Designing a climate-aware strategic asset allocation

As the momentum behind decarbonization and net-zero objectives builds around the world, asset owners are increasingly engaged in addressing the investment implications of climate change. Our Investment Strategy Team, in partnership with our Climate Research and ESG Research teams, has developed a framework for integrating climate change and its capital-market effects into multi-asset portfolios. This framework reflects our belief that climate change affects investment outcomes. It impacts macro-level variables, such as GDP growth and inflation; company-level dynamics, such as costs and future company activity; and decisions about regulation and fiscal policy. We believe these, in turn, all impact asset prices and asset allocation. What’s more, we think climate investment themes meaningfully expand the opportunity set for multi-asset portfolios.

Our climate-change framework may be of use to asset owners who are actively seeking to incorporate climate objectives into their portfolios via strategic asset allocation (SAA), as well as those who want to better understand the trade-offs they may face if they decide to specifically incorporate these objectives in their investment policy. The framework consists of three pillars. The first, which we wrote about previously, is focused on incorporating climate-related inputs (including transition risks and physical risks) into the capital market assumptions (CMAs) that underpin our SAA decisions. Our second pillar, the primary subject of this paper, is our climate-aware SAA approach — that is, adding relevant climate metrics to our asset allocation optimization process. By quantifying the climate risk embedded in various potential allocations, for example, we can undertake analysis around climate change similar to our approach to other risk exposures, such as portfolio volatility or maximum expected drawdown. In this paper, we also discuss the third pillar of our framework: implementation — the choice of specific climate-aware building blocks and strategies to express the desired asset allocation.

Below we outline the key research steps and findings that went into the design of our climate-aware SAA approach, including the following:

• After extensive testing, we selected a portfolio-alignment metric, implied temperature rise, as the climate variable for our optimization process, given its forward-looking nature.
• We believe top-down SAA optimization will likely need to be paired with building-block implementation in order to reduce turnover and pursue specific climate-change objectives (e.g., targeting groups of companies that have credible transition plans or offer climate solutions).
• We think fundamental, bottom-up research and active management, coupled with engagement, may play a crucial role in the implementation process. For example, the use of portfolio-alignment metrics relies on investee companies/countries executing on the targets they set out, so active monitoring is critical.

While this paper offers a model for climate-related asset allocation analysis, the process could be customized across multiple dimensions to meet an asset owner’s specific needs and priorities. Bear in mind, too, that climate-related metrics and calculations are evolving at a rapid rate, in both scope and methodology. Asset owners will therefore want to avoid a static approach and instead look to reflect the most up-to-date information.

Step 1: Selecting the climate variable

In choosing the climate variable for our optimization process, we considered a host of possibilities, including:

Science-based targets — Companies commit to reductions in greenhouse gas (GHG) emissions and can submit their targets for validation by the Science Based Targets initiative (SBTi). Targets are considered “science based” if they are deemed to be in line with the goals of the Paris Agreement (i.e., limiting global warming to well below 2°C above pre-industrial levels and pursuing efforts to limit warming to 1.5°C). While a growing number of companies have set SBTi-validated targets, small and mid-sized companies are lagging. Additionally, corporate bonds are covered but not other areas of fixed income, such as government bonds and securitized assets. From the standpoint of an optimization process, one of the key drawbacks to using SBTi would be the binary nature of the metric; we think optimization calls for more differentiation between indices than a simple determination of whether companies in an index have SBTi-validated targets or not. That said, these targets are a key input into projected emissions, which underpin the methodology behind implied temperature rise, our chosen variable, which is discussed in the next section.

Physical risks — As noted, we factor physical risks related to climate change into our CMA process. While we could have used the same underlying metrics in our climate-aware SAA process (e.g., country-level assumptions for future climate events developed with our partners at Woodwell Climate Research Center), this would have involved an element of double counting that we preferred to avoid. (Later in the paper we offer thoughts on how to integrate physical risk in an asset allocation via climate-adaptation-focused investments.)

“Green” indices and ESG scores — We determined that true “green” indices are far too narrow in scope for use in a wider optimization process. And while we could have leveraged our firm’s extensive ESG scoring framework, the scores are broader in scope than needed for a climate-specific exercise, given the inclusion of social and governance factors.

Weighted average carbon intensity (WACI) — One of the most widely used transition metrics, WACI offers simplicity, broad-based coverage, and relative transparency. However, our priority was to focus on more forward-looking metrics that incorporate absolute emission reductions.

Our chosen variable: Implied temperature rise 

Ultimately, we elected to use implied temperature rise (ITR), a metric expressed in degrees Celsius and designed to show the alignment between a company’s long-term emissions trajectory and the pathways to future temperature goals (less than 2°C). ITR measures global warming based on the assumption that the rest of the investment universe acts the same way as the single company being evaluated. ITR projects each company’s cumulative emissions (Scope 1, 2, and 3) by 2070, based on current emissions and emissions-reduction targets.¹ Each company’s cumulative projected emissions are then compared to a GHG budget, specific to its industry and regional exposure.

Aggregation at an index or portfolio level is weighted based on emissions using enterprise value, including cash as the base. We use only Scope 1 and 2 emissions in this exercise given that disclosure of Scope 3 lags significantly behind.

One of the key advantages of ITR is that it is forward looking and therefore can be used to either enable alignment with Paris-specific targets or gauge the level of unmanaged transition risk in securities or portfolios that are misaligned. ITRs (or a proxy) are also available for many securities, via a number of different providers. On the downside, the complexity of the calculations behind ITR makes it less transparent and straightforward than some other metrics, including WACI.

It’s worth noting that while ITR incorporates a company’s emissions profile, including stated climate targets, many companies lack such a target and, therefore, almost inevitably have a high ITR. We view this as a positive: Having a high ITR and knowing that asset owners will use that metric puts pressure on companies to establish targets or end up being excluded from climate-aware benchmarks and potentially facing a reduction in asset-owner demand.

In addition, because ITRs are calculated relative to a company’s respective industry or region, comparisons across different industries or regions can be counterintuitive. For example, a traditional oil and gas company can have a similar ITR to a renewables-focused company. To offer one illustration, a British multinational oil and gas company has an ITR of 1.5°C (although this rises to 2.4°C if Scope 3 is included), which is similar to the ITR of a Spanish utility (1.45°C) with a significant renewable energy business.²

Step 2: Integrating the climate variable into the optimization process

Having selected our primary climate variable, we chose data sources for different asset classes, including MSCI’s ITR data for corporate credit and equity exposures (Scope 1 and 2). For sovereigns, we opted for an equivalent statistic from FTSE Russell that uses ITRs based on countries’ net-zero targets for 2030.³ Note that all analysis in this paper is in US dollars.

Setting the baseline for climate-aware allocations

Next, we used ITRs as a measure of climate sensitivity to determine what they could tell us about a traditional multi-asset portfolio from an efficient frontier perspective. We first compared an efficient frontier using our non-climate-aware 10-year CMAs (dark-blue line in Figure 1) to an efficient frontier that integrates our climate-aware 10-year CMAs (light-blue line in Figure 1). The optimizations were run on an unconstrained basis. The negative impact of climate risks on the CMAs pulls the efficient frontier downward (i.e., less return). The yellow dot in the chart represents the baseline portfolio (based on the risk level of a 60% equity/40% fixed income reference portfolio) on the regular efficient frontier, while the light-blue and purple dots indicate the risk and return trade-offs for matching the baseline portfolio’s return and risk, respectively, on the climate-aware efficient frontier.

Figure 1

What were the ITR implications? There was effectively no difference between the baseline and climate-integrated portfolios; each had an ITR of 2.5°C. This speaks to the difference between running an optimization exercise based on CMAs and running one based on integrating climate objectives, such as temperature alignment. With the former, the impact on asset class returns is fairly muted. As a result, the rank order preferences for asset classes are affected but changes in the underlying assets required to move from one allocation to the other, while not insignificant, are meaningfully smaller than those required when integrating climate objectives. In our optimization exercise, the differences between the baseline portfolio and those targeting the same return or the same risk on the climate-aware efficient frontier were fairly minor. The main trade-offs were from emerging market (EM) equities to Japanese and European equities and from Japanese government bonds to global inflation-linked bonds (Figure 2).

Given the relatively muted impact of CMAs in our optimization exercise, it might be reasonable to ask whether climate-aware CMAs are even worth considering. We think so, and would bear in mind that while the difference between climate-aware and traditional CMAs might be relatively muted over a 10-year window, we would expect much more significant shifts over the longer term. In addition, asset allocation generally needs to address two goals when it comes to aligning portfolios around climate: 1) avoiding climate risks (e.g., transition risk), not just in the medium term but also in the longer term, and 2) meeting decarbonization objectives, including exposure to climate opportunities. From a climate-awareness standpoint, one could focus only on the first of these goals, but incorporating the second as well provides a more holistic approach. The “sequencing” of these goals also warrants consideration. For example, a country engaged in a significant transition effort could face a greater drag on real GDP in the short term (and thus, lower CMAs) than a country with a less aggressive transition approach. However, the former country’s climate leadership may mean a lower ITR over time, and therefore overweighting that country’s assets may be the right decision when it comes to decarbonization objectives.

Figure 2

How low can the ITR go?

For asset owners who must meet commitments on emissions reductions and improve their carbon footprint, targeting a lower ITR over time could be critical. With that in mind, for the next phase of the optimization, we wanted to target allocations with comparable levels of risk versus the baseline but with the lowest possible ITR. There are limitations, after all, since at a certain level of ITR, a particular asset allocation will not be viable. For example, a majority of companies (60% of the MSCI ACWI Investable Market Index⁴ based on Scope 1 and 2 emissions as of 10 February 2023) are not aligned with an ITR of 2°C or below, as reflected in the ITRs of the traditional indices (dark-blue bars) shown in Figure 3. That said, ITRs will continue to come down over time as more companies set targets, and that will result in a larger investable universe.

We next ran optimizations at 2.4°C, 2.3°C, and 2.2°C (below this level, an asset allocation would not be viable, based on current targets, as noted above) while leaving the risk unchanged (Figure 4). As shown in Figure 5, the changes across asset classes were far larger than those shown in Figure 2. For the 2.4°C portfolio, while the impact on return was marginal, the reduction in ITR came at the cost of a greater change in the underlying assets — about 15% of the base — though we would expect that many asset owners would find this an acceptable trade-off for the relatively small impact on return. For the 2.3°C and 2.2°C optimizations, the impact on return was marginal too, but the changes in underlying assets were greater, reaching 28% and 49%, respectively.

Figure 3

Figure 4

From a pure ITR perspective, European bonds and equities were the biggest beneficiaries of the efforts to reduce portfolio-level ITR, while assets such as EM equities, EM debt, and high-yield bonds provided sources of funds (Figure 5). Another notable result was the reduction in US Treasuries, given their relatively high ITR (the US scores poorly on a variety of climate metrics and policies versus other G20 nations, though we would expect this to change as the climate provisions of the Inflation Reduction Act of 2022 are implemented). For asset owners, this is an example of the importance of considering the practical trade-offs of integrating climate considerations into a strategic asset allocation — in this case, a large reduction in an allocation to the key global risk-free asset. Ultimately, the exercise shows a poor trade-off between ITR reduction and asset changes resulting from top-down optimization, and clear limits to the exercise — in this case, at an ITR of 2.2°C.

Figure 5

Step 3: Getting more granular through SAA implementation

Given the optimization “floor” of 2.2°C, we wanted to explore how asset owners could make changes on the margin, while preserving as much of a potential investable asset universe as possible. One method we considered was the use of Paris-aligned indices in the implementation stage of the SAA process, merely as a proof of concept.

To test this concept, we chose a set of MSCI Paris-aligned indices, which were set up in response to European Commission criteria aimed at aligning the trajectory of index-level carbon emissions with the Paris Agreement objective of limiting the rise in the global temperature to under 2°C. The indices target a “self-decarbonization” rate of more than 10% per year and a >50% reduction in carbon intensity (versus the parent index). They seek to underweight high carbon emitters (based on Scope 1, 2, and 3 emissions), target companies with a high proportion of green revenues, and reduce exposure to physical risk arising from extreme weather events by at least 50%. In pursuing these objectives, the indices tilt toward companies benefiting from transition opportunities and away from those exposed to transition risks, while targeting a low tracking error relative to the parent index.

The biggest advantage of the Paris-aligned indices is that their ITRs typically come closer to a 1.5°C – 2°C scenario. For instance, the ITR of the MSCI ACWI Climate Paris Aligned Index was 1.9°C as of 31 December 2022 (Scope 1 and 2 emissions), versus 2.5°C for the MSCI ACWI. However, some of these indices (e.g., the MSCI EM Climate Paris Aligned Index and some credit indices) do not meet the 2°C and under threshold due to the composition of the parent indices (as mentioned earlier, the majority of companies in major indices — and therefore any “child” indices — have an ITR above 2°C, and optimization can only go so far in reducing an ITR from a high starting point) and the low tracking error of the Paris-aligned indices. One additional disadvantage is that there are currently no Paris-aligned indices for government bonds (creating such an index has proved to be an incredibly complex undertaking for data and index providers). We would also note a key difference relative to ITR is that the indices do not incorporate forward-looking commitments, so there is a sector/absolute-emissions bias to their tilts.

We also examined key risk metrics for the Paris-aligned ACWI and EM indices (Figure 6). Note the reduced breadth in both indices versus the parent indices: about two-thirds of the constituent members are lost, a significant trade-off. Also, the two indices differ in both tracking error and risk composition. The ACWI Paris-aligned index has a lower tracking error (1.43%), with the variance against the parent index dominated by industry risk (about 62%). The EM Paris-aligned index has a tracking error of 2.44%, with fundamental and specific risk outweighing other factors.⁵

Figure 6

Moving to the implementation, we shifted the allocations to a blend of 50% Paris-aligned indices and 50% traditional indices. This reduced the previously noted ITRs (2.4°C, 2.3°C, and 2.2°C) to 2.24°C, 2.16°C, and 2.07°C, respectively. If we shift to 100% allocations to the Paris-aligned indices, the ITRs fall to 2.08°C, 2.02°C, and 1.95°C, respectively.

Figure 7 highlights the sector impact of this implementation approach — in this case, for the 2.4°C optimization. While the results should be intuitive (e.g., energy and materials decrease), the degree of change is worth noting. Also, bear in mind that we do not have sector CMAs and therefore do not optimize at the sector level, meaning that the shifts in sector weights are driven by regional and country shifts. Given that sectors play a much larger role in the design of Paris-aligned indices (as indicated by the industry-level risk shown in Figure 6), we see significant shifts in sector weights when we use the indices in the implementation process. For instance, in our initial optimization exercise earlier in the paper (using our climate-aware CMAs), moving from a 2.5°C to a 2.2°C ITR reduced the allocation to information technology, which is a consequence of a lower allocation to the US. But when we included the Paris-aligned index in the process, the allocation to information technology actually increased, as the sector as a whole is more aligned to lower temperatures.

Figure 7

Does an implementation approach add value to top-down optimization? 

Based on our analysis, we think many asset owners will find that it is critical to pair a top-down optimization process (e.g., using ITRs) with an implementation layer. The main advantage of the implementation layer is the ability to minimize asset-class-level portfolio changes while targeting improved climate outcomes. However, implementation through climate indices like those we’ve discussed may come at the cost of tracking error and reduced breadth, as shown in Figure 6. Another trade-off illustrated through our risk decomposition exercise is higher exposure to industry/sector risk. As shown in Figure 8, the MSCI Paris-aligned indices have a much different sector footprint than their parent indices. Thus, one should keep in mind that using a different implementation approach from broad indices can cause specific sector (or style) skews.

Figure 8

To preserve the broadest appropriate investable universe, we would advocate for including as many sectors as would be comfortable for the asset owner, as well as using tools like SBTi (discussed earlier) and active engagement to find target companies in indices that traditionally have poor climate risk scores (e.g., energy, metals & mining, agriculture).

Final thoughts

For any asset owner who wants to incorporate a climate constraint in the SAA process, there will necessarily be risk/tracking error relative to a portfolio that doesn’t incorporate climate. The question is how much of this risk should come from top-down AA decisions and how much from bottom-up implementation (e.g., choice of index). We think most asset owners will want some of each, but choices about the right blend will vary.

In many cases, we believe decisions about implementation may ultimately play the most significant role in helping asset owners achieve climate goals. It will likely be challenging to pursue these goals on a purely passive basis, given some of the index-related issues we discussed earlier (e.g., some Paris-aligned indices do not meet the 2°C and under threshold), and we believe fundamental research and active management will be crucial. Finally, we would highlight a number of top-down and bottom-up recommendations:

Top down 

• Establish what climate alignment means. In our research, we have incorporated climate in two complementary ways: 1) incorporating climate-related assumptions to improve the accuracy of our CMAs and 2) taking a forward-looking view of climate goals the asset owner might want to achieve, such as the integration of a Paris-aligned transition scenario with specific temperature alignment goals.
• Benchmark selection can blend traditional indices and sustainable, green, or Paris-aligned indices, although these come with sector or style skews.
• Climate-aware investing is not simply a risk-avoidance exercise. To pursue alpha potential from the effects of a warming world, asset owners may want to lean in to climate-exposed regions, sectors, or industries with a tilt toward climate solutions or adaptation. We think this is an area ripe for thematic investing approaches.

Bottom up 

• We think asset owners should consider implementing a preference for companies with SBTi targets and/or other net-zero approaches, as this may provide maximum flexibility in reaching climate goals. However, asset owners should be aware of trade-offs, including very different sector exposures than in broad benchmarks.
• ITRs are useful for relative-value decision making for similarly situated companies. Implementation within government bonds is currently a challenge, given the limited availability of suitable building blocks and benchmarks.
• Exclusions are an option, but we would advocate for engagement with companies instead. Exclusions are based on backward-looking data and often target companies that are critical to the low-carbon transition and in the process of implementing targets to improve their business models’ alignment.
• While we have focused mainly on transition risk and climate-change mitigation, underinvestment in climate-change adaptation constitutes a major market inefficiency. We think adaptation solutions will be a significant opportunity with quite a different footprint than mitigation solutions (see the article below).

Finally, whatever the approach to implementation, key performance indicators should be monitored over time. Not only will this help to create a clear connection between the asset owner’s investments and its broader climate initiatives, but it also provides reporting transparency that can ensure accountability and foster dialogue about trade-offs. 

To read more, please click the download link below.

¹Scope 1 emissions include direct emissions from a company’s owned or controlled sources. Scope 2 emissions include indirect emissions from purchased or acquired energy. Scope 3 emissions include all indirect emissions that occur in the value chain of a reporting company. | ²Example is for illustrative purposes only and not intended as an investment recommendation. | ³November 2022, “The COP26 Net Zero Atlas,” FTSE Russell. | ⁴The MSCI ACWI Investable Market Index captures large-, mid- and small-cap representation across 23 developed markets and 24 emerging markets. With 9,126 constituents, the index covers approximately 99% of the global equity investment opportunity set. | ⁵Fundamental risk is risk associated with exposure to common risk factors like size, value, growth, etc. Specific risk is associated with unique characteristics of individual stocks.

Important disclosures: capital market assumptions 

Equities 
General — Assumed market returns are based on the Investment Strategy Group’s expectations for future dividend yield, earnings growth, and valuation change. Assumed volatility and correlations are based on historical analysis of the representative indices. Indices used are as follows:

• Global equities: MSCI AC World
• DM equities: MSCI World 
• EM equities: MSCI Emerging Markets 
• US large-cap equities: S&P 500 
• Non-US DM equities: MSCI EAFE 
• Europe equities: MSCI Europe 
• Japan equities: MSCI Japan

Bonds
General — Assumed risk and correlations based on historical analysis of the representative indices. High-quality sovereign bonds – Return assumptions are based on starting yields and the expectation that yields move toward our estimate of a terminal interest rate over the time period. Using these inputs and the duration of the respective bill, note, or bond, we then calculate the income and capital gains/losses associated with these changes. We assume zero downward adjustment for downgrades and defaults for high-quality sovereign bonds.

Credit risk premia — For non-sovereign and corporate bonds, excess return assumptions are estimated. The excess return assumption is a function of excess spread, a downward adjustment for downgrades and losses, and reversion to median spread levels. The excess spread is readily observable in market pricing. The downward adjustment for downgrades and defaults is based on our proprietary research and the long-term historical experience. Indices used are as follows:

• Global bonds: Bloomberg Global Aggregate (USD Hedged) 
• Global Treasuries: FTSE World Government Bond (USD Hedged) 
• Global inflation-linked bonds: Bloomberg Global Inflation Linked (USD Hedged) 
• Global corporate bonds: Bloomberg Global Corporate Index (USD Hedged) 
• Global high yield: Bloomberg Global High Yield Index (USD Hedged)
• EMD: JPM EMBI Global Diversified 
• US cash: US 3-month T-bill

Currencies
Return assumptions are shown for unhedged currency exposure, unless stated otherwise.

Hedged — Hedged currency return assumptions are based on current and forward-looking estimates for interest-rate differentials.

Unhedged — Unhedged currency return assumptions are formulated based on forward-looking estimates of real carry returns, normalization of real exchange rates, and an adjustment for productivity growth.

General
Period — Intermediate capital market assumptions reflect a period of approximately 10 years. If we developed expectations for different time periods, results shown would differ, perhaps significantly. Additionally, assumed annualized performance and results shown do not represent assumed performance for shorter periods (such as the one-year period) within the 10-year period, nor do they reflect our views of what we think may happen in other time periods besides the 10-year period. The annualized return represents our cumulative 10-year performance expectations annualized. The assumed returns shown do not reflect the potential for fluctuations and periods of negative performance.

This analysis is provided for illustrative purposes only. This material is not intended to constitute investment advice or an offer to sell, or the solicitation of an offer to purchase shares, strategies, or other securities. References to future returns are not promises or even estimates of actual returns a client may achieve. This material relies on assumptions that are based on historical performance and our expectations of the future. These return assumptions are forward-looking, hypothetical, and are not representative of any actual portfolio, or the results that an actual portfolio may achieve. Note that asset-class assumptions are market or beta only (i.e., they ignore the impact of active management, transaction costs, management fees, etc.). The expectations of future outcomes are based on subjective inputs (i.e., strategist/analyst judgment) and are subject to change without notice. As such, this analysis is subject to numerous limitations and biases and the use of alternative assumptions would yield different results. Expected return estimates are subject to uncertainty and error.

ACTUAL RESULTS MAY DIFFER SIGNIFICANTLY AND AN INVESTMENT CAN LOSE VALUE. Indices are unmanaged and used for illustrative purposes only. Investments cannot be made directly into an index. This illustration does not consider transaction costs, management fees, or other expenses. It also does not consider liquidity (unless otherwise stated), or the impact associated with actual trading. These elements, among others, associated with actual investing would impact the assumed returns and risks, and results would likely be lower (returns) and higher (risk). Any third-party data utilized in the analysis is believed to be reliable, but no assurance is being provided as to its accuracy or completeness.

Related content

• Climate from the allocator's perspective
• Integrating climate change into capital market assumptions: our approach and findings