By Jim Weiss
Usage-based insurance (UBI) for auto is a major investment for any insurer. While the idea of using policyholders' actual driving behaviors for pricing and underwriting is intuitively appealing, the cost of telematics technology required to support UBI can significantly exceed that of other innovations such as insurance credit scoring. That begs the question: With all the money insurers are spending, what constitutes a successful investment? And by what measure?
The parameters for success are often unclear — expressed in vague terms such as "greater understanding of risk relationships" or "improved customer experience." While those are worthy aspirations, a more quantitative metric — specifically, return on investment (ROI) — would arguably better describe the success (or failure) of any capital venture. The definition of ROI is, loosely, the difference between the gains from an investment and the costs of that investment as a percentage of the costs. Analyzing each of those factors and their interdependencies can provide a useful road map for the business decisions necessary to make money on UBI.
In this article, Ubiquitous Insurance Company, or Ubico, serves as the subject of our case study. The chief actuary, Mr. Obi Dee II*, recently suggested three different program designs for the company's entry into the UBI market: one to grow the business profitably, another to stem deteriorating loss ratios, and a third to create a new revenue stream for Ubico. (See Figure 1.) Evaluating the options from an ROI perspective will allow Ubico to make "red light vs. green light" decisions on each and then evaluate other, less tangible factors more typically used to describe UBI.
The chief actuary's first suggestion is geared toward profitably growing the business. It involves offering a 5 percent premium discount to any new policyholder who agrees to install a telematics device in his or her vehicle and report monthly mileage over a one-year period. Insurers typically front the costs of telematics, and the cost of devices that merely record and communicate odometer readings should be relatively minimal. For example, estimate $50 per device for hardware and $2 per month for wireless data transmissions. Assuming a three-year useful life for the devices (that is, each unit can be deployed in three different vehicles over its service life), technology costs represent approximately 2.5 percent of premiums collected from the UBI book.1
On the surface, spending on technology simply to offer discounts doesn't seem like an advisable recommendation to the chief financial officer. However, Mr. Dee argues that Option 1 will reduce premium leakage by weeding out applications from fraudsters who falsify their mileage. The National Highway Traffic Safety Administration (NHTSA) recently estimated that 3 to 4 percent of vehicle odometers are "rolled back" during their lifetime. Moreover, Mr. Dee notes that in the fleet sector, Multiband Corporation estimated "the psychological effect of monitoring" to be a robust 16 percent reduction in accidents;2 the chief actuary believes Ubico can replicate that effect. That would translate to approximately an 8 percent reduction in accidents in the UBI book (not 16 percent, since device installation will be only temporary). If that's the case, then gains under Option 1 would outpace technology costs and premium discounts.
But it's not immediately clear how mileage monitoring would cause policyholders to reduce risky behaviors such as late-night driving. Perhaps the company could use more capable (but also more costly) devices, as in Option 2 below.
The next suggestion is geared toward stemming deteriorating loss ratios in Ubico's existing book of business. It involves offering premium discounts between 5 and 25 percent to policy;holders who operate their vehicles during less risky times of day over a one-year period, as demonstrated using telematics. The estimated average discount under Option 2 would be 10 percent. Technological spending will exceed that of Option 1 because the second alternative requires a greater level of data collection. Assume the technology necessary to support Option 2 costs $75 per device and $3 per month in wireless; that will consume approximately 4 percent of UBI premiums collected.
Best-case scenarios for UBI have shown accident reductions significant enough to cover both discounts and technology costs under Option 2. Mr. Dee observes that a U.K.-based insurer once offered similar discounts to youthful operators — a traditionally high-risk group — for staying off the roads during riskier times of day and was able to effect a 20 percent reduction in accidents. However, technology was costlier at the time, and the insurer eventually discontinued the program. Telematics is less expensive now, but there's no guarantee that Ubico can replicate the loss-side reductions of the U.K. insurer.
The bar graph illustrates the ROI trade-offs of each proposed option for implementing UBI. The left-hand bar for each option represents program costs, while the right-hand bar in each pair represents best-case gains under the program. More ambitious implementations such as Option 2 present greater advantages but also carry greater technology costs.
To enhance the viability of Option 2, Ubico could incorporate cost-cutting measures. For example, an observation period of less than one year would stretch the technological investment by allowing devices to be redeployed in many different vehicles. However, reducing the extent of monitoring might stem loss reductions and defeat the purpose of the initiative. The company may wish to experiment with a more conservative discount plan instead.
The third alternative is to create a new revenue stream for Ubico. It involves offering policyholders one year of value-added services, such as roadside assistance and access to an online portal where they can view their driving habits and post to social media. Mr. Dee believes consumers will pay $8 per month for those services, meaning Option 3 would generate new income equivalent to approximately 6 percent of UBI premiums. However, Option 3 would require more sophisticated devices than Options 1 or 2, costing $100 per unit and $5 per month for wireless. Even excluding discounts, those costs will almost completely negate new revenues.
Option 3 could conceivably effect loss reductions, although it lacks the "psychological impact" of Option 1 or the robust incentives of Option 2. The Monash University Accident Research Center estimated that access to automated crash notification (ACN) reduced accident fatalities by 11 percent.3 If roadside assistance can effect improvements even by a fraction of that number, it will move Option 3 toward profitability — and, more important, save lives.
There are various ways to further improve the ROI prospects of Option 3. For example, smartphones contain many of the same technological components, such as GPS chips and three-axis accelerometers, as the dedicated telematics devices used in many UBI programs. Ubico could eliminate hardware costs from the equation by developing a smartphone application to collect the information required for the portals. Smartphone data would be insufficient to support Option 1 (which requires odometer interaction) or Option 2 (which policyholders could outsmart by powering off during risky hours of driving), but Option 3 doesn't require data sufficient to support insurance rating.
Figure 1 compares the expected discounts and technological spending under each option with the expected loss reductions and new revenues (if any) created. None of the options is a clear "red light." All seem to present reasonable prospects for positive ROI.4 Once the viability of each option has been established, Ubico must decide which of the stated objectives underlying each option is most valuable. It's possible the company could combine aspects of all three proposals: Ubico could offer a discount to new policyholders, further discounts for operating vehicles at safe hours, and paid access to services. That would allow the company to achieve multiple objectives using the same technology and stretch the investment. With an increasing number of insurers deeming UBI financially viable, the one option Ubico and similar companies may not be able to afford — if they are to protect themselves from adverse selection — is to do nothing at all.
Jim Weiss, FCAS, MAAA, CPCU, is manager, Personal Automobile Actuarial, ISO Insurance Programs and Analytic Services.
*OBD (on-board diagnostics) refers to a vehicle's self-diagnostic and reporting capability. OBD-II is a standard introduced in the mid-1990s that provides almost complete engine control and also monitors parts of the chassis, body, and accessory devices, as well as the diagnostic control network of the car.
1. Ubico's average annual premium per vehicle before UBI discounts is assumed to be consistent with recent industry averages. UBI premiums refer to those collected from enrolled policyholders (before discounts) between the device installation period and the eventual discontinuation of Ubico's program, which is assumed to continue throughout the service lives of the technology.
2. Source: "Managing Driver Behavior with Fleet Telematics," http://analysis.telematicsupdate.com/fleet-and-asset-management/managing-driver-behavior-fleet-telematics.
3. Source: "The potential for automatic crash notifications to reduce road fatalities," http://www.ncbi.nlm.nih.gov/pubmed/19026225.
4. Note that there are many costs besides technology associated with setting up a UBI program that, for simplicity, are not considered in this article. One such cost is securing appropriate intellectual property protections on proprietary systems and methods used in rating.