The murkier side of transparency

On a recent visit to Toronto, I was able to admire the city’s rather fashionable pedestrian traffic signals, which clearly show you how long you have before the lights change. Crossing the street to a flashing “3 – 2 – 1 – 0” is like experiencing your own little rocket launch.

Toronto isn’t the only city with such signals: Manhattan has them, Washington DC has had them for many years, and London is beginning to introduce them. But it’s Toronto that provides the stage for a fascinating new study in unintended consequences.

From late 2006 through 2008, countdown signals were gradually installed across the city. It is important to understand that from the point of view of safety, the order in which they were introduced was arbitrary. (Some accident black-spots are temporary, the result of bad luck. Had the signals been installed at junctions with a history of accidents, they would have looked like brilliant safety measures simply because the run of bad luck ended.)

In effect, Toronto accidentally arranged a randomised trial of the new signals. By examining the accident rate at each intersection and how it changed with the new signals, economists Sacha Kapoor and Arvind Magesan got a detailed picture of the effect of the countdown.

You might well anticipate that the countdowns would make junctions less dangerous, by telling pedestrians whether or not they have time to cross in safety. Toronto’s traffic planners certainly seemed to believe that would be the case. They were wrong. The new signals caused more accidents.

How could this be? Kapoor and Magesan suggest three explanations. One is that the signals cause pedestrians to take more risks – but it seemed that fewer pedestrians were involved in accidents after each signal was installed.

The other two explanations rely on the fact that it’s not only the pedestrians who see the countdown: drivers can too. If a signal is about to turn red for pedestrians crossing at a junction, then drivers who are trying to get across the junction in the same direction are also about to get a red light. Since there was more speeding and more rear-end collisions after the countdown signals were installed, Kapoor and Magesan reckon the natural explanation is that some drivers were accelerating into the junction to avoid being delayed, just as other drivers were slowing down.

It’s not the first time that economists have discovered that what looks like sensible transparency can have unintended consequences. Ten years ago, David Dranove, Daniel Kessler, Mark McClellan and Mark Satterthwaite looked at the impact of mandatory “report cards” in New York and Pennsylvania, which published data on the performance of individual doctors, hospitals or both.

One might imagine that this information would, at the very least, be convenient. At best it should spur physicians to improve their skills because patients would seek out the very best. But the researchers looked at the impact on cardiac surgery, and found a tragic side effect: once doctors and hospitals knew that their success rates would be published, they had a strong incentive to operate on the healthiest patients. The best hospitals had their pick of the sick and selected easy cases. Meanwhile patients with more complicated conditions were more likely to have surgery postponed. The net result: more money was spent, yet more people died of heart attacks.

Publishing clear information is often a way to make the world a better place – but not always. Sometimes it pays to be selective. Doctors could benefit from report cards, provided their patients never find out what they said. And Toronto’s countdown signals would work perfectly if only they could be hidden from drivers.

‘The Undercover Economist Strikes Back’, by Tim Harford, is published by Little, Brown