What Bothers People About SuperFreakonomics?

We never know what parts of our books will agitate people enough that they will send us an email. In Freakonomics, the passages that inspired the most emails were the discussions of swimming pools vs. guns and real estate agents not always acting on behalf of their clients.

In SuperFreakonomics, far and away the most common subject of emails is drunk walking vs. drunk driving. In particular, every few days someone writes us to tell us that our analysis is wrong because we are comparing the rate of death per mile driven drunk versus the rate of death per mile walked drunk. Sure, they say, drunk walkers get killed more per mile. But since cars travel much faster, per hour, it is safer to drive drunk than to walk drunk.

It is true that if someone held a gun to your head and said, “If you don’t walk drunk for an hour or drive drunk for an hour, I will shoot you. You choose whether you would walk or drive,” then you might very well want to spend your hour walking drunk. However, in real life, that is virtually never the dilemma you face. Rather, you are drunk in one place and you want to get to another place. The distance you need to cover is what is constant, not the time you will spend traveling.

Thus the per-mile comparison we make is the most sensible one.

The other thing about that passage that makes people angry is that they interpret our arguments as condoning drunk driving, despite the fact that we cite my own research that shows that drunk drivers are 13 times as likely to cause a fatal crash. We end by telling people to take a cab.

When we wrote that people should take a cab, however, we never actually did the calculation. Is it really true that taking a cab is the right thing to do?

According to our estimates, there are 21 billion miles driven drunk each year, resulting in 13,000 fatalities. That works out to be about one fatality for every 1.6 million miles driven drunk.

Economists typically use a value per statistical life of $6 million. So, for instance, when trying to decide whether the benefits of a government program outweigh the costs, the benefit per life saved is calculated as being worth $6 million. If one person dies for every 1.6 million miles driven drunk, and the value of a life is $6 million, then the cost in extra deaths per mile driven drunk works out to be about $3.75.

That number is an average. Obviously, it will depend on how drunk you are, and it ignores other risks associated with driving drunk like getting arrested.

How much does it cost to take a cab? According to this web site, a three-mile cab ride will cost you about $8 plus tip in most major cities. After a tip, that is about $3 per mile — not too different than the implied cost per mile of driving drunk.

Obviously, I’ve left out all sorts of other costs and benefits in this simple analysis that could tip the balance one way or another, but I suspect most people will be surprised to see how close it is to a toss-up.

Ramsey H

The real problem with your analysis is that I don't care if some drunk idiot gets himself killed while walking. I *do* care if a drunk driver kills someone *else*.


I don't know about you, but my choice of where to go is normally constrained by my means of transportation. It's not a question of walking out of a bar and randomly being assigned to the Car or No Car group. When I have to walk home, I tend to stay closer to home; when someone is DDing, we feel free to choose places that are further afield.


Another consideration here is one of internalization -- the cost in drunk driving is more likely to be an externality, while the cost of a cab is internalized by the drunk. Cab use thus should make the intoxication-transportation market more efficient.


The dollar analysis is incorrect because you are comparing the government cost of a life vs. my out of pocket cost for a cab. Would you sacrifice your life or your spouse's or children's life for 6 million dollars? Probably not.

The government and I don't spend money on the same scale, so to compare the two is not valid.


Depends how drunk, doesn't it? What is the threshold you use for drunkenness? BAC of 0.08?

I bet the death rate rises more than linearly as BAC rises, so maybe driving at 0.08 is actually safer than cabbing it.


Considering that I typically don't walk distances over a mile (especially drunk), I don't see driving and walking as substitutes. But If I were to walk the distances that I might otherwise drive, I would become less drunk over time, making the long walk safer as the walk gets longer. And I'm still struggling to see why this is a topic worth discussing. I guess I should just move on.


While the distance is constant in either case, the time spent traveling should matter since the effect of alcohol diminishes over time.

Also as another commenter noticed...when driving it isn't always the drunk who pays the cost of an accident. Plenty of otherwise uninvolved people are killed by drunk drivers every year, and I'd be willing to bet very few are killed by drunk walkers.

Eric Valpey

Except the choice is rarely so simple.

Oftentimes, the decision of how to travel home informs the decision of where to go in the first place.

If one expects to walk home, one will only really be considering a local bar/pub within walking distance. Making arrangements for a designated driver or planning on public transportation alters the landscape of evening destinations.

"Walking home" may actually be some combination of walking and public transit (walk to a transit station - bus/rail to local station - walk home). Thus the number of miles actually walked is far fewer than the number of miles driven by car or by cab.

If we are talking about someone who is surprised to find themselves drunk at a bar who has traveled there by car and now must choose how to get home the options might include a nearby destination other than home like a friend's house or a late night diner where one might conceivably sober up; another example where "walking" might involve alternate paths.



"The real problem with your analysis is that I don't care if some drunk idiot gets himself killed while walking. I *do* care if a drunk driver kills someone *else*."

it's always someone *else* if you aren't the one dying, silly goose. adjust your world view accordingly.


I thought that the big assumption was that the percentage of miles walked drunk is equal to the percentage of miles driven drunk. I wouldn't be surprised if they were vastly different. (Although the 8X or 5X result means that there is a lot of leeway for that result to be wrong and still make the conclusion valid.)

Also, as other commenters noted, it depends what your objective is. If my objective is to avoid being killed, then the 8X number is most relevant. If the objective is that of a social planner that wants to reduce the number of people killed, then the 5X number is most relevent. However, if we want to make the decision from the viewpoint of a libertarian social planner, then there is no cost to walking drunk since you will only kill yourself, but driving drunk imposes a social cost by killing others with some probability.


It's an interesting analysis, that all hinges on the assumed cost of a life, which is a very tricky thing to estimate, and not one that I would pick the government to do the job of figuring out.

Obviously different people are going to come up with different estimates, if it's a loved one, or yourself, you'll probably come up with a much higher figure. And I'd guess that for a random stranger most people's actions would point towards a much lower figure.

Consider how much you would have to donate to charity, say, feeding the hungry to be likely to save one life. I'm sure that if you donated $1M dollars that would most likely save a number of lives, even modest amounts like $1K might be enough to save a few lives. If people really valued a random stranger's life at $6M then charity would seem to be quite a bargain.

But, for lack of a better way of estimating I'll accept that $6M is somewhere within an order of magnitude or so. In which case, the drunk should take the cab, don't stick us with your externalities.


Matty J

Phil (#4), the $6 million figure is not "the government cost of a life." It is a figure that economists use to compare and contrast potential benefits of social programs. Asserting that the figure is somehow derived from a price tag that our government puts on a human life is a misinterpretation of the article. The number is simply used fairly evenly across the board so different groups of people can plug a constant into the equations. It is a number, an average; it is not intended to represent the dollar amount at which any given person would sacrifice his/her spouse or children. Relax a little and enjoy the analytical process my man.

Mike B

I don't follow how having the distance being invariant on a per trip basis makes it the logical basis for comparison. The very fact that the per-mile death rate of WWI is greater than the per-mile death rate of DWI seems surprising indicates that you should probably examine things from another angle to determine exactly WHY those results are surprising.

It's the exact same situation as the comparisons between flying in a plane and driving in a car. On a per mile basis planes are by far the safest way to travel, yet people are far more terrified of flying than driving. Social scientists love to dive into this apparent irrational behavior and most of the Freakomonics brand is built on exposing how human intuition is often flawed, but I believe that your own intuition on the flaws of human intuition are clouding your judgment.

If one actually looks at the per-hour statistics between aircraft and ground vehicles, they are actually much more similar than one would expect just as if one compared the per-hour rates of WWI and DWI. Instead of just making an ASSERTION that distance is the proper metric, you should try to STUDY how risk is increased by time and distance and then see which dominates. Using aircraft as an example I believe that the answer is already hinted at by regulatory processes that measure almost everything about aircraft safety in terms of hours, not distance.

Of course I am sure that properly accounting risk has both fixed, time variable and distance variable components, just like a taxi meter (wow, I tied it in). Instead of phoning in another classic "Freakonomic" finding, you should have done the legwork to properly model how one's risk meter increments as they move about in the world. I personally think that drunk transport follows a similar model to walking out in a lightning storm. The chance of getting zapped is completely independent on how far I am walking in the storm, but instead how long I am walking in the storm.



The most important factor ignored in this column is the fact that with the first drink a person's judgement is impaired. So the decision to drive or walk must be made while sober, accompanied by support from individuals or a pre-arranged taxi pick-up. After that first drink, all bets are off.


I think this is one of the things I LOVE about your books. We are presented with a new way of looking at things that makes us uncomfortable and gets us to think (hopefully!) about our preconceptions and actions in a different way.

Thank you and I can't wait for the next book.


the issue isn't the time length of the drunk trip. It is that most drunk walks are under a mile. While most drunk drives are several miles. My hunch is that most people who drive to the bar live 5+ miles away and wouldn't consider walking. So, the question at the end of a night at the bar should be how does the risk compare between walking 1/2 a mile and driving 5 miles.

How does the cost/benefit of cabbing 5 miles compare with drunk driving 5 miles?

How does the cost/benefit of cabbing 1/2 miles compare with drunk walking 1/2 miles?


It would be interesting to see how drunk biking comes into all of this. I can't believe it that I have intelligent friends who think it's safer to ride a bicycle drunk than it is to drive.

Steve Austin

Pfft . . . most lives aren't worth $6 million.


You ignore two important facts about drunk driving: it has legal costs, and its fatality costs are partly external.

First, in addition to worrying about killing someone, I might be worried about getting arrested. I would think this is actually rather more likely to happen. While it won't cost six million, it will cost a fair amount, and the likelihood is probably higher enough that this should be factored in to the calculation.

Conversely, the value of a life is viewed from a unweighted social position. This relates to whether society should encourage people to take cabs vs. drive drunk; it's not terribly relevant to my own personal calculation, unless I practice totally unweighted utilitarianism, which no one does.


I reckon it's articles like these that make the economics profession unpopular among the general public. Equating a value to a life, even a "statistical" one, in order to calculate whether we should take a taxi or risk killing ourselves or someone else.

You also assume risk neutrality, whereas I would suppose most people are risk averse when it comes to their own life. Inebriated, however, they might even become risk loving - or at least risk negligent (did I just invent a new term for the economics jargon?).

I usually take the bus after drinking, by the way, but I suppose that's a viable option only in some European cities.