unlined roads, and dirty sensors—is extremely difficult and therefore unlikely to
be in use outside major cities before 2025,
or in the majority of cars before 2035.
The first key concept is that this slow
spread of autonomous cars is unlikely
to slow the demise of auto-insurance
profitability. This is because cars with a
range of current technologies (80 percent
cars), such as adaptive cruise control and
predictive braking, have been shown to
have crash rates 20 to 30 percent of fully
manual cars. The introduction of 80
percent cars is likely to have a substantial
impact on crash rates than the introduction of 100 percent cars, and 80 percent
technology is impacting on premiums for
these high-end cars now, leading to the
start of a decline in insurer cash-flow.
The industry, however, argues that
even this threat is minor as these 80
percent technologies will only be embedded in middle-end cars starting by 2021
and low-end cars starting in 2025. This
implies manual cars (non-80 percent) will
still be the majority of cars sold in 2023.
Some commentators argue telematics will
not be widely introduced because, based
on current black-box telematics, they are
too expensive for all but the highest risk
clients to install.
This ignores the fact cars increasingly
are being manufactured with telematics
pre-installed, and mass production is
causing the cost to drop sharply. When
Google first built its autonomous car,
the LIDAR system cost more than
$70,000. Today that same hardware can
be purchased for $250, and is expected to
fall to $90 soon, making it affordable for
even lower-end cars. In 2017, 83 million
cars with smart functions will be sold.
They are especially useful for commercial
trucking or taxi fleets, where they aid
It needs to be noted the inclusion of
Profitability vs. Premiums
telematics has increased the cost of cars, es-
pecially exposed components such as wing
mirrors or windscreens. The instillation
of these has become more complex and
expensive. This has led to a temporary rise
in crash payouts. However, by about 2020,
the rate of crashes will decline faster than
the component cost will rise.
The second key concept is the difference between the rate of spread of autonomous cars
and the decline of auto-insurance profitability. What needs to be clearly understood is
the demise of the car insurance industry is
not about percentages of cars with X or Y
type of technology, nor is it really about declines in premiums flows. Demise dynamics
depends instead on profitability. The lesson
from past episodes of disruptive innovation
is the point when premium cash flow starts
to decline by five or 10 percent per year, the
natural reaction from insurers will be to
cut premiums to retain market share. This
quickly leads to ‘destructive competition,’
where in order to increase market share to
hold premium income constant, companies
compete until profit margins turn negative.
Most incumbents have cost structures with
large distribution costs so they have little
flexibility in reducing per sale costs. This
means insurance profits disappear when
manual car numbers and premium flows
are at 70 to 80 percent of previous levels.
This implies decreasing auto insurance
profitability from about 2020 on.
In addition, since the richest customers will be the owners of the 80 percent
cars, an increasing proportion of premiums will come from insuring the lower
income owners of the remaining manual
cars. Insurers will thus have to compete
for a shrinking pool of decreasing quality
clients, or those few who choose to manually drive as a hobby. The only firms with
an incentive to stay in this type of market
are specialty insurers.
An additional trend in large cities will
be the use of car-sharing apps. Already
car ownership rates have peaked in key
demographic urban categories. The net
impact of this is still unknown, but since
profit margins on fleet insurance deals are
low, the impact could be noticeable.
The conclusion is the turning point
for car insurers does not occur when 100
percent autonomous cars arrive on our
roads in substantial numbers—which
is a decade away—but when a sufficient
number of 80 percent autonomous cars
exist to substantially cut crash rates, and
thus cut premiums and insurer cash flow.
Some commentators will argue, correctly,
car insurance will still exist for decades.
I’m sure it will; people will still continue
to drive manual cars for fun. My response
is horse insurance still exists, but I’m not
sure current insurers visualize their future
in that way.
The third key concept is the crash rate of
an individual car does not just depend
on its own level of technology; it also
depends on the technology of other cars.
The 80 percent cars not only avoid crashing themselves, they detect and avoid
straying manual cars. In a networked
environment, manually driven cars can
be identified well before the auto-car gets
to them and they will be given a wide
margin. Therefore, as the proportion of
80 percent cars rises, it becomes harder
for even erratically manual cars to crash.
This is not a linear process. Models
derived from diseases and financial crises
show there is a “tipping point” where the environment changes rapidly from one where
one type of behavior dominates to one
where another type of behavior dominates.
These models imply there will be a tipping
point when about 45 to 55 percent of current cars have 80 percent technology, which
will occur around 2023 to 2026. At that
point, cars with 80 percent or better crash
avoidance technology will only crash due
to software bugs or hacking, poor maintenance, roadside objects, or due to manual
drivers. Within an exponential change, a
forecast based on a linear projection of past
trends is dangerous, as insurers are likely to
be surprised and thus wrong-footed.
Electric and Automatic-drive
A related issue relating to turnover is
the rapidly decreasing drop in the cost