Congestion pricing is hard. The fact that it floats up and down with demand is confusing. As is the fact that someone paying a congestion toll is not, technically, paying for either road construction or road upkeep. He’s just paying for getting in everyone else’s way.
Where it gets really confusing, though, is that while congestion revenue is not collected to pay for the road, congestion revenue can be used to pay for the road. And when it’s done right, congestion — or, more precisely, the price people pay to avoid it — pays for the whole thing.
I’ve tried to explain this here before without much success. This time I’ll try an example.
Let’s suppose Gotham is a big metropolis and Suburb is Gotham’s aptly-named suburb. There’s no good way to get from Suburb to Gotham — they’re separated by a river and a deep canyon that a pack mule couldn’t cross, much less a car. The lack of a convenient route from Suburb to Gotham more or less defeats Suburb’s raison d’être. Suburb therefore decides to build a highway bridge to Gotham.
Suburb forms the Suburban Mobility Authority (SMA) and staffs it with smart transportation engineers and equips it with powerful computers. The city government is shot through with fiscal conservatives, though, and insists that the bridge be revenue neutral. It absolutely must pay for itself.
The transportation engineers get together, look at data, run simulations, consult oracles, and conclude that if they build a six-lane bridge from Suburb to Gotham and charge the drivers congestion tolls, the congestion revenue will just cover the cost of the bridge.
Two years later, the William Vickrey Memorial Bridge opens for business. The SMA charges a variable toll: $5 round-trip during peak hours, a little bit less during the near-peak hours, a little bit less before and after that, and so forth, with no charge at all when demand is light.
An average of 50,000 drivers a day begin taking the bridge during the peak commuting time. Others drive during the cheaper near-peak periods to enjoy the slightly lower tolls. Some drivers drive at weird times of the day to avoid tolls altogether. And some Suburbanites — those with no reason to drive at all if they can’t drive during peak periods — sit home, stare out the window, and mutter about gas taxes and the sons-of-bitches at SMA. But the drivers who do pay a toll pay just enough, collectively, to cover the bridge’s bond and maintenance costs. The transportation engineers have nailed it. They get written up in Traffic Engineer Monthly and everyone is happy (except the mutterers).
It turns out that it’s not the same 50,000 people driving at peak hours every day. Some days Joe is content to work from home. Some days he is willing to leave very early or (to his boss’s displeasure) very late to avoid the peak toll. And some days he is in a hurry and gladly forks over the money. But, on an average business day, 50,000 drivers are willing to pay the $5 peak toll.
Over the years, thanks to its enlightened government, Suburb’s population surges and eventually doubles. Now there are more people who are in a hurry some days, more people who don’t like leaving very early or very late for work, and more people who get a strange, visceral satisfaction from giving their money to a toll authority. The revenue generated by the congestion tolls creeps steadily upward until the SMA is charging $10 per round trip during the peak period rather than $5 as before.
The SMA is now raking in cash, far more than it needs to pay the bonds and maintenance. The SMA officials could blow the money on lavish office parties and good scotch, but they’re good bureaucrats and decide the extra money should be used to build more capacity. But their mandate remains: Any new capacity shall pay for itself. Whatever capacity they build must give commuters what they are willing to pay for but no more.
The SMA’s engineers break out their computers, slide rules and eyes of newt, and start figuring out the best way to add capacity that will pay for itself. They quickly rule out expanding the Vickrey. The six-lane span just can’t be widened, and the canyon’s topography precludes another wide bridge parallel to the Vickrey.
They could build another six-lane bridge upstream, but that would require building a five-mile long road to Farmington, erecting a new bridge across the river there, and then building (or asking Gotham to build) another five-mile long road back to Gotham’s downtown. They run the calculations and conclude that this roundabout highway would indeed draw some traffic. But it would not draw enough interest — meaning enough drivers to get in one another’s way — to pay for itself, even counting the surplus congestion revenue from the Vickrey.
They consider a third possibility. While there is not a feasible canyon crossing for a new six-lane highway bridge, there is a place suitable for a narrower, lighter bridge wide enough for two train tracks. Building a bridge for commuter rail would be cheaper than building a second bridge in Farmington and it would provide a shorter trip to boot. The engineers conclude that’s the best option. Three years later, the commuter line opens for business.
As predicted, some who were driving the Vickrey switch to the commuter rail to avoid the high peak and near-peak tolls. Some who were driving at weird times of the day to avoid tolls switch to rail trips, too. And some people who had never dreamed of going in to Gotham at all because they don't like either tolls or weird travel times start riding the rail. There’s plenty of demand. So much, in fact, that the rail would be congested at the peak hours if it were free. Long lines would form at the fare machines, the car doors would have to stay open for precious extra seconds to board all the passengers, and clean-smelling people would be packed in tight with the hygienically challenged. The SMA therefore charges a $2 round trip fare during peak hours to clear congestion. Even at this $2 round-trip fare, 75,000 commuters ride the train every day during peak hours because it is better than their alternatives.
The new commuter line also affects the Vickrey. Due to people switching to the train, the peak toll on the Vickrey drops even though it continues to carry 50,000 during the peak rush hours. It doesn’t drop all the way back to $5 per round trip — the car is still faster and some drivers just don’t like the looks of train riders — but it drops from $10 to $7 per round trip.
The transportation engineers, it turned out, did a good job with their projections. The total revenue collected by the Vickrey and the rail line are exactly enough to pay for the construction and maintenance of the rail line plus the cost of the Vickrey.
But they did not obtain that happy result by making the commuter rail pay for itself. After all, the Vickrey was paying for itself with a $5 toll; the SMA is now collecting a $7 toll. The Vickrey, in other words, is still producing "excess" revenue. Since the combined revenue from the Vickrey and the rail line is just covering the cost of both, the Vickrey’s drivers must be paying part of the cost of the rail. They are — horror! — subsidizing the commuter rail riders.
But both car drivers and train riders are better off than before. The 50,000 people who drive over the William Vickrey Memorial Bridge during peak hours are better off because they’re paying $7 per round trip rather than $10. The commuter train riders are better off because they prefer paying $2 for a train trip to $7 (and certainly $10) for a trip across the Vickrey during peak hours. Or to driving at inconvenient hours. Or to sitting home, staring out the window and muttering about gas taxes and the sons-of-bitches at SMA. Even the engineers who planned the system are better off because they not only get another nice write up in Traffic Engineer Monthly, they get a glowing review in Mass Transit Today. The government, of course, is glad it could help, especially since the new capacity was revenue neutral. Sure, there are still mutterers who stay home, but they were doing that anyway so they’re no worse off, either.
Now here’s the counter-intuitive point: Not only is every one better off than before, the SMA can’t make anyone better off by charging different fares. Let me repeat it: The SMA can’t make anyone better off by charging different fares.
To see this, let’s assume the SMA raises the train fare by $2 per round trip because the free-market fairy told it that transit is supposed to pay for itself.
Who benefits and who loses?
Some train riders continue to ride the train during peak hours. They’re now paying $4 per round trip rather than $2 per trip. They are clearly worse off.
Some train riders switch to riding the train at off-peak, inconvenient times. Some switch to driving the Vickrey at off-peak, inconvenient times. Some even switch to the Vickrey at peak hours and just pay the toll. We know that all of these people are worse off with the $4 fare, though, because they preferred to ride the train during peak periods when the fare was $2.
Moreover, their being worse off doesn’t make anyone else better off (except possibly the government) because they weren’t generating any congestion; we’d already figured out that $2 was the train’s congestion-clearing price during the rush hours. And they weren’t costing the transit authority any more money aside from the wear and tear their posteriors put on the seats. In other words, their marginal cost before the fare increase was exactly the $2 they paid.
But perhaps the peak-hour Vickrey drivers are better off. After all, they’re now providing a smaller subsidy to the train riders.
In fact, the Vickrey drivers are worse off, too. Some people who were riding the train have decided to switch back to the Vickrey — the car is faster and now relatively less expensive. Since more people are now chasing the same number of slots on the Vickrey, demand for the Vickrey rises. And so must the congestion-clearing price. Everyone driving across the Vickrey during peak hours must pay more for the privilege.
The fare increase thus makes the peak-hour drivers worse off. It makes the peak-hour train riders worse off. It makes the people who switch to off-peak commutes worse off. The people who were taking the train but now stay home and mutter about tolls and train fares are worse off. Even the transportation engineers are worse off, because the National Association of Transportation Engineers revokes their membership and the National Transit Association bars them from its annual convention at the Bellagio in Las Vegas.
What about the government? The government might collect more revenue or it might collect less. That depends on a lot of things that aren’t appropriate for a family blog. If the government collects less revenue, though, then it’s done a stupid, stupid thing, because it’s created a budget deficit for itself while making its residents worse off. If it collects more, it might be better off in a money-grubbing sense, but then it could have saved itself a lot of trouble by not building the commuter line in the first place. It chose to build exactly the amount of capacity its citizens were willing to pay for. It makes no sense to deny them the benefit of it.
So by trying to make the rail line pay its own way, the SMA has managed to make everyone worse off. The market has led us astray.
This is silly, of course. The fact is, the “market” does not require that each highway, street, or tram pay for itself. They are complementary parts of a network. What matters is that we get people where they want to go in the most efficient way possible, not pit one part of the network against another. If we want revenue-neutrality, we should demand that the network pay for itself. And that’s exactly what Suburb’s did before the ill-advised fare hike, even though some parts of the network generated “excess” revenue. That was the efficient outcome. It was spoiled by a misguided, ideological insistence that both components pay for themselves.
Now one could make a couple of objections to my little hypothetical. One is that the facts are never this way: that building the alternative road would have been more cost-effective even if it had had to go through Farmington; that SMA could have built a second highway bridge because there really is no canyon like that; or that drivers would never, ever switch to a train unless chased onto one by a pack mule. But the purpose of my little hypothetical is to illustrate how subsidizing transit can be efficient in theory. If your argument is that the facts are never like this, I don't believe it. I won't subject myself to that argument in any event.
A better, fairer, criticism is that transportation engineers in the real world never have enough information to get a perfect result like this. This is very true and I would be a fool to dispute it.
But the fact is that transportation engineers have to estimate capacity, ridership, demand, cost and many other unknowns before building any transportation project. When planning a flat-tolled road, they estimate how many drivers it will attract and how much revenue it will generate. (Sometimes, they can even convince bond buyers to bet money on their estimates.) When building transit systems, planners estimate ridership and revenue. The federal government believes this is possible because it has a law that makes them do so. Planners also know something about how and when drivers will substitute rail trips for car trips; they can study real-world commuter lines that run parallel to real-world highways.
The real world is messy and no one has perfect information. Transportation planners simply have to do the best that they can. They can’t do that by shooting at the wrong goal. A rigid insistence that transit always pay for itself is the wrong goal. The facts matter.
Finally, I realize that it is rare that a public transportation network -- highways included -- pay for itself. I don't seriously contend that we should only build new roads or transit when the congestion revenue will pay for them. We've got too many free roads for that to be realistic. Still, it is useful to focus on congestion pricing because it reminds us that (1) it's wasteful to build too much capacity and (2) that we should care about the efficiency of our network rather than the efficiency of a single highway or rail line.