Consider an ice cream stand. Like any business, it has revenues (ice cream sales), and expenses (cones, refrigeration, city permits). One can figure what a cone ‘should’ cost by adding up its components – raw materials, labor, utilities. For example, suppose ice cream costs $2 per gallon wholesale, there are 20 scoops per gallon, it takes a minute to scoop ice cream, labor is $12 an hour, cones cost five cents, and the stand itself costs $144 for a twelve-hour day. A two-scoop cone then ‘should’ cost $0.65, and that’ll be the price in a perfectly competitive market.

But we’ve missed one assumption. What if we aren’t scooping ice cream all the time – what if there are dry spells? We have fixed capital costs (the worker and the stand), which have to be amortized over a large number of sales. If there are fewer sales – if the worker, having nothing to do, pulls up a chair and gazes down the wide open road every now and then – the costs are split fewer ways, making each sale more expensive. If the stand is out in the boonies and only makes one sale per hour, that one sale has to cover the whole hour of worker time and stand time, and that two-scoop cone now has to cost $24.45. A large jump!

Standby costs are a large component of many businesses. Restaurants still have to pay rent at 3 PM, when no one is eating. A rural store only gets a few customers a day, making its products expensive. The railroad mostly runs at rush hour, but the trains are around 24/7/365…

One can be very successful by finding new ways of eliminating standby costs. RelayRides, for instance, makes money by employing capital (people’s cars) more hours per year. Ultimately, self-driving cars will do the same, on a much larger scale. And what a bang that will make.