In Oxford’s Michael Osborne and Carl Benedikt Frey’s hugely influential 2013 paper looking at the likelihood of automation for various professions, truck driving was one of the professions that were projected to be automated in double-quick time. Indeed, Frey and Osborne themselves thought that there was a 79% chance that truck driving would be automated within a decade or so.
Nearly a decade on from that influential paper, and with driver shortages across Europe still fresh in our mind, we may regard such predictions with a degree of scorn. The breathless predictions around the automation of trucking were no doubt driven by strong early results achieved by the likes of Otto, Uber’s self-driving truck project. Otto was the focus of Uber’s initial driverless ambitions, not least because they assumed that the sector would be easier to crack than the messy world of city center driving.
Alas, the company was beset by legal problems and the brand was eventually mothballed by Uber. A recent study from the University of Michigan and Carnegie Mellon aims to explore whether things have materially changed and whether drivers are at any more risk from automation now than they have been during the 9 years since Frey and Osborne’s paper.
At risk (kind of)
Despite the huge demand for long-haul truck drivers today, the researchers are just as bearish as previous academics, albeit with the fairly significant caveat that their pessimism is dependent upon the autonomous trucking technology improving such that it can successfully operate in the kind of weather conditions seen across the continental United States.
“Our results suggest that the impacts of automation may not happen all at once. If automation is restricted to Sun Belt states (including Florida, Texas and Arizona)—because the technology may not initially work well in rough weather—about 10% of the operator hours will be affected,” they explain.
The authors analyzed transportation data from the 2017 Commodity Flow Survey, which is generated by a consortium of the U.S. Department of Commerce, the U.S. Census Bureau, and the U.S. Bureau of Transportation Statistics. The data contained information on trucking shipments and the operator hours required to fulfill those shipments.
They also examined various scenarios for the automation of trucking, such as its rollout in sunny states or during the spring and summer months. Maybe just for journeys of more than 500 miles. They also considered a full deployment across the country.
“Our study is the first to combine a geospatial analysis based on shipment data with an explicit consideration of the specific capabilities of automation and how those might evolve over time,” they explain.
In the long haul
The researchers explain that a number of companies are working on a kind of “transfer hub” model, that would see automated vehicles operate on highways between hubs, with human drivers then undertaking the significantly more complex urban and suburban legs of the journey at either end.
They also explain that labor currently accounts for around 40% of the cost of trucking, so there are clear economic incentives to develop an automated option, but with around 3.5 million people employed as long-haul truck drivers in the U.S. the concerns around job losses in such a scenario are understandable.
“Because trucking is viewed as one of the few jobs that give folks with a high school education the chance to make a decent living, there is a concern that automation will eliminate these jobs,” the researchers explain. “Some people worry that all or most of the million or more trucking jobs might be lost.”
Given the scale of the sector, the researchers speculate that automation is likely to result in a few hundred thousand jobs being lost, although they hedge this by stating that many of these jobs are badly paid and generally unpleasant in nature.
“We think that it is possible that the number of operator hours lost at truck stops, because automated trucks will have no drivers who need to be served at truck stops, could be compensated by new employment opportunities at transfer hub ports,” they say.
Of course, if long-haul jobs are affected, there might be a corresponding increase in the short-haul jobs required to take goods from the hubs to homes, warehouses, and shops. The researchers analyzed this and found that while there is a degree of increase in the prospects of short-haul drivers, this is unlikely to compensate for the decline in long-haul prospects, not least due to the lower pay that short-haul jobs typically secure.
“We found that an increase in short-haul operation is unlikely to compensate for the loss in long-haul operator-hours, despite public claims to this effect by the developers of the technology,” the authors explain. “As a result of these conflicting claims, as well as the uncertainty over the technology itself and its limitations, there is little clarity on how automated trucking will be deployed and its economic and political ramifications, such as the impact on the long-haul trucking labor market. We hope to help resolve these controversies.”
Of course, the study rests on the likely developments in the sector, both from a technological perspective but also a regulatory perspective. It also assumes that technology will be capable of performing all of the tasks currently done by drivers including delivering the goods to customers, face-to-face communication, and so on.
Indeed, when we talk about automation in trucking, we typically talk of level 2 or 3 automation, which will still require human input and oversight. While level 4 is more powerful, industry insiders don’t see that taking hold for a couple of years, with dates of around 2025 commonly used.
Of course, that doesn’t mean that truck drivers should be relaxed, not least as fleets are likely to be among the first sectors to be automated, especially given the well-documented driver shortages across the sector. Whereas the current shortage has created a seller’s market for drivers, therefore, this boom time may not last forever.