U.S. Rail: The Spectre Of Autonomous Trucks

Includes: BRK.A, BRK.B, CN, CP, KSU, NSC, TSLA
by: Burt Rothberg

Autonomous trucks will likely be on the Interstates in a few years.

Will this cut into railroad intermodal business?

Yes, but not as much as you might think.

When Tesla (NASDAQ:TSLA) announced its electric semi, Elon Musk said that when driven autonomously, the cost per ton mile will be less than that of rail. Warren Buffett, whose Berkshire-Hathaway owns BNSF, has also said that autonomous trucks will be bad for his business. Is this the technology that kills railroad valuations? Let's look at the numbers.

Long haul trucking is a near-perfect use case for autonomous vehicles. Almost all the driving is on limited access highways. These roads have a small set of visual clues compared to urban/suburban ones, so the driving algorithm has less to worry about. In fact, manufacturers like Cadillac are already producing no-hands driving on these roads, and the tech is getting better all the time.

Another plus is the effect on driver supply. Right now trucking companies are starved for drivers. A big reason is that long haul drivers have to spend a considerable time away from home, sometimes for weeks. This is hard on anyone with a family. But if the algorithms take over driving on the Interstate, drivers will only be needed for local work. This is presumably a more attractive job.

Trucks are not likely to ever be competitive with rail hauling bulk commodities like coal. But railroads make a lot of money hauling around containers and trailers. They call it "intermodal". The railroad moves the trailer most of the way, say from a port in California to a warehouse in Ohio. It's then offloaded to a truck that takes it to a store in, say, Cleveland. This is efficient because RRs have a substantially lower cost per mile than trucks.

Intermodal already makes up a substantial fraction of RR carloads and revenue (data is from the railroad's latest available quarterly report).

Class I railroad intermodal revenue:

Intermodal Revenue % of Total
Union Pacific (UP) 18 %
Canadian National (CN) 24 %
CSX 15.5 %
Norfolk Southern (NSC) 23 %
Kansas City Southern (KSU) 14 %
Canadian Pacific (CP) 21 %

Moreover, it is a growing fraction. RRs have lost bulk business as coal and petroleum volumes have shrunk. Here's a table for the major North American RRs (data is from FRB St. Louis):

The orange line is total RR carloads; the blue line is intermodal ton-miles. Intermodal is the only real source of growth for North American RRs, so increased competition from trucks could be very bad.

So what will be the effect of driverless trucks on the Interstate? Will this destroy Intermodal? I'm going to use the American Transportation Research Institute as a data source. Here's a table from their 2017 report:

Cost per Mile of US Trucking:

2008 2009 2010 2011 2012 2013 2014 2015 2016
Driver Wages + Benefits $0.58 $0.53 $0.61 $0.61 $0.53 $0.57 $0.59 $0.63 $0.68
Fuel $0.63 $0.41 $0.49 $0.59 $0.64 $0.65 $0.58 $0.40 $0.34
Lease/Purchase Payments $0.21 $0.26 $0.18 $0.19 $0.17 $0.16 $0.22 $0.23 $0.26
All Other $0.23 $0.26 $0.27 $0.32 $0.29 $0.30 $0.31 $0.31 $0.32
TOTAL COST $1.65 $1.45 $1.55 $1.71 $1.63 $1.68 $1.70 $1.58 $1.59
Cost / ton mile $0.08 $0.07 $0.07 $0.08 $0.08 $0.08 $0.08 $0.07 $0.07

Note: Cost / ton mile assumes a trailer with 43K lb. capacity.

Now this is for all US trucking. We will have to make some adjustments for Interstate-only trucking. I don't have any detailed data on this, but let's conservatively assume that a truck on the Interstate runs at least twice the speed of a local truck. So a mile on the Interstate uses about half the driver-hours as a local mile. Moreover, it's unlikely that highway manpower will go to zero. The trucking companies still have to have roadside maintenance personnel and probably "backup drone pilots" at headquarters. Additionally, we can expect autonomous trucks to cost more (Somebody has to pay for all that R&D). I will assume 20% more. Here's my calculation:

Cost changes from driver-operated to autonomous

Current Costs
Driver $/mile Total $0.68
Driver $/mile Highway $0.34
Lease/Purchase Payments $0.26
TOTAL $0.60
Autonomous Costs
Driver $/mile Highway $0.00
Other Labor / mile $0.05
Lease/Purchase Payments $0.30
TOTAL $0.35
Autonomous Savings / mile $0.25
Autonomous Savings / ton-mile @ 43K lbs. $.011

So it looks like autonomy will reduce truck costs by about 1/7 to $.06 / ton-mile. How does this compare to RRs? Data on route rates is closely guarded for competitive reasons. And they are negotiable; I have spoken to two shippers and they gave me very different rates. So I'll have to make some educated guesses. If anyone reading this has firmer data, please comment.

I did a simple calculation: I took the operating expenses for the class I railroads I could find data on and divided by revenue ton miles. Here is what I came up with:

Operating costs of Class I Railroads (Source: Annual Reports):

Revenue ton-miles RR costs Costs / ton mile
UP 2015 485 13.761 $0.03
NSF 2016 191 6.814 $0.04
CN 9 mon 9/17 59.05 1.557 $0.03
avg $0.03

Now this is a very crude measure. It includes all trains. Naturally it costs less to move a ton of coal than a ton of a container. OTOH, the long hauls of most intermodal are less per ton/mile than shorter hauls. Nonetheless, it does show that trucking costs are about twice RR costs. This is about what I would have expected. The difference is the RR margin.

So in general, RRs will continue to have a cost advantage over trucks, but it will decline. This is negitive for RRs margins. Right now RRs costs are about half that of trucks, but the prices they charge are pegged to the truck competition. If trucking charges go down, so will RR margins.

There may also be a negative effect on volumes. The above calculations are for long-haul Interstate shipping. Actual shipping is almost always a mix of road types. The flexibility of trucks will increase their advantage for intermediate runs. Right now 800 miles is about the minimum for RR cost advantage. This number should go up. And of course, the tech continues to get better. Eventually we will have autonomous trucks for all roads, driving down costs even further.

This paints a somewhat negative scenario for RRs long term. However, politics may favor the other side. Just think: How would you as a driver (and voter) feel if you had your drive clogged up by a 15 truck autonomous train-truck. It would undoubtedly be going at lower speeds, and it would block exits as it goes by. Many Interstates are already at near capacity; this would make things worse. I have no way of forecasting how this will pan out, but it will become a major issue.

My view is that while this is a long term negative, the next couple of years could be quite good for North American RRs. I am long NSC. I'm writing up my thesis for SA right now, and will publish it shortly.

Disclosure: I am/we are long NSC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.