ACTEX Study Manual for CAS Exam 6 U.S. · ACTEX CAS Exam 6 U.S. Study Manual, Spring 2018 Edition ACTEX is eager to provide you with helpful study material to assist you in gaining

Learn Today. Lead Tomorrow. ACTEX Learning

ACTEX Study Manual for

CAS Exam 6U.S. Spring 2018 Edition Volume I

Victoria Grossack, FCAS

ACTEX LearningNew Hartford, Connecticut

ACTEX Study Manual for

CAS Exam 6U.S.

Spring 2018 Edition Volume I

Victoria Grossack, FCAS

Copyright © 2018, ACTEX Learning, a division of SRBooks Inc.

ISBN: 978-1-63588-235-3

Printed in the United States of America.

No portion of this ACTEX Study Manual may bereproduced or transmitted in any part or by any means

without the permission of the publisher.

Actuarial & Financial Risk Resource Materials

Since 1972

Learn Today. Lead Tomorrow. ACTEX Learning

ACTEX CAS Exam 6 U.S. Study Manual, Spring 2018 Edition

ACTEX is eager to provide you with helpful study material to assist you in gaining the necessary knowledge to become a successful actuary. In turn we would like your help in evaluating our manuals so we can help you meet that end. We invite you to provide us with a critique of this manual by sending this form to us at

your convenience. We appreciate your time and value your input.

Your opinion is important to us

Publication:

Very Good

A Professor

Good

School/Internship Program

Satisfactory

Employer

Unsatisfactory

Friend Facebook/Twitter

In preparing for my exam I found this manual: (Check one)

I found Actex by: (Check one)

I found the following helpful:

I found the following problems: (Please be specific as to area, i.e., section, specific item, and/or page number.)

To improve this manual I would:

Or visit our website at www.ActexMadRiver.com to complete the survey on-line. Click on the “Send Us Feedback” link to access the online version. You can also e-mail your comments to [email protected].

Name:Address:

Phone: E-mail:

(Please provide this information in case clarification is needed.)

Send to: Stephen Camilli ACTEX Learning

P.O. Box 715 New Hartford, CT 06057

ACTEX Learning Exam 6-US

TABLE OF CONTENTS

A. REGULATION OF INSURANCE AND UNITED STATES INSURANCE LAW

1. Kucera NAIC Public Hearing on Credit-Based Insurance Scores 1 2. McCarty Impact of Credit-Based Scoring on the Availability and Affordability of Insurance 5 3. NAIC Telematics Usage-Based Insurance and Vehicle Telematics 16 4. NAIC Price Price Optimization 33 5. Porter 1 – Ch. 2 Development of Insurance Regulation 48 6. Porter 1 – Ch. 3 Federal and Other Influences on Insurance Regulation 63 7. Porter 1 – Ch. 4 Roles of State Regulators and the NAIC in Insurance Regulation 78 8. Porter 1 – Ch. 5 State Department of Insurance Operations 87 9. Porter 1 – Ch. 6, 8 Licensing and Rate Regulation 94 10. Porter 1&2 – Ch. 12 Insolvency Regulation 101 11. NAIC Solv. Reg. Framework State-Based Regulation and Solvency Modernization Initiative 118 12. Feldblum Rating Agencies Rating Agencies 145 13. GAO Report Clarifications Could Facilitate States’ Implementation of LRRA 160 14. Baribeau Demystifying the Regulatory Web 166 15. Vaughan (Economic Crisis) The Economic Crisis and Lessons from (and for) U.S. Insurance Regulation 174

B. GOVERNMENT AND INDUSTRY INSURANCE PROGRAMS

16. King Flood National Flood Insurance Program: Status and Remaining Issues 185 17. Gov’t Insurers Government Insurers Study Note 203 18. Cook Personal Insurance 221 19. Webel 1 Terrorism Risk Insurance Issue Analysis 229 20. Webel 2 Terrorism Risk Insurance Legislation 240

C. FINANCIAL REPORTING AND TAXATION

21. Odomirok et al I-III (13) Financial Reporting Through Lens of a Property/Casualty Actuary, Parts I-III(13) 245 22. Odomirok et al III (14) Financial Reporting: Reinsurance in Schedule F 341 23. Odomirok et al III (15) Financial Reporting: Schedule P 377 24. Odomirok et al – IV-V Financial Reporting Through Lens of a Property/ Casualty Actuary, Parts IV-V 431 25. Odomirok et al – VI&VIII Financial Reporting Through Lens of a Property/ Casualty Actuary, Parts VI&VIII 523 26. Feldblum Surplus Statutory Surplus: Computation, Pricing, and Valuation 563 27. SSAP 5R Liabilities, Contingencies, and Impairment of Assets 579 28. SSAP 9 Subsequent Events 584 29. SSAP 53 Property and Casualty Contracts—Premiums 588 30. SSAP 55 Unpaid Claims, Loss and Loss Adjustment Expenses 592 31. SSAP 62R Property and Casualty Reinsurance 597 32. SSAP 65 Property Casualty Contracts 617 33. NAIC IRIS Property/Casualty Ratios 625 34. NAIC APPM Preamble Accounting Practices and Procedures Manual Preamble (2016) 656 35. Feldblum Loss Res. Disc. IRS Loss Reserve Discounting 668


D. PROFESSIONAL RESPONSIBILITIES OF THE ACTUARY IN FINANCIAL REPORTING

36. AAA Materiality Materiality, Concepts on Professionalism: 688 37. COPLFR PN Practice Note on Statements of Actuarial Opinion (2016) 707 38. ASOP 20 Discounting of P&C Loss and Loss Adjustment Expense Reserves 794 39. ASOP 36 Statements of Actuarial Opinion Regarding P&C Loss and LAE Expense Reserves 800 40. ASOP 41 Actuarial Communications 808 41. ASOP 43 Property/Casualty Unpaid Claims Estimates 814

E. REINSURANCE ACCOUNTING PRINCIPLES

42. Blanchard & Klann Basic Reinsurance Accounting—Selected Topics 822 43. ASC 944 Financial Guarantee Insurance Contracts 830 44. Freihaut & Vendetti Common Pitfalls and Practical Considerations in Risk Transfer Analysis 837 45. Klann Reinsurance Commutations 856

Notes on Past Exam Questions and Answers

Questions and parts of some solutions have been taken from material copyrighted by the Casualty Actuarial Society. They are reproduced in this study manual with the permission of the CAS solely to aid students studying for the actuarial exams. Some editing of questions has been done. Students may also request past exams directly from the society. I am very grateful to the CAS for its cooperation and permission to use this material. It is, of course, in no way responsible for the structure or accuracy of the manual. Readings are outlined, with the exception of some tables and spreadsheets. These, however, usually have questions based on them to aid with study. The readings are organized, more or less, by learning objective. The exams themselves usually follow this organization, with questions from learning objective A coming first. Exam questions are identified by numbers in parentheses at the end of each question. CAS questions have four numbers separated by hyphens: the year of the exam, the number of the exam, the number of the question, and the points assigned. A few recent exam questions were based on several readings but were only included after one of the readings. For those readings with no or few appropriate exam questions, review questions were created to aid students, numbered R1, R2, and so on. An attempt has been made to clean up irrelevant and redundant questions and answers. I have made a conscientious effort to eliminate mistakes and incorrect answers, but I am certain some remain. I encourage students who find errors to bring them to my attention. Please check our website for corrections subsequent to publication. ♦ ♦ ♦ I would like to thank Peter J. Murdza and Dean A. Westpfahl for their contributions to this manual, which include many summary outlines and past examination answers. To the students who make use of this manual, feedback is welcome. Good luck on April 30, 2018!

VAG

Kucera 1


Jeff Kucera, “NAIC Public Hearing on Credit-Based Insurance Scores,” April 30, 2009, American Academy of Actuaries

OUTLINE

I. INTRODUCTION A. Overview 1. Scores allows better segmentation of insurance risks

2. Most companies use for PPA and HO

3. Better matches premiums and risks’ costs

4. Prohibition of the use of scores will not affect overall premium but rather its distribution among risks B. Current Economic Circumstances 1. Current economic crisis includes severe tightening of credit markets

2. Economic problems raise questions of use of credit rating in insurance II. ISSUES IDENTIFIED BY THE NAIC A. Definition of What Constitute a Credit-Based Score

1. Insurance score – “numerical score or ranking assigned to an insurance risk (i.e., a prospective insured) based on that risk’s underlying characteristics”

a. Generates information for underwriting and pricing b. Provides a relative measure of an insured’s expected cost

2. Credit-based insurance score – insurance score that “utilizes various attributes found in a typical individual’s credit report”

3. Different models built by insurers and third-party vendors

4. Attributes seen as important a. Number of inquiries into opening new accounts b. Accounts 30 days or more past due

5. Strong correlation between scores and risk’s expected costs; statistically reliable

2 Kucera


B. Evaluation of How Insurers Use Credit-Based Insurance Scores 1. Standards for rates

a. Prohibition of ones that are excessive, inadequate, or unfairly discriminatory b. Should be actuarially sound estimates of the costs of risk transfer c. Overall rates should be sufficient d. Individual rates should reflect expected costs

2. Most insurers begin using scores in the period 1998-2001

3. Uses a. Determine whether prospective insured qualifies for underwriting b. More frequently used for risk segmentation 1) May constitute risk classification factor 2) May determine the tier to which an insured is assigned

4. Actuarial Standard of Practice No. 12: “Risk Classification” a. Risk characteristics should relate to expected outcomes b. Relationship demonstrated if variation in actual or reasonably anticipated experience is correlated

with the characteristics c. Rates are equitable if rate differences reflect differences in expected costs d. Should take into account interdependence of characteristics

5. Insurers contend that use of insurance scores, multivariate rating, and other new rating factors expand availability

C. Discussion of How Current Economic Conditions Have Affected Policyholder Premiums Related to

Credit-Based Insurance Scores 1. Impact on aggregate premium a. Scores not used to determine overall premium need but relationships between classes

b. If economic conditions affect all scores equally no change to the overall premium nor to rate relationships between insureds

2. Impact on individuals’ premium a. Regulators may have concerns that shift in scores could disrupt relative rates, whereas cost

differences do not change b. But such seen as common in insurance and are corrected by rate differential analysis c. But possible that harm is done before a new analysis makes corrections d. But so far no evidence of a dramatic shift in credit scores

Kucera 3


PAST CAS EXAMINATION & REVIEW QUESTIONS

1. A regulator is evaluating a rate filing in State X that incorporates the use of credit-based insurance scoring.

a. Define a credit-based insurance score.

b. Briefly describe two ways an insurer could use credit-based insurance scores. (11–6US–1a&b–.5/.5) R1. What, according to Kucera, should a shift in credit scores do to aggregate premium collected? R2. Describe two ways that insurers use individuals’ credit scores in underwriting. R3. When Kucera gave his speech (2009) there was a global economic crisis. Some people feared that an

economic crisis could impact the credit scores of individuals and thus their premium. What, according to Kucera, had been observed so far? What are two possible reasons for this observation?

R4. Kucera discusses two ways of examining the impact of an economic crisis via credit scores on premium.

What are these two different ways? R5. Several studies have already been done looking at the relationship between individual credit scores and

two lines of business. What are these lines of business? R6. Some insurers argue that the use of credit scores allows them to write more insureds. Defend this

argument.

4 Kucera


SOLUTIONS TO PAST CAS EXAMINATION & REVIEW QUESTIONS

1. a. A credit based insurance score is a ranking of the relative risk of a policyholder based on its characteristics and items found in a credit report. Examples of such items are the number of inquiries for new accounts and balances 20 days overdue. 1) Credit based insurance score is a score or rank assigned to a risk based on attributes commonly found

in a credit report. The purpose is to estimate the relative expected loss. 2) It’s a score based on insured’s credit history. 3) A credit based insurance score is a calculated value attempting to gauge an insured’s insurance risk

based on its credit history.

b. Segment risks based on insurance score, and charge higher premiums to risks with worse scores and lower premiums to risks with better scores. 1) Use insurance scores to determine whether or not to offer coverage to risks with worse scores. 2) Use insurance scores to slot risks into one of several underwriting companies. 3) Use insurance scores to target certain market segment in the marketing campaign.

R1. Let us assume that the insurer is using credit scores to rate individual insureds. In the long run, unless overall loss costs or expenses change, the aggregate premium should not change. This is because credit scores are used to assign individuals to different classes, but the overall premium needed is not changed.

R2. (1) An insurer may use an individual credit score in deciding whether or not to underwrite that individual. (2) An insurer may use an individual credit score to assign the individual to a particular tier of insureds. R3. At the time of his speech, the crisis had been about six months in the making. So far the impact on

insurance, according to anecdotal evidence, had been minor. One reason for this could be that most insurance is based on renewals and this would see less change in the rates. Another possible reason would be that the crisis needed more time to be felt in the overall premiums.

R4. He believes one should look at the (1) aggregate premium (2) individual premiums. R5. Private passenger automobile and Homeowners (see Appendix). R6. Instead of declining the insured, which they would have to do without the credit score, they can assign the

insured to a higher tier.

McCarty 5


Kevin M. McCarty, “The Impact of Credit-Based Insurance Scoring on the Availability and Affordability of Insurance,” May 21, 2008.

OUTLINE I. OVERVIEW A. Use of Credit-Based Insurance Scores in Personal Insurance Lines 1. Proponents a. Scores are predictive of future claims experience b. Serve as a necessary underwriting tool

2. Critics – scores are another example of discrimination against lower-income people

3. Studies do show predictive value but also disparate impact

4. Most insurance professionals undecided about whether use is unethical

5. Use requires examination of the acceptability of factors used in underwriting and rating

6. Wide variety of information is available that correlates with insurance claims but their usage is not necessarily fair and valid

B. Other Rating Factors Considered To Be Important

1. Different life expectancies by race can be used to justify different rates but produce negative results a. Violates equal protection for consumers b. Results in poor public policy

2. 1996 – Health Insurance Portability and Accountability Act outlaws usage of genetic testing for health insurance purposes

3. 2007 – FL holds hearing on whether the rating factors of occupation and education serve as proxies for race

II. PROBLEMS WITH THE USE OF CREDIT SCORES A. The Credit Reporting System

1. Inherent weaknesses in the credit reporting system are also a problem that may invalidate the usage of credit scores

2. Groups that are negatively affected by credit reports to a disproportionate degree a. Recent divorcees b. Recently naturalized citizens c. Elderly d. Disabled e. Those with certain religious convictions f. Younger individuals with no credit histories

3. Economic downturn may increase differentials in credit scores

6 McCarty


4. Results of empirical studies a. Differential only in the frequency of claims, not in their magnitude b. But frequencies of insured loss events may be the same even though frequencies of claims filed

differ; those with lower incomes file more claims c. No suggestion that claims filed are not legitimate or that rates charged excluding credit scores are

not actuarially sound

5. Opaqueness of methodology that produces credit-based insurance scores a. Variation by company b. Affected by financial decisions not linked to insurance risks, e.g., having an installment loan

B. Disproportionate Impact of Credit-Based Insurance Scores 1. Concern over the relationship between credit scores and the following: a. Race b. Ethnicity c. Income status

2. Though credit scores are not a proxy for race, relationship may be strong enough to prohibit their usage because of discrimination

3. Lower average credit scores found in the following groups: a. African-Americans b. Hispanics c. Young people and elderly as use credit less often d. Certain religious groups that do not use credit 4. Scores seen as not measuring event risk but indirectly measuring socioeconomic status III. THE RESPONSES TO THE PROBLEM A. Florida Actions Regarding Credit-Based Insurance Scores 1. 2003 – FL passes a law limiting usage of credit scores in PPA and HO insurance

2. Four industry challenges to the law a. FL regulators did not have authority to prevent usage as an underwriting/rating tool b. FL regulators did not have authority to define “unfairly discriminatory” c. Insurers lacked necessary data to show effect of scores on protected classes

d. Definition of “disproportionate impact” seen as too vague – only challenge upheld by the administrative law judge

B. Conclusion and 2007 FTC Report 1. Negative impacts seen to outweigh benefits of enhanced accuracy

2. Concern about similar variables, namely occupation and education

3. Fair and Accurate Credit Transactions Act (FACTA) of 2003 mandates an FTC report, which proves disappointing

a. Data may have been selected to present the best case for scores b. No premium used c. Economic advantages of the usage of scores seen as conjectures d. Industry, but not regulators, received advance copies of the report

McCarty 7


C. State Involvement 1. 48 states have taken legislative or regulatory action limiting the usage of credit scoring

2. Many adopted model legislation or a variation of it

3. Many legal provisions involve the following: a. Notification of consumers of its usage b. Allowing regulators access to the scoring model c. Restricting decisions based only on the model

4. Other legal provisions a. Disallowance of credit history as the sole basis for underwriting/rating decisions

b. Prohibition of usage of credit history for policy cancellation or nonrenewal or for rate increases

c. Banning credit history for underwriting or rating existing customers d. Banning credit history for underwriting or rating in all cases

5. Important the federal action not detract from states’ regulatory power

6. HR 5633 seen to have some favorable provisions a. FTC to examine in more detail relationship between credit scores and race/ethnicity b. But other agencies should also investigate

7. HR 6062 a. Agrees with its exemption of personal lines insurance from the Fair Credit Reporting

Act b. Recognizes that FTC report found a disparate impact on minorities and thus as a starting

point credit scoring should be eliminated c. Reliance on the Fair Credit Reporting Act consistent with separation of powers of

McCarran-Ferguson

8. Differing views on credit scoring found in other states a. Not a concern if one of many factors b. Any disparate impacts on race/ethnicity seen as coincidental c. Most consumers benefit from it d. No need to expand the analysis to occupation or education

8 McCarty


PAST CAS EXAMINATION QUESTIONS

1. A regulator is evaluating a rate filing in State X that incorporates the use of credit-based insurance scoring.

a. Briefly describe two reasons the regulator might support the use of credit information in ratemaking. b. Briefly describe three critiques of the use of credit-based insurance scoring that a consumer advocate

taking McCarty’s position might make. c. A recent downturn in the economy of State X may affect the usefulness of credit-based insurance scoring.

Propose one analysis that the company actuary could undertake to ensure the continued appropriateness of the overall rate level in State X. (11–6US–1c,d&e–.5/.75/.5)

2. a. Briefly describe two reasons that using credit scores in insurance may be appropriate. b. According to McCarty, describe three reasons why the use of credit score in insurance ratemaking may

not be appropriate. c. Assume that an insurer uses credit scoring in ratemaking. Discuss the impact an economic downturn

would have on Insurance consumers Insurance companies Regulators (14S-6US-5-2.75:0.5/0.5/1.75)

3. a. Briefly describe the cost-based condition for insurance rates to be considered equitable.

b. According to McCarty, briefly describe two conditions for insurance rates to be considered equitable to consumers.

c. An insurance company finds a significant correlation between a driver’s claim frequency and the

number of text messages sent. The insurance company proposes segmenting groups into low and high risk categories based on texting frequency.

Describe whether the insurer’s use of texting frequency as a rating variable would be considered equitable in each of the following contexts:

Within a risk classification system

From the perspective of an individual consumer

d. In a scenario where all auto insurance companies except for Insurer X are using texting frequency

as a rating variable, discuss a concern that a regulator might have regarding Insurer X’s financial stability.

e. Based on the concern discussed in part d. above, briefly describe two IRIS ratios that the

regulator should examine and the trend that the regulator should look for in each ratio to validate the concern.

McCarty 9


f. Assuming that Insurer X does not introduce texting frequency as a variable, describe one action

that Insurer X could take to remain competitive. (14F-6US-2-3.75:0.25/.5/1/0.5/1/0.5) 4. a. According to McCarty, identify the two age groups where insureds are disparately impacted by

the use of credit scoring and briefly describe the reasoning.

b. Other than age groups, identify one group of insureds that McCarty claims is disparately impacted by the use of credit scoring and briefly describe the reasoning.

c. Empirical studies have shown that those with different credit scores have an observed difference

in total insurance losses. Identify a key metric and describe how the metric could be driving these loss differences.

(15F-6US-1-2.25:1/0.5/0.75) 5. a. Assume that credit scores are significantly different between genders, and as a result, a law has

been proposed to ban the use of credit scoring in personal lines ratemaking.

Describe one argument in favor of and one argument against this proposed law.

b. Assume that personal lines insureds with better credit scores counterintuitively have a higher claim frequency. i. Propose an explanation for observed frequency difference ii. In the context of these results, describe one argument in favor of and one argument

against the use of credit scores in personal lines insurance. (16S-6US-1-2.5:1/0.5/1) 6. An insurance company uses credit-based insurance scoring as part of its personal auto rating plan. A

recent severe countrywide economic downturn is believed to be causing a lowering of credit-based insurance scores among consumers.

a. Describe one concern regulators may have regarding the response of the company’s rating plan to

the changes. b. Briefly describe and justify two ways in which the pricing actuary could reflect the impact of the

economic downturn on credit-based insurance scores in an upcoming rate filing. c. To protect consumers against the possibility of large rate swings due to deteriorating credit-based

insurance scores, the regulator requests that the insurer cap the renewal rate increase for each policyholder. Describe two ways this might violate the ratemaking principles outlined in the CAS “Statement of Principles Regarding Property and Casualty Insurance Ratemaking.”

(16F-6US-2-2.5:0.5/1/1)

10 McCarty


SOLUTIONS TO PAST CAS EXAMINATION QUESTIONS

1. a. Credit scores show the relative risks of the consumers correlated to the score ranges. They allow insurers to price insurance better based on the consumer’s relative risks. 1) Scores allow the insurer to compete better in the market, creating insurance availability. 2) A majority of policyholders benefit from the use of credit scoring, having lower rates than they would

have in the absence of credit scoring. 3) Allows insurer to write more business. 4) Allows insurer to price more accurately. 5) Credit based scores are predictive of future loss experience.

b. (Social)

1) Use of credit is counter to equal protection for consumers and not sound public policy. 2) Use of credit is not measuring insurance risk, but indirectly measuring socioeconomic status. 3) The disparate impact based on race or ethnicity is discriminatory. 4) Has a disproportionate negative effect on low income people and protected classes of citizens. (Data Issues) 5) Credit report data can often be inaccurate. 6) Credit reports can be adversely impacted by things outside the direct control of the insured (identity

theft, age, religious beliefs, disability, economic conditions). 7) Approximately 50% of credit reports contain errors. 8) Many groups of people do not use credit – example certain religious groups. 9) Errors in credit reports are widespread and may invalidate the use of credit history. (Opacity) 10) Credit scoring methodology is difficult for consumers to understand. 11) Resulting credit scores from models can be negatively impacted by sound financial decisions that

can’t directly be linked to insurance risk. 12) Credit scoring methodology varies significantly from company to company. 13) Consumers don’t understand how the scores are impacting them. 14) Calculation of credit scores is complex and not transparent to consumers. 15) Opaque to consumers who cannot link financial decisions to insurance prices. 16) Credit scores have only been shown to be productive of claim frequency, but not severity.

c. Determine if the relative magnitude between levels of scores is still supported; increase / decrease overall rate levels to balance any distributable shifts. 1) Determine if the relative magnitude between levels of scores is still supported; increase / decrease

overall rate levels to balance any distributable shifts. 2) Actuary should review the premium in the state to ensure that overall premium is adequate. Then,

redistribute premium/ rate base on the relativity of the insurance score / relative risk. 3) If all scores increase by a factor of say 5% the actuary could renormalize all scores so that the average

level remains unchanged. 4) Actuary should estimate the impact of economy downturn on overall insurance score. Then, apply it

to adjust pricing. If a person’s insurance score is dropping at average level, he should not be surcharged. He should only be surcharged if drop in score is more than average.

5) Using a multivariate model recalculate the relatives between various credit scores then off balance these to assure the correct overall rate level is achieved.

6) Adjust overall credit scores to have neutral affect this would cause no charge in overall premium. 7) The overall amount of loss should not presumably be affected, so the insurer should adjust the base

premiums so the overall amount is adequate, similar to adjusting for changes in home values in homeowners insurance.

McCarty 11


2. a. Two of the following: reflect differences in expected costs, loss costs, frequency, severity appropriate for risk differentiation increases availability won’t lower rate level overall if not used lowers pricing risk and thus overall costs objective and available measure difficult to manipulate

b. Three of the following: weakness in credit reporting system errors in reports due to identity theft disproportionately affected by certain life events: divorce, recent citizen, disabled discriminates against those who don’t use credit opaque to consumers disparate impact to select ages, race, and religious groups indirectly measures socioeconomic status privacy concerns reflects frequency not severity, claims reporting pattern studies didn’t use premium

c. Consumers: Credit scores may decrease (uniformly, in segments) Premium increases temporarily, assuming companies adjust accordingly Inherent riskiness hasn’t changed Premium increases permanently, assuming loss costs increase Might not be impacted if company does not rerun score for renewal Rate can go up if score decreases more than average (person lost job, etc.) Might drop coverage

Insurance Companies: Will see a change in credit score distribution, uniform or otherwise Will need to examine relativities among risks Will need to decrease/increase overall loss cost estimate, depending if loss costs are assumed to have

changed or stayed the same similar to AOI on homeowners Will reconsider relativities/loss costs due to competitive pressure, adverse selection, loss of market

share Might be temporary increase in profit due to increased rates

Regulators: Increased complaints from consumers Need to apply additional scrutiny to rate filings may reexamine use of credit scores in rating models May see availability decrease and increased residual market

12 McCarty


3. Note that this question pulls from multiple readings, covering subjects from learning objectives A1, A2 and C2

a. It gives lower risks the rate they deserve based on their expected loss cost and eliminates possible subsidization

b. Two of the following:

Rates should not be based on factors that correlate highly with race, ethnicity, religion and other protected factors.

Factors that the insured is charged for (factors used in rate calculation) are under the insured’s control.

Another condition is that rates actually correlate to the underlying risk. For example, McCarty expressed concerns that a downturn in the economy could cause abrupt and unjustified changes in credit-based insurance scores.

Should not be subject to inaccuracies, e.g., credit scores are often wrong.

c. Within a risk classification system: Using texting frequency will create a more refined classification system and make the rate more

aligned with true risk condition, thus equitable. Unequitable– if the texting data was self-reported it could be prone to error which would make it

unreliable. From perspective of an individual consumer: Not equitable as there is no clear causation between texting and driving Would have privacy issues. Meanwhile, will have disparate impact on certain groups of people, e.g.,

those sending messages frequently while not driving. This is not equitable. d. Due to anti-selection, high risk insured tend to purchase insurance from insurer X, thus may result in

insolvency issue for insurer X. e. Many possible answers, here are some examples:

2 year operating ratio: Is the insurer’s profitability decreasing due to adverse selection? Regulator should look for increasing trend and ratio > 100% to identify unprofitability

NWP to Surplus: Insurer X may see premium growth with smaller surplus increase (or even decrease), with ratio becoming higher over time.

Change in NWP: See if it’s grown in the last year. It’s an indicator if insurer did not take steps to actively grow business that it’s being adversely selected against.

Gross ∆ PHS o As the increase high risk insureds will raise liabilities and therefore decrease PHS, should look for negative trends.

f. Many possible responses

Find a proxy variable for texting: Insurer X could refine its rating based on other rating variables to avoid adverse selection by picking up texting variable with something correlated but more reliable.

McCarty 13


4. a. The elderly and young adults. Young adults typically have not had the time to build a good credit score or credit history. The elderly typically use less credit, and if they are on a fixed income they may also have lower income which also correlates to a lower credit score. If credit scoring is used to determine rates, these groups will be at a disadvantage.

b. One of the following (others possible): People of lower socioeconomic class – may have lower score due to lack of available lines of

credit b/c of their income, for example. This drives up premiums for something that is beyond their control.

Certain religious faiths which don’t support the use of credit / usury / interest, so they wouldn’t have fair credit scores – doesn’t tell about driving ability.

Race / minorities - McCarty claims credit scores are a proxy for race and socioeconomic status. Poorer minorities are more likely to have a low credit score and insurance companies are effectively increasing their rates.

Recently divorced – they had credit w/ their spouse that they most likely had to cancel. New citizen / new immigrants – those who just moved to the US have no or short credit

history, therefore probably won’t have high credit score. They would be disparately affected. Disabled / handicapped – more difficult to find jobs and keep steady income. Creditors may

be less likely to provide them loans. c. One of the following (others possible):

Frequency of claims / percent of accidents reported / willingness to file a claim / risk absorbing – people with high credit scores may pay for smaller claims out of pocket while low credit scores may file a claim with insurance company. Therefore those with high scores will have lower total losses since they are paying for some claims themselves.

Income / wealth / affluent / socioeconomic status / more resources / wage – Individuals with higher income tend to have higher scores. Higher income individuals are more likely to retain smaller losses as opposed to lower income insureds. This translates to lower total insurance losses for higher income (higher credit scores) insureds and vice versa for lower income (lower credit scores) insureds.

Responsible / carelessness / self risk control / managing credit score – people with lower credit scores are thought to be not very responsible in their finances. That could translate into irresponsible driving behaviors, which lead to higher expected losses. People with higher credit scores -> more responsible / careful drivers -> less expected losses.

Late payments / paying bills on time / pay off debt / number of days overdue / accounts overdue / balances over 90 days overdue – paying on time is a key metric to your credit score and it speaks to a person’s responsibility. Someone who is more responsible is generally more aware and avoids losses thus lowering their total losses.

Number of inquiries / attempts to open credit cards – one would seek to open new credit accounts to relieve one’s financial distress temporarily. This lowers the credit score and this person is more likely to file a claim since he/she is unlikely to be able to absorb further financial damages.

Length of time credit lines have been open – if an 18 year old driver pulls a credit score, its likely to be low since they probably haven’t had lines of credit open for too long. They also have limited driving experience due to age, which could drive up their total insurance loss.

Deductible – people with higher credit score tend to buy coverage with higher deductible. As a result their incurred losses are smaller because they are on net of deductible basis.

Location car is most frequently parked – minority groups appear to be disparately impacted by the use of credit scoring. Over time, it is argued that socioeconomic barriers, as well as unfair lending practices against these groups, have led to worse credit situations. Urban areas, which tend to have higher populations of minorities, also have higher population density, which leads to more frequent losses.

14 McCarty


5. a. In favor of banning: Credit scoring can be considered highly discriminatory since it can be based on region, affluence, race, gender, etc. Banning the use of this variable could be appropriate to avoid discriminating and to avoid public outcry over the matter. Social acceptance is an important aspect in selecting rating criteria.

Against banning: The scores are correlated with loss experience. By banning you force un-actuarial premiums to be charged. Low risks pay more than they should and high risks are being subsidized.

b. i. One of the following:

A high credit score driver may carry high debt on a car. Therefore, there might be a diminished sense of ownership leading to drive more recklessly. Whereas a low credit score driver may own the car without financing, leading to more cautious driving. (a)

People with better credit scores have better education and awareness of insurance, they tend to file a claim whenever they believe that they are covered. Whereas lower credit score people may not have the awareness or resources to file a claim.

The people with higher credit scores live in metropolitan areas. Metropolitan areas are highly populated and more accidents are likely to happen so they have more frequent losses than low income suburban families who live in less densely populated areas.

ii. Arguments for using credit scores (can be tweaked to suit part b.i. above): Credit scores provide insight into insureds habits (like spending habits, risk aversion or lack

thereof), so using credit scores to assign insureds to risk classes will predict loss expenses more accurately, which will increases availability.

If higher credit score insureds really do drive more and have higher frequency, it is fair for them to pay higher premiums so that low scores aren’t subsidizing them. Arguments against using credit scores (can be tweaked to suit part b.i. above):

Because credit scores produce a counterintuitive result, and studies have shown that more than 50% of credit scores contain errors, in addition to identity theft problems, credit scores should not be used in pricing.

Credit score is just a proxy for something else. Use urban versus rural instead; it’s more intuitive and understood by policyholders

6. a. Regulators would be concerned that insureds would have an increase in premium AND (one or

more of the following): without a change in inherent risk; it would be excessive or company would earn excessive profits; it would disproportionally impact protected classes or lower income people or lower socioeconomic insureds; would be unaffordable; would lead to less availability of insurance o it is unfair or inequitable or not actuarially sound

b. Two of the following (others possible):

If the current rate relativities between score classes remains valid, and CBIS scores are dropping in essentially a uniform fashion, the pricing actuary would respond to the distributional shift via an offsetting change to the base rate. There would be no long-term impact on the premium collected just from the CBIS shift.

Remove CBIS from rating by using a proxy to replace it or recalibrating other rating variables absent the CBIS

Incorporate the rising premiums into the premium trend selection, which will result in a decrease in the overall indication

Use CBIS only in accept/reject or tier placement underwriting decision making instead of in rating.

McCarty 15


Capping the overall premium change that insured would see as a result of their credit score change Remapping or changing the ranges of CBIS corresponding to certain factors in rating Freezing insureds' credit scores or using an average score over several years to limit the impact Restrict CBIS score changes from resulting in an increase in premium. Only allow the impact to be

premium neutral or result in a decrease. Change the rate differentials similarly to how Homeowners rating will change when the housing market

shifts. Calculate the overall premium after the downturn using CBIS from before the downturn and compare to

the total charged premium after the downturn.

c. Two of the following: Principle 1: A rate is an estimate of the expected value of future costs. Capping premiums results in the

expected future costs being higher than the rate charged. Principle 2: A rate provides for all costs associated with the transfer of risk. By capping individual

insureds and not adjusting the premiums of the other insureds, the company is failing to provide for all costs associated with the transfer of risk on an aggregate level

Principle 3: A rate provides for the costs associated with an individual risk transfer. Capping individual premiums prevents insurer from charging a rate that accounts for all costs associated with the transfer of risk on that individual.

Principle 4: A rate is reasonable and not excessive, inadequate, or unfairly discriminatory if it is an actuarially sound estimate of the expected value of all future costs associated with an individual risk transfer. Capped rates would be lower than the actuarial sound rate prescribed by the rating plan, leading to overall inadequate rates.

16 NAIC Telematics


“Usage-Based Insurance and Vehicle Telematics: Insurance Market and Regulatory Implications” NAIC White Paper by Dimitris Karapiperis et al

2015 pp. 1-16 and 42-60

OUTLINE

0. Glossary

DBD: Driving Behavior Data DOI: Department of Insurance ECU: Electronic Control Unit FHWA: Federal Highway Administration GPS: Global Positioning System OBD: On-Board Diagnostics PAYD: Pay-as-You-Drive PAYDAYS: Pay-as-You-Drive-and-You-Save PHYD: Pay-How-You-Drive PPP: Public-Private Partnership UBI: Usage-Based Insurance VMT: Vehicle Miles Traveled

I. Forward (by NAIC Staff)

A. Telematics-supported usage-based insurance (UBI): new era in automobile insurance. B. Study looks at

1. Technological advances 2. Changes in the insurance market 3. Analyze implications for insurers, consumers and state regulators

C. Vehicle telematics, integrated navigation, computer & mobile communication technology 1. Let insurers use true causal risk factors to develop precise UBI rating plans 2. Policyholders incentivized to adopt risk-minimizing behaviors 3. Benefits for consumers, insurance companies, and society 4. Insurance market expanding availability of telematics-based UBI programs. 5. CIPR survey of state departments of insurance (DOI): UBI in at least 42 states

D. US automobile liability insurance from its beginning unti l recently 1. No true causal data 2. Used group behavior-based demographic proxy factors affecting loss costs

a) Driver record, age, gender, marital status and territory. b) More recent correlated but controversial variables: education, occupation and

credit scores. 3. Individual driver level, the concepts of UBI not new at all,

a) U s e d mileage aa rating variables insurers historically b) Needed to trust consumers for mileage and other driving details

4. Value of real driving behavior data recognized in early days a) 1929 paper by Paul Dorweiler identified critical factors contributing to

frequency & severity b) Technology not available then

NAIC Telematics 17


5. Advancements in technology a) new digital technologies in cars during the 1980s allowed for the development of

increasingly electronic management and operation control sophisticated systems (engine management, suspension systems, braking, safety, etc.)

b) Telemetry systems began with high-tech race cars, moved to long-distance truck fleet operators in 1990s

c) GPS originally developed for military but then introduced to civilian d) Combination could provide detailed driving behavior to central behavior e) Started integrating into underwriting decisions but high costs of integrating new

technology interrupted/slowed down adoption f) But technology has improved and become cheaper g) More widespread: 38% new US cars in 2013; 80% new US cars projected for

2018 E. Telematics (for insurance) defined by SAS as “the use of wireless devices to transmit data in real

time back to an organization” where data can be used to: 1. Create more accurate pricing, 2. Improve granularity of risk management techniques 3. Reduce losses by enabling better claims assessments. 4. Get better predictive models for identifying and analyzing risks

F. Many U.S. insurers offer telematics-based UBI policies with significant discounts 1. Market survey: Consumers like the technology and the value it can offer. 2. Insurers’ telematics programs going beyond premium discounts to include other value-

added services aimed at increasing competitiveness and consumer loyalty. 3. ABI Research: global insurance telematics subscriptions to grow from 5.5 million (2013)

to 107 million (2018). G. UBI, PAYD, PAYS or PHYD – imprecise but factors affecting premiums are generally :

1. Where (location) the automobile is driven 2. How often (number of trips) 3. How far (mileage) and 4. How well (driver behavior.)

H. Key drivers for the rapid growth of telematics-based UBI are numerous benefits 1. Consumers: possible lower premiums, enhanced safety and claims experience, 2. Insurers: reducing claim costs, better risk pricing, mitigating adverse selection and

moral hazard, modifying risky behavior, and improving brand recognition and loyalty 3. Social benefits from incentivizing fewer miles: reducing accidents, congestion and fuel

consumption, carbon emissions, dependence on fossil fuels. I. Major barrier remains for public acceptance: privacy of data

1. Consumers question insurers’ ability to safeguard their data 2. Major corporate security breaches. 3. Consumers gradually less uneasy following insurers’ assurances of

a) Limited use and storing of private data (e.g., GPS–detailed data) b) Not sharing data with other third parties (e.g., police, marketers).

J. Consumer privacy issues also addressed in state legal frameworks 1. California prohibits use of private data for most insurance purposes 2. Transmission, storage and reporting key concern for state regulators 3. Also concerned about rating factors determining UBI premiums.

K. Regulation of UBI 1. States with and without active telematics UBI programs emphasize:

a) Rates not to be excessive, inadequate or unfairly discriminatory b) Need for public disclosure and transparency.

2. Survey shows some states have passed legislation on use of telematics

18 NAIC Telematics


a) If the rating factors in statute do not include the standard UBI PAYD behaviors (e.g. California) telematics UBI programs are in question.

b) States pro-telematics, UBI-specific legislation has been enacted affording confidentiality protection for insurers’ proprietary UBI solutions.

c) Novelty and regulatory challenges will keep state regulators focussed on (1) Safeguarding consumers’ rights while allowing (2) Development of new, more effective insurance plans.

3. Federal Highway Administration (FHWA) funding multiple promotion efforts

II. Telematics Technology in the Automobile Insurance Industry (by NAIC) A. Introduction

1. Data traditionally one of insurance industry’s greatest assets. 2. Historical challenge from:

a) Wireless connectivity b) Increasing sophistication of in-vehicle electronics c) Machine-to-machine (M2M) communication

3. Insurers investing in ability to collect, store, manage and analyze vast amounts a) Solving complex problems in order to remain competitive and profitable. b) Big data poised to change the business of insurance as we know it

4. Data sets large a) ~ 5MB to 15MB of data annually, per policyholder. b) Cost of technology, hardware, installation, maintenance and logistics: a limiting

factor to adoption of telematics. c) As technology becomes cheaper, expected to grow faster.

5. Significant challenges to insurers integrating into IT infrastructure a) Huge data demands in terms of storage and analytics b) Lack of standardization in telematics devices c) Main players—auto manufacturers, insurance companies and telematics service

providers—competing for market with own solutions and products. d) Choosing best technology for UBI only first challenge for insurers.

(1) Lack of publicly available driving behavior data mean high costs for launching and maintaining a telematics-based UBI program.

(2) Success for insurers means building effective and profitable program without passing costs to consumers

B. Current Telematics Technological Solutions

1. Telematics devices generally used are either a) Plugged into the on-board diagnostics (OBD-II) port of an automobile b) Integrated in original equipment installed by car manufacturers.

2. Data recorded and transmitted from the car varies according to a) Telematics technology chosen b) Policyholders’ willingness to share personal data. c) Simple: date, time, location and distance driven d) Complex: speed, lane changing, cornering, acceleration and deceleration.

3. Currently, four distinct categories of telematics solutions on market: a) Dongle:

(1) Self-installed device provided by the insurer to be used for a certain time, typically for six months.

(2) Preferred solution in U.S. due to low cost and high reliability (3) Installed by driver, re-usable, can be transferred to another vehicle, turns

on with ignition, high-quality and secure data on location and driving style, and can be bundled with other value-added services.

NAIC Telematics 19


(4) Weaknesses: only modern vehicles, could be tampered with, will soon be technologically obsolete.

b) Black box: (1) Professionally-installed black box, popular across Europe (2) Good for PAYD and PHYD, especially PHYD as can provide in-depth

and detailed data on driving behavior. (3) UBI-PHYD products require devices like black box to track performance

data like speed, g- forces in hard cornering and braking. (4) Black box can also use vehicle’s internal sensors by linking with its

electronic control unit (ECU). (5) Good for first notice of loss (FNOL) services

(a) Early notice in the event of theft (b) Good information for forensic crash reconstruction

(6) Driving behavior data (DBD) for inexperienced drivers. (7) Weaknesses: not portable and expensive

c) Embedded: (1) End of 2013, 11 car manufacturers with embedded telematics (2) Early embedded telematics services: diagnostics, navigation and

infotainment services (3) Now, connected to ECU, can deliver UBI services and record/transmit a

wealth of data (4) Strengths of embedded telematics: Product differentiation; Improved

customer relationship management; Potentially lower costs in the case of product recalls.

(5) Weaknesses: Comparatively high cost for consumer (most are subscription-based); Lack of standardization; Not always compatible with insurance solutions; Obsolescence (long product cycle for automobile)

d) Smartphones - latest tool in telematics (1) Stand-alone devices or linked to vehicles’ systems (2) Strengths:

(a) Equipped with a host of relevant sensors, such as GPS, accelerometers and gyroscopes. T

(b) Large data storage capacity (the cloud) (c) Superior communication capabilities. (d) No device, installation or connectivity costs to insurers (and no

additional cost to the consumers) (e) Computing power means some data processing can be done on

the device (f) Economies of scale (g) Rapidly improving technology

(3) Weaknesses: (a) Quality of data (b) Reliability of measurement data

C. Insurers’ PAYD UBI Telematics Program 1. Progressive’s Snapshot a wireless device plugged into an OBD II port

a) Records and transmits time, speed and harsh braking. b) Partnered with AT&T for network support. c) Does not record location as does not currently have GPS d) Drivers’ personal data received is not shared with third parties e) Snapshot only used to resolve a claim if the policyholder permits it

20 NAIC Telematics


f) Data not shared unless required to prevent fraud. g) Progressive’s telematics has many patents, challenged by competing insurers. h) March 2014: more than 10 billion miles driving data with Snapshot i) Exploring methods, mobile apps and GPS, to capture new driving factors.

2. Allstate’s Drivewise: telematics device installed in the vehicle’s diagnostic port. a) Also partnered with AT&T b) Device records time and location during trips, number of trips per day, speed,

hard breaking and mileage. c) Average driving performance does not earn policyholders discounts while more

exposure can increase cost. d) Drivewise participants can monitor their behaviors/discounts with an app.

3. State Farm’s Drive Safe & Save uses 3rd party technology a) Works with telematics already in vehicles: such as OnStar and SYNC b) With an In-Drive device provided by Verizon. c) Drivers must pay annual subscription after the first free year d) Data includes: miles, acceleration, hard braking, sharp turning, speeding and

time of the day e) Also provides roadside assistance, maintenance alert and stolen vehicle locator. f) 3rd-party solutions use GPS, but company only records general location and

does not share that information, except as required by law. 4. The Hartford’s TrueLane: device plugs into OBD-II port; collects and transmits drivers’

data to the company using cellular phone signal. 5. National General’s telematics UBI programs

a) Based on General Motor’s OnStar to confirm miles driven, b) Only available to those vehicles equipped with OnStar.

6. Nationwide’s SmartRide: plug-in device a) Collects only driving behavior data and GPS information b) Drivers can go online to track their discount and personalized feedback.

D. Data Challenges: other big technological challenge 1. Need critical mass of data for effective telematics

a) With tools and information infrastructure, can collect/analyze driving behavior b) Industry has not yet (for most part) moved to data about environment (i.e., road

type and conditions, traffic patterns, etc.) which impact risk 2. Need right data to understand and adequately model risky driving behavior and for UBI

programs to be successful 3. To be effective need to communicate right data to the end-user

a) Standardization telematics data collected and reported to insurers required for effective analytics and widespread telematics adoption.

b) Association for Cooperative Operations Research and Development (ACORD), a global standards development organization, working to standardize data elements

c) Given multiple representations of data, from many devices, ACORD engaged to ensure data delivered in standard format to all insurers.

4. Once telematics data delivered to insurers, need to understand. E.g., braking on highway not like braking on a rural road; risks different

5. Rich PAYD variables best way to understand a) How drivers behave under real conditions b) Sustain telematics UBI risk models over many years

NAIC Telematics 21


III. Telematics UBI Modeling and Analytics (Robin Harbage, Director, Towers Watson)

A. Introduction 1. Usage-based insurance has been in development since the 1990s.

a) Based on data from telematics devices professionally installed in automobiles (manufacturer or technician and aftermarket device).

b) After defined period of monitoring, insured gets new rate with driving experience as a part of the rating algorithm.

c) Almost no insurers base entire premium on just driving behavior, still largely rely on proxy variables approved in their jurisdiction.

2. Top key issue facing insurers is ability to build predictive loss cost models that identify behaviors indicative of unsafe vehicle operation.

B. Predictive Models – currently two main types

1. Less granular – predefined set of events a) Total mileage, time of day b) Events such as harsh braking, acceleration or cornering c) Event counter scores limited in their capability because based on assumption that

certain events are the universe of variables to predict loss costs 2. Much more granular data about vehicle use

a) Second-by-second basis or more granular as needed for accelerometers b) Research predictive power of characteristics in contextual basis. c) Example: observe distribution of g-force when changing heading by more than

45 degrees at greater than 45 mph. d) Series of thresholds may identify an event that adds to the predictive power of an

existing loss cost model, in context of road type, sunlight or darkness, weather, road speed limit, etc.

3. Distinct differences between these two approaches a) Granular data allows identification new predictive variables more quickly b) Can identify risky driving behaviors and coached drivers to correct c) Can also correlate driving behaviors with fuel consumption.

4. Key: combine granular data and context to create more accurate loss cost models 5. Challenge for regulators: balance privacy protection against letting data inform and

improve models that can coach behavior to reduce loss costs, fuel consumption and deaths.

IV. Insurer, Consumer and Societal Benefits of Telematics-Based UBI (NAIC Staff)

A. Introduction Telematics when paired with UBI offers many potential benefits for insurers, consumers, and society

B. Insurer Benefits

1. More accurate risk assessment and pricing practices.

a) Collected driving behavior data granular predictor of risk

b) Better segment drivers by their risk indicators c) Premium rates, deductibles and coverage appropriate for each segment.

2. Variable pricing within existing classifications (such as age, annual mileage and territory) can be a much better pricing model a) Traditional classifications based on indirect aggregated variables of past trends

and events. b) Need variables which enhance not duplicate existing model predictability

22 NAIC Telematics


3. Tying driver behavior to pricing lets insurers better control risk exposure, a) May raise their risk tolerances b) Let them reach new customer bases. c) Charging drivers less for safer driving habits gives incentive to consumers to

improve their driving behaviors to get lower premiums. 4. Competitive advantages.

a) Can identify lowest-risk drivers, and raise their retention levels b) Gain new customers by offering all drivers chance to pay less c) Good for younger, riskier drivers: modify behavior, get discount

5. Telematics gives insurers new policyholder communication channels. a) Can use to build stronger relationships. b) Tying premium to miles driven encourages less driving, lowering loss costs.

6. Claims management practices a) With real-time driving data (such as hard breaking, speed and time) from

accidents, can estimate damages better b) Can reduce fraud and claims disputes. c) Improve accident response time d) Track and recover stolen vehicles e) Monitor driver safety.

7. Industry direction a) Some studies predict market > 25 percent telematics by 2020. b) Early adopters have a competitive advantage due to history of driving behavior

data

C. Consumer Benefits 1. Benefit most by having ability to reduce auto insurance costs.

a) Insurer participation discounts b) Improved driving performance or voluntary reductions in mileage driven.

2. Reasons to switch to telematics, according to survey a) Discount an incentive b) Ability to control premiums

3. Gives consumers some control of premium costs a) Telematics UBI programs convert part of fixed costs into variable b) More direct link between driving behavior/ usage and policy pricing. c) Consumers can therefore save money, more benefit to lower-income HH

4. Eliminates cross-subsidy between higher risk and lower risk drivers a) Benefits majority of consumers b) 63.5% of HH with insured vehicles would save average of $496 a year c) Those who do not save can reduce premiums with better driving habits.

5. Incentivizes better driving habits. a) Safer drivers become even safer b) Riskier drivers educated to modify high risk behavior. c) Appeals to households with young divers:

(1) Parents get feedback (2) Young drivers get coached

6. Consumers benefit when everyone drives more safely 7. More communication with insurers

a) Faster emergency response time b) Road-side assistance c) Stolen vehicle recovery d) Fuel efficiency e) Vehicle maintenance support.

NAIC Telematics 23


D. Societal Benefits 1. Telematics-based UBI programs: incentive to reduce the miles driven

a) Less road congestion b) Lower infrastructure costs c) Lower overall fuel consumption and vehicle emissions

d) Fewer accidents safer roads for all citizens.

2. Brookings Institute: tying insurance costs to miles ~ 8% reduction in VMT

a) Policyholders seek out alternative transportation options or skip unimportant trips$

b) Saves 50 to 60 billion each year: reduced accidents & road congestion. c) Reduce fuel consumption proportionally d) Greater reduction on carbon emissions (considers total refining process) e) Good for national security

3. UBI programs could increase number of insured drivers by creating more affordable auto insurance options. a) Consumers can adjust driving amount to the premium they can afford. b) Good for lower-income, who may not purchase auto insurance otherwise. c) Brookings: savings can be significant for lower income HH, where insurance

costs a significant amount of income

d) Lower auto premiums and fuel consumption lower total transportation

costs, especially important for lower income households (HH) 4. More socially equitable

a) Eliminates cross-subsidy between low-mileage and high-mileage drivers b) Common non-driving rating factors (depending on state): marital status,

occupation, educational attainment, credit score and homeownership. c) Factors statistically valid predictors of risk d) May penalize young, poor, seniors, urban residents etc e) But urban drivers usually drive fewer miles, so could pay less.

5. Telematics PAYD UBI programs delivery societal benefits a) Need to change consumer behavior. b) Customers must understand link between behavior and pricing –

(1) But some algorithms complex (2) Other algorithms proprietary

c) Society benefits from more transparent programs

V. Consumer Concerns and the Promise of UBI (Birny Birnbaum,* Executive Director, Center for Economic Justice) A. Introduction power to transform:

1. Auto insurance – improve pricing 2. Auto safety - empower consumer to modify driving behavior

B. Policy Goals 1. Consumers see two overriding public policy goals for insurance.

a) Ensuring all consumers have access to essential insurance products. (1) Essential financial security tools (2) Especially important for low-income consumer

b) Core institution for loss reduction and risk mitigation. (1) Through the risk classification system, insurance reduces loss of life

and property through incentives for less risky behavior and 2. Insurers uniquely positioned to accomplish these goal

24 NAIC Telematics


3. Consumer advocates have long pushed for pay-by-the-mile auto insurance, as a fairer way to price insurance

C. Bright Future for Consumers?

1. Real-time feedback regarding risky driving 2. In exchange for sharing the data, premiums should be based on miles driven and driving

behavior and not socio-economic rating factors 3. Wrong turn:

a) Have not created transparency in auto insurance pricing b) Have not created new opportunities for loss mitigation c) Instead another black box rating factor

4. Concerns about the current state of telematics include: a) Privacy issues and use and of data for purposes other than loss mitigation and

pricing, including using information in claim settlements when helpful to insurers but not when helpful to consumers.

b) Disproportionate impact of offer and sale of UBI against consumers in low- and moderate-income and minority communities.

c) Failure to achieve meaningful loss mitigation because of a black box approach by insurers of collecting data for rating.

d) Just another data mining exercise like credit— penalizing consumers not because of driving behavior but because of where and when they drive as a function of work and housing segregation.

e) Limited regulatory oversight to date

D. Pushing Ahead 1. Insurers are telling caution regulators not to impede innovation with telematics: Code for

do not regulate. 2. Consumer advocates have seen results of past innovation:

a) Massive abuses in credit scoring in the 1990s early 2000s b) Counting inquiries as claims in Comprehensive Loss Underwriting Exchange

(CLUE) databases in the 2000s c) Price optimization (price gouging) under the banner of management pricing

discretion in the 2010s. 3. Interests of insurers do not align with interests of consumers.

a) If they did, programs would feature transparency, protection of consumer privacy and d riving data.

b) Insurers compete on basis of risk classifications and keep methods secret. c) Insurers defeat key function of risk classification: to provide incentives for less

risky behavior and disincentives for more risky behavior.

E. A Failed Promise? 1. Consumer and public policy perspective: telematics a market failure. 2. To improve consumer use of UBI need better regulation 3. Regulatory framework should:

a) Establish data ownership and privacy standards b) Establish standards for permitted and prohibited uses of consumer data c) Collect/analyze granular data on offers and sales of UBI based related to

prohibited risk classification factors, including race and income d) Require insurers to include variables for race and income in generalized linear

models. e) Establish standards for disclosure of telematics results and rating programs to

ensure consumer receive feedback necessary to alter behavior.

NAIC Telematics 25


f) Replicate analyses presented by insurers in summary form, not just loss ratio as outcome variable, but other analyses using other outcome variables.

g) Stop fiction of discounts only unless and until the rating factor can be associated with lower overall claims and not simply a redistribution of income.

4. Regulation and competition are not inconsistent and regulatory efforts to would increase consumer confidence and grow the market

VI. Regulatory Implications of Telematics UBI (Sandra Castagna, Associate Commissioner, Maryland

Insurance Administration (Retired))

A. Introduction 1. Insurance companies underwrite and price risks, and pay claims based on data. 2. Telematics-based UBI programs mean more data gathering than ever 3. The implementation of these programs and the collection, analysis and use of the data

present regulatory concerns and provide an opportunity to propose action to address them.

B. Data Collection 1. When first introduced to regulators, telematics-based UBI simple enough: mileage on

device determined PAYD discounts, from one percent (just for participating) to a maximum of 30% (very few miles driven).

2. Arrangement appeared straightforward, was understood by consumers, and any discount to the policy premium was easily computed.

3. Now have many technological methods, capture much more information, miles driven, and when, where and how, in many formats – now much more complicated

C. Technology Concerns (for regulators)

1. Method used by insurers to record, transmit, receive and report driving data. a) Third party more likely than in house b) Questions concerning third parties?

(1) How is data processed before being forwarded to insurer? (2) Does vendor scrub data for accuracy? (3) How will it be formatted, stored and protected from misuse by

internal and external actors? (4) Method(s) requires disclosure to and understanding by regulators

2. Different equipment based on the make and model a) May not record same data or record data in same manner b) Irrespective of method, regulators should confirm same data obtained for every

program participant c) Confirm all potential discounts made available to all participants who meet

established criteria. 3. Frequency and duration of data transmission another issue:

a) Some devices record/transmit data every 30 seconds or less b) A great deal of information is captured c) Some UBI programs collect continuously throughout the policy term, d) Others may limit to a specific period of time, such as 30 or 60 day – what is

credible? e) If data captured for shorter periods, complete, consistent measurement is

imperative.

D. The Need for Transparency 1. Credit history, occupation and education vs. driving as rating factors

a) Credit etc., questioned and in some states, limited or prohibited

26 NAIC Telematics


(1) Not transparent as a predictor of loss (2) Concern unfairly discriminatory

b) Telematics less controversial (1) Driving factors being measured (2) Driving behavior considered accurate predictor of risk.

2. Consumer and willingness to provide personal driving data for lower premiums a) Some more concerned with privacy b) Others OK as long as know what “good driving behavior” is c) If don’t understand measurement of good driving behavior, just another

untrustworthy black box 3. If make data-related information available to regulators and policyholders insurers could

demystify their telematics programs. a) Could be disclosed to regulators in the filing b) Could be disclosed to insureds via a UBI participation agreement. c) Insurer best practices and participation agreements should include:

(1) Which driving factors measured and why (2) Insureds have right to accept terms

4. Transparency can be improved by a) Access to apps and websites that track driving history and identify improvements

insureds can make b) State in agreement other data uses (sharing for marketing, claims, government)

5. Information about data processes should be available for filings with UBI. a) May seem irrelevant to rating but needed for support for discounts b) One-page reports generated by third-party vendors may or may not be sufficient

to support the application or removal of a UBI discount. c) Questions pertaining to underlying data should be asked during filing review

process to avoid later compliance issues: (1) Can data be retrieved easily when required? (2) Is it being secured safely in a protected environment? (3) Will it be retained in accordance with regulations?

E. Rating Considerations 1. Challenge for regulators:

a) Understand how recorded driving data is predictive of loss and reflected in the insurer’s rates

b) No factors prohibited by statutes c) Rates may not be inadequate, excessive or unfairly discriminatory

2. Good driving should result in lower premiums a) But what about time of day? b) People who work evening shifts in hospitals? c) Those with low income, education, credit? d) Actuarially sound or unfair discrimination?

3. Insurers and/or vendors use generalized linear models (GLMs) to quantify characteristics for better driving and lower losses. a) Rating models or algorithms may be defined as supplementary rate information,

subject to filing requirements b) Insurers may protest that models contain confidential information c) But without review, cannot tell if rates based upon them are compliant.

(1) What assumptions made regarding driving factors measured? (2) When considering specific driving data (braking & turning & speed),

what combination presents the least likelihood of loss and will result in greatest premium discount?

NAIC Telematics 27


4. Telematics-based UBI programs: tremendous amounts of data a) Insurers can identify and develop more granular risk classes. b) More complex models, nuanced rating plans and individualized rates c) Regulators must review data and rating plans for compliance, but easier said than

done in file-and-use or use-and-file regulatory environment. F. Availability and Affordability

1. Availability/affordability of auto insurance frequently studied/debated by many a) Various insurance industry groups b) Consumer groups c) Insurance departments and the NAIC. d) NAIC Auto Insurance (C/D) Study Group, joint working group of the Property

and Casualty Insurance (C) Committee and Market Regulation and Consumer Affairs (D) Committee, is studying the affordability of automobile insurance as it relates to low-income insureds.

e) Federal Insurance Office (FIO) requested comments on same subject. 2. Insurers say UBI programs can make auto insurance more affordable.

a) Discounts related to driving behavior are made available, b) Insurance premiums are reduced by demonstrating safe driving behavior and,

therefore, coverage becomes more affordable. c) Those with higher than average premiums can benefit most d) Maintaining competitive markets and providing policyholders with increased

options leads to premium savings will automatically ensue. 3. Regulators still have concern

a) Acknowledge merit to insurer position b) But still concerned about cross subsidization and certain rating factors c) Application of a discount to discriminatory rate does not substitute for

appropriate classifications of risk and actuarially sound rates G. Claims Management

1. Benefit of telematics: reduction in the frequency and severity of claims. a) Theory: people modify behavior when observed, so will drive better b) Better driving usually means fewer accidents.

2. GPS-enabled information a) Reduction in theft b) Reduction in fraud by combining driving data with mapping technology,

3. Insurers must establish protocols for use of telematics data a) Must not violate state unfair claims settlement practices acts. b) Must consider data consistently or invite administrative action. I c) Information may support denials or acceptances of claims

H. Next Steps 1. Challenges presented here

a) Neither inconsequential, nor insurmountable but deserve attention b) Vast amounts data now incorporated into rating plans c) Regulators must determine if rating plans comply with rating laws

2. Technological innovation will continue to affect rate development. a) Telematics began with PAYD and evolved into PHYD. b) Devices record data but may not know actual operator. c) Insurers and rating organizations already overlay multiple models, d) Combination could contain prohibited factors or produce rates that are unfairly

discriminatory. 3. If using UBI requires sale of data, is insurer engaged in unfair discrimination if

insureds in certain areas receive discounts or other promotions, and others do not? 4. When regulators see “telematics” in a rate filing, must ask questions.

28 NAIC Telematics


PAST CAS EXAMINATION AND REVIEW QUESTIONS

1. a. Describe two ways in which usage-based insurance, such as telematics, might unfairly

discriminate against lower income and protected classes of individuals.

b. Describe two ways in which usage-based insurance might benefit lower income and protected classes of individuals.

c. Describe two criticisms of usage-based insurance, other than the criticism that usage-based

insurance may be unfairly discriminatory against lower income and protected classes of individuals. (17S-6US-1-3:1/1/1

R1. What is PAYD and PHYD and what has enabled the progression of one to the other? R2. What, according to consume advocates, do insurers really mean when they tell regulators not to impede

innovation? R3. According to the Center for Economic Justice, consumers were eager for real-time feedback regarding

risky driving and ii exchange for sharing their data with insurers, having their premiums be based on miles driven and driving behavior instead of socio-economic factors. However, the Consumer for Economic Justice says that telematics-based UBI has not lived up to its promise. Describe two of the problems.

R4. Describe three potential benefits to society, other than lower insurance costs, that could be facilitated by

telematics UBI programs. R5. Describe three ways that telematics could lower insurance loss costs R6. The innovation of using telematics in auto insurance is frequently compared to the innovation of using

credit reports in auto insurance. a. Describe two similarities between credit and telematics b. Describe two differences

R7. Individual driving data can be used to determine what has happened during an accident.

a. What are consumer advocates concerned might happen with this data? b. What do regulators recommend?

NAIC Telematics 29


R8. As many insurance companies rely on third parties to gather driving behavior data, what questions should be raised about the data from these third parties? Describe two.

R9. Give two reasons why UBI programs are of particular interest to households with young drivers. R10. Describe five ways that consumers could benefit from UBI R11. Describe three competitive advantages that insurers might get from using telematics R12. There are two main predictive models being used with driving behavior.

a. Describe the two different types of models b. Describe two differences between the models

R13. What is ACORD trying to do regarding the data?

30 NAIC Telematics


SOLUTIONS TO PAST CAS EXAMINATION AND REVIEW QUESTIONS 1. a. Any two of the following (more possible):

Disproportionate impact of offer and sale of UBI against customers in low/moderate income and minority communities.

Telematics may find that people who drive at the early hours of the day present more risk. This could impact low-income insureds who work night shifts.

Telematics may find that people who drive in urban areas present more risk. When the telematics device is purchased by the consumer, those with low- income may not be able to

afford it. Low-income drivers are more likely to operate older vehicles that may not be able to use the telematics

device. Low-income drivers are more likely to operate vehicles in poorer condition or that don’t have new

automobile features. This could result in harder braking and more swerving.

b. Any two of the following (more possible): Progressive’s Snapshot does not use GPS, so urban drivers are not penalized by their urban location. UBI removes the subsidy for high-mileage drivers who account for the majority of miles driven but pay a

disproportionally lower premium. This increases affordability for lowermileage drivers many of whom are lower-income or elderly. This makes insurance more socially equitable.

UBI can potentially reduce the number of uninsured motorists by making insurance more affordable. Insureds are encouraged to become better drivers through the telematics program; a reduction in premium

would have a disproportionately positive impact on low-income drivers. The GPS technology in the telematics devices would help out in case of car theft, which is more likely to

happen in poorer neighborhoods. The UBI data could supplant or even replace other variables such as credit score, education, and

occupation, which are accused of being unfairly discriminatory. Seniors tend to drive during the day and not during commute times which could decrease premiums. The data insurers collect from UBI could speed up the claims process, making payments out to the

insureds quicker. This would be particularly helpful for low-income insureds that may need a claims payment immediately, especially if their vehicle is unusable as a result of the claim’s event.

Youthful (teen) drivers would have their driving habits tracked. If their parents have access to this data it could help keep them out of trouble

c. Any two of the following:

Telematics-UBI is just another black box rating factor for insurers. By keeping these black box items secret it defeats the key function of risk classification.

Limited regulatory oversight Using a “discount-only” model for installing the devices until a rating factor can be associated with lower

overall claims. Reliance on a third-party vendor which raises issues of privacy, data accuracy, and misuse of data by

internal and external parties. Use of different vehicle input devices may lead to issues where the devices may not record the same data.

Some devices are only available for newer vehicles. Inconsistent frequency and duration of data transmission to the insurers If insured is not privy to detailed information regarding the rating factors being measured and their

relationship to the receipt of the discount it is less likely that changes in driving behavior will result or premium reductions will be achieved.

Driving behavior is not always linked to the actual operator.

NAIC Telematics 31


R1. PAYD is an acronym for Pay As You Drive and has premium influenced by the number of vehicle miles

driven. PHYD is an acronym for Pay How You Drive and is based on the idea that it is not just the amount you drive, but how you drive that matters. Because of the increasing granularity of telematics, insurers can measure things like acceleration, cornering and braking.

R2. According to Birny Birnbaum, the Executive Director, Center for Economic Justice, “do not impede

innovation is simply code for “do not regulate” R3. Two of the following (more possible):

Have not created transparency in auto insurance pricing Have not created new opportunities for loss mitigation Instead another black box rating facto

R4. Three of the following (more possible):

Less road congestion Lower infrastructure costs Lower overall fuel consumption and vehicle emissions

Fewer accidents safer roads for all citizens

R5. Three of the following (more possible)

Driver behavior feedback encourages better driving and thus lowers the frequency and severity of accidents

By encouraging people to drive less, there are fewer accidents GPS makes it easier to track vehicles and so thefts are reduced Telematics can give better information about accidents and discourage fraud

R6. a. Two of the following (others possible):

Credit reports and telematics often rely on third-party vendors Credit ratings and telematics both have the reputation of being like a black box to consumers and

sometimes to regulators

b. Two of the following (others possible) Telematics is heavily dependent on changes in technology, whereas credit does not Telematics allows relevant driver feedback Although how telematics is used may be controversial, how much one drives and how well on drives

is less controversial than using socioeconomic variables for rate classifications R7. a. Consumer advocates are concerned that the insurance company (which has access to the data)

would only use the data when the results benefit them, in other words, to deny claims b. To avoid unfair claim settlement practices, regulators recommend that insurance companies come

up with consistent protocols for handling this data in all claims

32 NAIC Telematics


R8. Two of the following: How is data processed before being forwarded to insurer? Does vendor scrub data for accuracy? How will it be formatted, stored and protected from misuse by internal and external actors?

R9. Two of the following:

Can give parents feedback about kids’ driving Can incentivize safer driving (can be easier to get this information from a neutral party than a parent) Can potentially lower premiums

R10. Five of the following (others possible):

Benefit most by having ability to reduce auto insurance costs. Improved driving performance or voluntary reductions in mileage driven. Gives consumers some control of premium costs Eliminates cross-subsidy between higher risk and lower risk drivers Incentivizes better driving habits. Safer drivers become even safer Riskier drivers educated to modify high risk behavior. Appeals to households with young divers: Parents get feedback & young drivers get coached Consumers benefit when everyone drives more safely Faster emergency response time Road-side assistance Stolen vehicle recovery Fuel efficiency Vehicle maintenance support

R11. Three of the following (others possible)

Can identify lowest-risk drivers, and raise their retention levels Gain new customers by offering all drivers chance to pay less Good for younger, riskier drivers: modify behavior, get discount Can use to build stronger relationships. Tying premium to miles driven encourages less driving, lowering loss costs.

R12. a. (1) Less granular – predefined set of events, such as total mileage, time of day, events such as harsh

braking, acceleration or cornering. Assumes these events can predict loss costs (2) Much more granular data about vehicle use with second-by-second basis or more granular as

needed for accelerometers; research predictive power of characteristics in contextual basis. b. Distinct differences between these two approaches

Granular data allows identification new predictive variables more quickly Can identify risky driving behaviors and coached drivers to correct Can also correlate driving behaviors with fuel consumption

R13. Standardize it.

ACTEX Study Manual for CAS Exam 6 U.S. · ACTEX CAS Exam 6 U.S. Study Manual, Spring 2018 Edition ACTEX is eager to provide you with helpful study material to assist you in gaining

Documents