In 1975, Dr. Sam Peltzman hypothesized that more automobile safety regulations lead to offsetting behavior by participants; that is, as drivers gain comfort through safety enhancements, they drive more recklessly. This phenomenon is known as the Peltzman Effect, and with the Daytona 500 just a few days away, it's certainly relevant to race fans. In this essay, I examine whether NASCAR's safety enhancements actually cause competitors to behave more recklessly.
Originally published on NASCARnomics
The short answer is, "Yes; increased safety associates with more driver recklessness." This entry visualizes the relationship between safety and recklessness in NASCAR, quantifies the connection, and explains the importance of these results. I present my main findings:
- As driver-safety improves by 10%, competitors' recklessness increases by as much as 3.8%. More concretely, NASCAR's combination of the HANS device, SAFER barrier, and Car of Tomorrow results in a 3.6% growth in wrecked vehicles.
- Cup Series experience also influences the number of accidents in an event. A Cup field with average experience (7.2 seasons per driver) would reduce its number of wrecks by 10.3% if the entire field gained an extra year of starts. This partially explain the steady decrease of wrecks in NASCAR's top series over the past 7 seasons.
- Increased speeds positively influence reckless behavior as well. As the average pace of a Cup field (160mph) accelerates another 10mph, one expects the number of accidents to increase by 15.8%. Conceptually, drivers have less time to react to trouble ahead when events unfold quicker.
- Other characteristics held equal, road courses and restrictor plate facilities are responsible for more carnage than other configurations. After accounting for the slow pace and long courses, Sonoma and Watkins Glen are the most reckless tracks — almost double the number of damaged vehicles compared to non-restrictor-plate ovals, ceteris paribus.
Race-level characteristics in the Cup Series from 1994 through 2013 are utilized for this study. I exclude races that did not meet or exceed their scheduled distances; ad hoc reduction of an event's distance may bias driver aggressiveness. I assign each race's pole-winning speed as the speed for the race since the average pace for an event is normally skewed by caution flag periods and pit-stops. For races that did not hold a qualifying session, I use the track's previous event's pole speed. I multiply the speed by 3.5% for restrictor-plate events in order to account for drafting effects that are present in races but not in qualifying. After these minor alterations, my cleaned dataset includes 662 NASCAR races.
"We'll race like this until we kill somebody." — Carl Edwards, 2009.
To determine the effect of safety on recklessness, I first define the terms. While previous papers proxy recklessness with caution flags or retired entries, I provide an elegant measurement. Since all drivers involved in an accident identify themselves as part of the wreck, I tally the number of cars damaged in accidents at the race-level. The sanctioning body strictly enforces safety standards for all official races, so I define injuries as instances in which a driver retires early from a Cup Series race and does not attempt the following event on the schedule due to ailments. Finally, I calculate a driver's perceived safety as his probability of injury in a wreck; that is, the number of injuries divided by the number of wrecked cars in the previous 120 races. The following image plots driver safety advancements over the previous 20 years:
Safety has improved dramatically in NASCAR's top series. Whereas every driver in every wreck suffered over a 1% chance of getting hurt in the mid-1990s, recent years show an increased attention to safety — only 3 drivers (Ricky Rudd, Dale Earnhardt, Jr., and Denny Hamlin) have missed events due to injuries directly obtained in Cup Series races. (Of course, there are other methods for defining recklessness, safety, and injuries. To ensure robustness of my results, I apply alternative specifications of each term and encounter similar findings. These calculations are presented at the end of this essay.)
I sketch how driver recklessness relates to safety. Each dot represents the annual correlation between the number of wrecked vehicles and a driver's chance of injury. I insert a red trendline to illustrate the relationship:
A relationship between the two measurements is apparent. As safety in NASCAR improves, recklessness tends to increase. Points from the mid-1990s suggest that drivers were less safe compared to now — 30 average-sized accidents (2.4 cars) typically led to at least 1 injury. In contrast, recent years suggest that NASCAR's product is safer — the risk of injury is reduced (applicably, 0.0% in many years), while recklessness has increased. Of course, other factors might hinder the analysis, so further exploration is necessary. I employ a negative binomial regression to control for other characteristics to determine the marginal effect that safety has on recklessness in NASCAR:
While other characteristics significantly impact the number of damaged vehicles in a race, driver safety maintains its strong and positive effect on recklessness. Specifically, an environment that is 10% safer results in 2.5% extra carnage (or as much as 3.8%). An alternative model that gauges safety with the HANS device, SAFER barrier, and Car of Tomorrow reports that the 3 structures collectively lead to a 3.6% increase in reckless behavior (consistent with a report from 2010).
Other factors analyzed include field experience, speed of the race, the number of drivers competing, and fixed effects for tracks. As competitors earn more Cup-level exposure, they presumably advance their skills. The actual speed of race-cars lends to a greater number of accidents. Driving at higher speeds makes it more difficult to react to trouble ahead on the track. Driver density influences accidents, too. Assuming a field of 43 machines, cutting the track size in half nearly doubles the number of cars in accidents, ceteris paribus. Road courses are an anomaly. When controlling for their lengthy perimeters and relatively slow speeds, Sonoma and Watkins Glen result in the greatest number of trashed cars.
I also considered several characteristics which demonstrated no tangible impact on the number of cars in accidents. The size of an event's purse, Chase races, the number of lead changes, the number of tire providers, and NASCAR's short-lived "5 and 5 rule" carried no significant effects on recklessness.
Because of concerns of accident-reporting changes, I constructed several models to ensure that NASCAR's method of tracking accidents does not substantially effect my results. I present my findings for 8 unique measurements of recklessness and safety:
My reported values are consistent across a handful of various measurements. I confidently submit that the Peltzman Effect, indeed, exists in NASCAR's top national series. Drivers take full advantage of a safer environment and choose to behave more aggressively. Specifically, as driver safety improves by 10%, their recklessness (no matter the measurement) increases somewhere between 1% to 5%. Discretely, the HANS device, SAFER barrier, and Car of Tomorrow have resulted in an uptick in accidents by 2% to 5%.
So why is this important? NASCAR has successfully added safer elements to motorsports that instill confidence into its competitors. Because wrecks demonstrate a positive effect on television ratings and attendance, fans indirectly appreciate a safer atmosphere, too. Potentially dire negative externalities are present, however. As enhanced safety continues to take hold, more wrecks should occur — and the opportunity increases for a car to indavertently find a non-absorbent wall or a mangy catch-fence.
Here are some questions for which I've no answer yet (i.e., chime-in and let me know what you think). Will NASCAR's baseline concussion testing lead to more missed races, perhaps shedding light that too many drivers were "playing hurt" recently? Will the Cup Series' new policy which mandates a driver race weekly (barring a medical waiver) pressure drivers to compete no matter their injury? Has the Chase — a mechanism which elevates several drivers into championship contention — stressed drivers on mid-level race teams to compete every week in order to maintain their outside shot at the title?
Andrew Maness dabbles in economic research for a financial firm in Charlotte, NC. He incorporates his passion for motorsports with his quantitative expertise at NASCARnomics.com. Maness intends to build a career within auto racing as a research and analytical consultant.