Carbon Emissions Decomposition
Investors are increasingly looking to manage the carbon emissions (which represents the greenhouse gas emissions) of their portfolios (portfolio emissions thereafter) to meet new regulatory and stakeholders’ expectations. Since the Paris Agreement (2015), many initiatives have emerged to structure these practices. On the one hand, regulators are standardizing the extra-financial information that financial and non-financial companies must communicate (e.g., in EU through the sustainable finance disclosure regulation). On the other hand, collective investor initiatives are proposing frameworks to align portfolios with climate objectives (e.g., the Net-Zero Asset Owner Alliance and the Paris Aligned Investment Initiative). While early requirements have focused on reporting, frameworks increasingly include target setting related to emissions performance metrics, including absolute and asset-level value chain emissions. However, popular climate metrics, either assessing current or forward-looking performance, give investors limited understanding and control of the factors influencing the evolution of portfolio emissions over time in the absolute or relative to other portfolios.
Over the last 25 years, research in environmental economics has been confronted by similar issues at the macroeconomic level and has developed decomposition methods to understand the different drivers of global emissions. One of the most famous is the “Kaya identity” that expresses global emissions as the product of four factors: population, GDP per capita, energy intensity, and carbon intensity [1].
We propose to adapt these decomposition methods to the variables of interest and factors relevant for analyzing and controlling equity portfolio emissions. The proposed decomposition model makes it possible to distinguish between five factors that influence these emissions: sector allocation (weight of a given sector in portfolio), intra-sectoral allocation (weight of a stock in the sector), emissions intensity of the firms (expressed as tons of CO2e per million dollars of sales), sales and market capitalization. The model can be applied to analyze the different performance metrics recommended by the regulator and investor initiatives (intensity, footprint, or absolute emissions), and can be used for both historical and cross-sectional analysis.
By revealing the factors underlying the decarbonization of a portfolio, this framework allows investors to control the extent to which emissions and emissions trends are explained by i) sector biases, which will potentially increase the portfolio’s active risk with arguably a limited effect on climate mitigation [2], or ii) by selection of companies within sectors with lower and structurally decreasing emissions. This makes it possible to gain a more qualitative view into the decarbonization of a portfolio, and hence to limit the risk of “portfolio greenwashing” in the sense of Amenc et al. [3].
A Decomposition Method for Portfolio Emissions
Theoretical principles of the index decomposition analysis
A series of works on environmental economics identify the drivers of observed changes on environmental variables such as energy consumption or emissions. The decomposition methods used in environmental economics have origins in index number theory, which is about decomposing the difference of an aggregated value (e.g., the value of a basket of products) over a period into two factors: a price factor (e.g., consumer price index) and a quantity factor. Methods used in environmental economics are extensions of index number theory methods to more than two factors [4]. This sub-section introduces the index decomposition analysis (IDA) principles as it is the most easily applicable method for a financial portfolio.
Let us consider as an aggregate value of sub-categories (e.g., the sum of emissions associated with different instruments), and factors contributing to these sub-categories’ emissions (e.g. weight in portfolio, carbon intensity, sales, etc.).
The goal of IDA is to understand historical aggregate change from to as an (1) additive or (2) multiplicative operation between factors.
The reasoning behind the IDA is to derive the aggregate value formula over time and isolate the contribution of the factors. As developed in Ang [5], the additional effects of the factor is given by
where and is the emissions of the instrument (the proof is shown in Appendix A).
The reason to choose the additive or multiplicative method is a matter of presentation and it is possible to shift from one to the other. Additive decomposition is preferred for aggregate analysis of multi-year periods, while multiplicative decomposition is preferred to identify changes in trends.
The identity for absolute emissions is
The factors can be grouped into the following three categories:
Portfolio management choices: The portfolio manager can have a direct impact on the two following factors: , the sector allocation (weight of sector in portfolio ), and the intra-sectoral allocation (weight of holding in sector ). The control of the sector allocation factor is particularly important. Firstly, from a climate impact point of view, it can be seen to some extent as an artifice, allowing to reduce emissions simply by reducing exposure to emissive sectors. Secondly, from a financial risk perspective, inappropriate sectoral deviations from a benchmark can cause large tracking errors.
Emissions intensity: Companies can reduce their emissions intensity on the three scopes: (scope 1), (scope 2), and (scope 3). The selection of companies that improve their intensity can be the result of portfolio management choices. In the rest of the article, scope 1+2 emissions are reported emissions when available (either directly by companies or through the Carbon Disclosure Project) while scope 3 emissions are systematically estimated emissions (estimates from Bloomberg). This choice is motivated by the lack of comparability of scope 3 reported emissions ([6]) and by our goal to provide historical analysis. Historical analysis requires consistent data over time; however, the coverage of companies reporting scope 3 data reported in 2014 was very low and the reporting methodologies have changed significantly since then. Estimated data is more stable because it is adjusted in line with methodological developments and was therefore preferred.
Economic factors: These factors provide a link between relative emissions (intensity or footprint) and absolute emissions. As discussed in the previous section, more and more frameworks recommend setting absolute targets and monitoring the absolute emissions of a portfolio. Environmental economics have confirmed that demographic and economic factors (population and growth of GDP per capita) are the main drivers of global emissions. It is therefore essential to analyze absolute emissions and to understand the economic factors that influence them.
: The absolute emissions of a company could be explained from a change in its emissions intensity or its sales.
: the share of emissions attributed to an investor depends on the market value of a company (for a fixed amount invested in the company). Capital market volatility can contribute to significant short-term effects.
: as the market value of a portfolio increases, the investor's responsibility in terms of emissions increases.
Cross-sectional and historical analysis
Although emissions decomposition methods are mainly used for historical analyses in environmental economics, they can also be used for cross-sectional (e.g., between two countries) and forward-looking (e.g., between two energy scenarios) analyses [7]. For an equity portfolio, it may also be relevant to carry out a historical decomposition together with a cross-sectional decomposition, for example against a benchmark. Depending on this choice and the type of metric analyzed, only some factors will influence the decomposition model (Exhibit 1).
The cross-sectional decomposition is only influenced by two factors: sector allocation and intra-sectoral allocation. Therefore, the identity for carbon emissions decomposition on the Scientific Portfolio platform is
The performance of a portfolio, whether financial or non-financial, can be assessed in relation to a reference. For example, the BMR requires for an index to be considered as “Paris-aligned” that its “GHG [Greenhouse gas] intensity or, where applicable, absolute GHG emissions […], including scope 1, 2 and 3 GHG emissions, shall be at least 50 % lower than the GHG intensity or absolute GHG emissions of the investable universe" (from the Commission delegated regulation (EU) 2020/1818).
Conclusion
In response to the growing expectations of regulators and stakeholders for investors to manage the carbon emissions associated with their portfolios, we propose a method for decomposing these emissions. This method, inspired by those used in environmental economics, allows for cross-sectional analysis (of a portfolio compared to a benchmark at a given time) and historical analysis.
The model proposed enables us to distinguish the effect of three types of factors: i) those linked to the portfolio manager's allocations, ii) those linked to the emissions intensity of the companies, and iii) economic factors (sales and market capitalization) that influence absolute emissions. Understanding the role of each factor makes it possible to control whether a portfolio's emissions have been reduced effectively, i.e., without relying solely on sector allocation and therefore to avoid the risk of “portfolio greenwashing”.
This approach is thus complementary to existing forward-looking approaches. Metrics such as implied temperature rise may be relevant for communication but are limited for managing emissions associated with a portfolio [8]. It is therefore important to separate forward-looking methods for setting long-term alignment targets, including carbon emission reduction targets, from methods for assessing and monitoring the achievement of these targets. The decomposition method is part of this assessment objective and, to be as relevant as possible, it is necessary to have previously defined targets at the portfolio level that are compatible with global emissions pathways.
References
1 Y. Kaya. (1990). Impact of Carbon Dioxide Emission Control on GNP Growth: Interpretation of Proposed Scenarios. Paper presented to the IPCC Energy and Industry Subgroup, Response Strategies Working Group, Paris, France.
2 Alex Edmans and Doron Levit and Jan Schneemeier. (2022). Socially Responsible Divestment. SSRN Electronic Journal. 10.2139/ssrn.4093518.
3 Noël Amenc and Felix Goltz and Victor Liu. (2022). Doing Good or Feeling Good? Detecting Greenwashing in Climate Investing. The Journal of Impact and ESG Investing. 2(4): 57-68. 10.3905/jesg.2022.1.045
4 Paul de Boer and J.F.D. Rodrigues. (2020). Decomposition analysis: when to use which method?. Economic Systems Research. 32(1): 1-28. 10.1080/09535314.2019.1652571.
5 B. W. Ang. (2015). LMDI decomposition approach: A guide for implementation. Energy Policy. 86(undefined): 233-238. 10.1016/j.enpol.2015.07.007.
6 Frédéric Ducoulombier and Victor Liu. (2021). Carbon Intensity Bumps on the Way to Net Zero. The Journal of Impact and ESG Investing. 1(3). 10.3905/jesg.2021.1.013.
7 B. W. Ang and Tian Goh. (2019). Index decomposition analysis for comparing emission scenarios: Applications and challenges. Energy Economics. 83(undefined): 74-87. 10.1016/j.eneco.2019.06.013.
8 PRI. (2021). Forward Looking Climate Metrics. Discussion Paper, Principles for Responsible Investment.