InsurTech
Dr. William Quinn
What is InsurTech?
• InsurTech refers to the use of technology to generate efficiency gains
in the insurance industry
• Typically refers to:
• Technology that gathers additional data that improves insurance pricing
• Technology that analyses this data more accurately and efficiently
• Technology that reduces the real probability of a claim
• May also include online intermediaries, firms that make software
used by insurance firms, and automation in the context of insurance
The Problem
• The insurance industry provides conditional credit to customers in
exchange for up-front payments
• Allows people to significantly alleviate personal risks, at the expense
of a small reduction in expected future wealth
• But there are moral hazard and adverse selection problems:
• Once people are insured, they might take more risks
• People who want to take more risks may be more likely to seek insurance
The Problem
• Insurance companies resolve this by gathering as much information as
possible on potential customers
• E.g. for car insurance
• Age
• Occupation
• Years of driving experience
• Years of no claims
• Gender (now estimated via proxies for legal reasons)
The Problem
• They then use this information to estimate the probability and
expected cost of a claim
• This is actuarial science – it is notoriously complex and difficult
• Insurance companies all use highly sophisticated models, but the
range of prices offered is usually very large
The Problem
• How could it be improved by technology?
1. Additional data
• Reduce asymmetric information à less moral hazard and adverse selection
2. Improved data analysis
• Price risk more accurately à outcompete rival firms
3. Improved claim prevention
• Lower prices for customers that implement technologies that protect against
incidents that lead to claims
InsurTech – Additional Data
• What are some ways in which technology could allow insurance firms
to gather additional relevant data?
InsurTech – Additional Data
• Social media data is likely relevant for many kinds of insurance
• Car insurance – devices that measure speed, distance, routes
• Health/Life insurance – wearable technology measuring e.g. exercise
levels
• Home insurance – technology that detects when a door is left
unlocked
Black Box Car Insurance
• The major example of InsurTech in car insurance is the black box
• A device that attaches to a car, records driving behaviour, and sends it
to the insurance company
• This data is used to inform future insurance pricing
• Generally only used on an opt-in basis, but can lead to cheaper
insurance
Black Box Car Insurance
• Often doubles as a theft recovery device – so also provides some
claim prevention
• Has also been used to aid the detection of fraudulent claims, reducing
premiums for legitimate customers
“…the black box showed one of the cars involved had been parked for
20 minutes earlier in the day outside the address of a company owned
by the other person… it also showed the initial impact had taken place
on a country lane before the car was driven to the site of the alleged
accident.”
Wearable Technologies
• The main data-gathering innovation in health insurance is the use of
wearable technologies
• Measure factors such as heart rate, exercise levels, temperature, and sleep
quality
• This could lead to more accurate health insurance pricing while providing
new insights into the factors affecting health
• But not currently widely used, largely because the area is very ethically and
politically thorny
InsurTech – Improved Analysis
• The insurance industry may also benefit from the data analysis
methods discussed in the machine learning parts of the course
• Currently most likely to identify potential relationships in large
datasets that can be examined more closely by a human
• This is because identifying the causal mechanism is particularly
important for insurance applications
• Compliance and discrimination prevention
• Goodhart’s Law
Goodhart’s Law
• When insurance companies measure something, it is likely to become
a target
• Is this a desirable outcome for the insurer? Often, yes.
• Safer driving from black boxes
• Healthier living from (reliable) wearables
• At other times, Goodhart’s Law will cause operational, legal, and
ethical problems
InsurTech – Claim Prevention
• Other InsurTech is intended to actively reduce the probability of a
claim
• Use of this technology is monitored and rewarded with lower
premiums
• One example is the use of programmable security devices in home
insurance – your insurance company will know immediately if your
alarm system fails
InsurTech
• The insurance industry has a highly competitive market structure, with lots
of firms for consumers to choose from
• Pricing uncertainty is difficult, and feedback mechanisms are noisy
• The best-case scenario for InsurTech firms might involve producing
software-as-a-service that is sold to a large number of insurance providers
• The ceiling for tech-led insurance providers is likely lower than that of
payments firms or FinTech banks
InsurTech
• Do the theoretical benefits of InsurTech innovations collide with the
business model of most UK insurance firms?
• Many insurance firms currently operate by offering a low price for the
initial year, then increasing it in subsequent years, relying on customer
inertia
• As soon as they’ve gathered the first year of data, they’re immediately
undercut by a different provider
• Adoption of new technology may be less clean in practice than theory
would suggest
Conclusion
• Technology that allows for better data gathering, processing, and analysis
has many applications in insurance
• Comes with many positive opportunities, but also a familiar set of
questions about privacy and ethics
• Highly competitive nature of insurance industry makes high-profile
InsurTech unicorns rare
• But it also provides market pressure that large banks may not face – so we
may instead see rapid widespread adoption