Behavioral Claim Velocity Attribution

What is this report? When a physical hazard (hail, wind) damages a neighborhood, the actual insurance claims filed often differ from what the damage alone would predict. This analysis measures that difference — called claim velocity — and traces it back to specific neighborhood characteristics.

Baseline = 1.0×. A neighborhood at 1.0× files exactly as many claims as the physical damage model predicts. A neighborhood at 1.5× files 50% more claims than the damage alone would suggest. A neighborhood at 0.7× files 30% fewer.

Velocity is clipped at 3.5× to prevent extreme outliers from distorting pricing. The model is calibrated on Texas homeowner wind and hail claims (2018–2023) from the TX Department of Insurance.

Four neighborhood types (cohorts) are identified by unsupervised clustering: Stable · Transitional · Stressed · Assertive
TX TDI calibrated multipliers: Assertive 1.344× · Stressed 1.310× · Transitional 1.031× · Stable 0.865×

Part 1A — How Each Index Moves Claim Velocity by Cohort

The model uses four composite neighborhood indexes, each built from dozens of raw signals. This table shows how much each index raises or lowers claim velocity for each cohort — relative to the median-level neighborhood (which anchors at 1.0×). Hover over any signal name for a full explanation. Click any column header to sort.

Claim Frequency Effect (how often claims are filed)

Signal	Model Strength	Effect Direction	Stable	Transitional	Stressed	Assertive
Dispute Culture Index	+0.836	↑ raises	1.110×	1.000×	1.810×	2.648×
Maintenance Index	-1.216	↓ lowers	0.492×	0.991×	2.136×	3.500×^⌈

Claim Severity Effect (how large individual claims are)

Signal	Model Strength	Effect Direction	Stable	Transitional	Stressed	Assertive
Dispute Culture Index	-0.000	≈ 0 (ns)	1.000×	1.000×	1.000×	1.000×
Maintenance Index	+0.109	↑ raises	1.066×	1.001×	0.934×	0.862×

Part 1B — Individual Signal Contributions (Chain Estimate)

Each composite index is built from raw neighborhood signals. This table traces velocity contributions down to individual signals via chain propagation: a signal's effect = its weight in the parent index × the parent index's model coefficient. These are approximate estimates — not directly calibrated.

Source: Dallas–Fort Worth (1,559 census tracts). Hover signal names for definitions.

Signal	Parent Composite Index	Chain Strength	Stable	Transitional	Stressed	Assertive
Attorney × Complaint Interaction	Dispute Culture Index	+0.0585	1.009×	1.004×	1.005×	1.052×
CFPB Auto Complaint Rate	Dispute Culture Index	+0.0836	1.015×	1.021×	0.992×	1.072×
CFPB Complaint Rate	Dispute Culture Index	+0.2508	0.990×	1.076×	1.019×	1.203×
Claiming Culture Index	Dispute Culture Index	+0.2090	1.070×	1.055×	1.067×	1.261×
Civil Court Filing Rate	Dispute Culture Index	+0.0251	0.991×	1.003×	1.018×	1.009×
Dispatch Dispute Rate	Dispute Culture Index	+0.0418	1.000×	1.000×	1.000×	1.000×
State Insurance Complaint Rate	Dispute Culture Index	+0.1672	1.000×	1.000×	1.000×	1.000×
College Education Rate	Maintenance Index	-0.0392	0.957×	1.007×	1.043×	0.981×
Discretionary Services Index	Maintenance Index	-0.0784	0.981×	1.006×	0.995×	0.951×
Elderly Share (65+)	Maintenance Index	+0.0392	1.024×	1.010×	0.996×	0.988×
Home Improvement Loan Rate	Maintenance Index	-0.1961	0.727×	0.978×	1.070×	1.044×
In-Migration Rate	Maintenance Index	+0.0392	0.991×	1.001×	1.009×	1.044×
Long-Term Resident Share	Maintenance Index	-0.1569	0.910×	0.944×	1.045×	1.197×
Maintenance Response Index	Maintenance Index	-0.1177	1.000×	1.000×	1.000×	1.000×
Median Housing Age	Maintenance Index	+0.0784	1.015×	1.006×	1.000×	1.008×
Mobile Home Share	Maintenance Index	+0.0392	1.008×	1.018×	1.019×	1.005×
Owner Occupancy Rate	Maintenance Index	-0.1961	0.868×	0.966×	1.154×	1.354×
Veteran Share	Maintenance Index	-0.1177	0.950×	0.949×	1.050×	1.026×

Part 2 — Velocity Distribution Across All Tracts

Every census tract in the national dataset is run through the model to get a predicted velocity. This section shows how those velocities are distributed within each cohort — not just the average, but the full spread from the 5th to 95th percentile. The 90% confidence interval on the average is computed by resampling 1,000 times (bootstrap).

Cohort	Average Velocity	90% Confidence Range	Tract Range (P5–P95)	Median Tract	Tracts in Cohort
Stable	0.923×	0.915× – 0.931×	0.500× – 3.021×	0.500×	27,476
Transitional	1.427×	1.414× – 1.438×	0.500× – 3.500×	0.784×	26,434
Stressed	2.580×	2.568× – 2.593×	0.632× – 3.500×	3.336×	17,777
Assertive	3.291×	3.283× – 3.300×	1.746× – 3.500×	3.500×	11,321

Assertive cohort note: At the new ceiling of 3.5×, Assertive tracts may still saturate — the GLM predicts very high velocities for this cohort before any clipping. The TX TDI calibrated mean for Assertive is 1.344×, which is the empirically validated anchor. Individual-tract GLM predictions represent theoretical maximums, not observed averages.

Part 3 — Pricing Leverage by Signal

For each signal, all other signals are held at their median value while this one moves from low to extreme. This shows the standalone velocity curve for each signal — which one has the most pricing leverage, and where the non-linearity kicks in.

Pricing Leverage is Q95 velocity ÷ Q50 (median) velocity. A signal with leverage of 3.0× means an extreme-level neighborhood is priced at triple the velocity of an average one, based on this signal alone. The ⌈ marker indicates the model ceiling (3.5×) was hit — the true uncapped value is higher.

Protective signals (negative β) appear with leverage below 1.0× — meaning the Q95 tract has lower velocity than the median. Maintenance Index (β=−1.216) is the single strongest driver in the model, but its leverage ratio is inverted: a neighborhood at the 95th percentile for maintenance behaves far better than average, not worse. The leverage floor is just as valuable as the ceiling of risk signals.

Signal	Pricing Leverage Q95 ÷ Q50	10th pct low	25th pct below avg	Median baseline	75th pct above avg	95th pct extreme
Dispute Culture Index	3.50× (8.6× uncapped)	0.618×	0.695×	1.000×	1.857×	3.500×^⌈8.6×
Maintenance Index	0.50×	3.500×^⌈6.3×	2.610×	1.000×	0.500×	0.500×

Deferred: Multi-signal combination analysis (grid search, co-amplification) requires all 4 GLM predictors in the national dataset. Claiming Culture Index and Attorney & Adjuster Density are currently absent from the national geoparquet — they are only available in city-level feature matrices. Once added, the combination analysis will show whether high dispute culture and high attorney density together compound risk non-linearly.