of course a very good first step but (in my opinion) not yet empirically robust.
Finally, one important disclosure. While I was writing this week’s post, the authors released an updated version of the paper on SSRN. I used the most recent charts and tables to avoid data errors, but I haven’t covered the new contents. However, from skimming them, I am quite confident that the overall results doesn’t change. In fact, the authors mainly provide additional evidence for the permanent return premium on “bad” companies.
- AgPa #66: Machine-Learned Manager Selection (2/4)
- AgPa #65: Machine-Learned Manager Selection (1/4)
- AgPa #64: Fund Manager Multitasking
- AgPa #63: Fire the Winners and Hire the Losers
This content is for educational and informational purposes only and no substitute for professional or financial advice. The use of any information on this website is solely on your own risk and I do not take responsibility or liability for any damages that may occur. The views expressed on this website are solely my own and do not necessarily reflect the views of any organisation I am associated with. Income- or benefit-generating links are marked with a star (*). Please also read the Disclaimer.
|1||More information about the fund here.|
|2||For the mechanism, the specific reasons doesn’t really matter.|
|3||This is probably especially relevant when ESG ratings suddenly change.|
|4||As so often in economics and finance, both mechanisms are probably active at the same time.|
|5||Refinitiv is a high quality data provider, so there shouldn’t be any data issues. The authors also use Yahoo Finance to obtain FX-rates. This is not the most reliable database, but it should still do the job for FX-rates.|
|6||Look-ahead bias means that you hypothetically trade on information that wasn’t available at the time of trading.|
|7||The Carhart 4-factor model adds momentum to the Fama-French 3-factor model. Despite robust evidence and popular opposing views, Fama and French still don’t add momentum to their model.|
|8||49 of the 189 exclusions are US companies.|
|9||Various regulation forces many institutional investors to consider ESG in their investments. So if a company is on the exclusion list of the Norwegian oil fund, chances are high that other institutions will also not invest.|
|10||If ESG exclusions only avoid future losers, that would be a very attractive investment strategy and we would all become ESG investors, right? I think you see the argument…|
|11||Just to avoid any misunderstandings: this is not necessarily my personal view.|