Auto Rating Plans for the New Year (and Beyond!) Michael Chen Katey Walker Gary Wang February 18, 2016

Transcription

1 Auto Rating Plans for the New Year (and Beyond!) Michael Chen Katey Walker Gary Wang February 18, 2016

2 About the presenters Kathryn A. Walker, FCAS, MAAA, CPCU Consulting Actuary Chicago, Illinois Office 16 of years experience Michael K. Chen, FCAS, MAAA Consulting Actuary Des Moines, Iowa Office 12 of years experience Gary Wang, FCAS, MAAA Consulting Actuary Bloomington, Illinois Office 17 of years experience 1

3 Overview Personal auto insurance rating plans Recent changes: New rating concepts Innovative uses of common rating variables Predictive analytics Impacts from: Territory definitions Usage-based insurance Competitive intelligence And beyond! 2

4 What year did the first auto accident occur? A B C D Polling Question #1 Which state is considered the most liberal for auto insurance laws? A B C D Massachusetts Wyoming New Hampshire Mississippi Polling Question #2 3

5 Rating plan origins 1925 Connecticut implements the first Financial Responsibility law 1927 Massachusetts enacts the first compulsory insurance requirement for vehicle registration 1939 Age, mileage, and use incorporated as rating variables 1950 s Driving record included in rating plan 1980 s Companies are using tiers and multiple rating programs 1990 s Introduction of credit scoring 2000 s Predictive analytics, usage-based insurance, household views 4

6 Traditional ratemaking limited differentiation Companies featured: Similar coverage and product offerings Similar rating variables Similar class segmentation Focus on: Overall rate level Marketing / relationships / commissions Underwriting Target niches 5

7 Ratemaking evolution Traditional Me too Innovation New coverage options Non-traditional variables Refined class segmentation Behavioral-based elements Modern Proprietary models Product/portfolio management Customer incentives Lifetime value 6

8 Rating variables Drivers attributes Age Gender Martial status Driving record Driver training Driver status Driver behavior Occupation Education Vehicle attributes Location Use Mileage Age of vehicle Value Safety devices Performance Symbols by coverage Policy attributes Multiple vehicles Coverage limits Deductible Prior insurance Homeowner Payment methods Credit score Driver-vehicle matrix Household structure Shopping behaviors Long-term value Discounts Affinity rating 7

9 Enter predictive analytics Generalized Linear Models (GLMs) Multivariate framework Impact of nonintuitive variables Refined rating segmentation Third party data Weather Highway Loss Data Institute Census data Quote information Integration of customer behavior data Retention analysis Conversion analysis 8

10 Models and rating Predictive Analytics Part of the analytics solution Cannot replace human expertise with model output Good models don t give the right answers Answers specific questions Explains broader concepts 9

11 Incorporation of models Conversion & retention analysis Territory definitions Loss cost predictions frequency & severity UBI driver scoring Claims triage Agent commissions Business Strategy 10

12 When was the last time your company updated the territory analysis procedure? A B C D E Within the past year Within the past 3 years Within the past 5 years Within the past 10 years More than 10 years Polling Question #3 11

13 Territory Analysis

14 Geodemographic data sources Hazard data providers Weather Crime Census and other governmental data Housing density Crime statistics Population density Household structure Catastrophe model output 13

15 Bulking up the data HLDI Industry level data Membership includes all of the top 10 personal auto insurance companies For each calendar year, 10 latest model years data Key coverages available BI, PD, Comp, Coll, Med, PIP 14

16 Data adjustments Eliminate effect from all other rating variables Capping Smoothing Normalization Weighting together of various internal and external data sources 15

17 Smoothing Key smoothing variables Predictive value of local data Identification of complement data How many observations are required to smooth How far to allow smoothing search to continue Many equations are available to combine local data with surrounding information Exposure Weighted Average Straight Line Declining Distance formula Squared Declining Distance formula 16

18 Smoothing Data at the basic element level lacks credibility Smoothing process allows inclusion of more localized data rather than statewide information Results in a rate or rate relativity for each individual zip code based upon the data within that zip code modified as necessary to include a significant number of observations 17

19 Smoothing considerations State borders and corners Use of smoothing across state boundaries Potential separate smoothing of urban and rural areas Distance based smoothing process or contiguous based smoothing process 18

20 Smoothing impact 19

21 Optimal cluster evaluation Selection of Target Optimal Clusters for use in establishing territories based on analysis of variance data Goal Risks within territory very similar to each other Minimize within variance Risks outside territory different from those within Maximize between variance 20

22 Percent of Total Variance Optimal cluster evaluation 45% Within Variance / Total Variance 40% 35% 30% 25% 20% 15% 10% 5% 0% Number of Clusters 21

23 Territory definition selections Helpful to look at a variety of cluster sets to provide guidance when making judgmental changes, based on: Size of resulting territories Past events distorting results Competitive considerations C luster T o Review Proposed Terr: Exposure Weighted PP Exposure Zip Count Exposure Weighted PP Exposure Zip Count Exposure Weighted PP Exposure Zip Count

24 Usage-Based Insurance

25 What statement best describes your company s view on usage-based insurance? A B C D My company markets a UBI based personal auto insurance policy My company is developing a UBI based personal auto insurance policy My company is piloting UBI products My company is not interested in UBI Polling Question #4 24

26 Usage-based insurance Current market penetration Insurance Group 2014 NWP (000's) 1. State Farm Group 35,372, Berkshire Hathaway Insurance Group 20,552, Allstate Insurance Group 18,836, Progressive Insurance Group 16,374, USAA Group 9,824, Liberty Mutual Insurance Companies 8,445, Nationwide Group 7,258, Farmers Insurance Group 7,249, American Family Insurance Group 3,516, Travelers Group 3,389, Hartford Insurance Group 2,623, Erie Insurance Group 2,357, Auto Club Enterprises Insurance Group 2,348, MetLife Auto & Home Group 2,238, Mercury General Group 2,219, CSAA Insurance Group 2,142,687 Progressive Snapshot : 2.6 Billion 25

27 Keeping an eye on the leader Progressive Snapshot 2.6 billion premium written 12 billion miles of recorded miles 2.5 million participants Progressive Insurance has formally chosen Censio to develop the software for the insurance company's leading usage-based insurance program, Snapshot. The company will begin a pilot of the mobile app with select customers across the country in mid-september. -Sep 2,

28 Moving beyond the OBD-II port devices Mobile app solutions Allstate Progressive OEM systems Ford SYNC State Farm OnStar National General Insurance State Farm Progressive VERISK! Verisk Telematics Data Exchange Verisk insurance solutions announces GM as inaugural auto manufacturer to join telematics data exchange. -Sep 2, 2015 Expected rollout: summer

29 Considerations for a successful UBI pilot program Things to consider for rating: What data will you initially use in rating? Are those data elements collected by your device? Mileage Excessive acceleration Hard braking Speeding Cornering Time of day Location Do you plan on updating the rating once you collect more of your own data? Will additional data elements be used? Are those additional elements currently collected? 28

30 In the UBI of the consumer How can data on the UBI experience of consumers shape product development, pricing and customer service offerings? How important are UBI device options to the consumer? What are the expectations of the consumer regarding savings? What other services are important to consumers in a UBI program? 29

31 Description of analysis social media data Analysis sentiment of social media posts related to UBI programs Text mining Review of individual tweets Data Over 6.5 million insurance tweets total (January, 2012 to present) Keyword searches for terms related to UBI programs in US, UK and Canada UBI information collected since October 2012 Data fields Content of the tweet Specific tweet recipient Sender of the tweet Language of tweet Where the tweet originated Link to a picture of user Latitude and longitude of the user Date and time of tweet User profile 30

32 Non-customer negative responses US Progressive Snapshot 'bombshell': Device can raise users' rates Privacy 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 23.4% 18.9% 8.8% 18.2% Misunderstanding of Program 24.4% 15.8% Rate Can Go Up Concerned About Results 9.8% 14.1% 12.3% 13.4% 12.2% Pretty sure if I tried Snapshot, Progressive would triple my premium. 16.7% 2013 Study 2014 Study 2015 Study 31

33 Positive customer responses US 0% 10% 20% 30% 40% 50% 60% 70% 80% Savings 74% 76% 76% Excitement 7% 14% 14% Tracking 0% 0% 6% Better Driving 2% 4% 13% 2013 Study 2014 Study 2015 Study 32

34 Customer complaints US 30% 25% 24% 23% 22% 20% 18% 20% 15% 16% 16% 15% 14% 10% 8% 11% 10% 10% 5% 0% 4% 2% 1% 1% 0% Driving Habits Beeping Snapshot Evaluation Annoying Premium Change Installation 2013 Study 2014 Study 2015 Study 33

35 Tone Percentage for Each Discount Category Savings US Reported Snapshot Discount by Tone 100% 90% 80% 35% 15% 0% 3% 1% 6% 0% 70% 60% 6% 50% 96% 96% 94% 40% 85% 30% 59% 20% 10% 0% 2% 2% Zero Less than 10% 10% to 19% 20% to 29% 30% and Greater Reported Discount Positive Neutral Negative 34

36 Key conclusions Value proposition has to be compelling for customers Device options are not as important as ensuring that the options work Ultimate customer satisfaction is dependent on customer perception of driving ability relative to UBI evaluation While value added services may increase the chance of conversion, the most significant reason for trying and staying with UBI is savings 35

37 Competitive Intelligence

38 How is competitive quote information currently used in your company? (choose all that apply) A B C D E Validate existing rates Marketing reports Test rating plan proposal Define business strategy Metrics Polling Question #5 37

39 Competitive intelligence evolution Traditional Competitor rate filings Key quote analysis Agent feedback And Beyond! Big data Market basket Demographics Industry and research data 38

40 Business strategy Underwriting Product Development Marketing Strategy Rating Plans Competitive Intelligence 39

41 Consumer data & analytics Analytics Hindsight Insight Foresight Current State what carriers are doing today 75% 20% 5% How do I get market data? No market data; Buy from additional Buy manufactured data ask agents or rate filings outside sources What am I doing with market data? Internal data - conversion rates, premium, sales goals Compare to Peers: evaluate agents, profitability, marketing effectiveness, compare product performance Compare to Peers: Expansion plans, cut products, change pricing or product, do strategy responding to changing market What problems will I solve if I have more data? Internal data - evaluate agents, profitability, marketing effectiveness, compare product performance Expansion plans, cut products, change pricing or product, do strategy responding to changing market What will happen If / then analysis Sensitivity analysis 40

42 Rate segmentation/competitive position assessment Current factors Insurance score Multi-line discount Multi-car discount Driver/vehicle matrix Prior insurance Age/gender/marital status Potential additional elements Billing frequency and payment history Mileage (UBI and non-ubi) New business/disappearing discounts Affinity discounts Non-chargeable accidents and claims Vehicle history Vehicle symbol Use of MVR s 41

43 Leveraging consumer shopping behavior Advantages Provides behavioral element Real information Broader market perspective Industry variables Limitations Information often self-reported Cost of information verification Non-universal variables Timeliness & updates 42

44 Premium ($) Competitive intelligence 2,000 Average Policy Rate Level 1,800 1,800 1,600 1,500 1,600 Uncompetitive Competitive 1,400 1,460 1,425 Too Competitive Current Proposed Adj Proposed Target 1,200 Median 1,

45 Comparative rater data Conversion Indicator Vendor Batch Quote Data Date/Time State Carrier Reference Carrier Name LOB Premium Rank Number Selected Number Quoted High Premium Low Premium Average Premium Median Mode Mean Range Competitive Position Policy Number Quote Number Quote Date Effective Date Agency Status City State Zip Insurance Score Marital Status Birth Date Gender Rating Tier Occupation Good Student Driver Training County Territory Model Year Vehicle Make / Model Annual Mileage Vehicle Use Deductible Vehicle Symbols Anti-Theft Devices Anti-Lock Brakes Number of Claims Number of Violations Prior Insurance Insurance Score BI/PD Limits Comp/Collision Medical Payments Limit Uninsured Motorist Limit Payment Method Advanced Shopper PAC Credit Key Rating Variables Company Detailed Quote Data 44

46 Case study competitive analysis solution Average premiums by variable to identify opportunities and areas of concern Measure of current premium proposed premium competitor premium 45

47 Relative Difference Number of Quotes Output example vehicle age Vertafore Market Basket Analysis Arizona Newest Vehicle Age 6, , , , , , Newest Vehicle Age - Number of Quotes Average Premium Median Comp X 46

48 Other than rate indications you only have the resources to perform one other actuarial study for What would you choose to research? A B C D New variable development Territory analysis Competitive analysis Usage-based insurance Polling Question #6 47

49 Conclusions Future advancements: Autonomous vehicles Safety technology Connected technology Weather & CATs Challenges: Regulatory & social acceptability Ability to verify inputs vs cost Significant impacts of changes Change is hard agents, customers, employees Benefits: More accurate rates More meaningful segmentation 48

50 Join us for the March Apex 49

51 Final notes We d like your feedback and suggestions Please complete our survey For copies of this APEX presentation Visit the Resource Knowledge Center at Pinnacleactuaries.com 50

52 Thank you for your time and attention Katey Walker Gary Wang Mike Chen Commitment Beyond Numbers 51