big data and price optimization in general insurance
TRANSCRIPT
Session 171 PD, Big Data and Price Optimization in General Insurance
Moderator: Ann Weber
Presenters:
Amy Bach, J.D. Angela Nelson Mark Smith
Mary Jane Wilson‐Bilik, J.D.
SOA Antitrust Disclaimer SOA Presentation Disclaimer
Big Data and Price Optimization in General InsuranceOCTOBER 26, 2016
Panelists
2
Moderator:Ann WeberGovernment Affairs DirectorSociety of Actuaries
Panelist:Amy Bach, J.D.Executive DirectorUnited Policyholders
Panelist:Angela L. Nelson, AMCMDirector, Division of Market Regulationand Chief Industry LiaisonMissouri Dept. of Insurance
Panelist:Mark Smith, CPCU, AIS, APIAsst. Vice President National AffairsGovernment Relations DepartmentInsurance Services Office, Inc.
Panelist:Mary Jane Wilson‐Bilik, J.D.PartnerSutherland Asbill & Brennan LLP
Data mining’s impact on consumers in the context of insurance underwriting,
pricing and sales
Amy Bach, Esq., Executive DirectorUnited PolicyholdersEmail: [email protected]: www.uphelp.org
October 26. 2016Society of ActuariesLas Vegas, Nevada
United Policyholders2016 © ALL RIGHTS RESERVED
Dramatic yes, but…
• Legitimate cause for concern on many fronts
• Equal access to data seems a good approach
Unhealthy market segmentation
- Low/moderate income consumers easy to identify- Less attractive to insurers - Unfavorable pricing (CFA alleges low income
households are paying more)
How Big Data Enables Economic Harm to Consumers, Especially to Low-Income and Other Vulnerable Sectors of the Population, Nathan Newman, J.D., Federal Trade Commissionhttps://www.ftc.gov/system/files/documents/public_comments/2014/08/00015-92370.pdf)
Potential underground use of prohibited rating factors
• Ethnicity/Race• Political affiliation• Religion• Hobbies/Interests unrelated to risk of loss• Economic status
Insurance is different
• Mandatory and de facto mandatory purchase means free market competition is insufficient to protect policyholders
• Less “desirable” customers still have to buy insurance and deserve to pay a fair price
Benefits of big data skew heavily in insurers’ favor:
Access to much more detailed info about the risks they’re undertaking
Insurers can cherry pick at a granular level
Insureds don’t have equivalent new tools to compare quality of coverage/policies and performance of insurance companies
Unequal access to information
• Regulators can’t afford to buy the data
• Are smaller insurers that can’t either unfairly disadvantaged vis a vis risk distribution.
• Unfair competition?
Undermining effective regulatory review of rate filings?
• Regulators can’t penetrate to determine whether there are prohibited factors embedded in the underlying models
How much can we charge before the customer walks?
• McKinsey & Co: “Harnessing the flood of data available from customer interactionsallows companies to price appropriately—and reap the rewards.” See:
http://www.mckinsey.com/business-functions/marketing-and-sales/our-insights/using-big-data-to-make-better-pricing-decisions
?
How large a deductible can we add before the customer balks?
How much coverage can we carve out via exclusions before a regulator notices?
How can we harm our competitors with what we learn about their customers?
Price Optimization• Seems unfair when used in the insurance context – Whether
or not someone comparison shops is not a fair predictor of risk.
• Penalizes loyal customers– Who think they’re getting a benefit by being loyal and don't
know they may be overpaying– Who may have bundled to get a better deal but aren’t
getting one
State prohibitionsMaryland, Ohio, Washington, Vermont, Indiana, Colorado, California, Connecticut, Delaware, Minnesota, Montana, Missouri, Pennsylvania, and Florida, and Rhode Island and other states have restricted or banned the practice through administrative bulletins. Legislation under consideration in Illinois, Oklahoma, and Montana.
• Deem price optimization to be not a legitimate rating factor/no actuarial value as a predictor of risk.
Privacy concerns:• Pay as you drive products were tagged with privacy
concerns from the get-go.
• Consumers routinely give what they consider to be benign data to companies (e.g, social media apps) but don’t realize that it gets sold to third parties, including insurers
• Insurance is different b/c you have to buy it
The new redlining?• An increase in redlining by prohibited characteristics (race,
ethnicity, income) buried in models. Insurers always prefer clients who are wealthy and have more than one major asset to insure
• Consumers advocates are pushing for FIO data call on auto ins affordability – national data collection would be beneficial but state
regulators resisting Fed encroachment
NAIC whitepaper: http://www.naic.org/documents/committees_c_catf_related_price_optimization_white_paper.pdf
Wells Fargo – an illustration of how TMI can lead to abuses
• Data-driven targets/goals (8 accounts per customer) led to abuse of consumer personal information
The art of underwriting
• Flows with humanity as it evolves
• Impacted by social media
• Based on expertise of experts who know people and the math of probabilities too
• Very important to preservation of healthy risk transfer through insurance
Amy Bach, Esq.
• Co-founder of United Policyholders
• Career insurance consumer advocate
• Former insurance litigator, legislative staffer
• Member, Federal Advisory Committee on Insurance
• 7 term NAIC consumer representative
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
Big Data and Price Optimization:The NAIC White Paper and Beyond Society of Actuaries Annual MeetingLas Vegas, NV
Mary Jane Wilson-Bilik, PartnerOctober 26, 2016
2
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
Presenter
Mary Jane Wilson-BilikPartner, SutherlandWashington, [email protected]
3
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
Price Optimization: Regulators Respond
NAIC White Paper on Price Optimization- Adopted by the Property and Casualty Insurance (C) Committee at
the NAIC Fall Meeting in November 2015 Likely to remain controversial with advances in vehicle telematics
18 State have issued bulletins on price optimization - 12 bulletins have required insurers to revise their filings- Many have clarified what is NOT prohibited, including: Use of sophisticated data analysis Certain pricing practices applied on a group basis
Ongoing Focus on Big Data at the NAIC- Big Data (D) Working Group
Overview
4
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC Price Optimization White Paper
The advent of sophisticated data mining tools and modeling allows for more detailed quantitative info on aspects of rate setting traditionally left to judgment or anecdotal evidence PO = Process of maximizing or minimizing a business
metric using sophisticated tools and models to quantify business considerations: marketing goals, profitability and policyholder retention Ratebook; Individual and Hybrid Optimization
How does the NAIC define price optimization?
5
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC White Paper on Price Optimization
PO may improve rate stability and lower insurer’s long-term costs But voluminous rate information is difficult for regulators to distill Regulators do not have data and tools necessary for independent
evaluation Critics charge that PO is unfairly discriminatory
Insurers argue PO is a technological improvement to current practices - Also argue that criticism is aimed at “individual price optimization –
not ratebook optimization” used in setting rates
What potential benefits and problems with price optimization does the NAIC identify?
6
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC White Paper on Price Optimization
Regulators should address price optimization through state laws prohibiting rates that are “excessive, inadequate or unfairly discriminatory” Rating plans should be derived from sound actuarial
analysis and be cost-based Two customers with same risk profile should be charged
the same premium for the same coverage Capping and transitional rules may be appropriate
What policy recommendations does NAIC set forth?
7
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC White Paper on Price Optimization
Insurance rating practices that adjust the current or actuarially indicated rates should not be allowed when practice cannot be shown to be cost-based The following practices are inconsistent with statutory
requirement that rates not be unfairly discriminatory: Price elasticity of demand Propensity to shop for insurance Retention adjustments at the individual level and A policyholder’s propensity to ask questions or complain
Policy recommendations, continued:
8
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC White Paper on Price Optimization
Rating plans in which insureds are grouped into homogeneous groups should not be so granular that resulting classes have little actuarial or statistical reliability Use of sophisticated data analysis to develop finely tuned
methodologies with a multiplicity of possible rating cells is not, in and of itself, a violation of rating laws, so long as classes and factors are cost-based
Policy recommendations, continued:
9
© 2016 SUTHERLAND ASBILL & BRENNAN LLP / SUTHERLAND (EUROPE) LLP
The NAIC White Paper on Price Optimization
Issue a bulletin (18 states) Enhance requirements for rate filings
Define constraints: e.g., only apply models to specific class sizes – not so small as to be at the individual insured level or small group
Analyze models to ensure model adheres to state law and actuarial principles
What options for state regulatory responses does the White Paper present?
1SLIDE
Audience: October 4, 2016
Big Data: Regulator Perspective
Angela NelsonDirector, Market Regulation Division
and Chief Industry LiaisonMissouri Department of Insurance
2SLIDE
Audience: October 4, 2016
SLIDE
Audience: October 4, 2016 2
Regulator View on Big DataLack of transparency
FilingsMarket Conduct
Difficult to maintain regulatory balanceEncourage efficiency in underwritingAdequate consumer protection
Struggle with causality“P-Hacking”
3SLIDE
Audience: October 4, 2016
SLIDE
Audience: October 4, 2016 3
Source: http://www.tylervigen.com/spurious-correlations
4SLIDE
Audience: October 4, 2016
SLIDE
Audience: October 4, 2016 4
5SLIDE
Audience: October 4, 2016
SLIDE
Audience: October 4, 2016 5
Big Data Working GroupAddressing regulatory challenges
Regulatory transparency• Third party models
Disparate treatmentSegmentationResources for DOIs for complex model reviewsCoordinated regulatory reviews
© 2016 Insurance Services Office, Inc.. All rights reserved.
Session 171 Panel Discussion: Big Data and Price Optimization in General Insurance
Society of Actuaries Annual Meeting 2016October 26, 2016Mark Smith, CPCU
1
© 2016 Insurance Services Office, Inc.. All rights reserved.
Agenda•Evolution of data used for insurance
•Data throughout the product lifecycle
•Benefits of Big Data products
•Big Data Challenges
© 2016 Insurance Services Office, Inc.. All rights reserved.© 2016 Insurance Services Office, Inc. All rights reserved.
Evolution of data used for insurance
3
© 2016 Insurance Services Office, Inc.. All rights reserved.4All pictures are from istockphoto
How did the data get so “big”?
© 2016 Insurance Services Office, Inc.. All rights reserved.
How did the data get so “big”?
5All pictures are from istockphoto
• Evolution of computing power and storage capabilities
• Collection and digitization of all different types of data
• Traditional and non-traditional
• Correlation revolution• GLM and post GLM analysis
•Looking for the next big thing
© 2016 Insurance Services Office, Inc.. All rights reserved.6All pictures are from istockphoto
• Increased granularity of datasets and rating plans
• Matching losses to exposure
• Additional non-traditional variables that are highly correlated with risk
• Use of predictive variables in all parts of the insurance cycle
Looking for the next “big” thing
© 2016 Insurance Services Office, Inc.. All rights reserved.7
•Expansion of insurance data sources and use of external data sources
•Expansion of data set size and granularity
•Models to find correlations using non-traditional data
•Regulator uncertainty in understanding and reviewing, looking for help and reassurance
•Consumer group focus
“Big data” – Why now?
© 2016 Insurance Services Office, Inc.. All rights reserved.© 2016 Insurance Services Office, Inc. All rights reserved.
Data throughout the product lifecycle
8
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Underwriting, reserving and marketing tools
•Lifetime value of policy, cross-selling
• Investment and portfolio analysis
“Big data” informing decisions
© 2016 Insurance Services Office, Inc.. All rights reserved.10
•Risk Analyzer
•Telematics
•Fireline
•Xactimate
•Decision Net
•Cargo net
ISO products
© 2016 Insurance Services Office, Inc.. All rights reserved.© 2016 Insurance Services Office, Inc. All rights reserved.
Consumer Benefits of Big Data
11
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Emerging markets
•Precision marketing
•Risk segmentation
•Claims settlement
•Underwriting
“Big data” benefits
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Terrorism
•Flood
•Cyber liability
•Pet insurance
Emerging markets
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Customer experience
• Innovative services
•New business models
•Cross selling and up selling
•Product personalization
Precision marketing
© 2016 Insurance Services Office, Inc.. All rights reserved.
• Increased market share competition
•Decreased residual market share
•Specialization
•More clearly defined groupings
Risk Segmentation
© 2016 Insurance Services Office, Inc.. All rights reserved.
•More accurate pricing
•Expanded market availability
• Increased objectivity
•More efficient, less costly
Underwriting
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Expedite claims payments
•Reduce investigation and litigation expense
•Detect and prosecute fraud
•Predictive value
Claims
© 2016 Insurance Services Office, Inc.. All rights reserved.© 2016 Insurance Services Office, Inc. All rights reserved.
Challenges
18
© 2016 Insurance Services Office, Inc.. All rights reserved.
•Confidentiality
•Centralized review and regulation
•Constraints on innovation
•Blurred regulatory responsibilities
•Cost and compliance
“Big data” challenges
© 2016 Insurance Services Office, Inc.. All rights reserved.© 2016 Insurance Services Office, Inc. All rights reserved.
Questions?
20