Session 029 PD - Tools and Data in the Underwriting Process

Moderator:
Donna Christine Megregian, FSA, MAAA

Presenters:
Derek Michael Kueker, FSA, MAAA
Kevin Pledge, FSA, FIA
Zhe (David) Zhu, FSA, FCIA, Ph.D.

SOA Antitrust Compliance Guidelines

SOA Presentation Disclaimer
2017 SOA Annual Meeting & Exhibit
DONNA MEGREGIAN, FSA, MAAA
KEVIN PLEDGE, FSA, FIA
DEREK KUEKER, FSA, MAAA
ZHE (DAVID) ZHU, FSA, FCIA, PHD
Session 29, Tools and Data in the Underwriting Process

October 16, 2017

 SOCIETY OF ACTUARIES
Antitrust Compliance Guidelines
Active participation in the Society of Actuaries is an important aspect of membership. While the positive contributions of professional societies and associations are
well-recognized and encouraged, association activities are vulnerable to close antitrust scrutiny. By their very nature, associations bring together industry competitors
and other market participants.
The United States antitrust laws aim to protect consumers by preserving the free economy and prohibiting anti-competitive business practices; they promote
competition. There are both state and federal antitrust laws, although state antitrust laws closely follow federal law. The Sherman Act, is the primary U.S. antitrust law
pertaining to association activities. The Sherman Act prohibits every contract, combination or conspiracy that places an unreasonable restraint on trade. There are,
however, some activities that are illegal under all circumstances, such as price fixing, market allocation and collusive bidding.
There is no safe harbor under the antitrust law for professional association activities. Therefore, association meeting participants should refrain from discussing any
activity that could potentially be construed as having an anti-competitive effect. Discussions relating to product or service pricing, market allocations, membership
restrictions, product standardization or other conditions on trade could arguably be perceived as a restraint on trade and may expose the SOA and its members to
antitrust enforcement procedures.
While participating in all SOA in person meetings, webinars, teleconferences or side discussions, you should avoid discussing competitively sensitive information with
competitors and follow these guidelines:
• Do not discuss prices for services or products or anything else that might affect prices
• Do not discuss what you or other entities plan to do in a particular geographic or product markets or with particular customers.
• Do not speak on behalf of the SOA or any of its committees unless specifically authorized to do so.
• Do leave a meeting where any anticompetitive pricing or market allocation discussion occurs.
• Do alert SOA staff and/or legal counsel to any concerning discussions
• Do consult with legal counsel before raising any matter or making a statement that may involve competitively sensitive information.
Adherence to these guidelines involves not only avoidance of antitrust violations, but avoidance of behavior which might be so construed. These guidelines only
provide an overview of prohibited activities. SOA legal counsel reviews meeting agenda and materials as deemed appropriate and any discussion that departs from the
formal agenda should be scrutinized carefully. Antitrust compliance is everyone’s responsibility; however, please seek legal counsel if you have any questions or
concerns.

2
 Presentation Disclaimer

Presentations are intended for educational purposes only and do not replace
independent professional judgment. Statements of fact and opinions expressed are
those of the participants individually and, unless expressly stated to the contrary,
are not the opinion or position of the Society of Actuaries, its cosponsors or its
committees. The Society of Actuaries does not endorse or approve, and assumes no
responsibility for, the content, accuracy or completeness of the information
presented. Attendees should note that the sessions are audio-recorded and may be
published in various media, including print, audio and video formats without further
notice.

3
Purpose of the session
• The search is on for how to disrupt the underwriting process
to find or select the best risks. Can data ever be as good as
blood? There is an ever growing number of tools and
resources becoming available to insurance companies to
underwrite business. The presenters will present some of the
tools, data sources, uses and concerns in the underwriting
process.
• At the conclusion of the session, attendees will be able to:
• Identify some of the tools available,
• Compare the tools and their uses; and
• Evaluate the tools for usefulness within the underwriting process.

4
What is the driving force to disrupt the
underwriting process?
1. Reach new market
2. Cost of underwriting
3. Time to complete
underwriting 50%

4. Technology advances
5. Everyone else is doing it
17%
11% 10% 13%

1 2 3 4 5
Tools available
New tools to integrate into the UW process
Available risk assessment tools / data:

Electronic
Driving
Insurance health Lifestyle / Wearables
Prescriptions Record Credit
History records Social

Rules-based
Predictive models: Automated UW
Application
1) screen
& Tele- 2) risk selection Risk Class
Interview 3) smoker Manual UW

Medical Attending Income &

traditional data: Lab Physician financial
Results Statement info
7
Marketing and lifestyle data

Area Predictor Direction

Affluence Face amount ▼
Home value ▼
Client issue age - older ▼
Education ▼
Investment assets ▼
Lifestyle Catalog stationery buyer ▼
Healthy behavior change index ▼
Survey xxxxx music ▲
Survey xxxxx vacations ▲
Survey lotteries or sweepstakes ▲
Survey xxxxx diet ▼
xxxxx pet ownership ▲
Location Counties, grouped ▼

8
 Health records
1) The APS provided the most useful
information for modeling.
2) Various regression and machine learning
techniques including random forest,
gradient boosting and support vector
machines applied to predict probability of
decline.
3) Besides probability, model
provides keywords driving
prediction along with page
where found in text.
• axillary: 12, 28, 30
• prostate: 12, 56, 57
• svc: 33
• ancillary: 39, 40, 41, 43, 44
• prostatic: 57, 61
• influenza: 62, 63, 66
• jaundice: 59
• excised: 59, 61, 62
• jaundiced: 68
9
Wearables
Info input by user Data input by device
Age Step Count Incorporate with
application, UW data,
Height Minutes of moderate/vigorous activity new risk assessment
tools for:
Weight Sleep • Screening
• Risk class assignment
Waist Circumference Resting Heart Rate • Mortality model

Blood Pressure

Family History

Behavior (stress, happiness, smoking, alcohol)

Factors listed here are illustrative, not exhaustive, and vary by device;
provided by Vivametrica

10
Online Insurance
Immediate
Price Competitive
Fully Underwritten
NO HOME VISITS OR REQUESTS FOR:
ADDITIONAL INFORMATION
ADDITIONAL TESTS
BLOOD

11
How do you get the answers you need to
underwrite?

Ask questions

12
Application Process
Modify Application
Questions

Quote Apply Add’l Q’s Offer Issue

Underwrite Evidence

Quote
and
refer

Buy

Quote Apply
and UW

13
14
Reflexive Questioning
Have you been told that
No glaucoma is due to Yes
another disease?

Term: TI
Are you awaiting an
CI: TI
operation or any
WP: TI
investigations?
ADB: TI

Decision tree rules No

Has the glaucoma been

Yes

Term: +0
CI: Remove Blindess
treated? WP: D

• Closed questions No Yes

ADB: +0

• Combination rules Term: +0

CI: Remove Blindess
WP: D
ADB: +0
Is your vision now normal
in the affected eye or
completely corrected by
spectacles or contact
lenses?

No Yes

Term: +0
CI: Remove Blindess Did you have glaucoma in
both eyes?

Behavior Science
WP: D
ADB: +0

No Yes

Digital data Predictive Analytics Term: +0

CI: +50
WP: +100
ADB: +0
Term: +0
CI: Remove Blindess
WP: D
ADB: +0

15
Cost
Traditional Sale Online Sale

Call Quote
Appointments Apply
Sale (application) External Issue
UW Data Call
Issue

16
 What tools do we use (online only)
Reflexive questions predictive analytics /
behavioral science
Traditional Non-traditional
• MIB • Aggregate
• Rx Marketing data
• MVR • Credit score data
• Criminal Record check

17
Tool Selection
How are companies vetting these tools?
• How do you know you should use it
• How to put a value on the tool
• What are the cost/benefit trade offs

19
Data Considerations
 The future will be
Internal Data Industry data-driven
Management Validation
 We have only
scratched the surface
Protective
FCRA vs. on data solutions
Value &
Non-FCRA
Exclusivity
 Do your homework –
Predictive not all data is created
Models equal
 Internal Data Industry
Management Validation

Industry Validation Protective

Value &
Exclusivity

Predictive
FCRA vs.
Non-FCRA

Models

• Demonstrates the value of historical experience (e.g.,

mortality, lapse, etc.)
Retrospective • Provides a starting point in setting future assumptions
Analysis • Should be considered necessary for the validation of a
new data source

• Caution when using alone!

Distribution • Anything can segment your business
• Takes many years to validate experience
Analysis • Can provide great insight when combined with a
retrospective study
 Internal Data Industry
Management Validation

FCRA vs. Non-FCRA Data Protective

Value &
Exclusivity

Predictive
FCRA vs.
Non-FCRA

Models

FCRA
 Fair Credit Reporting Act (FCRA) Section 604 specifies permissible
purposes for use of consumer reports to a person “which it has
reason to believe intends to use the information in connection with
the underwriting of insurance involving the consumer…”
NON-FCRA
 Generally used in lead generation/target marketing
 Must understand whether the data is for an individual or household
 Data quality may not be as accurate as FCRA-compliant data
 Internal Data Industry
Management Validation

Data from Predictive Models Protective

Value &
Exclusivity

Predictive
FCRA vs.
Non-FCRA

Models

Understand the Development

• Target variables, input data, modeling technique, etc.
• More data inputs does not always indicate a better model
• Is your data source raw data or modeled data?

Model Should be Transparent

• Score from a model should also provide drivers behind the score
• Score from a model should have meaning
• Transparency is key … How will you communicate the results?
Ask the Tough Questions!
• How much data was used to build and validate?
• Was the data related to the data to be used going forward?
• Did the model really address the problem?
• What happens if a data element is not available?
• Did your modeler understand the business?
• Did the modeler have a stake in the success of the model?
Evaluating Protective Value & Exclusivity
Internal Data Industry
Management Validation

Protective Value Protective

Value &
Exclusivity
FCRA vs.
Non-FCRA

Predictive

 Determine the value the data provides to you. The Models

value should always outweigh the cost of the data.

Exclusivity

 Data may illustrate protective value on its own, but

the protective value may diminish with current
pieces of data in practice.
Internal Data Management is Paramount
Leave no stone unturned!
Internal Data Industry
Management Validation

• Look within your company for ALL Protective

How do we effectively potential data sources

Value &
Exclusivity
FCRA vs.
Non-FCRA

manage and expand Predictive

Models

our internal data Be inquisitive

assets in order to:
 Properly leverage existing
data assets
 Maximize the value of
existing data
 NOT get left behind
• Have we captured AND optimized all of
our data?
Consider investing in data
• Capture missing historical data
• Invest in new, emerging data assets
Data Management Strategy
• Develop a corporate strategy to ensure
enterprise alignment and minimize
duplicative efforts
Alternative data-based mortality scores
Our methodology:

• credible deaths in general population datasets

Classical A/E analysis • limited credibility in insurer-specific datasets; however there is
still enough consistency across carriers

• maximizes the insight gained from retrospective studies, even

Statistical analysis using with limited claim incidence
predictive models • isolates the effect of the scores and places confidence bands
around relative mortality estimates

26
 Alternative data-based mortality scores
Our analysis suggests the scores:
• Are a new non-medical based dimension of mortality risk
• Are not intended to mimic the same UW risk class assigned using traditional
medical underwriting
• Can improve the fit of mortality predictions on top of health-based underwriting
criteria
• Provide mortality segmentation even within similar levels of affluence

Common applications include:

• Set threshold for eligibility into AUW program
• Set threshold for cases that are Declined
• Set threshold for Preferred criteria
• Risk class shifting

27
 Alternative data-based mortality scores

Worse Better Worse Better

Mortality Mortality Mortality Mortality
Score Score Score Score

Statistical techniques isolate the impact of the scores and provides confidence bands around estimates

28
 Rx (prescription drug)-based mortality score

• Milliman is a the major provider of

Rx-based mortality score
• Effectively stratifies mortality risk
across attributes
• Particularly effective in identifying
high mortality risk when little other
medical data is available

Common applications include:

• Set threshold for eligibility into AUW program
• Set threshold for cases that are Declined
• Set threshold for Preferred criteria
• Combine with other mortality scores/UW criteria

29
Where are the tools/data taking us

30
Our reputation is on the line…
Can we enhance the UW process with external
data
Added value – better decision / fewer questions
Real time
Integration with existing rules
Other integration considerations
Cost

32
Cost
Traditional Sale Online Sale

Call Quote
Appointments Apply
Sale (application) External Issue
UW Data Call
Issue

33
What do companies need to consider before
using these tools?
• How it is used
• Regulation
• Consumer friendliness
• Underwriting impact
• Agent impact
• System impact

34
Regulatory Feedback: Actions to Date
 Visited 5 states on the topic of accelerated
underwriting and new data
 Presented to the ACLI
 Presented to the NAIC Life Actuarial Task Force
~ 30 states in attendance
 Submitted TrueRisk® Life model inputs and
reason codes to California and Texas
Regulatory Feedback: Underwriting Triage
 Is underwriting triage an underwriting decision?
• Feedback generally leaned towards not having a
distinction
between an underwriting decision and triage. This is
somewhat inconsistent with the market stance.
• Interest in FCRA compliance in general
could limit the data used even within
a triage environment

As a market, we need to continue to educate using a consistent message

Regulatory Feedback: Consumer Disputes and
Data Transparency
Primary concern of every state: consumer
interactions
• Need to ensure and in some cases prove that data is not discriminatory
• Requires a seamless and transparent dispute-resolution process
o Consumer must be able to understand what the data element is, where it
came from and how to contest it

States are taking a proactive approach to

protecting the consumer
• Many have requested inputs to the TrueRisk® Life model, which we
have provided Expect all data points to be scrutinized and model inputs and
consumer reason codes/communications to be requested
Regulatory Feedback: Ongoing Expectations

Models must continue to be tested

• Expectation is that additional years will be added to studies to ensure long-term
predictive nature
• Continuous monitoring expected, especially within new accelerated processes
• Clear guidance to users on model limitations (products, ages, etc.)

If you are using it on one product but not another,

there will be questions
New non-medical data sources are to be a
component of the decision, not the sole decision
maker
Regulatory Case Study - Education
Education to verify occupation

Education to deny coverage

We must analyze all data used in underwriting

through the lens of unintended discrimination.
If our ability to use new data sources was taken
away tomorrow, it would be due to:
A. Unexpected regulations resulting
from unexpected consumer
litigations
B. Unexpected regulations resulting
from an inappropriate use of a new
data sources
37%
C. Unexpected regulations resulting 29% 30%
from our lack of transparency to
regulators 4%
D. We won’t lose access to new data
sources – stop worrying A. B. C. D.
When do you feel there will be enough tools &
data available that blood will not provide more
value to the underwriting process?
A. 1-5 years
B. 6-10 years
C. 11-20 years
33%
D. 20+ years 29%
18%
13%
E. Never 7%

A. B. C. D. E.
Facial Recognition Case Study
 from Lapetus

While traditional underwriting can take up to a month,

CHRONOS delivers more accurate results in seconds

Used by

43
44
45
46
47
48
49
50
Demo premium shown here

51
Demo premium shown here

52
What if I used a
different picture?

53
54
No smiling

55
56
57
58
59
60
61
62
63
Picture does not seem to affect the results

64
65
Indicative premium shown here

66
Thank You

Donna Megregian David Zhu

VP and Actuary Actuary
RGA Re Munich Re

Kevin Pledge Derek Kueker

CEO Actuary
Acceptiv RGA Re

67
References
• Milliman Rx Risk Score: https://www.munichre.com/site/marclife-
mobile/get/documents_E937882213/marclife/assset.marclife/Documents/Publica
tions/Milliman-Rx-Risk-Score-WP_FINAL_8.31.17.pdf
• TrueRisk® Life Score: https://www.munichre.com/site/marclife-
mobile/get/documents_E-
921937629/marclife/assset.marclife/Documents/Publications/Transunion_TrueRis
k_IA_Final_6.15.17.pdf
• MVR study: http://www.rgare.com/knowledge-center/media/research/an-
analysis-of-motor-vehicle-records-and-all-cause-mortality

68