Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach
summary
The gist
Climate scenario choice affects model-implied default probabilities and the resulting risk-return tradeoff across obligors and sectors The paper develops a framework to assess loan portfolio budget…
In short
This study develops a framework to assess loan portfolio budget under NGFS climate scenarios over 3-5 years. It links sector-level climate shocks to firm default probabilities and capital signals using a structural credit risk model. The results show how different climate pathways significantly alter lending allocation patterns across various sectors.
Key concepts
- NGFS Scenarios
- These are short-term climate scenarios provided by the Network for Greening the Financial System (NGFS). They represent different future paths for global greenhouse gas emissions, which serve as the external shocks used to test how climate risk impacts financial stability and lending decisions.
- Structural Credit Risk Setup
- This is a modeling approach that treats firms as having underlying economic and physical risks. It uses stochastic differential equations to model firm value dynamics, incorporating climate-related transition risks (like regulatory changes) and physical risks (like extreme weather penalties).
- Default Probability Estimation
- The paper calculates the likelihood of a borrower defaulting by simulating firm values under specific climate scenarios using Monte Carlo methods. A default occurs when the firm's value drops below a set liability threshold, allowing for the estimation of scenario-specific PDs.
- Lending Allocation Optimization
- This involves formulating lending decisions as a constrained linear program. The goal is to maximize expected interest income while simultaneously penalizing high default risk and excessive capital usage, thereby determining how much capital should be allocated to each obligor.
Terminology used across episodes
This episode discusses
The paper
Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach · Read on arXiv
This paper develops a framework to assess loan portfolio budget in credit portfolios exposed to climate risk over a 3-5 year horizon. Using short-term scenarios from the NGFS, the approach translates transition and physical risk pathways into changes in obligor default risk. Firm performance is modeled within a structural credit risk framework where scenario-dependent climate shocks adjust the drift of firm value dynamics. Defaults occur when firm value falls below a liability threshold. Joint defaults are simulated using Monte Carlo methods with dependence captured via principal components. Resulting default-probability estimates inform a constrained optimization used to examine lending allocation across exposures. Climate scenario choice affects model-implied default probabilities and the resulting risk-return tradeoff across obligors and sectors. The results show scenario-specific differences in default-probability distributions, upper-tail risk, and lending allocation patterns. Under the benchmark synthetic portfolio, HWTP produces the highest mean and upper-tail PDs, while SWUC produces the lowest mean and median PDs. These rankings are model-implied and calibration-dependent rather than empirical rankings of climate-scenario severity. The framework remains transparent and can be refined as more granular climate and firm-level data become available. The framework helps banks link NGFS scenarios to sector-level credit risk and portfolio loss outcomes. It can support ICAAP and climate stress testing by highlighting exposures most sensitive to climate risk, adding a forward-looking overlay to existing models.
Transcript
Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.
Ruth: We're looking at the paper: "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach".
Warren: Here's the gist of the paper. Climate scenario choice affects model-implied default probabilities and the resulting risk-return tradeoff across obligors and sectors The paper develops a framework to assess loan portfolio budget in credit portfolios exposed to climate risk over a 3–5 year horizon This study embeds NGFS scenarios in a structural portfolio credit risk setting, linking climate narratives to firm-level default dynamics and capital signals in a tractable, scenario-analysis framework Modeling the Core Framework The paper presents a modeling framework that links NGFS climate scenarios to changes in firm-level default probabilities and lending allocation outcomes This approach uses a structural credit risk setup with scenario-adjusted firm performance to translate sector-level climate shocks into obligor-specific drift adjustments and simulates defaults via Monte Carlo methods The objective function for the lender is formulated as maximizing expected interest income subject to capital constraints, which simplifies to a constrained linear problem max x1,..., xN X N i=1 xiηi(1 − PDi) subject to X N i=1 xi = K, 0 ≤ xi ≤ κK for all i Climate Risk Translation The core mechanism for incorporating climate risk involves translating NGFS scenario shocks into firm-level drift adjustments, which are then used to adjust the drift of firm value dynamics Transition risk is modeled by defining an emission decision vector and cost function that includes quadratic forms for energy output and regulatory penalties/rewards based on benchmark emissions Physical risk is introduced as a penalty term in the log-production stochastic differential equation, where xi represents the amount of capital to be allocated to obligor i Default Probability Estimation Default probability estimates are computed by simulating firm value processes under scenario-specific parameters using Monte Carlo methods The default condition is defined when the firm’s value falls below a liability threshold L i(t), and the probability of default is defined as PDi t:= P Vbi t ≤ L i(t) Systemic risk dependencies are captured by applying Principal Component Analysis (PCA) to the covariance matrix of firm-specific default thresholds X t, allowing for efficient simulation of joint defaults Lending Allocation Outcomes The final step involves formulating lending allocation as a linear program to maximize a risk-adjusted objective that penalizes both default risk and capital usage The resulting allocation patterns differ significantly across the four NGFS scenarios, showing how scenario-adjusted default probabilities and diversification constraints jointly determine portfolio composition For instance, under DAPS, capital is concentrated primarily in Wind, while under SWUC it favors Clean Gas and CO2 Capture Limitations and Practical Use The framework has limitations because NGFS scenario shocks are mapped to obligors through sector-level assignments, implicitly assuming that firms within a sector face similar transition and physical risks The study suggests that the results should be interpreted as model-implied outcomes under the synthetic portfolio design, rather than as empirical rankings of climate-scenario severity It can be used to support ICAAP and climate stress testing by highlighting exposures most sensitive to climate risk This work presents a simulation-based framework for studying climate-adjusted credit risk and scenario-informed lending allocation across a large portfolio of obligors The simulation framework shows how NGFS scenarios influence firm solvency and portfolio risk Overall, the outputs offer a quantitative view of how climate policy uncertainty affects creditworthiness and portfolio resilience References
Auzepy and Bannier(2025): Auzepy, A.; Bannier, C.E.
Ruth: Integrating Climate Risks in Bank Risk Management and Capital Requirements; Wiesbaden, Germany, 2025
Bourgey et al.(2025): Bourgey, F.; Gobet, E.; Jiao, Y.
Warren: An Efficient SSPbased Methodology for Assessing Climate Risks of a Large Credit Portfolio.
Ruth: HAL Open Science Preprint 2025
Cucker and Smale(2002): Cucker, F.; Smale, S.
Title and authors: Ruth: So we're looking at this paper now titled "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach". What does that actually mean for us?
Warren: It means they’ve built a framework to figure out how climate risk changes the default probabilities on a loan portfolio over about three to five years. It links those big climate scenarios, the NGFS ones, directly to what happens with firm defaults.
Iris: And the core idea is using Monte Carlo simulations within a structural credit risk setup. They adjust how firm value dynamics drift based on those scenario-dependent climate shocks, and then they see where the money should go.
Ruth: Right, so if we're driving here and thinking about loans, this paper is trying to answer what happens when you factor in different climate futures into your lending decisions.
Warren: Basically, the model uses a constrained linear program to maximize expected interest income while keeping track of capital limits for each obligor. That objective function is max xone..., xN xi eta i (one minus PDi) subject to X N i=one xi = K, zero less than or equal to xi kappa K for all i.
Iris: And they are estimating those default probabilities using Monte Carlo methods by simulating firm value processes under the specific scenario parameters. The default condition is when the firm value falls below a liability threshold L i(t), and the probability of default PDi t is defined as P Vbi t less than or equal to L i(t).
Ruth: That sounds like they are taking a big, abstract climate narrative and translating it into something concrete for individual companies. How do they actually handle the transition risk versus the physical risk?
Warren: Transition risk is modeled by defining an emission decision vector and a cost function that includes quadratic forms for energy output and regulatory penalties or rewards based on benchmark emissions. Physical risk gets introduced as a penalty term in the log-production stochastic differential equation, where xi represents the amount of capital to be allocated to obligor i.
Title and authors: Iris: So they are using these adjustments to change the drift of firm value dynamics, which is their main channel for incorporating those NGFS sector shocks into the structural credit risk model. They also use Principal Component Analysis on the covariance matrix of firm-specific default thresholds X t, which lets them efficiently simulate joint defaults and systemic risk dependencies.
Ruth: That dependence capture through PCA is a smart way to look at joint defaults, because in reality, climate risks aren't isolated; they interact across sectors. So what do we get out of all this simulation work?
Warren: The final step is formulating lending allocation as a linear program to maximize a risk-adjusted objective that penalizes both default risk and capital usage. The resulting allocation patterns shift quite a bit depending on which NGFS scenario you choose. For example, under DAPS, capital gets concentrated primarily in Wind, while under SWUC it favors Clean Gas and CO2 Capture.
Iris: They found that the choice of climate scenario significantly affects the model-implied default probabilities and the resulting risk-return tradeoff across obligors and sectors. They also showed scenario-specific differences in default-probability distributions, upper-tail risk, and lending allocation patterns under DAPS, HWTP, SWUC, and DiRe scenarios.
Ruth: It sounds like this is really about showing how uncertainty in climate policy translates into real shifts in where banks decide to lend their money. What are the suggested improvements for this framework?
Warren: The paper suggests that the framework has limitations because they map NGFS scenario shocks to obligors through sector-level assignments, which means it implicitly assumes firms within a sector face similar transition and physical risks. It also says these results should be interpreted as model-implied outcomes under the synthetic portfolio design, not as empirical rankings of climate-scenario severity.
Iris: And they point out that the framework doesn't fully capture within-sector heterogeneity in things like firm-level carbon intensity, geographic exposure, or adaptation capacity. They say the framework remains transparent and can be refined as more granular climate and firm-level data become available.
Title and authors: Ruth: So what does this mean practically for institutions? The authors suggest it can support ICAAP and climate stress testing by highlighting exposures most sensitive to climate risk, adding a forward-looking overlay to existing models.
Warren: Exactly, they're providing a way for lenders and investors to explore how different climate pathways may affect relative sector risk and portfolio vulnerability. It gives you that forward-looking overlay you need.
Iris: And from an econometric standpoint, they are contributing to stress testing research by translating NGFS scenario shocks into obligor-level default probabilities and portfolio allocation outcomes, which connects to climate finance research showing how climate risk is priced through firms’ emissions exposure and transition vulnerability.
Ruth: So the paper title "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach" essentially gives us a tool to test how different ways the world warms or shifts policy will change a bank's loan portfolio structure.
Warren: Right, and it shows that under DAPS, for instance, capital is heavily weighted toward Wind because of those specific scenario-adjusted default probabilities. That’s a concrete result you can use to think about asset allocation.
Iris: And it also highlights the sensitivity of certain sectors to these changes, which is important because we need to see where the stress points are most exposed under different pathways.
Ruth: So, as we wrap up this discussion on "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach", this framework provides a quantitative view of how climate policy uncertainty affects creditworthiness and portfolio resilience. It's a simulation-based approach that links NGFS scenarios to firm solvency and allocation outcomes across the portfolio.
Warren: It really shows the interaction between scenario choice, default risk, and capital deployment in a structural setting. That’s what they were aiming for with this Monte Carlo approach.
Iris: It's useful because it helps institutions explore how different climate pathways may affect relative sector risk and portfolio vulnerability by providing that forward-looking overlay.
Ruth: We're moving on now, but this paper gives us a lot to consider about how we model risk when facing non-linear, scenario-based threats.
The paper's summary: Ruth: So, to wrap up what we just heard, this paper essentially creates a simulation to see how different climate futures—those NGFS scenarios—shake up a bank's lending decisions across its entire portfolio.
Warren: Right, it’s taking those big climate narratives and translating them into actual changes in the default probabilities for individual firms over three to five years. It moves the whole process from abstract risk to concrete capital allocation under uncertainty.
Iris: What’s interesting about this approach is how they link those sector-level climate shocks directly to firm-level drift adjustments in value dynamics, which then feeds into the default calculations via Monte Carlo simulations. It’s a very direct channel for translating physical and transition risks.
Ruth: Exactly, it connects the macro climate projection to the micro decisions of where capital goes. But what’s really new here is how they structure that allocation using a linear program that balances maximizing interest income against those scenario-adjusted default risks.
Warren: That optimization setup is key because it shows how much the strategy changes depending on whether you pick DAPS or SWUC, for example. It’s not just one static answer; it’s a whole map of where capital gets pushed under different climate rules.
Iris: And they use Principal Component Analysis on those default thresholds to handle systemic risk dependencies, which is important because you can't just look at firms in isolation when you're dealing with interconnected climate risks.
Ruth: So, what does this actually mean for the world outside of the technical details? It gives us a way to test how policy uncertainty around carbon pricing or regulatory change translates into tangible shifts in credit risk and portfolio resilience.
Warren: It’s about stress testing lending structures against plausible, but different, climate pathways. If you can see that under one scenario wind gets prioritized and under another clean gas, the strategic implications for asset allocation are pretty clear.
Iris: The limitation they point out is that they map these sector shocks to firms through existing sector assignments, so the model assumes all firms in a sector face similar transition and physical risks. That’s a big assumption you have to be aware of when interpreting the results.
Ruth: True, it’s an "model-implied" outcome based on their synthetic portfolio design, not necessarily an empirical ranking of how bad those climate scenarios actually are in the real world. It’s about what happens *within* this model structure.
Warren: So, the practical value is using this to highlight which exposures are most sensitive to these climate shifts so institutions can prepare for potential vulnerability under different pathways. That's a forward-looking stress test tool.
Iris: Ultimately, this framework offers a quantitative view of how policy uncertainty affects creditworthiness and portfolio structure by linking NGFS scenarios to firm solvency and allocation outcomes through simulation.
Ruth: It’s a powerful way to visualize the trade-off between maximizing returns and managing climate-adjusted default risk when you're trying to build a resilient loan book. Next up, we’ll look at how this framework handles those specific climate shock transmissions through firm value dynamics.
The paper's improvements: Tom: So, we’ve seen how this paper sets up the core simulation linking climate scenarios to lending decisions, but what are they suggesting to make this framework better?
Iris: They point out a few things where they could tighten things up. First, they note that mapping NGFS scenario shocks to obligors just through sector assignments is an assumption; it implicitly assumes all firms in that sector face the same transition and physical risks.
Ruth: That’s fair. The authors themselves flag that their results should be seen as model-implied outcomes from their specific portfolio design, rather than a direct measure of real-world climate scenario severity. It keeps the interpretation grounded in the math, not just a ranking of warming pathways.
Warren: And they also highlight that the framework doesn't fully capture all the heterogeneity inside those sectors. They mention things like firm-level carbon intensity or geographic exposure, which could be important for more granular analysis later on.
Iris: Exactly. The econometrician side is saying that if you want to falsify their main mechanism, you have to challenge that assumption of uniform risk across a sector when applying the shocks. It’s about identification, right?
Ruth: Right, and on the practical side, they suggest using this framework to support ICAAP and climate stress testing exercises by pinpointing exactly which exposures are most sensitive to climate risk under different pathways. That’s a concrete use for it.
Warren: It gives institutions a way to explore how different climate scenarios may affect relative sector risk and portfolio vulnerability, which is useful for those forward-looking stress tests you mentioned. It’s about exploring the downside across multiple possibilities.
Iris: They also suggest that the structure is transparent enough that it can be refined as more detailed data on firm-level climate and geography becomes available. It’s not a dead end; it’s an input for future work, right?
Ruth: So they aren't just stopping at the simulation; they are laying out how this model could evolve when we have better data to make those sector assignments more precise. It shows the path forward for integrating climate finance into credit risk modeling.
Warren: It sounds like the next step is taking these improved inputs and seeing what kind of dynamic portfolio selection strategies emerge under these scenario-adjusted default probabilities. That’s where the real trading implications come in.
Conclusion: Tom: So we’re wrapping up this session on "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach". Essentially, this paper provides a simulation framework showing how different climate scenarios directly influence default probabilities and where capital gets allocated across a large portfolio of borrowers.
Ruth: It really boils down to quantifying the trade-off between expected interest income and the specific climate risk embedded in those lending decisions under various NGFS pathways. The main implication is that policy uncertainty about climate change has a measurable impact on creditworthiness at the firm level.
Warren: And for us, as traders, it means we can stress test our portfolio structures against different scenarios before committing capital, seeing exactly how much our strategy might fluctuate based on the climate rules we choose to play by. It’s about understanding the potential drawdown from these systemic shifts.
Iris: From an econometric point of view, what this study does is establish a method for identifying that variation—the scenario choice—that drives the change in default probability distributions and allocation patterns. It’s a test of identification using Monte Carlo methods across different risk regimes.
Ruth: The caveats are still there, though: they admit the model relies on sector-level assignments when mapping those climate shocks to specific borrowers, which isn't perfectly granular. And it stresses that this is an AI-driven, synthetic portfolio design outcome, not a perfect empirical reflection of every real-world credit event.
Warren: So the practical change for someone listening is realizing that if you’re building a lending strategy, you need to run these kinds of simulations to see how robust your allocation is under DAPS versus SWUC. It helps shape the risk-return profile proactively.
Iris: And it suggests that while there are limitations in mapping shocks perfectly, the framework itself is valuable for supporting ICAAP and stress testing by highlighting exposures most sensitive to these climate factors. It’s a tool for sensitivity analysis.
Ruth: Exactly, so this paper offers a way to bridge the gap between broad climate policy uncertainty and specific credit risk metrics in a portfolio setting. It gives us a quantitative lens on how policy shifts affect solvency.
Warren: It’s a solid framework for exploring those forward-looking risks before they become realized losses in the market. We need this kind of scenario mapping to make sure our capital deployment strategy is resilient across different climate futures.
Iris: Ultimately, "Climate aware lending allocation under NGFS scenarios - A Monte Carlo approach" gives us a simulation method that connects NGFS scenarios to firm solvency and allocation outcomes in a tractable way. It’s a useful model for understanding credit risk under non-linear threats.
More episodes
- The day in conversation for 2026-11-19
- Daily Summary for 2026-11-19
- The day in conversation for 2026-11-18
- Daily Summary for 2026-11-18
- 2601.05274-On the use of case estimate and transactional payment data in neural networks for individual loss reserving
- 2009.00868-Loss-Given-Default Modeling by Post-Last Passage Time Process
- 2606.30070-Financial Resilience Evaluation: From Conditional Expectations to Dynamic Convex Risk Measures
- 2212.12687-Measuring price impact and information content of trades in a time-varying setting
- Daily Summary for 2026-11-17
- 2401.08323-Dynamic portfolio selection under generalized disappointment aversion