Spatial Risk Diffusion: Predicting risk linked to human behavior
-
Upload
accenture-insurance -
Category
Business
-
view
580 -
download
0
Transcript of Spatial Risk Diffusion: Predicting risk linked to human behavior
Spatial risk diffusion Predicting the
propagation of risk
inked to human
behaviour
Competing with digital disrupters
Traditional Insurance
Carriers:
• Cannot rely on only usual analytics methods, internal data
• Must adopt modern risk-assessment methods
Those that do not move out of their comfort zones will be left behind by their competitors
Digital Disrupters
exploit data to:
• Run efficient, agile, profitable operations
• Take marketplace leadership positions
They will use data from the digitalization of our lives to maximize their risk-assessment capabilities
The Spatial Risk Diffusion Model
Does not assess risk just as it relates to past events
Is unlike generalized linear modeling
Is a sophisticated predictive human behavioral model
Examines risk as a function of:
• Customers’ individual and collective behavior
• Linkages, influences of behavior across customer segments
The Spatial Risk Diffusion Model
At the heart of the model—the theory of risk as a social construct:
Societal norms are formed through social network interactions and influence risk-taking by individuals in the group
The Spatial Risk Diffusion Model
Scientific support: Peer-reviewed medical studies on smoking cessation, obesity by:
• Harvard Medical School • University of California – San Diego
At the heart of the model—the theory of risk as a social construct:
Societal norms are formed through social network interactions and influence risk-taking by individuals in the group
The Spatial Risk Diffusion Model
Our theory:
Applied to various behavior-driven risks, insurers
could understand the propagation of risk trends
throughout the population and predict:
• Which groups will influence
other groups’ behavior
• How long behavior will
take to spread
The Spatial Risk Diffusion Model
In partnership with researchers at
Stevens Institute of Technology
12-week research study tested:
• Theoretical foundation of the
model
• Our more expansive application
of the theory
• Predictive capabilities
of the model
The Spatial Risk Diffusion Model
The study:
Examined 12 years of NHTSA data on teen driving
Applied the model to teen driving behavior
Accurately predicted changes in driving behavior
Identified leaders, laggards in behavior among population groups
Population
data (Census)
Time series
analysis
Significantly correlated
Not significantly correlated
Pair-wise cross correlations and time lags between crash data time-series of states
Highly correlated
The Spatial Risk Diffusion Model
US
States
AL AK AZ AR CA CO CT DE FL GA HI Etc
AL 0 -3 0 0 0 0 0 0 0 0 0
AK 3 0 6 2 2 3 1 4 3 1 0
AZ 0 -6 0 0 0 -2 0 2 0 -1 -2
AR 0 -2 0 0 0 0 0 -2 -1 -1 -2
CA 0 -2 0 0 0 0 0 0 0 0 -2
CO 0 -3 2 0 0 0 0 1 0 0 0
CT 0 -1 0 0 0 0 0 0 0 0 0
DE 0 -4 2 2 2 0 -1 0 0 1 0
FL 0 -3 0 1 0 0 0 0 0 0 0
GA 0 -1 1 1 0 0 0 -1 0 0 0 Arizona lags Colorado by two years
HI 0 0 2 2 2 0 0 0 0 0 0 (Correlation 0.727)
Etc Colour represents correlation
Number shows lag time in years
(2000-2012, Age 15-19)
Fatal crashes
(NHSTA)
The Spatial Risk Diffusion Model
Significant potential impact on insurers’
financial performance
Teen risk attitudes becoming more conservative by
about 4% annually
Insurers could improve loss ratios by 60 to 100 basis points by
accounting for risk diffusion effects gleaned from state-level data
Loss-savings opportunity could be larger as analyses refined to
more granular zip code or county level
For example, in U.S. auto market:
The Spatial Risk Diffusion Model
Insurable behavior-driven risks that could
be better assessed include:
Workers compensation
Automobile
Burglary
Employers liability
Health
The Spatial Risk Diffusion Model
Loss reserving
Marketing
Public policy campaigns
Other areas the model can help:
Build behavioral-assessment capabilities needed to apply
spatial risk diffusion concept within regulatory framework:
The Spatial Risk Diffusion Model
Identify data points of most value, with highest predictive potential
Incentivize customers to share personal data to gain behavioral insights
Consider partnering opportunities
Conclusion
The Spatial Risk Diffusion Model would help insurers gain a competitive
advantage in risk assessment by allowing them to identify the population
groups whose behavior influences other groups. It also would enable insurers
to track how long that behavior takes to propagate throughout society.
They would become the industry’s
disrupters, rather than the disrupted.
With this advanced understanding of risk
diffusion, insurers would improve their
positioning against competitors—including
the industry’s digital disrupters—while
growing their business.
For more information and to
download the full report visit:
www.accenture.com/us-en/insight-spatial-risk-diffusion