PulsCar analysis of NHTSA public data, September 2026 Topic: catalytic converters - owner complaints, recall campaigns and manufacturer communications License: CC BY 4.0 (attribution: "PulsCar analysis of NHTSA public data, September 2026") SOURCES - NHTSA Office of Defects Investigation flat file COMPLAINTS (FLAT_CMPL), field dictionary https://static.nhtsa.gov/odi/ffdd/cmpl/CMPL.txt 777,040 owner complaints received 1 January 2015 - 8 September 2026. - NHTSA ODI flat file RECALLS (FLAT_RCL), dictionary https://static.nhtsa.gov/odi/ffdd/rcl/RCL.txt - NHTSA ODI flat file MANUFACTURER COMMUNICATIONS (FLAT_TSBS), dictionary https://static.nhtsa.gov/odi/ffdd/tsbs/TSBS.txt - Downloaded and indexed 17 September 2026; queries run 18 September 2026. QUERY - "a complaint about a catalytic converter" Step 1. Full-text search for catalytic OR "p0420" OR "p0430". -> 3,736 complaints. Step 2. Motorcycle and powersports makes removed. -> 3,727. Step 3. Strong-phrase test. The narrative must contain one of: "catalytic converter", "catalytic-converter", "cat converter", "P0420", "P0430", "catalyst efficiency", "catalyst below threshold". The bare word "converter" is NOT accepted, because in this database it far more often means a torque converter, a DC-DC converter or a converter clutch. -> 3,490 complaints, received 3 January 2015 to 8 September 2026. This is the published set. SAMPLE READING (required before publishing any keyword count) Thirty random matches were read in full. Twenty-eight were genuinely about a catalytic converter, a P0420/P0430 code, or the upstream fault an owner blamed for killing one. Two were about something else and mentioned a converter in passing. The observed false-positive rate is therefore about 5-7% in a sample of 30. The bare-word "converter" exclusion in Step 3 was written after a first pass in which torque-converter complaints were arriving in the set. FIGURES QUOTED IN THE ARTICLE (published set, n = 3,490) Fire 2.49% (87) against a 1.82% baseline over all 777,040 complaints. Crash 0.89% against 4.37%. Towed 4.96% against 5.11%. 51 injuries, 0 deaths. Mileage: median 100,000; quartiles 70,000 and 130,000 (n = 1,722 with a usable odometer reading; miles <= 0 or >= 400,000 dropped). Baseline median 70,000. 570 of the 1,722 (33.1%) are at or under 80,000 miles; 52.6% at or under 100,000; 14.0% at or above 150,000. IMPORTANT on the 33.1%: mileage alone does not establish eligibility for the federal emissions warranty in 40 CFR 85.2103, which expires at 8 years OR 80,000 miles, whichever comes first, counted from the date of delivery to the first purchaser. The complaint file carries no in-service date, so 33.1% is an upper bound on the share that could have been inside the mileage half of that window, not a count of covered repairs. The article states this limit in its own text. What the narrative blames (a complaint may name more than one): oil consumption or burning oil 448 (12.8%); coolant, head gasket or engine internals 323 (9.3%); an oxygen sensor 273 (7.8%); a misfire 238 (6.8%); a rich condition or fuel-trim code 169 (4.8%); an exhaust leak 50 (1.4%). Context: mentions a recall, extended warranty, service campaign or class action 1,409 (40.4%); describes a stall or loss of power 786 (22.5%); describes heat, smoke, a burning smell or fire 731 (20.9%); says coverage or a claim was refused 292 (8.4%); mentions the 8-year/80,000-mile federal emissions warranty 141 (4.0%); second or third converter on the same car 65 (1.9%); failed an emissions or smog test 49 (1.4%); describes theft of the converter 22 (0.6%). Owner-quoted dollar figures: every "$N" between $100 and $20,000 in a narrative was collected - 683 figures, median $2,000, quartiles $900 and $4,000, 53.4% at or above $2,000. These are NOT prices; they are amounts owners typed into a safety complaint. Concentration: Kia 916 complaints (26.2% of the set), of which the 2012-2016 Soul alone is 751; Ford 419 (12.0%); Chevrolet 416 (11.9%); Hyundai 315 (9.0%). Oil consumption is named in 46% of the 125 Hyundai Tucson complaints, 41% of the 97 Chevrolet Equinox complaints and 36% of the 56 GMC Terrain complaints, against 12.8% across the whole set. THEFT - WHAT THIS DATASET IS NOT Only 22 of 3,490 complaints (0.6%) describe a converter being stolen, and that is a fact about the dataset rather than about theft. NHTSA's complaint database exists for vehicle DEFECTS; a theft is a police and insurance matter and almost nobody reports one here. Every theft figure in the article therefore comes from a named outside source with its date, not from this analysis. QUERY - recall campaigns Rows in FLAT_RCL whose component names a catalytic converter, or whose defect or consequence text names one. Motorcycle and powersports makes removed. Counted DISTINCT on the campaign number. -> 12 campaigns, 1,187,137 potentially affected vehicles, of which 10 mention fire in the defect or consequence text (file converter-recall-campaigns.csv). Note that some of these campaigns are not about a defective converter but about something else that reaches a hot converter - Honda 12V-222 is a power steering hose leaking onto one - and the CSV keeps the defect text so that distinction stays visible. QUERY - manufacturer communications Rows in FLAT_TSBS whose summary names a catalytic converter, a catalyst, P0420 or P0430. Volkswagen's monthly "VIN4APIN" index documents removed. Counted DISTINCT on NHTSA_ID. -> 1,403 documents (file converter-bulletins-by-theme.csv). 629 (44.8%) put software, a reprogram or a calibration in the remedy; 367 (26.2%) name P0420 or P0430; 182 (13.0%) name a misfire; 138 (9.8%) name an oxygen or air-fuel-ratio sensor; 123 (8.8%) name a leak; 50 (3.6%) tell the technician the condition is normal or to check something before replacing the converter; 21 (1.5%) concern theft or a theft-deterrent shield; 10 (0.7%) name oil consumption. Separately: 11 documents in the whole file use the phrase "incorrectly illuminated" MIL together with P0420 or P0430. All eleven are Ford, published between 12 March 2025 and 10 August 2026, and each one's remedy is a PCM software update. LIMITATIONS - Complaints are self-reported, unverified, and are not a failure rate. There is no denominator, nothing here is normalized, and no make or model is called reliable or unreliable on the strength of these counts. - Kia's 26.2% share is one nameplate and one recall population (2012-2016 Soul with the 1.6L GDI engine, recall 19V-120 and the later SC176/SC176Y service actions). It says nothing about Kia's converters generally and the article does not use it that way. - Only 1,722 of 3,490 complaints carry a usable odometer reading, so every mileage figure describes that subset. - "What the narrative blames" is what the owner wrote, not a verified cause. Owners who know about their car's oil-consumption bulletin write about oil consumption; owners who do not, do not. - Bulletin theme counts are counts of what a one-paragraph summary says. The documents quoted in the article were opened individually as PDFs. FILES converter-complaints-by-make.csv complaints, median failure mileage, share at or under 80,000 miles converter-top-models.csv nameplates with 5+ complaints and the share naming oil consumption converter-what-owners-blame.csv upstream conditions named in the narratives converter-context-and-outcomes.csv context words and towed/crash/fire against baseline converter-owner-reported-amounts.csv owner-typed dollar figures, bucketed converter-recall-campaigns.csv 12 campaigns, with whether the text mentions fire converter-bulletins-by-theme.csv 1,403 catalyst/P0420 bulletins by theme and by make ------------------------------------------------------------------------------ CASE SENSITIVITY — SECTION ADDED 19 SEPTEMBER 2026 ------------------------------------------------------------------------------ NHTSA stores complaint narratives in ALL CAPS for records filed through roughly 2020 and in mixed case for records filed from 2021 on. A case-SENSITIVE search of the narrative field loses about 92% of the complaints filed after 2020, so this is the single largest way an analysis of this database can go wrong while still looking finished. Every query behind this file is case-insensitive. SQLite's LIKE and its FTS5 full-text index both ignore case by default, and no query in this work uses GLOB or any case-sensitive comparison. This was not taken on trust. On 19 September 2026 every published folder was re-checked against the same database, comparing the figures here with what a case-sensitive query would have returned. The result: the complaint set published here (3,736) is LARGER than the absolute ceiling a case-sensitive search of the same terms could have returned (1,383 narratives). A case-sensitive query cannot produce a bigger set than itself, so the bug is ruled out for this folder by arithmetic, not by assurance.