Usage Based Insurance: Is Braking Bad?

usage based insuranceIn the usage based insurance realm, braking events are an important indicator of risky driving behavior. If braking events per 100 miles are higher than average, the driver may be distracted (i.e. texting); the driver may not be maintaining proper vehicle distance; or the driver may have aggressive driving tendencies. In many claim situations, research has shown that frequency leads to severity. Likewise in driving, it’s a safe leap in thinking to believe that frequent braking events indicate a higher frequency of crash avoidance and perhaps a greater likelihood of an eventual accident.

While most auto insurers agree that braking is a good activity to monitor, there is debate about whether usage based insurance braking measurements are reliable. Questions linger about the difference between OBD and smartphone braking measurements, leading back to our original question … is braking bad?

Before I address this question, let me explain a few details about how braking events are measured:

  • Braking event: A braking event is defined as a reduction of vehicle speed of at least 7 mph per second. For example, if I’m travelling 65 mph at 12:59:01 and I’m travelling 55 mph at 12:59:02, a braking event should be recorded.
  • OBD device: A basic OBD device takes a vehicle speed reading from the car’s onboard computer approximately once every second. The OBD device has a couple of key limitations most notably that the OBD speed values don’t have timestamps attached. So while the readings are assumed to be one second apart, there’s no way to verify that they are. In some cases, there are small variations and the readings are actually .8 or 1.1 seconds apart and these small variations can result in under- or over-reporting of braking events. Also concerning is the fact that some vehicles send multiple speed readings at the same time and some vehicles report the speed as a truncated integer. Granted, some OBD devices are more sophisticated and include an accelerometer and/or GPS sensors. However, these types of OBD devices are more expensive, further inflating the cost of usage based insurance programs.
  • GPS sensors: Some people believe that a GPS device can detect braking events. However, GPS sensors tend to produce many false positives. That’s because the typical GPS has a low average accuracy of measuring 10 meters per one hertz. So, whether used in OBD devices or a smartphones, GPS sensors do not have high levels of accuracy when used independently.
  • Smartphone platform: For the purpose of this discussion, we’re referring to a smartphone UBI platform which constitutes an app combined with cloud analytics. True accuracy comes from the smartphone UBI platform’s ability to interpret and infuse data from multiple modes of measurement. During our studies, our smartphone platform merged data from the GPS, accelerometer, gyroscope and magnetometer. Sensor fusion gives the smartphone platform the intelligence needed to differentiate between an actual braking event and “noise” like when a phone is dropped. However, a remaining concern is that the phone may be turned off at times and some activity could be missed. With that in mind, a smartphone UBI program must have a built-in mechanism for accounting for time periods in which the phone is off, and removing those time periods from the average braking event per 100 miles calculation. This was true in our study.

Now back to the question … Are braking event measurements bad or good?

Earlier this year, we conducted a number of tests to compare smartphone UBI measurements to OBD measurements. As part of our testing, we compared the braking events per 100 miles as measured by a smartphone using GPS only, a complete smartphone platform and an OBD device. Below is the average number of events detected by each:

Results

Smartphone using GPS only

Smartphone Platform with Sensor Fusion

OBD Device

Braking Events per 100 miles 

22.9

7.2

6.1

Conclusion:
While it’s clear that a smartphone with GPS only does in fact create “bad” results with excessive false positives, the other two measurements were remarkably close. Our conclusion is that auto insurers can confidently proceed with a complete smartphone UBI platform as described above, eliminating the burden and cost of hardware, while maintaining braking measurement accuracy.  For more on the results of our study, see our recent article and sign up to receive the full study results when they become available.


Comments are closed.