Featured Image

Key Considerations for Stress Testing with Climate Risk Data

Risk, Performance, and Reporting

By Kristina Bratanova-Cvetanova  |  February 23, 2024

Stress tests have become an integral part of risk management in the last decade. They are now required in bank and investment fund regulations, derivative investments, and many other areas of the financial sector.

Stress tests are beneficial for risk managers because they:

  • Allow hypothetical scenarios, which may have never happened in the past and are not represented in standard risk analytics. This could include, for example, VaR and standard deviation, both of which are based on a historical period of time and the corresponding returns.

  • Provide snapshots of how portfolios will perform under shocks, which may or may not happen at the same time. This offers a multi-faceted view of portfolio performance under multiple scenarios or combinations of scenarios.

  • Provide investment decision-makers with insight into possible portfolio shocks and the subsequent potential for assets/groups of assets to lose or earn more in different scenarios.

Now there is a way to add the climate risk aspect to a stress test analysis, and below we do it in a couple of steps.

As a first step, we can compare the Europe broad equity index to its Climate Paris-aligned version. Climate Paris-aligned indices are built to represent resilient portfolios compatible with a 1.5ºC global warming climate scenario of the Paris Agreement as they transition to a net-zero economy. We selected one from a large benchmark provider.

For the analysis we also selected a few event-driven scenarios where the energy sector goes down by 20% in the US or Europe, as well as shocks replicating an increase in crude oil or natural gas of 30%. Our historical scenarios include periods where energy impact is notable or is the main driver for the shock:

  • Historical record high price of crude oil in July 2008

  • Oil price plunge between December 2015 and February 2016, reflecting the sustained excess of crude oil supply over global demand

  • The large price drop of crude oil from March 2020

  • The Russian invasion of Ukraine, which among other impacts also redirected flows of natural gas and crude oil supply

01-stress-test-percent-return-of-europe-equity-and-paris-aligned-europe-equity-index_updated Jan2024

The above chart illustrates that, under all scenarios and compared to the conventional version of the European index, the loss of the Paris-aligned index is lower and the return is close or slightly lower. This is related to the strategy of Paris-aligned indices to underweight high-impact sectors and companies in terms of carbon footprint.

Those stock returns would be most impacted under the energy-driven scenarios from the chart. Comparing the return of the conventional index to the Paris-aligned version helps identify that in the case of the above stress tests, the tilt towards companies with lower transition risks would lead to lower losses and slightly lower profits.

A second step is to sort and analyze groups of assets based on their adaptiveness to climate change (i.e., transition risk score). To do this, we applied Entelligent’s patented climate transition and non-localized physical risk data, known as T-Risk and available in FactSet Workstation, for Net Zero 2050.

  • A low T-Risk score indicates improved climate adjustments compared to peer companies

  • A high T-Risk score indicates high exposure (and higher risk) in a transition to a low-carbon economy

A negative T-Risk score indicates superior adjustments in a low-carbon, Paris-Aligned scenario (in which the global economy reduces carbon on a path to meet Paris Accord objectives) relative to Business-as-Usual, also known as Current Policies (in which the global economy keeps operating as it does now). For Carbon-Adjusted T- Risk Score, after estimating T-Risk at the security level, Entelligent adjusts the estimates by adding entity-level carbon footprint information, which improves the transition gamma of resultant T-Risk-based portfolios (also Gauss Rank-transformed).1

Companies classified with T-Risk scores are segmented into four categories—leaders, innovators, followers, and laggards—based on their performance.2

In the below stress test report in FactSet, the index is tilted toward leader companies (37%) and innovators (35%), while only 14% are allocated as followers and 12% laggards. The Paris-aligned equity index (referred to below as the Benchmark) has even higher allocation to leader companies (44%) and slightly lower in innovators (33%), and even lower to follower companies (8%), and laggard companies (11.9%).

On the top level we can also see the weighted average Carbon-Adjusted T-Risk Score of the Paris-aligned index shows a better transition risk profile compared to the standard version of the index. This corresponds to the strategy of the Paris-aligned index to select companies compatible with the 1.5ºC global warming climate scenario of the Paris Agreement as they transition to a net-zero economy.

02-stress-test-report-in-factset_updated Jan 2024

The plot below illustrates the Carbon-Adjusted T-Risk score for each company on the X axis and the active weight (portfolio weight – benchmark weight) on the Y axis. The companies with lowest Carbon-Adjusted T-Risk score would have the lowest climate transition risk. Dark blue dots are stocks in both equity indices, while light blue dots are included only in the Europe equity index.

There are two observations easily visualized in this plot:

  1. The Paris aligned index (dark blue dots) is allocated predominantly in companies with low Carbon-Adjusted T-Risk scores (green quadrant below) holding most of the dark dots from the plot. This indicates consistency with companies’ Carbon-Adjusted T-Risk scores.

  2. Stocks with lower T-Risk scores (and thus lower transition risk profiles, green quadrant) have negative active weights. This indicates they are overweighted in the Paris-aligned version, or scattered around 0, as they hold as much as the conventional index. For the companies with higher Carbon-Adjusted T-Risk scores, it is the other way round. Most of them have higher active weight, indicating they are either overweighted compared to the Paris aligned version or not included in it.


Source: Entelligent, FactSet

How does this relate to stress test performance? 

04-stress-test-updated_Jan 2024

Single-factor shocks

Above we see a few single-factor energy-related shocks: At the sector level, S&P Energy down by 20%, Europe Oil & Gas 20% down, Crude Oil 30% up, and Natural Gas 30% up. These are factor event-weighted scenarios from the FactSet Workstation.3

Under all scenarios with losses, the Percent Return column (return under scenario with no account of the portfolio weight) shows that leaders will lose consistently less than the rest of the groups across all scenarios. In addition, the loss of the Paris-aligned equity index is lower compared to the conventional version of the regional equity index.

Therefore, under these scenarios, selecting companies that are more resilient to transition risks could help reduce the possible losses. However, so do companies in the positive return scenarios, where leader stocks provide the lowest return when gas and oil prices increase. In a sense, leader stocks as a group appear to be less sensitive to changes in both directions.

Drilling down the analysis on asset level, we will show how to quickly identify stocks with the same T-Risk score—but with substantially different returns under a given stress test. Selecting stocks with lower loss or higher return and the same Carbon-Adjusted T-Risk score (dotted yellow line between the two) could help improve portfolio performance (in a scenario we feel is plausible) and preserve the longer-term transition risk profile. This could be done on the index/portfolio level or within a particular industry.

In the first chart below, we focus on the scenario of European oil and gas declining 20%. In that example, we could consider increasing the weight of a stock returning -5.5% (green outline) and reduce the weight of the ones with -27.5% loss (red outline), while preserving the T-Risk adjusted score at the index level, as both stocks have Carbon-Adjusted T-Risk of 0.1.

Alternatively, under the scenario we may select companies with the same return but with significantly different carbon adjusted T-Risk scores. This enables an improvement in the transition risk profile at the index/portfolio/industry levels while preserving the return under a given scenario. As an example, in the plot below we overweigh the stock with T-Risk score of -1 (green outline), while underweight the one with Carbon-Adjusted T-Risk score of 0.9 (red outline), as both provide a return of -13%.


Source: Entelligent, FactSet

The two examples below illustrate how to perform the same analysis within a given industry or sector.

In the plot below for the Financials sector, we easily spot companies with the same T-Risk adjusted score. For example, the companies with losses of -24% (red outline) and -6% (green outline).

By reducing the weight of the former and increasing the weight of the latter, we could stem the loss on industry and portfolio levels while preserving the same transition risk industry and portfolio profile.

When seeking to preserve the return and reduce the T-Risk score, we may consider a higher weighting for the company with a Carbon-Adjusted T-Risk score of -1 (green outline) and a lower weighting for the one with the score of 0.5 (red outline). That’s because both of them return -15% under this scenario (yellow vertical line).

Similarly, within the Health Care sector we may consider the stock that returns around -4% (green outline) to increase the weighting while decreasing the weight of the company with a loss of -20% (red outline), given both of them have a Carbon-Adjusted T-Risk score of -0.55 (yellow horizontal line).

If we want to preserve the return and improve the transition risk profile, we may overweight the stock with a carbon adjusted T-Risk score of -1 (green outline) and underweight the one with the score of 0.25 (red outline), as both of them return -10% under this scenario (vertical yellow line).


Source: Entelligent, FactSet

Extreme Event Stress Tests

Using extreme event stress tests in the FactSet Workstation4, we can replicate several historical scenarios: a record high price of crude oil in July 2008, an oil price plunge between December 2015 and February 2016, a large drop in the price of crude oil price since March 2020, and the Russian invasion of Ukraine in February 2022.


Source: Entelligent, FactSet

Across all of the event stress tests, the Paris-aligned equity index loss is less than the loss of the conventional equity index. We can see the loss for the leaders is consistently lower compared to the other groups. Thus, higher allocation to companies with lower T-Risk scores leads to lower loss and relatively better performance under replicated historical scenarios with energy prices and related shocks.

For the analysis on asset levels, we combine the two arguments from the above sections and show how to quickly identify stocks with better returns under a given scenario—and at the same time lower transition risk, as measured by T-Risk carbon adjusted score. This enables improved portfolio performance under a scenario we feel is plausible to happen and can help reduce the longer-term transition risk profile by overweighting stocks with lower loss/higher return and a lower T-Risk carbon adjusted score (green outline below) over stocks with higher loss/lower return and a higher transition risk (red outline area below).

This could be done at the index/portfolio level or within a particular industry. The first plot below highlights overweighting a stock with a return of -3% and a T-Risk carbon adjusted score of -0.9 (green circle) over one with a -19% loss and a Carbon-Adjusted T-Risk score of 0.5 (red circle). This would, if the oil price drop scenario repeats today, increase the T-Risk score at the portfolio level and reduce the expected loss.


Source: Entelligent, FactSet

The same analysis could be performed within a given industry or sector with two examples below, where we focus on the Materials and Consumer Discretionary sectors. Red areas depict companies with high transition risks and worse return under the selected scenario. Green areas outline companies with lower transition risks and lower losses under the event stress test.


Source: Entelligent, FactSet

As an example in the Materials plot above, we may opt to a) reduce the weight of the company with a T-Risk adjusted score of 0.5 and a return of -18% (red circle); and b) increase the weight of the company with T-Risk adjusted score of -0.50 and a return of -7% (green circle).

Among stocks in the Consumer Discretionary sector, we may consider the company with a Carbon-Adjusted T-Risk score of -0.6 and a loss of -5.5% (green circle) as a candidate to overweight. A candidate to underweight is the company with a Carbon-Adjusted T-Risk score of 0.35 and a loss of -17% (red circle). These considerations could improve the long-term transition profile and reduce the loss under the selected stress test.

In Conclusion

The examples above show one approach for incorporating climate transition risk data into your scenario analysis process. The perspective presented aims to show there is a way to analyze under different scenarios—historical or factor shocks, how would the portfolio and each of the Entelligent T-Risk score groups perform, and whether there are scenarios under which companies with higher or lower T-Risk score are more vulnerable to loss or lower return.

Based on this analysis and expectation for each scenario’s probability of realization, an investor might decide to reallocate the portfolio so that it performs better under a given scenario, improves its transition risk profile, or achieves both at the same time.

To learn more, visit Entelligent on the Open FactSet Marketplace.


This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.

1 EntelligentTop-down climate transition risk calculations, while valuable in assessing macroeconomic scenarios, have limitations. They work well in macroeconomic stress test environments that rely on aggregated data and broad economic indicators specific to energy prices and the energy-technology transition to a low-carbon economy. However, their effectiveness diminishes when applied to macro indicators that are unrelated to energy or some micro-level company assessments that are beyond energy dependency. Entelligent models primarily utilize climate and energy data, applying it downstream to publicly available financial data on companies. It may not, by design, fully account for the self-reporting benefits that can be exploited by companies to present a more favorable environmental image (e.g. short greenwashing). To address this limitation and provide greater accuracy, a double-verification system should be employed, involving financial practitioners who can evaluate a company's preparedness for climate and energy transitions. It is crucial to acknowledge that for many companies, there is substantial work required to align with climate and energy transition goals, and top-down calculations should be complemented by bottom-up analyses to create a more comprehensive and accurate risk assessment.

2 T-Risk (0/1/2/3: leader/innovator/follower/laggard) represents a quartile bin within each cohort. This intra-cohort ranking is consistent between T-Risk v2 and v3. However, the cohorts themselves are defined differently between the two versions. T-Risk v3: Cohorts are defined by 6 macro-regions (Asia/Pacific Ex Japan, Japan, North America, Latin America, Europe, Africa/Mideast ) and 32 RBICS L2 categories. T-Risk v2: Cohorts are defined by 6 macro-regions (Asia/Pacific Ex Japan, Japan, North America, Latin America, Europe, Africa/Mideast ) and 25 GICS Industry Groups. More information on T-Risk methodology is available in the T-Risk methodology white paper, which can be accessed for a detailed understanding of our classification process.

3 Factor event-weighted scenarios allow you to compare historical period changes to the supplied shock value saved with the stress test. Periods that have changes similar to the shock amount are weighted higher than non-similar periods. First you sort by the absolute value of the difference between historical change to the shock amount and then you exponentially weight the periods based on the saved decay rate.

4 Extreme Event Stress Tests take today's portfolio and hypothesizes what its return would be if an extreme event period were to happen again. Current factor exposures are used with factor returns from the event to derive the impact of this event on your portfolio returns.

Subscribe to FactSet Insight

Kristina Bratanova-Cvetanova

Ms. Kristina Bratanova-Cvetanova, CFA, is Senior Product Manager, ESG, Climate, Regulatory Risk, at FactSet, based in Sofia, Bulgaria. In this role, she is responsible for driving growth and development of regulatory risk solutions. Prior to FactSet, she spent over nine years at FinAnalytica in a few roles, most recently as a Head of Global Account Management and Client Solutions Director. Before joining FinAnalytica, she worked for three years at Financial Supervisory Commission analyzing the impact of regulatory framework on the market for capital market, pension, and insurance company sectors. Ms. Bratanova-Cvetanova earned a Master’s Degree in Finance and Banking and a Bachelor’s Degree in Economics from Sofia University St. Kliment Ohridski and is a CFA charterholder.


The information contained in this article is not investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.