The monthly employment report for the truck transportation sector is a tale of two markets.
Seasonally adjusted figures reported by the Bureau of Labor Statistics showed a drop in employment between April and May. This occurred even as virtually every trucking company under the sun has a Help Wanted sign out, along with a package of pay increases, bonuses and other incentives.
But the decline in seasonally adjusted figures was not matched by the report for not-seasonally adjusted figures. The not-seasonally adjusted figures for all sectors are considered by economists to be the poor stepchild of data, and news reports inevitably focus on seasonally adjusted data. But the not-seasonally adjusted data does exist, and not all economists ignore it.
A definition of seasonally adjusted data supplied by the BLS said that the process to produce that number “attempts to measure and remove the influences of predictable seasonal patterns to reveal how employment and unemployment change from month to month.”
The not-seasonally adjusted numbers are not a raw number that counts every job; the monthly employment report, particularly in its first release, always has an aspect of modeling to it. The OECD defines not-seasonally adjusted figures as data that has not been changed and therefore doesn’t remove “the effects of regular, or seasonal, patterns.”
The end result for May’s trucking data was a significant difference in employment levels for the truck transportation sector.
The seasonally adjusted figures for truck transportation reported that the sector had 1,479,700 jobs in May. That represented a decline of 1,900 from April; a revised figure for that month put the number at 1,481,600 jobs. The final figure for March was 1,481,000 jobs so on a seasonally adjusted basis, the number of truck transportation jobs in May has fallen 1,300 jobs in two months.
But look at the data for not-seasonally adjusted numbers and it tells a different story. That figure for the truck transportation sector was 1,478,600 jobs in May. That is up 10,800 jobs from April and up 18,000 jobs since March.
Jason Miller, an associate professor of logistics at Michigan State, does think the not-seasonally adjusted numbers have value. “I view seasonally adjusted data as saying, ‘Given the time series structure of these data (e.g., their degree of seasonality, magnitude of autocorrelation, trading day adjustments, etc.), what is the expected value that we would see for this series for this month?’” Miller said in an email to FreightWaves.
But as the saying goes, the past may not be prologue: “The issue is that a time series’ past structure may not hold after some massive shock like COVID-19. For some data that are very seasonal, such as new housing starts, the issue is that the seasonally adjusted data appear to jump around dramatically.”
Truck transportation was far from the only area that showed a large difference between seasonally adjusted and not-seasonally adjusted numbers. The overall employment report showed total U.S. job growth at 559,000 jobs. The not-seasonally adjusted figure was 973,000 jobs.
The number of jobs is one indicator of employment in the truck transportation sector, but it is not the only one. Aaron Terrazas, the director of economic research at Convoy, noted that the number of hours worked by employees in the sector rose considerably, both for all employees and nonsupervisory employees.
“The average weekly hours worked for nonsupervisory employees at trucking firms touched 43.3 in April, coming off of an earlier all-time high of 43.0 hours in March, which was one hour per week longer than the 2015-2019 average of 41.9 hours,” Terrazas wrote in a blog commentary. When you take those additional hours and translate them into employees, Terrazas notes that for the nonsupervisory employees, it works out to about 24,000 new workers.
“May data are likely to show an even larger increase in hours worked for specialized trucking firms given the temporary relaxation of hours-of-service regulations for fuel tank haulers during the mid-May Colonial Pipeline outage,” Terrazas wrote.
Max Farrell, the CEO of Workhound, has a bird’s-eye view of the retention and recruitment of truck drivers through the feedback received via the Workhound app. He said in an email to FreightWaves that jobs in the sector “are being added at a methodical pace.”
But the sector’s hiring is in “a rut,” he said. “As the labor market tightens, companies in the supply chain will need to continue to embrace new strategies to retain their best people,” Farrell said. “They will also need to keep finding ways to evolve the job to make it more attractive for the post-pandemic workforce.”
Other highlights from the month’s employment report:
— The cost of trucking is rising. That is not a surprise. But the producer price index for the truck transportation sector is showing just how much it is rising. In April, it stood at 158. That 0.6% increase from the prior month comes after the March figure rose 2.1% in one month, an enormous increase. In February 2020, just before the pandemic hit, the trucking PPI was 146.7. That is a 7.7% increase in the PPI in 14 months.
— As Terrazas notes, both all employees and nonsupervisory employees are working more hours. They are getting paid more also, but the increase is not enormous and still hasn’t caught up to the levels of earlier in the year. The average hourly wage for production and nonsupervisory employees in April was $24.56/hour, up 11 cents from the prior month. But in January, it was $25.01, so it still has not recaptured that level.
— The difference between seasonal and not-seasonal numbers was stark in the figures for warehouse workers. The seasonally adjusted total for warehouse workers, which has been trending down the past few months, showed an increase to 1,408,800 jobs, up 1,800 jobs from the prior month. But in the not-seasonally adjusted category, jobs dropped to 1,393,200 jobs, down 3,400 jobs from the prior month.
— Comparing figures from last year is starting to run up against the problem of comparing a pandemic to the ending of a pandemic, with the year-on-year comparisons showing huge increases. The seasonally adjusted truck transportation jobs figure was 50,600 jobs more than last year. Seasonally adjusted warehouse jobs were 156,100 jobs more than a year ago.
More articles by John Kingston
Did employment of truckers drop just because they can’t be found?
Trucking job numbers: stronger but tempered
Truck transportation jobs grow but still below year-ago levels
rn
@John K, Prof. Jason Miller, et al… writing on the (narrow) topic of relying on SA vs NSA timeseries datasets (only)… given the “it depends” in this article’s headline and having encountered similar headlines/sentiment elsewhere.
Is the “narrow” matter not resolved (or at least better understood) by foll. std. guidance* around using such datasets specifically & insight generation generally e.g. :
1) performing only y-o-y comparisons when using NSA data or similar (& avoiding sequential period comparisons like m-o-m)
2) taking the time to examine & understand the actual (current and ongoing) seasonal adj techniques used by the dataset publisher
3) understanding the dataset publisher’s guidelines on when/how often past historical underlying dataset revisions may occur… including changes to seasonal adj technique itself… [and be prepared to revise one’s past interpertation/analysis/insight fm such changes 🤦♂️😣😇]
4) never relying on just 1 dataset regardless of sa/nsa/other transformations
A Senior partner-level consulting colleague & mentor of mine once remarked “stop complaining about issues with the data & take a position… the client is not paying us to hear it at it-all-depends” 😊
{*List intended as illustrative only & not comprehensive.}