Smeed’s Law

7 September 2006

Smeed’s Law

By Gwynne Dyer

I was in a taxi in Beijing this morning, in heavy traffic moving very slowly, when a truck tried to change lanes and push in front of us. The taxi driver was having none of that, so he nosed forward to block the truck. The truck driver held his course, the taxi pushed forward again, and after about fifteen seconds of that came the inevitable crunch. Small accident, nobody hurt.

You see this sort of thing almost every day here, and you sometimes wonder why Chinese drivers are such idiots. But they aren’t. They’re just first-generation drivers.

China still has less than one car for every fifty people, but even ten years ago over 100,000 people were dying on its roads annually. The death toll is probably much greater now, because car ownership has grown at least fivefold since then. But I’ll bet that it is already falling in terms of deaths per vehicle-mile (vehicle-kilometre) driven. There is a national learning curve in driving, and China is already climbing it. So is the rest of the developing world.

Around the world, about 1.2 million people are killed in road accidents each year. An astounding 85 percent of those deaths happen in developing countries, although they own less than a fifth of the world’s vehicle fleet. There’s no getting round it: they are very, very bad drivers in China, India, Africa and the Middle East. (And they are almost as bad in South-East Asia, Eastern Europe and Latin America.)

Take Liberia, for example. If the Liberian death rate per million miles (kilometres) driven were transposed to the United States, six million Americans per year would be killed on the roads. Actual American road deaths are about 40,000 a year, so it is 150 times more dangerous to drive in Liberia than in the United States. You can’t blame all that on poor brakes and bad roads. But why is it like that?

Back in 1949, R.J. Smeed, Professor of Traffic Studies at University College London, proposed a statistical “law” — more a rule of thumb really — which was, to say the least, counter-intuitive. He said that a growing number of cars on the road leads to a DECREASE in the number of accidents per vehicle. A growing car population means a big, persistent annual fall in the death rate per million miles (kilometres) driven.

Common sense tells us the opposite. It says that, other things being equal, the number of single-vehicle accidents (driving into trees, running over pedestrians) ought to increase more or less in direct proportion to the number of cars on the road. Two-car crashes and multi-car pile-ups ought to increase as the SQUARE of the number of vehicles. But it doesn’t work like that.

The amount of road traffic in the United States has grown fourteen-fold since 1925. If the number of American deaths per million miles (kilometres) driven had stayed steady at the 1925 rate, there would now be 300,000 deaths per years on American roads, not 40,000. Americans have become much better drivers — and everybody else will, too.

Smeed offered no explanation for this phenomenon, but I think that there is a collective learning process as more and more people become experienced drivers, and particularly as the generations turn over and children grow up in families that already own cars.

In thirty years, the mass stupidity that was Mexico City’s road scene in the 1970s — make six lanes where there are only three, block the intersections, and blow your horn incessantly — has morphed into the relatively disciplined, relatively silent Mexico City traffic of today, which flows more smoothly now even with three or four times as many cars on the road.

Between 1925 and 1984, the US road death rate per million miles (kilometres) driven fell steadily by 3.3 percent per year. In Britain, between 1949 and 1974, it fell by 4.7 percent a year. You can’t just attribute it to safer cars, because modern cars in the hands of first-generation, Third-World drivers still achieve kill rates as high as those of 1920s Americans in their Model T Fords. Better engineered roads may make a bit of difference, but on the other hand there are far more cars on the road.

Eventually, you hit diminishing returns: in the last ten years US road deaths per million miles (kilometres) driven have fallen at less than two percent a year. When almost everybody is a third-generation driver, what’s left to learn? But Smeed’s Law still holds, because the number of miles (kilometres) driven per year in the United States isn’t growing fast any more either.

So this year’s 1.2 million traffic deaths will not soar to five or ten million when all the people in the developing countries get cars too. It will rise for a while, due to the huge surge of new drivers, and then it will fall back as they gain experience, perhaps even below the current figure.

Which leaves only two problems. The long-term one is that the world may go into climate meltdown because of those two or three billion cars. (The current global car population is about 500 million.) The short-term one is that Beijing, where an extra thousand cars are put on the road every day, will achieve total gridlock just in time for the 2008 Olympics.


To shorten to 725 words, omit paragraphs 7 and 10. (“Common…that”; and “In thirty…road”)