Actuarial Notes

Climate Risk

Study guide: Exam 6C

Climate Risk is the risk to an insurer from climate change, in two forms: physical risk — more frequent and severe weather events damaging insured property and the insurer's own assets — and transition risk — losses arising from the shift to a low-carbon economy, through policy, technology, litigation and changing preferences. OSFI's Guideline B-15 makes it a prudential matter, not a corporate social responsibility one.

  • Physical risk divides into acute (individual events: flood, wildfire, hail, windstorm) and chronic (gradual shifts: sea level, temperature and precipitation regimes, permafrost). Canadian insured catastrophe losses have risen by an order of magnitude over recent decades, and flood and wildfire dominate.
  • Transition risk reaches insurers through their investments (carbon-intensive assets losing value), their liability underwriting (directors' and officers' and professional liability exposure to climate litigation), and their business model (declining demand for coverage of activities being phased out).
  • The actuarial problem is non-stationarity. Ratemaking and catastrophe models assume the past is a guide to the future. If the hazard itself is changing, historical experience is a biased estimator of future cost, and "1-in-100" language describes a return period that no longer holds. This is the single most important technical point on the topic.
  • OSFI's expectations (B-15): governance and accountability at board level; climate risk integrated into the risk management framework and ORSA; scenario analysis using specified transition and physical pathways; and disclosure aligned with international frameworks. OSFI has also run standardised climate scenario exercises across the industry.
  • The insurability question. Rising hazard can make coverage unaffordable or unwritable in the most exposed locations — the Flood Insurance problem generalised. The policy response is the same trilemma: risk-based pricing, explicit subsidy, or a coverage gap absorbed by government assistance.
  • The IAA and CIA have both published on the actuary's role: climate belongs in pricing, reserving, catastrophe modelling, capital, investment and disclosure, and it is within the Appointed Actuary's remit rather than beside it.

Trending a Non-Stationary Hazard An insurer's water damage and flood loss costs per exposure over ten years show an average annual increase of 9\%. Three of those years contained a major flood event. The pricing actuary proposes projecting the 9\% forward for the next three years.

Critique the approach.

Answer The 9\% is a mixture of at least four distinct effects, and projecting the blend forward is wrong regardless of which direction it errs in.

Decompose it:

  1. Ordinary inflation in repair and remediation costs — persistent, and projectable.
  2. Exposure change — more finished basements, more expensive contents, more building in exposed locations. Persistent, and projectable, but it is a change in the exposure base, not in the hazard.
  3. Event randomness — three major events in ten years may be more or fewer than the underlying rate implies. Not projectable at all; it is process variance being read as trend.
  4. Genuine hazard change — the underlying frequency and severity of extreme precipitation increasing. Persistent, and the reason the past understates the future.

Why the naive projection fails. If much of the 9\% is item 3, the insurer is projecting bad luck forward and will overprice. If much of it is item 4, the historical average understates what is coming and the insurer will underprice even at 9\%. The two errors point in opposite directions, and a single blended trend cannot distinguish them.

The defensible approach:

  • Separate attritional from catastrophe experience, and model catastrophe losses from a model calibrated to current climate, not from ten years of the insurer's own event history.
  • Normalise historical losses to current exposure levels before fitting any trend, so item 2 is removed.
  • Use forward-looking hazard science rather than the insurer's own loss history for item 4 — this is the specific point at which climate change breaks the actuarial toolkit, and no amount of care with the insurer's own data solves it.
  • Test the sensitivity and put the range, not the point, in front of the pricing committee and in ORSA.

And say so in the report. Under the Standards of Practice the assumptions and their uncertainty must be communicated; a 9\% trend presented as a selection, with no statement that the underlying hazard is changing, is not adequate communication.