GPS: The Missing Link

By Christopher Sirota and Steven Craig

Many auto insurers are interested in using telematics to collect accurate data on driving behavior. Telematics requires that a vehicle have an onboard device and a means of data transmission, typically a cellular connection. A telematics device can collect data from the diagnostics port, its own accelerometer, its own GPS (Global Positioning System), or any combination of the three.

GPS is of particular interest because it provides accurate time/date stamps, speed, and direction. Even more important, it can provide location such as latitude and longitude, a valuable link to layers of geographic data known as GIS (geographic information systems) data. The general public is becoming more familiar with GIS and GPS because of personal navigation devices such as TomTom and Garmin and cell phone navigation applications that link GIS (maps) and GPS (location). Tele­matics devices take this one step further by recording and wirelessly transmitting frequent GPS locations, thus enabling users to share the locations of their driving journeys. Sharing GPS journey data concerns some privacy advocates, but if used appropriately by an insurer, such information can provide valuable policyholder services and be a unique source of data on personal and commercial driving behavior.

Linking Data Sets
An insurer that decides to use telematics to understand its book of business can choose not to add GPS to the telematics device. That way, the insurer avoids the location privacy issue. However, it forfeits a valuable data link. Figure 1 displays some of the information available with telematics and GIS data sets.

The telematics device can provide event and driving-behavior data, and map layers can provide GIS-based data that describes conditions. When linked together by the GPS data, the two sources reveal significantly more about the auto risk. By adding policy-level data such as garaging address to the mix, an insurer can accomplish even more thorough analysis and comparison. As another example, hard braking alone is not as revealing as hard braking during inclement weather, during rush hour, and on highways rather than on side streets. Only with GPS can such a data set be created.

What Can Location Immediately Tell an Underwriter?
Modeling with linked data sets is an exciting goal and challenge, but it will take time to collect a credible amount of data. Some analysis can be done much sooner. An insurer could compare the reported garaging address to a location that the GPS journey data reveals is more likely to be the actual garaging address. Journey data can be analyzed to determine accurate commuting distance and radius of operations. An insurer could further link each GPS journey to geographic loss costs, as shown in Figure 2, to analyze for riskiness. Note how the green GPS journey line passes through the different color sections. Each section's color represents a range of predictive loss costs. One method of using this analysis would be to calculate the weighted average of miles driven through each section. One could measure the difference between this result and the loss cost of the original garaging address.

Privacy
A policyholder should feel comfortable with telematics and recognize its value before voluntarily installing a device. Some services that use location-derived information are risk management of teens, risk management of seniors, breakdown towing, crash assistance, parking location, and stolen vehicle recovery. In addition, users can set geographic boundaries to an online map that reveal whenever the vehicle enters or leaves the boundary. For example, an alert can signal when a teen arrives at school or the vehicle is moved from home after midnight. Those services are event-based and reveal exact location for the benefit of the policyholder.

However, a policyholder may not want exact locations available for insurer needs, especially the location of every driving journey. That is why many insurers using telematics with GPS choose to collect only summarizations of driving journeys. An insurer then needs to consider levels of summarization, the time periods of data sets, and the frequency of the GPS recording. Too much summarization will yield less valuable data for modelers. A compromise is needed.

One method, also displayed in the above diagram, allows users to record the riskiness of locations driven by the insured vehicle without knowing the actual geographic locations. That's accomplished by dividing a state into sections, calculating the riskiness of each section (as shown above by color), and then grouping the sections into an appropriate number of bands for that state. With this method, an insurer that uses a telematics device with GPS can calculate how many miles are driven within each risk band (a "loss cost band" in the example above) and never need to know where the band is located geographically. That protects the privacy of the driver. In addition, the insurer can choose to combine other data elements, such as speed and time of day, into bands to protect privacy. This method requires a considerable amount of data to be credible for regulators.

Improving Claims Adjusting
Telematics devices can also have accelerometers that not only provide information on driving behavior, such as braking and steering, but — with the correct algorithm — can also indicate incidents of extreme vehicle movement. Such events can trigger a call to a service center to determine if the driver needs assistance. If the condition is unknown, a 911 call can be made.

When the location of such an incident is critical, GPS is a necessary link. GPS enables emergency service and provides an accurate time/date stamp for the incident. The device could quickly initiate a first notice, which in turn can reduce costs. Although an emergency incident will require an exact location, it is more acceptable because the policyholder benefits. In the long term, collecting incident data with linkable GPS data will provide further geoanalytics opportunities.

Christopher Sirota, CPCU, is product development and research manager at Applied Informatix®.

Steven Craig is general manager of ISO's Applied Informatix division and A-PLUSTM (Auto-Property Loss Underwriting Service).