October 2019

Man GroupESG: Bringing order to uncertainty

For investment professionals only. Not for public distribution.

© M

an 2

019

The current state of responsible investment

Source: Pensions & Investments, Sept. 18 2019. 1

Conflated approaches and unproven performance

© M

an 2

019

The ESG Data Value Chain

Source: Andreas G.F. Hoepner, Financial Data Science for Responsible Investors, 2018.Schematic illustration. Any descriptions or information involving investment process or strategies is provided for illustration purposes only, may not be fully indicative of any present or future investments, may be changed at the discretion of the investment manager and are not intended to reflect performance. 2

There are more than 1,200 non-traditional datasets (Eagle Alpha, 2018) The past decade has seen significant improvements in the collection of information and its contextualization We believe the key to a meaningful signal is how the information is organized and evaluated We expect machine learning and natural language processing to add multiple-speed signalling power

Data points(unstructured)

Knowledge

Signal

Information(structured)

Investment signal capable of enhancing returns or mitigating risk

Information that has been fact-checked

Data points in a structural context

Numbers, words, pictures….anything that can be discriminated

SASB, GRI, carbon accounting,natural capital accounting

Third-party ESG data providers

Unexplored

Unrefined

© M

an 2

019

The question of RI-driven performance

3

Evidence still remains tenuous

Is the “ESG doesn’t hurt performance but it doesn’t add performance” message enough? A growing body of work suggests ESG overlays can generate meaningful alpha But is it really ESG? What’s the evidence for a ESG premia?

Academic survey and meta-analyses supporting linkage between ESG and financial performance1

Practitioner work - relative performance per MSCI Europe ESG rating2

47.9%

6.9%

0.146

62.6%

8.0%

0.150

0%

10%

20%

30%

40%

50%

60%

70%

Share of PositiveFindings

Share of NegativeFindings

Weighted CorrelationLevel r in Studies

Vote-count studies

Meta-analyses

which 47.9% in vote-count studies and 62.6% in meta-analyses yield positive findings.Source (1): Gunnar Friede, Timo Busch & Alexander Bussen. Journal of Sustainable Finance and Investment, 5:4, 210-233. “ESG and Financial Performance: Aggregated Evidence from more than 2000 Empirical Studies”. Dec. 15, 2015. 90% of studies find a nonnegative relation between ESG and corporate financial performance, of Source (2): MSCI ESG Research, FactSet and Nordea Markets.

© M

an 2

019

The necessary exercise of defining ESG

4

The data problem

Scepticism is warranted around the quality of the data Cross-vendor correlations remain low, but why should ESG ratings converge anyways? Where are the returns from existing ESG ratings really coming from?

What factors are really driving existing ESG signals?1 What factors are really driving existing ESG signals?2

-0.5

-0.3

-0.1

0.1

0.3

0.5

0.7

0.9ESG Factor 1ESG Factor 2

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Spec

Ris

k

Liqu

idity

Size

Valu

e

Mom

entu

m

Yiel

d

Beta

Barra

Fac

tor E

xpos

ure

(Std

Dev

uni

ts)

Sust. ESGMSCI ESG

Source (1): Man Numeric, MSCI and Sustainalytics, 2019.Source (2): JP Morgan research, “ESG – Environmental, Social and Governance Investing,” 2018.

© M

an 2

019

The necessary exercise of defining ESG

5

Process, premia or both?

Lack of definition means ESG as perceived more as a discretionary process Process is subjective. How can we measure performance returns through attribution analysis? A true ESG factor remains widely unestablished due to greater data complexity and dimensionality

ESG as process1 ESG as style factor?2

Value

Growth

Size

Risk

Quality

- Price to Book- Sales/Price- Earnings/Price- EBITDA/Price- FCF Yield

- 5Y Sales- 5Y Earnings- 5Y Assets CAGR- 1Y Fwd Sales Growth- 1Y Fwd EPS Growth

- 1Y Volatility- 2Y Beta- 5Y CFFO Variability- Sales Dispersion- EPS Dispersion

- Total Assets- Sales LTM- Market Cap- EBITDA LTM- # EPS Estimates

- 3Y Avg ROE- Asset Turnover LTM- 3Y Avg EBITDA Margin- Net Debt/EV- Capex to Sales

- 1Y Total Return- 3M Stories Growth- Qtrly Sales Acceleration- 6M Target Price Change- 3M Sales Revision

ESG

Momentum

- ????- ????

Source (1): Man GLG, 2015.Source (2): Man Group Plc, 2019.

© M

an 2

019

MSCI v. Sustainalytics Correlation (within Numeric Global Universe)

Comparing ESG datasets

As of March 2019 Source: MSCI, Sustainalytics Man Numeric. MSCI ESG: MSCI World ESG Universal index. MSCI Env, MSCI Social and MSCI Gov are three sub-components of the MSCI World ESG Universal index. 6

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Aug-

09

Jan-

10

Jun-

10

Nov

-10

Apr-1

1

Sep-

11

Feb-

12

Jul-1

2

Dec

-12

May

-13

Oct

-13

Mar

-14

Aug-

14

Jan-

15

Jun-

15

Nov

-15

Apr-1

6

Sep-

16

Feb-

17

Jul-1

7

Dec

-17

May

-18

Oct

-18

Mar

-19

Cor

rela

tion

ESG Composite Environmental Governance Social

Skepticism is high, cross-vendor correlations remain low….that is a good thing

© M

an 2

019

ESG dataset comparison

Any descriptions or information involving investment process or strategies is provided for illustration purposes only, may not be fully indicative of any present or future investments, may be changed and are not intended to reflect performance. 7

Evaluated / trialed numerous datasets focusing on eight key criteria

Transparent methodology

Vendor 1 2 2 2 2 3 2 2 3 18Vendor 2 2 1 3 3 3 2 1 3 18Vendor 3 3 2 1 2 2 2 3 3 18Vendor 4 2 2 2 2 3 2 2 2 17Vendor 5 3 2 2 3 1 3 1 1 16Vendor 6 2 1 1 1 2 2 3 2 14Vendor 7 1 3 2 2 2 1 1 1 13Vendor 8 1 3 1 1 1 2 2 1 12

Data Frequency

Data Transparency

Market Adoption

Final ScoreData vendor

Data Quality

Data Delivery

Factor Granular

Factor Uniqueness

Data Coverage

Data quality – usable state, time-series consistency (backfilling, look ahead)Data delivery – ease of access

Granularity and breadth of factorsFactor uniqueness – items not found in other sources

Company, sector, factor coverage over time

Data vendor reputation and awareness

Frequency of data change

© M

an 2

019

8

(2) Sector, region, factor biases - environmental scores by region

(1) Data granularity – tail biased vs continuous distribution

(1) Data granularity: 12% of companies are legitimate zeros; 37% correspond to missing data listed as zero

(2) Many ESG data items have biases - ie, sector, region, factor which must be adjusted to make fair and accurate comparisons.

(3) ESG data providers often score sectors quite differently (ie, tobacco ranks poorly while controversial weapons is viewed favourably).

Overview

Challenges interpreting raw data, distributions and scoringInvestors need to understand the nuances behind the ESG datasets they are adopting

(3) Different approaches to ESG scoring among sectors

0%

10%

20%

30%

40%

50%

60%

70%

80%

0 1 10 15 20 24 25 30 33 40 49 50 60 66 70 74 75 80 90 95 99 100

% o

f Com

pani

es

ESG Provider Raw Score

Business Ethics Controversies or Incidents

Renewable Energy Programmes

Discrimination Policy

404550556065707580

Mar

-12

Sep-

12

Mar

-13

Sep-

13

Mar

-14

Sep-

14

Mar

-15

Sep-

15

Mar

-16

Sep-

16

Mar

-17

Sep-

17

Mar

-18

Sep-

18

Mar

-19

ESG

Pro

vide

r Sco

re

S&P 500 R3000 MSCI Europe MSCI Japan MSCI EM

* Average Sustainalytics by sector January 2013 – March 2019Any information involving investment process or strategies is provided for illustration purposes only, may not be fully indicative of any present or future investments, may be changed and are not intended to reflect performance.

0102030405060708090

100

0 20 40 60 80 100

SUST

AIN

ALYT

ICS

MSCI

Controversial Weapons generally viewed favorably

Tobacco names rank poorly by MSCI

12% of companies = legitimate zeros; 37% correspond to missing data listed as zero

© M

an 2

019

Measuring and managing ESG risk

Any descriptions or information involving investment process or strategies is provided for illustration purposes only, may not be fully indicative of any present or future investments, may be changed and are not intended to reflect performance. SIMULATED HYPOTHETICAL PERFORMANCE. The simulated performance data above reflecting hypothetical results is shown for the time period from December 31, 2012 to March 31, 2019. Simulated performance data and hypothetical results are shown for illustrative purposes only, do not reflect actual trading results, have inherent limitations and should not be relied upon. Please see the back of this presentation for additional information on simulated performance. 9

Leveraging quantatitive ESG research to better understand risk

Numeric ESG Model maps vendor data to 15 principles-based pillars Industry-focused framework, adjusted for country, sector and factor biases Simulations show better risk-adjusted performance than off the shelf data from vendors Proprietary ESG signal has historical predictive power in global, US and Europe

Standalone ESG Signal (Global Universe) Standalone ESG Signal (Region)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

Sustainalytics MSCI Numeric--RiskAdj

IR

Ann.

Ret

urn

Ann Ret Ann Vol IR (RHS)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

Global US Europe

IR

Ann.

Ret

urn

Ann Ret Ann Vol IR (RHS)

© M

an 2

019

ESG is complementary

Source: Man Numeric. Daily stock level correlation of ESG signals with other Numeric composite signals. Global Universe December 31, 2012 – December 31, 2018.This slide contains hypothetical or simulated model results that have certain inherent limitations. Unlike an actual portfolio record, simulated results do not represent actual trading. Also, since the trades have not actually been executed, the published results may have under-or-over compensated for the impact, if any, of certain market factors, such as lack of liquidity. Simulated trading programs in general are also subject to the fact that they are designed with the benefit of hindsight. There exist limitations inherent with model results. Results include simulated transaction costs, but do not include the impact of actual trading. 10

ESG signal has low correlation with other factor models

ESG models have low correlation with each other Near zero correlation between ESG and traditional models‒ Governance and Quality 0.01 correlation‒ Provides a unique source of information

Score Correlation of ESG Signals with other Models

Global Universe

ESG Composite Environment Social Governance

Global Combo Quality Value Momentum

Informed Investor

ESG Composite1.00

Environment 0.68 1.00

Social 0.70 0.32 1.00

Governance 0.75 0.26 0.28 1.00

Global Combo 0.04 0.06 0.02 0.02 1.00

Quality 0.03 0.07 0.04 0.01 0.44 1.00

Value 0.03 0.06 0.01 0.00 0.64 0.14 1.00

Momentum 0.00 0.00 0.00 0.00 0.48 0.06 -0.09 1.00

Informed Investor 0.01 0.01 -0.01 0.02 0.38 0.10 0.18 0.08 1.00

© M

an 2

019

Multi-factor ESG framework

Integration methods

11

Combining ESG factors within a multi-factor framework

All purely hypotheticalAlpha scores follow a 0 to 1 scoring systemAlpha scores are equally weighted

Pros─ Ability to combine multiple views of the company─ Explicit weight can be assigned (e.g. Blend ESG model at 25% to the overall Combined model) Cons─ Does not guarantee that a certain target ESG exposure is achieved─ Does not guarantee that obviously ‘bad’ names do not make it into the portfolio

Any descriptions or information involving investment process or strategies is provided for illustration purposes only, may not be fully indicative of any present or future investments, may be changedand are not intended to reflect performance.

© M

an 2

019

Next generation ESG exposure and risk analytics

Source: Man Group plc. Illustrative examples - for information purposes only. 12

Combines off-the-shelf ESG data with proprietary approach

Proprietary ESG analytics inform PM research with non-financial information sources Tool contributes to investment process, rather than purely reporting Scores are measurement, not targets

12