Thursday, January 30, 2014

How Pedestrian Countdown Signals Cause Auto Accidents

Pedestrian countdown signals at crosswalks show how much time is left before the light turns yellow, thus letting pedestrians know if they should rush to cross the street--or perhaps wait for the next light. But when these signals were introduced in Toronto, the rate of rear-end auto accidents was higher at the intersections with pedestrian signals compared to neighboring intersections. Sacha Kapoor and Arvind Magesan tell the story in "Paging Inspector Sands: The Costs of Public Information," which appears in the most recent issue of the American Economic Journal: Economic Policy (6:1, pp. 92–113).  (The title was obscure to me: I'll introduce Inspector Sands at the end.)

The story starts a few years back when the city of Toronto decided to change over its existing streetlights to a more energy-efficient variety. Then the city decided that while doing the change-over, it would also install pedestrian countdown signals at the same time. It would start in the places where it was cheapest to retrofit, and then work across the city. This history matter for the economic analysis, because the pedestrian countdown signals were installed for reasons and in an order that had nothing to do with whether an intersection was known to be unsafe or whether previous accidents had occurred. Thus, one can reasonably compare intersections with signals to nearby intersections without, and do so before and after the signals are installed.

"Our empirical analysis reveals that countdown signals resulted in about a 5 percent increase in collisions per month at the average intersection. The effect corresponds to approximately 21.5 more collisions citywide per month. The data also reveals starkly different effects for collisions involving pedestrians and those involving automobiles only. Specifically, although they reduce the number of pedestrians struck by automobiles, countdowns increase the number of collisions between automobiles. That the total number of collisions increased while collisions involving pedestrians decreased suggests that pedestrian countdown signals had a very significant effect on driver behavior. In fact, we find that collisions rose largely because of an increase in tailgating among drivers, a finding that implies drivers who know exactly when traffic lights will change behave more aggressively."
In short, the pedestrian countdown signals were good for pedestrians. But some of the drivers were watching the signals, trying to squeeze through before the light changed, and rear-ending other cars.

There's are some narrow lessons here about pedestrian countdown signals and a broader lesson about how information works. Here are two narrow lessons, which come out of a  more detailed analysis of the data: "The first is cities might benefit from installing countdowns at historically highly dangerous intersections and from not installing them at historically safe intersections. The second conclusion
is that while countdowns can improve safety in historically dangerous cities, they may be detrimental to safety in historically safe ones." Also, instead of having a pedestrian countdown signal that is visible to cars, it might make more sense to have a verbal countdown that could only be heard by pedestrians.

The broader lesson is that it's common to assume, without a lot of thought, that more information shared more broadly will make everyone better off. But the case of Toronto's countdown signals is an example of where making information available only to some (pedestrians) and not to others (drivers) is socially beneficial.

Another example involves the story of Inspector Sands in the title of the article. Kapoor and Magesan write: "Few know who Inspector Sands is, and no one has ever met him. This is for good reason. Theater companies in the United Kingdom are believed to use the code name “Inspector Sands” in order to alert ushers to pending emergencies, such as fires and bomb threats, without inciting panic among their patrons. When theater  staff learn of a fire, for example, they page Inspector Sands to the fire’s location. When ushers arrive they can put out the fire or help to evacuate the premises in a discrete and orderly manner. By ensuring the threat remains hidden from the public eye, the code name allows ushers to complete the tasks without having to deal with panicked crowds." Thus, Inspector Sands is a case, like pedestrian countdown signals, where information is revealed in a limited way to some, because revealing it to all would risk causing harm.

Full disclosure: The AEJ:EP is published by the American Economic Association, which also publishes the Journal of Economic Perspectives, where I work as Managing Editor.