More on the economics of transit use vs driving

Last week, we highlighted a recent report by the American Public Transit Association which claims that in Portland, an individual can save on average $859 per month–over $10k per year–if one holds a transit pass instead of owning an automobile. To be clear–this claimed savings requires not owning a car (or owning one less than there are drivers in a household); much of the savings include fixed costs that are incurred whenever you own a car, even if it is left in the garage. There were some questions about the automobile costs cited–while nearly $1000/month is a reasonable cost of ownership for a newly-purchased automobile of median value; it is easy to own and operate a vehicle in reasonable condition for far less money than that. On the other hand, at current fuel prices, the $88 charge for a transit pass is less than the cost of two tanks of gas.

Today, we look a bit more in detail on the economics of transit use vs automobile use, and how public policy might affect the equation going forward.
Fixed vs variable costs

When determining the cost of using (and having access to) a resource, it is useful to separate the costs into fixed costs (or access costs), which are incurred and must be paid simply to have access to the resource; and variable costs (or usage costs), which are incurred for incremental utilization of the resource. The combination of costs can range from entirely fixed to entirely variable.

In many cases, it is possible to arrange matters so that a different schedule of fixed and variable costs applies to a given resource. For example, you can choose a cell phone plan with unlimited minutes and a flat monthly fee (high fixed cost, low variable cost), or a plan with a lower monthly fee but which accrues per-minute charges (lower fixed cost, higher variable cost). You can choose to dine at the all-you-can-eat buffet (fixed cost for a meal), or go to a cafeteria and pay for your meal a la carte. And so on.

The ability to trade off fixed costs vs variable costs is an important consideration, for both parties to a transaction. For both sides, converting costs to fixed costs makes the supplier’s revenue stream more predictable; a benefit which often is used to justify lower prices. Fixed costs also reduce the need to process incremental payments for incremental usage. On the other hand, fixed costs may encourage overconsumption of a resource, particularly if there isn’t some other limit beside price involved–this is the justification phone companies give for recent moves to no longer offer unlimited data plans (whether you believe that is up to you). Important for this discussion, arrangements with high fixed costs reduce flexibility by encouraging lock-in. If A and B provide competing services; and you enter into arrangements where you pay a fixed price for free (or reduced-rate) consumption of A; this is a powerful disincentive to use B. Even if it might meet your needs under certain circumstances: You’ve already “paid for” A, so the rational choice is to use it.

And for many commuters, A stands for “automobile” and B stands for “bus”. If you already have a car sitting in the driveway, all gassed up and ready to go, buying a transit pass on top of it frequently makes no economic sense. As we live in an auto-centric society, one of the things limiting the potential of transit is the fact that many of us already have cars at our disposal. Given that much of the cost of that car is “paid for”, there is frequently little economic advantage for using transit, except for those trips where the marginal cost of driving is unusually high, either in dollars or in time. The daily commute is an example for many, particularly those living in the suburbs and working downtown.

Of course, if you already have a transit pass, and live in a dense urban environment where all the amenities you need can be reached by transit (or on foot), you may find you don’t need a car. Low density land use patterns (the UGB nonwithstanding) don’t make this an attractive option for many; but the economics of car ownership–which make occasional use of a car expensive–further make it difficult.

To see why, it is useful to examine the cost structures of both transit use and auto use. This article only discusses voluntary costs; general taxes needed to support both transit and the roads are excluded from this discussion. Also excluded are externalized costs such as pollution and congestion.

The cost structure of transit…

The cost structure of transit is pretty straightforward: You get on the bus, you need to pay the fare. In Portland, you can buy single-ride tickets for two bucks and change; you can buy a day pass for less than $5. A monthly pass is $88, and if you buy a year’s worth you get one month free. (Prices are for all-zone adult fares). The monthly pass represents a significant savings over two single-ride tickets per day; so is a no-brainer for anyone using transit regularly, even if only for commuting to work and back.

Transit also has the advantage of cost predictability. If you hold a monthly pass, you know with high certainty what your ride is gonna cost each month–there aren’t any (financial) surprises. If a bus breaks down, fixing it is on TriMet’s tab; and TriMet supplies the elbow grease to get the job done.

The question of time is where the analysis gets interesting. Transit has the advantage that when you’re a passenger and not a driver, you can do something productive with your time other than operating a motor vehicle. On the other hand, many journeys take longer on transit, and may involve the need to transfer between services, which can sometimes be unpleasant–particularly if the wait times are long and/or the transfer location is inadequately sheltered. In many parts of town, unfortunately, transit service is poor, and trips take significantly longer than they would by car. If you live in Tualatin or Happy Valley, for instance, TriMet is probably only an option if you are desperate–people who live in such places are auto-dependent.

There is one other important caveat to the low cost of transit claimed by APTA, however: it is per-person. If you have two or more adults (or older children) in a household who want to use transit, each will need a pass of their own. Young children can travel free with fare-paying parents. The costs associated with cars, on the other hand, do not vary with the number of passengers.

…and of driving

Costing models for automobile operation are far more complex. The APTA study uses a model published by the American Automobile Association, which includes the following. Fixed costs include depreciation, finance charges, insurance, licensing, and registration; variable costs include fuel, maintenance, and other consumables (most notably tires). The model doesn’t make any mention of parking or tolls, or the cost of being pulled over by Johnny Law (for those motorists who make moving violations a habit). The costs appear to assume a new car, as noted above, and different price points are provided for different levels of usage. The model does not appear to include the time cost of driving–when you are behind the wheel, that is time lost from more productive or enlightening activities.

A quick note on how the model accounts for the purchase price of a car. Car purchases are capitalized in the model–the base purchase price of a car is not considered an expense per se; though finance charges are. Instead, the depreciation of the vehicle–the loss of value over time–is booked as an expense. Note that for cars in the consumer market, amortizing depreciation over time rather than over mileage is an appropriate treatment; personal autos lose most of their value simply by virtue of sitting around–the calendar is just as important as the odometer in determining a car’s residual value.

Unlike with transit, where it is easy to jettison fixed costs and only incur variable costs simply by buying single-ride tickets rather than a pass; this is much harder to do with cars. If you don’t have a car, you can’t drive. Leasing a car is similar to owning one as far as fixed/variable costs go; automobile leases are typically structured to cover a fixed term for a fixed rate, with per-mileage charges only applying if a certain threshold is exceeded.

Other business arrangements for acquiring access to a car are considerably more expensive, and only make sense for occasional use. Traditional car rental services are generally time-based rather than mileage based (rates are per day or per week, with unlimited mileage); renting a car is often a high-friction transaction, and the car rental industry is infamous for lack of transparency, price discrimination, and creative ways to nickle and dime customers to death.

Car sharing services, such as Zipcar, offer more reasonable terms, including hourly rental, far less inconvenience, and none of the games and hassles that characterize the car rental industry (though some car rental firms, such as Hertz, are also getting into the car-sharing business). Zipcars can be rented on either a per-hour basis (about $7 per hour, depending on plan) or per day ($60-$65); and either rate includes gas, insurance, and maintenance. Car-sharing services are geared towards occasional users, however; it is generally not economical to use a Zipcar as a primary means of transportation.

Driving an automobile often involves other, unforeseen, costs and hassles–car repair being a major one. There’s also the cost of time spent behind the wheel, which cannot be spent on other productive activities. On the other hand, in a country optimized for automobile travel, with low average densities and ample free parking, travel times by car are often a fraction of transit times, especially when the suburbs are involved. Automobiles also have the advantage of being able to carry passengers (and cargo) essentially for free; that said, many of the objections that environmentalists and others have to single-occupant vehicles don’t apply (or apply less) to cars carrying multiple passengers. (A bus, after all, is essentially a car designed to hold 80 rather than five.)

Putting it together

Public transit has the advantage, generally, of being available at fare lower price points. As noted above, a single ride cost $2.35, and an annual pass is less than $1000; both of which are far cheaper than any comparable arrangements which can be made using cars. Unfortunately, the costs grow with household size, and several decades of car-friendly land-use policies have produced a suburban environment which is difficult to provide quality transit service to. Driving, on the other hand, has a high entry cost, but marginal costs per mile can be low for those who bite the bullet and decide to own one. Driving also makes more sense for larger families, and is further advantaged by the extensive network of roads, mandatory free parking, and other concessions to the automobile in our built environment.

Looking at the AAA costing model, it indicates that for a small sedan (a Toyota Corolla or similar), owning and driving it 10k miles will cost 58.6c per mile; at 15k miles, 45.1 c/mile, and at 20k miles, 38.1c/mile. Assuming a linear cost structure and doing the algebra, this works out to a fix cost of ownership of $4100 and a cost per mile of 17.6c. For a minivan, the numbers work out to 25.2c/mile, and a fixed cost of ownership of $5680. For a SUV, 27.8c per mile and $7040 of fixed costs. Figures for other car types (mid-size sedan, large sedan) are also presented–calculating their numbers is left as an exercise for the reader. Given that an annual transit pass is presently $968, for transit to make more economic sense (ignoring externalized costs and questions of time and convenience, as well as unaccounted-for costs such as parking and tolls) one would have to drive 5700 miles less if one owns a small sedan, ~3800 miles less for a minivan, or ~3500 miles less for a SUV. Assuming 250 workdays per year, that comes to 22.8 miles/day (total, not each way) for the sedan owner, 14.8 miles/day for the minivan, and 14 miles/day for the SUV.

Given all of that–the present mode split and demographics of transit and auto use aren’t too surprising; the primary users of transit are the poor (who frequently cannot afford automobiles), and urban dwellers and downtown commuters (who incur higher costs, both financial and otherwise, for automobile use). Outside these three groups, its frequently easier to just go buy a car or two–and once you’ve done that, the natural and economic thing to do is to drive it everywhere.

But–as noted above–this analysis ignores the external costs of transportation. For transit, the (uncaptured) external costs are fairly low; but for driving, they are quite high–operation of two-ton machines to transport individuals around is a gross waste of energy, and a primary source of pollution. Sound public policy is that which attempts to rein in externalized costs–given that, what can be done about this state of affairs?

The usual exhortations to improve transit service and reform land-use laws to discourage sprawling, auto-centric development obviously apply, but are outside the scope of this article. Instead, we look at ways to further migrate transit to a fixed-cost basis, and move driving to a more variable-cost basis–and to capture externalized costs as well. Unfortunately, one of the biggest cost drivers–deprecation–is highly dependent on the behavior of the automobile resale market, and as long as motorists perceive older cars to be less valuable than newer ones with similar mileage, treating depreciation as a fixed cost is likely to remain the proper way of accounting.

Reasonable policy actions include:

  • Family tickets/plans for transit. I’d like to see TriMet offer fares better geared towards families, particularly families travelling together. Right now, children under seven ride free with a fare-paying adult, and 7-17 year olds get a reduced rate. My recommendation? Bump up the free age to 10. Why 10? Because that’s the age at which it is legal to leave a child at home unattended. (The law is a bit ambiguous for ages below 10; but a 10-year old is presumed to be able to attend to himself). Barring that, it might also be useful to offer other family discounts on passes for members of the same household. By encouraging families to use the system, TriMet not only encourages off-peak business, but also encourages the next generation of riders.
  • Usage-based insurance, also known as “pay-as-you-drive”. Traditional auto insurance policies offer liability insurance for a fixed term, typically six months, and charge premiums based on prior driving record. An obvious inefficiency is present in this arrangement, as risk is proportional to how much the car is driven; liability policies in particular impose no risk when a car is non-operational. By offering usage-based insurance, motorists can pay for the coverage they need. Usage-base policies can be based on simple odometer readings (or policies which are only good for a certain number of miles), all the way up to use of sophisticated devices which electronically monitor driver behavior, including time and distance driven. Portland Afoot has a good summary of the state of mileage-based insurance in Oregon; no company presently offers true mileage-based insurance, where you purchase insurance for a set number of miles rather than a set time (a company called MileMeter offers such a program in Texas). Three insurers which do offer substantial mileage-based insurance products are include Traveler’s and Progressive, both of which supply motorists with onboard monitors to install in vehicles, and National General Assurance (formerly GMAC), which provides a similar service to OnStar customers. Progressive Insurance’s “Snapshot Discount” program is probably the most well-known usage-based insurance product, although it is a bit limited in its scope> a monitoring device is installed in a vehicle for 30 days; drivers who drive less, avoid dangerous driving times (rush hour or late at night), and avoid sudden stops are eligible for up to a 30% discount off a standard policy.
  • Mileage-based fees. Many of the taxes and fees paid to support the roads are either levied against everybody (such as local property taxes) or levied against motorists on a flat-rate basis. Registration fees are essentially de minimis in Oregon, and thus likely don’t contribute significantly to mode choice; but in many state, registering an automobile is a significant expense–and these states generally either levy a fixed fee, possibly based on vehicle classification, or a “personal property tax” based on the value of the vehicle. Oregon is one of a handful of states to impose a weight-mile tax on trucking; but presently no state imposes any distance-based tax levies (other than fuel taxes) on personal automobiles. However, ODOT recently ran a pilot program to determine the technical feasibility of implementing and collecting a “road user fee” from motorists based on mileage–a fee which would replace the motor vehicle tax. (Portland Transport coverage of the project here and here). The pilot program only looked at technical issues, not at the politics or the policy of the matter and produced the following report.

    This year, the Oregon House introduced HR 2328, to implement road usage fees. Unfortunately, the bill (which is making its way through the House and has not been taken up by the Senate) only covers electric or hybrid vehicles; not all vehicles, and is intended to “make up” for the fact that these vehicles don’t “pay their fair share” of gas tax. A fee of $0.0156 per mile is currently proposed within the bill; owners of hybrids who also purchase gasoline (subject to gas tax) would be eligible for a gas tax refund. One other element of the bill which I dislike is a “buyout option”; high mileage drivers can elect to pay a flat rate of $300 (equivalent to 19231 miles) rather than pay the per-mile rate.

  • Pollution fees to replace the flat DEQ testing charge. A road usage fee, properly implemented, could also replace/enhance the DEQ testing program as well. DEQ inspections carry a flat fee (levied on all non-exempt vehicles every two years); this fee is only intended to cover the costs of administering the program; not to capture the cost of vehicle pollution. A useful change might be to roll the cost of testing (and pollution charges) into a RUF, so that cars which are driven more pay more. While I’m on the subject, I’d expand the emissions categories of vehicles from “exempt”, “pass”, and “fail”–while vehicles which fail to meet minimum standards ought to be black-flagged as before; the “pass” range would be further divided into subranges, with vehicles that pollute more being levied higher fees. (Oh, I’d also end the ridiculous practice of exempting very old cars from the DEQ testing program program. If you like driving your beautifully-restored, cherry-red ’52 Filthbelcher around town, with its carbeurated engine and lack of modern emissions equipment, you should pay for the pollution this causes. No need to subsidize the hobbies of auto enthusiasts).
  • Peer-to-peer car sharing. While the praises of Zipcar is sung above, not everyone will find it a useful service. Another possibility is peer-to-peer carsharing, in which owners of automobiles can make their cars available for short-term rental by others. There are legal obstacles to this, but a bill (House Bill 3149) in the Oregon Legislature designed to enable such services by requiring auto insurance policies to “play nice” with carsharing programs was recently passed by the Oregon Senate by a wide margin. (A similar version was previously passed by the House; minor technical differences between the two bills must be ironed out before it can be sent to Governor Kitzhaber for his signature).
  • Per-mile leasing. One other idea which might be useful is changes to the automobile leasing market. Leasing cars has long been trickier than leasing real estate, in no larger part because vehicles lose value a lot faster than land and buildings do. The current common structure of a vehicle lease (the “closed-ended lease”)–with a fixed monthly rate and a fixed term, and a mileage threshold that the lessor must keep below or incur a penalty, came into being due to abuses prevalent with open-ended leases. Were auto dealers to offer mileage-based leases, with a larger component of the payment structure mileage-dependent (something more practical with modern technology); it might help bridge the gap between automobile ownership and car-sharing. OTOH, given the nature of the used car market–I can see why leasing companies wouldn’t be interested in going along.

36 Responses to More on the economics of transit use vs driving