3 - Esg - ML - Social Positioning Within The SP 500

Overall ESG &
Social Positioning
within the S&P 500
WHITE PAPER
by ORENDA’S QUANT RESEARCH TEAM
Orenda Software Solutions Inc.

211 Yonge Street, #502, Toronto, ON M5B 1M4 CANADA
https://www.orendasolutions.com
Orenda’s Overall ESG & Social Positioning within the S&P 500 | White Paper
© 2020 Orenda Software Solutions Inc. All rights reserved.
Disclaimer
The information provided in this document is provided "as is" without warranty of any kind.
Orenda Software Solutions Inc. disclaims all warranties, either express or implied, including the
warranties of merchantability and fitness for a particular purpose. In no event shall Orenda
Software Solutions Inc. be liable for any damages whatsoever including direct, indirect,
incidental, consequential, loss of business profits or special damages, even if Orenda Software
Solutions Inc. has been advised of the possibility of such damages.
Document Lifetime
Orenda Software Solutions Inc. may occasionally update online documentation between releases
of the related software. Consequently, if this document was not downloaded recently, it may not
contain the most up-to-date information. Please refer to https://orendasolutions.com/use-cases/
for the most current information.
From the Web site, you may also download and refresh this document if it has been updated, as
indicated by a change in this date: June 30, 2020.
Where to get help
Orenda Software Solutions Inc. support, product, and datasets licensing information can be
obtained as follows.
• Product information — For documentation, release notes, datasets updates, or for
information about Orenda Software Solutions Inc. products, licensing, and service, go to
the Orenda Software Solutions Inc. website at: https://orendasolutions.com/contact/
• Technical support — For technical support, go to https://orendasolutions.com/contact/
and email us with requests. Our support engineers will contact you in less than 4 hours
during 8AM to 5PM EST.
Your comments
Your suggestions will help us continue to improve the accuracy, organization, and overall quality
of our publications. Please send your opinion of this document to: eze@orendasolutions.com.
© Orenda Software Solutions Inc. 2|Page

Orenda’s Overall ESG & Social Positioning within the S&P 500
Ezequiel Machabanski* and Jose Jatar ** and Tanya Seajay ***
June 30, 2020
Abstract
In this paper, we demonstrate that publicly traded companies that are socially aligned with
stakeholders’ values and behave favorably with good environmental and social actions and strong
governance, as well as prioritizing social wellbeing, are virtuous investments. By holding these
assets in a quarterly rebalanced investment strategy, we outperform the underlying, predefined,
benchmark. To investigate our hypothesis, we employed social positioning scores, quantified by
Orenda Software Solutions, as a tool to select socially aligned companies within the S&P 500 for
the period of September 2017 to March 2020.
Keywords: social positioning, esg, s&p, s&p 500, factor investing, alternative data
* Ezequiel Machabanski, CFA, Head of Financial Modelling, Orenda Software Solutions, Inc.
Email: eze@orendasolutions.com
** Jose Jatar, MA, Senior Research Analyst, Orenda Software Solutions, Inc.
Email: jose@orendasolutions.com
*** Tanya Seajay, MCM, CEO and Founder, Orenda Software Solutions, Inc.
Email: tanya@orendasolutions.com
Special acknowledgment to Team Orenda: Rachit Mukesh Trivedi, MSc, Data Scientist, for
developing and constantly improving Orenda’s machine learning techniques as well as his direct
involvement in deploying models, integrating APIs, and undertaking a thorough quality assurance
process. And for the ongoing commitment and support of Cory Musgrave, Software Engineering
Manager, and Sondra Lahey, Project Manager.

1. Introduction
This paper analyzes how publicly traded companies that prioritize the maximization of
stakeholders’ wellbeing and the communities where they conduct business are positioned to
perform better than their peers and a predefined benchmark for security selection and portfolio
construction.
Consumers across the globe and especially those who belong to younger generations, make
purpose-driven buying decisions. “Millennials prefer to do business with corporations and brands
with pro-social messages, sustainable manufacturing methods and ethical business standards”,
(Sarah Landrum, Forbes). If they feel that companies are falling out of alignment with their shared
values, they desist from conducting business with such companies and seek alternatives they can
trust.
Trust is a deciding factor for buying decisions. Edelman Holdings reported that 81% of
consumers will buy from companies when “they are able to trust the brand to do the right thing”
(Edelman, 2020). This behavior also applies to crises periods, such as COVID19 where “to return
to previous spending behaviours, retailers have to understand that customers will want to see things
like plexiglass barriers, the use of PPE, frequent disinfection, a distanced layout and hand sanitizer
availability, to build trust” (Gooding, 2020) . When consumers feel their values have not been
met, trust erodes, and companies suffer financially from loss or lack of repeating business and this
is reflected in their stock price accordingly.
Trust is one of several relationship metrics that comprise an efficient Social Positioning
indicator, and one of several that accelerate favorable Environmental, Social and Governance
(ESG) scores. Although trust by itself is paramount, once a company has secured it for its brands,
it must meet other metrics to establish and be perceived as a highly socially positioned entity. This
includes continuous ESG commitments and to bring balance to the planet.
For this white paper, Orenda quantifies, in real time, companies’ social positioning from eight
everlasting relationship metrics by analyzing millions of online conversations. These metrics were
deemed to be critical for a healthy relationship to exist between 2 entities (Seajay, 2014), for
example, a corporation and its consumers or a corporation and its investors.

The following eight relationship metrics comprise Orenda’s Overall ESG Social Positioning
score and were equally weighted for this research paper.
1. Trust
2. Satisfaction
3. Corporate Social Responsibility
4. Commitment
5. Influence
6. Exchange of Benefits
7. Character
8. General
As these metrics increase for a corporation, so does its Overall ESG and Social Positioning.
The same is true if such scores decrease. Provided Orenda’s simple scoring methodology of 1 to
5, ranging from extremely negative to extremely positive, it is uncomplicated as well as plausible
to incorporate these datasets into multifactor models and ranking signals.
The Overall score was subsequently employed in a forthcoming financial model for stock
ranking and selection. We backtested its performance for the period of September 2017 through to
March 2020, with the objective of testing our hypothesis that highly socially positioned stocks are
superior investments and can lead to excess financial results.
We organized the remaining of this paper as follows. Section 2 further elaborates into Orenda’s
background, relationship metrics and Overall ESG and Social Positioning score. Section 3 defines
Orenda’s Social Positioning Risk Factor and contemplates traditional factor performance. Section
4 proposes the paper’s hypothesis, describes the employed data and model design for simulation.
Section 5 delivers empirical results of the backtested simulations. Section 6 presents the
conclusions. Section 7 discusses final remarks and next steps.
2. Orenda Social Positioning Scores

2.1 Orenda Background
Orenda Software Solutions was incorporated in 2015, after its founder, Tanya Seajay,
completed a 20-year communication experience and social science research journey that concluded
in 2014 at McMaster University, located in Hamilton, Ontario. The company’s first objective was

to leverage existing social science research by creating models that capture market perception of
companies and their respective brands.
Orenda launched its first Social Positioning product as a retail solution for marketing managers
and C-level executives. It subsequently transitioned into capital markets to close a gap that
sentiment and Environmental, Social and Governance (ESG) data providers were not fulfilling.
For example, ESG data supplied companies biased, self-reported metrics that might not conform
with public perception. Orenda reports ESG metrics from the opinion of such public. Sentiment
providers quantify fading emotions such as joy, anger or happiness, lacking the stability and
impactful effect on a company’s long-lasting reputation. In contrast, Orenda solves this problem
by quantifying relationship metrics that are needed to define a company’s true Social Positioning
among its peers.
2.2 Social Science
Orenda’s social science models are based on 8-relationship metrics that were deemed necessary
for a healthy relationship to establish and grow (Seajay, 2014). And although all eight metrics are
paramount to efficiently assess how stakeholders perceive companies’ social positioning, the
existence of trust is crucial, as no relationship can exist without it. This builds on a significant
body of research, including the six elements of a healthy relationship, as defined by Childers Hon
and Grunig (1999). Orenda’s relationship metrics include:
1. Trust: Measures the amount of integrity, dependability, and competence that the public
has in a company and its brand and considers whether people believe that your organization
has the ability to follow-through and deliver on its promises.
2. Satisfaction: Determines the amount of favorability the public expresses about a company
and its brand, and measures whether people’s expectations with a product or service are
positively reinforced by their experiences
3. Corporate Social Responsibility: Measures the level of confidence that people have in a
company, in relation to its actions to improve or uphold the social fairness and
environmental awareness of society.
4. Commitment: Measures the level of dedication that the public has to your company or
brand in order to receive their desired benefits, as opposed to filling these needs using other
products or services.

5. Influence: Analyzes the capacity that a company can affect the opinions of the public, and
measures how the public judges a company's leadership, transparency, and authority on
topics when as industry.
6. Exchange of Benefits: Measures the level of reciprocity the public expects from a
company for choosing its products or services rather than from competing companies.
7. Character: The distinct qualities that the public, according to industry standards, judges
the unique traits of a brand in comparison to its competition.
8. General: Measures any remaining important social factors that have not been covered in
any of the other categories.
Orenda weighs Trust along with the seven remaining relationship metrics to arrive at a
company’s Overall ESG and Social Positioning score. This is achieved by analyzing millions of
online conversations, for publicly traded and privately held companies, and processing as well as
quantifying such content through Orenda’s AI and proprietary dictionaries. Orenda’s technology
collects content and updates social positioning scores every 10 minutes, providing a total of 144-
daily updates. The 10-minute interval allows Orenda to collect enough content to deliver a
statistically meaningful score. The scoring methodology employs a range of 1 to 5, representing
an extremely low to extremely high Social Positioning score, respectively. Additionally, Orenda
collects the number of conversations each update is comprised of, representing the level of Social
Engagement. Figure 1 provides a sample Social Positioning dataset for Starbucks (NYSE: SBUX).
FIGI ID TICKER COMMITMENT SOCIAL RESPONSIBILITY INFLUENCE EXCHANGE OF BENEFITS GENERAL CHARACTER SATISFACTION TRUST OVERALL GENERATED TIME COUNTRY
BBG000CTQD87 SBUX 3.0541 3.3162 3.0408 3.0122 2.997 3.2861 3.2143 3.0726 3.1242 2015-03-31T10:30:00.000Z US
Figure 1

3. Orenda Social Positioning and Traditional Risk Factors
Orenda’s Social Positioning scores can naturally be incorporated into multifactor models. As
we will cover in subsequent sections, these scores are employed to select stocks that are, in the
public’s perception, aligned with their shared values. The opposite also stands true for poorly
aligned companies, since this will signal investors to divest, short or restrain from holding such
securities. This allows us to replicate a traditional risk factor exposure, but for this paper, we
restrict it solely to socially aligned companies.
Orenda’s Social Positioning risk factor comes at a time when traditional factor models are being
challenged by the market and academia. For example, Fama and French (1993), introduced an
extension to the Capital Asset Pricing model by contemplating the Small minus Big (size premium)
and High minus Low (value premium). Later, Fama and French (2015) built on their 3-factor
model by adding a Conservative minus Aggressive (investment premium) and a Robust minus
Weak (profitability premium). These models were globally accepted and employed but have come
under scrutiny for lack of efficient results, especially within the last decade. David Blitz found
that “over the most recent 2010-2019 decade, the return on each of the Fama-French factors fell
well short of its long-term average. The size and value factors even experienced a negative decade,
with the return of the value factor being particularly poor. It is not just the size and value factors
which have had a hard time, however. Over the past decade, the premium on the investment factor
also failed to materialize, with a return close to zero. Only the profitability factor generated a
positive return, but the magnitude of this premium is only about half its pre-2010 level” (David
Blitz, 2020).
Traditional factors can experience periods of long-lasting disappointing results. Asset managers
with discretionary power must decide between patience or exposing their portfolios to different
risk factors that would help them secure higher risk-adjusted results. In contrast, Orenda’s Social
Positioning risk factor aims at capturing a systematic risk premium for companies that are
perceived as socially aligned with stakeholders’ values, providing an opportunity for superior risk
adjusted returns.

4. Research
4.1 Hypothesis
This research paper contemplates the hypothesis that constructing a paper portfolio based on
Orenda’s Social Positioning risk factor provides superior risk adjusted results when compared to
a predefined benchmark (Invesco S&P 500 Equal Weight ETF (NYSE:RSP)) and it yields positive
alpha after controlling for traditional market, size, value, investment and profitability risk factors.
4.2 Data & Benchmark Selection
We have secured Social Positioning and Consumer Engagement data from Orenda Software
Solutions for current and historical constituents of the S&P 500. Pricing data for current and
historical constituents as well as for the benchmark was obtained from S&P Global. Daily and
monthly Fama and French factor data was downloaded from Fama-French library.1 The sample
period for Social Positioning and Consumer Engagement covers from June 30th, 2017 to April 30th,
2020. The benchmark selection was based on the universe of stocks Orenda quantifies for Social
Positioning and the equal weights employed in the investment simulation strategy. We found that
Invesco S&P 500 Equal Weight ETF (NYSE: RSP) provided all of the desirable features of a
suitable benchmark - it is unambiguous, measurable, investable, appropriate, and it was specified
prior to developing and testing the model.
4.3 Model Design
The model design aims at presenting an unconvoluted path to engage Orenda’s datasets. With
this objective in mind, we delineated a model that allows us to rebalance a paper portfolio
quarterly. In our opinion, this provides the minimum amount of time that a security needs to
materialize good or bad actions into stock price performance. Portfolio creation took place on
September 30th, 2017 (Q3, 2017). Thereafter, we rebalanced with only socially aligned companies
as dictated by the model detailed below.
First step consisted of averaging quarterly Social Positioning (1) and Social Engagement data
(2) for all current and historical constituents of the S&P 500.
1 https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

∑(𝑂𝑣𝑒𝑟𝑎𝑙𝑙 𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔) 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 t

𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 = # 𝑜𝑓 𝐷𝑎𝑦𝑠 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 t
(1)
∑(𝑂𝑣𝑒𝑟𝑎𝑙𝑙 𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡) 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 t

𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 = # 𝑜𝑓 𝐷𝑎𝑦𝑠 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 𝑡
(2)
We proceeded by computing the natural log of quarterly change for each variable (3 & 4).
𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡

ln ( ) (3)
𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑡−1
𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡

ln ( ) (4)
𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑡−1
We repeated the process for quarterly price change by security (5).
𝑃𝑟𝑖𝑐𝑒 𝑡
ln ( ) (5)
𝑃𝑟𝑖𝑐𝑒𝑡−1
The final step consisted of computing the excess change of social positioning over the change
in price per unit of change in Social Engagement to arrive at Orenda's Social Positioning (6).
𝑂𝑟𝑒𝑛𝑑𝑎 𝐹𝑎𝑐𝑡𝑜𝑟
𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑡 𝑃𝑟𝑖𝑐𝑒
𝑙𝑛 ( ) − 𝑙𝑛 (𝑃𝑟𝑖𝑐𝑒 𝑡 ) (6)
𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 t−1 t−1
=
𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡
𝑙𝑛 ( )
𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 t−1
4.4 Selection Methodology

The stock selection methodology consisted in holding securities in the top 10, five, and one
percent of Orenda’s Factor distribution and to compare the results against the benchmark as well
as the bottom 10, five, and one percent. We did not conduct a portfolio optimization strategy and
modelled an equally weighted portfolio approaching the allocation of the predefined benchmark,
Invesco S&P 500 Equal Weight ETF (NYSE: RSP). The universe of stocks the model selected
from conform to the current and historical constituents of the S&P 500.
© Orenda Software Solutions Inc. 10 | P a g e

5. Empirical Evidence and Results
This section presents the empirical evidence, performance and risk adjusted results for Orenda’s
top 10% selection (portfolio A) and contrasts such results against the benchmark and bottom 10%.
We repeat this exercise for the top five percent (portfolio B) and one percent (portfolio C) in
sections 5.2 and 5.3, correspondingly. Results reflect those of paper portfolios assuming no
arbitrage opportunities or transaction costs.
To validate the existence of positive alpha in the proposed model, after controlling for
traditional risk factors, we employ time-series regressions on Orenda’s daily excess returns against
Fama-French 3 and 5 factor model (7 & 8). Fama and French 3 factor model includes risk, size
and value premiums while the 5 factor adds profitability and investment premiums.
𝑅𝑂𝑟𝑒𝑛𝑑𝑎,𝑡 − 𝑅𝑓,𝑡
̂ (𝑅𝑚𝑡 − 𝑅𝑓𝑡 ) + 𝛽̂
= 𝛼𝑓𝑓3 + 𝛽_𝑚𝑘𝑡 ̂
𝑠𝑚𝑏 (𝑆𝑀𝐵𝑡 ) + 𝛽ℎ𝑚𝑙 (𝐻𝑀𝐿𝑡 ) (7)
+ 𝑒𝑖𝑡
𝑅𝑂𝑟𝑒𝑛𝑑𝑎,𝑡 − 𝑅𝑓,𝑡
̂
= 𝛼𝑓𝑓3 + 𝛽 ̂ ̂
𝑚𝑘𝑡 (𝑅𝑚𝑡 − 𝑅𝑓𝑡 ) + 𝛽𝑠𝑚𝑏 (𝑆𝑀𝐵𝑡 ) + 𝛽ℎ𝑚𝑙 (𝐻𝑀𝐿𝑡 ) (8)
+ 𝛽̂ ̂
𝑟𝑚𝑤 (𝑅𝑀𝑊𝑡 ) + 𝛽𝑐𝑚𝑎 (𝐶𝑀𝐴𝑡 ) 𝑒𝑖𝑡

5.1 Portfolio A
Table 1 captures the coefficient estimates of Fama and French 3 factor model against Portfolio
A daily excess return for the period of September 30th, 2017 to April 30th, 2020. As observed,
Jensen’s (daily) Alpha is positive and significant at a 95% confidence level (P-value of 0.0366).
The presented empirical results also show that Portfolio A returns are principally systematic as
measured by the statistically meaningful market beta of 1.081154 and, to a lesser extent, to value
and size factors with relatively low coefficients of 0.255629 and 0.296940.
Portfolio A TOP 10%

Panel A
Residuals:
Min 1Q Median 3Q Max
-0.0209862 -0.0019492 -0.0000534 0.0017509 0.0298935
Panel B
Coefficients:
Estimate Std. Error t value Pr ( >|t|)
(Intercept) 0.000331 0.000158 2.094 0.0366 *
Mkt_RF 1.081154 0.010874 99.423 <2e-16 ***
SMB 0.255629 0.025274 10.115 <2e-16 ***
HML 0.29694 0.021942 13.533 <2e-16 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Multiple R-squared: 0.9466
Adjusted R-squared: 0.9463
F-statistic: 3808 on 3 and 645 DF, p-value: < 2.2e-16
Residual standard error: 0.004002 on 645 degrees of freedom
Table 1
We repeated this empirical test against Fama and French 5 factor model. As shown in Table 2,
Jensen’s (daily) Alpha is positive and significant at a 95% confidence level (P-value of 0.04197).
Although we have obtained similar results for market, size and value premiums, the investment
factor (CMA) coefficient is not statistically related to Portfolio A indicating that returns for
socially aligned stocks do not behave as those of companies with aggressive investment strategies.
A possible explanation is that socially aligned companies are more selective at times of deploying

capital for investments since stakeholders’ wellbeing, as opposed to solely shareholders’ wealth
maximization, is a factor in the investment decision-making process. Profitability factor (RMW)
is statistically meaningful with a lower coefficient representing the least relation to Portfolio’s A
returns among the four meaningful factors.
Portfolio A TOP 10%

Panel A
Residuals:
-0.0207551 -0.0019992 -0.0001058 0.0018963 0.0299353
Panel B
Coefficients:
(Intercept) 0.0003205 0.0001573 2.038 0.04197 *
Mkt_RF 1.0744903 0.011561 92.941 <2e-16 ***
SMB 0.253624 0.02645853 9.586 <2e-16 ***
HML 0.3063083 0.02608 11.745 <2e-16 ***
RMW 0.111004 0.0427643 2.596 0.00965 **
CMA -0.0835608 0.0541149 -1.544 0.12305
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 2

Similarly, we tested for empirical alpha for Orenda’s bottom 10%. As expected, and captured
in Table 3, companies that ranked low in Social Positioning scores produced non-statistically
meaningful Jensen’s (daily) alpha at 90% confidence level. Returns for these stocks were largely
systematic with all 3 Fama and French factors meaningful at 99% confidence. However, in line
with Orenda’s hypothesis, companies that are not perceived as virtuous are challenged to achieve
above market returns.
BOTTOM 10%
Panel A
Residuals:
-0.0216902 -0.0020772 -0.0000227 0.0020654 0.0178242
Panel B
Coefficients:
(Intercept) 0.000202 0.0001585 1.275 0.203
Mkt_RF 1.0568036 0.0109043 96.917 <2e-16 ***
SMB 0.3016119 0.025343 11.901 <2e-16 ***
HML 0.3389382 0.022002 15.405 <2e-16 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 3

Table 4 presents the results against Fama and French 5 factor model for the bottom 10%. Similar
to our preceding empirical test, Jensen’s (daily) alpha is not meaningful at a 90% confidence level.
We now find that all 5 factors, including CMA, are statistically related to the returns of the bottom
10%.
BOTTOM 10%
Panel A
Residuals:
-0.0193945 -0.0020747 0.0000597 0.0021833 0.0175736
Panel B
Coefficients:
(Intercept) 0.0001908 0.000155 1.231 0.21865
Mkt_RF 1.0375712 0.0113905 91.091 <2e-16 ***
SMB 0.2794983 0.0260678 10.722 <2e-16 ***
HML 0.3867642 0.0256954 15.052 <2e-16 ***
RMW 0.1466556 0.0421336 3.481 0.000534 ***
CMA -0.2492974 0.0533169 -4.676 3.57e-06 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 4

Additionally, we have computed the annualized Sharpe ratio for the predefined benchmark,
Orenda’s top and bottom 10%, for the period of September 30th, 2020 to April 30th, 2020. This
permits performance comparison on risk adjusted basis. Table 5 captures Orenda’s top selection,
outperforming the bottom 10% and benchmark. Furthermore, and as validation of the previous
presented results, on risk adjusted basis, the bottom 10% underperformed the benchmark.
TOP 10% BOTTOM 10% BENCHMARK
0.357 0.125 0.166

Table 5
To visualize performance on absolute terms, we computed the growth of a hypothetical $1

portfolio invested in the benchmark as well as top and bottom 10%. Figure 2 presents the general
outperformance of highly socially aligned companies over the benchmark and poorly socially
aligned companies. It is interesting to observe how the paper portfolio holding the top 10%
recovers more quickly and more robustly from crises or corrections when compared to the
benchmark or bottom 10%. This signals that companies that enjoy desirable Social Positioning
levels can recover faster than the mean and low, socially aligned companies. A $1 investment in
portfolio A would have grown to $1.20 vs $1.06 for the benchmark.
Figure 2

5.2 Portfolio B
As introduced earlier, Portfolio B comprises the top five percent of socially aligned companies.
Table 6 presents the coefficient estimates of Fama and French 3 factor model against Portfolio B
daily excess returns. Here, we observe that Jensen’s (daily) Alpha is statistically meaningful at
90% confidence (P-value of 0.0857) and larger than Portfolio A (0.0003514 vs. 0.000331). Similar
to Portfolio A, Portfolio B’s returns are primarily systematic as measured by the statistically
meaningful market beta of 1.108 and, to a lesser extent, to value and size factors with relatively
low coefficients of 0.2568 and 0.3021.
Portfolio B TOP 5%
Panel A
Residuals:
-0.029416 -0.002567 -0.000149 0.002384 0.042011
Panel B
Coefficients:
(Intercept) 0.0003514 0.0002042 1.721 0.0857
Mkt_RF 1.1083167 0.0140487 78.891 < 2e-16 ***
SMB 0.2568782 0.0326512 7.867 1.53e-14 ***
HML 0.3021009 0.0283468 10.657 < 2e-16 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 6

Table 7 captures the regression of Portfolio B’s excess returns against Fama and French 5 factor
model. Once again, Jensen’s (daily) Alpha is statistically meaningful at 90% confidence and is
larger than Portfolio A (0.0003442 vs. 0.0003205). A key difference in these newly presented
results is that the profitability factors (RMW) factor is not statistically meaningful at 90%
confidence level and excess returns are primarily related to market, value size, respectively.
Echoing prior results, the investment factor is not statistically meaningful.
Portfolio B TOP 5%
Panel A
Residuals:
-0.029758 -0.002563 -0.00011 0.002398 0.042006
Panel B
Coefficients:
(Intercept) 0.0003442 0.0002042 1.685 0.0925
Mkt_RF 1.104786 0.0150114 73.596 < 2e-16 ***
SMB 0.2573067 0.0343544 7.49 2.29e-13 ***
HML 0.3051667 0.0338637 9.012 < 2e-16 ***
RMW 0.0746771 0.0555274 1.345 0.1791
CMA -0.0435311 0.0702656 -0.62 0.5358
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 7

We repeat the empirical test for the bottom five percent. Table 8 presents the coefficients
estimates against Fama and French 3 factor model. Mirroring the 10% bottom, (daily) alpha is not
statistically meaningful at 90% confidence level (P-value of 0.577) and all 3 Fama and French
factors are statistically related indicating that returns are largely systematic.
BOTTOM 5%
Panel A
Residuals:
-0.049265 -0.002738 -0.000157 0.0029 0.02506
Panel B
Coefficients:
(Intercept) 0.0001189 0.0002133 0.557 0.577
Mkt_RF 1.0825696 0.0146777 73.756 <2e-16 ***
SMB 0.3925586 0.0341132 11.508 <2e-16 ***
HML 0.3357664 0.029616 11.337 <2e-16 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 8

Table 9 presents the coefficients against the Fama and French 5 factor for the bottom five
percent. Supporting our preceding results that (daily) alpha is not statistically meaningful at 90%
confidence level. All 5 factors are significant at 99% with the exception of RMW, which is
statistically meaningful at 90% confidence.
BOTTOM 5%
Panel A
Residuals:
-0.046282 -0.002824 -0.000029 0.002865 0.02466
Panel B
Coefficients:
(Intercept) 0.0001143 0.000209 0.547 0.5847
Mkt_RF 1.0540978 0.0153606 68.623 < 2e-16 ***
SMB 0.3495086 0.0351537 9.942 < 2e-16 ***
HML 0.4196982 0.0346515 12.112 < 2e-16 ***
RMW 0.1073807 0.0568191 1.89 0.0592
CMA -0.3741967 0.0719003 -5.204 2.62e-07 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 9

The Sharpe ratio for the predefined benchmark, Orenda’s top and bottom 5%, is presented on
Table 10. Orenda’s top five percent displays a Sharpe ratio of 0.365, while the benchmark and
bottom five percent achieved a 0.166 and 0.086, respectively. These findings support our previous
results that top socially aligned companies outperform the benchmark and low socially aligned
companies on risk-adjusted basis.
0.365 0.086 0.166

Table 10
Figure 3 presents the growth of $1 invested in the top five percent socially aligned stocks over
the benchmark and bottom five percent socially aligned companies. Performance on absolute,
relative and risk adjusted basis was superior to bottom five percent, benchmark as well as Portfolio
A, with a similar recovery pattern from crisis and corrections. A $1 investment in portfolio B
would have grown to $1.21 vs $1.06 for the benchmark.
$1 DOLLAR PERFORMANCE (TOP 5% - BOTTOM 5% - BENCHMARK)

$1.55
$1.45
$1.35
$1.25
$1.15
$1.05
$0.95
$0.85
$0.75
$0.65
RSP TOP 5% BOTTOM 5%
Figure 3

5.3 Portfolio C
The final sets of empirical tests were conducted on Portfolio C comprised of the top one percent
of Orenda’s socially aligned companies. Table 11 captures the coefficient estimates of Fama and
French 3 factor model against Portfolio C daily excess returns. Once again, we observe that
Jensen’s (daily) Alpha is positive, robust and statistically meaningful at 99% confidence (P-value
of 0.00689). Additionally, alpha is materially larger than Portfolio A and B (0.0010675 vs.
0.000331 and 0.0003514). Similar to Portfolio A and B, Portfolio C’s returns are primarily
systematic as measured by the statistically meaningful market beta of 1.1859 and, to a lesser extent,
to value and size factors with relatively low coefficients of 0.5258421 and 0.2484471.
Portfolio C TOP 1%
Panel A
Residuals:
-0.062642 -0.004962 -0.000794 0.004399 0.072631
Panel B
Coefficients:
(Intercept) 0.0010675 0.0003938 2.711 0.00689 **
Mkt_RF 1.1859556 0.0270991 43.764 < 2e-16 ***
SMB 0.5258421 0.0629821 8.349 4.2e-16 ***
HML 0.2484471 0.0546791 4.544 6.6e-06 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 11

Table 12 presents the regression coefficients of Portfolio C’s excess returns against Fama and
French 5 factors. Once again, Jensen’s (daily) alpha is statistically meaningful at 99% confidence
and is materially larger than Portfolio A and B (0.0010533 vs. 0.0003205 and 0.0003442,
respectively). The investment (CMA) and profitability (RMW) factors are statistically meaningful
at 95 and 90%, respectively, while all remaining factors are related to Portfolio C's excess returns
at 99% confidence level.
Portfolio C TOP 1%
Panel A
Residuals:
-0.059788 -0.005146 -0.000951 0.004402 0.072209
Panel B
Coefficients:
(Intercept) 0.0010533 0.0003922 2.685 0.00743 **
Mkt_RF 1.1621569 0.02883 40.311 < 2e-16 ***
SMB 0.4988125 0.065979 7.56 1.39e-13 ***
HML 0.3072026 0.0650365 4.724 2.85e-06 ***
RMW 0.1850317 0.1066424 1.735 0.08321
CMA -0.3083204 0.1349479 -2.285 0.02265 *
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
F-statistic: 462.2 on 5 and 643 DF, p-value: < 2.2e-16
Table 12

As captured on Table 13 and approaching the results for the bottom 10 and five percent, the one
percent bottom selection displays insignificant Jensen’s alpha (P-value of 0.174) at 90%
confidence level after controlling for Fama and French 3 factors. Once again, the bottom one
percent returns are systematic, provided the statistical meaningful coefficients for market, size and
value.
BOTTOM 1%
Panel A
Residuals:
-0.051554 -0.004549 -0.000155 0.004366 0.041058
Panel B
Coefficients:
(Intercept) 0.0005045 0.0003703 1.362 0.174
Mkt_RF 1.0606428 0.0254843 41.619 < 2e-16 ***
SMB 0.2804869 0.0592291 4.736 2.69e-06 ***
HML 0.3432156 0.0514209 6.675 5.35e-11 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 13

Table 14 replicates the empirical test against Fama and French 5 factor model. With a P-value
of 0.21, daily Jensen’s alpha is insignificant at 90% confidence level. All factors are related to the
bottom one percent except for CMA displaying a P-Value of 0.708.
BOTTOM 1%
Panel A
Residuals:
-0.049447 -0.004528 0.000111 0.004532 0.04142
Panel B
Coefficients:
(Intercept) 0.0004584 0.0003657 1.254 0.21
Mkt_RF 1.0555674 0.0268768 39.274 < 2e-16 ***
SMB 0.3119351 0.061509 5.071 5.17e-07 ***
HML 0.3083691 0.0606303 5.086 4.80e-07 ***
RMW 0.4354029 0.0994175 4.38 1.39e-05 ***
CMA -0.0472245 0.1258053 -0.375 0.708
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Panel C
Table 14

The Sharpe ratio for the predefined benchmark, Orenda’s top and bottom one percent are
presented on Table 15. Orenda top one percent displays a Sharpe ratio of 0.818 while the
benchmark and bottom one percent achieved 0.166 and 0.426, respectively. These findings
support our previous results that top socially aligned companies outperform the benchmark and
low socially aligned companies on risk adjusted basis.
0.818 0.426 0.166

Table 15
Figure 4 presents the hypothetical growth of $1 invested in the top one percent socially aligned
stocks over the benchmark and bottom one percent socially aligned companies. Performance on
absolute terms was materially superior to bottom one percent, benchmark as well as Portfolio A
and B with a similar recovery pattern from crisis and corrections. A $1 investment in portfolio C
would have grown to $1.99 vs $1.06 for the benchmark.
Figure 4

5.4 Additional Supporting Results

Figure 5 presents a correlation matrix among Fama and French five factors and Portfolio C
excess return. As expected, Orenda’s socially aligned portfolio has a relative low correlation to
traditional factors, below 0.30, and a negative correlation coefficient of -0.224 with CMA. This,
coupled with prior findings, suggest the benefit of adding Orenda to multifactor models as an
efficient way to achieve diversification and higher expected risk adjusted returns. Similar results
were found for Portfolio A and B.
Figure 5
Continuing with Portfolio C, we ran a Bayesian regression model on 10,000 iterations with the
objective of analyzing density functions for each of the regression coefficients. Figure 6 indicates

Jensen’s alpha (Intercept coefficient) to be positive and above zero and only negative in extreme
events located at left-end-tail of the distribution.

Figure 6
6. Conclusion
We live in a digitally empowered world and shared values spread globally and continuously by
younger generations that seek to interact with companies they trust and are socially aligned. These
companies benefit by capturing recurring business and capital from a population of investors that
seek maximization of stakeholder’s wellbeing instead of stockholder’s wealth.
Companies that follow good social behavior are considered virtuous investments and having
the ability to segregate those that enjoy a robust Social Positioning from companies that are
considered socially unethical can lead to profitable investment strategies.
To test this hypothesis, we have employed Orenda’s datasets as an efficient tool to select
securities that are highly aligned with stakeholders’ values. We conducted this exercise by creating
3 paper portfolios, from the S&P 500 universe of stocks consisting of the top 10, five and one
percent of the Social Positioning distribution. The returns were subsequently compared, to a
predefined benchmark and bottom 10, five, and one percent distribution of the least socially

aligned securities, on risk adjusted basis and after controlling for Fama and French 3 and 5 factor
models.
Our primary empirical findings demonstrate that the portfolios of stocks selected based on high
Social Positioning do provide alpha after controlling for Fama and French 3 and 5 factor models,
which include market, size, value, profitability and investment premiums. We also found
empirical evidence that companies that are poorly socially aligned do not provide above market
returns after controlling for the traditional factors listed previously.
All three socially aligned portfolios outperformed the benchmark and the bottom distribution
on risk adjusted and absolute basis, as captured by the Sharpe ratio and hypothetical $1 growth.
Further validating our hypothesis, the bottom 10 and five percent also underperformed the
benchmark indicating that companies that are situated in the bottom distribution of Social
Positioning scores are challenged to generate above market returns. We can also conclude that as
portfolio selection narrowed from top 10 to one percent, we have seen improvements in empirical
alpha, Sharpe ratio and $1 performance, suggesting that within a basket of highly socially aligned
companies, as dictated by Orenda Social Positioning factor, the top still outperforms the bottom.
7. Final Remarks
Orenda’s Social Positioning dataset is unique and powerful. The presented model to arrive at
the Social Positioning risk factor demonstrates one simple way, to employ Orenda’s dataset for
security selection. There are many more methods available. It can be included and modelled in
more complex scenarios allowing for end users with different capabilities to adjust the input as
they see fit.
As next steps, Orenda will release a series of papers employing our Social Positioning and ESG
scores in diverse models for data validation. We welcome the audience to provide feedback on this
paper and to get in contact with us with general comments or ideas for future publications. From
our team, we thank you!

References
Childers Hon, L., & Grunig. J. (1999). Guidelines for measuring relationships in public relations.
Institute for Public Relations. Retrieved from www.instituteforpr.org.
Edelman Trust Barometer, Special Report: Brand Trust and the Coronavirus Pandemic., 2020
Fama, Eugene F., and Kenneth R. French. 1993. “Common Risk Factors in the Returns on Stocks
and Bonds.” Journal of Financial Economics, vol. 33, no. 1 (February): 3–56.
Fama, Eugene F., and Kenneth R. French. 2015. “A Five-Factor Asset Pricing Model.” Journal of
Financial Economics, vol. 116, no. 1 (April): 1–22.
Myles Gooding, Lead National Retail Consultant at PwC Canada.,

https://www.macleans.ca/economy/what-will-shopping-in-canada-look-like-in-the-aftermath-of-
coronavirus/
Sandra Landrum, “Millennials Driving Brands To Practice Socially Responsible Marketing.”,

March 17, 2017
https://www.forbes.com/sites/sarahlandrum/2017/03/17/millennials-driving-brands-to-practice-
socially-responsible-marketing/#1f6cdf074990
Tanya Seajay, Connecting Right From the Start: A study on the value of engaging stakeholders on
large-scale projects. Major Research Paper Submitted by: Tanya Collier MacDonald, MCM (c)
Department of Communication Studies and Multimedia Faculty of Humanities, McMaster
University May 2, 2014

3 - Esg - ML - Social Positioning Within The SP 500

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

3 - Esg - ML - Social Positioning Within The SP 500

Uploaded by

Copyright:

Available Formats

Overall ESG &

Orenda Software Solutions Inc.

© 2020 Orenda Software Solutions Inc. All rights reserved.

© Orenda Software Solutions Inc. 2|Page

Ezequiel Machabanski* and Jose Jatar ** and Tanya Seajay ***

June 30, 2020

© Orenda Software Solutions Inc. 3|Page

© Orenda Software Solutions Inc. 4|Page

2. Orenda Social Positioning Scores

© Orenda Software Solutions Inc. 5|Page

© Orenda Software Solutions Inc. 6|Page

© Orenda Software Solutions Inc. 7|Page

3. Orenda Social Positioning and Traditional Risk Factors

© Orenda Software Solutions Inc. 8|Page

© Orenda Software Solutions Inc. 9|Page

∑(𝑂𝑣𝑒𝑟𝑎𝑙𝑙 𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔) 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 t

∑(𝑂𝑣𝑒𝑟𝑎𝑙𝑙 𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡) 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 t

𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡

𝑆𝑜𝑐𝑖𝑎𝑙 𝐸𝑛𝑔𝑎𝑔𝑒𝑚𝑒𝑛𝑡 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝐴𝑣𝑒𝑟𝑎𝑔𝑒𝑡

We repeated the process for quarterly price change by security (5).

4.4 Selection Methodology

© Orenda Software Solutions Inc. 10 | P a g e

5. Empirical Evidence and Results

© Orenda Software Solutions Inc. 11 | P a g e

Portfolio A TOP 10%

© Orenda Software Solutions Inc. 12 | P a g e

Portfolio A TOP 10%

© Orenda Software Solutions Inc. 13 | P a g e

© Orenda Software Solutions Inc. 14 | P a g e

© Orenda Software Solutions Inc. 15 | P a g e

TOP 10% BOTTOM 10% BENCHMARK

0.357 0.125 0.166

To visualize performance on absolute terms, we computed the growth of a hypothetical $1

© Orenda Software Solutions Inc. 16 | P a g e

© Orenda Software Solutions Inc. 17 | P a g e

© Orenda Software Solutions Inc. 18 | P a g e

© Orenda Software Solutions Inc. 19 | P a g e

© Orenda Software Solutions Inc. 20 | P a g e

TOP 5% BOTTOM 5% BENCHMARK

0.365 0.086 0.166

$1 DOLLAR PERFORMANCE (TOP 5% - BOTTOM 5% - BENCHMARK)

RSP TOP 5% BOTTOM 5%

© Orenda Software Solutions Inc. 21 | P a g e

© Orenda Software Solutions Inc. 22 | P a g e

© Orenda Software Solutions Inc. 23 | P a g e

© Orenda Software Solutions Inc. 24 | P a g e

© Orenda Software Solutions Inc. 25 | P a g e

TOP 1% BOTTOM 1% BENCHMARK

0.818 0.426 0.166

© Orenda Software Solutions Inc. 26 | P a g e

5.4 Additional Supporting Results

© Orenda Software Solutions Inc. 27 | P a g e

© Orenda Software Solutions Inc. 28 | P a g e

© Orenda Software Solutions Inc. 29 | P a g e

© Orenda Software Solutions Inc. 30 | P a g e

Myles Gooding, Lead National Retail Consultant at PwC Canada.,

Sandra Landrum, “Millennials Driving Brands To Practice Socially Responsible Marketing.”,

© Orenda Software Solutions Inc. 31 | P a g e

You might also like

Ezequiel Machabanski* and Jose Jatar and Tanya Seajay *