Nick Janusch looks at everything from an economist’s perspective. For more, visit his blog, “Appreciate Yourself

Bay Area commuters are experiencing new toll charges at their Bay Area bridges this month (July 2010). One of the more interesting toll changes is the congestion pricing schedule implemented at the Bay Bridge. 

The pricing schedule works out where for single drivers, the toll Monday through Friday is $6 during commute (peak) hours and $4 during the non-commute (off-peak) hours.  Carpools will now be charged $2.50.

The single driver toll is $5 all day on weekends.  The idea is reduce congestion during the peak hours by price discriminating commuters to incentivize them to decide and use alternative modes of transportation or to cross the bridge during the off-peak hours.

Yet something interesting occurred the first day of the pricing scheme, as reported by the San Jose Mercury News reports (and also reported here), moments before 10 a.m. when the toll switched from $6 to $4 drivers started to slow down or stop.  The San Jose Mercury News reported that:

Motorists slowed up considerably, even stopped, as they approached the toll gates at 9:57 a.m., three minutes before tolls were scheduled to drop $2.

“You bet I waited,” yelled out one of those drivers, a guy in a beat-up red Oldsmobile as he inched ahead after tolls fell from $6 to $4. “I’m saving a couple of bucks.”

I am curious how widespread this is during this hour and if it occurs daily.  If drivers are aware of the peak pricing schedule, I am sure no one wants to be the last car and arrive at 9:59:59 a.m. (yes, early by one second) and pay the peak price toll.*  As such, the peak pricing schedule appears to be flawed system during this time period when the price to travel across the bridge depends on a matter of minutes or seconds.  Will many drivers then react daily by doing what the driver in the Oldsmobile did by slowing up and causing congestion for all the drivers behind him?  Further,  how about when the price is scheduled to switch from $4 to $6?  Would some drivers then react by aggressively speeding to the toll both before the toll switches to the higher rate?  Seems like the benefits from the peak pricing schedule would be negated by the behavior of drivers during these periods when the toll price changes.**

Unlike peak pricing schedules, the San Jose Mercury news article also reports of a different congestion pricing system that will be implemented on Interstate 680.  It is a responsive system, and the article reports:

But South Bay commuters won’t have such an obvious strategy for timing their commute when congestion pricing comes to Interstate 680′s southbound carpool lane later this year. That’s because the form of pricing there may vary every few minutes for the 14-mile drive from Highway 84 to Calaveras Boulevard.

It could be a $3 trip at 7 a.m. for solo drivers to buy their way into the diamond lane. But another driver entering the same lane a few minutes later could pay more as traffic stalls, or less if the commute suddenly eases.

With less warning on 680 and other freeways like 580, 85 and 101, where congestion pricing will be introduced in carpool lanes over the next several years, it’ll be harder to figure out what time to best avoid higher tolls and still get a faster drive.

Thus the toll used on 680′s HOT (High Occupancy Toll) lane will be real-time pricing and will be based on current demand and not by predicted demand pricing like the Bay Bridge’s peak pricing system.  This logic follows the late Nobel Laureate William Vickrey work on responsive pricing where he argued that real-time responsive pricing is a more effective to reduce congestion by allowing for an immediate feedback loop enabling users to respond to unexpected demand and supply shocks (Vickrey, 1971).  A peak pricing system would prohibit users to adjust their behavior during unpredicted market fluctuations since the prices would be fixed and set in advance.  So real-time pricing then is preferable by avoiding the inefficiencies created by having prices generated without an immediate feedback loop.  Expectations of price changes and other information play a central role in the behavioral decisions of users.

Yet I am not sure if the randomness of a responsive price system would be acceptable to Bay Bridge commuters, but at least it would be a more effective way to manage and measure congestion.  Moreover, it would eliminate this apparent unintended consequence of drivers at the toll plaza slowing or speeding up during the moment a scheduled toll price change occurs.  Besides if the responsive price system was implemented and the price ended up always near its upper limit and never near the minimum price, then at least policy makers concerned about congestion can justify an increase in the upper limit of the responsive price system.

*Would (some) drivers then hope for more congestion before they arrive at the toll plaza when the price changes downward to assure them of the cheaper toll?

**This peak pricing issue could also be applied to the peak pricing systems currently being applied to on-street parking (parking is one of my favorite topics).  Drivers wanting to park and avoid the peak price charge would do similar behavior, as mentioned above, where they would wait either in their parked car (if they are lucky to find a spot) hoping that a meter maid does not ticket them or might drive aimlessly around the block before parking to avoid a peak pricing charge and/or being ticketed for arriving early.  If the driver decided to drive aimlessly to pass the time to avoid the peak price, then that driver is not only imposing costs on others by increasing congestion, wasting fuel, and putting pedestrians at danger, but also imposing costs on him or herself .  The driver would have rather not waste their time driving aimlessly but actually reach their destination and not wait in their car.

Vickrey, W. (1971) “Responsive Pricing of Public Utility Services.” The Bell Journal of Economics and Management Services, 2 (1), 337-346.