Physical climate risk is visible. Action is the problem

Physical climate risk is increasingly understood across financial institutions.

It is far less often acted on.

In practice, institutions are not waiting for a fully closed chain from hazard through to capital. Many are already working with what is decision-useful and regulator-defensible today, even where that means operating with partial translation.

Across the recent Banking and Capital Markets Strategy I developed for Swiss Re and Fathom, Phase 1 focused on capturing a ‘voice of the market’ across banks, asset managers, insurers and data platforms. One theme came through with unusual consistency. The debate has moved on from whether physical risk matters. It is now about whether institutions are actually able to act on what they are seeing.

This is where the constraints start to show.

Most large financial institutions are not designed to make decisions on the basis of uncertain, forward-looking and evolving risk signals. They are built to operate on relatively stable assumptions, supported by models that can be validated, explained and repeated through governance frameworks.

Physical climate risk challenges that model.

It introduces uncertainty that does not collapse neatly into a single number. It evolves in ways that are not always well captured by static scenarios, and it often sits outside the historical datasets that underpin existing risk frameworks.

Turning that insight into action is not straightforward.

In practice, many institutions are taking a pragmatic approach. Rather than waiting for a fully integrated chain from hazard through to loss and capital, they are embedding decision-useful components where they can, screening portfolios, segmenting exposures, applying scenario overlays, and building a view incrementally in a way that can be explained and defended.

But this does not remove the underlying constraint.

As those insights move closer to core decisions, lending, pricing, allocation, they encounter the realities of how institutions actually operate.

Credit teams are not incentivised to reprice based on forward-looking hazard signals. Investment teams are not rewarded for moving ahead of the market. Model risk functions are required to challenge anything that cannot be clearly evidenced or back-tested.

So even where the signal is clear, acting on it is not straightforward.

This plays out in practical ways. Analysis is often produced at an asset level, while decisions are taken at portfolio level. Outputs are expressed as hazard or damage estimates, while decisions require impacts on PD, LGD or valuation. Risk teams may have a view, but the commercial decision sits elsewhere. In many cases, no single function owns the end-to-end translation.

The result is not inaction, but dilution.

The signal weakens as it moves through the organisation.

Recent commentary from institutions such as the Bank for International Settlements and the European Central Bank has started to acknowledge this more explicitly, highlighting the gap between risk identification and integration into core financial processes. Work from McKinsey & Company and Deloitte points to the same challenge, embedding climate risk into decision-making, not just measuring it.

There is also a structural timing issue.

Much of the climate risk framework used today is built on long-term scenarios. These are useful for understanding direction, but harder to apply to decisions that operate on much shorter horizons. A credit committee is not making a decision about 2050. It is making a decision about the next refinancing cycle, the next covenant test, the next few years of performance.

At the same time, the underlying hazard is not static. It is evolving, sometimes gradually, sometimes abruptly, and not always in line with scenario assumptions.

Much of the current framework still assumes that risk evolves in a relatively smooth and observable way. In practice, physical climate risk does not always behave like that. It can be non-linear, with critical thresholds, step changes and compounding effects that are difficult to anticipate and even harder to price.

This becomes more relevant as the industry starts to engage more seriously with non-linear shifts, where the timing and magnitude of change are deeply uncertain and, in some cases, irreversible.

This creates a different kind of challenge for decision-makers.

Financial institutions are generally not designed to act on risks that do not move incrementally. Credit, pricing and portfolio decisions tend to assume that change is gradual, observable and can be incorporated over time. Non-linear shifts disrupt that assumption, introducing the possibility that exposure moves more quickly than governance, models or market pricing can respond.

That does not make the risk invisible. But it does make it harder to act on with confidence.

In many cases, it reinforces existing behaviours, a preference to wait for clearer signals, stronger evidence or broader market alignment before making a move.

There is also a broader question of organisational maturity.

The challenge is not only about data or models. It is about whether institutions have the governance capability and internal alignment to act on what the analysis is telling them. Decision readiness evolves over time. The actions an organisation is willing, and able, to take today may look very different in a few years’ time.

This is not a problem with a fixed end state.

There is also a human dimension.

Acting on physical climate risk can create immediate consequences, for client relationships, for revenue and for portfolio composition. The downside of acting early is visible and near-term. The downside of acting late is uncertain and further out.

So decisions tend to follow incentives.

Few institutions want to move pricing in a way that is difficult to justify externally. Few want to take decisions that cannot be defended through existing governance frameworks. And few are willing to introduce volatility into processes designed to be stable and repeatable.

The next phase of the market will not be defined by better models alone.

It will be defined by whether institutions can adapt their decision frameworks to accommodate this type of risk, incrementally, pragmatically and in a way that builds confidence over time.

That may not require a fully closed chain from hazard through to capital in one step.

But it does require clarity on where risk enters decisions, what level of insight is sufficient to act, and how that action is governed and repeated.

The underlying issue is no longer visibility.

It is action, in a system that is not designed to absorb how this risk actually behaves.

Until that changes, physical climate risk will continue to be understood in analysis, but only partially reflected in the decisions that move capital.

That is where the tension now sits.