Sunday, September 20, 2009

GPS Evidence-Revolutionizing Auto Risk Analysis?

Traditionally, auto insurance premiums have been determined by a number of general factors - driver age, drivers address, past drivers record, and a few more. Wikipedia lists the general process here. However, recently, several insurance companies have sharpened their risk analysis using GPS evidence. The new type of insurance, called Usage Based Insurance, is described by Wikipedia here. Insured persons or organizations typically pay based on the actual miles driven, combined with identified risk factors for that driver. Usage Based Insurance can potentially save the insured a great amount of premium over traditional risk rating systems.

Other systems use on board telekenitics reporting systems that report g-forces and other vehicle information without GPS location reporting. These factors report a lot about the vehicle driven, and perhaps the driver, but not much about the environment it is being driven in.

Combinations of these systems, together with some outside data on real physical risks of the environment, promise a pinpoint risk assessment system that can allow an insurer to charge no more than the actual risk the insured incurs. Rather than rating an entire town or region for risk, insurers could rate individual intersections and road segments, measure the insureds travel on those road segments and through those intersections, compare the insureds skill at driving (in near real time) and assess the premium.

The problem with usual rating systems- As I drive down the road, I am charged with the risk of the 2 a.m. bartime accidents, the "late to work" rush hour accidents, the noon time "hurry to lunch & back" accidents, the teen "after school rush" accidents, dusk "visibility problem" travel, even if I only drive at 10 a.m and 2 p.m. Even if I only travel residential streets, I could be charged with the risk of interstate highway travel, high volume intersection travel, sporting event travel, and the like. Even though might I travel through quite neighborhoods with little traffic, I could be charged with traveling through downtown, industrial area, or dangerous sections, if any of these are in my rating territory. Although I may only drive short distances relatively rarely, my insurer can only give me a small discount because some drivers might lie about their actual mileage, and the insurer take a risk if it has too small a bottom bracket. In short, a rating territory is (and has to be) quite large, and some drivers bear the costs of risks other drivers are taking. Usage Based Insurance balances and corrects those inequities.

The benefit the using both GPS and telemetry systems: The telemetry tells how the insured driver is driving; the GPS location + geocoded accident data (and the time of day) tell how the other drivers in the same area are driving.

If such systems were combined with real time feedback from the insurer, like a talking copilot to warn of dangerous intersections, or detected speed violations for the road in question, the insured could reduce his own premiums and the roads could be made safer for all. Never before could an insurer take positive actions in real time to reduce risk to the insured, both as public service and as a huge money saver.

Here's how it could work (and may work, as the author hasn't reviewed all the systems involved.) The following is just my opinion of what a GPS/telemetric system is capable of:-
  1. Insurers install, for a price break, GPS units on "study vehicles" in a given territory.
  2. Simultaneously, insurance rating organizations geocode (Wikipedia definition) the accident locations for that territory for a given time span, say, the last ten years. These geocodings include a cronological element, giving the time of day the accident occurred. A final code would indicate the severity of each accident, including whether the parties carried insurance for the loss.
  3. Using the geocoded information, insurers assign a relative risk rating to each street segment and intersection hour by hour. If necessary, data can be analyzed by seasons, as accidents will likely increase during winter or rainy seasons.
  4. When the system is generally deployed, insureds pay standard rates while driving with a GPS & telemetry data analyzers for a time period. This allows the insurer to "calibrate" the system and determine the accuracy of the GPS rating system. Ultimately, what rate to charge in specific situations will be a judgment call-however, that rate can be constantly adjusted by my current driving record, my personal driving risk (assessed by telemetry systems in the car), where I am driving, what time I'm driving, the kind, type, severity and time of accidents that have occurred on my route, and the length of time I'm exposed to those risks (by mileage and/or time.)
  5. Finally, the full system is in place, and the insurer starts charging its insureds by mileage driven or time the vehicle is traveling at a rate based on the factors above.
  6. Picture a drive under Usage Based insurance:
    • Sitting in my garage, my vehicle is still under some risk, say, of fire, so I'm charged a very low rate while garaged. Were my vehicle on the street, the risk might be higher, because of theft and weather threats.
    • I drive out onto the street. My onboard GPS tells the insurer where I am. Either in real-time (as it is happening) or later, depending on if there is a link to cellular phone system/ or the internet, the GPS would constantly update my position. The insurer's computers would correlate this position with accident histories on my street, assign a risk score, and charge a premium for my trip so far.
    • I come to a dangerous intersection- no traffic controls and I must cross four lanes of traffic. My GPS comes with "feedback", and warns me of the danger coming up, and suggests a safer route. (This "feedback" information need not come from the insurer in real time - insurers could send CDs, DVDs, or embed the information in the GPS memory for most "feedback" information where the insured is likely to drive. Insureds driving outside a covered area could get realtime "feedback".) As I drive down the road, my insurer can compare accident types, times of accidents, how much those accidents cost and assess how likely I am to have an accident on this trip (with my driving record, age, and driving habits.).
    • If I'm out in the country, its a beautiful day and a good song comes on the radio, my GPS will alert me that I'm going over the speed limit for this. Of course, it will take note of my slip and charge me a slightly higher premium, but the "feedback" stops me from continuing the risky behavior and having a much more expensive accident. The realtime "feedback" saves me money in the long run, as I'm much less likely to have an accident in the future.
  7. When a new accident actually happens somewhere to some GPS tracked driver (and it will) the GPS record and telemetry will help the insurer quickly determine fault, pay the claim if necessary, or litigate the claim if need be, with confidence. If both cars have GPS and telemetry, the additional information will make the investigation much easier. This is not to say GPS will solve all cases- a car crossing over the center line involves feet and inches, not yards & miles- but GPS & vehicle telemetry will narrow the issues like speed, location and timing, depending on how accurate the GPS is set to report and record. GPS's could record data to two areas-Standard and highly-accurate-rotating-buffer areas. The buffer would record, for instance, 20 seconds of very accurate data, continuously writing over that data, but "dumping" the buffer to permanent memory when high G-force conditions were detected. This would "freeze" the data from an accident situation.

If enough drivers have enough GPS "feedback" over time, they will change their habits, become better drivers, and thus, lower their premiums. Bad drivers, on the other hand, will be located quickly:
  1. Bob Baddriver, after an evening in a bar, gets in his vehicle and starts it up. The GPS notes its 2 a.m., bar time in most places, and a dangerous time of day even for the sober. While the GPS doesn't know Bob Baddriver was in a bar (it could, with enough correlation, but that would be both intrusive and prone to erroneous assumptions - maybe Baddriver is the designated sober driver) it can tell if there were accidents in this area around this time, and start charging an appropriate premium.
  2. Bob Baddriver, it turns out, has been drinking. He speeds. He ignores the feedback. His premiums are shooting higher as he goes. He crosses the center line. Realizing it, he locks up the brakes. He takes a deep breath, and drives home as carefully as possible. However, his insurer has noticed. His premium bill comes the next month. He will either pay for his risky behavior or have his policy canceled.
  3. Leaving the same bar Sammy Goodfella is a designated driver, using Usage Based Insurance. While he pays a little more for driving this time of night, the telemetry notes his anticipation of road hazards and safe driving, and his use of low risk routes. He pays a little more for being the designated driver.
  4. Bill Boring, home in bed, doesn't pay anything to cover the elevated risk of these two drivers.
This kind of analysis can go on for any motor vehicle, train, plane, car, bus, or commercial carrier.

As more drivers go to Usage Based Insurance, the remaining pool of traditional insured drivers will likely be the more risky drivers who want to hide their behavior. Premiums are likely to climb among them, driving more to Usage Based Insurance. Eventually, while standard insurance might be available, it is likely to cost so much that even the risky insureds will go with Usage Based Insurance, and the risk the insurer knows is probably less than the risk of the unknown.

No comments:

Post a Comment

Comments are always moderated-your comment will not appear unless it is, in the eyes of the author, relevant.