3D Investing: Implications

for Net Zero

3D INVESTING: IMPLICATIONS
FOR NET ZERO
Clint Howard
Quantitative Researcher, Robeco, Rotterdam, the Netherlands
Mike Chen
Head of Next Gen Research, Robeco, Rotterdam, the Netherlands

Traditional mean–variance portfolio optimization is based on the premise

that investors care only about risk and return. Some investors, however,
also have nonfinancial objectives, such as sustainability goals. Central to
these goals, such as working toward net-zero emissions, is the question of
how to incorporate such objectives into an investor’s portfolio. We show how
an extended mean–variance–sustainability optimization can incorporate
sustainability goals into a portfolio, particularly aligning the portfolio with
the net-zero transition set out in the Paris Agreement. Importantly, we
compare various methods for integrating sustainability goals in investor
portfolios and highlight the implications of such approaches on investor
outcomes.

Introduction
Numerous approaches have challenged the standard risk-and-return portfolio
framework. All of them focus on making investment decisions based on
objectives that are not strictly risk or return based, such as impact investing,
socially responsible investing (SRI), or environmental, social, and corporate
governance (ESG) investing. Accordingly, investment practice has evolved to
incorporate sustainability objectives into the investment problem, including
metrics related to carbon footprint, ESG characteristics, and sustainability
development goals (SDGs). In this chapter, we explore potential applications and
implications of the 3D investing framework from Blitz, Chen, Howard, and Lohre
(2024) in the context of net-zero transition alignment, as outlined in the Paris
Agreement, adopted at the UN Climate Change Conference (COP21) in Paris on
12 December 2015.

The Paris Agreement is a landmark treaty in which 195 nations committed to

limit global temperature rise this century to less than 2°C above preindustrial
levels and pursue efforts to target an increase of less than 1.5°C. In 2018,
the Intergovernmental Panel on Climate Change (IPCC) stated that carbon
emissions need to reach net-zero neutrality by 2050 to limit global warming
to 1.5°C (IPCC 2018). Achieving these ambitious climate and decarbonization

Author’s note: This chapter is based on the article “3D Investing: Jointly Optimizing Return, Risk, and Sustainability”
in the Financial Analysts Journal (Blitz, Chen, Howard, and Lohre 2024), with an extended discussion around
potential net-zero implications and applications of the original article. The views expressed herein are not
necessarily shared by Robeco or its subsidiaries.

© 2024 CFA Institute. All rights reserved. | 1

Investment Innovations Toward Achieving Net Zero: Voices of Influence

goals requires investors to integrate net-zero transition objectives alongside

traditional risk and return considerations, necessitating flexible portfolio
construction frameworks.

Considering these ambitious climate and decarbonization goals, academics and

practitioners have started developing new frameworks and toolkits to address
the urgent need to decarbonize. At the center of this work is the concept of
decarbonization pathways and trajectories toward net zero. These concepts
can be seen as an evolution or extension of “low-carbon” portfolios, which aim
to reduce exposure to assets with high carbon footprints at the moment of
investment. Net-zero portfolios additionally aim to help transition the economy
from “brown” to “green,” which is inherently a more challenging forward-
looking problem. Barahhou, Ben Slimane, Roncalli, and Oulid Azouz (2022)
argue that constructing a net-zero portfolio is more complex than constructing
a decarbonized portfolio because of the multi-objective nature of reducing
portfolio carbon and financing the transition. At its core, the desire to construct
net-zero-aligned portfolios is a multi-objective optimization problem.

Blitz et al. (2024) show how portfolio decarbonization can be achieved using
both constraints and an objective function term and highlight how, for
ambitious targets with low active risk budgets, the objective function term
outperforms. The study’s results show that for portfolios that seek to track
the benchmark closely while outperforming it, ambitious sustainability goals
are better implemented using a direct objective function term rather than a
portfolio-level constraint. The objective function term allows for a rewarded
time-varying trade-off of a stock’s expected return and the stock’s contribution
toward the sustainability objective. It is this flexibility to decide at the portfolio
construction’s run time when it might be better to go for expected return
vis-à-vis sustainability that gives the superior result of the objective function
approach. In this chapter, we relate the concept of 3D investing to that of
net-zero investing and the many-dimension problem of integrating net-zero
objectives into a portfolio.

In recent years, the construction of net-zero portfolios has received considerable

attention from both academics and practitioners. Bolton, Kacperczyk, and
Samama (2022) propose a framework to align portfolios with a carbon budget
that aims to keep global temperature rise below 1.5°C. This approach aims to
maintain minimum tracking error to a market index while demonstrating the
importance of time for reducing emissions. Le Guenedal and Roncalli (2022)
survey how asset managers measures climate risk and constructs portfolios
based on these climate risks. They highlight the importance of considering the
impact of different carbon emission scopes and the challenges of integrating
these objectives into the portfolio. Importantly, they highlight the nuance
between portfolio decarbonization and portfolio alignment with Paris Aligned
Benchmarks and net-zero carbon objectives. Jondeau, Mojon, and Pereira da
Silva (2021) provide methodologies for constructing benchmark portfolios
where the component companies’ carbon footprint decreases over time.
In this chapter, we explore the applications and implications of a 3D investing

2 | CFA Institute
 3D Investing: Implications for Net Zero

framework for the pressing challenge of constructing net-zero-aligned

portfolios.

One of the key considerations with net-zero investing is balancing the long-term
objective of reaching net zero by 2050 with the short- to medium-term
objectives and incentives around balancing risk and return. Constructing
net-zero portfolios is inherently a multi-objective problem, weighing
decarbonization against financing the transition, risk, and return. Investors are
balancing the urgency of decarbonizing the portfolio with the need to maintain
the return and risk profile of the portfolios that they manage. Such a balance
naturally requires a multi-faceted optimization approach that can incorporate
numerous objectives alongside risk and return.

Specifically, in the context of net-zero investing, one mechanism could be to

incorporate a forward-looking net-zero metric into the objective function and
encourage the portfolio optimizer to take exposure to stocks based on expected
returns, risk, and forward-looking net-zero expectations. If one considers
incorporating Paris Aligned Benchmarks, these benchmarks effectively
require a 50% carbon-intensity reduction relative to the benchmark based on
current emissions, 7% year-on-year decarbonization, and adherence to several
exclusions and exposure constraints. Meeting such objectives can naturally be
achieved with both constraints and objective function terms. Blitz et al. (2024)
show that for more ambitious carbon footprint reductions and lower tracking
error targets, the objective function term helps reduce turnover and increase
expected net outperformance.

Given the strict requirements of Paris Aligned Benchmarks, one could apply
a portfolio construction paradigm that consists of portfolio-level constraints
on current emissions, an objective function term on current emissions, and an
objective function term on expected future emissions. Such an approach could
allow for meeting the immediate-term requirements while also allowing the
portfolio to take on greater exposure to decarbonization when it is “cheap” from
an expected return or risk perspective. For example, if investors’ expected return
forecasts about highly emitting stocks are currently very negative, then they
may be willing to take a larger underweight in such stocks if they also derive
additional “net-zero utility” from such a position. Given that reducing current
emissions is more valuable from a net-zero perspective than reducing future
emissions, as shown by Daniel, Litterman, and Wagner (2019) and Fearnside,
Lashof, and Moura-Costa (2000),1 having a portfolio construction framework
that can dynamically trade off return, risk, and net-zero objectives may lead
to superior after-cost performance while meeting all stated objectives for
integrating net-zero goals into the portfolio.

The question of how to integrate environmental objectives into an investment

decision has been studied extensively. Repetto and Austin (2000) propose a

1
This is the so-called time value of carbon. See the Wikipedia page on the topic: https://en.wikipedia.org/wiki/
Time_value_of_carbon.

CFA Institute | 3
Investment Innovations Toward Achieving Net Zero: Voices of Influence

methodology to integrate environmental issues into the analysis of individual

companies, using a scenario-based approach to evaluate the impact of
emerging environmental issues on a company’s operations. Barber, Morse,
and Yasuda (2021) show how, in recent years, investors have begun to derive
nonpecuniary utility when investing in dual-objective venture capital impact
funds. They argue that investors are willing to sacrifice returns in pursuit
of these alternative objectives.

Many approaches that strive to incorporate more general sustainability

objectives into a portfolio have been proposed in the literature. These include
excluding undesirable stocks from the investment universe (Diltz 1995;
Kinder and Domini 1997; Naber 2001), constraining the portfolio’s exposure
to such objectives (Boudt, Cornelissen, and Croux 2013), and incorporating
sustainable targets into the return/alpha component of the objective
function (Steuer, Qi, and Hirschberger 2007; Bilbao-Terol, Arenas-Parra, and
Cañal-Fernández 2012; Hirschberger, Steuer, Utz, Wimmer, and Qi 2013;
Utz, Wimmer, Hirschberger, and Steuer 2014; Chen and Mussalli 2020).

The key tension of net-zero portfolio construction is the desired urgency of

decarbonizing while meeting core risk and return objectives. All portfolio
construction methods have different positives and negatives in considering
these specific tensions. For example, divesting from high-carbon-emitting
companies may significantly improve the immediate carbon profile of a
portfolio, yet these companies may be best positioned to help develop and
implement transitional technologies. Similarly, excluding a substantial portion
of stocks may introduce significant added risk to a portfolio that is not within
the risk budget. The investor’s core focus is to balance these dimensions, and
toolkits such as 3D investing can provide insights into how these dimensions
interact in a portfolio.

In this chapter, we explore how a 3D investing framework could be applied to the

challenge of constructing investment portfolios aligned with net-zero emission
goals. Building on the work of Blitz et al. (2024), we show how integrating
forward-looking climate metrics and emission pathway constraints into a
multi-objective portfolio optimization could help investors navigate the complex
trade-offs between decarbonization, performance, and risk. A 3D investing
framework can allow for dynamic exposure to climate leaders and laggards
based on return expectations and sustainability characteristics while adhering to
decarbonization pathways. As investors grapple with the urgency of the net-zero
transition, frameworks such as 3D investing will be useful tools for helping align
portfolios on multiple dimensions.

The remainder of this chapter is organized as follows: In the next two sections,
we outline the general multi-objective optimization framework and illustrate the
use of 3D investing for climate objectives. Then, we explore the implications and
applications for net-zero portfolios. Finally, we provide concluding remarks.

4 | CFA Institute
 3D Investing: Implications for Net Zero

Multi-Objective Optimization Framework

We begin by introducing the portfolio optimization framework that we work
with. First, we specify the common mean–variance optimization framework,
where the investor trades off maximizing expected returns while jointly
minimizing risk. We then expand this optimization paradigm to a multi-objective
optimization framework.

Standard Mean–Variance Optimization

Equation 1 shows the standard mean–variance optimization formula:

γ
max λw ′µ − w ′Σw,
w 2 (1)
s.t. w ′e = 1,

where

w is an N × 1 vector of asset weights

µ is an N × 1 vector of expected returns
Σ is the N × N variance–covariance matrix
e is an N × 1 vector of ones
λ and γ are scalar coefficients

Portfolios generated under Equation 1 are mean–variance optimal in that they

achieve the maximum expected return for a given level of risk. This framework
can be extended to include additional dimensions, such as constraining the
portfolio relative to some benchmark (Jorion 2003), incorporating transaction
cost penalties (Taksar, Klass, and Assaf 1988; Ledoit and Wolf 2022),
penalizing turnover (Hautsch and Voigt 2019), or enforcing positive asset
weights (Jagannathan and Ma 2003). Ibbotson, Idzorek, Kaplan, and Xiong
(2018) explore a popularity asset pricing model (PAPM) where they introduce
additional “popularity” characteristics into the standard CAPM framework.
Such an approach generalizes the standard mean–variance optimization
problem to any number of alternative objectives. Steuer, Qi, and Hirschberger
(2007) derive analytical solutions for an efficient portfolio surface with three
criteria, using portfolio liquidity as an example. They extend the classical
two-mutual-fund theorem to a three-mutual-fund theorem and show how the
obtained three-dimensional efficient surface has paraboloidal/hyperboloidal
structures.

CFA Institute | 5
Investment Innovations Toward Achieving Net Zero: Voices of Influence

A Multi-Objective Optimization Framework

It is straightforward to extend the mean–variance optimizer from Equation 1

to construct portfolios on an efficient frontier surface in three (or more)
dimensions. In the case of additional sustainability considerations, Equation 1
can be extended to three dimensions as follows:

γ
max λw ′µ + (1 − λ)w ′µ SI − w ′Σw,
w 2 (2)
s.t. w ′e = 1, w ∈Ω,

where µSI is an N × 1 vector of any (discrete or continuous) sustainability metric,

λ becomes the relative preference between the return and sustainability
objectives, and Ω is the set of feasible solutions, which includes any portfolio
constraints. This formulation is general and can accommodate the incorporation
of common sustainability characteristics. These include commercial ESG
metrics from vendors, such as MSCI and Sustainalytics; carbon footprint; SDG
scores; and climate transition scores. The only requirement here is that the
sustainability metric is ordinal.2

Targeting a Climate Traffic Light

To illustrate how the 3D investing framework can easily integrate forward-looking
climate measures, we use the simulation framework of Blitz et al. (2024) with
the Robeco Climate Traffic Light (CTL) scores (Robeco 2022).3 To summarize,
we use an MSCI World Index developed markets universe alongside a simple
expected returns model and variance–covariance matrix to conduct benchmark-
relative portfolio optimization exercises.4 Our sample consists of MSCI World
constituents at the end of every month from December 1989 to December
2022.5 We source stock returns and fundamental data from Refinitiv.

We use a portfolio optimization setting that mimics the construction of a

real-life investment portfolio applying realistic portfolio constraints and settings.
We construct portfolios with tracking errors of 0.5% because it represents
the challenging multi-objective scenario of delivering high expected returns
and sustainability goals with a limited risk budget. The portfolio exposure to
regions (defined as North America, Europe, and Asia Pacific) and Global Industry
Classification Standard (GICS) first-level sectors are restricted to ±0.5% of the
benchmark market-capitalization-weighted value. Portfolios must be long only.
The maximum trade size is limited to 25% of a stock’s average daily volume
over the past 65 trading days (ADV). The maximum stock weight relative to

2
For practical considerations on the sustainability metric, µSI, see Chen and Mussalli (2020).
3
We additionally use the data simulation approach of Blitz and Hoogteijling (2022) to produce a longer history
of carbon footprint data and SDG data. Note that any potential forward information leakage is of little concern
because we are comparing two portfolio construction approaches using the same data. We aim to illustrate the
broad application of our methodology on a representative set of sustainability data.
4
For full details on the portfolio implementation, see Blitz et al. (2024).
5
Prior to 2001, we use constituents of the FTSE Developed Markets index as a proxy for MSCI World constituents.

6 | CFA Institute
 3D Investing: Implications for Net Zero

the benchmark (i.e., active weight) is ±0.5%. The maximum active share of the
portfolio is 40%. The portfolio must be fully invested. We assume that the funds
under management grow with the realized market return, and we design the
simulations such that the final fund size at the end of 2022 is EUR4 billion. We
incorporate a turnover penalty into the objective function, which is the sum of
the squared absolute trade sizes.

As we target specific tracking errors, we transform the weight vector of

Equation 2 from absolute asset weights to benchmark-relative weights:6

w new = w p − w bm .

Our portfolio optimization problem for a single time step is then given by

γ
max λ1w ′new µ + λ 2 w ′new µ SI − w ′new
w 2 ∑w new
− κ w new − w old , (3)

where wold represents the portfolio weights immediately before the rebalance,
κ is a scaling parameter for the turnover penalty (we set κ = 1), and we incorporate
the previously described constraints. We use a base set of portfolio construction
constraints and settings across our simulations, and then we permute the expected
return coefficient (λ1), the risk aversion coefficient (γ), and the sustainability
coefficient (λ2) in each different optimization. Lastly, we introduce an additional
optional constraint on either carbon footprint or SDG scores (e.g., the portfolio
carbon footprint must be less than or equal to the benchmark carbon footprint.)

As inputs of expected returns µ, we use a simple equal-weighted multifactor

score (denoted QMV) consisting of value, quality, and momentum signals. For
value, we use an equal-weighted combination of book to price and 12-month
forward earnings to price, ranked within GICS sectors. For quality, we use
an equal-weighted combination of return on equity and debt to assets. For
momentum, we use the previous 12-minus-1-month return. Each of the four
underlying signals is first rank standardized between -1 and +1. The signals
are then combined into a single multifactor score. We aim not to construct the
best multifactor score but rather to construct a simplified score that represents
common choices and implementations of multifactor investment strategies.

As for expected risk, we use a standard variance–covariance (VCV) matrix (Σ)

that follows a latent factor model approach where we apply principal component
analysis (PCA) with 20 components to the sample VCV matrix estimated using
60 months of daily return data. We use five-day overlapping returns to account
for market asynchronicity (Burns, Engle, and Mezrich 1998).

Exhibit 1 shows the ex ante view of expected returns, ex ante tracking error, and
CTL improvement over the benchmark as of December 2023. By mapping out a
3D surface of these elements, we can see how the objective of taking on more

6
We use the same benchmark, the MSCI World, when constructing portfolios and evaluating financial and
sustainability objectives.

CFA Institute | 7
Investment Innovations Toward Achieving Net Zero: Voices of Influence

Exhibit 1. Climate Traffic Light Efficient Surface

Expected Return High

Medium

Low
Positive
Tr
af
fic
Li

Zero
gh
tI
m

High
pr

ror
ov

g Er
em

Medium ackin
Tr
en

Negative nte
t

Low Ex A

Note: This graph plots the ex ante expected return/tracking error/sustainability surface for Robeco’s climate traffic light. The solid black line
corresponds to the ex ante expected return/tracking error efficient frontier (i.e., the traditional case where only risk and return are considered).
The surface is shaded based on the y-axis variable (climate traffic light relative to the benchmark), where green corresponds to a higher
improvement and magenta corresponds to a lower improvement. This surface was calculated using data as of December 2023.

exposure to positive forward-looking climate stocks affects the risk and return
characteristics of the optimal portfolios. In line with expectations, as the desire
to integrate an alternative objective (which is not necessarily correlated with
expected returns) into the portfolio increases, this integration requires either
increasing tracking error or reducing expected returns.

Exhibit 2 compares the historical CTL profiles of portfolios constructed using

different optimization approaches. It illustrates how the time-varying nature
of a 3D investing approach can vary in comparison to a strict constraint. The
dark blue line at the bottom represents an unconstrained portfolio that seeks
to maximize expected excess returns without any consideration of CTL scores.
This exposure is identical to the CTL improvement that is at least as good as the
benchmark (“2D Constrained at 0%” yellow dotted line), suggesting that this
constraint is not binding at any time. The “2D Constrained at 40%” bright blue
line represents a portfolio that targets a minimum 40% CTL improvement relative
to the benchmark at each rebalancing date, using a 2D optimization approach
with a hard constraint on the minimum CTL score. The “3D Objective” orange
line represents a CTL improvement using a 3D optimization approach. The “3D
Constrained at 40%” gray line represents a portfolio that targets a minimum 40%

8 | CFA Institute
 3D Investing: Implications for Net Zero

Exhibit 2. Climate Traffic Light Improvement to MSCI World

under Various Optimization Scenarios
70.00%

60.00%
Traffic Light Relative to Benchmark

50.00%

40.00%

30.00%

20.00%

10.00%

0.00%
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
Unconstrained 2D Constrained at 0% 2D Constrained at 40%
3D Constrained at 40% 3D Objective

Note: This figure plots the percentage improvement of the portfolio’s climate traffic light exposure over the MSCI World climate traffic light
exposure using different 2D and 3D portfolio construction approaches. We report results for a portfolio with a tracking error target of 0.5%.

CTL improvement using a 3D optimization approach. This approach allows for a

flexible trade-off between the competing objectives because the optimizer can
choose to exceed the 40% minimum CTL improvement if doing so is expected
to enhance returns or reduce risk. Further, in the 1999–2000 period, we can
see what happens when a constraint cannot be satisfied. At this point, the “2D
Constrained at 40%” bright blue line is unable to meet the 40% constraint and
thus is forced to deviate to find a portfolio that satisfies this constraint.

These illustrative examples show how one can simply model the incorporation
of an alternative objective into portfolio optimization. This outcome can be
achieved by changing the expected return forecast for a stock or simply adding
the term into the objective function with a prespecified parameter. As shown
in Exhibit 2, both the 2D and 3D approaches that target a minimum 40% CTL
improvement achieve this objective consistently over time. The 3D approach,
however, exhibits greater variability in its CTL profile, occasionally exceeding
the 40% minimum by a significant margin, because the 3D approach allows
the optimizer to prioritize CTL improvement more heavily when it is expected
to be beneficial from a risk–return perspective. The results presented in
Exhibits 1 and 2 demonstrate the flexibility and effectiveness of the 3D investing
framework in incorporating forward-looking climate metrics into the portfolio
construction process.

CFA Institute | 9
Investment Innovations Toward Achieving Net Zero: Voices of Influence

It is important to note that the specific results presented here are based on a
particular set of assumptions and data inputs and may not be representative
of all scenarios. The appropriate trade-off between expected returns, risk, and
climate alignment will depend on an investor’s specific preferences, constraints,
and investment horizon. Nevertheless, the 3D investing framework provides
a useful tool for exploring these trade-offs in a systematic and transparent
manner and can be adapted to incorporate a wide range of forward-looking
climate metrics and optimization objectives.

Implications and Applications of 3D Investing

for Net-Zero Portfolios
The CTL example is a simple application of the 3D investing framework of Blitz
et al. (2024) but does not present anything new. Rather, it demonstrates how
incorporating a simple forward-looking climate measure into the objective
function is a trivial process, and the decision one must make concerns the
relative risk–return cost of integrating this objective. Naturally, the question
that someone using such a framework must answer is, What forward-looking
climate measure do I want to target? This is a key challenge of the net-zero
framework: The required forward-looking nature of both financing the transition
and decarbonizing means that there is uncertainty around how to measure and
model the required decarbonization pathway. Nevertheless, in this section, we
elaborate on some of the implications of net zero for portfolio construction and
present potential mechanisms for integrating net-zero goals into the portfolio
construction problem.

Implications of Net Zero for Portfolio Construction

The transition to a net-zero economy has significant implications for portfolio

construction because investors must navigate the complex trade-offs between
achieving long-term climate goals and maintaining short-term financial
performance. Traditional portfolio optimization frameworks, which focus solely
on expected returns and risk, must be extended to handle the multi-objective
nature of net-zero investing. One of the key challenges in constructing net-
zero portfolios is balancing the need to reduce portfolio emissions in the short
term with the objective of financing the transition to a low-carbon economy in
the longer term. It requires investors to consider not only the current carbon
footprint of their holdings but also the forward-looking emission trajectories and
transition plans of the companies in which they invest.

The 3D investing framework provides a tool for navigating these trade-offs

by allowing investors to explicitly incorporate both short-term emission
reduction targets and long-term net-zero alignment objectives into the
portfolio construction process. By including a term in the objective function
that minimizes the portfolio’s current carbon footprint, investors can ensure
that their portfolios are aligned with the urgent need to reduce emissions
in the near term. At the same time, by incorporating forward-looking metrics

10 | CFA Institute
 3D Investing: Implications for Net Zero

such as Implied Temperature Rise or transition readiness scores, investors can

position their portfolios for the long-term transition to a net-zero economy. This
forward-looking perspective is important for identifying companies that are well
positioned to thrive in a low-carbon future and avoiding those with elevated
risks of being left behind.

Another key implication of net-zero investing is the need to consider the

real-world impact of portfolio allocation decisions. Although traditional
portfolio optimization focuses solely on the financial outcomes for the investor,
net-zero investing requires a broader perspective that considers the impact
of investment decisions on the overall decarbonization of the economy.
The 3D investing framework can accommodate this broader perspective by
incorporating metrics that capture the alignment of portfolio companies with
science-based emission reduction targets or the contribution of portfolio
holdings to the financing of low-carbon solutions. By explicitly considering these
real-world impact metrics alongside financial objectives, investors can ensure
that their portfolios not only are aligned with net-zero goals but also support the
transition to a low-carbon economy.

Constructing net-zero portfolios using a 3D investing framework presents some

challenges, however. One key issue is the need to specify the relative weights
of the various objectives in the optimization process, which can be a complex
and subjective exercise. Investors must consider their own preferences and
constraints when setting these weights, as well as the potential trade-offs
between short-term and long-term objectives. Another challenge is the need
for robust and reliable data on the emission trajectories and transition plans of
portfolio companies. Although a growing number of companies are disclosing
this information, the quality and comparability of these disclosures vary, making
it difficult for investors to accurately assess the net-zero alignment of their
portfolios. Naturally, any portfolio construction technique will grapple with
similar challenges around data quality.

Despite these challenges, a 3D investing framework provides a valuable starting

point for investors seeking to align their portfolios with net-zero objectives.
By explicitly incorporating emission reduction targets and forward-looking
transition metrics into the portfolio construction process, this approach enables
investors to navigate the complex trade-offs between short-term and long-term
objectives while also considering the real-world impact of their investment
decisions. As the data and methodologies for net-zero investing continue
to evolve, the 3D investing framework can serve as a foundation for further
innovation and refinement in this critical area of sustainable finance. Although
3D investing provides a useful toolkit, investors face complex decisions around
how to appropriately weight different objectives, which will require careful
consideration of their specific constraints and objectives.

CFA Institute | 11
Investment Innovations Toward Achieving Net Zero: Voices of Influence

Incorporating Forward-Looking Net-Zero Metrics

Forward-looking metrics go beyond simple measures of current carbon footprint

and aim to capture a company’s alignment with future net-zero trajectories.
By incorporating such forward-looking measures, investors can construct
portfolios that may be better positioned for the transition to a low-carbon
economy. The quality of the forward-looking measure and what it aims to
capture specifically will influence the characteristics of any portfolio that
integrates such a measure.

The climate traffic light we discussed is one example of a forward-looking

climate metric. Investors may have a preference for other metrics, however, and
our proposed framework accommodates any ordinal measure. The following
are other examples of forward-looking net-zero metrics that could be integrated
into a 3D investing framework:

● Implied Temperature Rise: This metric estimates the global temperature rise
associated with a company’s emission trajectory, providing an indication of
its alignment with the Paris Agreement goals. A company with an Implied
Temperature Rise below 2°C would be considered aligned with net-zero
objectives.
● Science-Based Targets initiative (SBTi) portfolio coverage: This metric
estimates the percentage of a portfolio’s holdings that have set emission
reduction targets validated by the SBTi as consistent with the Paris
Agreement goals.
● Transition readiness scores: These scores assess a company’s preparedness
for the low-carbon transition based on such factors as its decarbonization
strategy, capital allocation plans, and climate governance. Higher scores
indicate better positioning for the net-zero transition.

To incorporate these metrics into a 3D investing framework, an investor could

modify the objective function in Equation 2 as follows:

γ
λ1w ′µ + λ 2 w ′µ ITR + λ 3 w ′µ SBTi + λ 4 w ′µCTL − w ′Σw,
2

where µITR, µSBTi, and µCTL are vectors of the chosen forward-looking net-zero
metrics for each asset. The λi parameters control the relative importance of
each forward-looking metric alongside expected returns (µ) and risk (Σ) in the
optimization process. The choice of values for the λi parameters will depend on
an investor’s specific net-zero goals and risk–return preferences. One approach
could be to set these weights based on each metric’s perceived importance and
potential financial materiality. Alternatively, investors could use optimization
techniques to identify the combination of weights that best aligns with their
overall objectives, subject to tracking error and other constraints. As with any
optimization input, sensitivity analysis will be important to understanding the
impact of these choices.

12 | CFA Institute
 3D Investing: Implications for Net Zero

By incorporating forward-looking net-zero metrics in this way, the 3D investing

framework allows investors to construct portfolios that are not only aligned with
current carbon reduction goals but also positioned for the long-term transition
to net zero. This forward-looking perspective is crucial for investors seeking to
manage the risks and opportunities associated with the low-carbon transition
while still achieving their financial objectives.

Implementing Net-Zero Pathways

The 3D investing framework can also be used to construct portfolios that align
with specific net-zero emission pathways or glidepaths over time. For instance,
an investor could modify Equation 2 to include an additional constraint:
Eactual(t) ≤ Etarget(t), where Eactual(t) is the portfolio emissions at time t and Etarget(t)
is the target emissions level at time t prescribed by a net-zero pathway. The
3D optimization would then produce the portfolio that maximizes alpha and
sustainability objectives and minimizes risk while also satisfying the net-zero
glide path constraint. This approach ensures alignment with a long-term net-
zero trajectory while allowing time-varying exposures based on expected returns
and sustainability characteristics. Such a constraint could also trivially be added
to any portfolio optimization problem and is not unique to a multi-objective
framework.

Bolton et al. (2022) demonstrate how it is possible to achieve a net-zero

portfolio that tracks major indexes7 with limited tracking error, even if the
underlying reference benchmark’s carbon emission stays at the 2020 level. The
authors did not consider the potential for alpha generation in such a portfolio.
We use their portfolio construction as a starting point but now consider how
one may incorporate alpha considerations in such a portfolio.

Following Bolton et al. (2022), we consider the total cumulative carbon budget
of 268.5 gigatons (Gt) of carbon dioxide (CO2) as of 2021 to meet the 1.5°C
target by 2050. With this starting point of total emission, different pathways
to the 1.5°C target exist, dependent on both the start date and level of
decarbonization.8 Regardless of the pathway chosen, we define the following:

● The net-zero investor’s chosen target pathway portfolio emission at year t is

Etarget(t).
● The actual portfolio emission at year t is Eactual(t).

∑
t
● The cumulative target pathway emission as of year t is Ctarget (t ) = E (i ).
i =0 target

∑
t
● The cumulative actual emission as of year t is Cactual (t ) = E (i ).
i =0 actual

7
Bolton et al. (2022) considered the MSCI All Country World, MSCI Europe, and MSCI Emerging Markets indexes.
8
Bolton et al. (2022) explicitly state “starting in 2021, with a geometrical rate of emission reduction, the path can
be either an immediate 25% reduction in carbon footprint, followed by an 85% decrease, or a constant annual
10% reduction. With a linear rate, the pathway can be either a 25% initial reduction, followed by an annual 3.2%
reduction, or a constant annual 4.6% reduction. All these paths are structured so that the entire carbon budget of
268.5 Gt CO2 is spent.”

CFA Institute | 13
Investment Innovations Toward Achieving Net Zero: Voices of Influence

The problem of jointly optimizing alpha and risk and satisfying a net-zero path
becomes
γ
max λw ′µ + (1 − λ)Eactual
−1
(t ) − w ′Σw,
w 2 (4)
s.t. w ′e = 1, w ∈Ω, Cactual (t ) ≤ Ctarget (t ).

The objective function in Equation 4 is set up to jointly optimize alpha, risk,

and actual annual carbon emission. The objective function will aim to minimize
the actual carbon emission, but it is allowed to go above the target pathway
emission, Etarget(t), if doing so will yield more attractive expected return or
risk profiles. At the same time, the cumulative actual emission, Cactual(t), is
constrained to stay below the target pathway emission, Ctarget(t), at each point
in time. That is to say, the optimization problem will allow the actual annual
emission to go above the target pathway annual emission only if there have
been excess emissions “saved up” in previous years. We know that there is a
temporal dimension to the impact of emissions on climate change (see Daniel
et al. 2019; Fearnside et al. 2000). A ton of CO2 does more damage to climate if
released into the atmosphere now compared with the same ton of CO2 released
into the atmosphere later, all else equal. This means that with the constraint
Cactual(t) ≤ Ctarget(t), the optimized portfolio will strictly follow a net-zero path
presented in Bolton et al. (2022) while jointly optimizing the immediate alpha,
risk, and emissions considerations.

This formulation also has some limitations. One key drawback is that it requires
specifying the net-zero pathway, Ctarget(t), ex ante, which may not be optimal if
new information emerges over time that suggests a different pathway would be
more appropriate. Additionally, the use of a hard cumulative emission constraint
may lead to suboptimal portfolios in some cases because it does not allow for
any trade-off between emissions and other objectives once the constraint is
binding. Thus, there is an element of path dependency, which any portfolio
construction approach targeting a pathway will be exposed to. It is important to
understand the implications of such constraints on the risk and return objectives.

To address these limitations, investors could consider several extensions to

the formulation in Equation 4. For example, the cumulative emission constraint
could be complemented with a penalty term in the objective function that
imposes a cost on deviations from the target pathway. This situation could allow
for a more flexible trade-off between current emissions, cumulative emissions,
and other objectives while still ensuring alignment with the net-zero pathway.

It is important to note that the emission pathway constraint in Equation 4

operates independently of any other sustainability metrics in the objective
function. In some cases, these objectives may be in tension—for example,
favoring companies with strong transition plans could lead to short-term
deviations from the desired pathway. Investors will need to carefully balance
these considerations and may wish to fine-tune the relative weights in the

14 | CFA Institute
 3D Investing: Implications for Net Zero

objective function over time as new information becomes available. The 3D

framework provides the flexibility to explore this balance, but the onus remains
on investors to define their priorities and manage these trade-offs.

Finally, although a 3D investing framework provides a conceptual toolkit for

navigating the complexities of net-zero portfolio construction, its practical
implementation (and that of any portfolio construction approach) depends
on the availability of high-quality, consistent, and comprehensive data.
Investors seeking to incorporate forward-looking metrics such as Implied
Temperature Rise, science-based targets, and transition readiness into their
portfolio optimization face continuing data challenges. Many companies still
do not disclose their full Scope 1, 2, and 3 emissions, let alone more granular
information on their decarbonization strategies and capital allocation plans.
Even among firms that do report this information, many methodologies and
metrics lack standardization, making comparisons difficult. Moreover, the
reliability of self-reported data can be questionable, highlighting the need
for more robust auditing and verification processes. An important area is the
continued development of comprehensive, standardized, and reliable datasets
on corporate climate performance and risk management. Progress on this front
will require a concerted effort from regulators, standard setters, investors,
and companies to improve the quality and comparability of climate-related
disclosures.

Conclusion
As the world grapples with the urgent need to decarbonize the global economy
and achieve net-zero emissions by 2050, investors face the challenge of how
to construct portfolios that align with these ambitious climate goals while still
delivering on risk and return objectives. This chapter explores the value of the
3D investing framework as a tool for constructing net-zero-aligned portfolios.
By explicitly incorporating sustainability metrics into the portfolio optimization
objective function, 3D investing allows for dynamic trade-offs between expected
returns, risk, and climate outcomes based on an investor’s unique preferences
and constraints. We show how the framework can be extended to incorporate
forward-looking climate metrics and emission pathway constraints, enabling
investors to pursue short-term decarbonization while preserving long-term
alignment with net-zero targets. We also acknowledge, however, the inherent
tensions in net-zero investing, such as balancing short-term performance with
long-term climate goals, and the need for investor discretion in navigating these
trade-offs.

Our analysis provides insights into applications of portfolio construction

paradigms, but we recognize several limitations and areas for future research.
A 3D net-zero investing framework must assume a forward-looking climate
metric that captures the nuances of companies’ decarbonization trajectories and
potential contributions to real-world emission reductions. Future work could
also explore how 3D investing could be adapted to optimize for climate impact

CFA Institute | 15
Investment Innovations Toward Achieving Net Zero: Voices of Influence

beyond individual portfolio alignment, although quantifying this impact remains

challenging.

Ultimately, translating these research insights into implementable net-zero

investment solutions will require close collaboration between academics and
practitioners. As climate goals evolve and data availability improves over time,
investors will need to continually adapt and refine their approaches to net-zero
portfolio construction. A 3D investing framework provides a framework for
this ongoing innovation, offering the flexibility and rigor needed to face the
challenge of aligning investment portfolios with the net-zero future.

References
Barahhou, I., M. Ben Slimane, T. Roncalli, and N. Oulid Azouz. 2022. “Net Zero
Investment Portfolios—Part 1. The Comprehensive Integrated Approach.”
Working paper (13 October). doi:10.2139/ssrn.4283998.

Barber, B. M., A. Morse, and A. Yasuda. 2021. “Impact Investing.” Journal of

Financial Economics 139 (1): 162–85. doi:10.1016/j.jfineco.2020.07.008.

Bilbao-Terol, A., M. Arenas-Parra, and V. Cañal-Fernández. 2012. “A Fuzzy

Multi-Objective Approach for Sustainable Investments.” Expert Systems with
Applications 39 (12): 10904–15. doi:10.1016/j.eswa.2012.03.034.

Blitz, D., M. Chen, C. Howard, and H. Lohre. 2024. “3D Investing: Jointly
Optimizing Return, Risk, and Sustainability.” Financial Analysts Journal
80 (3): 59–75. doi:10.1080/0015198X.2024.2335142.

Blitz, D., and T. Hoogteijling. 2022. “Carbon-Tax-Adjusted Value.” Journal of

Portfolio Management 48 (5): 121–37. doi:10.3905/jpm.2022.1.343.

Bolton, P., M. Kacperczyk, and F. Samama. 2022. “Net-Zero Carbon Portfolio

Alignment.” Financial Analysts Journal 78 (2): 19–33. doi:10.1080/00151
98X.2022.2033105.

Boudt, K., J. Cornelissen, and C. Croux. 2013. “The Impact of a Sustainability

Constraint on the Mean-Tracking Error Efficient Frontier.” Economics Letters
119 (3): 255–60. doi:10.1016/j.econlet.2013.03.020.

Burns, P., R. F. Engle, and J. J. Mezrich. 1998. “Correlations and Volatilities

of Asynchronous Data.” Journal of Derivatives 5 (4): 7–18. doi:10.3905/
jod.1998.408000.

Chen, M., and G. Mussalli. 2020. “An Integrated Approach to Quantitative

ESG Investing.” Journal of Portfolio Management 46 (3): 65–74. doi:10.3905/
jpm.2020.46.3.065.

16 | CFA Institute
 3D Investing: Implications for Net Zero

Daniel, K. D., R. B. Litterman, and G. Wagner. 2019. “Declining CO2 Price Paths.”
Proceedings of the National Academy of Sciences of the United States of
America 116 (42): 20886–91. doi:10.1073/pnas.1817444116.

Diltz, J. D. 1995. “Does Social Screening Affect Portfolio Performance?” Journal

of Investing 4 (1): 64–69. doi:10.3905/joi.4.1.64.

Fearnside, P. M., D. A. Lashof, and P. Moura-Costa. 2000. “Accounting

for Time in Mitigating Global Warming through Land-Use Change and
Forestry.” Mitigation and Adaptation Strategies for Global Change 5: 239–70.
doi:10.1023/A:1009625122628.

Hautsch, N., and S. Voigt. 2019. “Large-Scale Portfolio Allocation under

Transaction Costs and Model Uncertainty.” Journal of Econometrics
212 (1): 221–40. doi:10.1016/j.jeconom.2019.04.028.

Hirschberger, M., R. E. Steuer, S. Utz, M. Wimmer, and Y. Qi. 2013. “Computing

the Nondominated Surface in Tri-Criterion Portfolio Selection.” Operations
Research 61 (1): 169–83. doi:10.1287/opre.1120.1140.

Ibbotson, R. G., T. M. Idzorek, P. D. Kaplan, and J. X. Xiong. 2018. Popularity:

A Bridge between Classical and Behavioral Finance. Charlottesville, VA:
CFA Institute Research Foundation. https://rpc.cfainstitute.org/research/
foundation/2018/popularity-bridge-between-classical-and-behavioral-finance.

Intergovernmental Panel on Climate Change (IPCC). 2018. Global Warming of

1.5°C: IPCC Special Report on Impacts of Global Warming of 1.5°C above Pre-
industrial Levels in Context of Strengthening Response to Climate Change,
Sustainable Development, and Efforts to Eradicate Poverty. Cambridge, UK:
Cambridge University Press. doi:10.1017/9781009157940.

Jagannathan, R., and T. Ma. 2003. “Risk Reduction in Large Portfolios: Why
Imposing the Wrong Constraints Helps.” Journal of Finance 58 (4): 1651–83.
doi:10.1111/1540-6261.00580.

Jondeau, E., B. Mojon, and L. A. Pereira da Silva. 2021. “Building Benchmark

Portfolios with Decreasing Carbon Footprints.” Swiss Finance Institute Research
Paper No. 21-91. doi:10.2139/ssrn.3987186.

Jorion, P. 2003. “Portfolio Optimization with Tracking-Error Constraints.”

Financial Analysts Journal 59 (5): 70–82. doi:10.2469/faj.v59.n5.2565.

Kinder, P. D., and A. L. Domini. 1997. “Social Screening: Paradigms Old and New.”
Journal of Investing 6 (4): 12–9. doi:10.3905/joi.1997.408443.

Le Guenedal, T., and T. Roncalli. 2022. “Portfolio Construction with Climate Risk
Measures.” Working paper (3 January). doi:10.2139/ssrn.3999971.

CFA Institute | 17
Investment Innovations Toward Achieving Net Zero: Voices of Influence

Ledoit, O., and M. Wolf. 2022. “Markowitz Portfolios under Transaction Costs.”
University of Zurich Department of Economics Working Paper No. 420.
doi:10.5167/UZH-221804.

Naber, M. 2001. “Catholic Investing: The Effects of Screens on Financial Returns.”

Journal of Investing 10 (4): 58–65. doi:10.3905/joi.2001.319487.

Repetto, R., and D. Austin. 2000. Pure Profit: The Financial Implications of
Environmental Performance. Washington, DC: World Resources Institute.

Robeco. 2022. “Climate Traffic Light.” www.robeco.com/files/docm/docu-

robeco-paris-alignment-assessment.pdf.

Steuer, R. E., Y. Qi, and M. Hirschberger. 2007. “Suitable-Portfolio Investors,

Nondominated Frontier Sensitivity, and the Effect of Multiple Objectives on
Standard Portfolio Selection.” Annals of Operations Research 152: 297–317.
doi:10.1007/s10479-006-0137-1.

Taksar, M., M. J. Klass, and D. Assaf. 1988. “A Diffusion Model for Optimal
Portfolio Selection in the Presence of Brokerage Fees.” Mathematics of
Operations Research 13 (2): 277–94. doi:10.1287/moor.13.2.277.

Utz, S., M. Wimmer, M. Hirschberger, and R. E. Steuer. 2014. “Tri-Criterion

Inverse Portfolio Optimization with Application to Socially Responsible
Mutual Funds.” European Journal of Operational Research 234 (2): 491–98.
doi:10.1016/j.ejor.2013.07.024.

18 | CFA Institute