2016-16536. Monitoring Availability and Affordability of Automobile Insurance

AGENCY:

Federal Insurance Office, Departmental Offices, Treasury.

ACTION:

Notice; advising adoption of methodology to monitor affordability of personal automobile insurance.

SUMMARY:

The Federal Insurance Office (FIO) of the U.S. Department of the Treasury (Treasury) issues this notice pursuant to its authority to monitor the extent to which traditionally underserved communities and consumers, minorities, and low- and moderate-income (LMI) persons have access to affordable personal automobile insurance. In July 2015, FIO sought comments from stakeholders, including state insurance regulators, consumer organizations, representatives of the insurance industry, policyholders, academics, and others regarding: FIO's proposed working definition of “affordability” in relation to personal automobile insurance; the key factors FIO should use to calculate an affordability index for Affected Persons (e.g., premium, income, and other metrics); and how best to obtain appropriate data to monitor effectively the affordability of personal automobile insurance for Affected Persons. After carefully considering all the comments received in response to this and a previous solicitation, in conjunction with additional research and consultation, FIO has adopted a method to measure the affordability of automobile insurance for Affected Persons: FIO will calculate its Affordability Index by dividing the average (or mean) annual written personal automobile liability premium in the voluntary market by the median household income for U.S. Postal Service ZIP Codes (ZIP Codes) identified as being majority-minority or majority-LMI. FIO will presume that personal automobile liability insurance is affordable for Affected Persons if the Affordability Index is less than or equal to 2 percent.

To undertake the study of the affordability of automobile insurance for Affected Persons, FIO will collect and analyze premium data received and aggregated by statistical agents. In addition, FIO will use data publicly available through the U.S. Census Bureau. In combination, these data sources should facilitate analysis necessary for FIO to monitor the affordability of personal auto insurance for Affected Persons. FIO will report its findings annually, and note, among other things, the trend of the Affordability Index relative to each of the ZIP Codes analyzed.

FOR FURTHER INFORMATION CONTACT:

Lindy Gustafson, Federal Insurance Office, 202-622-6245 (not a toll free number).

SUPPLEMENTARY INFORMATION:

I. Background

Subtitle A of Title V of the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 (the Wall Street Reform Act) established FIO in Treasury and provides it with a number of authorities, including the authority to monitor the extent to which traditionally underserved communities and consumers, minorities, and low- and moderate-income (LMI) persons (collectively, Affected Persons) have access to affordable insurance products regarding all lines of insurance, other than health insurance.^[1]

In notices published in the Federal Register by FIO in April 2014 (April 2014 Notice) ^[2] and July 2015 (July 2015 Notice),^[3] FIO explained the reasons it is monitoring the availability and affordability of personal automobile liability insurance for Affected Persons. They are:

1. Nearly all jurisdictions of the United States generally require a driver or owner of a motor vehicle to maintain automobile liability insurance or financial security that may be satisfied by automobile liability insurance and that is applicable at the time of an accident, while operating a motor vehicle, or at the time of registering a motor vehicle;

2. On a nationwide basis, the percentage of uninsured motorists was approximately 14 percent between 2002 and 2009, before decreasing to 12.3 percent in 2010, 12.2 percent in 2011, and 12.6 percent in 2012;

3. Owning an automobile gives low-income commuters greater access to jobs since public “transit only enables [low-income commuters] to reach less than one-third of metro-wide jobs within 90 minutes . . . while the automobile enables them to reach all jobs in the 51 largest metropolitan areas within 60 minutes;” ^[4] and

4. Although some stakeholders have asserted that automobile insurance has become more affordable over time, representatives for consumers continue to assert that automobile insurance has become less affordable for Affected Persons.

A. The April 2014 Notice

In the April 2014 Notice, FIO requested comments regarding, among other things: A reasonable and meaningful definition of affordability of personal automobile insurance, and the metrics and data FIO should use to monitor the extent to which Affected Persons have access to affordable personal automobile insurance.^[5]

B. The July 2015 Notice

In the July 2015 Notice, FIO sought comments from the public on a framework for measuring the affordability of automobile insurance for Affected Persons. Based on comments submitted in response to the April 2014 Notice, FIO proposed a working definition for affordable personal auto insurance based on an affordability index. To do that, the July 2015 Notice set out in sequence: (1) A proposed definition of affordability; (2) a proposed definition and proposed calculation of an affordability index; (3) a proposed calculation of average premium; (4) a proposed definition of the market scope for an affordability index; and (5) a proposed definition of Affected Persons.^[6] Based on its consideration of those elements, FIO proposed the following working definition of affordable personal auto insurance:

A personal auto[mobile] liability insurance policy is affordable if the annual premiums are within the financial means of most people as measured by an affordability index for Affected Persons in the standard market. Personal auto[mobile] liability insurance is presumed to be affordable if, with respect to household income, the affordability index does not exceed two percent for Affected Persons in urban areas, for LMI persons within a specific geographic area (including rural areas), or for all individuals in majority minority geographic areas.^[7]

i. The Definition of Affordability

In developing its working definition of affordability, FIO considered three definitions submitted by commenters on the April 2014 Notice and ultimately proposed adopting the definition of “affordability” derived from a dictionary and submitted by one commenter: “being within the financial means of most people.” ^[8] FIO explained that this “common sense definition may be used to develop `a practical and effective approach to monitoring access to affordable personal auto[mobile] insurance.' ” ^[9]

ii. Use of an Affordability Index

FIO observed that some federal agencies use an index to measure affordability and provided examples. For instance, the U.S. Department of Housing and Urban Development (HUD) has a publicly available location affordability index that estimates the percentage of a family's income dedicated to the combined cost of housing and transportation in a given location.^[10] Additionally, the Consumer Financial Protection Bureau (CFPB) has a definition of “qualified mortgage” based, in part, on the ratio of the consumer's total monthly debt to total monthly income.^[11] Given the use of indices by other federal agencies, and FIO's statutory authority to monitor affordability for Affected Persons, FIO endorsed the concept of an affordability index for personal automobile insurance and proposed to calculate an affordability index for personal automobile insurance for Affected Persons.^[12]

iii. Average Premium

FIO stated that an affordability index for Affected Persons may be derived from a broad set of criteria, such as the average premium for personal liability insurance, personal injury protection, comprehensive insurance, collision insurance, uninsured motorist insurance, and underinsured motorist insurance; or more narrow criteria, such as the average premium for personal automobile liability insurance for a given year.^[13] FIO proposed to limit the calculation of an affordability index to the average annual personal automobile liability insurance premium for Affected Persons after considering comments to the April 2014 Notice. FIO chose this approach because states generally require the purchase of personal automobile liability insurance as a condition of driving or owning a motor vehicle.^[14]

FIO noted that the affordability of personal automobile insurance may be calculated by an examination of the average premium calculated as either (1) the total annual written premium for all insurers writing personal automobile insurance divided by the total number of policies; or (2) the total annual premium quoted by a sample of insurers writing personal automobile insurance divided by the number of insurers in the sample. FIO proposed to use one or both of these average premium metrics for annual premium depending on available data sources.^[15]

iv. Market Scope for an Affordability Index

FIO explained that an affordability index may be calculated for the entire market for personal automobile liability insurance or a specific market within personal automobile insurance because, historically, the automobile insurance market has been divided into three segments: (1) The standard market; (2) the non-standard market; and (3) the residual market. FIO described the residual market as generally comprised of the highest risk drivers, i.e., drivers who do not qualify for personal automobile insurance offered in the standard market or non-standard market; the non-standard market as comprised of high risk drivers, such as new drivers, drivers with moving violations, drivers with a rare or unusual motor vehicle, or drivers with a high automobile insurance policy cancellation or non-renewal rate; and the standard market as comprised of all other drivers. FIO reported that generally annual premiums for personal automobile insurance are highest in the residual market, followed by the non-standard market, and, finally, the standard market.^[16] Accordingly, FIO proposed to limit the calculation of an affordability index for personal automobile liability insurance to the standard market in order to diminish the impact of the annual premiums charged to the highest risk drivers.

In describing the framework that would be applied to determine whether personal automobile insurance is affordable, FIO examined the level of a person's income that should be devoted to that expenditure and cited to the suggestion by at least one commenter to the April 2014 Notice, that personal automobile insurance is affordable if it does not claim more than 2 percent of a low-income family's take-home pay.^[17] FIO also cited another study of the affordability of personal automobile insurance that found the national average insurance expenditures divided by national median income has been below 2 percent since 1995.^[18] In addition, FIO also cited to a Current Employment Statistics (CES) report that found the average expenditure for all households for automobile insurance and the average income after taxes for all households, based on 2013 data, indicated that all consumers spent about 1.6 percent of average income after taxes on automobile insurance.^[19] Based on this analysis, FIO proposed to presume personal automobile liability insurance is affordable if, for Affected Persons, the affordability index is less than or equal to 2 percent of household income.^[20]

v. Definition of Affected Persons

FIO is statutorily authorized to monitor the extent to which traditionally underserved communities and consumers, minorities, and low- and moderate-income persons have access to affordable insurance products. FIO adopted the term “Affected Persons” to describe traditionally underserved communities and consumers, minorities, and low- and moderate-income (LMI) persons.

FIO initially proposed to use “urban area,” as defined by the U.S. Census Bureau (Census Bureau), as a proxy for traditionally underserved communities and consumers.^[21]

FIO then proposed to define LMI by adapting the definitions used by the Federal Deposit Insurance Corporation (FDIC), which defines low-income as “individuals and geographies having a median family income less than 50 percent of the area median income” and moderate-income as “individuals and geographies having a median family income of at least 50 percent and less than 80 percent of the area median income.” ^[22] “The area median income is: (1) The median family income for the [metropolitan statistical area]; or (2) the statewide non-metropolitan median family income, if a person or geography is located outside a [metropolitan statistical area].” ^[23] FIO proposed to adapt this definition by using median household income as defined and identified by the Census Bureau,^[24] instead of median family income, in its study of affordability of personal automobile insurance. Accordingly, FIO proposed to define LMI persons as “individuals living in areas where the annual income of the geographic area is less than 80 percent of the median household income of a metropolitan statistical area or state.” ^[25]

FIO noted that the term “minorit[y]” is defined by law as “Black American, Native American, Hispanic American, or Asian American.” ^[26] It proposed to use ZIP Codes in which the minority population exceeds 50 percent as the standard for majority-minority geographic areas.

vi. Data Source and Request for Comments

FIO concluded the July 2015 Notice by describing the data needed to conduct its study, and sought opinions on how best to collect that information. FIO explained that it considered the currently available data relating to premiums for personal automobile insurance and concluded that the data is inadequate for FIO to monitor the extent to which Affected Persons have access to affordable personal automobile insurance.^[27] FIO stated that insurers have the most complete and accurate information that would allow it to perform its function of monitoring the extent to which Affected Persons have access to affordable automobile insurance and would be able to provide accurate price quotes for a given profile of a driver, including for a specific geographic area.^[28] In addition, FIO noted, insurers have the information to calculate the average annual premium for liability coverage for personal automobile liability insurance in the standard market for urban areas, and areas where the majority of residents are minorities or LMI persons.^[29]

Finally, FIO again requested that commenters provide feedback on the following:

1. FIO's proposed working definition of “affordability” in relation to personal automobile insurance;

2. The key metrics FIO proposes to use to calculate an affordability index for Affected Persons (e.g., premium, income, and other metrics); and

3. The best approach for FIO to obtain appropriate data to monitor effectively the affordability of personal automobile insurance for Affected Persons.^[30]

II. Final Working Definition of Affordable Personal Auto Insurance

After considering all the comments received—to both the April 2014 and July 2015 Notices ^[31] —and after undertaking additional research and stakeholder consultation, FIO has adopted a final working framework to study the affordability of personal auto insurance for Affected Persons. Personal auto liability insurance is presumed to be affordable if using an affordability index that is calculated by dividing the average annual written personal automobile liability premium in the voluntary market by the median household income for ZIP Codes identified as being majority-minority or majority-LMI, the Affordability Index does not exceed 2 percent.

In adopting this final working definition, FIO has made some changes to the proposed working definition from the July 2015 Notice based on comments and additional research. First, FIO will use the average annual written personal automobile liability premium in the voluntary market to calculate the Affordability Index. Second, FIO has adopted a different method of defining and accounting for Affected Persons to reflect issues with measuring traditionally underserved communities. Third, FIO has clarified that, to calculate the Affordability Index, FIO will use median household income data for ZIP Codes identified as majority-minority and majority-LMI areas. Finally, FIO has concluded that, based on comments and its additional research and consultation, all other aspects of the working definition are adopted as proposed.

A. Elements of Working Definition of Affordable Personal Automobile Liability Insurance for Affected Persons

For its final working definition, FIO has adopted an index to measure the affordability of automobile insurance for Affected Persons. FIO's Affordability Index will be calculated as the average annual written personal automobile liability premium in the voluntary market divided by the median household income for the ZIP Codes identified as majority-minority and majority-LMI.

i. Affordability Index

Based on comments received in response to the July 2015 Notice, insurers generally oppose the concept of using an affordability index to measure affordability for each category of Affected Persons. The Financial Services Roundtable (FSR) commented that a “mathematical index . . . attempts to reduce a myriad of complex factors into a single `one-size fits all' formula,” and “is inappropriate, insufficient, and perhaps even misleading as a measure of auto insurance affordability.” <sup>[32]</sup> The Property and Casualty Insurers Association of America (PCI) commented that an “affordability index does not consider that insurers have little or no control over the costs that drive auto insurance premiums and the `pass through' nature of the insurance mechanism.” ^[33] Meanwhile, the National Association of Professional Insurance Agents (PIA) commented that “attempts to define affordability as a fixed measure of income, [do] not give an accurate assessment of the non-insurance related factors—such as state tort law and highway safety measures—that impact insurance prices.” ^[34] Two groups of consumer advocates that provided comments—the Consumer Federation of America (CFA) and New Yorkers for Responsible Lending (NYRL)—support the creation and use of an affordability index to define affordability.^[35] CFA commented that it supports “an affordability index that defines affordability as a . . . percentage of a household's annual income.” ^[36] NYRL commented that “an affordability index is an effective way to evaluate the affordability of personal auto insurance” and “should be based on the cost of auto insurance as percentage of income.” ^[37] The Insurance Research Council (IRC) also offered support, acknowledging that an affordability index can be a useful method for monitoring affordability over time.^[38] In an August 2015 IRC Study, Trends in Auto Insurance Affordability (the IRC 2015 Study), the IRC used “expenditure and income data to form the IRC's expenditure-to-income ratio.” ^[39]

FIO acknowledges the various objections to adopting an affordability index as a tool to measure and evaluate the affordability of personal automobile insurance. FIO recognizes that some commenters view an index as reducing a myriad of complex factors into a single formula,^[40] or that an index, potentially, disregards non-insurance factors such as state tort law and highway safety measures.^[41] However, FIO is influenced by the established practices of other federal agencies that use indices to measure affordability, and for other purposes. Significantly, HUD created the Location Affordability Index to provide estimates of the percentage of a family's income dedicated to the combined cost of housing and transportation in a given location.^[42] Furthermore, FIO notes that the Bureau of Labor Statistics (BLS) in the U.S. Department of Labor (DOL) has long produced the Consumer Price Index (CPI), the most widely used measure of inflation, which provides information about price changes in the U.S. economy,^[43] while the U.S. Department of Commerce, Bureau of Economic Analysis produces the Personal Consumption Expenditure Price Index (PCE),^[44] generally thought to be “the single most comprehensive and theoretically compelling measure of consumer prices.” ^[45] And, even within the private sector, the National Association of Realtors produces the monthly Housing Affordability Index, which provides a way to track over time whether housing is becoming more or less affordable for the typical household.^[46] Finally, the IRC produces its own automobile insurance affordability index.^[47]

FIO notes the persuasive precedent of federal agencies using indices to measure affordability, among other economic measures, and agrees with those commenters who assert that an affordability index is an effective and meaningful way to measure and evaluate the affordability of personal automobile insurance. Furthermore, as FIO discussed in the July 2015 Notice, other federal agencies use indices to measure other kinds of affordability. Accordingly, given FIO's statutory authority to monitor affordability for Affected Persons, FIO confirms the adoption and use of the Affordability Index. FIO recognizes that an index does not address affordability for any individual consumer but that it is a tool that will help monitor over time the changes and trends in automobile liability insurance premiums for Affected Persons as a group. Consistent with its statutory authority, FIO will limit the application of the Affordability Index and evaluate affordability only for Affected Persons.

ii. Average Premium

FIO stated in the July 2015 Notice that an affordability index may be calculated using the average annual written personal automobile liability premium.^[48] The July 2015 Notice sought comment on the appropriate method of calculating the average premium and the types of policies included in the calculation.

Three commenters specifically addressed the appropriateness of using the average premium price at all. The American Insurance Association (AIA) commented that average premiums should not be used to calculate an affordability index because doing so would reflect a population, even among Affected Persons, who choose to buy higher limits, adjust their deductible, or have multiple household drivers or vehicles on a single policy.^[49] In contrast, the American Academy of Actuaries (AAA) took the opposite view and commented “that an appropriate measure of affordability of automobile insurance would be to compare average premium to average income.” ^[50]

In its comment, the PIA recommended using the median rather than the average, as it is a more precise measure because “[e]ven when limiting consideration to personal auto[mobile] liability in the standard market, consumer choices and insurance practices will skew average results in certain areas of the country.” ^[51] Although a median might be a more precise measure than an average, it would require collection of data that is not readily available and that therefore might place an undue burden on the collecting agencies, insurers, and others.

After reviewing the comments, and taking into consideration the varying perspectives on whether to use the average premium cost, FIO has concluded that using an average premium price is appropriate to calculate the Affordability Index. FIO will use average premium price data for the purpose of calculating the Affordability Index because of the following factors: (1) Average premium data is more frequently collected; ^[52] (2) additional conversations with industry participants have mitigated concerns of skewness in the premium distribution; and (3) using average premium data will reduce the reporting and computational burden on participating insurers and statistical agents.

In the July 2015 Notice, FIO proposed two ways to calculate average premium: (1) Average annual written premium for all insurers writing personal auto insurance or (2) average quoted premium for a sample of insurers. Allstate was the only industry commenter to specifically address the issue of using written or quoted premiums. In its comment, Allstate recommended “using actual total premiums written” because that information is collected by state insurance departments and the National Association of Insurance Commissioners (NAIC). It stated further that “[t]he collection of quote information . . . would necessitate the development of `hypothetical' customers who may or may not be representative of the people purchasing insurance in a particular area.” ^[53] Consumer advocates objected to the use of written premium over quoted premium, expressing concerns that using the actual prices paid for coverage, i.e., written premiums, does not provide a good measure of affordability because some consumers will not purchase insurance upon receiving quotes that are too expensive.^[54] Accordingly, consumer advocates recommended that FIO analyze data to reflect the premiums actually offered or presented to, rather than the premiums paid by, Affected Persons.^[55]

FIO will use written premium, not quoted premium, in this final working definition. One commenter supporting using quoted premium has previously acknowledged its drawbacks. In a September 2014 study, the Consumer Federation of America opined that collecting premium quotes from Web sites has several limitations such as (1) not all insurers' Web sites provide quotes; (2) a quote may be higher or lower than the actual price a consumer would pay depending on credit record; and, (3) because quotes must be collected manually, it is difficult to collect premium information for a large number of geographies or driver profiles.^[56] Although commenters make a reasonable argument for gauging affordability based on quoted premiums, the drawbacks identified in the 2014 study and by commenters dictate use of annual written premium, not annual quoted premium, in the calculation of the Affordability Index.

Commenters were divided on FIO's proposal to limit its analysis to only the premium for liability coverage and not consider comprehensive, collision or other costs associated with personal auto insurance. Comments from consumer advocates expressed concerns about this approach's exclusion of comprehensive and collision coverage costs from an affordability index calculation.^[57] Both CFA and NYRL commented that premiums for comprehensive and collision coverage should be included in calculating an affordability index because “a rising number of low- to moderate-income drivers have car loans that require additional insurance coverage.” ^[58] In addition, CFA noted that this coverage costs “approximately the same amount as the basic liability policy offered by a company,” ^[59] while NYRL commented that the cost of comprehensive and collision coverage “puts an additional burden on the driver who may make just enough to make the car payment” ^[60] and, therefore, should be included in calculation of an affordability index. The National Association of Mutual Insurance Companies (NAMIC) commented that limiting the scope of an affordability index to liability insurance would lead to data quality problems because state minimums vary and some states require personal injury protection (PIP).^[61]

Other commenters—AAA, Allstate, PIA, and PCI—submitted comments supporting FIO's view that an affordability index should measure only the cost of mandatory liability coverage. The AAA commented that the “optional [c]omprehensive and [c]ollision coverage should not be included” in an affordability index.^[62] Allstate commented that “[i]nsurance expenditure should be adjusted to reflect the minimum coverage required; [because] it is likely that Affected Persons purchase lower coverage limits, which reduces the amount they spend on insurance relative to the average insurance consumer.” ^[63] PIA commented that FIO should consider “only personal auto liability insurance in the standard market.” ^[64] PCI commented that only “the mandatory personal auto liability ” coverage for bodily injury and property damage should be included because “states generally require only the purchase of liability insurance as a condition of driving or owning a motor vehicle.” ^[65] As explained in the July 2015 Notice, because liability coverage (or financial responsibility limit) is the only requirement imposed by states as a condition of driving or owning an automobile, FIO concludes that liability coverage should be the basis for calculating the Affordability Index. Many variables affect consumers' decisions on the amount of collision and/or comprehensive coverage to purchase. For example, risk-averse consumers or consumers seeking asset protection may purchase the maximum amount of coverage available, while risk-tolerant consumers may purchase only the mandatory minimums. By including collision or comprehensive coverage in its calculation of the Affordability Index, FIO would introduce unnecessary confounding variables unrelated to affordability into an already complex analysis. For these reasons, FIO will limit the calculation of the Affordability Index solely to premiums for mandatory liability coverage.

iii. Market Scope

Commenters were split on the issue of limiting the calculation of affordability to the standard market only. As explained by FIO in the July 2015 Notice, an affordability index may be calculated for the entire market for personal automobile liability insurance or a specific market within personal automobile insurance.^[66] FIO explained that generally, annual premiums for personal automobile insurance are highest in the residual market, followed by the non-standard market, and then the standard market.^[67] FIO proposed to use only premiums in the standard market in order to diminish the impact of the higher annual premiums charged to the highest risk drivers in the other markets.

Consumer advocates opposed the use of only data from the standard market and, rather, proposed including data from the non-standard and residual market as well.^[68] CFA commented that residual and non-standard market should be included in an affordability index “because both of those markets serve, to some extent, good drivers who are Affected Person.” ^[69] NYRL commented that data should include residual market and non-standard premiums because “good drivers are being placed in non-standard markets as a result of socioeconomic factors.” ^[70] Other commenters supported FIO using only standard market data. The AAA and PIA stated that only data from the standard market should be considered in calculating an affordability index.^[71] The PIA commented that using data in the standard market will “diminish the impact of annual premiums charged to high-risk drivers as well as state laws and other requirements.” ^[72]

Notwithstanding the conflicting views, FIO notes that insurers generally use varying methodologies to rate policyholders who qualify for standard market premiums versus those who do not. For this reason, the exact size of the standard and the non-standard auto market is hard to calculate. The potential impact of excluding premium data for the non-standard market—estimated at 30 to 40 percent of the total private passenger auto insurance market ^[73] —when calculating the Affordability Index is significant. Accordingly, FIO will use data for both the standard and non-standard market to calculate the Affordability Index. As a result, FIO will capture relevant data, while addressing the concerns of consumer advocates that “good drivers are being placed in non-standard markets as a result of socioeconomic factors.” ^[74] For present purposes, FIO will refer to the standard and non-standard market collectively as the “voluntary market,” to distinguish it from the residual market and state assigned risk pools.

iv. Affected Persons

FIO has revised its definition of Affected Persons. In the July 2015 Notice, FIO adopted the term “Affected Persons” to collectively refer to “traditionally underserved communities and consumers, minorities, and low- and moderate-income persons.” FIO then proposed an approach to account for such persons in its working definition. First, FIO proposed to use “urban area” as the proxy for defining “traditionally underserved communities and consumers,” following the Census Bureau definition of urban area, “as densely developed territory that encompasses at least 2,500 people of which at least 1,500 reside outside institutional group quarters.” ^[75] Second, adapting the FDIC definitions for low-income and moderate-income, FIO proposed, for purposes of its definition of LMI, to consider individuals living in areas where the annual income of the geographic area is less than 80 percent of the median household income of a metropolitan statistical area or state.^[76] In explaining its decision, FIO noted that the FDIC defines low-income as “individuals and geographies having a median family income less than 50 percent of the area median income” and moderate income as “individuals and geographies having a median family income of at least 50 percent and less than 80 percent of the area median income.” ^[77] Third, FIO noted that “minorit[y]” is defined by law as “Black American, Native American, Hispanic American, or Asian American,” which is the definition incorporated by reference in the Wall Street Reform Act.^[78] In addition, FIO proposed to use ZIP Codes in which the minority population exceeded 50 percent as the standard for majority-minority geographic areas.^[79]

FIO received several comments in response to the July 2015 Notice regarding its proposed definition and parameters to account for Affected Persons. One comment encouraged FIO to define Affected Persons broadly in order to include communities that are marginalized because of factors beyond income.^[80] Two commenters opined that using geographic areas to identify Affected Persons may be the most practical way to approach the affordability analysis, with one of those respondents suggesting the use of ZIP Codes as the measurement of geographic area.^[81] Another commenter cautioned that the use of “urban areas” as a proxy for “traditionally underserved communities” would create a statistical category covering over 80 percent of the U.S. population, as over 250 million people live in “urban areas.” ^[82] Another commenter stated that the proposed definition for Affected Persons would be unmanageable because it would combine populations (LMI and minorities) with multiple and overlapping geographic units (i.e., ZIP Codes and Census Bureau “urban areas”).^[83] Relatedly, a letter by the Ranking Member of the U.S. House of Representatives Financial Services Committee, Congresswoman Maxine Waters, to FIO Director Michael McRaith, cautioned against the use of “urban areas” as a proxy for “traditionally underserved communities” because that term would exclude rural areas and could unduly skew data because of the presence of high-income households in high-density urban areas.^[84] Finally, a commenter warned that many states prohibit insurers from collecting data on income, race, religion, national origin, sex, familial status, or disability; insurers do not want to collect such data; and any requirement that insurers collect such data could create conflicting regulatory requirements.^[85]

FIO agrees with the commenters who suggested that its earlier proposal to use the Census Bureau-defined term “urban areas” as a proxy for identifying “traditionally underserved communities (including rural areas) and consumers” would fail to adequately capture and account for Affected Persons. Using “urban areas” as a proxy raises two significant concerns. First, the proposed proxy is over-inclusive because “urban areas” account for over 80 percent of the U.S. population.^[86] This level of coverage could capture numerous communities and consumers that would not meet any reasonable definition of traditionally underserved. Second, the proxy would exclude rural communities. The CFA commented that FIO could attempt to use ZIP Codes with high levels of uninsured motorists as a proxy to identify “underserved” areas, but conceded that even that data is not easily obtained, and noted that “LMI ZIP Codes and majority minority ZIP Codes” sufficiently capture those communities that would be properly considered “underserved” in this context.^[87]

The Wall Street Reform Act does not provide a definition of “traditionally underserved communities and consumers” or a methodology for identifying such communities or consumers. Likewise, the legislative history of the statute does not establish a clear or specific Congressional intent as to the meaning of the phrase.^[88] Given the lack of a statutory definition and an acceptable working definition and parameters for “traditionally underserved communities and consumers,” FIO reexamined the approach to the definition and parameters for Affected Persons as a whole and agrees with the observations of CFA about the challenges of defining “underserved” areas. Accordingly, in lieu of using urban areas as a proxy for identifying underserved communities as previously proposed, FIO adopts the approach recommend by CFA and instead will use “LMI ZIP Codes and majority minority ZIP Codes” ^[89] to capture those communities that would be considered underserved.

Based on stakeholder comments and its own research, FIO affirms the validity of the definition and parameters it adopted for identifying minority and LMI populations subject to the refinements discussed below. FIO will use the definition of minority set by law as “Black American, Native American, Hispanic American, or Asian American.” FIO has revised its adaptation of the FDIC methodology it proposed in the July 2015 Notice for identifying LMI persons. Following more precisely the practice of the FDIC, FIO will use median family income for designating LMI geographies instead of using median household income.^[90] FIO makes this change for two reasons: (1) Aggregated non-MSA ^[91] median household income data is not readily available, and (2) existing regulatory frameworks tend to use median family income data instead of median household income when analyzing geographic areas. For example, the Federal Financial Institutions Examination Council produces annual data tables by MSA, metropolitan division (MD), and non-MSA using family income for the Community Reinvestment Act (CRA) examination of banks.^[92] As noted above, the FDIC uses median family income to designate low- and moderate-income individuals and geographies. The lack of aggregated household income data for non-MSA areas would pose a challenge for FIO to readily identify rural LMI areas. Therefore, FIO will use median family income and not median household income to identify LMI geographies.

Accordingly, FIO will adopt the revised definition and parameters in its final working definition to account for Affected Persons as (1) persons living in majority-minority ZIP Codes, and (2) persons living in majority-LMI ZIP Codes.^[93] FIO believes that this approach results in a more workable framework while still reflecting the intent of the statute to monitor “traditionally underserved communities and consumers.”

The Affordability Index is calculated as premiums divided by income. In essence, it measures insurance expenditure expressed as percentage relative to income. While FIO's authority is to monitor the availability and affordability of insurance for Affected Persons, an automobile insurance premium study is most useful if linked to geography. This fact supports using majority-minority ZIP Codes and majority-LMI ZIP Codes as parameters to account for Affected Persons. FIO does not have ready access to individual insurance premium experiences and corresponding personal demographics data and, as commenters have pointed out, it is unlikely that insurers and statistical agents have this demographic data. Consistent with the reasoning in the July 2015 Notice, aggregate geographic areas can act as useful proxies to account for Affected Persons. In lieu of obtaining individualized data that may not be maintained by insurers, ZIP Code provides the closest proxy for observing the experiences of Affected Persons within discrete measurable geographic areas for which data is collected and available.^[94] Insurers acquire data to set premiums and, in so doing, capture policyholders' addresses, including ZIP Codes, for account billing, marketing, and other purposes. Accordingly, FIO will use ZIP Codes to define the geographic areas for calculating the Affordability Index because ZIP Code premium data is available and has (1) greater capacity to show variance across populations and geographic regions than counties and states; and (2) lower margins of errors than demographic data based on census tract. Incorporating these attributes of ZIP Codes has a positive impact on FIO's Affordability Index by providing a more detailed view of Affected Persons' automobile insurance experience than using state and county level data, and a more precise view than using census tract level data. This approach is consistent with prior reports studying the affordability of U.S. automobile insurance which analyzed ZIP Code-driven geographic areas.^[95] Focusing analysis on a ZIP Code basis allows areas with high concentrations of Affected Persons to be specifically evaluated, thereby facilitating understanding of the insurance experiences of Affected Persons across the United States and compensating for the lack of individualized data about Affected Persons.

In its July 2015 Notice, FIO proposed defining majority-minority geographic areas as those ZIP Codes in which the minority population exceeds 50 percent.^[96] Although FIO proposed that, for purposes of its working definition, it would define LMI individuals as those living in areas where the annual income of the geographic area is less than 80 percent of the median household income of a metropolitan statistical area or state,^[97] it did not provide the parameters for establishing the geographic areas for LMIs. As explained above, using a ZIP Code as a unit of analysis allows FIO to match demographic data for Affected Persons to aggregated data already collected by insurers, including ZIP Code-level data regarding average premiums. Additionally, income data is readily available at the ZIP Code level. Both the CFA and NYRL commented that ZIP Codes should be considered in the identification of Affected Persons. The CFA commented that FIO should refine the proposed definition of “LMI people” to focus geographic areas “explicitly on LMI `ZIP codes.' ” ^[98] The NYRL commented that “the focus should be placed on zip codes identified as populated by low- to moderate-income individuals and zip codes with predominantly non-white populations,” resulting in more targeted areas for FIO to “develop a more accurate evaluation of accessibility and affordability of personal auto insurance.” ^[99] Based on the views expressed by commenters and stakeholders, and FIO's own analysis, FIO will use majority-minority ZIP Codes and majority-LMI ZIP Codes as parameters to ensure that the Affordability Index more accurately captures the experiences of Affected Persons.

The use of “majority-LMI ZIP Codes” in the final working definition adds specificity to the proposed definition's use of “specified geographic area” as the parameter for reflecting LMI persons in the calculations. For purposes of the final working definition, majority-minority ZIP Codes are those in which the minority population exceeds 50 percent, consistent with the proposed definition, and majority-LMI ZIP Codes are those in which LMI persons exceed 50 percent of the population. FIO is mindful of the IRC's comment that this approach could still result in an overlap of the categories of Affected Persons within the same ZIP Code. Thus, a majority-minority ZIP Code may also be a majority-LMI ZIP Code. FIO will keep this potential complication in mind when identifying majority-minority and majority-LMI ZIP Codes.

B. Definition of Affordability and Application of the Affordability Index

In developing its definition of affordability, FIO considered three definitions submitted by commenters in response to the April 2014 Notice, and ultimately proposed adopting the definition of “affordability” derived from a dictionary and submitted by one commenter: “being within the financial means of most people.” ^[100] FIO explained that this “common sense definition may be used to develop `a practical and effective approach to monitoring access to affordable personal automobile insurance,' ” and proposed that it will presume automobile liability insurance is affordable for Affected Persons if the affordability index is less than or equal to 2 percent of household income.^[101]

The FSR commented that generally the proposed definition is “an acceptable construct” but “strongly disagree[d] that it can be reconciled with the factors and criteria delineated under the proposed affordability index,” and that “it is impossible to address the issue of affordability without openly referencing the concepts of consumer choice.” ^[102] The AIA commented that the “proposed definition is vague and ambiguous, and does not consider variations among states in required liability limits, . . . mandated personal injury protection (PIP), or claim and litigation environments” and “suggest[ed] that a reasonable definition . . . is one that recognizes relativity and consumer choice.” ^[103] NAMIC commented that although it understood “the difficulty in attempting to create such a definition,” it found the proposed definition of affordable confusing because of “the juxtaposition of `most people' and `Affected Persons,' ” and that “[i]t is not clear what `most people' means” in the context of the definition.^[104]

CFA commented that affordability “must be precisely defined rather than defined loosely as `within the financial means of most people,' ” and “that two percent of the household income of an Affected Person is the appropriate standard.” ^[105] Further, CFA stated that assessment of affordability should be relative to the purchasing capacity of low- and moderate-income persons, because “it is essential that affordability is gauged against the ability of low-wealth drivers to purchase insurance.” ^[106]

As these varying comments from the insurance industry and consumer advocates illustrate, there is not one generally acceptable method or definition of affordability. Rather, there are differing views, approaches, and methodology. Accordingly, FIO has considered all the comments provided, and adopts an objective standard as its first formal measure and definition of affordability of automobile insurance for Affected Persons. For the reasons explained in the July 2015 Notice, and reiterated below, FIO presumes that personal automobile liability insurance is affordable if the Affordability Index is less than or equal to 2 percent in the areas used to account for Affected Persons. In explaining its proposal, FIO cited a study of the affordability of personal automobile insurance that found the national average insurance expenditures divided by national median income has been below two percent since 1995.^[107] FIO also cited a report that found, based on 2013 data, that consumers spent about 1.6 percent of average income (after taxes) on auto insurance.^[108]

In comments to the July 2015 Notice, consumer advocates generally favored the 2 percent benchmark, while insurers and industry representatives opposed the adoption of a fixed numerical value as a measure for affordability. Both the CFA and NYRL stated that 2 percent is consistent with previous analysis of basic household budgets.^[109] On the other hand, insurers and others generally opposed adopting the 2 percent metric. The AIA stated that the 2 percent is artificial.^[110] The AAA stated that the 2 percent is only a single measure, and using it alone may be ill-advised because it could over simplify the complex task of defining “affordability.” ^[111] Allstate expressed concerns with the 2 percent, stating that FIO should monitor actual cost rather than make subjective assessments using a threshold.^[112] The FSR indicated that 2 percent is a misrepresentation of the term “affordable” and is unjustifiably low; and that it could create a perception that automobile insurance coverage is an inexpensive service whose price can easily be altered to meet particular needs and situation of each particular consumer.^[113] The IRC said the 2 percent is arbitrary in that an external reference or standard does not exist to support it; ^[114] while NAMIC stated that a reasonable basis for a 2 percent standard does not exist, and that it raises the question of how much the expenditure may deviate from the specified percentage before automobile liability insurance is deemed “unaffordable.” ^[115] The PIA said that relying on a metric to define affordability in terms of a percentage could lead to the desire to “fix” the problem by some kind of a subsidy; ^[116] and the PCI said the 2 percent is weighted heavily towards the higher income groups because LMIs, by definition, will spend a higher percentage of their income on automobile insurance as would be the case for other necessities.^[117]

FIO has carefully considered the views expressed by the commenters on this subject, including those who oppose using the 2 percent measure. Nevertheless, for purposes of monitoring the affordability of personal auto liability insurance, FIO will presume that insurance is affordable if the Affordability Index is less than or equal to the 2 percent benchmark. Based on the final working definition, the Affordability Index is the average annual premium divided by median household income.^[118]

In adopting this threshold, FIO considered that the overall cost of living varies considerably across the nation and that variation is reflected in part by the variation in household income. By basing the threshold on a specific percentage of household income, the measure will adjust, at least in part, for the variations in the overall cost of living and income levels from region to region. Using household income at the ZIP Code level is superior to other approaches because it (1) applies to more of the population than family income, (2) lessens the effect of outliers that could skew averages, (3) avoids the complexity of residual income approaches that could be biased due to high cost areas, and (4) is a widely accepted and used component to analyze affordability of other consumer products.^[119]

In settling on the 2 percent benchmark, FIO was most persuaded by the data in the Consumer Expenditure Survey, as produced by the Census Bureau and the BLS, which showed that the average household spent 2 percent of its income on automobile insurance.^[120] FIO notes that other key consumer goods and services already have an established affordability threshold that is expressed as a percentage of household income. For example, the affordability threshold for housing is 30 percent, healthcare is 9.56 percent,^[121] and residential running water is 2 percent.^[122]

Therefore, FIO adopts a 2 percent Affordability Index as a reasonable empirical benchmark for monitoring affordability and for the study to compare the cost of automobile insurance for Affected Persons. FIO acknowledges that the Affordability Index does not account for all circumstances which may be relevant to an individual consumer's cost of personal automobile insurance. Affordability for any individual consumer can be assessed accurately only within the context of that consumer's circumstances.

C. Data Sources

In the July 2015 Notice, FIO specifically requested input on how to best obtain appropriate data to monitor effectively the affordability of personal automobile insurance for Affected Persons. After considering stakeholder comments and potential information services, FIO intends to collect and analyze data received and aggregated by statistical agents. In addition, FIO will use data publicly available through the Census Bureau.^[123] In response to FIO's request, consumer advocate commenters suggested that FIO issue a data call to the 100 largest insurers in each state in order to obtain vehicle data and to reflect the premiums actually offered to Affected Persons.^[124]

Contrary to comments from consumer advocates and the views expressed by Ranking Member Waters,^[125] industry stakeholder comments objected to FIO issuing any data calls or other mandatory collections. Many argued that FIO could obtain information it needed from existing sources. The AIA commented that FIO should consult with the Automobile Insurance Plan Service Office (AIPSO) for data,^[126] while the AAA commented that FIO should use data available from statistical agents such as the Independent Statistical Services (ISS), Insurance Services Office (ISO), and the National Institute of Statistical Sciences (NISS). Allstate commented that FIO should use data available from the NAIC, the Insurance Information Institute (III), and IRC.^[127] The FSR expressed concerns about the substance, workability, cost, and administrative burden of a data call.^[128] The IRC commented that FIO should conduct an analysis of existing data before initiating research requiring new and costly data reporting and collection efforts.^[129] NAMIC, IRC and FSR's comments averred that FIO should first analyze and report existing studies and other data already available.^[130] In addition, NAMIC's commented that the term “monitor” should not be interpreted as authority for FIO to collect data directly from insurers.^[131] Finally, PCI stated that FIO should use BLS and Census Bureau data, and if FIO were to issue a data call, then it should rely upon third parties—statistical agents like ISO, ISS, and NISS—to aggregate that data.^[132]

FIO has reviewed and evaluated the comments received from stakeholders on whether to collect data directly from industry to support this work, and respects concerns about duplicative information gathering. FIO intends to avoid unnecessary burdens or expenses on stakeholders. FIO will exercise all reasonable efforts to use existing available information. Accordingly, at this time, FIO will not collect data directly from insurers through a data call as proposed in the July 2015 Notice.

For its initial affordability study, FIO will use data currently available from the Census Bureau,^[133] statistical agents, and certain states. In this regard, 20 states require insurers to report ZIP Code-level automobile premium data to one of three statistical agents (ISO, ISS, and NISS) who collect and aggregate this data.

For purpose of its next study in 2017, FIO will 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.

For 2017, FIO will request that large insurers who do not already report ZIP Code-level premium data voluntarily provide that data to the statistical agents with which the insurers typically work. FIO will ask that insurers covered by this request provide the statistical agents the following information: (i) ZIP Code-level premium data, (ii) for liability coverage at the financial responsibility limit, (iii) for the voluntary market.

In combination, the data sources described above are expected to provide sufficient data to support the objective analysis necessary for FIO to monitor the affordability of personal auto insurance for Affected Persons. If, however, FIO receives incomplete data, or if insurers or statistical agents are not responsive to this request, FIO may collect information directly from those insurers in the future.

Going forward, FIO will rely upon the methodology and the data described above to calculate the Affordability Index it will use to monitor the affordability of automobile insurance premiums in majority-minority or majority-LMI ZIP Codes. FIO will publicly report its findings annually and note, among other things, the trend of the Affordability Index relative to each of the analyzed ZIP Codes.

Michael T. McRaith,

Director, Federal Insurance Office.

Footnotes

1.  31 U.S.C. 313(c)(1)(B).

2.  Monitoring Availability and Affordability of Auto Insurance, 79 FR 19,969 (Apr. 10, 2014) (April 2014 Notice).

3.  Monitoring Availability and Affordability of Auto Insurance, 80 FR 38,277 (Jul. 2, 2015) (July 2015 Notice).

4.  Clifford Winston, “On the Performance of the U.S. Transportation System: Caution Ahead,” Journal of Economic Literature, Vol. 51, No. 3 at 805 (2013) (citations omitted), available at https://www.aeaweb.org/​articles?​id=​10.1257/​jel.51.3.773.

5.  April 2014 Notice, supra note 2, at 19,970.

6.  July 2015 Notice, supra note 3.

7.  Id. at 38,280.

8.  Id. at 38,279.

9.  Id. (quoting Property and Casualty Insurers Association of America, at 1 (June 9, 2014), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2014-0001-0020).

10.  Id. at 38,279 & fn. 33 (citing HUD, “Location Affordability Portal,” available at http://www.locationaffordability.info/​lai.aspx).

11.  12 CFR 1026.43(e)(2)(vi).

12.  July 2015 Notice, supra note 3, at 38,279.

13.  Id.

14.  Id. at 38,278. See also Insurance Information Institute, “Compulsory Auto/Uninsured Motorists” (June 2016) (listing automobile financial responsibility limits and enforcement by state), available at http://www.iii.org/​issue-update/​compulsory-auto-uninsured-motorists. New Hampshire is the only state that does not require the purchase of personal automobile liability insurance; however, drivers must be able to demonstrate they are able to provide sufficient funds to meet New Hampshire Motor Vehicle Financial Responsibility Requirements in the event of an “at-fault” accident. See State of New Hampshire Insurance Department, “Your Guide to Understanding Auto Insurance in the Granite State,” at 1, available at http://www.nh.gov/​insurance/​consumers/​documents/​nh_​auto_​guide.pdf.

15.  July 2015 Notice, supra note 3, at 38,279.

16.  Id. at 38,820 & fn. 38, noting that, in 2011, of the 330 insurers that wrote personal auto insurance in the standard and non-standard market, 95 wrote personal auto insurance in the non-standard market. Of the 95 insurers in the non-standard market, 15 also wrote in the standard market. See StoneRidge Advisors, LLC, “Non-Standard Auto Insurance Market Overview & M&A Trends,” View from the Ridge (August 2012), at 2, available at http://stoneridgeadvisors.com/​Content/​View_​From_​The_​Ridge_​August_​2012.pdf.

17.  July 2015 Notice, supra note 3, at 38,278.

18.  Id. at 38,280 (citing Insurance Research Council, Auto Insurance Affordability (November 2013), at 7).

19.  Id. at 38,280. Each month the Bureau of Labor Statistics' CES program surveys approximately 146,000 businesses and government agencies, representing approximately 623,000 individual worksites, in order to provide detailed industry data on employment, hours, and earnings of workers on nonfarm payrolls. See BLS, “Current Employment Statistics—CES National,” available at http://www.bls.gov/​ces/​.

20.  July 2015 Notice, supra note 3, at 38,280.

21.  Id. (proposing to define urban area as densely developed territory that encompasses at least 2,500 people, of which at least 1,500 reside outside the institutional group quarters. See Census Bureau, “2010 Census Urban Area FAQs,” available at https://www.census.gov/​geo/​reference/​ua/​uafaq.html).

22.  July 2015 Notice, supra note 3, at 38,280 & fn. 41 (quoting FDIC, “Community Reinvestment Act (CRA) Performance Ratings,” available at https://www5.fdic.gov/​crapes/​peterms.asp).

23.  Id.

24.  Id. at 38,280 & fn. 43, noting that household income includes income received on a regular basis by the householder and all other individuals 15 years of age and older in the household, whether related to the householder or not. It does not include capital gains or noncash benefits. According to the Census Bureau, “respondents report income earned from wages or salaries much better than other sources of income and that the reported wage and salary income is nearly equal to independent estimates of aggregate income.” Census Bureau, “About Income,” available at https://www.census.gov/​hhes/​www/​income/​about/​.

25.  July 2015 Notice, supra note 3, at 38,280.

26.  31 U.S.C. 313(c)(1)(B) (incorporating by reference the definition established in 12 U.S.C. 1811, note).

27.  July 2015 Notice, supra note 3, at 38,280.

28.  Id. at 38,281.

29.  Id.

30.  Id.

31.  Eighteen comments were submitted in response to the April 2014 Notice and 11 submitted in response to July 2015 Notice. All comments are available through www.regulations.gov.

32.  FSR, at 2, 7 (August 31, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0011 (FSR Comment).

33.  PCI, at 2 (August 13, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0006 (PCI Comment).

34.  PIA, at 2 (August 28, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0004 (PIA Comment).

35.  CFA, at 1 (August 31, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0014 (CFA Comment): NYRL, at 1 (August 31, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0010 (NYRL Comment). “CFA” includes all signatories to the comment letter—10 national groups (Americans for Financial Reform; Consumer Action; Consumer Federation of America; Consumers Union; NAACP; National Association of Consumer Advocates; National Consumer Law Center, on behalf of its low-income clients; National Council of LaRaza; U.S. PIRG, and United Policyholders) as well as 39 state groups from Alaska, California, Delaware, Florida, Georgia, Illinois, Indiana, Kentucky, Maine, Maryland, Massachusetts, Minnesota, Mississippi, Montana, New Jersey, New York, North Carolina, Ohio, Oregon, Pennsylvania, Utah and Virginia.

36.  CFA Comment, supra note 37, at 2.

37.  NYRL Comment, supra note 37, at 1.

38.  IRC, at 1 (August 28, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0005 (IRC Comment).

39.  IRC 2015 Study at 17, summary available at http://www.insurance-research.org/​research-publications/​trends-auto-insurance-affordability.

40.  FSR Comment, supra note 34, at 2.

41.  PIA Comment, supra note 36, at 2.

42.  Additional details about the HUD Location Affordability Index are available at http://www.locationaffordability.info/​default.aspx.

43.  Additional information about the Consumer Price Index is available at http://www.bls.gov/​cpi/​home.htm.

44.  Additional information about the Personal Consumption Expenditure Price Index is available at http://www.bea.gov/​faq/​index.cfm?​faq_​id=​518.

45.  Craig S. Hakkio, “PCE and CPI Inflation Differentials: Converting Inflation Forecasts,” Economic Review, at 51 (Federal Reserve Bank of Kansas City 2008), available at https://www.kansascityfed.org/​publicat/​econrev/​pdf/​1q08hakkio.pdf.

46.  Additional information about the Housing Affordability Index is available at http://www.realtor.org/​topics/​housing-affordability-index.

47.  IRC 2015 Study, supra note 41.

48.  July 2015 Notice, supra note 3, at 38,279.

49.  AIA, at 3 (Aug. 31, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0009 (AIA Comment).

50.  AAA, at 1 (Aug. 31, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0012 (AAA Comment).

51.  PIA Comment, supra note 36, at 3.

52.  A recent study on auto insurance affordability similarly focused on average premiums because national and state insurance expenditure data was “only available as an average.” Patrick Schmid, “Auto Insurance Affordability,” Journal of Insurance Regulation, vol. 33, no. 9, at 4 & fn.5 (2014), available at http://www.naic.org/​documents/​prod_​serv_​jir_​JIR-ZA-33-09-EL.pdf.

53.  Allstate, at 6 (August 27, 2015), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0003 (Allstate Comment).

54.  See CFA Comment, supra note 37, at 6-7; NYRL Comment, supra note 37, at 5-6.

55.  Id.

56.  Tom Feltner, Stephen Brobeck, & J. Robert Hunter, The High Price of Mandatory Auto Insurance for Lower Income Households: Premium Price Data for 50 Urban Regions, at 3-4 (Consumer Fed. of America Sept. 2014), available at http://www.consumerfed.org/​pdfs/​140929_​highpriceofmandatoryautoinsurance_​cfa.pdf.

57.  CFA Comment, supra note 37, at 3; NYRL Comment, supra note 37, at 3-4.

58.  NYRL Comment, supra note 37, at 4. See also CFA Comment, supra note 37, at 3.

59.  CFA Comment, supra note 37, at 3.

60.  NYRL Comment, supra note 37, at 4.

61.  See NAMIC, at 5 (August 31, 2015) available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2015-0005-0007 (NAMIC Comment).

62.  AAA Comment, supra note 52, at 1.

63.  Allstate Comment, supra note 55, at 2.

64.  PIA Comment, supra note 36, at 2.

65.  PCI Comment, supra note 35, at 2.

66.  July 2015 Notice, supra note 3, at 38,280.

67.  Id.

68.  CFA Comment, supra note 37, at 5-6; NYRL Comment, supra note 37, at 4.

69.  CFA Comment, supra note 37, at 5.

70.  NYRL Comment, supra note 37, at 4.

71.  AAA Comment, supra note 52, at 1; PIA Comment, supra note 36, at 2.

72.  PIA Comment, supra note 36, at 2.

73.  Andrea Wells, “Nonstandard Auto Insurance Market Is Not For Everybody,” Insurance Journal (April 13, 2015), available at http://www.insurancejournal.com/​news/​national/​2015/​04/​13/​364065.htm.

74.  NYRL Comment, supra note 37, at 4.

75.  July 2015 Notice, supra note 3, at 38,280; Census Bureau, “2010 Census Urban Area FAQs,” available at https://www.census.gov/​geo/​reference/​ua/​uafaq.html .

76.  Id.

77.  Id. (citing FDIC, “Community Reinvestment Act (CRA) Performance Ratings,” available at https://www2.fdic.gov/​crapes/​peterms.asp).

78.  Id. (citing 31 U.S.C. 313(c)(1)(B) (incorporating by reference the definition established in 12 U.S.C. 1811 note)).

79.  Id.

80.  NYRL Comment, supra note 37, at 1.

81.  Allstate Comment, supra note 55, at 6; CFA Comment, supra note 37, at 4-5.

82.  FSR Comment, supra note 34, at 8.

83.  IRC Comment, supra note 40, at 2.

84.  Ranking Member Waters letter to Director McRaith, re FIO's efforts to monitor the availability and affordability of automobile insurance (November 19, 2015) (Waters' Letter).

85.  AIA Comment, supra note 51, at 3.

86.  FSR Comment, supra note 34, at 8-9 & fn.22.

87.  CFA Comment, supra note 37, at 5.

88.  FIO notes that the CFPB has adopted a definition for “underserved.” According to the CFPB regulation at 12 CFR 1026.35(b)(2)(iv)(B): A county is “underserved” during a calendar year if, according to Home Mortgage Disclosure Act data for the preceding calendar year, no more than two creditors extended covered transactions, as defined in § 1026.43(b)(1), secured by a first lien, 5 or more times in the county. FIO has not adopted this approach because it is not well suited to insurance.

89.  CFA Comment, supra note 37, at 5.

90.  This definition is based on the definition used in the Community Reinvestment Act examination and accepted and implemented by the Community Development Block Grant program, FDIC, Federal Reserve Board of Governors, Office of the Comptroller of the Currency, and the Federal Financial Institutions Examination Council.

91.  An “MSA” is a metropolitan statistical area as defined by the Director of the Office of Management and Budget.

92.  See, e.g., 12 CFR part 345, 12 CFR 228.12, and 12 CFR part 25.

93.  This definition will capture Affected Persons in both rural and urban areas.

94.  FIO considered but decided not to use census tract data for the Affordability Index because insurers do not sort data by census tract, but instead by ZIP Codes.

95.  See Tom Feltner and Douglas Heller, “High Price of Mandatory Auto Insurance in Predominantly African American Communities” (Consumer Federation of America November 2015), available at http://consumerfed.org/​wp-content/​uploads/​2015/​11/​151118_​insuranceinpredominantlyafricanamericancommunities_​CFA.pdf?​source=​externa;​ Stephen Brobeck and J. Robert Hunter,, “Lower-income Households and the Auto Insurance Marketplace: Challenges and Opportunities” (Consumer Federation of America, January 2012), available at http://consumerfed.org/​reports/​cfa-report-title-forthcoming/;;; NAIC, “NAIC Insurance Availability and Affordability Task Force Final Report” (January 1998), available at http://www.naic.org/​documents/​prod_​serv_​special_​iaa_​pb.pdf.

96.  July 2015 Notice, supra note 3, at 38,280.

97.  Id.

98.  CFA Comment, supra note 37, at 4.

99.  NYRL Comment, supra note 37, at 4.

100.  July 2015 Notice, supra note 3, at 38,279 fn. 29 (citing PCI, at 1 (June 9, 2014), available at http://www.regulations.gov/​#!documentDetail;​D=​TREAS-DO-2014-0001-0020).

101.  July 2015 Notice, supra note 3, at 38,280.

102.  FSR Comment, supra note 34, at 4.

103.  AIA Comment, supra note 51, at 2.

104.  NAMIC Comment, supra note 63, at 2.

105.  CFA Comment, supra note 37, at 1.

106.  Id.

107.  July 2015 Notice, supra note 3, at 38,280 (citing IRC, “Auto Insurance Affordability,” (November 2013), at 7).

108.  Id. at 38,280. See also BLS, Current Employment Statistics, supra note 19.

109.  CFA Comment, supra note 37, at 3; NYRL Comment, supra note 37, at 3.

110.  AIA Comment, supra note 51, at 3.

111.  AAA Comment, supra note 52, at 2.

112.  Allstate Comment, supra note 55, at 2.

113.  FSR Comment, supra note 34, at 5-6.

114.  IRC Comment, supra note 40, at 2.

115.  NAMIC Comment, supra note 63, at 3-4.

116.  PIA Comment, supra note 36, at 2.

117.  PCI Comment, supra note 35, at 2.

118.  More specifically, as described above, the Affordability Index will be calculated using the average annual written premium for personal automobile liability insurance in the voluntary market, divided by median household income for areas which are majority-minority or majority-LMI, i.e., Affected Persons exceed 50% of the population.

119.  See, e.g., “America's Rental Housing: Evolving Markets and Needs,” (Joint Center for Housing Studies of Harvard University (2013), available at http://www.jchs.harvard.edu/​sites/​jchs.harvard.edu/​files/​jchs_​americas_​rental_​housing_​2013_​1_​0.pdf;​; and, New York City Rent Guidelines Board, “2015 Income and Affordability Study” (April 2015), available at http://www.nycrgb.org/​downloads/​research/​pdf_​reports/​ia15.pdf (using household income in rental housing affordability study).

120.  BLS Consumer Expenditure Survey, “Table 1110. Deciles of income before taxes: Annual expenditure means, shares, standard errors, and coefficients of variation” (2014), available at http://www.bls.gov/​cex/​2014/​combined/​decile.pdf.

121.  Shared Responsibility for Employers Regarding Health Coverage, 79 FR 8544 (Feb. 12, 2014), available at https://www.gpo.gov/​fdsys/​pkg/​FR-2014-02-12/​pdf/​2014-03082.pdf;​; 26 CFR 601.105, available at https://www.irs.gov/​pub/​irs-drop/​rp-14-62.pdf.

122.  The cost of water/wastewater is considered unaffordable when it exceeds 2% of median household income. See U.S. Environmental Protection Agency, Memorandum re: Financial Capability Assessment Framework for Municipal Clean Water Act Requirements (Nov. 24, 2014), available at http://www.epa.gov/​sites/​production/​files/​2015-10/​documents/​municipal_​fca_​framework.pdf.

123.  See Census Bureau, “American Fact Finder,” available at http://factfinder.census.gov/​.

124.  CFA Comment, supra note 37, at 6; NYRL Comment, supra note 37, at 5-6.

125.  Waters' Letter, supra note 86.

126.  AIA Comment, supra note 51, at 5.

127.  Allstate Comment, supra note 52, at 2.

128.  FSR Comment, supra note 34, at 3.

129.  IRC Comment, supra note 40, at 3.

130.  NAMIC Comment, supra note 63, at 4-5; IRC Comment, supra note 40, at 3; and FSR Comment, supra note 34, at 3-4, 10-11.

131.  NAMIC Comment, supra note 63, at 5.

132.  See PCI Comment, supra note 35, at 5.

133.  See Census Bureau, “American Fact Finder,” available at http://factfinder.census.gov/​.