Sunday, June 13, 2010

Don't you hate traffic? Part 2

I have a few different loose ends to tie up from the first installment of Don't you hate traffic? .

First, since I've gotten thousands of emails asking about the title of this post, I'll address that first. It's a reference to The Simpsons (clip here) from back when the show was funny. (Yeah, I'm dating myself here.)

Second, how do you deal with privacy issues involved with tolls? If we use EZ-Pass, do we really want Big Brother knowing where you've been driving? I think there's a good way to get around this, which some politicians might not like: Sell EZ-Pass tags the same way we sell pre-paid disposable cell phones. If you buy an "anonymous" pre-paid unit with $100 on it and throw it away when you're done, then you haven't really given up much personal information.

Now finally, let's throw out this assumption that everyone is the same. Different people dislike traffic to varying degrees. Traffic makes me angry and agitated, but other people are in no rush to get anywhere.

So let's say we have a heterogeneous population and roads that aren't sufficient to accommodate everybody. One day I look outside and see 1000 cars on the road, which translates into 2000 units of traffic, using my super-secret traffic grading system.

Presumably, there are some people on the road who really value the road and/or don't particularly mind traffic. Perhaps they're willing to deal with up to 5000 units of traffic at this time. Every single one of them is on the road right now, because after all, there are only 2000 units of traffic out there.

There are also some people who are willing to deal with 3500 units of traffic, some people who can tolerate 2200 units of traffic, etc. and they're all driving now too.

But then there's a guy (call him Marginal Matt) who will only tolerate 2001 units of traffic. He's still driving (since there's only 2000 units) of traffic, but he's getting very little advantage out of having the roads available to him.

Here's the problem: If he would get off the road, there would be a little less traffic for the other 999 people. By deciding to drive, Marginal Matt is creating 2 units of traffic that all of those 999 other drivers has to suffer through and he's only getting 1 unit of advantage to himself.

In other words, he's creating thousands of units of harm and only getting one unit of benefit. Clearly, we need a pricing scheme that gets him off the road.

The optimal toll would balance these two effects. Switching to dollars, if the optimal toll was $10/mile, then there would be much fewer people on the road -- say 600 people instead of 1000. And if the toll is set correctly, then adding that 601st person would create exactly as much economic benefit to him and it would collectively take away from all the existing drivers.

One pleasant aspect of this is that only the people who value the roads the most will use it, and they'll be the ones paying for it. The people who walk to work and then pay an insane city income tax to help pay for roads are getting a raw deal under the current system. Their taxes could be reduced and we could have the drivers pay for the road. The result would be would be less traffic, more business activity (thanks to the lower taxes) and the roads would provide much more value to the residents.

Now I don't know what the optimal toll is, but I know that people do study this sort of thing and have attempted to put a price on traffic.

This system ideally would also end the "need" to subsidize mass transit. It drives me crazy to hear people say that we need to subsidize the subway in NYC since the alternative would be even more nightmarish traffic. When the idea of congestion pricing on roads is suggested as an alternative, people complain that that would increase the price of doing business, which would "get passed on to consumers".

While I don't have the time to delve into all the fallacies embedded in this nonsense, suffice it to say that if we priced roads correctly, a lot of people would take mass transit because it would be much, much cheaper than driving, and in the case of buses, it would be pretty fast (since traffic would be less severe).

In any case, how is it fair that my friend who walks to work has to subsidize the subway? How 'bout charging me for the subway ride that I take to work and charge Marginal Matt a buttload for the amount of congestion he's causing during rush hour?


ESM said...

Good ideas, but I don't quite understand the application of your super-secret traffic grading system.

Just because Marginal Matt will stop driving if traffic volume is equal to or greater than 2001 units doesn't mean that a traffic volume of 2000 units has 1 unit of value to him. His driving utility function could be (and probably is) highly non-linear with traffic volume, possibly even with a step at 2000 units.

The non-linearity of driving utility vs traffic volume presents some problems for toll pricing, and means the pricing system needs to be more sophisticated and condition-dependent.

Fortunately, we already have an analog which we can use as a basis for toll pricing: the electricity market and its very intelligent use of local marginal prices (LMPs) based on computer modeling of the entire transmission grid.

Subject matter for another post I think.

For electricity,

Coupon_Clipper said...

Yeah, you're right. To really do things correctly, we'd have to do everything in terms of subjective value in dollars terms. (Utils wouldn't even be the right approach if we're trying to achieve Marshall Efficiency.)

I think the non-linearity actually makes the tolls an even better idea. (I'm not saying you disagreed; I just think it's an interesting point.) If having one more driver on the road makes traffic reach some tipping point, then it's really really valuable to keep that next guy away. Everyone else gets tremendous value from grounding him, and they'll pay up for that right.

Coupon_Clipper said...

I was thinking.... if I weren't so long-winded I could've just said this in place of the entire post: If roads are a scarce resource, then there's someone on the road who's just barely getting any amount of value out of driving. He's creating traffic for everyone else, so it's almost certainly a net loss.

So you want him off the road. And you probably want a lot more people off the road (though I don't know the exact number). A pricing system achieves this. And that's all there is to it.

ESM said...

Something I haven't seen mentioned before is the importance of driver education in improving traffic. I think 95% of drivers are blissfully unaware that even a brief episode of erratic driving (e.g. slamming on the brakes because you almost went past your exit) can create an instability which propagates and grows into a major slowdown event.

I think if drivers were more aware of the non-linear impact of their driving decisions on traffic behind them, they would be more careful. I certainly try to be.

A way to enforce this is to have periodic driver's education classes which focus on what causes traffic jams. Drivers who show up for the classes would get a discount on their E-Z Pass tolls. I think the effect would be significant. Even getting people to avoid rubbernecking would be a huge help.