When auto insurance executives talk about claims severity trends across the industry, a common theory they advance is that repairs on vehicles with expensive advanced driver-assistance systems technology is going to drive costs skyward.
Kay Wakeman, Director of Insurance Outreach for IIHS-HLDI, thinks there’s another factor at play. Speaking during a Society of Insurance Research webinar titled “ADAS, Automation, and the Impact on Auto Insurance,” early this month, Wakeman pointed to “mean shifting” as another influence to consider for certain collections of accident claims, such as front-to-rear collisions.
After revealing results of the Highway Loss Data Institute’s research on eight collision avoidance technologies and their impact on claims frequencies for all auto insurance coverage parts, Wakeman turned to severity questions. She explained the concept of “mean shifting” by going over hypothetical, and then real-world examples.
For the hypothetical, she depicted a pool of insurance claims graphically, with vertical bars representing the cost of each individual claim—some as high as $10,000, and many less than $2,000. A horizontal line across the middle of the graph was meant to represent the average level of all the individual claims–$4,600 in her hypothetical.
“What if suddenly we introduce a safety system on these vehicles that prevents low-dollar claims? They’re going to drop out of the system entirely,” she said, showing a second graph without vertical bars for the low-value claims missing. This causes the average claim severity for the remaining claims to go up “despite the fact that none of the individual claims increased in size.”
“It’s just that we eliminated [some] claims from happening at all in the first place,” she said. On the graph of the hypothetical pool of claims, the horizontal line representing the average moved up to just over $6,000.
“There’s this idea out there that all of these technology systems are really expensive and they’re going be driving severities through the roof. And there may be an element of that. But there’s also an element of mean shifting happening,” she said.
Wakeman went on to observe that the mean cost per claim can also shift down. “What if we’ve got a system that only works at high speeds and it’s going mitigate high-dollar claims? They’re not going go away, but they’re going be less severe because some speed is going to get taken out of the crash.”
“Then our severity is going to shift down.”
Most of the real-world examples of driver-assistance technologies she presented took out low-value claims, shifting overall collision and property damage liability average claim severities upward. One of them was a GM parking system that includes a parking sensor, a rear-camera system, and a rear automatic emergency braking (AEB) system. All three features reduced low severity collision claims. The rear automatic braking feature, in particular, reduced the frequency of claims less than $2,000 by more that 20 percent. With those low-dollar claims dropping, HLDI saw an overall collision claim severity increase in its insurance data base.
“The reason we think that this is mean shifting is that we see the same pattern hold for [both] collision and PDL,” she said, referring to the property damage liability coverage part. “PDL is dealing with the struck vehicle. So it has nothing to do with the cost of the technology on the striking vehicle.” Still, the automatic emergency braking was “really good at reducing low-to-mid PDL claim frequency, giving us an overall PDL claim severity increase,” she said as she displayed a graph indicating nearly a 40 percent drop in the frequency of PDL claims less than $1,500 involving an insured striking vehicle with AEB, and a more than 20 percent drop in the frequency of mid-severity PDL claims (costing $1,500-$6,999). The average severity across all sizes of PDL claims studied rose almost 20 percent, the graph showed.
After showing similar changes in collision and PDL claim frequencies by size of claim and overall claim severities for cars equipped with BMW parking systems, Wakeman reviewed HLDI research results for front crash prevention systems—one from Mazda that works only at low speeds and one from GM that works only at high speeds. Like the parking systems, the Mazda system eliminated low-dollar claims entirely, again increasing overall PDL claim severity. The GM system, equipped with forward collision alert and lane departure warning systems that work at high speeds, had the opposite impact. Because it was good at reducing the frequency of high-cost PDL claims, the overall PDL claim severity shifted downward.
“All of this is to say, yes, some of these ADAS systems can be expensive. They might be impacting severity. But we need to be taking a look at the bigger picture because we’re changing the pool of claims that are still happening,” Wakeman stressed.
AEB Works
Wakeman began her presentation with a summary of the effects of eight different types of crash avoidance technologies on collision, PDL, bodily injury (BI) liability, med pay and personal injury protection claims frequency figures across several auto manufacturers included in HLDI’s research. The eight technologies are: front collision warning, front autobrake, curve adaptive headlights, lane departure warning, blind spot warning, parking sensors, rear camera, rear autobrake.
According to a bar graph summary displayed during the virtual webinar as Wakeman spoke:
Referring to the fact that front collision warnings and front AEB had more impact in lowering PDL claim frequencies than collision claims counts, Wakeman said, “That’s because of the types of crashes these systems are designed to prevent. They’re designed to prevent front-to-rear crashes. That is the single most common PDL claim.”
“The BI claim frequency reductions are even larger than PDL because if we do away with PDL, we’re going to do away with BI. And even in the event that a crash is not avoided, hopefully the driver has at least slowed down, which really lessens the risk of injury,” she added.
Comparing the effectiveness of the two frontal systems—forward collision warning and front autobrake—in lowering PDL and BI claim frequencies, front autobrake was better. “Taking a driver out of the system and allowing the vehicle to brake for itself is giving us larger benefits,” she said.
Moving to crash prevention systems at rear of the vehicle, Wakeman reported that collision and PDL claim frequency reductions for rear autobrake “was the single largest we’ve seen of any technology—much larger than just parking sensors or rear cameras, which rely on the human driver to take that information in and do something with it to avoid the crash.”
Importantly, she said that 20 automakers have committed to making autobrakes standard by September 2022.
One technology that did not show any clear benefit was lane departure warning. “What we found is that people were much more likely to leave their front crash prevention system turned on than they were to leave their lane departure warning turned on,” Wakeman reported, explaining the lack of any statistically signal related to claim frequencies one way or the other. Driver willingness to leave lane departure warnings on depended on “lane maintenance intervention level,” she added, noting that drivers were more likely to leave on active lane-keeping systems that keep them centered in a lane than simple alert systems.
Wakeman devoted a portion of her presentation to sharing claims frequency changes associated with front crash prevention systems of specific manufacturers, including BMW, Nissan and Tesla. And at one point during the presentation, she also reported on findings from joint research with a commercial driver monitoring company called SmartDrive, looking into the effectiveness of ADAS systems on large trucks. Data for over 2 billion vehicle miles traveled from SmartDrive revealed that AEB and forward collision warning were just as effective on large trucks as they were for passenger cars, she said, noting that large trucks showed a 40 percent reduction in front-to-rear police reportable crash rates.