What Peltzman Found

by Russ Roberts on November 15, 2006

in Regulation, Risk and Safety

There’s some confusion about what Sam Peltzman said/found in his work on automobile safety discussed in the comments to this post.

Peltzman argued that mandatory safety devices such as seat belts reduced the probability of harm to the driver if the car crashed. That in turn would encourage people to drive more recklessly. The effect on the number of deaths was an empirical question. Which effect would be larger—the harm from the increase in the number of accidents or the reduction in harm when an accident occurred?

Holding other factors constant that might change the number of accidents (and this is never easy but he did the best he could with the data at hand), Sam found that mandatory seat belts did indeed cause more accidents. But this effect was roughly the same as the effect in the opposite direction, that accidents were less harmful. So the net number of fatalities of drivers was unaffected by the law. Sam found some evidence that the effect of the law might be to reduce driver fatalities. Unfortunately, because drivers were more reckless, there were more accidents involving pedestrians and cyclists. So their death rate due to cars increased. Total deaths were unchanged.

You may find this intuitive or unintuitive. You can challenge the empirical analysis. But at the time, the findings were shocking (and distasteful) to the safety community and to academics who had confidently predicted that safety regulations would produce dramatic significant reductions in driver deaths. These engineering predictions ignored any possible behavioral response by drivers. Whether Sam measured these behavioral responses accurately is always an open question. But his insight that people respond to incentives is always a very good place to start.

The second issue that came up in our podcast was airbags. In general, airbags have the same effect as seatbelts—they reduce the likelihood of injury and death from an accident which in turn increases reckless driving and the number of accidents. But in the early days of airbags, they could kill you, especially if you were a small person or a child. That led to some tragic deaths of kids in the passenger seat and small women in the driver seat. That in turn led to the government changing the requirements on the speed and force of the airbag deployment. Sam pointed out in the podcast that this is one of the problems that can arise when a national standard is mandated. Unlike the case where a car company might try something like this and learn from its mistake or a state might impose such a regulation, here everyone was forced to provide airbags, and the deaths from a badly designed regulation were magnified.

I encourage readers to read Sam’s original article. Unfortunately, the only online copy I can find is at JStor.

Be Sociable, Share!



28 comments    Share Share    Print    Email


Robert Coté November 15, 2006 at 4:45 pm

Didn't control for traffic density or extent. Worthless.

Safety devices save lives in oh so many ways. It is weird to see even questioning of the results. Many more people travelling many more miles at slightly higher average speeds and net fatalities drop a lot nevermind per capita or per million passenger miles. Where's the issue?

Bruce Hall November 15, 2006 at 7:09 pm

My argument with the idea of letting one or several states set regulations that could be used nationally is they are often motivated by something other than good science, economics, whatever….

See this: http://www.bdlaw.com/news-84.html

I also discussed the way people make economic decisions with my son who is a safety engineer consultant with a firm in Ann Arbor, MI. His take was that people generally miscalculate risks and, consequently, don't necessarily make the best decisions when applying money against those risks. Seatbelts and airbags are just two examples of those economic miscalculations that required legislation to overcome… and they do save lives. Peltzman is wrong despite the fact that there were injuries associated with the first generation of airbags.

Jacko November 15, 2006 at 11:36 pm

Let's be totally politically incorrect:

I believe that all this evidence is superceded by the data that shows that neither seatbelts nor airbags have brought passenger safety back to levels when we drove larger cars.

If you want safety, stop penalizing and regulating the size of vehicles, and rezone towns and cities so they can walk a couple blocks to small corner bakeries/butcher shops/etc.

That will decrease the number of accidents and decrease total fuel consumed. And it will further decrease crime, because we don't steal stuff from people we know.

Tristan November 16, 2006 at 4:54 am

That's completely intuitive to me. Especially in the light of research I came across in my teens which said that people appear to have an inbuilt risk level which they tend to hit, so safety feature make you take risks in other ways.

A similar thing is traffic management, recently some traffic lights near me broke, and the traffic was never worse than it is with them working and was often better, makes sense to me (but not to my economist fiancee). It is purely anecdotal of course.

Bruce Hall: Government never takes the best decision on economic or safety grounds – they take decisions on political grounds (and they're people, they're crap at assessing risk).
You just need to read Bruce Schneier to realise that people do not understand risk and security…

Jacko: Perhaps abandoning zoning would be best (personally I'd have a Land Value Tax which will encourage most efficient use of the land).
Planning regulations seem to cause more problems than they solve (especially in the UK where they're largely responsible for the lack of housing and the ever rising prices).

joan November 16, 2006 at 7:01 am

Last summer I went on a road trip in the UK and was astounded by beauty of the country side. But all things have a price, and I would like to know more about the housing shortage, is it just London or is it country wide. How much is the land use regulation was due to fear of food shortages after WWII. In the US people spend about 30% of their income on housing, adjusting the amount of space to the local price for housing. Is this also true in England or do the spend more. From what I saw people seem to live in smaller houses than the US. Any additional details would be of interest.

spencer November 16, 2006 at 8:26 am

I would love to read the article but to not have access to JSTOR.

My big question is the data age adjusted.

In the 1960s — the period in question –
because of the babyboomers moving into the driving age we saw an extremely large and unusual increase in young drivers (under 25)
both in terms of absolute numbers and
share of miles driven. Baby boomers started in 1946 so 16 years later in 1962 they started driving and they did not reach 25 till 1970. Young drivers accidentt rate is several times that of older drivers. So just for this demographic factor we would have seen an increase in the other accidents with motocycles, pedestrians and cycliest. But in the article I read Peltzman claims these other accidents confirms his thesis that it is because of reckless driving.

Unless he "age adjusted" the accident data it seems obvious to me at least is that it was the jump in young drivers, not the Peltzman effect that caused the other accidents.

tarran November 16, 2006 at 8:28 am

OH, one little anecdote to add to the mix:

My first car was a Corolla built in 1991 without an airbag. Consequently, due to government regulations, they had a shoulderbelt attached to the door, which would automatically retract into the restraining position when the door was closed.

However, the lap belt still had to be manually buckled.

I have never met anyone who liked the system over the traditionall manually buckled shoulder belt.

The interesting thing to me was the fact that if one was in an accident without the lap belt, they stood a significant risk of being seriously injured by the shoulder belt, which tended to garote or decapitate people if it caught on the neck, although in one case it eviscerated some tall teenager.

So, you has a system that did not improve safety (the same manual action was required as for the system it replaces), was more expensive to produce, and more expensive to maintain.

Despite universal appreciation of the idiocy of the idea, for several years, every car without an airbag that was sold in the U.S. had one of these things, since a manufacturer who tried to sell one without these things would face kidnapping and confiscation and destruction of their stock. Understandably, most manufacturers chose to put them in, to the detriment of their business and their customers.

spencer November 16, 2006 at 8:44 am

Tarran — hate to spoil your party but there were no governmnet requirements that required the type of seatbelt attached to the door that you described.

that type of seatbelt was completely the idea of the private sector and was marketed as a way to save lives because people would not have to fasten the belts– it would be automatic.

spencer November 16, 2006 at 9:42 am

I read his 2004 speech regulation and the
natural progression of Opulance.

I have no problem accepting the concept that you have offsetting behavior, what I have trouble accepting is the magnitude of his findings on seat belts and airbags.

Moreover, I still do not see why the assumption that airbags and/or seatbelts would reduce the accident rate is true.
Tell me why.

Steve November 16, 2006 at 11:15 am


I am almost 80% certain that you are wrong in your contradiction of Tarran's claim regarding the retracting seat belt. As I recall, this was a feature of cars for a few years after a regulation took effect requiring "passive safety restraints" but before the airbag regulation became law. In addition to my best recollection, I have a relative that was injured by one of these "passive" seat belts and lost in court on the grounds of the "regulatory requirement"

Does anyone have definitive data on this?


Will November 16, 2006 at 1:37 pm

Auto manufacturers had to have either a driver side air bag or the automatically retracting shoulder belt, or some similar measure. So, if the vehicle was built without a driver side airbag, it had to have the shoulder belt rigamarole. It was a U.S. government rule. I don't know if the rule is still on the books or just obviated since driver side airbags are the norm now. I remembers GM's way out of this was to integrate the seat belt completely into the door. Of course, it was up to you to buckle it or not. Chrysler's way out was to make driver side airbags standard, and score marketing points along the way. Reminds me of the old seat belt interlock system of the 1970s.

triticale November 16, 2006 at 7:52 pm

When one particular Ferrari model was imported into the US in the early '90s, it had to be retrofitted with a three point lap and shoulder belt, because the factory supplied five point harness did not meet Federal regulations.

John Thacker November 17, 2006 at 2:19 pm

Many more people travelling many more miles at slightly higher average speeds and net fatalities drop a lot nevermind per capita or per million passenger miles. Where's the issue?

That it makes things more dangerous for pedestrians, bikers, and motorcyclists, for one thing. The other issue is that, no, net fatalities did not drop a lot.

Last summer I went on a road trip in the UK and was astounded by beauty of the country side.

The US, of course, has some astoundingly beautiful countryside as well. One difference is that the UK has legislated "greenbelts" surrounding urban areas that keep unspoilt countryside very near the urban areas. In the US, by contrast, the population density is much lower, but to get to unspoiled countryside it's much more common to drive through suburbs and watch the population density slowly decrease, rather than hit a sudden wall of legislated countryside.

The downside, of course, is that the legislated open space does drive up housing costs.

rmark November 17, 2006 at 4:30 pm

The safest car ever built should be the circa 1960 Corvair – a spear-like steering column unprotected by a front mounted engine, combined with a tendency for the rear end to swing around in a skid – remember to drive defensively.

Previous post:

Next post: