Using proxy exposure units to measure commercial auto risk

With increased access to sophisticated data analytic strategies, there is growing opportunity to jump into the commercial auto insurance business with sound risk analysis and decision-making.

Commercial auto has long been a troubled line of business for property and casualty insurers.

Since it suffers from historically poor performance, many insurers refrain from writing certain businesses in the line entirely, which eliminates a potential revenue stream and plagues those trying to create a profitable commercial auto portfolio.

Even though insurers would like to have access to driver and telematics data, they recognize many businesses and related service providers are unwilling to turn over telematics information. And even if they could access such information, the data may be incomplete because of high driver turnover.

It also can be hard to sell top management on the purchase or investment of data sources.

Commercial auto insurance obstacles

Additionally, as insurance companies venture into usage-based commercial auto insurance for the first time, they are trying to make inroads in understanding the existing data structures. They are trying to scale up across various distinctions of commercial auto business.

They are looking for quick success to showcase the skill and build a case for future investment. This contributes to the reluctance to write commercial auto insurance.

With ever changing access to external data and sophisticated data analytic strategies, however, there is growing opportunity to jump into the business with sound risk analysis and decision-making.

What are the alternatives?

In situations where the behavioral data of the drivers in not available, it is possible to use vehicles as proxy for drivers to derive a manual premium value. This approach involves application of ISO document rules on the basis of vehicle class code and territory information to create on level premium.

This solution is feasible, and works, as it has been cited in many rate-making papers.

For example, geography is considered one of the primary drivers of claims experience. Consequently, it is one of the more established and widely used rating variables.

Companies typically define territories as a collection of small geographic units (e.g., postal/zip codes, counties, census blocks) and have premium rate relativities for each territory.

The territorial boundaries and associated rate relativities can vary significantly from insurer to insurer. Therefore, the vehicle’s geography can be used as driver’s proxy. This information can be used to do loss ratio modeling and have more informed underwriting and pricing techniques instead of focusing on average costs and expected number of claims that will be received during a given time (frequency severity modeling).

ISO data methodology

As you look to adjust premiums to current rate level (extension of exposures is a requirement), the ISO document method considers the following factors mentioned in the ISO manuals based on class code characteristics of a vehicle:

Rating territory

Insurers or rating agencies often divide states into geographical subdivisions called rating territories.

Fleet vs. non-fleet

Under a commercial auto policy, vehicles are subject to either fleet or non-fleet rates.

Fleet rates are typically lower than non-fleet rates. A fleet normally consists of five or more self-propelled autos (such as trucks or tractors). A trailer is not self-propelled, so trailers are not included in the vehicle count.

Size class

Under the ISO classification system, trucks are categorized into size classes based on their gross vehicle weight (GVW). The GVW is determined by the manufacturer. It is the weight of the truck when loaded to its capacity with people and cargo.

Size Class	Gross Vehicle Weight (Pounds)
Light Truck	Up to 10,000
Medium Truck	10,001 to 20,000
Heavy Truck	20,001 to 45,000
Extra-Heavy Truck	Over 45,000
Heavy Truck-Tractor	Up to 45,000 (GCW)
Extra-Heavy Truck-Tractor	Over 45,000 (GCW)

Radius class

Another factor that affects commercial auto premiums is the radius of use. This is the distance a vehicle normally travels each day from the place where it is stored. The distance is calculated using a straight line from the origin to the destination. A commercial vehicle is assigned to one of the following classes:

• Local Up to 50 miles
• Intermediate 51 to 200 miles
• Long Distance More than 200 miles

Physical damage

The premium charged for physical damage coverage on a commercial vehicle is based on the vehicle’s age and its current cost.

Private passenger types

The rating of private passenger vehicles is much simpler than that of trucks. The liability premium for a car is typically based on a flat charge. This charge varies depending on the rating territory.

All the variables above along with additional variables will gives us a combination of factors which can be used to calculate the on level premium of the commercial fleet.  Typically, for a commercial vehicle this can be represented by a simple mathematical equation

On Level Premium= Factor 1* Factor2 * Factor 3

Where, Factor 1, Factor 2, Factor 3 are combination of various hidden ISO factors received from Verisk.

Benefits of using ISO data methodology

Certainly, it is advantage to have actual driver data, but the fact remains it is possible to complete an on level premium calculation of commercial auto insurance.  In addition, this calculation has other benefits including:

1. Massive cost saving; gone is the need to buy telematics data and driver behavior data;
2. Repeatable process, which is easier to implement across different tools like R and SAS, and its usage is not confined to Big commercial insurance players;
3. Data knowledge gains;
4. Due to this on level premium, one can get extra variables for example additional information on nature of vehicles for a policy like age of vehicles, etc.

Limitations of this approach

There are also limitations to be aware of as you look at employing the ISO data methodology. These include:

1. This is not applicable to an entire line of business.
2. May not the most reliable as many policies are defined by driver characteristics and not by vehicle characteristics.
3. Incomplete vehicle data. Vehicles also keep changing for a policy, so the calculation needs to be fixed for a time period. This has a great impact for non-current year policies, it represents data at a point in time (mid-term/ end term of policy).
4. Absence of certain variables during methodology.

The use of data analytics is both necessary and inevitable for implementing successful lines of commercial auto insurance.

Even with stated limitations, the ISO data methodology demonstrates how insurers can set premiums in the absence of critical data such as drivers and telematics data while showing a cost advantage as it avoids buying behavioral data.

Dheeraj Pandey is assistant vice president at EXL Analytics, a provider of data analytics solutions to financial organizations including P&C Insurance firms. To reach this contributor, send email to Dheeraj.pandey@exlservice.com.

See also:

8 ways telematics will shape insurance agencies

A look at the evolution of insurance telematics

5 reasons to embrace telematics for the connected car