The Cost of Driving

(You can ignore the “per gallon” parts, they cancel in the division, but for some reason my mind doesn't want to do that. I have to make them go away by thinking about a single gallon of gasoline.)

Once you know the cost per mile, you can use that to calculate the cost per trip. Suppose I want to drive up to Rocky Mountain National Park to go hiking. That's about 40 miles, so it will cost me $4.80, or about $5.00. I like to think of that as the price of a ticket, a one-way ticket from Boulder to RMNP. That brings it down to earth for me and makes it easy to compare to other things.

The cost per trip isn't exactly like a ticket. If there are passengers, they can share the cost instead of having to buy their own tickets. But, I can't think of any commonplace good that it's more like. It's very much like taking a taxi, but that's not commonplace for me, and also it seems a bit circular. Yes, the cost of traveling by car is a lot like the cost of traveling by car.

That's about all I have to say about the cost of gasoline. Now let's step back and look at all the other things that contribute to the cost of driving.

1. There's the one-time cost of acquiring a car. This cost can take several different forms. If you pay cash, the one-time cost is the entire price. If you finance, the one-time cost is the down payment. If you lease, I don't know, is there an up-front payment for that? And possibly a second one-time cost if you opt to buy the car at the end of the lease? I'm not even going to talk about car sharing. I'm not opposed to it, I just imagine it has a completely different cost structure.
2. There are per-time costs. Some are monthly, like car payments (principal plus interest) or lease payments. Some are yearly, like registration and insurance. And, some may occur at other intervals. For example, every two years I have to take my car in for an emissions test. If you want to take the long view, you can even think of the cost of acquiring a car as a per-time cost, but I prefer not to since it's not required to occur at any fixed interval.
3. There are per-mile costs. The cost of gasoline is one, but we've already discussed that. The cost of tires is another. As it happens, I just bought some fairly nice tires last year. The total cost divided by the warranty mileage comes out to around 0.8 cents per mile, so the cost of tires is just a small correction to the cost of gasoline. I don't know any other costs that are clearly tied to mileage.
4. There's maintenance. If you follow the manufacturer's recommendations, maintenance should be done as a function of both time and mileage. For example, here's what my owner's manual says about changing the oil.

   Engine oil and the oil filter should be changed together every 6 months or 7,500 miles (12,000 km), whichever occurs first.

   I'll also include repairs in this category. Repairs are different from maintenance because it's chance whether they're needed, but the chance probably accumulates with time and mileage in much the same way.

5. There are per-use costs. Tolls and parking fees are the main examples, although if you pay them every day (or every weekday), they become per-time costs instead. If you live in a big city, you might even pay monthly rent for a parking space. Parking tickets are essentially probabilistic parking fees, so we can include them here, and once we've done that, we might as well include speeding tickets too. That's not a complete list of per-use costs, but the only other one I want to mention is the cost of going to the car wash.

With that lovely overview in hand, we're now equipped to tackle one of the great mysteries: the IRS standard mileage rate. For reference, here's the current value.

The 2014 rate for business use of a vehicle is 56 cents per mile.

The rate values have always seemed absurdly large to me, but then they're supposed to account for basically the entire cost of driving (not including interest and maybe a few other little things like tickets and car washes), so maybe I've been wrong about them all this time. Let's look at some numbers and find out.

1. I bought my gold 1990 Acura Legend used in 1998 for $8,000, and I've driven it about 80,000 miles. So, the one-time cost is 10 cents a mile. We can also calculate the useful fact that I drive an average of 4,700 miles per year.
2. For me, registration and insurance are the main per-time costs. Adding up what I pay per year and dividing by 4,700, we get a number that's well under 10 cents a mile.
3. We already know that gasoline and tires cost me 12.8 cents per mile. Let's go ahead and round that up to 13.
4. I keep all my maintenance and repair paperwork in a single file folder, so it wasn't too much trouble to go through the invoices for the last few years and add up the amounts. The total divided by the number of years and then by 4,700 comes out to a bit more than 10 cents a mile.
5. There aren't any tolls or parking fees that I have to pay regularly, so the total per-use cost is negligible.

Adding up all the categories, we get a grand total of about 40 cents per mile.

However, that number irritates me—it's artificially high because I don't drive much. If I drove more, all the costs besides the true per-mile ones would be spread out over a larger number of miles, and the grand total would be smaller. Now, a quick search tells me that the average American drives 13,500 miles per year, which is about 3 × 4,700. So, if I drove that much, the grand total would be 13 + (40−13)/3 = 22 cents per mile, plus a bit to account for the extra maintenance and repairs.

So what does it all mean? If the IRS number represents the average cost of driving, how much more is the average American spending than I am? Maybe we should compare 56 to 40 and say 40% more? Or maybe we should compare 56 to 22 and say two and a half times as much? Or maybe we should subtract out the true per-mile costs, since those are about the same for everyone, and compare 43 to 27 (60% more) or 43 to 9 (five times as much)? All the answers are plausible, so I guess I have to take back what I said about the rate values being absurdly large.

However, the real conclusion that I want to draw here is that trying to reduce the entire cost of driving to a single per-mile number is stupid and confusing. I think my five categories are a reasonable way to think about the cost of driving, but if you're determined to reduce something, here are two supercategories you might like.

1. Costs caused by the decision to have a car. In other words, this is what you pay for the privilege of being able to drive whenever and wherever you want. This covers categories 1 and 2 and most of category 4. In economic terms, this is the fixed cost.
2. Costs caused by the individual decisions to make individual trips by car. This covers categories 3 and 5 and the rest of category 4. In economic terms, this is the variable cost, made up of marginal costs.

Finally, here are a few random thoughts.

Apparently I've had notes for this essay sitting around since the year 2000, when gasoline was $1.50 per gallon and the IRS rate was 31 cents per mile.

Since I've been talking about the cost of gasoline, of course I have to mention The Age of Transportation. It seems silly right now because prices have been low recently, but I don't see that the fundamental situation has changed.

When I was describing the five categories, I tried to make everything sound as simple and exact as possible, but if you dig into the details, it turns out that a lot of the numbers are still only approximate.

2. Per-time costs work pretty well as long as the time scale is yearly or monthly. Above that, rates change unpredictably. Below that, you get into the usual calendar-related troubles. For example, for a monthly payment, the per-day cost is higher in February because there are fewer days. To put it another way, a stream of monthly payments is not a uniform flow of cash. The payments may occur on different days of the month, and even if they don't, they're still not uniformly spaced because of the calendar. And, if we ignore the calendar, they still can't be uniformly spaced because 365 and 366 aren't divisible by 12.
3. The per-mile cost of gasoline equals the price of gasoline divided by the mileage in miles per gallon, but neither the price nor the mileage is constant. The price is widely different in different parts of the country, and slightly different at different local gas stations. It depends on all kinds of factors including the global oil market. The mileage depends mostly on what kind of car you have and what kind of driving you're doing (city or highway), but also on whether you're stuck in traffic, whether you're going uphill or downhill, and whether the engine is warmed up. It probably even depends on the air temperature and thus on the weather.
4. When is maintenance really needed? It all depends on the kind of wear being put on each part of the car. Think about a manual transmission. Driving along in gear produces one kind of wear (per time, on that particular gear), shifting produces another kind, and grinding the gears another. Or think about the shock absorbers. They receive wear when the car hits a pothole, and that depends on how many potholes there are (which depends on the city government) and how careful you are about avoiding them. So, maintenance isn't just a function of time and mileage, it's also a function of shifts and potholes and so on.

Also, speaking of calendars and approximations and tiny details, see An Example.