Telematics - Casualty Actuarial Society

Report
Marty Epstein, FCAS, MFE
Global Auto Actuary, AIG
Casualty Actuaries of Greater New York (CAGNY)
December 6th, 2012
or

Broadest definition: the technology of
collecting, sending, receiving and/or storing
information via telecommunication devices

The three most popular Telematics products
in use: in vehicle navigation systems, stolen
vehicle recovery systems and fleet
management systems

A distant 4th most popular product is what
current discussion focuses on most: Pay How
You Drive, Pay As You Drive, or User Based
Insurance

Telematics products are most common in
countries with high rates of theft (Italy, Brazil,
South Africa) or in expensive, sophisticated
markets (UK, Western Europe, United States)

Penetration of Telematics products is globally
very low but expected to grow gradually until
product is commonplace (10-20 year
timeframe)

Too expensive for most markets
 Hardware is pricy, though becoming cheaper
 Data plans must be purchased for each device

Too complicated
 Operationally complex
 Data volumes can be massive, challenging to interpret
 Pricing implementation is a challenge
Intellectual property/patent claims on existing
technologies act as a barrier to entry
 Most insurers are taking a wait and see approach or
investing smaller amounts in pilot projects to test
program viability


Smartphone’s are now being piloted to enhance or replace dedicated
boxes for the collection and transmission of driving data
 Mobile only solution is a fraction of current operational costs with the
advantage of piggybacking on the customers own mobile plan
 Also, the user experience can be dramatically enhanced as context relevant
information can be displayed immediately after completing a trip and a
variety of additional services can be made available
 Mobile phone solutions have their own drawbacks, such as estimating the risk
of the phone owner which may not be the same as the vehicle risk

Some countries have plans for mandating Telematics
 Singapore plans to mandate use Telematics devices to determine fees for
driving on public roads
 Brazil seeks to reduce theft through mandatory Stolen Vehicle Recovery
systems

Risk models that use Telematics data will soon consider the context
of driving events, advancing sophistication
 Current models adopted in the US market that determine price credits from
mileage, braking, accelerating and speed data may be considered crude relative to
their potential accuracy
 Some examples of driving context:
▪ How do you evaluate a drivers risk if she exceeds the speed limit but travels at the average speed of
other drivers on a particular road segment at a particular time? What if she is driving in inclement
weather?
▪ Is a driver more risky if you learn that he often brakes midway through a turn rather than before
the turn?
▪ What is the risk difference between a driver who rarely texts while driving and one who frequently
texts while driving? What if the “texter” only does so while the vehicle is not in motion?

Psychological profiles are being mapped to driving data as a way to
understand and explain driving behavior but also to predict
behavior of customers in situations outside of driving

Most common incentive for enrolling in Telematics in the US is the
potential for premium reduction:
 Progressive: “you could save up to 30% extra for your good driving”
 Allstate: “rewards safe drivers with the big savings they truly deserve”
 State Farm: “you have control of how much you save, with the opportunity for
discounts up to 50%”

Additional services are offered by some programs, for example, State
Farm’s In-Drive program (monthly fee based):
 Get help locating your vehicle if stolen
 Vehicle diagnostic reports
 Emergency response notification
 Monitor high-risk drivers in household, such as teen or elderly drivers

Pay as you drive offers a way for low income and uninsured drivers to
acquire insurance

Context specific driver education

Praise for things you did right, guidance for things you could improve
Eco driving (feedback on your CO2 footprint and how to improve)
 eCall (in vehicle emergency service call button)




Planned for the EU in 2015
Claim submission and accident reconstruction

First notice of loss (identify location and time of crash, allow user to attach photos of damage)

Verify facts such as speed prior to impact, location of impact(s), estimate likelihood of
property damage and soft tissue damage; reduces fraud risk
Gameification / social networking

Prizes for “safest driver”
User portals with detailed information on driver behavior
 Targeted driver safety training and alerts for families with inexperienced or
elderly drivers

Insurers will have an opportunity to dramatically
change the interactive experience with their
customers
 What can be learned about your driving behavior will
be used to help predict your behavior in other
situations (subject to user agreements)

 Insurers will have an opportunity to understand their
customers behavior in ways only dreamed of until now
 Human behaviors to be measured such as:
aggression/calmness, focus, consistency, thrill seeking,
avoidance, pacing
Pandora’s box of nearly unlimited, intimate data is
open, whether we like it or not
 Insurers have been loathe to use the stream of data
out of fear of being called “big brother” or seen as
taking advantage of privacy standards
 Technology focused companies are jumping in to
driver safety space with new ideas that have the
power to transform the insurance business
 How will customers make a value judgment as to
whether they should share their data?

 Worst-case scenarios have mostly been speculative so far

Will the addition of Telematics data to loss cost
analysis result in a significant or only marginal lift in
predictive power?

Does the use of Telematics data bring us too close to
individual risk rating? What might be the unintended
consequences?

Telematics models indicate that many drivers ought
to receive a surcharge rather than a credit for their
riskier than expected driving – US insurers currently
only provide price discounts (no discount for riskier
drivers) -- is this sustainable long-term?

similar documents