Consumer Organizations Call on Federal Insurance Office to Update Auto Insurance Affordability Study and Dig Deeper Into Sources of Unfairness in Market

WASHINGTON, July 28 (TNSRes) (TNSCapV) -- The Consumer Federation of America issued the following joint news with 19 organisations:

In a comment letter sent to the agency on Monday, July 26, Consumer organizations urged the Department of Treasury's Federal Insurance Office (FIO) to update a 2017 study on the affordability of auto insurance in historically underserved communities and to conduct a deep and thorough investigation of unfair discrimination in auto insurance markets. While state laws in all states but New Hampshire require drivers to purchase auto insurance, it is often unaffordable for lower-income drivers. Industry pricing practices that rely on non-driving, socio-economic characteristics of customers disproportionately penalize people of color with higher premiums and fewer choices in the market.

Citing President Biden's note from a February 2021 town hall that "[i]f you live in a black neighborhood, you're going to pay a higher premium on your car," the 22 consumer and community organizations submitting the comments wrote:

As the only product that most Americans are required to purchase by law, it is particularly important that auto insurance is available, affordable, and priced fairly in the marketplace...Even drivers with unblemished driving records may find that the cost of coverage in their community and for people with their socio-economic characteristics far exceeds their family budget...

The groups argue that FIO should prioritize an immediate update of its 2017 study that found that about 18 million Americans live in ZIP codes where auto insurance is unaffordable, and they urged FIO to produce this report annually as was originally proposed. The report, however, was not updated at all during the prior administration. In addition to updating the 2017 study, the groups urged FIO to produce additional complementary research, including:

* collecting data about uninsured drivers, drivers with force-placed auto insurance coverage, and enforcement of mandatory insurance laws;

* analysis of insurers' use of socio-economic factors -- including employment status and occupation, level of education, homeownership status, credit history, and marital status -- and other non-driving data in auto insurance marketing and pricing as well as claims handling and anti-fraud efforts; and

* testing underwriting and rating factors for disparate impacts on communities of color.

* * *

July 26, 2021

To: Alex Hart, Senior Insurance Regulatory Policy Analyst, Room 1410 MT, Department of the Treasury, 1500 Pennsylvania Ave NW, Washington, DC 20220

Submitted electronically

Re: Consumer and Community Organizations' Response to Federal Insurance Office Request for Information: Monitoring the Availability and Affordability of Auto Insurance: Assessing Potential Evolution of the Auto Insurance Market Dear Mr. Hart:

The following comments of the undersigned organizations address the Federal Insurance Office (FIO)'s request for information (RFI) concerning the updating of the 2017 FIO Auto Insurance Affordability Study and other research topics and methodologies for FIO to consider regarding auto insurance affordability and availability.

Auto insurance is unique. As the only product that most Americans are required to purchase by law, it is particularly important that auto insurance is available, affordable, and priced fairly in the marketplace. However, the market for auto insurance is, as FIO's 2017 study illustrated, often not affordable to low- and moderate-income Americans and people of color. Even drivers with unblemished driving records may find that the cost of coverage in their community and for people with their socio-economic characteristics far exceeds their family budget. This problem runs headlong into the dual public policy goals of ensuring that all drivers maintain coverage and that the market for coverage is fair and affordable. In these comments we lay out measures that FIO can take to provide more information about the state of the auto insurance market and address inequities such as the one President Biden highlighted at a February 2021 CNN Townhall when he noted that "[i]f you live in a black neighborhood, you're going to pay a higher premium on your car."/1

The 2017 FIO Auto Insurance Affordability Study provided a first-of-its-kind and useful snapshot of the auto insurance market in some American communities that have been historically underserved by the insurance industry or have otherwise faced barriers to obtaining coverage. While there are other important research projects regarding availability and affordability of personal auto insurance that FIO should undertake, FIO's top priority should be to update the 2017 study for the following reasons:

1. An update is relevant - FIO can identify changes in affordability compared with the 2017 results using a consistent methodology.

2. An update is timely - there has been much public discussion of historic and persistent inequities in financial services - and FIO could complete an update this year.

3. The infrastructure - methodology and data collection process - for updating the study is in place.

4. An update does not compromise other research by FIO into personal auto insurance issues or the development of a more robust data collection and analysis in the future.

To be clear, it would not be sufficient for FIO to simply issue such an update and conduct none of the other research contemplated in the RFI. We urge FIO to recognize that our interest in seeing the regular updating of the 2017 study does not diminish the priority we place on the more robust research we will discuss in Part 2 of our comments below.

Part 1. Update the 2017 Auto Affordability Study Annually

With respect to updating of the 2017 study, the chief value of updating it is in pursuit of a longitudinal study of auto insurance affordability in America. As it is a straightforward report,/2 this study could be updated annually, as was suggested in the introduction to the original 2017 study./3 Over time, this dataset will:

* document changes in the affordability and availability of auto insurance in the American marketplace;

* help identify communities where problems with affordability are persistent, as well as communities where the problem is growing or receding; and

* help interpret the impact of changes in law, regulation, and the marketplace on historically underserved communities.

There are two aspects of the 2017 methodology that should be improved in order to provide a better analysis of the underlying research questions about affordability in certain underserved and vulnerable communities. Nothing in these suggestions alters the comparability of the data with the 2017 study, and they would increase the long-term value of this regularly updated report. We offer the following improvements:

1. FIO should collect more complete data from insurers to better implement the availability and affordability methodology. As explained in the July 2015 Notice advising of FIO's methodology, the original report relied on a limited set of data. FIO planned, for its next report, to "request data from insurers who have a statutory surplus greater than $500 million as of December 31, 2015, and who annually collect more than $500 million of premium for personal automobile insurance."/4 For the next iteration of this study, the identification of insurers to report their experience should be expanded and streamlined in the following ways:

a. "Insurers" should explicitly refer to an insurance group and not individual insurers;

b. the reporting threshold should be $100 million for an insurance group, and should include all companies in the group that write private passenger auto insurance;

c. the surplus criterion should be eliminated. Some insurers who write direct personal auto insurance serve as fronting companies for other insurers or reinsurers. As such, these insurers write the policies but maintain little ongoing exposure and, consequently, require and have little surplus; and

d. FIO should rely upon its statutory authority to require reporting by companies to ensure more comprehensive experience reporting.

2. FIO should collect data concerning the average premium for comprehensive and collision coverage in addition to the premium for liability coverage at the financial responsibility limits required by each state. An estimated 79% of policies include comprehensive coverage and 75% include collision coverage.5 For the many millions of low- and moderate-income Americans who hold car loans, auto insurance affordability cannot be measured by the cost of liability coverage alone, and the FIO report should include in its analysis data indicating how the need to purchase physical damage coverage impacts the affordability of auto insurance.

Part 2. Produce a Robust Analysis of the Auto Insurance Market, Particularly

as It is Experienced by Underserved Communities

Having noted the value of a discrete, reproducible study based on the 2017 report above, we now turn to the further inquiry and analysis that FIO should pursue separately in its effort to better understand and describe the auto insurance market, particularly with respect to lower-income drivers and communities of color. Below we present three avenues of complementary research into affordability and availability of auto insurance and then address the two additional areas of inquiry FIO has outlined in questions 3 through 12 of the RFI, namely:

* Non-Driving Related Factors in Personal Auto Insurance Underwriting and Pricing; and

* Structural Market Changes in Personal Auto Insurance

A. Complementary Research into Affordability and Availability

The longitudinal study will serve an important role in identifying affordability trends in the auto insurance market, but it does not paint a complete picture. Before we address the questions regarding the use of certain rating factors and market structure changes, we present three paths of inquiry - other indicators of availability and affordability - that would complement the 2017 study update.

(1) Prevalence of Uninsured Motorists. Higher levels of uninsured and underinsured motorists in a community suggest less affordability and availability of auto insurance. However, there is no publicly available research that details uninsured driving at any territory smaller than the state level. /6 Comparing Uninsured Motorist Bodily Injury (UMBI) claims in a ZIP code (or smaller geographic territory) with the total of bodily injury and UMBI claims would provide further insights about where auto insurance affordability poses the most severe challenges to drivers and how public policy responses to uninsured motorists might be altered to improve access to coverage. For example, if higher levels of uninsured motorists are concentrated in lower-income communities, this would suggest the value of a reduced coverage limits policy for low-income drivers akin to California's Low-Cost Auto Insurance Program./7

(2) Prevalence of Uninsured Motorist Citations. Collecting data on the variety of penalties imposed on drivers for failing to maintain insurance and the geographic distribution of the citations would provide multiple insights related to the auto insurance market. These data would add to an assessment of the level of uninsured driving in different communities and could shed new light on disparate policing practices that may lead certain communities to face enforcement of state financial responsibility laws more often and with more severe penalties.

(3) Prevalence of Lender-Placed Auto Insurance (LPI). The placement of coverage by lenders on vehicles for which they have provided a loan would provide more information about both the level of uninsured motorists (but for LPI) and the broader implications of unaffordable private passenger coverage. Noting that LPI can be very expensive, determining the percentage of vehicles in a ZIP Code or smaller geographic territory with LPI would add new insights into a way in which the unaffordability of auto insurance in underserved communities leads to even higher costs for impacted consumers.

(4) Disparate Impact Study. FIO should carry out a comprehensive and in-depth analysis of the use of underwriting and rating factors in the selling and pricing of private passenger auto insurance to assess the extent to which the use of certain factors produces a disparate impact on communities of color in America.

B. Non-Driving Related Factors in Personal Auto Insurance Underwriting and Pricing

i. Socio-economic, Non-driving Factors

As the RFI notes, auto insurers use several non-driving factors in underwriting and pricing. These factors - which include a customer's employment status and occupation, level of education, homeownership status, credit history, and marital status - are closely correlated with demographic characteristics, and insurers' use of these non-driving factors disproportionately harms people of color. Black, Latinx, and Native Americans are disproportionately represented in the tiers of each of these non-driving factors that face the highest premium penalties. These socio-economic factors, whether intentionally or not, are proxies for income/wealth and race/ethnicity. Consumer Federation of America (CFA) has conducted a number of studies on the plight of low- and middle-income consumers and people of color who confront auto insurance pricing models that include such proxies./8

This concern, however, should not just be limited to underwriting and rating. Auto insurers have, in recent years, utilized a growing number of new data sources, some of which are known to the public and some of which are not, for marketing, claims settlement, and anti-fraud efforts. Extending the analysis beyond pricing to marketing, for example, is critically important because insurers' ability to use new - and unregulated - data sources to micro-target consumers means that insurers can identify "high-value" and "low-value" customers before the application process begins and channel certain consumers toward affiliates with limited or higher-priced offerings.

We suggest that FIO start this non-driving factors line of inquiry by reviewing data collected by the National Association of Insurance Commissioners (NAIC) regarding insurers' sources and uses of data for personal auto and, if necessary, issuing a survey to insurers to fill in any gaps. As part of the collection of sources and uses, FIO should identify which data and sources are subject to Federal Credit Reporting Act (FCRA) requirements and which raise concerns about discriminatory impacts - proxy discrimination and disparate impacts - in traditionally underserved communities. FIO should also identify and review studies that have examined these issues, such as, for example, the studies by CFA cited above. By gathering existing research into the use of socio-economic factors and obtaining and analyzing non-public information regarding use of new data sources for marketing, claims settlement and anti-fraud, FIO would provide a valuable resource for future study.

Notwithstanding the meta-analysis described above, we also urge FIO to deploy resources in support of a disparate impact study that tests whether and how these rating factors disproportionately harm people of color. Further, if FIO determines it has the capacity to conduct its own collection regarding the use of non-driving factors, we suggest that FIO develop a matrix of drivers with different socio-economic characteristics but similar driving characteristics and gather data that will determine the extent to which socio-economic characteristics influence the cost of auto insurance. Following are three approaches to filling out this matrix that FIO might consider:

a) FIO could order insurance companies of a certain size - consider the $100 million annual premium threshold discussed above - to fill out a prescribed matrix for one FIO-selected ZIP Code in each state./9,/10 Using the rate plan and manual in effect on a certain date, each company would provide the premium quote that would be offered to a driver who fits the socio-economic description in each box of the matrix, with all other characteristics (e.g., driving safety record, type of vehicle) held constant.

b) FIO could acquire premium data derived from company rate and rule filings from a third-party vendor that already collects these data. Using such a vendor would reduce the challenge of data collection and reduce the data management burden of this research, while providing FIO with a large data set - based on the filed rate plans of insurers - that could be used to fill out the socio-economic matrix described above.

c) FIO staff could conduct "secret shopper" testing of rates using insurance company websites to get a range of quotes for drivers by varying the non-driving factors during the quote questionnaire process. This would be limited to those carriers that provide online quotes and would require a significant deployment of staff time.

ii. Geographic/Territorial Rating

The ZIP code in which a driver lives, and sometimes an even smaller territory known as a census tract block group, is one of the most significant determinants of a driver's premium, and often the most significant. We address geographic rating separately from the other socio-economic factors because a) there is a more directly causal relationship between geography and exposure to risk than is evident with the socio-economic factors, and b) geography is laden with very specific connections to structural racism in American society and history.

ZIP codes are associated with financial and social status in ways that cannot be ignored when looking at auto insurance through the lens of racial equity and economic justice. But, unlike the non-sequitur relationship between non-driving factors, such as one's job title, and driving risk, it is not unreasonable to assert that different driving environments create different exposure risks. Driving in the city center leads to more fender benders, while driving on the open roads of more rural areas results in fewer but more severe accidents.

However, this does not mean that the current application of territorial rating serves consumers well or fairly. FIO research in this regard would be of great value to the public, policymakers, regulators, and insurers. Indeed, at his CNN Town Hall meeting in February 2021,/11 President Biden expressed the problem at the heart of territorial rating:

I don't know what home you live in, but if you go ahead and you want to get insurance, and you're in a black neighborhood, you're going to pay more for the same insurance that I'm going to pay for the exact same home. Your car -- you never had an accident in your car. If you live in a black neighborhood, you're going to pay a higher premium on your car.

These comments, which reflect concerns that have been aired by residents of Black communities for decades, are borne out in research. CFA has shown numerous instances where living on the "wrong" side of a ZIP code boundary can increase auto premiums dramatically (including liability premiums), even when drivers in the adjacent neighborhoods will be driving the same roads during their daily commutes./12 Understanding the breadth of this problem and any patterns to it are crucial to addressing it.

Using data from an expanded version of the data collection described in Part 1 of these comments, FIO could provide clarity about the severity of the problem President Biden described. Matching premium data collected at the ZIP code or census tract level to demographic data regarding race and ethnicity would provide important insights into the extent to which territorial rating leads to higher premiums in communities of color.

By applying mapping techniques, FIO's research could reveal the extent to which current territorial rating mimics or differs from the old lending redlining maps that played major role in enforcing structural racism. FIO would also be able to determine whether there were significant changes in rates among geographically proximate but demographically different neighborhoods. Determining if there are patterns to this, or concentrations of the problem in one type of setting and not another (or, for example, in some cities and not others), would help move forward the long-standing and contentious discussion around geography and auto insurance that the President understands and echoed in his comments.

Given the discussion above, FIO should, as part of its auto insurance research outside of the updating of the 2017 study, examine territorial rating, including:

* the granularity of current insurer rating practices and the size of premium differences across adjacent rating territories;

* the correlation of current geographic rating with demographic characteristics, including race and ethnicity;

* the correlation of current geographic rating with historical lending redlining practices;

* geographic rating for uninsured motorist coverages to determine if communities with less availability and affordability of auto insurance are further penalized with high UM/UIM premiums; and

* a review of any state laws or regulations that aim to mitigate the disparate impact, affordability, or availability problems that arise from territorial rating.

With respect to question six of the RFI about the propriety of the socio-economic rating factors, we believe that there are two central requirements that must be met before any auto insurance underwriting or rating factor can be used: the factor must have actuarial value, and it must meet public policy goals and constraints. Importantly, a factor that satisfies the first requirement but not the second should be treated as an unacceptable factor and should be prohibited.

The actuarial test must show that the factor is a reasonable method for capturing information with a meaningful relationship to risk of loss. The public policy test must recognize that virtually every state, by law, has the dual public policy objectives of a) universal coverage for all drivers (with severe penalties for those who do not comply) and b) no unfair discrimination./13 This means that actuarial support cannot overcome practices that create barriers to coverage for people in a manner that disproportionately harms people of color and lower-income individuals or discriminates against any consumer on the basis of a protected class.

The non-driving socio-economic rating factors discussed above do not overcome this public policy standard, so they are not viable, whether or not they exhibit some relationship to risk of loss (a fact that is, in itself, disputed). With respect to the use of geography, it is apparent that a disproportionate burden exists, and we believe it needs to be addressed. Unlike the other non-driving factors, we believe there may be ways to minimize the disproportionate impacts of territory while still recognizing the regionality of risk. FIO research could present opportunities for greater discussion of strategies with this aim, such as limiting premium variation among contiguous ZIP codes, increasing the minimum size of territories, and using territory factors that combine local territorial risk and statewide risk to account for, among other things, the effects of commuting into a city from a suburb or exurb.

Having a national picture of the use and impact of these factors will provide a firm foundation for future policymaking concerning these aspects of auto insurance underwriting, rating, and other insurance processes in which non-driving factors are being used.

II. Structural Market Changes in Personal Auto Insurance

In several questions of the RFI, FIO solicits input about the impact of changes in the auto insurance market and their impact on consumers. Below we discuss some of those for which FIO research could benefit consumers.

A. The Impact of the Pandemic on Auto Insurance Consumers.

As it did with so many segments of the economy, the COVID-19 pandemic exposed substantial weaknesses in auto insurance regulation and an unwillingness among insurers to take sufficient steps of their own to address the impact of the pandemic on consumers. In fact, the pandemic produced a sizeable increase in profits for auto insurers. Due to stay-at-home orders, closures, and the pandemic's economic impact, driving plummeted, along with auto crashes and claims. While most auto insurers returned some premium in the forms of refunds to consumers, these givebacks were woefully inadequate; the refunds and premium credits averaged about 15% per month and generally lasted for only two months during spring 2020. Only California, Michigan, New Jersey, and New Mexico mandated premium refunds; only California required refunds for the duration of the pandemic; and earlier this year, California's insurance commissioner notably reported that companies' refunds were not sufficient and ordered more to be returned./14

This failure to return the excess premium generated by the pandemic not only should be addressed by state regulators (particularly since "excessive rates" are illegal in almost every state) - in recent weeks insurance commissioners in Washington and New Mexico have joined California in calling for further review of pandemic premiums/15 - but it also should be the subject of federal inquiry. FIO should use the experience of the pandemic to help prepare policymakers, regulators, and industry to respond to national crises in the future. In particular, FIO should articulate expectations of and guidance for providing appropriate refunds when exposure to risk falls precipitously and for extended periods of time. FIO should also evaluate the extent to which carriers allow consumers to make mid-policy changes to their anticipated mileage, so they can get premium reductions based on their individual change in risk. Better understanding of the willingness or unwillingness of insurers to adapt to sudden declines in risk exposure will help policymakers determine whether new laws or rule changes are needed to ensure that consumer premiums reflect their actual risk, especially in the wake of a crisis such as this pandemic.

B. The Potential for Expansion of Auto Insurance Telematics Programs

Related, to some extent, to the transformation of driving habits during the pandemic is the expansion of usage-based insurance, or telematics, in the market. As drivers have reduced their driving, insurers have increased their telematics offerings, though it is not clear whether this apparent expansion is in response to the pandemic or simply coinciding with and made more urgent by it. Through these telematics programs, insurers obtain consumer-generated driving data from automobiles via mobile apps and devices. These programs have the potential to reduce auto insurance premiums, shift pricing away from non-driving factors to factors more indicative of risk, and to encourage safer driving behavior. However, telematics also poses risks of unfair pricing, abuse of personal consumer information, racial and ethnic discrimination, and data insecurity, among other concerns.

Most states do not have adequate regulations and oversight of these programs. California has regulations for "pay as you drive" programs that only allow for the collection of mileage. New York has adopted guidelines for telematics programs, but most states have no specific laws or regulations governing telematics. Research by FIO that would develop a clearer picture of all the data collected by insurers and how the data are used both for insurance and non-insurance purposes would help policymakers and regulators determine how state and federal rules governing insurance premiums, data sharing, and consumer premiums should be adapted to better protect consumers as telematics programs gain traction in the market. FIO should also seek to determine if, and the extent to which, the use of telematics is supplanting the use of non-driving factors or being used in conjunction with those factors.

As a starting point, we share the following standards for telematics oversight, which were developed by CFA in its recent white paper:/16

* Require insurers to demonstrate and explain the actuarial basis for data that are collected and used as part of telematics programs;

* Require insurers to obtain informed consumer consent before they use consumers' data and prohibit insurers from using, selling, renting, or sharing that data for non-insurance purposes;

* Require insurers to test for and minimize disparate impact on protected classes such as race and ethnicity in telematics programs;

* Ensure that third party vendors who develop and participate in telematics are well regulated and subject to proper scrutiny; and

* Require insurers to protect consumer privacy and make sure telematics programs have strong consumer protection standards.

Conclusion

FIO, through its Request for Information, has rightly identified the affordability and availability of auto insurance as a critical avenue of inquiry, one that was prioritized by the agency's founding legislation and identified by President Biden. In 2017, with its initial affordability study, FIO issued what was then thought to be the first of an annually updated study about the cost of auto insurance in America's underserved communities. That it was not updated over the past four years increases the importance of getting that research back on track by updating the report immediately and enshrining it as an annual report.

Therefore, we propose that FIO take two paths in the wake of this RFI: first, updating the 2017 study and, separately, conducting more expansive research into various aspects of the auto insurance market. While the extent of this secondary research will depend upon agency capacity, we have outlined a variety of avenues of study including research regarding uninsured motorists; socio-economic rating factors; territorial rating; the use of non-driving characteristics in marketing, claims handling, and anti-fraud processes; best practices for addressing suddenly excessive rates associated with the pandemic; and the opportunities and potential pitfalls associated with growth in the use of auto insurance telematics.

Whether FIO consider some or all of these items, we ask that consumer and community organizations continue to be included in the decision-making process and the development of research methodologies. Please contact Michael DeLong of Consumer Federation of America ([email protected]) with any questions about these comments, or if you would like contact information for any of the organizations signing below.

Sincerely,

Center for Economic Justice

Center for Justice & Democracy

CPAN

Consumer Action

Consumer Federation of America

Consumer Federation of California

Consumer Reports

Consumers for Auto Reliability and Safety

Demos

Georgia Watch

Latino Action Network

Los Angeles County Department of Consumer and Business Affairs

National Community Reinvestment Coalition

New Yorkers for Responsible Lending

Oregon's Stop the Debt Trap Alliance

Texas Appleseed

The One Less Foundation

United Policyholders

U.S. Public Interest Research Group

Vehicles for Change

Footnotes:

1/ "Remarks by President Biden in a CNN Town Hall with Anderson Cooper." February 16, 2021. Available at https://www.whitehouse.gov/briefing-room/speeches-remarks/2021/02/17/remarks-by-president-biden-in-a-cnn-town-hall-with-anderson-cooper/.

2/ In essence the study consists of 1) the identification of ZIP codes that are considered vulnerable to being underserved; 2) a calculation of the average auto insurance premium in each of those ZIP codes; and 3) the calculation of the average premium as a percentage of ZIP code median income with those above two percent deemed unaffordable.

3/ Study on the Affordability of Personal Automobile Insurance." Federal Insurance Office, U.S. Department of the Treasury. January 2017. "As more data is analyzed and year-to-year trends become visible, annual study results are likely to provide more insights into industry trends and developments." [p.2]. Available at https://home.treasury.gov/system/files/311/FINAL%20Auto%20Affordability%20Study_web.pdf.

4/ Monitoring Availability and Affordability of Auto Insurance, 81 Fed. Reg. at 45381.

5/ National Association of Insurance Commissioners, February 2021. Auto Insurance Database Report 2017/2018. Percentages represent ratio of Total Comprehensive and Collision Exposures to Total Liability Exposures.

6/ The oft-cited Insurance Information Institute data report rates of uninsured motorists at the state level. See https://www.iii.org/article/background-on-compulsory-auto-uninsured-motorists.

7/ See Heller, D. (2019). An Auto Insurance Lifeline for Safe Driving, Lower-Income Marylanders. Abell Foundation, Abell Brief (Winter 2020) for a discussion of the California program and its applicability in other markets. See, also, the California program website: https://www.mylowcostauto.com for details on the program.

8/ "CFA Studies on the Plight of Low- and Moderate-Income Good Drivers in Affording State-Required Auto Insurance. Available at https://consumerfed.org/cfa-studies-on-the-plight-of-low-and-moderate-income-good-drivers-in-affording-state-required-auto-insurance/.

9/ As some companies use census tract block groups rather than ZIP codes, it might be necessary to identify a block group and the associated ZIP code for this data call.

10/ Because these factors are generally not tied to a territory factor, choosing one ZIP code affords FIO the ability to determine the relative impact of the factors in each state, though it will not necessarily provide a picture of something like the "average" premium, depending upon whether the premiums in the selected ZIP code are closer to or further from the statewide mean premium.

11/ "Remarks by President Biden in a CNN Town Hall with Anderson Cooper." February 16, 2021.

12/ See, for example, "Auto Insurers Often Charge Identical Neighbors Considerable Higher Premiums Because of ZIP Code Differences." Consumer Federation of America. October 15, 2018. Available at https://consumerfed.org/press_release/auto-insurers-often-charge-identical-neighbors-considerably-higher-premiums-because-of-zip-code-differences/.

13/ It is important to note that the public policy goal of "no unfair discrimination" is used here not just in the actuarial sense but also as a civil rights imperative of public policy.

14/ "Commissioner Lara Finds Auto Insurance Companies Overcharged Drivers As Accidents Plummeted During the Pandemic." California Department of Insurance. March 11, 2021. Available at http://www.insurance.ca.gov/0400-news/0100-press-releases/2021/release030-2021.cfm.

15/ On July 15, 2021, New Mexico's Superintendent of Insurance issued a data call to "determine whether additional refunds were necessary," (https://www.osi.state.nm.us/index.php/news/auto-insurance-data-call/) and on June 25, 2021, Washington's Office of the Insurance Commissioner issued a "Special data call relating to claims activity and premium refunds, credit, or reductions." (https://www.insurance.wa.gov/sites/default/files/documents/wa-oic-ppa-survey-cover-letter.pdf).

16/ "Watch Where You're Going: What's Needed to Make Auto Insurance Telematics Work for Consumers." Consumer Federation of America. May 26, 2021. Available at https://consumerfed.org/press_release/new-white-paper-released-auto-insurance-telematics-programs-require-new-consumer-and-privacy-protection-standards-to-achieve-safety-and-pricing-promises/.