Backtesting ESG Factor Investing Strategies
Backtesting ESG Factor Investing Strategies
Abstract
This paper takes an in-depth look at socially responsible investing and problems associated with
it. One of the main problems with ESG factor investing is caused by data. Firstly, we obtained
unfiltered ESG data from OWL Analytics. Secondly, we reviewed two strategies based on ESG
Factor investing: ESG Factor Momentum Strategy and ESG Level Factor Strategy. After we
tested both of these strategies on our data, we concluded factor strategies based on ESG scoring
seem to be profitable during the last several years. Our findings confirm the broader trend of
recent outperformance of strategies based on ESG. Unexpectedly ESG level had better
performance than the ESG momentum. Performance of the momentum was positive, although
low; compared to the notable performance of ESG level. ESG scoring seems to be applicable
to use in the portfolio as an addition to other better-known factors. Nonetheless, we still
recommend caution because of the inconsistency of the ESG data among data providers.
Authors:
Radovan Vojtko, CEO & Head of Research, Quantpedia
Daniela Hanicová, Quant Analyst, Quantpedia
Introduction to ESG
A reader who knows what the ESG investing is can skip this part and go directly to backtesting,
as we will shortly explore few ESG basics in the next few paragraphs.
How can we measure if any company is responsible to the environment and overall society?
Simply, by grading it – the ESG score measures the firm’s quality in three categories:
environmental, social and governance. Each category is affected by different things. The
environmental score takes into account carbon emissions, low resource consumption, pollution,
innovations aiming to improve environmental protection and much more. Social score focuses
on human rights, business ethics, safety standards for workers, cash donations, protection of
public health and so on. Factors that affect the governance score can be rights of shareholders
or the company’s financial and non-financial goals.
The question of why would someone use ESG score as an indicator has multiple answers.
Firstly, there might be a non-financial motivation, for example, a religious belief. Certain
religions may want to exclude sectors like the gambling, alcohol or armament industry from
their portfolio. Secondly, the philanthropic point of view that seeks green, socially responsible
investments is gaining popularity in recent years. High ESG score is seen as appealing for those
investors. And lastly, high ESG score can signal forward-thinking and efficient company, and
those attributes can signal better financial performance.
Of course, there are a lot of research papers which are focused on the applicable ESG investing
strategies; they examine if using ESG score enhances, harms or has no effect on the
performance. If researchers confirmed the performance-enhancement, it would mean that ESG
scores could be utilized as factors, and used in practice. Also, if there was no correlation
between ESG score and performance, an investor, who seeks sustainable investing could be
interested in it (as his performance would not be penalized). On the other hand, the focus on
the ESG scores could lead to investing in unprofitable companies or reduce an investment
universe in a way that wouldn’t be diversified.
The separate problem with ESG score is that it’s hard to measure it objectively. There are
multiple data providers with huge databases, but each provider can value different elements of
Related literature
In a paper written by Fulton, Kahn and Sharples: Sustainable Investing: Establishing Long-
Term Value and Performance [5], researchers have found that firms with high ESG ranking can
better finance their activities since they have a lower cost per capital. Stocks of these firms have
a much lower risk than others. This paper also states that "88% of studies of actual SRI fund
returns show neutral or mixed results "and "we have found that SRI fund managers have
struggled to capture outperformance in the broad SRI category, but they have, at least, not lost
money in the attempt. "
Another research examined the relationship between ESG and performance in the years 2010 –
2017. According to Bennani, Le Guenedal, Lepetit, Ly, Mortier, Roncalli and Sekine: How
ESG Investing Has Impacted the Asset Pricing in the Equity Market [1], ESG investing has
become profitable since 2014. During 2010 and 2013, the article did not find any indication of
the profitability of ESG integration. According to the study, ESG scores in stock picking
strategies would have led to neutral or negative results. It is different in the following part of
the study. In more recent years, there is a notable change, as most indicators turned positive
between 2014 and 2017.
Other pieces of researche such as Huppé: Alpha's Tale: The Economic Value of CSR [8] suggest
that corporate social investing alpha arose because investors have continuously undervalued
information about social responsibility. Despite, omitting such relevant information it leads to
a surprise of the performance of the corporate social responsibility leaders and the re-evaluation
of their choices. Research by Clark, Feiner and Viehs: From the Stockholder to the Stakeholder:
How Sustainability Can Drive Financial Outperformance [3] studied over 200 sources and
found a correlation between sustainable industries and economic performance. In the paper,
88% of reviewed sources declare that sustainable practices are applied to better cashflows,
where the mechanism is simply a better operational performance.
Secondly, there is still a problem with the various data sources for the ESG scores. Research
by Berg, Kölbel and Rigobon: Aggregate Confusion: The Divergence of ESG Ratings [2]
analyzed data from five different rating agencies (KLD, Sustainalytics, Vigeo-Eiris, Asset4,
and RobecoSAM). The paper has found out that the average correlation between scores is 0.61
and that this value ranges from 0.42 to 0.73. This makes it difficult or even impossible to price
ESG scores in the market correctly. Besides, it is hard for companies to improve their perception
and sustainability, since according to one ESG score, the company could be one of the better
firms, while the other ESG score can register something very different. This results in the fact
that any study on the relationship of the scores and performance is impacted by the choice of
the data provider.
Another possibility is not to lose the lowest ESG stocks from the investment universe, but to
tilt towards better-rated stocks. The study conducted by Nagy, Kassam and Lee: Can ESG Add
Alpha? An Analysis of ESG Tilt and Momentum Strategies [9] examined such strategy, using
the stocks of the MSCI World index during 2007-2015. ESG ratings were obtained from the
MSCI (IVA scores). The strategy was profitable, but the performance was only 1.06%.
A somewhat different approach is applied in the paper by Dorfleitner, Utz and Wimmer: Where
and When Does It Pay to Be Good? A Global Long-Term Analysis of ESG Investing [4], a
source paper of ESG Factor Investing Strategy in the Quantpedia's screener. This paper uses
data from the Asset4 and stocks that are scored by this firm in the various regions (North
America, Europe, Japan and the Asia Pacific). The period used in this paper is from 2002 til
2011. We also used this strategy, and its detailed description is in the latter part of this paper.
The most promising results from this paper are for the North America and Europe region, which
is not a surprise as the ESG investing is most common in these parts of the world. The theory
expects that in the short term, firms with high ESG scores would have abnormal returns of zero,
but in the long run, theory expects positive abnormal returns. On the other hand, firms with low
ESG scores are expected to have negative abnormal returns. Additionally, outcomes indicate
that financial markets are inefficient in terms of ESG scores and are not capable of sufficiently
pricing different levels of corporate social performance in the short run and particularly in the
long run.
Paper presented by Nagy, Kassam and Lee: Can ESG Add Alpha? An Analysis of ESG Tilt and
Momentum Strategies [9] has also tested a momentum strategy. Authors propose that a firm
which has improved the ESG the most is expected to outperform. Although the advantages of
a better-rated ESG portfolio are anticipated to be evident only in the long term, according to
the paper, the market could likely react to a change in rating in a relatively short period. Given
the rise of ESG investing, we may see a reduction in the response period. Therefore, the strategy
is to overweight, relative to the MSCI World Index, companies that increased their ESG ratings
most during the recent past and underweight those with decreased ESG ratings. Lastly, the
strategy is profitable, but the profit is low similarly to our conclusions.
Does ESG scoring add any incremental information? Can we build equity factor strategies
based on it? We decided to test it for ourselves.
We went to look for the ESG data provider and obtained an unfiltered ESG data from OWL
Analytics.
The data set we worked with contained monthly data from March 2009 to October 2019; for
each month, we had from 5000 (at the beginning of the sample) up to almost 30 000 companies
to work with. For each company, we got its ISIN, shareClassFIGI, region, and over 200 detailed
parameters for the E, S, and G score and of course, the total ESG score.
Because we wanted to focus on the U.S. market, we filtered companies from that region, and
through their ISIN, we paired them with their ticker and price for each month. Then we deleted
small firms and those to which we did not get the ticker or price from our data set. In the end,
we were left with around 700 companies.
The first strategy we decided to test was the ESG Factor Momentum Strategy. This strategy is
based on overweighting companies, whose ESG score increased in the recent past and
underweighting companies, whose ESG score decreased. The increases and decreases are based
on a 12-month ESG momentum, and the strategy is usually rebalanced monthly.
There are multiple methodologies of how factor strategies can be built. We decided to follow
the standard approach of building equity long-short factor strategies – sorting into deciles and
going long the best and going short the worst decile.
Our implementation of momentum is as follows; we took our finalized data, and we analyzed
the percentage shift in ESG scores in the last twelve months. We took into an account the two-
month delay in the data. Then we divided the scores into deciles; we go long on the 10% of the
stocks with the greatest ESG momentum and short on the 10% of the stocks with the lowest
ESG momentum. The portfolio was value-weighted, so the higher the market capitalization the
stock had, the greater weight we gave it. That solved the problem of small stocks having a great
impact on the performance of our portfolio. The holding period was three months, unlike the
one month that is usual for this strategy. Because of the two-month delay (skip period), we
wanted some time for the momentum to manifest; therefore, one month holding period after
two-month skip period seemed too short to us, and we picked three months instead. We also
implemented tranching – each month, we rebalanced 1/3 of the portfolio, so in three months,
the whole portfolio is rebalanced.
The second strategy we tested was the ESG Level Factor Investing Strategy. This strategy, like
the ESG Factor Momentum Strategy, is value-weighted but is rebalanced on a yearly basis.
Our implementation of this strategy is as follows: Firstly, we divided the data into quintiles, we
went long on the top 20% of the stocks with the greatest aggregate ESG level and short on the
20% of the stocks with the lowest aggregate ESG level. We decided to use the aggregate ESG
score for our strategy to keep things simple. But data from OWL Analytics offers more detailed
scores for each of the environmental, social and governance scores. Therefore, a more
sophisticated strategy built from multiple sub-strategies based on each of the E, S and G sub-
scores can be probably built.
We decided to hold stocks for 12 months. The reason for this is that ESG level is similar to
value or other fundamental factors, which take longer to manifest the information to the price,
unlike the ESG momentum which is closely related to other momentum-based factors. Because
of the long holding period, we had to divide the portfolio into twelve parts, and each month we
rebalanced 1/12 of the portfolio, so the entire portfolio was rebalanced in one year.
Conclusion
After we tested both of these strategies on our data, we can conclude that factor strategies based
on ESG scoring seem profitable during the last several years. Our findings confirm the broader
trend of recent outperformance of strategies based on ESG.
What surprised us is that the ESG level had better performance than the ESG momentum.
Performance of the momentum was positive, although low even after the ten year period.
However, ESG level performed unexpectedly well. ESG scoring seems to be applicable to use
in the portfolio as an addition to other better-known factors (like value, momentum,
low - volatility, etc.); however, we still recommend caution. Because of the fact that the ESG
data are not consistent among data providers, the reproduction of a certain approach may lead
to different results if the same dataset is not used. A possible solution is to work with as many
ESG datasets as possible. But we hope that we have presented the glimpse into the possible
application of ESG scores from the quantitative point of view. Either for profit-enhancing
properties or producing more socially responsible strategies.
[2] Berg, Florian and Kölbel, Julian and Rigobon, Roberto, Aggregate Confusion: The
Divergence of ESG Ratings (August 17, 2019). MIT Sloan Research Paper No. 5822-19.
Available at SSRN: https://ssrn.com/abstract=3438533 or
http://dx.doi.org/10.2139/ssrn.3438533
[3] Clark, Gordon L. and Feiner, Andreas and Viehs, Michael, From the Stockholder to the
Stakeholder: How Sustainability Can Drive Financial Outperformance (March 5, 2015).
Available at SSRN: https://ssrn.com/abstract=2508281 or
http://dx.doi.org/10.2139/ssrn.2508281
[4] Dorfleitner, Gregor and Utz, Sebastian and Wimmer, Maximilian, Where and When Does
It Pay to Be Good? A Global Long-Term Analysis of ESG Investing (October 7, 2013). 26th
Australasian Finance and Banking Conference 2013. Available at SSRN:
https://ssrn.com/abstract=2311281 or http://dx.doi.org/10.2139/ssrn.2311281
[5] Fulton, Mark and Kahn, Bruce and Sharples, Camilla, Sustainable Investing: Establishing
Long-Term Value and Performance (June 12, 2012). Available at SSRN:
https://ssrn.com/abstract=2222740 or http://dx.doi.org/10.2139/ssrn.2222740
[6] Gibson, Rajna and Krueger, Philipp and Riand, Nadine and Schmidt, Peter Steffen, ESG
Rating Disagreement and Stock Returns (August 6, 2019). Available at SSRN:
https://ssrn.com/abstract=3433728 or http://dx.doi.org/10.2139/ssrn.3433728
[7] Gloßner, Simon, The Price of Ignoring ESG Risks (May 18, 2018). Available at SSRN:
https://ssrn.com/abstract=3004689 or http://dx.doi.org/10.2139/ssrn.3004689
[8] Huppé, Gabriel A., Alpha's Tale: The Economic Value of CSR (July 30, 2011). Principles
for Responsible Investment Academic Conference, September 2011. Available at SSRN:
https://ssrn.com/abstract=1969583 or http://dx.doi.org/10.2139/ssrn.1969583
[9] Nagy, Zoltán and Kassam, Altaf and Lee, Linda-Eling, Can ESG Add Alpha? An Analysis
of ESG Tilt and Momentum Strategies (2016). Available at:
https://www.semanticscholar.org/paper/Can-ESG-Add-Alpha-An-Analysis-of-ESG-Tilt-and-
Nagy-Kassam/64f77da4f8ce5906a73ffe4e9eec7c49c0960acc
[12] Varma, Abhishek and Nofsinger, John R., Socially Responsible Funds and Market Crises
(September 6, 2012). Journal of Banking and Finance, Forthcoming. Available at SSRN:
https://ssrn.com/abstract=2142343 or http://dx.doi.org/10.2139/ssrn.2142343
[13] Verheyden, Tim and Eccles, Robert G. and Feiner, Andreas, ESG for All? The Impact of
ESG Screening on Return, Risk and Diversification (2016). Journal of Applied Corporate
Finance, 28(2), 47-55. Available at SSRN: https://ssrn.com/abstract=2834790
[14] Vojtko, Radovan and Padysak, Matus, Quant’s Look on ESG Investing Strategies
(January 7, 2020). Available at SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3504767