Big Data
an actuarial perspective
Decavi-KPMG,18/11/2015
MateuszMaj
Chairmanof IABE
BigDataWG
mat@motosmarty.com
What is Big Data?
Internet of Things
What is Big Data?
What is Big Data?
What is Big Data?
source:www.hp.com
Why Big Data
Working Group?
Discuss:
•  Impact of Big Data on insurance sector and the actuarial
profession;
•  Present challenges and good practices when working with Big
Data;
•  Educate actuarial profession about Big Data through CPD
courses
Insurance value chain:
undewriting
Covers different
Underwriting
360degree
customerview	
  
Combine different sources and apply analytics to
create comprehensive customer view and:
•  Maximize profitability of the current portfolio
•  Detect cross-sell and up-sell opportunities;
•  Increase customer satisfaction and loyalty;
•  Acquire new profitable customers and reduce marketing costs.
Underwriting
Underwriting
Tescogroup–UK
Motor – 1M
Pet – 0.45M
Travel - 0.175M
Life – 0.175M
Home – 0.4M	
  
•  Insurance prevention program with discounts and rewards for good
driving
•  ‘Phased’ approach:
•  Phase 1: combine data from different sources i.e. traditional channels, online channels,
external service providers, Tesco group warehouses;
•  Phase 2: Identify the right customers within Tesco network;
•  Phase 3: Provide initial offer and reward drivers with initial rewards from Tesco group;
•  Phase 4: Iterate and provide personalized insurance offers.
Underwriting
makinginsurancesexy
Pricing
Pricing
Ratingtrends
1980s Now
Profession
Engine power
Coverage
Bonus-malus
Coverage
Bonus-malus
Claims history
Traffic violation
history
Age of vehicle
Use of vehicle
Make of vehicle
Purchase price
Parking place
Occupation
No. of drivers
Age of drivers
Maritial status
Real estate
Driving license
Mileage
Registered owner
Credit rating
…
Do we need additional factors? Is telematics necessary?
Univariate
basis
Risk
modelling
Technical
premium
modelling
Scenario
testing
Price
optimisation
Extra data
sources
Telematics
data?
Pricing
Ratingtrends
•  New rating factors;
•  Flexible, dynamic risk pricing;
•  New modelling techniques like machine learning;
•  New, disruptive insurance offerings like Usage-
Based Insurance.
Pricing
Pricing
Usage-BasedInsurance(UBI)
UBI is the scheme where insurance premiums are
calculated based on dynamic causal data, including
actual usage and riskier driving behavior.
Pricing
Usage-BasedInsurance(UBI)
Full	
  UBI	
  –	
  niche	
  market	
  segments	
  
	
  
Motor	
  insurance	
  telema7cs	
  –	
  mass	
  market	
  
Works	
  well	
  with	
  niche	
  segments:	
  
•  Young	
  drivers	
  
•  People	
  living	
  in	
  risky	
  regions	
  
•  Low-­‐mileage	
  drivers	
  
Insure	
  the	
  Box,	
  UK	
  
New	
  disrup=ve	
  model	
  for	
  low-­‐mileage	
  drivers	
  with	
  prepaid	
  
miles	
  (similar	
  to	
  prepaid	
  mobile.	
  No	
  punishment	
  for	
  bad	
  
driving.	
  	
  Self-­‐selec=on	
  +	
  individualized	
  premiums.	
  
Progressive,	
  US	
  
“Prince	
  of	
  pricing”	
  with	
  MyRate	
  (pricing	
  variable	
  innova=on)	
  
&	
  Snapshot	
  (1st	
  telema=cs-­‐based	
  insurance)	
  offers	
  
Self	
  selec=on	
  &	
  individualized	
  insurance	
  premiums	
  
Intesa	
  SanPaulo	
  Assicura,	
  IT	
  
Telema=cs	
  play	
  key	
  posi=on	
  –	
  car,	
  bike	
  home	
  
Created	
  viable	
  risk-­‐based	
  pricing	
  model	
  with	
  phased	
  
approach	
  for	
  Viaggia	
  con	
  me:	
  	
  
Phase	
  1	
  (tradi=onal	
  variables),	
  Phase	
  2	
  (Banking	
  parameters),	
  
Phase	
  3	
  (Purchase	
  and	
  use	
  variables),	
  Phase	
  4	
  (Behaviour	
  
variables)	
  
Insurance value chain:
undewriting
Covers different
Claims management &
Fraud detection
Insurers loose 5% of the annual revenue due to fraud
Coalition Against Insurance Fraud (US) in the 2014 report has
stresses that technology & Big Data plays a growing role in
fighting fraud
Claims management
Examples-UBI
From high to low loss ratios
UnipolSai	
  -­‐	
  IT CoverBox	
  &	
  Carrot	
  -­‐	
  UK
Telema=cs	
  champion	
  (2.2M	
  ac=ve	
  boxes)	
  
Best	
  prac=ce	
  claims	
  management	
  incl.:	
  
•  FNOL	
  -­‐	
  quick	
  accident	
  response	
  
•  Vehicle	
  loca=on	
  in	
  case	
  of	
  of	
  theW	
  
•  Accident	
  reconstruc=on
Further improvement of the operational
efficiency including:
•  Crash data combined with video footage to fight
fraud
•  Better prediction methods to reduce claims duration
and cost i.e. no need for expert, efficient accident
reconstruction
•  Prove innocence
Covers different
Legislation
EU-widelaw
underconstruction
	
  
Insurance
Canitbesexy?
Health insurance, US
Oscar is a health insurance company that employs technology, design,
and data to humanize health care.
Technology as differentiator (telemedicine) - system connects doctors and
patients with help of mobile app, website, bills, free fitness-tracking devices – easy
to use and understand.
Insurance
Canitbesexy?
Home Insurance, IT
Habit@t is the first Connected Home Insurance by Cardiff;
Well targeted customers – young, tech-savvy, from big towns;
IoT based - combining home telematics with home insurance;
Superior customer relationship - user-friendly & seamless usage via
smartphone app, multichannel with strong support;
Proactive alerts and risk managements.
EU-widelaw
underconstruction
	
  
Role of actuaries
Datascientists	
  
MateuszMaj
Chairmanof IABE
BigDataWG
mat@motosmarty.com
Q&A	
  

Innovation and Big Data in Insurance

  • 1.
    Big Data an actuarialperspective Decavi-KPMG,18/11/2015 MateuszMaj Chairmanof IABE BigDataWG mat@motosmarty.com
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
    What is BigData? source:www.hp.com
  • 7.
    Why Big Data WorkingGroup? Discuss: •  Impact of Big Data on insurance sector and the actuarial profession; •  Present challenges and good practices when working with Big Data; •  Educate actuarial profession about Big Data through CPD courses
  • 8.
    Insurance value chain: undewriting Coversdifferent Underwriting 360degree customerview  
  • 9.
    Combine different sourcesand apply analytics to create comprehensive customer view and: •  Maximize profitability of the current portfolio •  Detect cross-sell and up-sell opportunities; •  Increase customer satisfaction and loyalty; •  Acquire new profitable customers and reduce marketing costs. Underwriting
  • 10.
    Underwriting Tescogroup–UK Motor – 1M Pet– 0.45M Travel - 0.175M Life – 0.175M Home – 0.4M  
  • 11.
    •  Insurance preventionprogram with discounts and rewards for good driving •  ‘Phased’ approach: •  Phase 1: combine data from different sources i.e. traditional channels, online channels, external service providers, Tesco group warehouses; •  Phase 2: Identify the right customers within Tesco network; •  Phase 3: Provide initial offer and reward drivers with initial rewards from Tesco group; •  Phase 4: Iterate and provide personalized insurance offers. Underwriting makinginsurancesexy
  • 12.
  • 13.
    Pricing Ratingtrends 1980s Now Profession Engine power Coverage Bonus-malus Coverage Bonus-malus Claimshistory Traffic violation history Age of vehicle Use of vehicle Make of vehicle Purchase price Parking place Occupation No. of drivers Age of drivers Maritial status Real estate Driving license Mileage Registered owner Credit rating … Do we need additional factors? Is telematics necessary?
  • 14.
  • 15.
    •  New ratingfactors; •  Flexible, dynamic risk pricing; •  New modelling techniques like machine learning; •  New, disruptive insurance offerings like Usage- Based Insurance. Pricing
  • 16.
    Pricing Usage-BasedInsurance(UBI) UBI is thescheme where insurance premiums are calculated based on dynamic causal data, including actual usage and riskier driving behavior.
  • 17.
    Pricing Usage-BasedInsurance(UBI) Full  UBI  –  niche  market  segments     Motor  insurance  telema7cs  –  mass  market   Works  well  with  niche  segments:   •  Young  drivers   •  People  living  in  risky  regions   •  Low-­‐mileage  drivers   Insure  the  Box,  UK   New  disrup=ve  model  for  low-­‐mileage  drivers  with  prepaid   miles  (similar  to  prepaid  mobile.  No  punishment  for  bad   driving.    Self-­‐selec=on  +  individualized  premiums.   Progressive,  US   “Prince  of  pricing”  with  MyRate  (pricing  variable  innova=on)   &  Snapshot  (1st  telema=cs-­‐based  insurance)  offers   Self  selec=on  &  individualized  insurance  premiums   Intesa  SanPaulo  Assicura,  IT   Telema=cs  play  key  posi=on  –  car,  bike  home   Created  viable  risk-­‐based  pricing  model  with  phased   approach  for  Viaggia  con  me:     Phase  1  (tradi=onal  variables),  Phase  2  (Banking  parameters),   Phase  3  (Purchase  and  use  variables),  Phase  4  (Behaviour   variables)  
  • 18.
    Insurance value chain: undewriting Coversdifferent Claims management & Fraud detection Insurers loose 5% of the annual revenue due to fraud Coalition Against Insurance Fraud (US) in the 2014 report has stresses that technology & Big Data plays a growing role in fighting fraud
  • 19.
    Claims management Examples-UBI From highto low loss ratios UnipolSai  -­‐  IT CoverBox  &  Carrot  -­‐  UK Telema=cs  champion  (2.2M  ac=ve  boxes)   Best  prac=ce  claims  management  incl.:   •  FNOL  -­‐  quick  accident  response   •  Vehicle  loca=on  in  case  of  of  theW   •  Accident  reconstruc=on Further improvement of the operational efficiency including: •  Crash data combined with video footage to fight fraud •  Better prediction methods to reduce claims duration and cost i.e. no need for expert, efficient accident reconstruction •  Prove innocence
  • 20.
  • 21.
    Insurance Canitbesexy? Health insurance, US Oscaris a health insurance company that employs technology, design, and data to humanize health care. Technology as differentiator (telemedicine) - system connects doctors and patients with help of mobile app, website, bills, free fitness-tracking devices – easy to use and understand.
  • 22.
    Insurance Canitbesexy? Home Insurance, IT Habit@tis the first Connected Home Insurance by Cardiff; Well targeted customers – young, tech-savvy, from big towns; IoT based - combining home telematics with home insurance; Superior customer relationship - user-friendly & seamless usage via smartphone app, multichannel with strong support; Proactive alerts and risk managements.
  • 23.
    EU-widelaw underconstruction   Role ofactuaries Datascientists  
  • 24.