Operating data for the week of May 26 highlighted the first of the four big public holidays that typically accelerate the U.S. Class I networks, due to the temporary reduction in volume pressure. That was, of course, Memorial Day, which had these specific effects:
- Union Pacific’s network accelerated 2.1% from the week prior and ran 6.6% faster than the same week in 2024.
- BNSF gained 1.9% in speed, sequentially, and Memorial Day week was 3.8% faster this year vs. last.
- Norfolk Southern gained a handy 4.2% in speed sequentially and was faster YoY by 2.6%.
- CSX, most interestingly, accelerated by a very significant 5.0%, but that was still 1.6% lower than the same week last year. Clearly more upside potential here.
While these holiday effects are temporary, it’s more than just a one-week boost as worked-down car backlogs in yards and repositioned power and crews typically allow speed to remain elevated for a week or two following the week in which the public holiday occurred—the “slingshot effect.”
Fourth of July is next: a similar tailwind but with the cross-current headwind from seasonally peaking train crew vacations (weeks 26, 27, 28).
CPKC IT Systems Cutover Pushes Up Terminal Dwell
Some of you will already have noticed the recent spike in terminal dwell times at CPKC over the past two reported weeks, through May 30. You can see in the chart below the full system impact is now on par with major weather events and last year’s brief strike/lockout in Canada. The cause of the disruption is CPKC’s long planned cutover from the old KCS U.S. operating systems to the Canadian Pacific ones, which occurred on May 3. It’s just the old KCS systems in the U.S. that have been decommissioned. The KCSM systems in Mexico remain in place and will be transitioned to the CP systems later (beyond 2025).
Railroad operating IT system cutovers are notoriously fraught and include a couple of renowned debacles: Union Pacific-Southern Pacific in 1996, and the Conrail integration by CSX and Norfolk Southern in 1999. Why are these things so difficult?
- A widely dispersed user base with highly varying degrees of IT competency.
- You can’t stop the trains; they’re still running 24/7 so any problems will have an immediate impact.
- You’re cutting over the IT layer on physical operating networks that are inherently fragile, temperamental, and subject to extreme domino effects. If a fire starts in one place it will quickly spread.
If you’re an IT professional, this is an extraordinary tough assignment.
How do we track it? In terms of available public data, dwell for four of the former KCS U.S. yards is reported: Kansas City, Mo.; Laredo, Tex.; Shreveport, La.; and Jackson, Miss. Dwell at the Kansas City yard is still in the normal range, and Laredo dwell is heavily distorted by run-through trains and naturally volatile, so it’s really Shreveport and Jackson that are the two particular canaries in this coal mine. Here they are in the chart below, with Shreveport currently running at 28 hours vs. a pre-cutover YTD average of 18, and Jackson at 32 hours vs. 20 hours. There’s not yet a material inflection to the better in the data. Let’s hope one materializes soon.
