Property Cat Reinsurance Down 14%: How to Recalculate Your Cat Load

Guy Carpenter's U.S. Property Catastrophe Rate-on-Line Index fell 14% through the April 2026 renewals, the sharpest annual decline since 2014 and a clear inflection from the 2023 peak. Yet rates remain roughly 66% above the 2017 soft-market trough. For the pricing actuary preparing a homeowners or commercial property rate filing, the question is not whether the catastrophe load should decrease, but by how much. The answer depends on how you allocate reinsurance cost changes across layers, how you blend modeled losses with market-implied pricing, and whether your attachment points shifted at renewal.

From tracking reinsurance treaty renewals over the past four cycles, the mistake most pricing actuaries make during softening is applying the headline rate change uniformly to the cat load. A 14% reduction in the gross reinsurance spend does not produce a 14% reduction in the filed catastrophe provision. The math is more nuanced, and getting it wrong in either direction creates problems: understating the reduction invites regulatory pushback, while overstating it erodes the margin cushion that the 2023-2025 hard market built.

The Cat Load Formula: Decomposing the Components

The catastrophe provision in a standard rate filing contains two components. The first is the modeled expected annual catastrophe loss, net of reinsurance recoveries. This is the output from your vendor model (Verisk, Moody's RMS, or CoreLogic) after applying the reinsurance program structure. The second is the allocated cost of the catastrophe reinsurance program itself: the ceded premium the company pays for its treaty, allocated to the line of business and state being filed.

The filed cat load equals:

Cat Load = (Net Modeled AAL + Allocated Reinsurance Cost) / Subject Earned Premium

When reinsurance costs decline, both components can move. The allocated reinsurance cost decreases directly. But the net modeled AAL may also change if the attachment point, co-participation percentage, or limit structure shifted at renewal. In 2026, attachment points remain elevated relative to pre-2023 levels even as rates soften, which means the primary insurer retains a thicker first-dollar layer than it did during the last soft market. That retained layer directly affects the net modeled AAL.

Layer-by-Layer: Why 14% Off the Treaty Does Not Mean 14% Off the Cat Load

Consider a simplified three-layer property catastrophe excess-of-loss program with a total limit of $225 million:

Layer	Structure	Expiring ROL	Renewal ROL	Expiring Premium	Renewal Premium	Savings
Layer 1	$50M xs $25M	18.0%	15.5%	$9.00M	$7.75M	-13.9%
Layer 2	$75M xs $75M	8.5%	7.3%	$6.38M	$5.48M	-14.1%
Layer 3	$100M xs $150M	3.2%	2.8%	$3.20M	$2.80M	-12.5%
Total				$18.58M	$16.03M	-13.7%

Even with a uniform 14% market decline, the actual cost reduction across the program is 13.7% because the savings are not evenly distributed. Lower layers carry higher absolute rate-on-line, so the dollar savings concentrate there. Higher layers, where rate-on-line is already thin, produce smaller absolute savings even at the same percentage reduction. The blended program reduction will always differ from the headline index figure.

Three additional factors complicate the walk from headline to filed:

Reinstatement premiums. Most property cat treaties include one or two reinstatements at pro rata or percentage-of-original rates. If your 2025 treaty had one reinstatement at 100% of the original premium and the 2026 renewal shifted to two reinstatements at different percentages, the total expected reinsurance cost for the policy period changes independently of the base rate-on-line decline. The pricing actuary must model the expected reinstatement cost using the cat model's exceedance probability curve, not simply apply the base premium reduction.

Co-participation (retained lines). Many 2023-2025 placements required the cedent to retain 10% to 20% of each layer as co-participation. If that co-participation percentage declined at the 2026 renewal (a common negotiation outcome when capacity is abundant), the ceded portion of losses increases, reducing the net retained modeled AAL. This can partially or fully offset the reinsurance cost reduction in the filed cat load.

The $25M retention itself. Although attachment points softened modestly in April 2026, most mid-size cedents still retained more than they did pre-2023. Guy Carpenter's data shows that the average attachment point for a $1 billion homeowners carrier is approximately 45% higher than the 2019 average. That retained layer represents catastrophe losses the primary carrier pays before any reinsurance recovery, and it flows directly into the net modeled AAL. If your attachment did not decrease at renewal, the reduction in the filed cat load will be smaller than the headline reinsurance cost decline suggests.

Credibility Blending: Model Output vs. Market Pricing

Some pricing actuaries use a credibility-weighted blend of two signals when setting the catastrophe load: the vendor model's expected annual loss output and the reinsurance market's implied catastrophe cost.

The logic is straightforward. Cat models carry parameter uncertainty, model risk, and demand surge assumptions that may not reflect current market conditions. Reinsurance pricing reflects the collective judgment of dozens of underwriters with proprietary loss data. Neither signal is fully credible on its own.

A complement-of-credibility approach assigns weight Z to the cat model output and weight (1 - Z) to the market-implied cat cost:

Blended Cat Load = Z × Model AAL + (1 - Z) × Market-Implied Cat Cost

The credibility weight Z typically depends on the company's own loss history relative to modeled expectations, the number of years of data, and the geographic concentration of the book. A Florida-concentrated homeowners carrier with 30 years of data might assign Z = 0.60 to the model, while a Midwest carrier with limited hurricane exposure and shorter history might use Z = 0.40.

When reinsurance rates decline 14%, the market-implied component drops, pulling the blended cat load down. But if the carrier's own experience in 2024-2025 was adverse (severe convective storm losses exceeded model expectations in both years for many carriers), the model component may move in the opposite direction after recalibration. The pricing actuary must re-derive both components, not just the market side.

Patterns we have seen in recent filings suggest that regulators are increasingly asking for documentation of the blend weights and the basis for the credibility assignment. ASOP No. 39 (Treatment of Catastrophe Losses in Property Insurance Ratemaking) requires the actuary to disclose the methodology for incorporating reinsurance costs into the catastrophe provision, and a shift in blend weights between filings without explanation will draw a data request.

Cedent Behavioral Shift: Frequency Covers Complicate Net Savings

The 2026 renewal introduced a behavioral wrinkle that complicates the cost allocation. With headline rates declining, many cedents reinvested the savings into supplementary covers. Howden Re's January 2026 renewal report and subsequent April data show that third-event and fourth-event frequency protections saw the highest growth in buyer interest since the product was introduced. Cedents also expanded aggregate excess covers and bought down retention on occurrence programs.

These additional covers reduce tail volatility, which is valuable for capital efficiency, but they add cost back into the total reinsurance spend. If the base occurrence program declined 14% but the cedent spent 40% of the savings on a new third-event cover, the effective program cost declined roughly 8%, not 14%. The pricing actuary must use the net program cost (base layers plus supplementary covers) as the numerator in the allocated reinsurance cost calculation.

Regional variation adds another layer of complexity. Howden Re reported that US programme-wide decreases ranged from 10% to 20%, Europe saw 10% to 20% declines, and Asia Pacific loss-free programs fell 10% to 20%. But these are market averages. The filed cat load in a Florida homeowners filing should reflect the cedent's own Florida-weighted treaty costs, not the national composite. A carrier with 60% of its cat exposure in Southeast wind will see a different treaty renewal outcome than a national writer with diversified exposure across convective storm, earthquake, and hurricane perils.

Time-Weighting for Mid-Term Treaty Renewals

Most property catastrophe treaties renew on January 1 or June 1/July 1. If the rate filing uses a prospective policy period that spans the treaty renewal date, the pricing actuary needs a time-weighted blend of the expiring and renewing treaty costs.

For example, if the filed rates take effect October 1, 2026 through September 30, 2027, and the treaty renews January 1, 2027, the cat load should reflect three months of the 2026 treaty cost and nine months of the 2027 treaty cost, weighted by earned premium exposure in each period. The formula is:

Blended Reinsurance Cost = (3/12 × 2026 Treaty Cost) + (9/12 × 2027 Treaty Cost)

The challenge is that the 2027 treaty cost is unknown at filing time. The actuary must select a projected reinsurance cost for the second period. In a softening market, the conservative approach is to assume the current rate decline persists (using the 14% decline as the starting point and adjusting for expected capacity trends). The aggressive approach is to assume further softening. Most state DOIs expect the actuary to document the basis for the projection and to use a reasonable central estimate rather than cherry-picking the most favorable scenario.

Historical context informs the projection. At the 2016 soft-market floor, the ROL Index declined only 7% annually. The current 14% decline is historically anomalous and occurred alongside record reinsurer capital of $785 billion (up roughly 10% year-over-year) and record third-party capital of $136 billion. Whether this pace of softening continues through January 2027 depends on hurricane season losses and capital market conditions. The actuary should use a multi-year average of reinsurance cost trends rather than extrapolating the most recent data point.

Putting It Together: The Adjusted Cat Load Walk

The complete adjustment from the expiring cat load to the filed cat load involves six steps:

Recalculate the layer-by-layer reinsurance cost using actual renewal terms, including reinstatement provisions, co-participation changes, and any new supplementary covers.
Re-derive the net modeled AAL using the updated attachment point, retention, and limit structure. If the attachment point did not move, the net AAL changes only if co-participation percentages shifted.
Update the market-implied component of the credibility blend using the new treaty cost and document the blend weight selection.
Time-weight the expiring and prospective treaty costs if the policy period spans a treaty renewal date.
Allocate by state and line using the cat model's geographic distribution of expected losses, not a flat pro rata allocation.
Document the walk from the prior filing's cat load to the current filing, showing each component's contribution. Regulators reviewing a cat load decrease want to see that the reduction is mechanically justified, not simply a pass-through of a market headline.

For a mid-size homeowners carrier with $500 million in subject premium and a $225 million cat program, the illustrative walk might look like this: the expiring cat load of 8.2% decomposes into 4.5 points of net modeled AAL and 3.7 points of allocated reinsurance cost. The renewal reduces the allocated reinsurance cost by 1.0 point (from 3.7 to 2.7, reflecting the 13.7% blended program decline partially offset by new frequency cover purchases). The net modeled AAL decreases 0.2 points due to a modest reduction in co-participation. The filed cat load moves from 8.2% to 7.0%, a 1.2-point or 15% decline in the filed provision, even though the headline reinsurance market declined 14%.

The numbers will differ for every carrier, state, and peril region. The methodology should not.

Sources