NAIC Homeowners Data Call Brings ZIP-Level Scrutiny to Territorial Rate Filings

From tracking NAIC property data initiatives since the 2020 wildfire season, the 2026 Homeowners Market Data Call represents the most consequential regulatory data collection ever directed at homeowners pricing. On March 25, 2026, the NAIC issued letters to homeowners insurers in 50 participating jurisdictions directing them to submit policy-year data for 2018 through 2025 at the ZIP-code level, with peril-specific loss splits, deductible distributions, and mitigation discount detail. The original June 15 deadline was extended to July 15, 2026 on May 21, acknowledging the submission's unprecedented scope (NAIC). Every insurer writing at least $50,000 in direct written premium on homeowners lines (Annual Statement Line 4, or forms DP-1 through HO-8) in any participating jurisdiction must submit. Florida Insurance Commissioner Mike Yaworsky stated the initiative will "help equip us with even more information, tools, and resources to not only speed resilience but also increase preparation before severe weather hits."

The NAIC plans to publish a comprehensive public report in early 2027. Our companion article covered the regulatory structure and compliance dimensions of the data call in detail. This article focuses on what the resulting dataset means for territorial ratemaking methodology and how it shifts the information balance between carriers and regulators in rate filing reviews.

How Territorial Ratemaking Works Today

Traditional territorial ratemaking in homeowners insurance follows a four-step process that has been the actuarial standard for decades.

Step 1: Define rating territories. Carriers group ZIP codes with similar expected loss characteristics into territories. The grouping uses clustering algorithms (k-means, CART-based spatial clustering), geographic contiguity rules, or both, aiming for homogeneous loss potential within each territory and maximum separation between territories. A typical state homeowners rating plan might contain 15 to 50 territories, each comprising dozens to hundreds of ZIP codes. The grouping decision balances actuarial precision against practical constraints: more territories produce better risk differentiation but thinner data per territory.

Step 2: Calculate territory-level loss costs. For each territory, divide incurred losses by earned premium (or earned house-years) to produce an observed loss cost. This calculation requires enough exposure within the territory to produce a credible estimate, which is where smaller or newly defined territories encounter data limitations.

Step 3: Apply credibility weighting. The standard limited fluctuation credibility formula is:

Z = min(1, √(n / k))

where n is the territory's claim count and k is the full-credibility standard. Using the classical approach with ±5% accuracy at the 90% confidence level, k is typically 1,082 claims. A territory with 200 claims receives Z = √(200/1082) = 0.43, meaning 43% weight goes to the territory's own experience and 57% to the complement of credibility (usually the statewide or countrywide loss cost). Only territories generating more than 1,082 claims receive full credibility and stand entirely on their own data.

Step 4: Select territorial relativities. Most state regulations cap the year-over-year change in any single territory's relativity to prevent rate shock, which means selected factors lag indicated factors during periods of rapidly shifting geographic loss patterns. Pricing actuaries balance the actuarial indication against transition constraints, filing the maximum supportable change within regulatory limits.

This framework operates under a critical assumption: that regulators reviewing filed territorial factors have less granular data than the carrier. That assumption is about to break.

What ZIP-Level Data Changes for Pricing Actuaries

The data call collects premiums, claims, and losses at the individual ZIP-code level across all participating carriers. Once the NAIC aggregates this data, regulators gain three capabilities they have never had.

Testing territory homogeneity. Regulators can now check whether a carrier's territory groupings genuinely reflect homogeneous loss potential or mask geographic cross-subsidies. If a carrier groups five ZIP codes into a single territory and assigns a relativity of 1.20, but the all-industry ZIP-level data shows two of those ZIPs running at 0.80 and three at 1.50, the regulator has evidence that the territory definition itself is problematic. The low-risk ZIPs are subsidizing the high-risk ZIPs within the same territory, and the regulator can quantify the cross-subsidy precisely.

Benchmarking territorial factors against all-industry data. Carriers currently file territorial factors supported by their own data, sometimes supplemented by ISO or AAIS advisory loss costs. Regulators review those factors primarily against the carrier's own historical experience. With the data call, regulators can reconstruct all-industry ZIP-level loss costs and compare any individual carrier's territorial factors against the industry benchmark. A carrier whose wind territory factor in coastal ZIP codes is 30% below the all-industry figure will face pointed questions about selection methodology and whether the territorial plan is producing adequate rates for the exposure.

Identifying outlier patterns across carriers. By comparing peril-specific territorial patterns across carriers, regulators can spot carriers whose territorial factors diverge significantly from industry norms. Such divergence may indicate adverse selection (the carrier is attracting risk that other carriers have priced more accurately) or inadequate geographic granularity (the carrier's territories are too coarse to reflect actual loss variation). Cross-carrier comparison at the ZIP level was simply not possible before this dataset.

The Credibility Calculus Shifts

The data call's most subtle effect is on credibility weighting. Under the traditional framework, a carrier's own ZIP-level data rarely achieves full credibility because individual ZIP codes produce too few claims. A carrier with 50 policies in a single ZIP code might see zero or one claim per year, far below any reasonable full-credibility standard. That lack of credibility is precisely what forces carriers to aggregate ZIP codes into territories in the first place.

But the all-industry dataset from the data call pools every carrier's experience within each ZIP code. If 15 carriers each write 50 policies in a given ZIP, the all-industry pool contains 750 policies and perhaps 30 to 40 claims per year. Over the eight-year observation window (2018 through 2025), that ZIP accumulates 240 to 320 claims in the pooled dataset. At the full-credibility standard of 1,082, the eight-year pooled credibility for a moderately populated ZIP reaches Z = √(280/1082) = 0.51. For densely populated ZIPs where multiple carriers each write hundreds of policies, the pooled credibility approaches or exceeds full credibility at the individual ZIP level.

This creates an information asymmetry that runs in the opposite direction from the historical norm. Regulators holding moderately credible to fully credible ZIP-level data will review carriers whose territorial definitions aggregate those ZIPs into broader, less granular groups. The burden of proof shifts: instead of the carrier demonstrating that its territorial factors are supported by its own data, the carrier may need to explain why its territory groupings differ from what the all-industry ZIP data suggests. Patterns we have seen in recent rate filing review cycles in auto suggest regulators will use benchmarking data aggressively once they have it.

Peril-Level Splits Enable Two-Stage Cat Load Analysis

The data call requires carriers to decompose losses by peril: wind, hail, fire (excluding wildfire), wildfire, water damage, and other. This decomposition enables a two-stage catastrophe load analysis that regulators could not previously perform with standardized cross-carrier data.

Stage 1: Separate catastrophe losses from attritional losses at the ZIP level. Regulators can isolate wind and wildfire losses (the peril categories most correlated with catastrophe events) from water, fire, and liability losses (predominantly attritional). This separation, done at the ZIP level across eight years of data, produces peril-specific attritional loss costs that are relatively stable year to year and catastrophe loss costs that are volatile but empirically observable.

Stage 2: Compare the carrier's filed cat load against empirical catastrophe loss experience. Currently, rate filers support cat loads using modeled losses from vendors such as Verisk, Moody's RMS, or CoreLogic. Regulators evaluate those modeled cat loads primarily on methodological grounds: Is the model appropriate? Are the parameters reasonable? Were the demand surge and loss amplification factors adequately calibrated?

With the data call, regulators can now compare the modeled cat load against the actual catastrophe loss experience aggregated across all carriers at the ZIP level. If the all-industry data shows $15 per thousand of insured value in annual wind catastrophe losses for a coastal ZIP grouping, and the carrier's modeled cat load produces $8, the regulator has a concrete empirical basis for questioning the discrepancy. This verification layer sits on top of the existing model review process and adds independent loss data that has not previously been available in standardized form across jurisdictions.

Few public datasets currently support this type of peril-by-geography analysis. ISO and AAIS advisory rates provide some benchmarking capability, but not at the peril-by-ZIP granularity the data call will deliver. The NAIC dataset, once published, will be the first nationally consistent source for this type of verification.

Deductible and Mitigation Data Add Filing Complexity

Two additional data elements complicate the rate filing landscape for pricing actuaries.

Deductible distribution data will enable regulators to assess whether carriers' deductible relativities reflect actual loss elimination ratios. The data call collects the count of policies in six deductible buckets for each of three peril categories (all perils, hurricane/named storm, and wind/hail). If a carrier offers a 15% premium credit for a $5,000 wind/hail deductible but the all-industry data shows that the loss elimination ratio for that deductible level is only 8% in the relevant ZIP codes, the carrier is effectively cross-subsidizing high-deductible policyholders at the expense of standard-deductible policyholders. This analysis requires the ZIP-level deductible distribution data the call collects, combined with the peril-specific loss data.

Mitigation discount data creates accountability for credits that may currently lack sufficient actuarial support. Several states require or encourage carriers to offer premium discounts for mitigation measures: fortified roofs, impact-resistant windows, wildfire defensible space. The data call collects both the count of policies receiving each discount type and the average discount percentage across five categories. Combined with peril-specific loss data, regulators can test whether discounted policies actually experience lower losses in aggregate. The NAIC's parallel development of the Strengthen Homes Act model law makes this mitigation data doubly relevant, as that framework requires actuarially justified premium discounts for qualifying improvements and the data call provides the empirical basis for evaluating those justifications.

Non-Renewal Data Connects Pricing to Availability

The data call also collects cancellation and non-renewal counts at the ZIP level. While not directly a pricing variable, this data connects to rate adequacy in ways regulators will use in filing reviews. ZIP codes with high non-renewal rates but actuarially adequate filed rates suggest carriers are managing exposure through underwriting rather than pricing. ZIP codes with both high non-renewals and suppressed rates suggest that regulatory rate constraints are forcing carriers to exit rather than price the risk accurately.

This linkage between pricing adequacy and market availability is precisely the analytical connection the NAIC's Homeowners Market Data Call Task Force was designed to establish (Insurance Journal). Pricing actuaries should expect rate filing reviewers to cite non-renewal data when evaluating whether filed rates are adequate, excessive, or unfairly discriminatory under state rating laws.

Why This Matters

The 2026 Homeowners Market Data Call will produce the first nationally consistent homeowners loss cost database. Its 50-jurisdiction scope enables cross-state benchmarking of territorial factors, catastrophe loads, and trend selections at a ZIP-level granularity that has never been possible before.

For pricing actuaries preparing rate filings in 2027 and beyond, three implications are immediate. First, territory construction methodology will face scrutiny from regulators who can reconstruct ZIP-level loss costs from the all-industry data and question territory groupings that mask significant within-territory loss variation. Second, catastrophe load support will require reconciliation against empirical peril-level data, not just catastrophe model output alone. Third, deductible and mitigation discount adequacy will be testable against industry-wide loss experience rather than accepted on methodological arguments alone.

The data call shifts the regulatory dynamic from "show me your methodology" to "explain why your results differ from what the data shows." For carriers with well-supported territorial plans, tightly calibrated cat loads, and empirically grounded deductible relativities, that shift is manageable. For carriers whose territory definitions are coarse, whose cat loads rely entirely on unvalidated model output, or whose deductible credits exceed empirical loss elimination ratios, the 2027 filing cycle will demand a level of quantitative reconciliation that previous cycles did not require. The early 2027 publication timeline means pricing actuaries have roughly one year to prepare for a regulatory environment where the information gap between filer and reviewer has narrowed substantially.