Texas insurance claims outlook, Q3 2026: what changed and our forecasts to 2027

The forecasts in brief

1. Quarter in numbers

Three things stand out.

Severity is moving, frequency is not. Texas road deaths fell 9.38% in 2025 while miles driven rose 1.79%, taking the fatality rate to 1.20 per 100 million miles, the lowest in our 2016 to 2025 series [10]. The pressure is in what each claim costs.

Legal and hospital costs lead. Legal services prices rose 7.4% (adverse for insurers) and hospital services 5.4% (adverse for insurers) over the year to September, against 1.5% for medical care overall [1][2]. Bodily injury severity follows these lines more closely than it follows headline medical CPI.

Weather was above normal. SPC logged 333 (adverse for insurers) preliminary hail and thunderstorm wind reports in Texas in Q3, more than any Q3 since 2016 except 2023 (520) [4][5]. Wind made up 284 (adverse for insurers) of them, against a Q3 average of 190.8 [4][5]. Hail was below normal at 49 (favorable for insurers) reports, against 64.6 [4][5].

July flooding was extreme. NOAA's Storm Events database records 160 (adverse for insurers) flood and flash flood events in Texas in July 2026, after 186 (adverse for insurers) in July 2025 [6]. No other July from 2016 to 2024 had more than 99 [6]. NOAA's property damage estimate for July 2026 flood and flash flood events is $172.4 million, and it records 4 direct deaths from flash flooding [6]. NOAA damage estimates are partial and are not insured loss figures. FEMA declared a major disaster, DR-4935, on 1 September for 26 counties in the Hill Country, South Texas and the Edwards Plateau [7].

2. Texas claims pressure scorecard

The scorecard is a simple, open index. Each component is scored against its own history. A score of 50 means the component is at its normal level. Each 10 points above or below 50 is one standard deviation from normal. The Texas score is the plain average of the components.

Sources: [1] to [6]. Baselines: CPI components use September readings for 2018 to 2025, PPI components use 2017 to 2025, storm and flood components use 2016 to 2025.

Texas versus the U.S.

Six components can be measured for both Texas and the U.S. Four use national price indexes for both, because BLS does not publish Texas indexes for repair, parts, legal services or construction materials. The two that differ are medical cost (Dallas-Fort Worth against the U.S.) and severe storms (Texas reports against all U.S. reports).

On these six shared components Texas scores 54.4 (adverse for insurers) and the U.S. scores 52.8 (adverse for insurers). The gap comes from medical costs: Dallas-Fort Worth medical inflation scored 57.7 (adverse for insurers) against 45.2 (favorable for insurers) for the U.S. Severe weather was above normal in both, with 11,966 preliminary U.S. hail and wind reports in Q3 against a 2016 to 2025 average of 9,690.9 [4][5]. Flood is excluded from the comparison because we do not yet have a like-for-like national July series.

Pressure by line

3. Forecasts for Q4 2026 and 2027

The review above covers what happened. This section covers what we expect next. All forecasts use public data and methods we can show in full. Each one gives a central number and an 80% range: we expect the outcome to land inside the range four times in five.

3.1 Claim cost and severity

We forecast each cost index with two simple models and average them. The first carries the last 12 months' average rate of change forward. The second is an ARIMA model, a standard statistical model that learns from the series' own recent pattern. We tested the combined forecast on every month from January 2012 and set each range from the errors it made in that test (162 test points at the longest horizon).

Sources: [1][2][3][15]. Inaza forecasts. Ranges are 80%.

What drives these numbers:

• Repair peaks in December, then eases. December 2025 was a weak month for the repair index (439.8, down from 445.3 in November), so the year-on-year rate jumps in December 2026 on that base alone [1]. After that, repair inflation settles at about 5% by the end of 2027. Parts stay close to flat. Labor and calibration, not parts, carry repair costs.
• Legal costs ease on paper, not in level. The legal services index jumped from 316.1 in December 2025 to 336.5 in February 2026 [2]. When those months drop out of the comparison, the annual rate falls to about 4% in mid 2027. The level stays high and the rate climbs back toward 6% by December 2027.
• Hospital costs stay near 5%. Hospital services inflation has run between 3.2% and 7.9% since January 2023, and our range stays inside that band [1]. Headline medical CPI, at about 2%, understates injury claim costs.
• Used vehicle prices are the wild card. They set total loss values and are the most volatile series here. The range for December 2027 runs from -12.4% to +11.2%.
• Construction materials cool but stay above wages. Materials inflation should ease from 12.0% in December 2026 to about 7% by late 2027, with a very wide range. Construction wages are steadier at 3.6% to 3.9% [3][15].

By line. We combine these into one claim cost basket per line. Weights are set out in the methodology.

3.2 Frequency

Miles driven. Texas drove 314.0 billion vehicle miles in 2025, up 1.79% [10]. U.S. miles for January to August 2026 were 0.53% above the same months of 2025 [13]. Texas mileage grew faster than U.S. mileage in 17 of the 22 years from 2004 to 2025 [10][13], so we forecast Texas from the U.S. figure with a simple regression on 2004 to 2025 (excluding 2020 and 2021). Our forecast for 2026 is 317 billion miles (311.5 to 323 billion), growth of 1.1% (-0.8% to +3.0%). For 2027 we expect 321 billion (310 to 331 billion).

Injury rate. TxDOT counted 244,534 people injured in reportable crashes in 2025, or 77.9 per 100 million miles [12]. That rate has fallen about 2.1% a year since 2016 (80% confidence: 1.9% to 2.4% a year), excluding 2020 [12]. We forecast 76.9 (favorable for insurers) in 2026 (74.5 to 79.4) and 75.3 (favorable for insurers) in 2027 (72.8 to 77.9).

What it means for claims. Fewer injuries per mile and modestly more miles leave the number of injured people roughly flat: about 244,000 in 2026 (235,000 to 253,000) and 242,000 in 2027 (230,000 to 253,000). Per mile driven, we expect injury crash frequency to fall about 2.1% (favorable for insurers) in 2027 (range -5.3% to +1.2%). That helps the loss ratio, but it offsets only a third to a half of the severity trend in section 3.1.

The 2025 fatality rate of 1.20 per 100 million miles was the lowest in our 2016 to 2025 series [10]. One good year is not a trend, so we do not forecast a further drop in deaths. We use the injury rate, which tracks claim counts more closely.

3.3 Catastrophe and weather

A very strong El Niño. NOAA's Climate Prediction Center expects the current El Niño to strengthen further, with an 80% chance of at least a strong event lasting through January to March 2027 [19]. The Oceanic Niño Index reached +2.16 for July to September 2026 [20]. For November to January, CPC favors above-normal precipitation across the central and southern Plains and equal chances of above, near or below normal temperatures over parts of the southern Plains [19].

Hail: plan on a normal spring, which is a big one. We tested whether El Niño changes Texas hail counts. We fit a regression of Texas hail reports on the winter El Niño index and a time trend for 1996 to 2025 [5][20]. The El Niño effect was close to zero: -1% per degree for Q1 (80% confidence: -13% to +12%) and -1% for Q2 (-6% to +3%). The three strong El Niño winters in that period gave Q1 counts of 474 (1998), 244 (2016) and 199 (2024) [5]. So we base the hail outlook on 2016 to 2025.

Q4 2026 adds October's preliminary count (11 hail reports for the full month, 5 of them 2 inches or larger) to November and December outcomes from 2016 to 2025 [4][5]. Q1 and Q2 2027 use the median and the 10th to 90th percentile of 2016 to 2025 [5].

Q2 is when Texas hail losses happen. April to June has averaged about five times as many hail reports as January to March [5]. Large hail of 2 inches or more, the size that damages roofs and vehicles, averaged 168.6 reports in Q2 [5]. One large hail event over a major metro area can outweigh the rest of the quarter.

Hurricane season close-out. No hurricane made landfall in Texas in 2026 through 30 October. Tropical storms Arthur in June and Edouard in early September brought heavy rain and flooding to southeast Texas [22]. In our scan of NOAA's HURDAT2 landfall records for 1851 to 2025, no tropical storm or hurricane has come ashore on the Texas coast after 17 October [21]. The latest hurricane landfalls were on 16 October, in 1912 and 1989 [21]. We put the chance of a Texas landfall before the season ends on 30 November at close to zero.

Flood. July flooding was extreme in both 2025 and 2026 [6]. With CPC favoring a wetter than normal winter in the southern Plains, we would expect more flood and wet-road claims than a typical winter. We do not put a number on this. Flood claim counts depend on where rain falls more than on seasonal totals.

3.4 Premium and rate trend

Price indexes tell the same story from three angles. Claim costs are rising faster than premiums for all three lines.

Sources: [1][14]. Inaza forecasts using the same method as section 3.1.

Texas personal auto filings point to a turn. TDI's statewide average personal auto rate change was +4.8% in 2024 and +0.9% in 2025 [17]. In TDI's rate filing data, personal auto filings received from January to 30 October 2026 had a median change of +0.5%, and 56% were increases [16]. That compares with -0.1% and 50% in 2025 [16]. The 42 filings with renewal effective dates in Q4 2026 had a median change of +2.1%, and 71% were increases [16]. These are counts of filings, not weighted by premium, so they show direction rather than the market average.

Our read. Personal auto prices bottom out in Q4 2026 and start to rise in 2027, but slowly. The ranges are wide because insurance price indexes swing with the underwriting cycle. Commercial property premiums keep rising at about 3%, behind rebuild costs.

3.5 Regulatory items with known dates

What we expect. Claim costs and litigation will be on the 2027 session agenda. Senate Bill 30, which would have limited medical damages evidence in injury suits, failed in 2025, and TDI's Speak Up, Texas participants named affordability and claims handling as their top problems. We cannot forecast the outcome of any bill. We do expect the SB 458 appraisal workflow to be the first new compliance cost in 2027.

4. What it means for loss ratio and combined ratio

Combined ratio is loss ratio plus expense ratio. We estimate the loss ratio effect of 12 months of cost, frequency and price trends for each line, using the forecasts above. We apply that to each line's 2024 Texas loss ratio from TDI, the latest available [18]. This is a trend estimate, not a loss ratio forecast. It leaves out catastrophe losses, reserve changes and the lag between rate changes and earned premium. If expense ratios hold, the combined ratio moves by the same number of points.

Sources: Inaza model on [1][2][3][10][12][14][15]; base loss ratios from TDI [18], earned to incurred, loss adjustment expense excluded. Commercial property uses the commercial fire and allied lines ratio.

Personal auto

Loss ratio: the gap is widest now. In the 12 months to December 2026, claim costs rise 6.6% (adverse for insurers). On price, the motor vehicle insurance CPI falls 4.8% while the private passenger auto insurance PPI rises 0.6%, and the model draws equally from the two [1][14]. Frequency per mile falls about 2.1% (favorable for insurers). Together that is about 4 points (adverse for insurers) of loss ratio pressure on a 65.6% base. In 2027 the pressure shrinks to about 1 point as price indexes turn up, but the range runs from -4.8 to +7.3 points because price is so uncertain.

Expense ratio: new compliance work. The SB 458 appraisal rules, adopted on 6 November, require appraisal notices in claim decision letters and deadline tracking for policies from 1 March 2027 [25]. Since 15 October, TDI has published personal auto nonrenewal and declination reasons by ZIP code under HB 2067 [26]. Legal costs, forecast at +8.4% in December, also raise defense and loss adjustment costs [2].

Commercial auto

Loss ratio: highest base, steady pressure. Commercial auto starts from the highest loss ratio of the three lines at 78.1% [18]. Its claim cost basket rises 6.3% to December 2026, with legal services at +8.4%, while the commercial auto insurance price index falls 1.1% [2][14]. That is about 4 points of pressure. In 2027 we expect about 1.7 points, with a 68% chance of a rise.

Expense ratio: verification work. Fictitious pickups, where deception is used to obtain a shipment, made up about a quarter of U.S. and Canadian cargo theft incidents in Q3 2026 [8]. Verifying carriers and documents adds handling time. Fleets that operate alongside the 3,164 authorized automated vehicles may face new liability questions [9].

Commercial property

Loss ratio: rebuild costs outrun premium. Our rebuild cost basket rises 7.9% to December 2026, against 2.1% for the commercial multiple peril price index [3][14][15]. That is about 3.7 points of pressure before any catastrophe losses. In 2027 it narrows to about 0.8 points, with a wide range because materials prices are volatile. Spring hail is the bigger swing factor, and it is not in these numbers.

Expense ratio: catastrophe surge. Flood and hail events bring claim volume peaks that strain adjusters and vendors.

Summary

The point estimates come from our model. The ratings and the operational levers are our judgment.

5. Methodology

• Scores. Score = 50 + 10 × (current value minus baseline mean) ÷ baseline standard deviation. Price components use the September year-on-year change. Storm components use counts. All components have equal weight. Line scores are plain averages of the components listed for each line.
• Data cutoff. Forecasts use price indexes through September 2026, U.S. miles driven through August 2026 and SPC preliminary storm reports for the full month of October 2026.
• Missing CPI months. BLS did not publish October 2025 CPI, and some series also lack November 2025. We filled missing months by straight-line interpolation of the log index. No published figure in this report uses an interpolated value. We do not forecast body work separately because 13 of its recent months are missing.
• Cost and price forecasts. For each monthly index we average two forecasts of the log index: (1) the latest value carried forward at the average monthly change of the last 12 months, and (2) an ARIMA(p,1,q) model with drift, with p and q from 0 to 2 chosen by AIC on the full series. Ranges are the 10th to 90th percentile of the errors the same combined method made on every forecast origin from January 2012, measured in year-on-year percentage points at each horizon. History runs from January 2007.
• Claim cost baskets. Personal auto: 40% physical damage (75% vehicle repair CPI, 25% used vehicles CPI), 30% liability vehicle damage (vehicle repair CPI) and 30% injury. Commercial auto: 15% physical damage, 25.5% liability vehicle damage and 59.5% injury. The physical damage and liability shares follow Texas direct losses paid in 2025 [18]. The 50/50 split of personal auto liability between vehicle damage and injury, and 30/70 for commercial auto, are our assumptions. Injury is an equal mix of hospital services CPI, physicians' services CPI and legal services PPI. Commercial property rebuild cost is half construction materials PPI and half average hourly earnings in construction.
• Frequency. Texas miles driven growth = 0.43 + 1.66 × U.S. growth (2004 to 2025, excluding 2020 and 2021; R² 0.66). U.S. 2026 growth is the January to August year-to-date growth, with uncertainty from how far full-year growth has differed from January to August growth since 2012. The injury rate forecast is a log-linear trend on 2016 to 2025, excluding 2020 (R² 0.94), with an 80% prediction interval.
• Hail. Counts are Texas hail reports in the SPC severe weather database. The El Niño test is a negative binomial regression of quarterly counts on a time trend and the December to February Oceanic Niño Index, 1996 to 2025. Outlook ranges are the 10th to 90th percentile of 2016 to 2025.
• Hurricanes. Texas landfalls are HURDAT2 landfall records ("L") for tropical storms and hurricanes between 97.6°W and 93.75°W and 25.8°N to 30.2°N.
• Loss ratio pressure. Change = base loss ratio × [(1 + severity) × (1 + frequency) ÷ (1 + price) - 1]. Severity is the line's claim cost basket. Frequency is the per-mile injury trend for the auto lines and zero for commercial property. Price is the line's insurance price index; personal auto draws equally from the CPI and PPI measures. We combine the ranges by simulation (50,000 draws), treating each forecast as independent. Real outcomes are often correlated, so the true range may be wider.
• Limits. Most cost and price indexes are national. Storm report counts measure meteorological activity, not insured losses. Filing counts are not premium-weighted. Nothing in this report uses Inaza client data.