Planes vs. Cars

As usual, I'm later than everyone else, so I'm just now getting around to reading Freakonomics by Steven Leavitt and Stephen Dubner. The book grew out of an article for the New York Times Magazine by journalist Dubner about economist Leavitt. Leavitt (who is here at the University of Chicago) is a rising young star in the profession, who had previously garnered considerable publicity for his work showing the real reason behind the dramatic drop in crime rates during the 1990's. It wasn't stricter enforcement, or a better economy, or innovative policing strategies; it was Roe v. Wade. Leavitt argues that the availability of abortions prevented a large number of childred from being born to mothers who didn't want them or were unable to take care of them, and that these at-risk kids are exactly the people likely to commit crimes as teenagers. The theme of the book, if there is one, is the attempt to tease out the counterintuitive structures of incentives and pressures that lay behind a wide variety of patterns in our daily lives. And there is, of course, a blog. But I was happy to see a mention, if only very briefly in passing, of an issue I've long wondered about: the relative safety of air travel vs. automobiles. It's a well-worn piece of wisdom that, despite the potential for spectacular accidents, air travel is actually safer than car travel. I've never been quite sure how seriously to believe this claim, since it was never spelled out how "safer" was being defined. The facts are the following: many more people die in auto accidents each year in the United States (about 40,000) than in airplane crashes (less than 1,000). But that certainly doesn't answer the question by itself. People spend a lot less time in airplanes than in cars, on average. In fact, it turns out that your risk of death per hour is about the same in a car as in a plane. So, what's the answer? Does that mean that air travel and auto travel are about equally dangerous? No, of course not. Nobody plans their trips by saying "I would like to spend x hours traveling today." (At least, putting aside the infrequent trips we take purely for the pleasure of being in a moving vehicle.) Rather, there is some place we want to go -- i.e., some distance that must be traveled. So the correct figure of merit is the risk of dying per given trip that you'd like to take, not per hour. Airplanes, of course, travel much more rapidly than cars do, so we spend less time in the plane than in a car for a given journey. In this case, the conventional wisdom is true -- air travel really is safer. Leavitt and Dubner don't actually mention that point, but it's a crucial feature of risk assessment, or for that matter all sorts of planning. When you are trying to weigh the merits of different ways of allocating resources (money, safety, or whatever), we always have to ask "per what?" That is to say, it's the rate of resource expenditure that matters, not the instantaneous value. If I want to spend $500 on a spiffy new espresso machine (and I do), I shouldn't simply say "Wow, that's a lot of money" -- I should carefully compare the cost of the machine plus coffee beans to whatever I'd be paying in small installments at my local Starbucks over the life of the machine. It's the cost per cup of coffee that matters, not the one-shot price. Or am I just trying to talk myself into something?