What is the right p and mc to use in evaluating transit operations? (Following up on a comment from @trnsprtst)

David Levinson, who probably is really the smartest urbanist (he’s really, genuinely smart, people) piped up on Facebook in response to my post about transit passes with waitaminute…shouldn’t we thinking in terms of average costs, and not marginal costs, since the marginal cost is often zero? It’s a very good point. He’s probably right. I wasn’t being particularly careful with how I used terms in the blog post, and even if he isn’t right or I wasn’t wrong (you’ll see why I hedge in a minute), he’s raised an important point I skipped in my discussion.

Cost curves in transit are not smooth. Once you make a decision to send out a vehicle, your cost of adding an additional passenger is low–I won’t agree to zero, but low–until you hit another meaningful margin (a term my old mentor Tom Pogue used) when you really do have to think about taking on sizable new capital investment, like expanding capacity for electricity generation–or in our case, another bus, another train car, or another entire train. These margins occur at both the variable costs and total cost levels: another vehicle means another driver, and the marginal cost depends on whether we’ve got a driver available we have to pay already (or not.) We also have marginal conditions to consider when thinking about adding new vehicles to the fleet (the total cost question).

My default thinking tends to center on those meaningful margins, not on the small increments in between. By using an average cost, Levinson spreads the costs of those big margins and small ones to get a more general picture of how much it costs to serve passengers. I think both ac and mc measures are useful, depending on the context, so the right mc or ac to use depends on what kind of operating decision you are going to make.

I tend to think about the big margins and big units simply because I don’t think it’s generally important to the agency how much a given passenger pays or costs; I think in terms of whole buses or whole trains–the costs per revenue hour, and the revenue per revenue hour. That’s where clarity comes on whether or not it’s possible to stay in the black running the service, even if you have to aggregate across all the lines you run to get to the meta question of “how much service is worth supplying?”

That said, I also am dubious of the idea that the marginal cost is nothing when adding to a vehicle that is below capacity. It’s a low cost, certainly, but even then the marginal cost is likely positive, for buses anyway. Crowded buses are an irritation not just because you are wedged in like a sardine, but because the more passengers, the more stops, and that time cost becomes considerable, both for the passenger and the agency. (The more stops a bus makes, the longer it takes to cycle through the route, the more buses we will probably need in order to make good on the posted headways).

Not much blogging here late, and not much content here, either. I’m tired. I worked myself into a bit of a state early in the summer, and now I think I need to rest. More when I feel better.