Scientific Beta: How to build a more forward-looking climate change screen
![Erik Christiansen (photo archive Scientific Beta) Erik Christiansen (photo archive Scientific Beta)](/l/nl/library/download/urn:uuid:1165928f-0886-4169-b7d3-7ae4e726c45d/erik+christiansen+%28scientific+beta%29+980x600.jpg?scaleType=1&width=980&height=600)
By Erik Christiansen, Head of Investment Solutions, Scientific Beta
Measuring and controlling the carbon intensity of a portfolio is usually part of a climate investing strategy, and serves two main purposes:
- Monitoring carbon intensity and reducing it through a combination of investment and engagement decisions is a core part of investors' net zero commitments: what gets measured gets managed.
- According to the TCFD (Task Force on Climate-related Financial Disclosures), a portfolio’s carbon intensity remains a rough proxy for its exposure to climate transition risks, which many investors wish to lower, whether they believe these risks are adequately priced in by markets or not.
However, carbon intensity can be cut down in many ways, and the ‘how’ matters more than the ‘how much’.
The ‘how’ and the ‘how much’ of decarbonisation
The carbon intensity of a portfolio can be reduced in different ways. Recently the two main net zero coalitions, the Net Zero Asset Owner Alliance (NZAOA) and the Institutional Investors Group on Climate Change (IIGCC), have been working on guidance on how to better measure these different drivers of decarbonisation. This decomposition seeks to verify that the most relevant levers are at the source of the reductions displayed.
Moreover, in May 2023 the IIGCC published its recommendations on Enhancing the Quality of Net Zero Benchmarks, in which it argues that these indices ‘can have their greatest real economy impact when they adopt a process based on emissions reductions in the real economy, in other words, where the dominant influence comes from organic declines in emissions or allocating capital to industry best performers rather than ‘declines on paper’, which stem purely from weighting changes that result in inter-sector reallocations.’
This means that portfolio construction techniques that aim at real-world decarbonisation, not just at portfolio-level, should reduce the capital allocation to laggards within sectors, instead of pulling the financing from entire sectors, indiscriminately.
Measuring climate transition risks without black boxes
But how good is decarbonisation at reducing climate transition risks?
The task of the TCFD (2017) is to produce ‘recommendations for disclosing clear, comparable and consistent information about the risks and opportunities presented by climate change’ and Weighted Average Carbon Intensity (WACI) remains their primary, specific disclosure recommendation for investors. Nevertheless, in this risk perspective, WACI has two main limitations:
- WACI relies on emissions that have already occurred, several months or even a couple of years ago. Current emission intensities are unlikely to change radically in the short term, but the closer one gets to the 2050 horizon built into net zero frameworks, the less the recent past will be a guide to the distant future. While a company’s carbon track record is certainly of interest to investors, past performance is no guarantee of future results, as they are often reminded of.
- Carbon intensity may in aggregate, at portfolio level, be indicative of companies’ transition risk exposures: increased carbon prices would presumably cost high emitters more than efficient ones. But it provides less insights at individual stock level: a manufacturer of air conditioners may have the same carbon intensity as one building windmills, but while the former might actually benefit from unbridled global warming, the latter’s so-called climate solutions would stand to benefit from tighter climate regulations.
Academics, investors, and data providers have therefore sought to create more forward-looking climate transition metrics. In this search, three broad categories of approaches can be distinguished, which we term the Crystal Ball, the Rule of Thumb and the Wisdom of Crowds approaches.
Crystal Ball
One common approach is to build a proprietary forecasting model. As a first step, such models typically involve estimating companies’ short-term and long-term carbon trajectories and comparing these to benchmark trajectories for different technologies, sectors, or regions. However, projecting a company’s emissions long into the future is fraught with uncertainty, and, to top it off, there are a multitude of conflicting benchmark scenarios to choose from.
Moreover, this first step of assessing whether companies align themselves on different transition scenarios only partly answers the objective of having a risk estimate.
Moving from alignment to risk assessments adds a second layer of uncertainty, most notably when estimating and incorporating the future costs of carbon. At the end of the day, somewhat disappointingly, the level of convergence across providers for climate risk metrics is no more reassuring than the substantial divergence observed between more qualitative, not to say subjective, ESG scores, which we analysed in a recent paper (Can we make ESG scores great again? Scientific Beta, 2024).
Rule of Thumb
In their comparison of different crystal balls, Bingler et al. (2021) find that ‘convergence between metrics is significantly higher for the firms most exposed to transition risk’ and that ‘the risk distribution in the energy sector is significantly higher compared to the other sectors.’ This conclusion is not earth-shattering. Few doubt that companies producing fossil fuels are most at risk if humankind decides to transition away from carbon-rich energy: that is a tautology baked into many such risk models.
To address such concerns, investors can more straightforwardly rely on fossil fuel divestment policies, especially related to the two least carbon efficient forms of fossil fuels, namely coal and tar sands. These fossil fuel screens need to evolve over time since what is at stake is a transition, not a disruption. For example, at Scientific Beta we apply a gradual year-by-year phase out of coal-fuelled electricity by 2030 in developed markets, and 2040 globally.
Wisdom of Crowds
An alternative to relying on the wisdom of expert models, is to exploit the wisdom of crowds. More precisely, the idea is to extract from market prices, which represent a sort of consensus between a multitude of investors, relevant information on which companies are deemed to be hurt or to benefit from shifts in climate transition risks. Two types of such market-based climate risk models have been proposed:
- Maeso and O’Kane (2023) apply a technique called sentiment analysis, that is measuring how a company’s stock price reacts to climate related news, testing a variety of language models, and gathering text from several high-quality newspapers.
- Others have proposed to measure stock price sensitivities to climate risks – often called climate betas – by regressing a stock’s returns on a long/short portfolio of high-risk versus low-risk stocks. For example, Görgen et al. (2020) propose to build the long/short portfolios by ranking stocks on ten variables related to their current emission intensities, the public perception of these and their policies towards them. While most of the variables are composite scores sourced from four different providers, the rankings remain primarily driven by current carbon intensity inputs, which are assigned fully 70% of the weighting in the final ranking.
At Scientific Beta we use a similar factor-based approach to compute our Climate Transition Risk Beta (CTR Beta), but we use only carbon intensity inputs at stock level, refraining from relying on proprietary climate scores, which are opaque and divergent across providers, as shown in our paper on ESG scores.
We rank stocks based on their emissions intensity within sectors. However, because exposures to climate transition risks vary widely from sector to sector, we apply a sector classification system that is used for example in the European Central Bank’s climate stress test framework.
In the other sectors, where climate policies are not deemed relevant, i.e. those less exposed to climate transition risks, stock prices are presumably too noisy to effectively capture small differences in company risk exposures. These sectors are therefore not included when defining the long/short portfolios against which stocks are regressed.
The resulting CTR Betas are expected to reflect the market wisdom on each company’s exposure to the climate transition: a CTR Beta above zero means an above average risk estimate and since investors determine stock prices from their projections of future cash flows and the riskiness of these, the CTR Betas are forward-looking.