Congestion Pricing

Road tolls that vary directly with congestion have reduced peak-hour traffic volume in Singapore and Stockholm by forcing commuters to internalize congestion cost.

It is commonly known that commuting time during peak hours takes much longer than during off-peak hours. If time is money, varying commuting time serves as some kind of congestion pricing. But because no out-of-pocket cash is involved, this de facto congestion pricing cannot discourage people who are time richer and money poorer from joining the commuting crowd. Nor does it allow commuters who are money richer but time poorer to price out the time richer but money poorer.

But can a cash congestion-toll that varies directly with traffic volume actually reduce peak-hour traffic congestion? Absolutely. Peak-hour traffic volume into the restricted zone in Singapore declined by more than 30% since the electronic road pricing system was installed in the late 1990's (Chin). And in the recent 7-month trial of congestion pricing in Stockholm, traffic volume in the restricted zone declined by 22% during peak hours (WSJ 8/29/2006).

In the Stockholm experiment, the toll charge ranges from $2.76 per pass during peak hours to free after 6:30 pm. Such a sliding toll-charge scale was enough to reduce the commuting time during peak hours from 3 times to 2 times the time of an off-peak ride.

An additional bonus is the reduction (14%) of exhaust emissions in the inner city as ridership for all forms of public transportation jumped 6%. Even the commuting time of bus riders was significantly reduced by the lower traffic congestion.

We can look at public roads without congestion tolls as a commons resource with limited carrying capacity. Free access encourages higher traffic volume than is optimal for peak hours. The additional commuting time and exhaust emissions are external costs that have not been internalized. Congestion pricing effectively forces commuters to internalize these external costs and price out commuters with less urgent needs during peak hours.