1 Introduction

In this article I explain how climate risk impacts traditional financial risk categories. I explore this impact on the price of financial assets (market risk), on bonds and loan commitments (credit risk) and on the ability of companies to refinance themselves (liquidity risk). Rather than representing a new type of risk, I consider climate risk via new risk drivers within existing risk categories.

There are specificities that need to be taken into account when it comes to climate risk. The novel nature of climate risk leads to the use of forward-looking scenarios and to projecting variables on long-term horizons. Doing so, it departs from traditional historical risk analysis and from the usual traders/asset managers’ investment horizons. The aim is to answer questions such as ‘if this scenario happens in the future, what is the consequence?’ In such an analysis, typically mid-term means 2030-2050, long-term means 2100. For each scenario, climate risk stems from a trade-off between non-stationary (evolving with time) physical risk drivers and transition risk drivers. Finally the revaluation of massive amounts of stranded assets poses new challenges. Climate analysis also identifies non-linearities when thresholds are crossed, leading to a series of cascading effects, the so-called tipping points. 

As a consequence, even though the general risk landscape is familiar, market infrastructure as well as internal and regulatory supervisory frameworks, need to evolve further to capture these specificities. New models, new datasets and new tools emerge to support management of climate risk in finance. This includes for instance Merton-like models such as the CERM for credit risk. I will address this topic more specifically in future articles. It also includes Atoti+ existing modules for climate risk monitoring. Here I discuss where and how Atoti+ is fit for purpose.

2 A New risk paradigm

2.1 Scenario Modelling

Forward-looking climate scenarios explore the many possible futures over the 21st century under changing assumptions and provide various more or less granular output variables. They rely on IPCC reports, on Integrated Assessment Models (e.g. REMIND from the Potsdam Institute for Climate Impact Research PIK) and on macroeconomic models (e.g. NiGEM used for economic forecasting). Climate scenarios reflect the trade-off between physical risk and transition risk and incorporate standardized sets of socio-economic assumptions (GDP, population, urbanization…). The Network for Greening the Financial System (NGFS, a group of 100 central banks and supervisors) or the International Energy Agency (IEA) provide databases of scenarios that can be used by financial participants. Scenario modelling is now a tool broadly used in climate finance.


Description automatically generated

Figure 1 – The NGFS scenarios – Scenarios are indicated with bubbles and positioned according to their transition and physical risks

I’m using the pyam module and the python API to integrate data series into atoti/Atoti+. Selected models, scenarios, variables with their temporal and spatial dimensions are then available for exploration. This is extremely useful to derive new measures, for instance to calibrate models and sensitivities to risk drivers. Pyam and atoti both come with free licenses so you can easily test this integration on your own.

2.2 Physical risk drivers

Physical risk relates to direct climate change impacts, either from acute hazards such as floods, storms and wildfires or from chronic hazards such as average rain or temperature changes. Analyzing physical risk requires a bucketing of assets in terms of geography, as well as data from geo scientists. This risk grows with time, and is ultimately controlled beyond 2050 under scenarios where we meet the ambitious Paris agreement targets without breaching tipping points (see 1.2.5) or if benefitting from massive negative emissions technology. Protecting assets from physical risk is called adaptation.

As a first step, one can start with ND-Gain index datasets to explore physical risk and build maps in atoti.

2.3 Transition risk drivers

Transition risk relates to disruptive changes required to decarbonize. To reach the Paris goal, we have to go from circa 55 gigatonnes of CO2 equivalent per year to 0 within the next 30 years. This implies massive transformations across all areas of the economy leading to massive value adjustments. This risk has very material impacts on the 2020-2050 time window. A proxy of exposure to transition risk is the carbon footprint, traditionally sub-divided into Scope 1 (direct emissions), Scope 2 (emissions via electricity used) and Scope 3 (value chain emissions). Protecting assets from transition risk is called mitigation.

Figure 2 – The 6 scenarios from the NGFS in Atoti+ – CO2 emissions MtCO2e/yr. Data series are integrated via the python API using the pyammodule.

One can start with the CDP GHG Emissions dataset to build dashboards and heatmaps in atoti. In France Scope 1 and 2 data are available from non-financial disclosures. Scope 1, 2 and 3 advanced data sets  are available from specialized providers such as Carbon4 Finance or MSCI.

2.4 Stranded Assets

The International Energy Agency defines stranded assets as investments which are made but which, at some time prior to the end of their economic life, are no longer able to generate an economic return as a result of changes in the market and to the regulatory environment. Reaching decarbonation goals requires keeping a large proportion of existing fossil fuel (oil, gas and coal) reserves in the ground, impacting related assets’ impairments (IFRS 9 measures). CECL and IFRS9 impairments as well as related Stages and changes between period can be consolidated and analysed in Atoti+. 


Description automatically generated

Figure 3 – Impairments Analysis in Atoti+

2.5 Tipping Points

tipping point in the climate system is a threshold that, when exceeded, leads to large and often irreversible changes in the state of the system. Tipping points have been identified in the physical climate system and in ecosystems, which will have severe and irreversible impacts on emissions when crossed. Well known examples are Permafrost unfreezing soil and the West Antarctic ice sheet collapse. It is very difficult to assign a probability to such events or to properly model related domino effects. Because of the tipping points, orderly (starting now and progressing smoothly) and disorderly (beginning late and abruptly) scenarios to net zero may have very different consequences.

3 Risk Types

3.1 Market Risk

Market risk is the risk of loss due to movements in market variables. It occurs when the aggregated value of a group of assets falls and/or the aggregated value of liabilities rise. It arises from large, sudden, positive/negative price adjustments via any asset class such as interest rates, foreign exchange, equity prices, commodity prices, credit spreads, implied volatilities, asset correlations, etc.…

In the context of climate change, the forward-looking consideration of non-stationary physical and transition risks brings new information about the cost of risk and the value of real and financial assets, resulting in potential price shocks, with an increase in market volatility and a sharp breakdown in correlations. For instance, price shocks may stem in a disorderly transition to a low-carbon economy, via physical climate events or via shifts in supply and demand, which may then impact market prices for the most exposed companies, sectors or countries. 

Specific market risk is also associated with a new asset class of instruments, via the growing market of Green Bonds (bonds whose proceeds are used to fund specific “green” projects) and Sustainability Linked Bonds (performance-based bonds whose payoff is indexed to the achievement of pre-defined sustainability targets). This market has grown to more than $1 trillion in outstanding issuance in 2021 and requires ad-hoc analysis in terms of pricing and risk.

Atoti+ market risk modules can be used to assess the exposure to market risk adjustments according to what-if fire-sale scenarios. The analysis can be applied across jurisdictions and legal entities to create a full picture of consolidated market risk.

3.2 Credit Risk

Credit risk is the risk of loss resulting from a counterparty failing to meet a payment obligation, either on a debt instrument or on a derivatives contract, or from collateral value adjustment. Climate risk is now considered a key determinant of banks’ balance sheet quality and of an asset’s creditworthiness.

Climate risks — specifically transition risks — have started to impact default probabilities, credit ratings and spreads, with very concrete consequences on loan and bond losses. Banks have started to integrate climate-related risks into their assessments of borrower credit risk. At deal inception, depending on the bank’s preferred scenario, borrowers with high carbon intensity are charged more than those with low carbon intensity to compensate for the additional default risk. The cost of risk assessment comes from expected risk, as well as from the cost of capital buffers required to cover tail risk

As an example, Natixis introduced the Green Weighting Factor, a mechanism that allocates capital to financing deals based on their climate impact. Under the mechanism, analytical risk-weighted assets (RWA) are reduced by up to 50% for green deals, while facilities that have a negative environmental and climate impact see their analytical RWA increased by up to 24%. Another example is Crédit Agricole CIB, who implemented the Liquidity Green Supporting Factor: to encourage green financing by its business lines, the bank grants projects tackling climate change with an internal premium of 5 bps (2020).

In aggregate, it is unclear how markets weigh climate scenarios and take climate risks into account when determining the value of financial assets. If the probability of adverse scenarios increases, spreads and capital buffers might well be below what is appropriate to cover lenders’ exposure, threatening financial stability and the transition via the abrupt market risk adjustments discussed in 3.1.

Atoti+ can be used both for credit risk and for collateral monitoring. I will present this specific use case in my next blog article.

3.3 Liquidity Risk

Liquidity risk is the risk that a market participant is unable to come up with the cash needed to meet its immediate obligations, either from a contract payment or from collateral.

For borrowers from certain sectors and locations, climate-related considerations impact the availability and accessibility to new capital. It can be a positive impact for transition sectors or a negative impact for fossil sectors or for industries depending on supply chains in exposed coastal areas. Climate risk also impacts the behavior of agents. ESG (environment, social, governance) filters are applied by the buy side in the investment processes. New regulations, such as SFDR (Sustainable Finance Disclosure Regulation), are pushing further in that direction. Banks and financial institutions are holding massive amounts of stranded assets that might become un-saleable in the future. Under some scenarios climate risk impacts banks through their ability to liquidate or to refinance such assets.

New climate-related criteria can be factored into the bucketing of assets for central banks collateral eligibility impacting refinancing capabilities and liquidity ratios calculation. 

Atoti+ enables users to monitor liquidity risk. They can instantaneously calculate hypothetical liquidity shocks on some categories of products and quickly and clearly show the results. They can drill down and access granular level details within the balance sheet such as position level data on the loan book and have a comprehensive enterprise wide view of liquidity ratios and profiles as well.

4 Risk models and Tools

4.1 Climate-related Stress Testing

Traditional, backward looking VaR is not a catch-all for climate risk. Forward-looking climate-related stress tests are used to aggregate exposures arising from physical and transition risks. Conditional to climate scenarios, stress tests are expressed in terms of economic variables and market data shifts. They are then applied to a large range of assets, reflecting the impact of climate change across sectors, countries and regions. On top of these shifts the bank factors in assumptions on the balance sheet evolution (fixed, constant credit quality, constant credit/climate quality…). For instance the Bank of England uses a “late action” scenario on a 30-year horizon where the balance sheet is fixed and where the risk premium rises: house prices fall by 20%, BBB spreads rise by 100BPs, equity prices fall by 20%, VIX increases by 50%… 

Stress testing involves a very large number of cash flows that need to be monitored constantly and constitute a huge volume of data. Atoti+ is especially well-suited for this task. As an elastic analytical platform with strong scale-up and scale-out capabilities, it enables users to aggregate and analyze results in an efficient way and is already used for FRTB, xVA and liquidity stress testing. In those areas, Atoti+ scales to Tier-1 bank volumes and allows a user to perform multi-dimensional analysis on very large data sets, containing up to 200 to 300 dimensions. 

Graphical user interface, chart

Description automatically generated

Figure 4 – Atoti+ for liquidity stress tests

4.2 Climate extended Credit VaR

From a modelling perspective, the Basel III ASRF formula on risk weights can be reviewed to reflect climate risk and to provide new metrics. This is what the CERM model does. I have implemented an illustrative version in atoti/Atoti+ that will be explained in our next article in this series.


No doubt climate change will push many other changes to market infrastructure and to market regulation, beyond what is described in this short article. This is a fast moving area requiring the ability to factor in regular updates in key dimensions: governance, strategy; risk management; metrics, targets, and limits; scenario analysis; and disclosures. 

To build quickly new metrics for climate finance one needs good and flexible tools that can handle data and new aggregations, including non-linear aggregations (quantiles). atoti and Atoti+ are really great for that. They provide risk modules across most areas mentioned above, allowing to track climate and transition impacts across the full spectrum of risk. They are powerful and scalable. In a context of active R&D, an additional benefit is that they are directly actionable from Jupyter Notebooks.

In the next blog post, I will present a few tips on the CERM implementation for credit risk. Stay tuned!