seatbelts, economists and speed

There’s an oft quoted study that when people are forced to wear seatbelts they drive faster. The notion being that seatbelts force reduce the cost of risky driving, so to speak, and people move the risk back up to their tolerance level by driving more quickly. I recently read about some research that debunked this. However, an econ prof just said it again.

Anybody know the whereabouts of this study and any revisions, or replications, of it?

It’s BS.

this analysis (warning: PDF file) cites several studies indicating that speeding (and passing other vehicles) is associated with not using seat belts.

However, both speeding and wearing seat belts were more prevalent in highly educated populations.

Thanks! I did find a few more in the category of the original argument. If you have more either way, please let me know. Thanks again.

The original study was done waaay back in 1975 by Sam Peltzman. There has been quite a literature about it since, and as is often the case, the empirical results are fairly mixed. Cites to the original article and discussion of the literature can be found in the following entertaining little paper:

http://www.be.wvu.edu/div/econ/work/pdf_files/03-01.pdf

(The paper itself is an interesting attempt to apply the theory to, of all things, NASCAR.)

Nametag, the study you cite isn’t quite relevant, because it’s a cross-sectional study (how do different people act within a given environment of rules and regulations?) but the question posed is a time-series one (how do people alter their behavior in response to a change in the rules and regulations?)

Thanks. I guess the issue of why people drive faster is hard to know without some of the information in the paper linked by Nametag.

This is called ‘risk homeostasis’ and there have been several studies supporting the theory.

One study even found that people who have airbags installed in their cars drive more aggressively than those who don’t because they feel safer, so they take more risks-

http://www.ncpa.org/ea/easo96/easo963a.html

Think how carefully you’d drive if your steering wheel came equipped with an 8" steel spike pointed at your chest.

One of the articles mentioned the problems with that point, Sam. It is reversing the problem to solve it and isn’t strictly logical. Besides, these are attempts at empirical verification, not just arm chair speculation articles (not that there’s anything wrong with that!). :wink:

Surreal, the really bad editing of the summary you point to on the right-wing website makes it hard to believe they got it right. As the other article linked here noted, there’s lots of evidence both ways.

You wouldn’t drive at all. Assuming that the slope of risk/driving caution is a constant is kind of silly; there’s no reason to believe the risk aversion formula is a straight positive correlation. Some people are only vaguely aware of whether or not their cars even HAVE air bags.

well i always wear a seatbelt when i drive fast, but if i am not wearing it, then its cuz i am driving slow on purpose, such as i am looking for parking. if somebody forced me to put a seatbelt on, i still would be looking for parking at low speed, not 80 mph.

now if somebody forced me to remove the seatbelt and turn off the airbag i would definitley go easier on the highway :slight_smile: thats cuz i have made use of that airbag twice already in other two cars.

I was being a bit flippant. I’ve seen the study, and I’m aware that it’s controversial.

I can’t remember - does it control for things like impovements in car handling?

I wonder if cars with airbags tend to be a bit more upscale, and therefore tend to also handle better, leading people to drive them faster?

The OP hits the nail on the head in viewing this as a question of economics – there are costs associated with driving at speed, one of these is risk. If risks are reduced then the overall costs are reduced, and hence we should expect an increase in “demand”.

This is such a well-observed phenomenon (decreased costs -> increased demand) that we should be surprised if risk compensation did not occur.

However, the jury is, at least, “still out”:

This article claims

But this (an article on the subject of rock-climbing(!)) claim

My own opinion is that risk compensation is inevitable and undeniable – consider the natural response to slow down as lanes narrow, or in the act of cornering, can anyone propose a satisfactory explanation that does not amount to a description of risk compensation?

Sam Stone’s “flippant” suggestion above, is another fine example:
steel-spike -> insanely-greater-risk -> massively reduced driving speed.

Imagine driving with a precious and/or fragile object in the boot (trunk) of your car? Would you not drive slower and more carefully? Why is that?

Hell! What is there to argue about? The examples are myriad, the conclusion to be drawn unambiguosly point to the existence of rick compensation.
Much more reading can be found Googling for “risk compensation” seat belts.

There is much to argue about…empirical evidence is needed not just armchair speculation. I prefer Galileo’s methods to Church Decreees anyday. In this case, you are taking the oft noted stance that “econ theory explains it” therefor lets not debate it.

I think the point that the relationship between perception of safety enhancements and change in behavior is not linear is telling. Here’s another example: people who drive SUVs are driving less safe vehicles, I’m told, than other choices. Are SUVs going slower on the roads? I haven’t noticed it. It would be interesting to see if people driving safer vehicles based on car crash ratings drive faster than people in cars with lower such ratings, assuming there’s a way to hold constant other variables indicating the risky taking attitudes of buyers of certain brands or types of vehicles.

I guess ideally, you want to examine people driving the exact same model in the same state, but one with air bags, or whatever, and one without.

Any other thoughts on the ideal experiment to answer this question?

The first link The Great Unwashed provides leads to a large number of articles and letters on the bike helmet issue. This just goes to show how complicated it is to study these issues and how political, and nasty, some of it becomes.

The first link The Great Unwashed provides leads to a large number of articles and letters on the bike helmet issue. This just goes to show how complicated it is to study these issues and how political, and nasty, some of it becomes.

Pish and tosh! I gave plenty of “empirical” evidence, and if I was the sort to take umbrage I’d be bent double under the weight of it at your comparison of my appealing to widely received and studied theories of economics to “Church Decrees”.

Whatever, by SUVs I take it you mean Sports Utility Vehicles? – are they less safe (for the occupants) than other vehicles (cite please)? More importantly, do the drivers of those vehicles perceive themselves to be less safe?

Your question of whether drivers of “safer” vehicles drive faster points to a problem with analysing the data. Those who elect to buy cars based upon their safety performance are predisposed to see safety as a critical issue, these people tend to be risk averse, this group is not randomly selected.

So a statistic like:
Drivers of “Safety Class 1” cars do not drive faster than drivers of “Safety Class 3” cars
would be meaningless unless we can account for the “selection” principle above.

There are other issues that would tend to muddy the statistics: the data cited in the BMJ Journal did not seem to me to be wholly reliable, for instance

indicated that there were other factors surrounding the introduction of compulsory wearing of helmets, namely, “multifaceted educational campaigns”, which could equally well explain the apparent success of the legislation.

It is worth noting, (and continuing the economics supply and demand analogy) that something akin to income- and substitution- effects should be anticipated, that is, we should not expect the entirety of the risk to be adjusted for, and that at least some of the benefits of increased safety will be consumed as increased safety.

As I said above, the reasons to suppose that risk compensation exists are strong enough to require compelling statistics to dissuade us from accepting it as fact.

Thoughts on experiment design? Yes – 8" steel spikes pointing at the driver’s chest, and just see if they don’t slow down. Risk compensation, perfectly illustrated, voila!

You’re just repeating what you/we have gone over. The armchair comments don’t solve the problem and I would think most studies would start with the null-hypothesis that there is not a difference, not that there is. Yes, the studies are hard to do. I and you agree. Yes there is selection bias, that’s why I mentioned ruling it out.

Please read more carefully.

PS People won’t slow down in an absurdly dangerous car, they wouldn’t drive it! There is also some point at which people in safer and safer cars can not drive much more recklessly, or at least not without getting in constant recks and thus raising the costs (of repairs) and making travel for them impossible. Driving, afterall, is not just about living, it’s about getting somewhere within a certain cost. That cost is not just in safety, but time consuming accidents, gas (faster wastes more), tickets, cultural responses, repairs from accidents, etc., etc. Since speeding up only really cuts down a bit of time, but increases risk of harm (offset by safety), risk of time in an accident, risk of ticket, burning of more gas, etc. it is possible that the amount of room for decreasing safe driving is very little.

I guess in the end, if the vast majority of cars today are more safe, is ther just a general trend to more accidents per car, per mile drive, per whatever? Not an answer to the main question, but just interesting in its own right.

Since we’re being so picky about requiring hard empirical evidence for risk homeostatis, I would like some good hard evidence for the claim that SUVs are less safe.

The two ‘facts’ always pointed to by people who make this claim is that SUVs often don’t perform as well in crash testing, and they have an increased risk of rollover.

Let’s accept those two as true (although SUVs have gotten much better in crash protection in the last couple of years, and the best of them are as good as cars). The problem with making a sweeping conclusion based on those is that they don’t account for all accident situations. For example, what percentage of injury accidents are due to rollovers? Even if SUVs were twice as likely to roll over as a car, that might be meaningless if the difference meant there would be .01% rollover instead of .005%. I don’t know what the true numbers are. I’ve seen a lot of accidents, and I almost never see vehicles on their side or back.

Second, crash tests are suspect as a measure of SUV safety because they typically involve crashes into immobile barriers. In the real world, most crashes are into things that ‘give’, like other cars, parked cars, signs, railings, etc. In those cases, the greater mass of SUVs should be a significant advantage.

Then there is the height of SUVs. This gives several advantages. One is better visibility. I have a Ford Escape, and man, sitting in that thing gives me a commanding view of everything around me. Best visibility I’ve ever had in any vehicle. Obstacles on corners like mailboxes and pedestrians can be looked over, whereas in my old sports car they would create blind spots.

The other thing about the height of the vehicle is that if you hit things that fold over or fly up (deer, signs, etc) they’re less likely to fly up and through the windshield. I had two uncles who were killed on a hunting trip when a moose went through the cabin of their car.

So, when you add it all up, are SUVs really less safe?

I’m just basing this on what the guys on the radio say on Car Talk and the postings inside my friends SUVs. It might not be true, but it was to make a point. That people driving what might be less safe than their previous car might not drive more safely…again, very hard to study.

(Regarding crash tests, aren’t some with cars hitting cars? I think the side impact ones do this.)