Predictive Modeling:
Rules of Thumb for
Communicators
Predictive Modeling Seminar
Insurance Marketing Communications Association
Chicago, IL
September 18, 2007
Download at
http://www.iii.org/media/presentations/predictivemodeling/
Robert P. Hartwig, Ph.D., CPCU, President
Insurance Information Institute  110 William Street  New York, NY 10038
Tel: (212) 346-5520  Fax: (212) 732-1916  bobh@iii.org  www.iii.org
PREDICTIVE
MODELING:
The Basics
Predictive Modeling:
Communications Challenges
• Predictive Modeling Can Be Complex
 Actuaries/Economists use a variety of statistical techniques
 Understanding how they work requires formal statistical training
 Underwriters apply them, usually as part of an already sophisticated and
automated underwriting process
•
•
•
•
•
Use of Some Predictive Factors/Models May Not be Intuitive
Usage Often Not Explained or Even Revealed to Communicators
Benefits Not Well Articulated to Communicators or Customers
Failure to Recognize & Enlist Agents as Communicators
Communications Obstacles in the Regulatory Context
 Regulators may have difficulty understanding
 Tendency is to react negatively
 May seize on issue for political gain
• Models Maximize for Statistical Accuracy
 Some May Feel Models Are Too Impersonal
 Invasion of Privacy Concerns?
Predictive Modeling:
What is It?
• What is Predictive Modeling?
 While people (even within the insurance industry) tend to view it as new, it
is in fact quite old—as old as insurance itself.
 DEFINITION: Predictive modeling is a process used to create a statistical
model of future behavior. In insurance, predictive models are primarily
concerned with forecasting probabilities, trends and relativities.*
 A predictive model is made up of a number of predictors, variable factors
that are likely to influence future behavior or results.
 In auto insurance, for example, a customer's gender, driving experience,
type of vehicle, driving record, miles driven, etc., help predict the
likelihood and cost of future claims. To create a predictive model, data is
collected for the relevant predictors, a statistical model is formulated,
predictions are made and the model is validated (or revised) as additional
data becomes available. The models may employ a simple or extremely
complex and employ a wide variety of statistical techniques.
• Use of Some Predictive Factors/Models May Not be Intuitive
*Adapted and modified by the Insurance Information Institute from www.searchdatamanagement.com
accessed Sept. 16, 2007.
Predictive Modeling: Why Do We
Hear So Much About it Today?
• Insurers rewrote their entire auto and homeowners book of
business beginning in the later 1990s/early 2000s in response to
huge losses in both of these key lines (which together account for
nearly 50% of industry premiums)
• This re-underwriting process was effectively a re-evaluation of
risk presented by each policyholder and the adequacy of the
premium paid by the policyholder to transfer that risk.
• In most cases the premium was inadequate and premiums rose
• Re-underwriting process included the use of sophisticated new
models designed to better match price with risk
• By definition, these models included more and better rating
factors as well as new statistical methodologies for gauging
interactions between these factors.
• Policyholders and regulators incorrectly associated new factors in
the models as being solely responsible for the increase
• Credit-based “Insurance Scores” are the best known example
Private Passenger Auto (PPA)
Combined Ratio
PPA is the profit
juggernaut of the p/c
insurance industry today
110
105
107.9
104.2
103.5
101.7 101.3101.3
101.1
101.0
Auto insurers have
shown significant
improvement in
PPA after reunderwriting entire
book of business in
early 2000s
99.5
100
95
109.5
98.4
Average Combined
Ratio for 1993 to 2005:
101.0
94.3
95.1 95.5
90
93
94
Sources: A.M. Best; III
95
96
97
98
99
00
01
02
03
04
05
06
Predictive Modeling: Why Now?
• Predictive modeling is not new—big issue in most industries
• Some form of it has been around since the earliest days of
insurance—used in personal and commercial lines
• In recent years the cost of data storage and acquisition have
declined as has the cost of computing power
• More data is available to insurers today at lower cost
• Powerful computers make analysis (mining) of the this data
easier, faster and more fruitful
• Public and regulators have pushed for more individualized rates
(and less reliance on factors like territory)
• Insurers responded by accelerating trend toward individual risk
ratingsmaller pools of increasingly homogeneous individuals
• Consequently, rating systems becoming fairer & more accurate
• Implies that subsidies are being removed from system
• Recipients of subsidies don’t like their removal nor do regulators
who view insurance as an extension of the social welfare system
Insurance Scores:
The Perfect Example of a
Communications Breakdown
• Insurers began to implement use of credit-based insurance score
in the early/mid-1990s, but not on a large scale until late 1990s
very early 2000s.
• Insurers had found that scores were among the most accurate of
all rating factors for predicting future loss.
• Roll-out and use of credit was not communicated to most key
personnel who come in contact with customers, regulators or
media
• Why credit works was not intuitive for most people (e.g., what
does credit information have to do with my driving ability?)
• Agents dislike having to explain why premiums rose due to credit
factors
• Special cases warranted special treatment abounded: No credit,
life-changing events, identity theft
• Consumer protections formalized only later (e.g., NCOIL)
• Race issue became (and remains) big (but is red herring)
PREDICTIVE
MODELING:
JUST PART OF THE
RATEMAKING &
UNDERWRITING PROCESS
Predictive Data Can Be Historical,
Class or Individual Specific
• Historical Information: Used to identify trends in data
 Actuaries use a variety of statistical techniques; get base rate
• Class Rating
 Data are adjusted for geographic, industry-specific factors or other factors
statistically correlated with risk of future loss
 E.g. Urban zip codes = greater accident frequency
 E.g. Occupation in workers comp
• Individual Risk Rating
 Policyholder-specific risk factors are taken into account
 E.g., Model of car; wood frame vs. masonry home; office vs. construction
worker
 Credit profile
 “Black box” data;
 FUTURE: GPS Tracking (on voluntary basis)
• Experience Rating
 Adjustments made to premium based on policyholder’s past claim filing
activity
UNDERWRITING:
Key to Accurate Risk
Assessments & Rates
What is Underwriting?
• Underwriting
Process by which insurer determines whether policy
should be issued and on what terms
• Complex Process
Many market and individual factors considered
All relate to riskiness/likelihood of loss
• Insurers All Use Underwriting Guidelines
 Helps keep insurers focused, disciplined, profitable, solvent
 E.g., no writing risks within 5 miles of coast, no high-rise
construction risks, no limits above $1 million, no sportscars
• Underwriting Tools
Objective is to improve accuracy of loss forecasts
Creates a more fair, equitable rating system for all
Premium is more closely associated with risk
RATING FACTORS
Helping to Match
Premium Charged to
Risk Assumed
Categories of Typical Auto
Insurance Rating Factors/Criteria
• Vehicle Type Factors
• Use of Vehicle Factors
• Location (Territorial) Factors
• Driving History
• Prior Insurance
• Personal Factors
• Other
Typical Auto Insurance
Rating Criteria
• Vehicle Type Factors
Number of vehicles to be insured on policy
Number of operators in household
Make, model & body style of each vehicle
Age of vehicle (model year)
Safety features (e.g., airbags, anti-lock brakes)
Anti-theft devices
• Use of Vehicle Factors
Distance driven annually
Commuting distance
Number of days per week used to commute
Who drives vehicle the most?
Years of driving experience (youthful operator?)
Use of vehicle for business purposes
Typical Auto Insurance
Rating Criteria
• Location (Territorial) Factors
Location where vehicle is kept
Garage or street parking
• Driving History
Accidents
Moving violations
Convictions (e.g., DUIs)
Personal claims history
• Prior Insurance Factors
Currently insured?
Number of years with current insurer?
Current Bodily Injury limits
Typical Auto Insurance
Rating Criteria
• Driving History
Accidents
Moving violations
Convictions (e.g., DUIs)
Personal claims history
• Prior Insurance Factors
Currently insured?
Number of years with current insurer?
Current Bodily Injury limits
Typical Auto Insurance
Rating Criteria
• Personal Factors
Marital Status
Gender
Occupation
Education
Student?
Homeowner?
• Other Factors
Information from credit reports
Drivers education, defensive driving course taken
Examples of
Relationships Between
Underwriting Criteria
& Losses
Example 1:
GENDER & AUTO
INSURANCE
Sex of Drivers Involved in All Auto
Crashes, 1994-2003
20
Millions of Accidents
Males are involved
Female
in 50% more
Millions of Accidents
18
18.6
Male
accidents on average
16
15.2
14.3
14
12.7
12.4
12.7
11.4
12
10.6
9.9
9.6
8.6
7.6
7.0
11.6
10.6
10
8
12.1
7.5
8.6
8.4
7.4
6
94
95
96
97
98
99
00
01
02
Source: National Safety Council; Insurance Information Institute 2005 Fact Book, p. 109.
03
Fatality Rate by Sex of Drivers
Involved in Auto Crashes, 1994-2003
Fatalities per Billion Miles Driven
30
Fatalities per Billion Miles Driven
27
25
25
27
27
25
24
20
17
17
17
15
Male
22
22
24
18
16
16
14
15
12
10
Female
13
13
Males are involved
in 61% more likely
to be killed in an
auto accident
5
94
95
96
97
98
99
00
01
02
Source: National Safety Council; Insurance Information Institute 2005 Fact Book, p. 109.
03
Example 2:
DRIVER AGE
2.9%
6.5%
3.6%
Teens are by far the most likely
to be involved in accident than
the elderly (but elderly more
likely to die in crash)
8.2%
4.8%
10%
Teens account for
just 5% of drivers
but 22% of
accidents! But
people 35-44
represent 21% of
drivers but just 16%
of accidents
7.0%
13.3%
12.5%
8.4%
15%
5%
Share of Accidents
20.7%
20.8%
15.8%
20%
17.3%
17.8%
25%
Percent of Total Drivers
18.3%
22.1%
Accidents by Age of Driver, 2003
0%
Under
20
20-24
25-34
35-44
45-54
55-64
65-74
75+
Source: National Safety Council; Insurance Information Institute
Example 3:
INSURANCE
SCORING (CREDIT)
Importance of Rating Factors
by Coverage Type
Coverage
Factor 1
Factor 2
Factor 3
BI Liability
Age/Gender Ins. Score
Geography
PD Liability
Age/Gender Ins. Score
Geography
PIP
Ins. Score
Geography
Yrs. Insured
Med Pay
Ins. Score
Limit
Age/Gender
Comprehensive
Model Year
Age/Gender
Ins. Score
Collision
Model Year
Age/Gender
Ins. Score
Source: The Relationship of Credit-Based Insurance Scores to Private Passenger Automobile Insurance
Loss Propensity Michael Miller, FCAS and Richard Smith, FCAS (EPIC Actuaries), June 2003
(Presented at June 2003 NAIC meeting).
Texas Auto: Average Loss per Policy
(by Credit Score Decile, Total Market)
Average Loss = $695
Avg. Incurred Loss per Policy
$1,000
Interpretation:
$918
$900
$846
$791
$800
Those with poorest credit scores
generated incurred losses 65% higher
those with the best scores
$707 $703
$700 $668
$681
$631
1st Decile = Lowest Credit Scores
$600
$584
10th Decile = Highest Credit Scores.
$568 $558
$500
No
Score
1st
2nd
3rd
4th
5th
Score Range
Source: University of Texas, Bureau of Business Research, March 2003.
6th
7th
8th
9th
10th
Indicated Relative Pure Premium
by Insurance Score (PD Liability)*
Interpretation:
0.4
33%
Relative Pure Premium
0.3
18%
0.2
0.1
10%
9%
Those with poorest credit scores had loss
experience 33% above average while
those with the best scores had loss
experience that was 19% below average
3%
0%
0.0
-0.1
-7%
-11%
-0.2
-14% -15%
-19%
-0.3
No
Hit/Thin
File
Source: EPIC Actuaries, June 2003
607
659
693
722
748
Score Range
774
802
837
894
997
Example: Credit Discount Can
Save $100s per Year*
•Credit discount
lowered annual
premium by 14.7%
Safety/Anti- Total Annual Savings from Discounts: $820
Theft
CreditDiscount
Related
•Policyholder saved
19%
nearly $300
Discount
$154
36%
•Credit was single
$296
largest discount
•Opponents of
credit will force
people to pay more
for coverage
$196
Good Driver
Discount
24%
*Annualized savings based on semi-annual data from example
Source: Insurance Information Institute
$174
Multipolicy
Discount
21%
Example 4:
WORKER AGE
(A Workers Comp Example)
THE AGEING
WORKFORCE
Age Could be Used a
Predictor of
Occupational Injury and
Loss, But it is Not
U.S. Workforce is Aging: Significant
Implications for Workers Comp
Median Age of U.S. Worker
42
40.6 40.7
40.5
39.0
40
Older and less
healthy workforce
38
34
32
38.0
36.6
35.8
36
39.4
35.2
34.3
The median age of US workers as the Baby Boomer begin
to retire is about 41 years. Immigration will hold this
number down and may even lower the figure.
30
1962
1970
1975
1980
1985
Year
Source: US Bureau of Labor Statistics, 2004.
1990
1995
2000
2005
2008
Fatal Work Injury Rates
Climb Sharply With Age
Fatal Work Injuries per 100,000 Workers (2006)
8
Fatality rates for workers 65 and older
are triple that of workers age 35-44. The
workplace of the future will have to be
completely redesigned to accommodate
the surge in older workers.
6
4.9
12
10
4
2
2.7
2.7
3.2
3.6
10.8
4.0
Age is not used as a an underwriting
factor in WC—should it be?
0.8
0
16-17
18-19
20-24
25-34
35-44
45-54
55-64
Source: US Bureau of Labor Statistics, US Department of Labor; Insurance Information Institute.
65+
Example 5:
WORKER WEIGHT
(Another Example
Relevant to Workers
Comp that is Not Used)
THE OBESITY
EPIDEMIC
Major Cost Driver that
WC Has Yet to Address
200
180
160
140
120
100
80
60
40
20
0
The most obese workers file twice as
many WC claims and 13 times more lost
workdays than healthy weight workers
10.80
183.63
11.65
8.81
5.80
10
8
75.21
6
60.17
40.97
14.19
12
117.61
7.05
5.53
14
Obesity is not a rating
factor, but it is an
identifiable cost factor
4
2
0
BMI <18.5
(Underweight)
18.5-24.9
(Healthy
Weight)
25-29.9
30-34.9 (Obese 35-39.9 (Obese
(Overweight)
Class I)
Class II)
Lost Workdays
40+ (Obese
Class III)
Claims
Source: Ostbye, T., et al, “Obesity and Workers Compensation,” J. of the American Medical Association, April 23, 2007.
Claims per 100 FTEs
Lost Workdays per 100 FTEs
WC Claims and Lost Workdays
by Body Mass Index (BMI)
60,000
10,000
$7,503
$5,396
20,000
$7,109
$3,924
30,000
BMI <18.5
(Underweight)
18.5-24.9
(Healthy
Weight)
$13,338
$13,569
40,000
$19,661
$23,633
50,000
$51,091
$59,178
70,000
Med claims costs are 6.8 times
higher for the most obese
workers and indemnity costs
are 11 times higher
$23,373
$34,293
Medical & Indemnity WC
Claims Costs by BMI
0
25-29.9
30-34.9 (Obese 35-39.9 (Obese
(Overweight)
Class I)
Class II)
Medical Claims Costs
40+ (Obese
Class II)
Indemnity Claims Costs
Source: Ostbye, T., et al, “Obesity and Workers Compensation,” J. of the American Medical Association, April 23, 2007.
Example 6:
TERRITORY
Baltimore Relativity to
State Loss Cost, 2001-2003
3.5
3.0
BI Liability costs in Baltimore are
more than double (2.11 times) the
state overall (i.e., 111% higher)
2.5
2.11
2.0
PD Liability costs in
Baltimore are 47% higher
than the state overall
3.00
PIP costs in Baltimore are
triple the the state overall
(200% higher)
1.47
1.5
1.0
0.5
0.0
Bodily Injury Liability
*ISO territories 33, 35, 36 and 39.
Source: ISO.
Property Damage
Liability
Personal Injury
Protection
Baltimore Relativity to
State Loss Cost, 1988
3.0
2.5
2.37
2.0
Costs in Baltimore were
well above average back in
1988 too—still are today
and will be in the future.
This is permanent feature
of most major urban auto
insurance markets
2.52
1.47
1.5
1.0
0.5
0.0
Bodily Injury Liability
*ISO territories 33, 35, 36 and 39.
Source: ISO.
Property Damage
Liability
Personal Injury
Protection
Are There Limits to What Predictive
Modeling Can or Should Do?
• Predictive Modeling Increases Accuracy, Equity
in Rates
Incumbent on insurers to use this information subject to
limits imposed by policymakers
• Advances in Data Storage, Retrieval, Computing
Will Advance the Frontier of Predictive Models
• Concern that Individual Risk Rating Will
Replace Risk Pooling is Absurd
No model will ever be 100% accurate
Some degree of pooling will always exist
• Societal Boundaries Will Always Exist
Predictive modeling will never be used to its full
potential
Privacy/”Big Brother” concerns
Predictive Modeling: 6 Rules of
Thumb for Communicators
1. EDUCATE: Educate Yourself to Develop Understanding of
How Products Work
 Get to know actuaries and underwriters in your company
2. PARTICIPATE: Get Communications (not just Marketing)
Involved at a Much Earlier Stage of Product Cycle
3. ANTICIPATE: Potential Communications Challenges
Before Rollout
4. IDENTIFY: Subject Area Experts as Technical Resources
5. DISSEMINATE: Create Plan to Help Employees with
Customer, Regulator & Media Contact Understand How
Product Operates
6. COORDINATE: Ensure Marketing, Government Affairs,
Customer Service, Agents all Operating from Same
Playbook
Insurance Information
Institute On-Line
If you would like a copy of this presentation, please give me
your business card with e-mail address, or dowload at:
http://www.iii.org/media/presentations/predictivemodeling/