NASCAR Drivers’ Risky Behavior and the Peltzman Effect

The introduction of automotive safety innovations is usually accompanied by concern about the side-effects of those innovations. For example, when seat belts were introduced, people worried that the belts would keep them from getting out of a car quickly enough if they needed to. When HANS devices first became available, drivers also expressed worry that the protective devices would keep them from getting out of the car fast enough, especially if there were a fire.

On the race track, those worries were quickly put to rest when drivers wearing HANS devices were able to escape their cars just about as quickly as they could without the potentially life-saving devices on.

But there is another side-effect of safety innovations that has become known via a theory called risk compensation. This theory say that people typically adjust their behavior according to the perceived level of effect. For example, when it’s icy outside, people tend to walk more slowly and watch where they’re going more closely because the risk of falling is greater. When it rains or snows, you drive more slowly because you know it will take you more time to stop.

The opposite side of this effect is that when you make cars safer, people will drive more recklessly. This is embodied by the Peltzman Effect, which is named after University of Chicago Professor Sam Peltzman (shown at left). This gentleman’s research in business focuses on the interactions between the public sector (government) and the private economy. Professor Peltzman is a distinguished researcher, having served as senior staff economist for the President’s Council of Economic Advisors.

Peltzman was interested in whether mandatory seat belt laws actually had any impact on decreasing injuries or fatalities. Peltzman came to the conclusion that the increase in safety was essentially entirely offset by and increase in risky behavior by drivers (driving faster, not paying as much attention, etc.) Peltzman came to the conclusion that regulation was “at best useless, at worst, counterproductive”.

Not everyone buys Peltzman’s arguments.  There were a lot of technical criticisms about the data set he used and the particulars of the analysis. As I’ll mention in a moment, trying to get an appropriate data set for this is a lot trickier than you might think. You can find papers in the literature that both support and don’t support the Peltzman effect.

But the big sticking point is estimating the magnitude of the effect. I think about it this way:

 

When I make the car (or track, or personal protection equipment) safer, I decrease the probability that a driver is injured in an accident. But if my doing that increases the number of accidents in some way (for example, drivers try riskier passes, take more chances), then I could conceivably offset the improvements. Let’s consider three cases. I’ll make the same safety improvement each time, but we’ll look at how behavior changes vary.

CASE 1: The safety innovation prompts much riskier behavior and the number of accidents increases a lot. Even though you’re safer in the car, if there are more accidents, more people get injured. Think about everyone you know who thinks anti-lock brakes make them invincible in a snowstorm.

CASE 2: The safety innovation prompts riskier behavior, but the magnitude of the riskier behavior exactly offsets the safety innovation. There’s no change.

CASE 3: The safety innovation prompts riskier behavior, but not a large increase – so the net effect is that the number of injuries decreases, even though there is some riskier behavior happening as a result of the safety innovation.

 

As usual, I’ve oversimplified this terribly. It’s not that easy a thing to measure. Let’s say that I want to look at the effects of wearing seat belts in Charlotte and compare before  mandatory seat belt laws and after mandatory seat belt laws. There are records of accidents I can access and I can follow up on who was injured and how badly.

BUT… (there’s always a but)

• What about weather?
• What if people weren’t actually wearing their seat belts?
• What if people were drunk when they had their accidents?
• What about injuries in little sports cars vs. injuries in pick up trucks?
• What about the world’s stupidest animal and accident causer – the deer?

We call these confounding variables and they make trying to get sense out of your data really difficult. The world is just not well-enough controlled for some analyses.

But you know what is?

NASCAR.

Think about it. The cars are similar. Safety equipment is mandated. You’re not going to have to worry about Dale, Jr. unhooking his safety harness in the middle of a race because he needs to stretch. We don’t race when it’s wet (mostly). The drivers are of comparable quality (I know some of you would argue this, but if you look on the grand scale of all the drivers in the country, these guys are all in the top echelon compared to us.) And mostly, the animals stay off the track.

In a paper entitled “Automobile Safety Regulation and the Incentive to Drive Recklessly: Evidence from NASCAR” in the Souther Economic Journal, Russel S. Sobel and Todd M. Nesbit analyze publicly available NASCAR race and accident data from 1972-1993 with the goal of quantifying how much more riskier driver behavior has gotten as major advances in safety were implemented in NASCAR.

Racing is all about knowing when to take risks. If you’re going to chance wrecking your own car, that’s a high penalty for risky behavior. Injury is also a consequence. But if you know that you’re likely to walk away from an accident without injury, you’re going to be much more likely to take that risk that there’s just enough room for you to squeeze up into the next lane between two other cars.

I won’t go into the data analysis because there’s a lot of math things like differential equations and matrices and regressions. I will tell you that I learned a new fancy Latin phrase from reading this paper:  ceteris parabus, which means “all other things equal”. Sure, you can say ‘all other things equal’, but doesn’t it sound so much smarter in Latin?

Anyway, Sobel and Nesbit’s conclusion is that, for NASCAR drivers

“… a 10% improvement in NASCAR automobile safety results in approximately a 2% increase in reckless driving”

Then go on to point out that the increase in reckless driving isn’t enough to produce a larger number of injuries, but is large enough to prove that there is a negative driver response to safety innovations in the form of riskier driving.

They analyzed five drivers who drove from 1972 to 1998 (Yarborough, Parsons, Bobby Allison, Dave Marcis and Richard Petty) in an attempt to get away from some confounding variables, like people who only ran a few races. Even at this micro-level, they found that this group of drivers did get into more accidents as the car got safer, again confirming the authors’ earlier finding that safety innovations encourage risky behavior.

They point out something interesting, based on the idea that there are a group of race fans who want to see crashes, but don’t want to see people hurt. The riskier behavior produced by the safety improvements should actually help NASCAR attract those fans. More accidents, but fewer injuries.

This paper was from well before the Chase, but the current Chase format incentivizes drivers toward even riskier behavior. Even with higher speed and more at stake, remember that the last deaths in one of NASCAR’s top three series was in 2001.

And that got me thinking.

When they switched driving from the left side of the road to the right in Sweden in 1967, there was a drop in crashes and fatalities. For six weeks after the change, the number of car insurance claims were down by 40%. But after those six weeks, the insurance claims returned to normal. And after two years, the fatalities returned to where they had been.

I started wondering how many drivers on the track today were there at the Daytona 500 when Dale Earnhardt, Sr. died? Then I broadened that a little to look at who of that era was still active on track.

I took a pretty deep data dive and managed to confused myself a couple times, so let me explain exactly what I did.

The blue (taller) bars represent the total number of drivers who drove each season. If you look at 2006, you’ll see that out of the original 70 drivers who drove in 2001, only 36 of those drivers were still driving in 2006. The problem with that number is that it includes the road course ringers, the one-offs for people who only drive the Daytona 500, the retired drivers who do a few races a year and the like.

So I went ahead and looked at how many of those drivers were full time drivers – the ones who are the backbone of the series. That’s what the orange bars represent. In 2006, out of the 36 drivers who had been driving in the Series in 2001, 17 of those were driving full time in 2001.  You can also compare the full-time drivers from 2001 (24) with the 17 in 2006. That’s how many drivers who ran full time retired (or ran reduced schedules) between 2001 and 2006.

You can see a pretty clear and steady decrease (as you’d expect) as time goes on. In 2015, there were only fifteen drivers who had been driving (at all) in 2001 and are still around. Six of those were full-season drivers in 2001.

• Dale Earnhardt, Jr.
• Jeff Gordon
• Matt Kenseth
• Tony Stewart
• Bobby Labonte
• Michael Waltrip

Note that of the last two of these drivers, Waltrip has run 3 races and Bobby Labonte 4 races.

A couple other familiar full time in 2015 names and how many races they ran in 2001:

• Kevin Harvick (35)
• Kurt Busch (35)
• Ryan Newman (7)
• Jimmie Johnson (3)

So there are 8 current full time drivers in the Sprint Cup Garage today (out of 43+) who were racing in the era where drivers lost their lives on the track.  Out of those 8, only four (Stewart, Gordon, Kurt Busch and Dale Earnhardt, Jr. if I counted right) were on track when Dale Earnhardt, Sr. lost his life in the 2001 Daytona 500.

If I had a little more time, I’d go look at how the accidents for the next couple races compared to those from the year before in which there hadn’t been a fatal crash at the Daytona 500.

All this data is confusing, I know. So I thought about how to make it a little more accessible and came up with this interactive infographic! I looked at only the 43 drivers who competed in the most number of races each year in an attempt to get rid of the road course ringers and the one-offs. For each of the gauge charts, the blue bar tells you who ran full time in 2001 and ran most of the races in each year indicated. The yellow bar gives you the number of drivers running part-time in 2001.

There’s no orange bar in 2001, which is because the orange bar tells you how many drivers were not driving in the Sprint Cup Series in 2001.

What I’d like you to take away from this is that, as time goes on, the collective memory of the days when drivers lost their lives on a racetrack disappears. I’ve said over and over here, on the radio, everywhere, that there is no way to make racing 100% safe. Despite the best attempts of racing sanctioning bodies, there will be deaths on the racetrack. It’s a matter of the odds and the inherent danger of racing.

That’s why the current trend of ‘I’m mad at someone, I’ll crash him out’ is so disturbing. The probability of a freak accident is independent of intent. How many stories have you heard of people goofing off and something unexpected and tragic happens? It doesn’t matter it’s Martinsville and a slow track. Something in the car breaks and safety measures fail and we have a serious injury or maybe a death.

For my money, here’s the best commentary on the incident.

Tony DiZinno, who consistently provides smart, well-thought out and well-written commentary on motorsports of all types.

Ricky Craven is always on target. He’s got the driver’s perspective, he’s always measured in his thoughts and balances the practical with the ideal.

Nate Ryan and Bob Pockrass react to the Kenseth suspension. Two of the smartest guys on the NASCAR beat.

 

https://medium.com/@TonyJWriter/kansas-martinsville-deference-and-trust-8d20065e7288