What Insurers Know About Canada’s Coastal Risk

Deciding what to do about coastal risk is not optional for the insurance industry, and not because insurers are virtuous. It is because they cannot defer the reckoning. When a risk is real and recurring, insurers must put a price on it, restrict coverage, or leave the market, and each of those decisions becomes visible to everybody else. They have no mechanism for absorbing losses indefinitely and calling it policy. That constraint makes insurance market behaviour a leading indicator of where ocean risk is concentrating in the Canadian economy, and the signals it is sending are not being read clearly enough by the other institutions that share the same exposure.

Access to flood coverage in Canada has widened sharply over the past decade. Overland flood insurance barely existed here in 2013, and by July 2026 the Insurance Bureau of Canada put roughly 94 percent of residential properties in the country in reach of it, with more than 30 insurers writing some form of the coverage. Standard home and business policies still exclude flood by default, so the coverage arrives as an endorsement a homeowner has to buy, and purchase rates climbed from 26 percent of households in 2017 to 71 percent in 2023. The number of homes unable to obtain overland coverage at all has fallen by close to 45 percent in seven years, from approximately 1.5 million to roughly 850,000. What remains is concentrated. The bureau reported that about 2 percent of residential properties account for more than half of Canada’s potential flood losses, and the 6 percent that remain ineligible tend to be in high-risk areas where coverage is unavailable or unaffordable. That count is national and covers riverine, urban, and coastal flooding together. Coastal storm surge is a narrower problem again, and coverage for it remains limited even where overland flood coverage is freely available.

That uninsured risk does not disappear, and it does not all land in the same place. The federal flood insurance task force’s 2022 report put annual flood damage in Canada at approximately CAD 1.5 billion, roughly CAD 800 million of it uninsured, with residential property owners carrying about 75 percent of the uninsured amount each year. What households do not absorb falls to businesses, municipalities, and provincial and federal disaster assistance budgets. Canada’s Disaster Financial Assistance Arrangements have paid provinces and territories more than CAD 14 billion since 1970, across floods, wildfires, storms, and everything else the program covers, and averaged CAD 881 million a year between 2010 and 2024. The Parliamentary Budget Officer projects the program total rising to CAD 1.8 billion a year over 2025 to 2034. Flooding is the largest single component of that projection at CAD 1.2 billion a year, ahead of wildfires at CAD 325 million and storms at CAD 258 million. Quebec’s 2019 spring floods produced approximately CAD 390 million in public expenditure against CAD 186 million in insured losses. Hurricane Fiona generated approximately CAD 660 million in insured losses with the majority of total economic damage falling to governments and individuals without coverage. The DFAA totals cover flooding of every kind, and the 2019 Quebec floods came down rivers. Hurricane Fiona is the ocean event in that set. What connects the three is how the cost was absorbed. Each was processed as a disaster requiring a public response, and cumulatively they represent a transfer of flood risk, including the ocean-driven part of it, from private markets, which have priced it out of reach or excluded it entirely, to households and public balance sheets, which have absorbed it without equivalent repricing of the underlying exposure.

This is where the moral hazard problem becomes analytically important. When public disaster assistance covers some of the losses that private insurance does not, the repricing signal that should reach insurers, developers, municipalities, and property owners is blurred. The assistance is conditional. Federal reimbursement generally excludes losses where adequate insurance for the hazard was available in the affected region, provinces set their own eligibility and caps on top of that, and the money goes to the province rather than to the property owner. The backstop is neither automatic nor complete. What it produces is a cost with no number attached to it until after the event, which is the condition under which a municipality weighs coastal development, a developer sites a subdivision, and an owner decides against an endorsement. The public backstop is not a failure of disaster policy. It is a structural feature of how coastal risk is currently managed in Canada, and it can weaken or delay the adjustment that risk pricing might otherwise produce.

The catastrophe models that underpin insurance pricing add a further layer of complexity. Catastrophe models generate thousands of simulated events against historical data to estimate annual average losses and probable maximum losses at various return periods. They are the mechanism through which insurers translate physical ocean conditions into financial assessments. The problem is that these models were built on historical data that is becoming less reliable as ocean conditions change. The concept of a one-in-one-hundred-year event, meaning a 1 percent probability of occurring in any given year, is being destabilized by a climate that is producing what were once extreme events with increasing frequency. Major reinsurers including Swiss Re are integrating new climate model outputs and extending their event catalogues to capture outcomes beyond historical experience. But this effort is ongoing and incomplete, and in the intervening period, models that rely on historical loss data are likely understating actual risk in Canadian coastal conditions.

Global reinsurance markets set part of the price Canadian insurers pay. Canadian property insurers transfer significant shares of their catastrophe exposure to that market, where the cost of the transfer answers to worldwide catastrophe losses, the supply of capital, competition among reinsurers, and the position of the market in its cycle. Swiss Re put global insured losses from natural catastrophes at USD 141 billion in 2024 and USD 107 billion in 2025, a year that included the largest insured wildfire loss on its records. Reinsurance prices still fell. Howden Re put the risk-adjusted decline in global property catastrophe rates at 14.7 percent at the January 1, 2026 renewals, the largest annual reduction since 2014, on the strength of record capital and competition for business. Insurance pricing is therefore informative about coastal hazard without being a map of it. A Canadian insurer can raise deductibles, tighten underwriting criteria, or decline a particular high-risk property in the same year its own reinsurance gets cheaper.

OSFI and the Autorité des marchés financiers found in their standardized climate scenario exercise that property and casualty insurers were well ahead of deposit-taking institutions at locating exposures on a map, and that only 12 percent of the deposit-takers required or collected flood insurance information on the properties they held as collateral. Consideration of flood risk in mortgage underwriting was limited. Desjardins provides one example of that information reaching a lending decision. On February 1, 2024, under the zone framework Quebec used at the time, it stopped writing new mortgages on homes in the zero to twenty year flood zone, the band carrying about a 5 percent annual chance of flooding. At the time, it did not sell flood coverage in those zones either, so an owner in Baie-Saint-Paul could find one institution declining both the policy and the loan.

Against that picture of constraint and withdrawal, the most interesting development in ocean-related insurance is the use of pre-agreed triggers to get money moving before a conventional claim could be adjusted. The Mesoamerican Reef Fund and the broker WTW built a parametric policy covering seven reef sites across Mexico, Belize, Guatemala, and Honduras, with AXA Climate and Munich Re providing the insurance capacity. The policy pays on measured wind speed at a covered site, with no adjuster and no assessment of the damage. When Hurricane Lisa crossed Belize’s Turneffe Atoll in November 2022, WTW calculated wind intensity at 70 knots, which met the policy condition and released USD 175,000 to MAR Fund’s emergency fund within fourteen days. The money reached the response brigades within 48 hours of arriving, and within a month thirteen brigade members had stabilized and repositioned 178 fragments of staghorn coral at two Turneffe sites. The Nature Conservancy has developed equivalent parametric reef policies in Hawaii and Mexico. Rare, a conservation NGO, launched parametric income insurance for small-scale Filipino fishers in 2025, with payouts triggered by fishing days lost to extreme weather. These are not large instruments by the standards of global insurance markets, and none of them stops a storm or rebuilds a reef on its own. What they settle in advance is what will be paid and on what signal, so that ecological recovery, and the household income that depends on the same water, is funded in the weeks that matter instead of the months.

The underlying logic of this approach, that intact ecosystems reduce insured losses and that insurers therefore have a financial interest in ecosystem health, is beginning to attract serious attention in actuarial and reinsurance circles. A global meta-analysis of wave studies found that coral reefs reduce incoming wave energy by an average of 97 percent, with the reef crest alone accounting for 86 percent. Salt marshes and mangrove forests provide buffering functions of their own. Lower expected damage is the kind of input a catastrophe model can carry, and an insurer able to quantify what a reef, marsh, or mangrove prevents would have a financial reason to care whether it stays intact. Whether that reaches a premium depends on the resolution of the modelling, the credit an underwriter will give for it, and the competitive position of the insurer writing the policy. The regulatory and data infrastructure to carry that value consistently into insurance decisions does not yet exist at scale.

What insurance knows about Canada’s coastal exposure, the location and magnitude of unpriced risk, the gap between insured and uninsured losses, the inadequacy of models calibrated to historical conditions that no longer describe the present, and the potential value of intact ecosystems as risk-reducing assets, is useful to the other institutions that share that exposure, and underused by them. Governments that design disaster assistance programs, lenders that underwrite coastal mortgages, developers that site infrastructure in coastal zones, and municipalities that plan for long-term capital needs are all operating in the same risk environment that insurers are already required to price. The gap between what insurance knows and what the rest of the system has priced is where coastal financial risk is likely to become harder to ignore.