Peltzman on Regulation

by Russ Roberts on November 13, 2006

in Podcast, Regulation, Risk and Safety

Here’s the latest episode of EconTalk: an interview with Sam Peltzman. We discuss automobile safety and his startling finding that making cars safer can lead to more death. We discuss how the requirement that drug makers prove drug efficacy kills people. And we talk about the political economy of regulation—why bad laws persist. Two parts of the interview I liked:

• Sam’s point that a national safety requirement like air bags is particularly harmful when it turns out to produce unintended consequences (killing women and children, initially, because of the power of the airbags). If one state or one manufacturer had made air bags too powerful, we’d have learned from it and fixed it. But more people die when it’s nationally mandated.

• Sam’s insight that the mobilization of the AIDS lobby to get the FDA to give people who are dying access to life-saving drugs ahead of normal approval times proves how dysfunctional the current system is—you have to have an enormous political effort in the midst of a horrific tragedy to get redress. The rest of us just have to wait.

Here’s an earlier post on Sam’s overview of the relationship between opulence and regulation.

Be Sociable, Share!



40 comments    Share Share    Print    Email


Bruce Hall November 13, 2006 at 10:00 am

If there were 50 separate, government regulations for one aspect of any product, it would be safer… because it would be non-existent!

Give us a break.

anon November 13, 2006 at 11:47 am

I love the well-SOMEBODY-has-to-regulate-it tone of that comment.

spencer November 13, 2006 at 1:30 pm

How can you look at the data on the drop in deaths and injuries from auto accidents over the last quarter century and give any credence to such absolute stupidity that he argues.

Show me a single example of auto safety regulation that generated an increase in the death and/or injury rate.

anon November 13, 2006 at 2:35 pm

Even if airbags result in a net decrease in auto fatalities, it's indisputable that women and children who otherwise would not have died have been killed by them. It also seems beyond dispute that nationally mandating airbags resulted in more fatalities than would have occurred if development of the technology had proceeded through smaller-scale experimentation at the state-wide or manufacturer-specific level.

Don't you guys find the fact that government intervention resulted in the deaths of women and children who otherwise would have lived even the least bit troubling?

Bruce Hall November 13, 2006 at 2:50 pm


It is certainly likely that some women and children were killed/injured by airbags that were deployed with too much force.

But let's take Peltzman's conjecture that it would have been better to use a smaller sample or one manufacturer's data to then develop a safer introduction to airbags (or other consumer products where safety is an issue).

First problem: which state is going to introduce the regulations? California, Maine, Nebraska? What if several states introduce different regulations that then get challenged in court for years, thereby delaying introduction of safety features and resulting in significantly more deaths overall?

Second problem: Which manufacturer is going to introduce their unique system that will subject them to endless litigation 5 or 10 years down the road (no pun)?

My oldest son is a safety engineer who works with a national association. They wrestle with design and specifications endlessly. It is not as simple as an economist's pronouncement. Technologies change and design capabilities change accordingly. If every improvement were dependent on perfection… nothing would change.

You must weigh the overall benefit of a change versus some unforeseen cost or damage. If a negative occurs, you fix it. If someone has a better idea, it will be copied.

spencer November 13, 2006 at 3:30 pm

anon — show me the source of the data that supports your assertion that seat belts and air bags has caused the death of women and children who would have lived if seat belts and air bags were not used.

You are assuming something that is not in evidence and claiming that it is a fact.

It is not indisputable.

I dispute it. Now show me I am wrong.

Russ Roberts November 13, 2006 at 4:04 pm


You might want to look at his paper on automobile safety before you assume he's wrong.

It's true that the last quarter century has seen a steady decline in deaths, controlling for miles driven.

The quarter century before that also showed a similar decline.

As a starting point, there is no dramatic effect on safety from safety regulation in either direction. That doesn't mean that there isn't an effect in either direction, just that it isn't very dramatic or if it is, it is exactly canceled out by some other effect.

The same is true, by the way, with safety on the job. Accidents on the job have fallen steadily since OSHA was put in place. They fell steadily before OSHA was in place.

What Peltzman finds in the case of automobile safety is that seatbelts made people drive more recklessly. So the probability of an accident was higher. But the probability of being killed in an accident was lower, because of the seatbelts. He found that these effects canceled each other out. So the effect on fatalities for drivers was roughly zero. Unfortunately, pedestrians and cyclists were more likely to be killed.

spencer November 13, 2006 at 4:23 pm

I still doubt his results.
It does not meet the common sense test.

When I look at the national data I find that the number of pedestrian and motorcycle
death and/or injuries has fallen at rates some 3 to 4 times that of auto occupants over the last 25 years — 9% for autos occupants vs, 29% for motorcycles, and 40% for pedestrians.

Where can I find his data?

Moreover, the government has been issuing safety requirements on autos essentially since day one. so the fact that there was no break in the data when seat belts were mandated does not demonstrate anything in and of itself.

spencer November 13, 2006 at 4:25 pm

Bruce Hall — there are hundreds of safety regulations on every autos.

so I guess your statement is incorrect.

Bruce Hall November 13, 2006 at 5:51 pm


I believe you misinterpret my comments. I realize that there are many regulations regarding automotive safety. My argument was that it does not make sense from a safety, deployment or litigation perspective to have 50 separate STATE regulations for something like seat belts.

Regulations accomplish two things:

First, they accelerate the deployment of a safety feature such as seat belts and air bags that otherwise would meet significant market resistance and manufacturers' unwillingness to add expensive content to a vehicle that is not generally perceived to be cost-effective by the public (note I said "generally").

Secondly, they offer the manufacturers who meet those standards a certain level of protection against those who demand "perfect" performance (perfect always being defined by hindsight).

While Peltzman might argue that the death rate would have fallen by itself, that is a vast stretch of the imagination. Without regulation, the deployment of those features would have been much slower (just as would emission controls) and significantly more lives would have been lost.

Peltzman is looking at things through rose-color economist's glasses… not the real marketplace. It is not the "logic" that is wrong… it is simply that the comparative DISADVANTAGE for a manufacturer to add expensive safety (and pollution control) features if others do not is too great. Volvo made a living on "safety", but others didn't follow until forced to do so.

spencer November 13, 2006 at 6:23 pm

Ok Anon you have shown me "a" case where seat belts caused a death. But the issue is still one of cost vs benefits. and the evidence is still overwhelming that seat beats and/or airbags have saved massively more lives then they cost. Moreover, the evidence is that government regulation has increased the use of seat belts/airbags and the net benefits of government regulation on this issue is overwhelmingly positive.

If I applied your standards I would be opposed to free trade because it caused a
single individual to lose their job.

Moreover, if I applied your standards I would be opposed to techonology because it causes individuals to lose their jobs even if it makes everyone better off.

Debashish November 13, 2006 at 9:44 pm

Prof Roberts,

I find EconTalk very interesting, and was wondering if transcripts of the interviews/discussions are available? I couldn't find them on the EconTalk website. While audio certainly has its own advantages, I feel that I often understand the content better if I read it (either on screen or a hard copy).

Thank you.

Russ Roberts November 13, 2006 at 10:13 pm

I have removed a recent comment on this post that was off-subject. (The same comment can also be found on another recent post.) We appreciate all points of view here at the Cafe as long as they are civil and about the subject at hand.

spencer November 14, 2006 at 8:28 am

What do air bags and/or seat belts do?

On average, in case of an accident they reduce the severity of injury or likelyhood of death.

Unlike better brakes, improved steering, brigher lights, or superior roads they
do not impact the performance of a car.

So there is no reaons to think that airbags and/or seatbelts have any impact on the accident rate.

So what Peltman has found is a nonexistent
phenomenon –improved accident rates –that is offset by an imaginary human response.

No wonder the net impact is zero.

Tom November 14, 2006 at 10:02 am

I would not dispute that seat belts make me safer, but I hate that I can be ticketed for not wearing one. I don't even mind paying for tha added safety, but whether or not I use it is MY business.

Bruce Hall November 14, 2006 at 10:18 am

Two comments:

I don't believe the issue was accident rate as much as fatalities and injuries. Systems that improve performance will certainly reduce both the accident rate and the death/injury rate per accident.

When accidents happen (and they do – last year there were 43,000+ deaths and 2.7 million injuries) safety systems can and do reduce the number of those killed and injured.

Your choice not to wear a seat belt (or disable an airbag) may be your way of expressing some anarchist's disdain for regulation and authority, but it does place you at a higher risk of injury and death. And should that happen, you will be among those who affect insurance rates that the rest of us pay.

I always tell my sons that "stupidity has its own rewards."

anon November 14, 2006 at 11:08 am


I am fine with individuals making that cost/benefit analysis. I'm even fine with them making it for their children. I'm not fine with the government doing it for them. Maybe it's quaint, but I still find significance in the public/private distinction.

Steve November 15, 2006 at 12:27 pm


When I'm wearing shoes, I walk through a parking lot differently than when I'm not wearing shoes. I don't worry so much about stepping on broken glass or other sharp objects, so I pretty much amble about as I please. I'm sure my rate of stepping on broken glass goes way up. Why do you think it's impossible that people may similarly modify their driving behavior after taking safety precautions? Would you argue that I should be required to wear shoes when taking out the trash? Why can't I determine the whether it's worth the cost (maybe I left my shoes at the beach 1/2 mile away)?

spencer November 15, 2006 at 2:10 pm

Steve– you do not get my point.

The Peltman argument is that airbags and/or seatbelts have two impacts.
1. reduce accidents
2. cause drivers to drive more recklessly
and this causes more accidents.

I do not agree that airbags and/or seat belts reduce accidents.
They reduce the severity of injury and or chance of death after an accident occurs.
But there is no reason to think they reduce the number of accidents.

The Peltman argument is that they reduce accidents x% and this is offset by an equal and exactly increase in accidents because of more reckless driving.

But if the initial impact of airbags and/or seatbelts is zero, the impact of reckless drivin also has to be zero.

So he has found two nonexistent forces.
No wonder they sum to zero.

spencer November 15, 2006 at 2:49 pm

Moreover, I find it interesting that a paper cited by Peltzman was by Liran Einav. This paper, "The Effects of Mandatory Seat Belt Laws on Driving Behavior and Traffic Fatalities" found that this theory is not supported by the data.

Cohen, Alma and Einav, Liran, "The Effects of Mandatory Seat Belt Laws on Driving Behavior and Traffic Fatalities" . Review of Economics and Statistics, Vol. 85, pp. 828-843, 2003 Available at SSRN: or DOI: 10.2139/ssrn.293582

I guess all this proves is that if you search hard enought and torture the data enough you can demonstrate any point you want to.

Previous post:

Next post: