The unit economics in carsharing are extremely challenging. I had an interesting conversation with Ezra Goldman yesterday, who made the case that the costs are as follows, in order of cost:

– Parking
– Insurance
– Inventory

I’ve heard arguments that Zipcar’s challenges in unit economics resulted from cost-prohibitive insurance. I’ve heard others claim that it was from holding inventory. But apparently a fleet of cars involves high startup costs, but maintenance costs (holding inventory) are actually low, and ROI on the inventory is not that bad. In fact, all major rental companies are actually used-car salesmen on the side. And the maintenance is a pain, but it’s not prohibitive. By timing when to buy and sell the cars they have figured out the way to minimize the depreciative aspects of car-ownership.

And insurance is expensive, but usage-based insurance (UBI) models in the incumbency (particularly Progressive), alongside some very interesting startups (MetroMile as a notable example) are creating underwriting practices that will ultimately change how we think of car insurance. It’s a race to the bottom with UBI, and as the practice is more widely adopted and the model gains traction, insurance costs will drop.

But in the places where carsharing works best — cities — parking is a central piece of the puzzle, and a particularly challenging one. For the demand experience to be a positive one, the supply needs to be close enough to any given user that it’s worthwhile for as a viable mode of on-demand transportation. This means that at scale a platform needs to invest in parking spaces that are accessible during reasonable driving hours at least, if not 24 hours. It means that these parking spaces need to be permanently assigned to a car, so that the experience is sufficiently convenient to beat alternatives. In a large dense, city, that is requires an enormous amount of (inefficient) capital because on the weekends, the Financial District cars are sitting unused in parking spots that cost a fortune, while at 2pm on a Tuesday the residential parking spots are similarly unused. This is the beast, and until this nut is cracked, the industry will have a tough time.

The newer peer-to-peer carsharing platforms don’t hold inventory, and only deal with parking in limited amounts, so one would think there is an opportunity for them to capitalize there, but in the case of a Getaround or a RelayRides, the problem is less with the unit economics (which are very challenging, in and of themselves), but with the sheer volume of supply that is required to service an on-demand experience on the scale of Zipcar or CityCarShare. Since it’s hard to know when I’ll want to use my car, you have to create a model that has a statistical buffer for when people use their own cars. And with revenue promise on the supply side being somewhat limited, it’s unclear how you get that much supply quickly.

It’s clear to me, though, that intra-city transportation and shared mobility are going to continue to explode. I eagerly await the self-driving overlords, who will solve the parking challenge, and probably a few others.