Let’s say that tomorrow, Toyota starts selling fully driverless cars. And for nothing more than the regular cost of the car! You can buy a Camry that can completely drive itself, with or without a human in it.
So, obviously, you need one. But on your way to the car dealership, a guy wearing a black-and-white U logo says, “Hey, you shouldn’t buy that Camry. My organization has bought 20,000 such Camries, and we operate in your city — you can just use your smartphone to hire a ride in our driverless Camries. Obviously, it’s way cheaper than our old driver service.”
Is what he’s proposing a good deal? Let’s look at it in detail.
The Cost of Owning a Car
There are two general types of cost for a car: per year prices and per mile prices. That is, suppose you buy a car and let it sit in the garage for five years, then sell it. Your car will have depreciated in value (quite significantly, as it happens). And you’ve paid for the parking (for the price of the garage, at least). But your car will have cost you a lot less than if you buy it and use it heavily for those five years — both because the car will be in worse shape, and because you’d have paid a lot for fuel and maintenance in that time.
Here are the costs I can think of for owning a car:
Dt = Depreciation of the car’s value over time (as opposed to mileage)
Pp = Cost to park the car per year.
Dm = Depreciation of the car’s value over miles.
Mn = Maintenance per mile
F = Fuel per mile
I = Insurance cost per mile
Multiply those by time (t) and mileage (m), and you get:
t * (Dt + Pp) + m * (Dm + Mn + F + I)
Cost to Rent a Car
Uber has to buy, fuel, maintain, and park their cars just the same as you. You ultimately pay for their costs. But there are some differences. Your car sits around unused a lot — you aren’t a trucker, so on almost all days you drive it at most a couple of hours a day. The rest of the time, it sits around slowly losing value and costing to park. Uber can presumably get more use out of a vehicle over any given time. Let’s say that it can get four times as much use out of the car as you can. That means that all the time-based costs are divided by 4.
On the downside, you have to implicitly pay the costs for an Uber car not just while you’re driving in it, but while it’s driving over to you. In other words, there’s more milage on an Uber car that you have to pay for. How much is unclear. In a perfect world, you’d hail an Uber right when it was dropping off someone half a block away, and it’d be almost no extra milage. In a very imperfect world, you might have a car drive two miles to you in order to give you a trip of only one mile. Let’s assume that Uber has dense demand in your city, and so the milage that Uber cars drive is only 5% more than the milage that a privately owned car drives.
So now the deal is that the cost of an Uber car is:
0.25t * (Dt + Pp) + 1.05m * (Dm + Mn + F + I)
But that’s the cost. Uber has to make a profit as well. Let’s assume that it wants a (very reasonable) 10% marginal profit. So the total price to you is:
1.1 * (0.25t * (Dt + Pp) + 1.05m * (Dm + Mn + F + I))
Which comes out to:
0.275t * (Dt + Pp) + 1.155m * (Dm + Mn + F + I)
So you’re looking to see what the conditions are in which this inequality is satisfied:
0.275t * (Dt + Pp) + 1.155m * (Dm + Mn + F + I) < t * (Dt + Pp) + m * (Dm + Mn + F + I)
And we already have an implicit ratio of t to m, so we can just take them out of the equation:
0.275 * (Dt + Pp) + 1.155 * (Dm + Mn + F + I) < Dt + Pp + Dm + Mn + F + I
And so this equaltion simplifies very easily to:
0.155Dm + 0.155Mn + 0.155F + 0.155I < 0.725Dt + 0.725Pp
Or, in other words, is 15.5% of the mileage costs less than 72.5% of the time-based costs? If so, Uber is a better deal. If not, owning your own car is a better deal.
Let’s try to quantify those costs. We’ll be assuming 20,000 miles per year.
F: Camries get about 20mpg. 20,000 miles means 1,000 gallons. A gallon of gas costs let’s say $3.00. So that’s $3,000.
I: Insurance probably is a pretty low cost in a driverless car world. Let’s say it’s $200 per year.
Pp: Parking is one of those things that varies a lot. Let’s say that it costs $100 per month, so $1,200 per year, and come back to this one.
Mn: Maintenance costs are probably something on the general order of $1,000 per 20,000 miles.
That leaves depreciation. I plugged a 2010 Camry into Kelley Blue Book, and then checked for how much I should expect to sell it for after 100,000 miles ($9k) and after 5,000 miles ($14k). So after 5 years at 20,000 miles per year, that implies that I have about $5k of milage-based depreciation (Dm), and about $7,500 in time-based depreciation (Dt). That’s after 5 years and 100k miles, we’ve been doing things in units of 1 year and 20k miles, so divide by five: Dt = $1.5k, Dm = $1k
0.155 * ($1k + $1k + $3k + $0.2k) < 0.725 * ($1.5k + $1.2k)
0.155 * $5.2k < 0.725 * $2.7k
$806 < $1,957.50
And the answer is yes, Uber is cheaper. By about $1,000 per year.
…Under the above assumptions.
Let’s take a closer look at those assumptions.
The biggest part of the milage costs is fuel. Obviously, Toyota isn’t going to actually start selling driverless Camries this year. If you believe that electric cars are coming, and that the driverless car will also be the electric car, then we would expect fuel costs to be much lower. This might also lower the maintenance costs, as electric cars are at least arguably easier to maintain. In the electric car scenario, you expect another $100 or so per year of advantage to the Uber side.
Parking varies really wildly in cost. If you live in a non-space-constrained area, then a garage for your house might have a very low marginal cost, and free or very cheap parking might be plentiful. That could potentially make owning very competitive with rental.
On the other hand, if parking is very expensive, as it is in say Manhattan, you might find that it’s hundreds or thousands of dollars per year more expensive to own than the above assumptions suggest.
Depreciation and car cost
Car value depreciation is a major factor on both sides of the equation. I have to say that I was a little surprised by the KBB answer that says that only $5k of the Camry’s cost depreciation is explained by 95k miles. But that seems like the closest thing to an objective source on depreciation that I could find.
Lots of things could affect how a driverless car depreciates. The sensors on the car might be quickly obsoleted — or they might not. The status and style value of a car might change dramatically in a world in which ownership models were in flux. Driverless cars might wear differently (more gracefully? Less?) than normal cars. Electric cars might depreciate differently from gasoline cars. If a driverless car costs much more than a normal car, even if it depreciates by the same percent and in the same ratio of milage to time, then it’s a win for Uber (that said, if a driverless car costs too much, then people will just stick with their normal cars).
I said that an Uber car is roughly four times more utilized than a personally owned car. This is conceivably an underestimate — maybe an Uber might serve five passengers, or six. But as long as most people are going to work at roughly the same time, you won’t see an Uber serve 10 or 20 passengers. Demand for vehicles is very spiky. If the car can be utilized by six people instead of four, then that’s worth an extra few hundred per year in savings to the Uber customers.
On the other hand, maybe an Uber will only serve three people. That’s worth an extra $200 or so per year in savings to the owners.
The most sensitive estimate I made here is how much extra milage an Uber vehicle incurs. Each extra 5% that the vehicle incurs erodes about $300 of Uber’s cost advantage. If an Uber vehicle incurs a 25% milage surcharge (all else being equal), it is the more expensive option.
Uber Profit Margin
10% profit margin on each car is not very much! In our scenario, Uber is a high-volume, low-margin business like Amazon is today. And it’s also a capital-intensive business. It’s not at all clear that Uber could operate at anything nearly as low as 10% unit margin. If we increase its profit margin to 20%, then almost half of the cost savings to the Uber user vanishes.
Long story short, there are a lot of assumptions here. If you go in wanting to believe something about whether it will be more advantageous to rent or own, you can twiddle the dials until it shows anything. I guess you’ll have to take me at my word that I tried to think up stuff that seemed reasonable and then didn’t go back and change anything after my numbers came out.
Of course, there are differences between owning your own car and renting from Uber besides price. If you own your own car, you can leave your stuff in your car. You will never find a car is not available or is surge-priced out of your price range. If you have kids, you won’t have any additional hassles about getting a car with an appropriate child seat or installing said seat. Your car may be a status symbol. You might have specific preferences about the car (perhaps you’re very tall) that are inconvenient in a commodity vehicle from a fleet.
On the other hand, if you have not sunk a large amount of money into a fixed-price car, you can more affordably get a car for your specific situation — a big car for when your family is going together, and a small car for when you commute alone. You could rent a fancy car for date night and an economical car for your day-to-day life. And so forth.
Many people are expecting an overwhelming swing to a rental model of car use in the driverless car age. Those people are largely speaking failing to understand how much of the costs of a car are based on milage, not time.