Autoresearch: Cass August 2026 truckload linehaul and shipments — did the supply-led volume-down / rate-up signature hold? FMCSA/government primary on non-domiciled CDL / ELP scale

Generated by /autoresearch on 2026-09-19. Synthesized across 3 rounds from 11 successfully fetched pages (2 government HTML fetches on transportation.gov failed), anchored by Grokipedia Truckload_shipping. See Provenance. Treat as raw material — review before promoting into a project or thread. Context: vault/projects/stock-market

Summary

The July 2026 Cass pair that the mechanism page treats as the independent-index “supply-led” signature — truckload linehaul (ex-fuel, ex-accessorial) +8.6% y/y with shipments −4.8% y/y — is confirmed on Cass’s own July report (Cass July 2026). The August 2026 print, dated in trade press to Monday 14 Sep 2026 (FreightWaves / Todd Maiden), does not preserve that pair. Linehaul accelerated to index 153.9, +11.3% y/y and +0.7% m/m; shipments turned positive at index 1.038, +2.1% y/y, +5.6% m/m (+5.0% seasonally adjusted) — the first y/y shipment gain after a 42-month downturn (Cass August 2026). Cass/ACT still hesitates to call the volume bounce a major demand improvement. ATA’s August 2026 for-hire tonnage is not published as of this pass (latest official release is July, 18 Aug 2026).

On driver-supply scale, this pass updates the vault’s prior reading of the FMCSA rule. The 13 Feb 2026 final rule on govinfo does contain a driver-supply RIA: FMCSA estimates ~200,000 non-domiciled CDL holders and ~20,000 CLP holders from annual program reviews; remaining ~194,000 “will exit the freight market”; that stock is ~5% of 3.8 million active interstate CDL holders in 2024; exit is assumed over approximately five years as credentials come up for renewal (FR 2026-02965 HTML). That is FMCSA’s arithmetic in the final rule, not only J.B. Hunt’s 5–12% / 214k–437k two-to-three-year model (the latter is still not in this government text). Realized enforcement counts from a DHS release quoting DOT, 31 Aug 2026: over 30,000 illegally issued licenses canceled; over 28,000 ELP out-of-service orders since June 2025 (DHS 31 Aug 2026). A separate ELP NPRM (comments due 9 Oct 2026) estimates ~9,000 incremental annual OOS events in U.S.–Mexico border commercial zones from narrowing the remaining exception — not a nationwide licensed-pool census (FR 2026-16288 HTML).

Findings

Cass August 2026: linehaul still up, shipments no longer down

Cass Information Systems’ August 2026 Transportation Index Report table (cassinfo.com August 2026):

The Truckload Linehaul Index “isolates the linehaul component of full truckload costs from other components (e.g. fuel and accessorials)” and is “a mix of contract and spot rates” from invoices processed by Cass, not necessarily the shipment month; January 2005 is the linehaul base (Cass TL Linehaul Index definition). The shipments index covers all domestic modes, with truckload “more than 50%” and LTL “about 25%” (Cass Freight Index definition; same mix restated on the August report).

Compared with Cass’s own July 2026 table: shipments 0.983, −4.8% y/y, −2.6% m/m (−2.2% SA); expenditures 3.518, +9.1% y/y; linehaul 152.9, +8.6% y/y, +2.3% m/m (Cass July 2026). July’s linehaul and shipment y/y figures match the mechanism-page recap that had been sourced to FreightWaves/Todd Maiden.

FreightWaves (Todd Maiden, Monday 14 Sep 2026) adds two claims not in the Cass table prose: August marked 20 consecutive y/y linehaul increases and the largest y/y increase since June 2022; the index was “up 70 basis points from July” (FreightWaves). Cass’s own August prose says the linehaul rose to 153.9, “up 0.7% m/m and up 11.3% y/y,” that the sequential increase is “in line with expectations and as indicated by the spot market,” and that “even as spot rates slow with modest sequential declines, the much larger contract market is adjusting higher.” Cass: “This index reflects the whole for-hire truckload market, both spot and contract rates.” With no further change, Cass says the 2026 linehaul index is “on pace for a 7% increase” after −10% in 2023, −3.4% in 2024, and +1.8% in 2025 (Cass August 2026).

On the signature question: the rate-up leg held and strengthened (8.6% → 11.3% y/y; 19th consecutive in July becomes 20th consecutive in the FreightWaves recap). The volume-down leg did not hold on a y/y basis. Cass writes that August shipments rose 2.1% y/y, “marking the first y/y gain since January 2023. This ends a 42-month downturn by this measure, the longest on record.” SA shipments rose 5.0% m/m, “essentially reversing declines in June and July.” Cass then qualifies: “As this roughly offsets the declines in the past few months, we hesitate to describe this as a major improvement in freight demand.” The 2-year stacked shipments change remains −7.4% (Cass August 2026). July’s Cass note had said the normal seasonal trend “would put the shipments component … down about 3% y/y in August” (Cass July 2026); the actual print was +2.1% y/y.

Expenditures accelerated to 18.7% y/y (Cass body copy rounds this to “19% y/y”) at 3.72, “driven mostly by higher shipments.” SA, expenditures +6.0% m/m after −2.1% in July; Cass infers that a 5.0% SA shipment gain “imply[s] a ~1.0% increase in overall rates” (Cass August 2026). FreightWaves attributes part of the expenditure jump to diesel “up 46% y/y and 10% sequentially” in addition to the shipment inflection (FreightWaves). That diesel figure is FreightWaves’, not Cass’s table.

ACT Research’s Tim Denoyer, who writes the Cass report, says “the bottom is probably in” and “while likely modest, freight growth should continue,” citing “a restock likely beginning with ocean volumes rising and tariff refunds happening.” In the same August ACT block: “Class 8 tractor sales rose above replacement levels in July and August, allowing the fleet to expand for the first time after tightening for the past 18 months.” He also writes that “capacity constraints from new regulations and broker liability law have raised barriers to entry and will limit the industry’s ability to add capacity,” while “the trajectory of volumes is likely to be supported by an expanding fleet” (Cass August 2026). That is Cass/ACT’s own hedge: regulation still caps how fast capacity can return, but Class 8 above replacement is named as a supply offset that was not in the July “Godot” write-up.

FreightWaves’ “why it matters” line: “August data … highlights a continued escalation in TL contract rates even as demand remains cool. The dataset confirms the impact strict regulatory enforcement is having on non-compliant capacity.” Maiden also notes “road-to-rail conversion has weighed on the trucking-centric dataset in recent months as fuel prices have surged” (FreightWaves). Cass July had already said “rail intermodal is gaining share from trucking this year, also pressuring this index” and that “to a large extent, volumes are still soft because capacity is declining” (Cass July 2026). August does not repeat the July sentence “volumes are still soft because capacity is declining.”

Cass’s public archives page, fetched 19 Sep 2026, still lists July 2026 as the latest monthly blurb and has no August 2026 archive summary (Cass archives). The dedicated August report URL is the primary.

Grokipedia’s Cass Freight Index page restates the publisher methodology (January 1990 = 1.00 for shipments; ~$37B annual freight spend; TL >50% / LTL ~25%) but does not carry the August 2026 print (Grokipedia: Cass Freight Index).

ATA August 2026 for-hire tonnage: not published

As of this pass (19 Sep 2026), ATA has not posted an August 2026 For-Hire Truck Tonnage Index release. The ATA News & Insights index, which includes items dated 16 Sep 2026, has no August-2026 tonnage headline; the latest tonnage item on that list remains July (ATA News & Insights).

The latest published ATA print is 18 Aug 2026: advanced seasonally adjusted For-Hire Truck Tonnage Index 113.5 in July (2015=100), −1% from June’s 114.7, −0.5% y/y; not-seasonally-adjusted 117, −0.9% from June’s 118; YTD vs 2025 +1.4%. ATA Chief Economist Bob Costello: “Aside from a couple pockets of strength, including the boom in data center construction for AI, freight has been lackluster. It is also true that the industry is seeing a recovery, but that is nearly all due to excess capacity leaving the market.” (ATA 18 Aug 2026).

A 23 Sep 2025 ATA release titled “ATA Truck Tonnage Index Rose 0.9% in August” is last year’s August, not 2026 (ATA 23 Sep 2025). Search-index pages that list a 21 Sep 2026 deadline for the August 2026 print were not fetched as ATA primary; they are consistent with the watch date of 2026-09-22 and with the absence of a 2026 August release on ATA’s news list. Do not treat August 2026 ATA tonnage as printed.

FMCSA final-rule RIA: ~200,000 non-domiciled CDL holders; ~194,000 “will exit” over ~five years

The 13 Feb 2026 final rule “Restoring Integrity to the Issuance of Non-Domiciled Commercial Drivers Licenses (CDL),” effective 16 Mar 2026, reaffirms the 29 Sep 2025 IFR: non-domiciled CLP/CDL eligibility limited to specified employment-based nonimmigrant statuses (H-2A, H-2B, E-2 as discussed in the rule) (govinfo FR 2026-02965). Preamble safety/compliance language still uses “tens of thousands” of non-domiciled CDLs issued contrary to federal regulations by more than 30 States, with sampled error rates including California ~25%, New York 53%, Texas 49% (same document; this matches the vault’s earlier primary check).

What is new relative to the mechanism page’s “no driver-supply economic analysis” note: the Affected Entities / Drivers and freight-market discussion in this final-rule RIA now state, in FMCSA’s voice:

• “FMCSA gathered information on current CLP and CDL holders during the APRs … and estimates that there are approximately 200,000 non-domiciled CDL holders, and approximately 20,000 non-domiciled CLP holders.”
• After estimating remaining eligible flow (H-2B “approximately 500 to 600” intending to operate a CMV per year; a simplifying H-2A assumption of 1/3 of BLS’s ~15,000 agricultural heavy/tractor-trailer drivers; E-2 “would not exceed 300”), “FMCSA estimates that SDLAs will issue approximately 6,000 non-domiciled CDLs per year. The remaining roughly 194,000 current non-domiciled CDL holders will exit the freight market, which is discussed in more detail in the cost section.”
• “There are roughly 200,000 non-domiciled CDL holders, which is approximately five percent of the 3.8 million active interstate CDL holders in 2024.”
• “FMCSA anticipates that these drivers will exit the market over approximately five years as their credentials come up for renewal, and that the market will respond to this change in capacity as it has in the past.”
• Carrier impact: “a maximum of 194,000 (or 25 percent) could be impacted” if each affected carrier employs one such driver; FMCSA immediately calls that “an extreme upper bound” because it assumes no carrier employs more than one (govinfo FR 2026-02965).

Caveats that are in the same government text: (1) the 6,000 figure is an annual issuance estimate for remaining eligible categories, subtracted from a stock of ~200,000 to produce ~194,000 — flow mixed with stock; (2) FMCSA’s freight-market discussion argues absorption, citing post-COVID carrier/driver entry, then falling rates and volumes, the Cass shipment and linehaul indexes, and BLS truckload employment, and stating that “current conditions in the freight market are conducive to just this type of adjustment” because of excess capacity since 2022, including a 2024 J.B. Hunt shareholder letter about the freight recession (same govinfo HTML). That is the opposite of treating 194,000 as an immediate 5% licensed-pool shock. (3) Commenters, not FMCSA, also used “almost 200,000” / “approximately 200,000 workers” in the comment-response section; those are commenter figures unless restated in the RIA block above.

What this RIA still is not: it is not J.B. Hunt’s published 5% to 12% of CDL holders (214,000–437,000) over two to three years combining non-domiciled CDL and ELP. That range is not in the FR 2026-02965 passages fetched here. The RIA’s ~5% of 3.8 million and ~194,000 apply to non-domiciled CDL stock, over ~five years, with FMCSA arguing the market can adjust.

FMCSA newsroom HTML was not fetched (www.fmcsa.dot.gov is a documented persistent-failure 403). FAQs encouraging states to revoke non-compliant unexpired non-domiciled CLP/CDLs appear in search-indexed FMCSA FAQ HTML; that FAQ file was dumped by search but is not treated as a clean WebFetch of fmcsa.dot.gov. The rule text on govinfo is the load-bearing primary.

Realized government counts (Aug 2026): >30,000 licenses canceled; >28,000 ELP OOS — not 194,000 yet

A 31 Aug 2026 DHS news release, in a section labeled “DEPARTMENT OF TRANSPORTATION OPERATIONS,” states: “In the past year-and-a-half, DOT has knocked over 28,000 drivers off our roads for failing to speak English, forced states to cancel over 30,000 licenses illegally issued to foreign drivers, and purged over 8,000 unqualified training schools from our FMCSA registry.” The same release separately says CDL skills tests must be given in English, “yet more than 28,000 drivers have been placed out of service for ELP violations since June 2025.” Same day, FMCSA announced emergency removal of more than 110 ELDT providers associated with more than 5,000 drivers who failed ELP tests; July investigations of nearly 400 providers in 40 states produced more than 160 notices of proposed TPR removals, with drivers from those 160+ providers “linked to 239” CMV-related fatalities (DHS 31 Aug 2026).

These are press-release counts, not a Federal Register table. They are government primary (DHS quoting DOT/FMCSA). They are much smaller than 194,000 current-stock “will exit,” and ELP OOS is not CDL cancellation: the 28,000 figure is roadside out-of-service since June 2025.

A 4 Aug 2026 DHS release on Operation Highway Shield (roadside wave 28–30 Jul 2026) reports that three-day operation’s own tally — 51 detained, including 21 drivers with non-domiciled CDLs issued by California and New York; 766 drivers/vehicles OOS; 86 arrests; 36 ELP violations; ~$1 million stolen cargo recovered — and does not, in the fetched DHS HTML, restated the cumulative 30,000 / 26,000 national figures that search snippets attributed to the parallel transportation.gov briefing (that DOT URL timed out / Access Denied on fetch) (DHS 4 Aug 2026). Do not treat snippet-only 26,000-from-May-1 or 28,000-from-May-1 as verified in this pass; the verified later cumulative is the 31 Aug DHS pair above.

ATA, 31 Aug 2026, supported a joint DOT/DHS operation “to permanently close 270 fraudulent CDL schools” (ATA 31 Aug 2026). ATA does not give a driver-stock number. DHS’s same-day release describes a synchronized surge at more than 200 training schools across 23 states plus the 110 emergency removals — not the 270 figure. The 270 is ATA’s wording of the operation, unverified against a fetched DOT/DHS count in this pass.

ELP NPRM (10 Aug 2026): ~9,000 incremental annual OOS in border commercial zones, not a licensed-pool census

FMCSA NPRM “English Language Proficiency; Out of Service Criteria,” 10 Aug 2026, RIN 2126-AC99, comments due 9 Oct 2026, would codify that a 49 CFR 391.11(b)(2) ELP violation is an OOS violation, aligning FMCSR with CVSA North American Standard OOS Criteria and with an 19 Apr 2026 CVSA revision that narrows the remaining exception for trips that stay inside U.S.–Mexico border commercial zones. Estimated annual motor-carrier cost approximately $14.4 million (govinfo FR 2026-16288).

The RIA’s driver count is not the 194,000 / 28,000 figures. FMCSA “estimates that approximately 9,000 drivers would receive ELP violations and also be placed OOS annually within the border commercial zones,” at $800/day downtime and two days to replace, “9,000 × $800 × 2 = $14.4 million.” The agency asks comment on whether that methodology under- or overestimates. Regulatory Flexibility text: “approximately 9,000 drivers could be placed OOS annually as a result of the changes in the proposed rule”; “FMCSA anticipates that all impacted entities would operate in and around the U.S.-Mexico border commercial zones” (same govinfo HTML). FMCSA also says the proposal “would not change the cost of non-compliance with the existing ELP requirements because the proposed rule merely codifies existing FMCSA guidance and CVSA’s … OOS Criteria,” except insofar as the contemporaneous border-zone tightening is scored.

So: 9,000 = incremental annual border-zone OOS from this NPRM’s analytical frame. 28,000 = DOT/DHS cumulative nationwide ELP OOS since June 2025 (press). They are not the same statistic and neither is a count of CDLs canceled.

Grokipedia’s CDL page notes recent federal tightening of non-domiciled processes and state-audit non-compliance but does not quantify 194,000 / 30,000 / 28,000 (Grokipedia: Commercial driver's license). Use it only as vocabulary, not as the scale source.

Contradictions and open questions

• Did the July supply-led signature hold? Rate-up: yes, stronger. Volume-down: no, on y/y. Cass’s own author still refuses a “major improvement in freight demand” reading, and 2-year stacked shipments remain −7.4%. Whether August is a one-month SA bounce (reversing June–July) versus a durable demand inflection is not settled by one print. Cass’s own July seasonal guide had pointed to ~−3% y/y in August; the miss is large enough that a follow-up September Cass print (Cass says normal seasonality would then be ~+1% y/y) is the next independent check (Cass August 2026).
• Class 8 above replacement vs regulatory capacity cap. Cass/ACT says tractor sales in July–August allowed the fleet to expand for the first time in 18 months and that new regulations plus broker-liability law still limit adding capacity. Those two sentences pull in opposite directions on how durable the rate-up is. This pass did not fetch a primary Class 8 OEM/ACT table; the claim lives in the Cass August ACT block only.
• 194,000 is now in the FMCSA final-rule RIA, which changes the vault’s “rule has no driver-supply analysis / 194k is only J.B. Hunt” provenance. It does not automatically validate J.B. Hunt’s 5–12% over 2–3 years including ELP. FMCSA’s own timing is ~five years, and FMCSA argues excess-capacity absorption. RXO-style “FMCSA estimates 194,000” is now literally true of the RIA sentence and still easy to over-read as an immediate 5% shock.
• Stock vs flow in the 6,000 / 194,000 arithmetic. FMCSA subtracts estimated annual remaining eligible issuances from current stock. Whether ~6,000 of the current 200,000 would actually requalify (H-2A/H-2B/E-2, or conversion of mis-typed LPR “non-domiciled” credentials to regular CDLs, which the preamble says happened in some states) is not independently tabulated in the passages used here.
• Realized vs modeled. Verified government press as of 31 Aug 2026: >30,000 licenses canceled; >28,000 ELP OOS since June 2025. That is the same order of magnitude as the vault’s older “early wave ~13,000” plus subsequent state revocations — not 194,000. How many of the 30,000 were active long-haul truckload drivers (vs CLP, vs already-expired, vs LPR paperwork corrections) is not in the fetched releases.
• OOS ≠ removal from the licensed pool. ELP OOS lasts until the condition is remedied; the NPRM scores delay cost in days, not CDL cancellation. Using 28,000 ELP OOS as a substitute for 194,000 non-domiciled CDL exits would mix two instruments.
• ATA August 2026 tonnage should exist around 21–22 Sep 2026 if the usual lag holds; it did not exist on 19 Sep 2026. July ATA still described recovery as “nearly all due to excess capacity leaving the market,” which rhymes with July Cass and has not been tested against Cass’s August shipment sign-flip.
• transportation.gov briefing-room HTML could not be fetched (timeout / Access Denied). Counts from May 1 “one-year” recaps that appear only in search snippets (including an earlier 28,000 licenses figure) stay unverified here; later DHS HTML is used instead. fmcsa.dot.gov newsroom (including a widely indexed California 17,000 item) was not fetched, per the reliability tracker.

Provenance

Rounds run: 3 of 3 (full)

Sub-questions by round:

Round 1 (broad survey):

1. What did Cass print for August 2026 TL linehaul (ex-fuel) and shipments, versus July +8.6% / −4.8%?
2. Has ATA published August 2026 for-hire truck tonnage?
3. Is there new FMCSA/govinfo primary on non-domiciled CDL / ELP scale other than J.B. Hunt’s 5–12% / 194k model?
4. How does Cass/ACT attribute August (capacity vs demand)?

Round 2 (drill-down):

1. Is 194,000 in the FMCSA final-rule RIA, or only in trade press / J.B. Hunt? — targeting the mechanism page’s “no driver-supply analysis” note.
2. What government primary supports “28,000 revoked”? — targeting realized vs modeled scale.
3. How does the 10 Aug 2026 ELP NPRM’s 9,000 relate to press 26,000 / 28,000 OOS? — targeting instrument mix-up (OOS vs CDL cancellation; nationwide vs border-zone).

Round 3 (resolve remaining uncertainty):

1. Are there later official CDL-revocation / ELP counts after May 2026? — targeting whether 28k was superseded.
2. Confirm ATA August 2026 still unpublished on 19 Sep 2026.
3. Does Cass’s archives page already carry an August 2026 blurb, or only the dedicated report URL?

Anchor source (Grokipedia, fetched before round 1):

• Truckload shipping — 12,017 chars extracted (API, truncated at --max-chars 12000) — FTL definition, equipment types, FMCSA cargo-securement citations; no August 2026 Cass numbers; no CDL-removal RIA.
• Additional encyclopedia fetches (counted in round budget, not the free lead-topic bonus): Cass Freight Index — 4,017 chars — methodology only; Commercial driver's license — 3,017 chars — program history plus qualitative non-domiciled-tightening note.

URLs fetched (11 successful, 2 failed):

Round 1:

• Cass Transportation Index Report | August 2026 — official / index publisher — August table + ACT commentary (primary for the print).
• Cass: TL rates jump 11% in August, freight shipments turn positive — trade press (Todd Maiden, 14 Sep 2026) — 20 consecutive y/y, largest since June 2022, diesel +46% y/y, “demand remains cool.”
• Cass Transportation Index Report | July 2026 — official — confirms +8.6% / −4.8% and July’s “capacity is declining” / ~−3% August guide.
• ATA Truck Tonnage Index Fell 1% in July — industry association — latest published ATA print (18 Aug 2026).
• English Language Proficiency; Out of Service Criteria (FR 2026-16288) — federal primary (NPRM) — 9,000 incremental annual border-zone OOS; $14.4M; comments due 9 Oct 2026.

Round 2:

• Restoring Integrity … Non-Domiciled CDLs (FR 2026-02965) — federal primary (final rule + RIA) — ~200,000 / ~194,000 / ~5% of 3.8 million / ~five-year exit; “tens of thousands”; NY 53% / TX 49% / CA ~25% sample error rates.
• ATA News & Insights — association index — no August 2026 tonnage item through 16 Sep 2026 dated posts.
• [Failed: https://www.transportation.gov/briefing-room/trumps-transportation-secretary-sean-p-duffy-celebrates-one-year-delivering-american] — timeout then curl Access Denied (snippet-only 28,000 licenses / May 1 2026 not used as verified).

Round 3:

• DHS, 4 Aug 2026, Operation Highway Shield — federal primary (press) — 28–30 Jul 2026 wave counts only.
• DHS, 31 Aug 2026, trucking-fraud crackdown — federal primary (press quoting DOT) — >30,000 licenses canceled; >28,000 ELP OOS since June 2025; 110 ELDT emergency removals / >5,000 associated ELP-fail drivers.
• ATA Supports Latest Crackdown on CDL Mills — association — ATA’s “270” schools figure (not matched to a fetched DOT/DHS count).
• Cass Freight Index Report Archives — official — July 2026 still the last archive blurb.
• [Failed: https://www.transportation.gov/briefing-room/us-transportation-secretary-duffy-us-homeland-security-secretary-mullin-white-house] — timeout; same-day DHS HTML used instead.

Not fetched (reliability routing): www.fmcsa.dot.gov newsroom (persistent-failure 403); www.federalregister.gov document HTML (hard-blocked; govinfo HTML used). PDF of FR 2026-02965 is on the *.gov whitelist but HTML was sufficient.

Tools used: WebSearch, WebFetch, grokipedia-fetch (python3 /workspace/.claude/skills/_lib/grokipedia.py fetch). No --include-x. Generated: 2026-09-19 ~21:00 UTC