Fun Facts, Fatalities and Big Data

Today’s fun fact comes from State Farm. U.S. drivers are just as likely to have a claim involving a collision with deer, elk or moose as they were last year, according to new claims data from State Farm. The odds drivers will have a claim from hitting one of those animals is 1 out of 169, the company says, same as it was in 2014. The likelihood of such a collision doubles during October, November and December.

For the ninth year in a row, West Virginia tops the list of states where a deer collision auto insurance claim is most likely to occur. The odds a driver in the Mountain State will have a claim did improve to 1 in 44, up from 1 in 39 in 2014, an 11.4% decrease. Hawaii rounds out the bottom of the list also for the ninth year in a row with odds of 1 in 8,765.

I mention this data analysis because it is a great public service that State Farm is performing in sharing this data. In order to produce this analysis, State Farm has to look at its own claims data across all fifty states, a process which is not always digital in nature and, as a result, is always time consuming and expensive.

But this kind of analysis is precisely the kind of life-saving analysis needed to mitigate the miserable level of highway fatalities in the U.S.

The slaughter of human beings on highways in the U.S. has been compared to the equivalent of a daily plane crash or terrorist mass killing event – with approximately 100 fatalities occurring every single day. The number of fatalities is exceeded only by the numbers killed daily in China, India and Brazil – with Brazil only recently surpassing the U.S. (But take heart, fatality fans, the U.S. tally increased in 2014, according to the National Highway Traffic Safety Administration.)

Automotive safety gadfly, Lou Lombardo, a former NHTSA analyst, notes that the total number of highway fatalities in the U.S. to date is 3,685,564 – three times the number of U.S. fatalities in all wars since 1776. As a result of these staggering figures, the historical and daily death toll, there is a cottage industry around mitigation strategies.

The latest contribution to the fatality reduction knowledge base comes from Rand Corporation which looks at state-level strategies for finding the most cost-effective ways to reduce highway fatalities. The report is enlightening, perplexing and disappointing – concluding as it does that measures for reducing drunk driving represent the single most cost effective way to reduce highway fatalities.

This position from Rand contrasts with NHTSA’s emphasis on reducing driver distraction, blamed for approximately 3,000 highway fatalities annually. Rand identifies 11 interventions which, if universally implemented across the U.S. would reduce the annual highway fatality total by 3,939 at a cost of $2B.

But Rand does not stop there. The report ultimately concludes that the ideal cost benefit ratio and recommended path forward prioritizes a handful of state-level interventions likely to produce the maximum positive outcome for the lowest expenditure.

In the words of the report’s concluding paragraphs: “Some state-intervention combinations have very high cost-effectiveness ratios under our assumptions. In fact, 10 intervention-state combinations have cost-effectiveness ratios greater than 100, meaning that the monetized benefits in terms of reduced injuries and fatalities is at least $100 for each $1 spent on implementation. We find that implementing these 10 state-intervention cmbinations would cost $2.1M and would prevent 170 fatalities. That works out to a collective cost effectiveness ratio of 167. When less cost-effective interventions are added to the mix more deaths are prevented, but the cost of doing so gets very high.”

https://tinyurl.com/zjgl8or - Using Cost-Effectiveness Analysis to Prioritize Spending on Traffic Safety

The problem I have with the report is that it focused almost exclusively on intrusive enforcement propositions including ignition interlocks, license plate and/or vehicle impoundment, and traffic and speed camera enforcement – with speed cameras promising the single greatest reduction in fatalities of more than 1,000 annually. The report appears, to me, to assume that all roads are universally safe as are all vehicles. The suggestion is that the sole source of traffic fatality remediation lies with sanctioning or limiting drivers.

This is in keeping with the increasingly expressed theme emanating throughout the automotive industry that drivers are responsible for 80%-95% of highway fatalities, depending on the commentator. This assumption, then is rapidly being disseminated and accepted as conventional wisdom feeding the long-term transportation thought processes around the development of autonomous vehicles or automated driving – two very different things.

At stake is nothing less than the right, privilege and ability to drive a car. All of us human drivers are being blamed for nearly all of the highway fatalities. It doesn’t seem right.

I hate to be a bubble burster, but more careful aggregation and analysis of vehicle crash data including location, time of data, weather conditions, etc. ought to lead to far more clever strategies for reducing highway fatalities. One third of all fatalities in the U.S. occur at intersections. I have little doubt that it is possible to identify the most dangerous intersection and take appropriate measures (lighting, signage, cameras, beacons) to reduce collisions at those locations.

Similarly, speeding the closure of outstanding recalls and fostering the adoption of advanced safety systems ought to make cars safer and less prone to collide. There is a role for the insurance industry in encouraging the adoption of such systems as automatic emergency braking, lane keeping, blindspot detection and pedestrian collision avoidance.

With respect to Rand, it is hard to argue with primary seatbelt law enforcement, motorcycle helmet requirements and limits on diversion and plea agreements, but applying big data strategies to the reduction of highway fatalities is a strategy that has yet to be applied. Rand’s strategy of looking at the highway fatality problem on a state-by-state basis is relevant here as there is limited or no aggregation of vehicle crash data.

Insurance companies like State Farm may actually represent the best hope to mitigate highway fatalities standing as they do at the gates to Big Data insights. State Farm last mounted a major effort to raise consumer awareness and stimulate state level funding of collision mitigation strategies more than 10 years ago with its now defunct “Dangerous Intersections Initiative.”

The Dangerous Intersections Initiative leveraged State Farm’s own claims data to identify dangerous intersections and provided matching funds to municipalities interested in trying to reduce collisions at those locations. But State Farm’s altruistic gesture was not greeted warmly.

https://tinyurl.com/opsjd2m - State Farm Dangerous Intersection Initiative

In the words of one anonymous State Farm-er: “We studied our claims data and then for the top 10 most dangerous intersections gave the ‘owning’ government entity $100K to use towards improving the safety of the intersection. It was received with mixed emotions – some people got very defensive because it made them look like they had either designed a flawed intersection or weren’t paying attention on their own to proactively improve it.

“We have tons of auto claims data, but for those efforts they were paper claim files that we had to review manually to extract location data to use. Even today we’re not getting precise location data that is readily available. There would need to be some human involvement to ensure quality data.”

The best hope for applying big data strategies to mitigating highway fatalities may lie with the auto companies. Like the insurance companies, auto makers are in a position to look down and across the entire landscape of vehicle accidents and have a vested stake in better understanding how their vehicles and the drivers performed in or contributed to a collision.

Connected cars offer maybe the best chance to achieve reductions in highway fatalities, but someone, somewhere will have to be looking at, cleaning up and trying to make sense of the data. In the wake of record recall rates and heightened NHTSA vigilance car companies are looking more closely than ever before. The outcome of that ongoing analysis may help to determine whether you and I will be allowed to continue to drive our cars.