Skip to content Skip to sidebar Skip to footer

OSFI’s Climate Modelling Matters — But Should Not Be Misinterpreted

Updated September 5, 2026

1. Why This Matters 

Canada’s financial regulator, the Office of the Superintendent of Financial Institutions (OSFI), is about to release a climate modelling exercise that assesses how climate-related risks—the transition away from fossil fuels as well as some physical impacts—might impact the lending, underwriting, and investment portfolios of our biggest banks, pensions, and insurers.

We get it: financial modelling isn’t exactly thrilling dinner-table conversation. But this kind of technical work can help shape big, behind-the-scenes decisions about how—and whether—money continues to flow into fossil fuels, or gets redirected toward safer, more sustainable investments.

If we are far off with the modelling and fail to see its limitations, we risk lulling ourselves into a dangerous sense of security. Modelling can only ever produce speculative results, and we must continue to improve it, while being clear about its limitations. This ensures that as we interpret its results, we apply an appropriate level of precaution.

As Dr. Yu (Nina) Chen, former US federal banking regulator and current senior visiting fellow at the Center for Economic Transition Expertise at LSE, so eloquently and succinctly puts it:

We need a warning label on such [climate scenario analysis] outputs: “Partial estimate. Likely underestimates. Use for education, not risk quantification.”

Below, I explain why this is the case with our upcoming Canadian climate scenario modelling exercise.

2. What the OSFI Exercise Is

OSFI’s latest climate risk modelling exercise builds on its 2021 pilot project, with the goal of helping financial institutions and regulators better understand how climate-related transition and physical risks could impact portfolios and, by extension, financial stability. It is an ambitious project to measure systemic societal risk.

The new version is a mandatory exercise for all federally-regulated financial institutions (our major banks, insurance companies, and pensions), which requires them to model certain future climate risk impacts on higher risk areas of their portfolios (e.g. fossil fuels, real estate, transportation, etc). The results are forwarded to OSFI, which publicly reports aggregated findings. This is a step forward for Canada’s financial system supervision.

The exercise helps OSFI assess the financial system’s resilience to potential future climate risks, and, if necessary, mitigate these with its corrective powers. The exercise also educates financial institutions about potential climate risk pathways and impacts to their portfolios so that they can take their own corrective measures. Depending on industry projections for systemic change–usually self-serving–is problematic, as the financial sector tends to work on a relatively short time horizon that follows the real economy, and the real economy only tends to change with appropriate government policy. 

The voluntary  OSFI climate risk pilot modelled the probability of default for Canada’s refined petroleum sector as potentially rising up to 550% by 2050 under a delayed transition scenario. It also projected oil and gas asset values dropping by as much as 100% below current prices under a net-zero scenario. Although based on a smaller sample of institutions (two banks, two pensions, two life insurers, and two P&C insurers), this gave Canadian financial institutions credible evidence to begin acknowledging and integrating climate risk into their planning. Nonetheless it did not lead to practical changes in financing behaviour. Between 2023 and 2024, Canadian banks significantly increased their financing of fossil fuel expansion.

The new iteration of the model includes some physical risks alongside transition risks—an important and necessary evolution. For example, it attempts to model the effects of tidal flooding and wildfires in Canada on financial portfolios. This iteration also extends the assessment to include real estate. This is a valuable step toward understanding how climate-related disruptions might ripple through our financial system.

3. What the OSFI Exercise Isn’t

Despite these advances, the exercise has significant limitations outlined in its publicly available methodology. OSFI deserves credit for being transparent; it clearly states that the results do not provide a comprehensive estimate of climate risk, and recognizes that more robust modelling is still needed. Nonetheless, there remains a risk of misleading decision-makers if the extent of these limitations are not properly understood.

First, the scope of physical risks it models is narrow. It only includes direct impacts from tidal flooding and wildfires in Canada. This means it excludes a swath of costly physical impacts, including:

  • Flooding from rain;
  • Ice storms, wind storms, and freeze-thaw cycles;
  • Multiple or sequential physical events;
  • Physical risks to international assets; and
  • Interactions between physical and transition risks.

The few physical risks that it does model are based on scenarios designed by the Network for Greening the Financial Sector, which the Chairperson herself acknowledges underreport risks. Partly this is due to the fact that it is difficult to project environmental tipping points–like the abrupt melting of an ice cap or collapse of a major coral reef–so the model does not try. Neither does it capture second- or third-order financial impacts. For example, a primary impact of a policy to require all new buildings to be net-zero by 2035 might be to tank the value of older, inefficient homes. 

However, primary financial impacts can lead to second- and third-order impacts like increased mortgage defaults, housing market instability, and ripple effects across construction, insurance, and real estate sectors. Similarly, a wildfire would have a direct impact on physical assets, like buildings and machinery, but second- and third-order order impacts, like business disruptions and value chain impacts are real risks that are largely missing from the model. As a final illustrative example, second- and third-order impacts can include geopolitical conflict triggered by extreme weather

Furthermore, it is not realistic to assume a future scenario in which climate transition risks occur without increased climate physical risks, yet the model does not consider these together. In fact, it models climate transition impacts in a future with a continually rising GDP, which is unlikely with increasingly acute and frequent weather events. 

Finally, even the manner the model considers transition risk is significantly limited. It applies an incremental increase in carbon price to represent all transition risk drivers. This has already proven to significantly underrepresent the economic impact of technological disruptions, like sharply decreasing battery costs, which have ripple effects through energy and transportation value chains.

These gaps are not just technical—they’re consequential. As the International Financial Stability Board warned in 2022, such models can create the illusion that climate risks are limited to a few sectors and are, for now, “manageable.” The Institute and Faculty of Acturaries at the University of Exeter further highlighted this issue in 2023, 

These outputs may provide false comfort to institutions and advisers. They may be particularly dangerous for regulators seeking to understand systemic risk, as an aggregation of benign results may result in misplaced confidence regarding the threat of climate change to financial resilience.

In other words, this modelling exercise only captures a small fraction of what could realistically happen in a warming world. But that’s not reality. That’s a modelling constraint.

4. What We’d Like to See Next Time

There’s no doubt this modelling work is complex and evolving. But rather than letting imperfect models dictate complacency, financial regulators must adopt a precautionary approach.

That means acknowledging the limitations of current modelling—not just in its methodology, but in framing headline findings—and taking meaningful action to manage climate risks, even when their financial risks can’t be fully quantified. For example, employing its monetary policies (e.g. capital buffers for loans and investments) to acknowledge the higher risk of transition-exposed assets (e.g. recent developments by the European Central Bank). Also, bringing into force all elements of OSFI’s Climate Risk Management Guideline B-15, such as its transition plan requirements, which the UK has recently begun efforts to implement.

Improved modelling is essential and should continue. In addition to addressing the modelling gaps identified above–for example, analyzing both direct and indirect impacts, considering transition and physical risks in tandem, and considering non-linear rates of physical and transition risks–OSFI may also consider adopting narrative scenarios, which can be a decision-useful tool for projecting the scope of future climate risks. 

In the meantime, we must not wait for perfect financial risk forecasts—which will never arrive—before regulating in line with the scientific reality that climate change poses systemic, escalating, and destabilizing risk to Canada’s financial sector.

Join Our Mailing List