Asset owners want and need to understand the potential effects of climate change on their portfolios. Traditional investment analysts spend their days examining the risks and opportunities that can change the value of an investment. But until recently, it has been difficult to gauge the potential impacts of climate change on a company’s valuation or portfolio performance. Climate change scenario analysis can potentially help investors understand the effects of climate change on physical assets, as well as the opportunities and challenges that a company might face as it transitions to a lower-carbon economy.
But this is a nascent undertaking, and growing pains are everywhere. Each provider has a unique approach to the data and analysis, with different strengths and weaknesses. Over time, there will likely be some convergence as regulators, clients, beneficiaries and other stakeholders demand that investors report on portfolio climate risk and opportunity exposure. It’s a daunting task for asset managers.
Academic and Investor Collaboration
AllianceBernstein (AB) has formed an ongoing partnership with Columbia University’s Earth Institute. Our team collaborated with the Earth Institute to create and deliver climate risk training to internal AB investors and stakeholders. Next, those investors and stakeholders worked with the Earth Institute to produce our first Taskforce on Climate-Related Financial Disclosure (TCFD)-aligned climate statement. As a signatory to the UN Principles for Responsible Investment (PRI), AB will eventually be required to implement scenario analysis for our investments and operations, and we believe that our partnership with the Earth Institute will benefit that analysis greatly.
Scenario analysis, as defined by the TCFD, is an exercise to help understand how the physical and transition risks and opportunities of climate change might affect a business over time. Physical risk measures the cost to a company’s physical assets due to climate change. This might include factories endangered by coastal flooding, the impact of temperature extremes on infrastructure, increased major storms or other tangible damage. Transition risk measures both the effect of policies such as the cost of emissions-reduction requirements, as well as opportunities that might arise from transitioning to a low-carbon economy, such as newly developed low-carbon technologies or other services or products that could help society mitigate or adapt to a changing climate.
AB is working to understand, assimilate and, over time, improve climate scenario analysis. Building meaningful climate-change scenario analysis for reporting and investment decision-making is a process that will be continuously refined as it becomes more commonplace.
Our analysis of the various providers and their offerings stems in part from our in-house experience. Specifically, AB has a proprietary climate-change-scenario model that is used to manage a strategy of Australian equities based on the ASX 200 Index. Our experience with this model—and the massive resources needed to appropriately analyze climate change scenarios for those 200 stocks—informed our evaluation of external scenario-analysis providers whose output we can combine with the insights of active portfolio management for larger investment universes.
AB asked 11 providers to respond to a Request for Proposal. Firms and their climate-change models were initially judged on nine criteria (Display).
Each provider had gaps in asset-class and market coverage. For example, some firms cover only developed markets, some also offer emerging markets, and none offers comprehensive frontier-market coverage. Some providers cover equity only, and others cover corporate or sovereign debt with or without equity. No firm covered options, and coverage of other asset classes was limited or nonexistent. Though it is still a very new industry, we have already seen providers attempting to close gaps through acquisition and consolidation, a trend that we expect will continue.
The next hurdle is the data used in each model. Data are provided by the companies being analyzed and compiled by climate change-model providers. Since most companies are not required to provide climate disclosure—and there are no specific standards—the data set is incomplete and inconsistent from the outset.
The depth of disclosure also varies: Is it data for one business unit or several business units? Plus, disclosing companies unitize their data in different ways—or they may provide only highlights of key drivers rather than complete information.
To compensate for missing data, providers either use third-party data or apply complex assumptions based on country, sector and industry trends. The assumptions are often not disclosed and can vary wildly. Clean, standardized and more thorough data are necessary to improve the relevance of climate-change models; where data are not available, assumptions should be disclosed.
Finally, the scenario-analysis models are complex, but many need to be more sophisticated to provide reliable output. For example, the models measure risk in different ways. Some look at a company’s business by sector and location, and then use generalized information to translate that into the potential impact. Others go further, looking beyond simple industry or geographic exposures into the businesses themselves, accounting for how current financials might be affected by climate change—slimmer margins, for example, coming from higher costs that include carbon taxes, and even how much mitigation work might already have been done.
The depth of analysis also varies widely. Obviously, analyzing the climate change risks and opportunities for a multinational company is complicated. Different jurisdictions have implemented different carbon prices, taxes, markets or reduction goals, which theoretically should be applied pro rata to a multinational company’s assets within those jurisdictions. Yet capturing that information depends on clarity and detail in corporate disclosures, which is often lacking. Since climate change-model providers have limited resources to perform in-depth issuer-level research and verification, our research suggests that the application of already-compromised data can be indiscriminate.
Comparing outputs of competing model providers was also difficult. Some providers use a quantitative measure of a company’s value at risk from climate change, a few offer a risk level (low, medium, high), and still others give a score that allows comparisons between companies, but without much context. This variety in metrics reflects the nascency of climate scenario analysis intended for investors and will likely evolve over time as providers and their clients increase their mutual understanding of, and familiarity with, use cases through practical experience and decision-making.
So which model is best? That is not an easy question to answer, as the selection is highly dependent on the needs and nuances of the end user. From the group of 11 that submitted RFPs, we short-listed four companies and compared their product offerings across eight critical categories (Display). Each had its strengths, and our eventual choice came down to a best fit for our needs, combined with the potential to collaborate with the provider to enhance the analysis for our purposes as the industry matures and changes.
During the second phase of our analysis, each provider was asked to evaluate a fixed income, equity and multi-asset portfolio to assess its exposure to climate-change risk and opportunity. Ratings for the four providers were converted to standardized scale for ease of comparison, and each provider supplied ratings for physical risks as well as net transition risks for each company. The data emphasized the challenges presented by this type of modeling.
While the total portfolio assessments were quite similar, the scoring for individual positions and risks varied wildly, highlighting how the models captured different levels of detail. In our review, for example, many companies scored extremely strong in one provider’s assessment but were judged to be exceptionally weak by another.
So why were overall portfolio assessments similar? Dilution. Despite large differences in individual investment scores, when combined based on their position size within the portfolio, outlying scores are muted. Still, the individual investment scores are of utmost importance in the context of portfolio management and construction and can make a significant difference in portfolio returns. We believe that this highlights the advantage of active investment managers who not only know their securities well but also are aware of the quirks and idiosyncrasies of the scenario-analysis model that they use in portfolio and investment decisions.
Case Study: First Quantum
Perhaps the best way to understand differences in climate models is to contrast and compare the scores for one company across providers. First Quantum, a Toronto-based company operating copper and gold mines around the world with significant assets in Zambia, received meaningfully different scenario-analysis results from two providers. These results highlight the differences between providers and their underlying data and assumptions, as well as the necessity to be able to critically analyze scenario-analysis outcomes—something that only active managers with intimate knowledge of their portfolio holdings can provide, in our view.
Provider A considered 12 company sites when evaluating the physical risks of climate change for First Quantum. It determined that the company would benefit by 0.22% in extreme cold but be harmed by 2.21% in extreme heat, for a net physical value at risk of –1.96%. Provider A’s model assigned no effect to the company from extreme precipitation, heavy snowfall, severe wind, coastal flooding or tropical cyclones.
In contrast, Provider C’s model measured physical risks of coastal flooding (–1.13%), river flooding (–0.70%) and chronic impacts (–0.28%). These are slow-onset impacts, such as those observed changes in cost and productivity of labor under increased heat stress, for example. It also included a benefit from adaptation (+0.55%), for a net physical value at risk of –1.56%. Even though each provider approached the problem from a different perspective and scored different factors, the net value at risk was moderately comparable.
However, our portfolio managers, analysts and ESG team noted that both providers failed to include in their analysis the ongoing drought in Zambia. Approximately 90% of Zambia’s power generation is hydroelectric, and there is routine power rationing that will only get worse as drought continues. This is a major, and overlooked, threat to the company’s operations in that country.
Net transition risk calculations exposed significant differences. Both providers included direct greenhouse gas emissions from company-owned power sources (Scope 1 emissions), but only one provider included emissions from purchased power sources (Scope 2 emissions) and downstream emissions in the company’s value chain (partial Scope 3). For most companies, the bulk of greenhouse gas emissions comes from Scope 3.
Provider C, which gave First Quantum a total transition risk score of –62.81%, broke down the score among various elements of its transition risk factors, such as abatement, scope emissions and demand destruction and response. Meanwhile, Provider A assigned a single total transition risk score of –72.35%, which conveyed the underlying elements of operations and business model risk, scope emissions and “one price globally rising over time” for those emissions but did not break out the scores in these areas or how they contributed to the total transition risk score.
The two models are also less aligned when measuring opportunities from climate change. Provider A looks specifically at company patents to measure future opportunities, leaving patent-less First Quantum without potential benefits from climate change. Provider C’s valuation, in contrast, includes an expected +10.7% for increased demand for copper as the transition to electric vehicles continues and assumes +65.6% carbon cost passthrough for a total transition opportunity of +76.3%. When comparing the totals for net transition risk, Provider C assigns a net value at risk of +11.9%, versus –2.0% from Provider A’s more limited opportunity assessment. An active manager with fundamental knowledge of the company is likely able to reconcile these two evaluations to come up with a more appropriate appraisal of transition risk and opportunity.
The Importance of Personal Insight: Woolworths
We also studied the ratings from scenario-analysis providers versus our previously mentioned proprietary scenario analysis for Woolworths, an Australian grocery and supermarket retailer. Our team estimates that the physical risks from supply-chain disruption will be much worse than the estimates from third-party scenario-analysis models. In fact, the AB physical risk rating ranked High as opposed to Low for all four providers. One of our specific concerns is the potential for climate change to affect food supply, leading to high food inflation, which, in turn, could cause the government to limit the return on invested capital for grocers in order to protect consumers.
On the other hand, the scenario-analysis providers all ranked the transition risk for Woolworths as Medium to High, judging that the majority of carbon exposure for most companies comes from Scope 3 fossil fuel emissions (i.e., within a company’s supply chain, customers, etc.). Looking through the Scope 3 lens, a grocery chain’s massive number of suppliers and customers would cause a higher-than-normal value at risk. What those models did not include was the transition opportunity of more environmentally conscious consumers seeking more climate-friendly products, services and places to shop, which caused AB to rate the transition climate change risk for Woolworths as Low.
The Road Forward for Scenario-Analysis Models
These examples show that identifying and quantifying climate risk and opportunity through scenario analysis is difficult work. Trying to do so for an entire portfolio is fraught with even more challenges and opportunities for improvement. But scenario analysis is in its infancy, so substantial change is likely as data and models both become more sophisticated.
The push for better data is real. More countries and institutions have begun to require the inclusion of climate-change data and scenario analysis into public disclosures and reporting. UK regulators, for example, are phasing in this reporting, and it is expected to be required for UK-listed companies, pensions and investment managers by 2025. Australia and New Zealand already require asset owners to provide the information. And the same data are also starting to be required for corporate issuers. While the PRI won’t score signatories on compliance with this recommended practice for the next year or so, we believe that it is coming.
AB is combining the detailed company and sector knowledge of our analysts as well as our extensive issuer engagement on climate issues with the climate risk data available from third-party providers. We believe that this combination should generate better insight than relying solely on external climate-change data and should also provide stronger investment decisions and reporting to clients.
Understanding the discrepancies in this first generation of commercially available scenario analyses is the first step in developing better tools. We expect substantial improvements to models and data as asset owners press their investment managers for reporting that is significant, relevant and incorporated into investment analysis. Firms that actively engage early in this process can influence and shape the development of better models and can better prepare their portfolios and investors for the long-term risks and opportunities that issuers face from climate change in the years to come.