Autonomous Cars Could Impact Nearly 16 Million Jobs In U.S., Commerce Department Finds

The technology required to enable fully autonomous cars is not here yet. You'll get no argument from us on that point. 

Despite the loftiest of wishes from companies like Uber and Tesla, for now autonomous cars can't seem to stop running red lights...which is a slight issue.  And that says nothing about the societal transformation required to fully adopt such technology which will likely span a generation.  Let's face it, just like grandma refused to adopt the e-commerce revolution, there are certain people who will simply never trust a computer to drive them around.

All that said, it is inevitable that, at some point in the future, autonomous vehicles will be the norm.  And, when that day comes, it will undoubtedly wreak further havoc on a U.S. job market where 95 million people have already decided they would rather sit at home than look for a least according to a new study from the U.S. Commerce Department

According to the study released last week, nearly 4 million jobs in the U.S. could be completely eliminated by autonomous vehicle technology while closer 16 million will be radically transformed.

The expected introduction of autonomous, or “self-driving,” vehicles (AVs) promises to have a potentially profound impact on labor demand. This paper explores this potential effect by identifying the occupations most likely to be directly affected by the business adoption of autonomous vehicles.


In 2015, 15.5 million U.S. workers were employed in occupations that could be affected (to varying degrees) by the introduction of automated vehicles. This represents about one in nine workers.


We divide these occupations into “motor vehicle operators” and “other on-the-job drivers.” Motor vehicle operators are occupations for which driving vehicles to transport persons and goods is a primary activity, are more likely to be displaced by AVs than other driving-related occupations. In 2015, there were 3.8 million workers in these occupations. These workers were predominately male, older, less educated, and compensated less than the typical worker. Motor vehicle operator jobs are most concentrated in the transportation and warehousing sector.


Other on-the-job drivers use roadway motor vehicles to deliver services or to travel to work sites, such as first responders, construction trades, repair and installation, and personal home care aides. In 2015, there were 11.7 million workers in these occupations and they are mostly concentrated in construction, administrative and waste management, health care, and government. Other-on-the-job drivers may be more likely to benefit from greater productivity and better working conditions offered by AVs than motor vehicle operator occupations.


So which professions will be hit the hardest?  Well, the Commerce Department says that 65% of the most obvious job losses will likely come from the long and short-haul commercial delivery businesses.


Of course, the direct driving job losses say nothing about the 2nd-derivative losses that will also inevitably come.  For instance, consider our post from almost exactly one year ago in which we argued that autonomous cars could double the capacity utilization of passenger vehicles thus cutting a 'normalized' auto SAAR in half (see: Ford Announces Plans To Self-Destruct Starting In 2021).

So what do we mean when we say an autonomous car pretty much ensures Ford's demise?  To be clear, we're not specifically targeting Ford...the whole auto industry is in serious trouble when truly autonomous driving arrives.  Below is a little math to help illustrate the point.


Right now there are roughly 250mm light-duty passenger cars on the road in the U.S. (that's about 1 car per driving age person, btw, which is fairly astounding by itself).  American's travel roughly 3 trillion miles per year in aggregate which which means that each car travels an average of 12k miles per year.  Now if you assume the average rate of travel is 45 miles per hour then you'll find that each car is implied to be on the road for an average of about 45 minutes per day.  That's a capacity utilization of about 3% (see table below for quick math).


Capacity Utilization


A 3% capacity utilization ratio is, needless to say, fairly terrible.  We don't imagine too many CFOs would model capital allocation decisions based on a 3% capacity utilization for fixed assets.  That said, individuals are forced to underwrite car purchases to a 3% capacity utilization because they have no choice.  People have to get to work and 100% reliance on public transit options as just not feasible for most people in this country.


That is, until the arrival of completely autonomous vehicles.  The problem with mass transit is that people still need a car to get back and forth to the train station or bus stop.  The problem with Ubers/Taxis is that they're expensive for daily use due primarily to the labor overhead that's built into your per mile rate.  But fully autonomous vehicles solve both those problems.  Now, people will have the option of a vehicle at their beck and call without having to fund the upfront capital cost of a purchase and/or the per unit human capital costs inherent in taking an Uber.  In other words, the per mile rental rate of a fully autonomous car should be competed down to a level that provides an adequate return solely on the cost of the wages, no benefits, none of the typical hassles associated with employing people.  Or, said another way, taking an Uber is going to get really freaking cheap.


But the best part is that capacity utilization with fully autonomous cars can skyrocket driving per unit costs even lower for passengers.  For example, when you drive to work right now your car sits there all day until you drive home.  In the autonomous car world, that car will drive you to work then go pick up multiple other people to do the same thing.  Now, if capacity utilization doubles from just 3% to 6% all of sudden half the number of cars are required in the US which means annual SAAR goes from ~17mm to ~8.5mm...which means Ford and GM likely find themselves in another bailout situation.


Normalized SAAR


And, if you're making half the cars, guess how many people you need to make them?

Oh well, at least there's always that McDonalds job to fall back on...oh wait...



The full Commerce Department study can be viewed here:


tmosley musicgem Mon, 08/14/2017 - 21:51 Permalink

NOT easy. If you start school today, by the time you finish, your field will probably be obsolete.Luckily, that is actually a good thing. Prices for most goods and services will plummet towards zero as the labor involved in their production and delivery are automated, much like how people used to have to work almost all day to get enough money to buy flour to make their daily bread, yet now even the lowest paid workers earn enough to buy a loaf in minutes. All thanks to farm and baking automation.

In reply to by musicgem

Stuck on Zero Never One Roach Mon, 08/14/2017 - 22:50 Permalink

This is al BS. Nearly all motor vehicle operators also perform auxilliary chores. Bus drivers collect fares, settle fights, help old ladies aboard and sing tenor for the passengers. Most delivery vehicles require the driver to help load and unload cargo e.g. concrete trucks. Food delivery trucks require the drivers to carry the merchandise into the stores and stock it. Limo drivers have to look snazzy, open the doors for the guests, and polish the hood when idle.

In reply to by Never One Roach

Wile-E-Coyote Stuck on Zero Tue, 08/15/2017 - 01:31 Permalink

Yeah, and in the old days' horses had to be fed, rubbed down and stabled, how did that work out. All your problems above all have current solutions.The problem I see is a mass scramble to adopt this new technology it won't be a slow process, it will be forced by the end user any competitive advantage will be grasped at like a drowning man gasping for air. It will be brutal.Of course, there will be employment opportunities for cleaners someone has to clean the sick out of the back of the car.

In reply to by Stuck on Zero

Justin Case tmosley Mon, 08/14/2017 - 23:49 Permalink

Electrician with 432 Certificate and PLC knowledge, Maintenance Millwright Certified 433, plumber, HVAC, robotics, aircraft mechanic,  construction, doctors, teachers etc. All good prospects today. Pilots are becoming redundent by the sofisticated electronic systems like drones, no pilots. Boeing can actually control their aircraft remotely already for 5 yrs or more. Flight MH370 - Boeing 777.Automation still requires electronc control systems and ASRS systems require mechanical maintenance as well as electrical controls.

In reply to by tmosley

pitz Justin Case Tue, 08/15/2017 - 00:06 Permalink

Modern aircraft still require a lot of intervention from the pilots to set up the automation, to ensure that all the requirements are adhered to, to cycle circuit breakers, to arrange for maintenance, etc.  It would be a giant mistake to think that planes just fly themselves.

In reply to by Justin Case

Tbirdthree pitz Tue, 08/15/2017 - 05:51 Permalink

Don't forget, Air France 447's  Auto Flight Control System went into a tizzy and packed it in, basically telling the Pilots, "Ok, I give up, you fly it, I don't know what I am doing.." when its Airpseed sensors were frozen over while the crew tried to navigate  around an impenetrable line of thunderstorms.The crew itself then misshandled the situation and crashed anyway, but the aircraft was flyable and should have been recoverable  had the manufacturer/ Aviation Authorities not bamboozzled the industry into thinking such failure modes were too remote to warrant training the pilots to handle.  Since that incident and others, there is renewed emphasis in recurrent training given to manual flying skills and jet upset recovery in order to add another layer of redundancy and mitigate for any other failure modes that the programmers and Certification Standards cannot even foresee or imagine is possible. To say nothing of the fact that every day pilots intervene and work problems out to land safely, without any of it being written up or documented, with only the accidents where they failed to prevent disasters being thoroughly investigated, thus the skewed conclusion is reached that "pilot error" causes  a majority of accidents...whereas often, the crew are just the last guys in a long chain caught holding the ball.For myslef, I can tell you that a Pitot Static system failure (like AF447) that occured to me while I was taking off in a 737 would have sent a fully automated 737 into an uncontrolled climb/stall or dive with no chance to recover at such low altitude. But, maybe by the grace of God that day, my co-pilot and I managed to take the beast around the patch and Landed fuss.... a simple incident report was filed and forgotten, never investigated by anyone.... and no slap on the back either. Wanna see R2D2 do that. 

In reply to by pitz

Wile-E-Coyote Justin Case Tue, 08/15/2017 - 01:39 Permalink

You will always need technicians.............. but that is even being dumbed down by on board diagnostics it's eroding the skill base. There will always be a need for highly skilled individuals but they are only utilized when a real problem arises that the computer can't give a solution. It's already happening and has been for a number of years it's having a dampening effect on wages already. Apprenticeships are being shortened more semi skilled people are being utilized and sooner or later automation of these tasks will arrive.

In reply to by Justin Case

Crazy Or Not Justin Case Tue, 08/15/2017 - 06:09 Permalink

The difficulties include: Robots don't pay taxes.Robots don't consume.Robots are "Slave systems" (multiple 'bots to 1 operator). Tech is here and now, and on rollout.Robots include heirarchal controls so can be generically updated/hackedWe're not prepared for the scale and diversity of the impacts.Vehicle below could equally be Construction /military or Semi Trailer Tractor unit.…

In reply to by Justin Case

Duc888 Mon, 08/14/2017 - 21:41 Permalink

Something tells me we are still years away from this tech running flawlessly. In our litigious society....I can not imagine this being rolled out smoothly.

TuPhat Duc888 Mon, 08/14/2017 - 22:10 Permalink

The article forgets that when I rent a car for a day I'm not paying for the driver and it's actually very expensive so why will autonomous cars that cost more to produce be cheap?  They failed to figure that out properly.  T mosley has some kind of brain that accepts all this without question but most people will not.  I think ZH should publish an article telling us how they lost their jobs to the robot writers that put out this trash.  That is if there are any humans left at ZH.  I have been on ZH for years and still like this site but the quality is going down hill.  I guess we can blame it on the robots.

In reply to by Duc888

undertow1141 musicgem Mon, 08/14/2017 - 23:47 Permalink

So if we're sharing a robo-car and I go camping for a month long vacation, how do you get to work? Who pays the vehicle registration and insurance? Will they come in 4x4 or pickup models so I can haul trash or go to Lowe's/Home Depot for lumber?When the automated delivery truck gets to the store, who takes the product inside? 

In reply to by musicgem

Mr Pink Mon, 08/14/2017 - 21:42 Permalink

My neighbor drives over night for FedEX. He thinks all of the cameras on his truck are watching his driving. I told him that they don't care about his driving...he is training his replacement

cheeseheader Mon, 08/14/2017 - 21:42 Permalink

Sooo, let's not proceed with autonomous cars. Elon, bless your heart, I'm filing a 1099 on this one...that'll be $350,000. For now.   Pleasure doing bidness with ya.  You have my number for the next conundrum, bud.

nmewn Mon, 08/14/2017 - 21:42 Permalink

Let me help this police state mindset along (for the nuuubs)."Now see! With autonomous cars the commie chick & all the innocents being run over by Jihadis would still be alive! Its really about our national security!" ;-)

Deplorable Mon, 08/14/2017 - 21:45 Permalink

Lets do a little logistics math:Assume we have a mid sized metropolitan area with 100,000 commuters who travel 15 miles daily from the suburbs to downtown, and it takes aproxiamately 1/2 hour for each one way commute. In the course of a normal day, these 100,000 commuters can all arrive at their destinations within the 2 hour rush hour windows beginning at 7 AM and 5 PM.Now lets assume these commuters all sell their vehicles and begin using a network of 10,000 Ubers to make their daily commutes.On Monday morning at 7 AM the commuters start calling Uber to summon their rides for the commute downtown. The first 10,000 commuters are successful in reserving their ride, and the remaining 90,000 are placed in the Uber queue to wait for the next availabe ride. Meanwhile, the first set of 10,000 commuters is successfully delivered to their destination at 7:30 AM, and the Uber fleet returns to the suburbs around 8 AM to pick up the next group of 10,000 commuters and take them downtown. Repeat this cycle 10 times, and Uber will have successfully delivered all the commuters downtown by 5 PM.Now lets reverse the cycle and send all of these commuters home at the end of the day, starting at 5 PM. If we follow the same assumptions as the morning commute, it should take another 10 hours to get everyone home by 3 AM Tuesday morning.Mission accomplished!  We have successfully replace 100,000 personal vehicles and eliminated the aggrevation of ownership for all those commuters.  They will no longer have to pay for gas, maintenance, or insurance for a personal vehicle that sits in the garage 95% of the time.Does anyone see a problem with this scenario? The only way you can deliver 100,000 commuters, within the two hour rush hour time frames, is if you have a pool of 50,000 Ubers that are 100% dedicated to moving rush hour commuters.    

Duc888 Deplorable Mon, 08/14/2017 - 21:54 Permalink

  Uber is a weird gig.  Once an area is saturated with Uber drivers thigs go all pear shaped quickly.  I've got a buddy who was between jobs and did the Uber Black thing in the New Haven area.  He drove a cab a few years back and is a bit OCD with charts and graphs and stuff like that.... so he charted UBER.  First month averaged 900 a week take home, month two and three...$800, month 4-5......down to about $500-$650.  It was getting to the point where he had to spend more and more time out on the road to make less and less money.  Last month he averaged about $6 an hour.  No shit.  When he was in the middle of NH in an intersection he would turn the app off, I guess there's a function on there that lets one see where all the Uber drivers are locally.   In a big 4 way intersection there were typically 10-12 Uber drivers within eyesight down all the streets...pure saturation. Luckily he started a week ago driving an oil truck for $22 an hour and all the OT he can handle...

In reply to by Deplorable

TuPhat tmosley Mon, 08/14/2017 - 22:18 Permalink

You are a one eyed dreamer.  Never happen.  I lived in a foriegn country and the few stop lights were totally ignored.  Traffic was much worse, much more dangerous and commute times much longer.  If the auto car can get through the mess faster that means you are at a higher risk.

In reply to by tmosley