Stephen Brobeck and J. Robert Hunter
Auto insurance is important for low- and moderate-income (LMI) households.Nearly all car owners are required by state law to purchase liability coverage, all those financing purchases are required by lenders to have collision and comprehensive coverage, and many car owners without financing would benefit from the latter. These insurance coverages are relatively costly. The federal government’s Consumer Expenditure Survey suggests that, in 2010, LMI households spent $30 billion on auto insurance premiums. This expenditure dwarfs LMI spending, in the same year, of $4 billion for automobile financing and $6 billion for life and other personal insurance premiums. It also greatly exceeds the estimated $9 billion in payday loan interest and fees paid by all consumers two years earlier. LMI auto insurance premiums were even two-thirds of the amount of all LMI spending on mortgage financing ($45 billion) in 2010.
This paper attempts to summarize what is known about LMI participation in auto insurance markets based on these sources and some new research.In doing so, it identifies and discusses key policy issues related need, access, and equity. These issues involve:
- State-Mandated Liability Coverage: Considering the importance of auto transport for most LMI households, should states require all car owners to purchase insurance liability coverage that mainly protects other motorists? If so, should this coverage be minimal, taking into consideration the ability of households to afford the coverage? And should society help subsidize the purchase of this coverage by lower-income households?
- Lender-Required Collision/Comprehensive Coverage: Can the relatively high rates of forced place coverage, which is purchased by lenders to protect their security interest when borrowers do not have their own collision and comprehensive coverage, be justified by lender and insurer costs? Should these rates receive greater attention by state regulators?
- Availability and Rates: For LMI households that wish to purchase more extensive liability coverage and/or collision and comprehensive coverage, are needed coverages affordable? And are these coverages priced fairly to LMI drivers? More specifically, are some insurers charging higher rates for less coverage? Are factors such as territory, education, occupation, and credit rating, which clearly have disparate impacts, being given too much importance? And are factors such as miles driven, which favor LMI drivers, being given too little importance?
- Claims Handling: Do insurers engage in disparate treatment of LMI claimants? Are there disparate impacts?
- Special LMI Programs: Should all states create special programs to allow safe LMI drivers to purchase minimal coverage at low rates? Is it important that, in the aggregate, these rates cover losses, or should society help subsidize the rates? Should all states make available easily-accessible information and advice to LMI households about how to purchase and maintain auto insurance most affordably?
This paper also seeks to identify the most effective and politically feasible approaches for meeting needs and mitigating problems. And it discusses research that would support these approaches.
The paper is organized in the following way:
Overview of Auto Insurance Marketplace
Annual Consumer Expenditures
Types of Coverage
LMI Automobile and Auto Insurance Needs
Need for an Automobile
Need for Auto Insurance
LMI Affordability of Auto Insurance
Auto Insurance Costs
Disparate Treatment of and Impact on LMI Households
Force Placed Coverage
Basis for Disparate Treatment
Uninsured and Underinsured LMI Households
Practical Policy Solutions
Reduced Liability Coverage Requirements
LMI Insurance Programs
More Effective Regulation
Development of Pay-Per-Mile Programs
More Effective Information and Advice
Summary and Conclusion
Overview of Auto Insurance Marketplace
Annual Consumer Expenditures
What U.S. households with auto insurance spend on this coverage can only be estimated. It is not even certain what all U.S. households, those with and without insurance, spent on this coverage. In 2007, according to industry sources, all households spent $160 billion on private passenger auto insurance premiums, nearly two-thirds of all personal insurance premiums and an average of $1379 per household. In the same year, the federal government’s Consumer Expenditure Survey (CES) reported average household auto insurance expenditures of $1071. Probably the most important factor accounting for this discrepancy between the two figures is underreporting by CES participants of their expenses.
Of course, not all households own cars, and not all car owners carry auto insurance, so the average cost for insured households is higher than that for all households. But since it is not certain how many households carry auto insurance, we cannot be sure how much higher. There is information, collected by the National Association of Insurance Commissioners, about the average premium per car. In 2009, that figure was $901, with a state range from $631 in Iowa to $1270 in Louisiana.And the Federal Reserve Board’s Survey of Consumer Finances reported that, in 2007, 87 percent of all households owned a car, though most of these households own at least two cars. Yet, as a fuller discussion about the uninsured later in this paper indicates, it is not certain what percentage of these vehicles are insured.
Considering varying estimates of uninsured motorist rates, the proportion of U.S. households carrying auto insurance may range from 70 to 80 percent. Adjusting the $1379 figure for all households upward would lead to annual estimated costs averaging $1724 to $1970per insured household. Adjusting the $1071 figure upward would result in annual costs averaging $1339 to $1530 per insured household.
Types of Coverage
Auto insurance coverage can be categorized broadly as collision/comprehensive or as liability. Both collision and comprehensive coverage pay for damage to the insured’s vehicle. However, liability coverage is more diverse and complex. The data on these coverages reported by the NAIC include sixteen different coverages, with several existing in only one state. There are, however, four major types of liability protection:
- Bodily injury liability, which pays medical, funeral, and legal expenses of those injured or killed by insureds who are at fault.
- Property damage liability, which paysfor damage to another driver’s vehicle (and other property damage) when the insured is at fault.
- Medical payments coverage, which pays for medical treatment of insureds for accident-related injuries, regardless of fault.
- Uninsured/underinsured motorist coverage, which pays for medical treatment of insureds and their passengers, if they are hit by uninsured or underinsured motorists (with some states also allowing some payments for property damage to the insured’s vehicle).
According to the NAIC, in 2009 on average consumers spent slightly more per insured vehicle for liability coverage ($474) than for collision and comprehensive coverage ($426). While both figures can vary depending on a number of factors, the latter will vary considerably depending on the value of one’s vehicle. In fact, many consumers choose not to purchase collision and comprehensive coverage on old cars worth less than $3000 to $4000. It also should be noted, and is discussed more fully below, that almost all states require purchase of at least some liability coverage.
Most auto insurance is sold by a few large companies. In 2006, there were 389 companies licensed to sell this insurance. However, in 2009 according to the NAIC, the largest four sold 45 percent of all private passenger liability (and PIP) premiums written. These companies (and their market shares) were — State Farm (18.0%), Allstate (10.2%), Berkshire Hathaway (GEICO) (8.7%), and Progressive (7.7%). In most states, even fewer companies are dominant. In 40 states four companies sold over 50 percent of all liability premiums, and in eight of these states the top four sold over 60 percent. In the District of Columbia, the top four share was 70 percent.
Most of the largest auto insurers sell directly through their own agents. That was not the case several decades ago, when companies like Hartford, Travelers, and Liberty sold almost exclusively through independent agents. But largely because they could not control costs as effectively as the direct sellers, these “indirect” sellers have lost market share. Today, some compete most effectively by winning contracts to sell insurance exclusively to members of large organizations, e.g., Hartford marketing to AARP members.
The reality, however, is that consumers, even members of these groups, often have a limited number of companies from whom they can purchase auto insurance. And, as will be noted later, these companies are not always interested in selling insurance to certain consumers in their service territories.
To a large extent, insurance premiums are based on insurer assessment of insured risk. And it is the function of underwriters employed by insurers to assess this risk. However, society has chosen to constrain risk-based rates. In fact, if rates were based entirely on risk assessments that were 100 percent accurate, risks would not be pooled, and policyholders would effectively be self-insured.
Society has decided that regulators should use equity considerations to modify risk-based rates. No states, for example, permit the use of race or income in rate-making. At the other extreme, all states agree that factors motorists largely control and also affect losses — notably type of vehicle, miles driven, and driving record — are appropriate factors to use in rate-setting, even though some, such as miles driven, are at present not easy to measure practically. A third set of factors, though, remain a continuing source of debate and controversy in many states. These factors include occupation, education, residence, credit history, and even age since, for example, pure risk-based rates for teenage male drivers would not be affordable for many families. These factors are discussed more fully in the section on disparate treatment.
Auto insurance rates and premiums, however, are based on more than insurer risk-assessment. They also are affected by how insurers pay claims. And they reflect the administrative expenses and profits of insurers, which for some companies can represent nearly one-half of all premiums collected. In 2010, according to industry data, the loss ratios of the 25 largest auto insurers ranged from 54.9 percent (Farmers) to 76.3 percent (State Farm). Public debates about the fairness of rates often involve insurer claims settlement, efficiency, and profit rates as well as the equity of insurance underwriting.
The U.S. insurance regulation system developed in the early 1800s when frequent insurance company failures and abusive treatment of customers persuaded states to establish commissions to regulate the industry, and most had done so by mid-century. In 1871, states created the National Association of Insurance Commissioners to help better coordinate their efforts. The states were permitted to regulate the industry until 1944, when the U.S. Supreme Court ruled that insurers were subject to federal law, including antitrust law. The next year, in response to the rulingCongress passed the McCarran-Ferguson Act, which not only delegated most insurance regulation to the states, but also granted a limited antitrust exemption to insurers. Despite legal and legislative challenges, including an antitrust provision of Proposition 103 approved by California voters in 1988, this antitrust exemption continues to allow the industry to engage in practices in most of the nation, such as the pooling of information through the Insurance Services Office (ISO), that would be considered anti-competitive and be illegal in most other industries.
There is no serious debate about whether the insurance industry should be regulated. Its essential role in the economy, its importance for consumers, the dependence of customers on its solvency, and the difficulty that individuals have evaluating the value of complex policies, let alone the solvency of their issuers, help explain the broad consensus of the need for regulation.
This consensus begins with solvency regulation. Insurers collect premiums that they invest then, at a later date, pay out in claims. In the case of life insurance policies sold to young adults, this date is usually decades later. Government regulation of insurers helps ensure not only that insurers remain solvent but also that they retain the confidence of their customers. The adoption by the NAIC of the accreditation program, which requires states to meet minimum standards for solvency regulation to be certified as compliant, has greatly improved the quality of insurance solvency regulation in America.
This consensus also extends to the regulation of market conduct by insurers. Regulators have the responsibility to prevent and remedy unfair and deceptive sales practices and also to see that customers have adequate information to make decisions about relatively complex products, often including information about typical rates charged by major insurers.This regulation, and restraint exercised by larger insurers concerned about reputational risk, help ensure that blatant, widespread consumer abuses — such as the sales abuses associated with several major life insurance companies in the 1990s — are infrequent. Consumer advocates and others, however, frequently complain about abuses that are less obvious and/or more controversial. These issues often relate to rate-setting and claims settlement. No accreditation sort of program exists and market conduct regulation by the states is significantly weaker than solvency regulation. Market conduct issues affecting LMI households are discussed later in the paper.
Also controversial is state regulation of insurance rates. The previous section noted disagreement about whether or the extent to which certain factors should be permitted in insurer rate-making. Just as controversial is whether or to what extent states should regulate rates. One state, Wyoming, allows insurers to use rates without filing them with the insurance commission. Several states allow insurers to use rates before actually filing them. Still other states permit “use and file” but limit increases or decreases within a range or “flex band.” Some states require rates to be filed before they are used — “prior approval” — with some of them also having “flex band” limits. One of these states is Massachusetts which, until several years ago, prescribed rates.
At present, largely becauseof Prop 103, the most extensive state regulation of insurance is by California. This initiative mandated a 20 percent premium rollback, instituted prior approval rate regulation, subjected insurers to state antitrust law, repealed anti-rebate laws for agents, provided for a “good-driver discount,” and limited rating factors such as sex and zip code. While most of the industry regards this regulation as burdensome and intrusive, advocates have argued that it represents model regulation for all states.
Aware of the importance of driving and the near-universal requirement that vehicle owners carry liability coverage, all states and the District of Columbia have created residual markets for owners who cannot purchase, or have difficulty purchasing, policies in the private marketplace. In these markets, auto insurers are required collectively to provide this auto insurance. State residual markets can be categorized as Automobile Insurance Plans (AIP), Joint Underwriting Associations (JUA), and Reinsurance Facilities, with AIP being utilized by a large majority of states.
An Automobile Insurance Plan, also called an Assigned Risk Plan, distributes car owners who cannot obtain coverage in private markets on a pro rata basis to auto insurers in the state. Thus, for example, if State Farm writes one-fifth of the premiums in a state, they are assigned one-fifth of the participating owners for whom they write policies, service these policies, and absorb related profits or losses.
Joint Underwriting Associations are organizations of auto insurers doing business in the state. The JUA helps design and set rates for the related auto insurance policy. A few companies are selected to administer the system, but underwriting losses are borne by all insurers based on the size of premiums written in the state.
Under Reinsurance Facilities, auto insurers must accept all applicants for coverage, then service these customers, including claims settlement. But insurers can cede customers to the reinsurance facility, then share underwriting losses and profits on the basis of premiums written in the state.
Although residual markets are intended to help car owners who cannot obtain reasonably priced insurance in the private marketplace, participating owners are usually charged premiums that are much higher than premiums charged in the mainstream marketplace. In fact, it is not unusual for these participants to be charged premiums that are two or three times higher, as will be shown later.
Participants in residual markets are often referred to as “high-risk drivers.” And many of them have poor driving records featuring speeding tickets and at-fault accidents. But these drivers also include many with excellent driving records who are young, poor, center city residents, those holding blue collar or service jobs, and/or those with poor credit records. In five states — New Jersey, New York, Massachusetts, New Jersey, and Rhode Island — between about four and seven percent of car owners participate in the residual market system, and in one — North Carolina — more than 20 percent are involved. But in most states, less than one percent of car owners participate.
Between 1994 and 2004, according to the Insurance Information Institute, the percentage of owners participating in residual markets declined from about four percent to 1.6 percent. The largest reductions were in Massachusetts, Michigan, New York, Pennsylvania, South Carolina, and Virginia. Important factors here have been the growth of substandard risk auto insurance markets and the increasing willingness of companies such as Progressive, GEICO, and some smaller companies to write these risks.
Assigned risk premiums are usually much higher than “standard” premiums, as suggested by information from New York and Maryland on typical premiums charged by four large insurers — Allstate, GEICO, Progressive, and State Farm — to a typical safe, middle-aged, female driver. For New York, in Hempstead, the assigned risk premium is $1607 while the other four premiums range from $538 to $1540; in Newburgh, the assigned risk premium is $1174 while the other four premiums range from $485 to $899; and in Rochester, the assigned risk premium is $733 while the other four premiums range from $158 to $508. For Maryland, in Montgomery County, the Maryland Auto Insurance Fund premium is $2034 while the other four premiums range from $614 to $1032; and in Prince George’s County, the MAIF premium is $1194 while the other four premiums range from $698 to $1118.
LMI Automobile and Auto Insurance Needs
Need for an Automobile
A large majority of LMI households own cars. The most detailed recent research on individual transportation was completed by the U.S. Department of Transportation using survey data collected in 2001. This research reported that nearly three quarters (73.6%) of households with incomes below $20,000, and nearly all (95%) of those with incomes between $20,000 and $40,000, owned a car. More recently, in 2007, the Fed’s Survey of Consumer Research indicated that only 65 percent of households with incomes below about $20,000 (lowest income quintile) and only 86 percent of those with incomes between about $20,000 and $40,000 (next income quintile) owned an automobile. Moreover, the comparable figures it reported for 2001 were 59 and 82 percent respectively.One reason for discrepancies between the DOT and Fed data is that, because incomes were lower in 2001 than in 2007, in 2001 nearly one-quarter of households had incomes below $20,000 and nearly another quarter had incomes between $20,000 and $40,000. Another reason may be that because the DOT’s survey was conducted on a one-time basis by a private contractor with a 41 percent response rate, albeit with 26,600 households, the well-established Fed survey may provide more reliable data about vehicle ownership. Regardless, both surveys reported that a large majority of both low- and moderate-income households own cars. Many without vehicles are households, often with low incomes, with an adult or adults who are not able to drive because of age or disability.
For most LMI households, not having a car imposes severe constraints on life choices. As one government report put it: “Overall…the limited mobility of lower-income men…affects access to potential employers, and may restrict access to health services, education, shopping at discount stores, and a vast array of recreational activities.”That is especially the case for employment. There is academic literature on “spatial mismatch,” the increasing difficulty people have getting to work because of the increased geographic dispersion of both jobs and residences. This research has found that access to a car is crucial to getting and holding the best jobs for which one is qualified. As one study concluded, “transportation problems predict employment outcomes.” Or as another study stated more specifically, “the importance of the automobile in providing employment access to lower-skilled, low-waged labor can hardly be overstated.” These findings have not been challenged.
For most LMI households, public transportation does not provide viable alternatives. Rural areas cannot sustain fixed-route, fixed-schedule transportation services, and as residents grow more dependent on auto transport, these services become even less sustainable. Almost all urban areas have some type of public transportation system. But except in a few large cities, these systems cannot meet all the transport needs of LMI residents. The systems often do not provide adequate access to residences, on the one hand, and workplaces, shopping centers, hospitals, and churches, on the other. Moreover, even when accessible, transit systems usually offer less flexibility, frequency, comfort, longer travel times, and more difficulty transporting heavy or bulky loads. All these reasons help explain why, according to the DOT research, low-income households take three-quarters of trips by car and only 5 percent by public transit. Most remaining trips represent short walks.
Need for Auto Insurance
Almost all LMI drivers are required to purchase auto insurance. All states but New Hampshire require drivers to carry liability insurance. The minimums required are below that of coverages recommended for most households with assets — $100,000/$300,000 bodily injury limits and $50,000 property damage liability (typically cited as 100/300/50). The lowest minimums permitted are the $10,000/$20,000 bodily injury limits in Florida and the $5000 property damage limits in California, Massachusetts, New Jersey, and Pennsylvania. By far the most common bodily injury limits are $25,000/$50,000 while two-thirds of property damage limits are either $10,000 or $25,000.
Historically, most states have not rigorously enforced their mandatory liability laws, but recently, an increasing number are doing so. Nearly four-fifths of states require drivers to have valid evidence of their policy in their vehicle at all times and to show this proof if stopped by the police. About the same number of states require drivers to produce evidence of insurance when they are involved in a crash. And, about half of states require evidence of liability coverage when a vehicle is registered.
Most states also require insurers to notify the motor vehicle department when a policy is cancelled or not renewed. In some, insurers are required to verify the existence of insurance in the event of an accident. In other states, companies are provided lists of randomly selected auto registrations, which they must then match up with insurance policies that drivers said were in effect. More recent laws, called computer data laws, require insurers to submit all automobile liability policies to a state agency such as the motor vehicle department. 
Auto lenders, as well as state governments, may require the purchase of auto insurance. Their interest, however, is protecting the value of their loan security, the vehicle itself, so they require borrowers to carry adequate collision and comprehensive coverage. To what extent does this requirement affect LMI households? At any one time, according to the Fed’s Survey of Consumer Finances, about one-fifth of low-income households (lowest income quintile) and one-third of moderate-income households (next lowest income quintile) are financing automobiles.
According to industry sources, when consumers finance the purchase of a car from a dealer, almost all carry or purchase coverage from an insurer. But when this coverage is dropped during the term of loan, and lenders learn about the termination of coverage, they purchase “force placed” collision and comp coverage whose charges are added to monthly loan payments. This coverage is almost always much more expensive than the borrower could purchase in the insurance marketplace.
LMI drivers, however, may need auto insurance not just because state government or lenders require it; they may feel the need to protect their property or their health. If they have attachable property, they may wish to protect it in the event they cause an accident, and the other driver sues to recover medical expenses and repair or replacement of their vehicle. According to the 2007 Survey of Consumer Finances, for those households who owned a car or cars, this asset was typically (median) worth $5600 to low-income households and $9200 to moderate-income households.
Also, if they do not have adequate health insurance, lower-income families may wish to purchase coverage that pays for their own medical costs when they were at fault or another at fault driver carries no or inadequate liability coverage. In this situation, LMI drivers may also feel they need collision and comprehensive coverage to help them afford repair or replacement of their own car. Some experts think that this risk should be borne when annual collision and comp premiums exceed ten percent of the value of a car.Still, some LMI households may, for example, prefer to pay $400 or more for this coverage, plus the deductible, instead of the $1500 or $2000 for major repairs or replacement of their vehicle.
On the other hand, there are many low- and moderate-income households — especially those with old cars, government health insurance, and insufficient assets to attach — whose only insurance benefits are compliance with the law and protection of the health and assets of other motorists. These households derive little if any direct benefits from purchasing state-required liability insurance coverage.
LMI Affordability of Auto Insurance
Especially during the recent recession, most LMI households have faced severe income constraints that make it difficult for them to afford auto insurance. All households in the lowest-income quintile have incomes below about $20,000 and average incomes, according to the 2010 CES, of just under $10,000. And all households in the second lowest-income quintile have incomes of about $20,000 to $40,000, and average incomes of just under $27,000. 
To understand precisely how these income levels constrain spending, it would be most useful, for each household, to estimate necessary expenditures as a proportion of income. However, researchers have concluded that both problems of definition — for example, the proportions of spending on food (eating out?), housing (air conditioning?), and transportation (5 vs. 10 year old car?) that are necessary and discretionary — and the variability of needs among lower-income households, related to factors such as household size and location of residence,makes this difficult and, thus recently, rarely attempted.
One useful effort, however, was undertaken in 2005 by a researcher at the Economic Policy Institute, who compared incomes and necessary expenditures for six types of working families residing in 400 communities. She estimated necessary spending for these households by computing “basic family budgets” using figures for specific types of expenditures that were based either on government estimates (e.g., low-cost plan for food at home) or typical LMI spending (e.g., 40th percentile rents). Under these assumptions, 30 percent of working families have incomes below basic family budget levels. The figures rise to over half for minority families (African American and Hispanic) and much higher than that for families with only one adult and one or more children.
Another indicator of income constraints is survey data on the proportion of those who say “they struggle to afford the necessities,” a question periodically asked by the Pew Research Center. In December 2010, 62 percent of those with household incomes under $30,000, but only 26 percent of those with incomes over $75,000, said they struggled to pay for heat and electricity. And 44 percent of the former, but only 11 percent of the latter, said it was difficult to afford food. While these percentages were inflated somewhat by lingering recession impacts on LMI households, in pre-recession February 2008 for all households the percentage who said they struggled to pay for heat and electricity was only four percentage points lower and that for food was only two percentage points lower.
Auto Insurance Costs
According to the 2010 Consumer Expenditure Survey, the average annual auto insurance spending per household in low-income households (lowest income quintile) was $535 and in moderate-income households (second income quintile) was $708. Yet, since many of these households did not own a car or carry insurance, these costs were higher for those who did. Adjusting these numbers, using car ownership statistics in the 2007 Survey of Consumer Finances, yields average annual premiums of $823 both for low-income and for moderate-income car owners. Since some of these car owners carried no insurance, the annual expenditures of those who did were even higher.
These auto insurance costs, however, include some liability coverage beyond the minimum required and some collision and comprehensive coverage. What would be the annual expense if low-income motorists with only one car chose to forego all coverage of their losses — payments for their medical expenses and repair or replacement of their own car — and purchase just required minimal liability coverage to protect the losses of other drivers?
In 2006, one researcher priced this coverage offered by three major insurers — Allstate, GEICO, and Progressive — for a low-risk driver living in a low-income area in twelve different metropolitan areas. These annual premiums ranged from $356 in Pittsburgh to $1660 in New York with an average of $831. In ten of the twelve areas, all but Pittsburgh and Indianapolis, the premiums were at least $600.
The average annual premium in Los Angeles was $802. But certain low-income drivers now qualify for a much less expensive option. Under California’s Low Cost Auto Insurance program, motorists with income less than $27,000 to $55,000, depending on family size, who have driven at least three years with a clean record and own a vehicle worth less than $20,000, qualify for minimal liability coverage (10/20/3) at relatively low rates. The annual premium in Los Angeles effective December 20, 2010 was $358 (the highest premium the program charges in the state). At the end of this year, there were 11,615 policies in force.
All these premiums for minimal liability coverage, though, are for low-risk drivers. Young drivers, especially males, pay much higher rates. In 2009, Insurance.com’s RateWatch issued a report that suggested adding a teen driver to a policy increased annual premiums by $1200 to $4900 a year. Even in the California low-cost program, single male drivers who are 19 to 24 years of age are surcharged 25 percent.
The higher rates for teenage drivers reflect the fact that they cause relatively high accident losses. But motorists of any age with a poor driving record pay higher premiums. For example, for three major insurers — Allstate, GEICO, and State Farm — the cost of minimal liability coverage in three Texas cities — San Antonio, Houston, and Dallas — is significantly more for drivers who have had one at-fault accident than for those with no accidents. Information available from the state insurance department indicates that premiums are 21-22 percent higher at State Farm, 25 percent higher at GEICO, 62-63 percent higher at Allstate, and 70-74 percent higher at Progressive.
In most states, insurers also charge higher premiums to motorists with “bad credit.” In the three Texas cities, for example, the premiums are 25-26 percent higher at State Farm, four percent higher at Allstate, 11-13 percent higher at Progressive, and the same level at GEICO.
How age, gender, residence, and driving record can combine to influence premiums can be illustrated by typical liability premiums charged two California drivers — a single female at least 30 years old who has been licensed 6-8 years, drives 7,600-10,000 miles a year, and has had no traffic violations or accidents, and a single male under 30 years old, who has been licensed 3-5 years, drives 7,600-10,000 miles a year, and has had one traffic ticket and one at-fault accident. For liability coverage, at the four major companies, the woman will be charged annual premiums of $694 to $1039 in Compton, Los Angeles, a low-income area, and premiums of $570 to $1058 in Sunnyside, Fresno, a moderate-income area. For the same coverage at three of the companies — no quotes from State Farm — the man will be charged annual premiums of $1628 to $2353 in Compton and premiums of $1334 to $1734 in Sunnyside. These high prices help explain why many motorists in these communities choose to risk fines for driving without insurance.
A survey of Latino participants in the Los Angeles auto insurance market revealed the impact of these high costs. The survey estimated that one-quarter of Latinos drive without insurance, primarily because coverage is too expensive. Some respondents reported that their auto insurance payments were (or would be) higher than their car payments.
In most urban areas, then, LMI motorists must pay annual premiums of at least $600, and sometimes more than twice this much, for minimal liability coverage that covers the expenses of other drivers but not their own. They must spend far more if they purchase standard coverage including collision and comprehensive, as typical premiums charged the man and woman from California suggest. For this broader coverage on an inexpensive new car, at the four major companies, the woman will be charged annual premiums of $2007 to $2618 in Compton and premiums of $1754 to $2352 in Sunnyside. For the same coverage at the three companies — again, no quote from State Farm — the man will be charged annual premiums of $5670 to $7511 in Compton and premiums of $4676 to $7552 in Sunnyside. These high prices help explain why so many insured motorists in low- and moderate-income communities choose to drive older cars and, if they buy insurance, purchase only the minimum liability coverage required by law.
Disparate Treatment of and Impact on LMI Households
These data on premiums suggest disparate impact on, and possibly disparate treatment of, LMI households who wish to purchase auto insurance. This disparate treatment and impact, however, can reflect differences not just in annual premiums but also in insurance availability and claims treatment. This section will discuss disparities.
There is evidence that reasonably priced insurance is less available in low-income areas, and to a lesser extent in moderate-income areas, than in higher-income areas. For example, throughout California in 1995, underserved communities included 16 percent of the state’s population and 13 percent of registered vehicles, but only six percent of auto insurance policies sold. Major insurers such as State Farm (2.6%) and Allstate (5.2%) maintained proportionately fewer offices in these underserved areas than throughout the state. By 2007, the percentage of policies in underserved areas had grown to ten percent, and State Farm now had 5.7 percent of their offices in these areas while Allstate had 4.7 percent.
These differences in availability from neighborhood offices are seen even more clearly in two cities studied. In the District of Columbia, there were 80 insurance offices distributed throughout eight wards, all of which had roughly equal populations (71,000-80,000). Yet, only four percent of the offices (3) were located in the two wards with the lowest median household incomes while 56 percent of the offices (45) were found in the two wards with the highest incomes. In Chicago, five insurers — State Farm, Allstate, American Family, Farmers, and Safeco — maintained nearly three-quarters (72%) of all 460 insurance offices in the city. Yet, only five percent of these offices (24) were located in the quintile of zip codes with the lowest median household incomes, while the other four zip code quintiles each included between 19 and 28 percent of the insurance offices (88 to 127).
Another reason for limited auto insurance availability to LMI households relates to whether policies are even offered to certain drivers and how these policies are priced. It has already been noted that State Farm, with the largest number of California policies, apparently will not sell one to a young man with a poor driving record who lives in Compton or Sunnyside. In fact, the state insurance department’s database of premiums suggests that State Farm will not sell a policy to this young man anywhere in the state.
Selective Pricing: A more common way for individual insurers to effectively deny auto insurance, though, is to grossly overprice it. For example, liability premiums for the Compton woman from 49 insurers were less than $1500, but Unigard’s premium was $2800. Similarly, premiums for standard coverage for the same woman from 48 insurers were less than $3100, but Viking’s premium was $4409, and Unigard’s premium was $4682.
Another practice of some insurers, which tends to discriminate against LMI car owners, is charging higher premiums for minimal liability coverage than for standard coverage. Price data available on several state websites allow comparisons that hold all factors constant but the extent of liability coverage. In several of these states — including Texas, Arizona, and Arkansas — some consumers are charged more by several companies for minimal liability coverage than for standard coverage. For example, in Texas two major insurers would charge a single female, age 25-64 living in low-income areas in Dallas, San Antonio, and Houston and driving a 2007 Toyota Camry with no traffic violations, more for 30/60/25 coverage than for 100/300/100 coverage. Allstate would charge $481 annually for minimal coverage vs. $454 for standard coverage in Dallas, $412 vs. $385 in San Antonio, and $481 vs. $454 in Houston. Nationwide would charge $563 for minimal coverage vs. $504 for standard coverage in Dallas, $427 vs. $380 in San Antonio, and $673 vs. $598 in Houston. This pricing pattern also exists for a married female and a young male. These differences suggest disparate treatment of LMI households who are far more likely to purchase minimal liability coverage than are higher-income households. While the differences may reflect, all or in part, actual losses, it seems unconscionable that insurers or their agents would offer more expensive policies with less coverage to a specific individual.
Rating Factors: More costly to LMI motorists than these practices, though,is the widespread use of rating factors — such as residence, education, occupation, and credit rating — that act as proxies for household income.The 2006 price comparisons for Allstate, GEICO, and Progressive in twelve metropolitan areas (cited earlier) found significant differences not only between areas but also within areas. A policy sold in lower-income neighborhoods ranged from eight to 94 percent more expensive, depending on the metropolitan area, than the same policy offered in upper-middle neighborhoods in the same urban area. In a 2006 release based on documents revealing GEICO’s extensive use of education and occupation as the basis for insurance rates and eligibility, the Consumer Federation of America showed that those with less education and less skilled occupations would pay premiums that were, on average, 40 percent higher.The use of credit rating in rate-making was illustrated earlier in this paper with examples from Texas.
Some, including most insurers, argue that these factors are highly correlated with insurer risk. And much research, including a 2007 report by the Federal Trade Commission concluding that credit scores are effective predictors of insurance claims, support this contention. There have also been efforts to develop explanations for the correlations. Those most convincing to regulators relate to residence: In urban areas, because low- and moderate-income motorists tend to live in densely-populated neighborhoods with fewer garages, their cars are more likely to be damaged either driving or at home.
Critics counter that these rating factors not only are flawed but are also inequitable and socially unwise. They argue that, without a convincing logical explanation for the correlations, they represent an arbitrary basis for rate-making and may be surrogates for income. The textbook example of the difference between correlation and causation is eating ice cream and death by drowning. While both activities are associated because they tend to occur during warm weather, one cannot reduce the chances of drowning by eating less ice cream.
Even if there are causal relationships, though, critics argue that rating factors are unfair to individuals for two types of reasons. First, the factors may accurately predict group behavior, but not individual behavior, and thus be unfair to members of the group with below-average risk for that factor. Second, the factors are often not measured adequately. For example, in a detailed 2005 study of the use of zip codes by major insurers, the California Insurance Department found that “the choice of individual zip codes as an appropriate building block in constructing territory is questionable” and that “industry wide pure premiums [loss ratios] do not strongly support the company zip code relativities [the relation between premiums charged and territory factors].”
Moreover, there is some agreement, even among free marketeers, that because auto insurance is needed and legally required by most lower-income households, it is neither fair nor socially sensible to force these families to spend much more than higher-income households for the same coverage. That is why no state permits household income to be used directly as a rating factor. The principle disagreement here is whether high-risk, lower-income households should be subsidized mainly through lower rates, usually resulting in higher rates for other insureds, or through special state-funded programs.
There is also some agreement that rate-making should be influenced largely, if not entirely, by factors over which individual motorists have some control, such as the cars they drive, and how far and how safely they drive them. That was an important part of California’s Proposition 103. There is also some awareness, though, of the limitations of these factors in predicting risk. Most motorists, for example, do not have sufficient accident experience to allow adequate differentiation of risk. However, a new emphasis on and ability to measure how far and safely a car is driven, though controversial, does offer potentials, discussed later, to more accurately link rates to individual risk.
Surcharges: Recently in California, a major insurer tried to win approval of a new type of surcharge that critics said would represent disparate treatment and result in disparate impacts. Through a ballot measure, Proposition 17 of 2010, Mercury Insurance sought to modify an existing law, which prohibits insurers from charging higher rates to customers on the basis of having been uninsured at some point in the previous five years. Consumer advocates claimed that the change would tend to raise rates on young, lower-income, and economically insecure motorists who were most likely to have had a lapse in insurance coverage in the past. The insurer argued the initiative would increase competition by authorizing companies to use a new discount, which they called a “continuous coverage discount,” to entice customers to switch companies. Despite a $16 million campaign by Mercury that promised policyholders a “$250 discount on their auto insurance,” voters sided with consumer advocates and defeated the measure.
The most important service an insured receives, or does not receive, from an insurer is claims settlement. There is some, but not conclusive, evidence that LMI insureds receive less satisfactory service than do higher-income policyholders. A large majority of the 34,580 consumer complaints about auto insurance, received in 2009 by state insurance departments and aggregated by the NAIC, involved claims settlement. But since individual companies were not identified, there is no way to correlate complaint ratios with LMI market shares.That is possible with the justified auto insurance complaints reported by the California Department of Insurance. In 2009, on average, the eleven companies with complaint ratios under 1.0 had 7.7 percent of their business in underserved communities, while the eleven companies with complaint ratios over 3.0 had 17.6 percent of their business in these communities. However, these ratios were based on a pool of only 328 complaints, which raises questions about the significance of the complaint ratio difference.
More informative than these complaints are ratings given by several thousand consumers, and several dozen body shop owners and managers, of the quality of service provided by major insurers in the Washington, DC area to Checkbook Magazine. Both sets of ratings reveal great variation in the quality of service. For example, 91 percent of surveyed customers of both Amica Mutual and USAA, but fewer than 60 percent of several other companies, rated their company “superior” for adequacy of claims payment. And three companies, including Amica and USAA, were mentioned favorably by at least 97 percent of auto body shops, while a couple companies, Progressive and Encompass, were mentioned favorably by two percent and zero percent respectively.
One of the lower-rated companies for service was Allstate, which may well reflect their implementation of the computerized Colossus system in an attempt to reduce claims paid by 20 percent. Most major auto insurers are making similar efforts, and these succeed in part because, according to an expert witness in a U.S. Supreme Court case, 70 percent of the insureds whose claims are denied take no action to pursue their claim. Most importantly, according to a former defense attorney who in disgust quit working for insurers and became a plaintiff’s attorney, in the claims process “it’s easier for insurers to pick on the sick, the weak, and the poor than someone who is big and tough.”
Force Placed Insurance
To protect their loan security, auto lenders require that car purchasers carry sufficient collision and comprehensive coverage. When borrowers allow this coverage to lapse, lenders purchase their own coverage on the secured vehicle. In the seven-year period from 2004 to 2010, consumers paid $24 billion in premiums.
Force placed policies are much more expensive than normal policies. In some cases, that is because lenders have purchased more coverage than the borrower originally had, because this coverage pays off loans in default, or because the coverage only pays claims when the car has been repossessed. It is also because of large commissions to lenders and preferential arrangements with captive reinsurers. These and other abuses were brought to the attention of insurance regulators by two consumer groups in 1996 and were the basis for a series of lawsuits around the same time. More recently, in the debate on financial services reform the Center for Economic Justice and Consumer Federation of America submitted evidence to Congress that loss ratios on creditor placed auto insurance were less than 25 percent.
Regulator attention and litigation may have curbed some of these abuses though it is not clear to what extent. In part, this is because the force placed auto insurance market is relatively small, so receives little attention. A large credit union estimates that they purchase this coverage on only one percent of their auto loans and that this coverage is often for relatively short periods of time because its expense motivates borrowers to find their own collision and comp coverage. Balboa Life and Casualty Company, which sells more than half of force placed auto insurance in the U.S., has annual premium revenues for this coverage of only several hundred million dollars. Moreover, few consumers complain about this coverage, and most who have done so recently, based on information in consumer complaint websites, express dissatisfaction mainly with auto lenders.
Basis of Disparate Treatment
Like other private enterprises, auto insurers seek to earn money, and they do so most profitably by selling standard policies to consumers with the most expensive cars, not by selling minimal liability insurance to those driving old vehicles. If the household owns more than one car, as most higher-income households do, so much the better. These households may well pay more than $2000 a year in premiums compared to most households with just liability coverage who spend under $1000 a year. Insurers also recognize that collision and comp are usually more profitable than liability coverage.
The economics of the industry explains much about how companies behave. Historically, they opposed required liability coverage because they believed the benefit of new customers, most with lower incomes, would be more than offset by increasing social obligation and regulation imposed on the industry, which has occurred. So insurers, prevented from using income in rate-making, limited sales and service in LMI markets through decisions about office location, product pricing, and claims settlement.
Uninsured and Underinsured LMI Households
For all the reasons previously discussed — their low incomes, their relatively large insurance costs and modest benefits, disparate treatment and impacts, and lax enforcement of mandatory coverage laws –a significant minority of LMI households drive without any insurance coverage, while additional LMI households drive with inadequate coverage to protect either themselves or other drivers.
There is better information about uninsured than underinsured LMI households, but even estimates of the number of the uninsured are controversial. There are two principal methods for estimating the proportion of drivers who are uninsured. The first, which is frequently used by states, seeks to compare the number of cars registered, using state data, with the number insured, using insurer data. One limitation of this method is that it fails to take into account those vehicles that are driven but are not registered. Another challenge is matching vehicles in the state and insurer databases, in part because of inaccuracies in the registration data in many states.
A second method, used most prominently by the Insurance Research Council (IRC), compares the frequency of claims paid under uninsured motorist coverage with bodily injury claims paid under insured liability coverage. The uninsured estimates represent the proportion of the first set of claims to both sets of claims. This method has been criticized for overestimating the number of insured motorists by underestimating the number of bodily injury claims relative to uninsured motorist claims. It has been suggested that insurers are more likely to pay uninsured motorist claims to their insured than liability claims from drivers they do not insure. It has also been suggested thata number of potential small bodily injury claims are often settled by drivers who do not report them to their insurers.
The potential difference between statistics computed using the two methods is illustrated by fairly recent criticism by Illinois insurance officials of the IRC estimates for that state. The officials claimed that the IRC’s 16 percent insured motorist estimate was much higher than their estimate, based on a comparison of state and insurer databases for a sample of drivers, of only five percent. On the other hand, when Texas officials made this comparison in 2002 they claimed that the IRC had underestimated the percentage of uninsured motorists.
Nevertheless, there is a broad consensus that a significant portion of LMI drivers are uninsured. The IRC’s latest estimate (2007) of the national uninsured motorist rate is 14 percent. They estimate that the lowest rates are in four New England states — Massachusetts, Maine, Vermont, and NewYork — and in North Dakota, all of which have rates that are six percent or lower. The highest rates are found in the Southern states of Florida, Tennessee, Alabama, Mississippi, Oklahoma, and New Mexico, all of which have rates between 20 and 30 percent. The estimated rate for California is 18 percent and for Texas, 15 percent. The IRC also estimated that the recent recession has increased uninsured motorist rates.
Researchers agree that there is a significant inverse relationship between income and the uninsured motorist rate — the lower their incomes, the less likely drivers are to carry auto insurance. This finding is well illustrated by research carried out by the California Insurance Department over the past fifteen years. Since 1995, this agency has estimated the uninsured motorist rate for “underserved communities” — those with below-median incomes, large minority populations, and insured motorist rates at least ten percentage points above the state average. In 1995, the department estimated this rate as averaging 39 percent, and more than a decade later, both in 2006 and in 2008, it estimated the rate as 40 percent. Moreover, it found that in a number of underserved communities, more than three-fifths of cars driven were uninsured.
To the percentages of uninsured motorists must be added percentages of underinsured drivers. A variety of sources report that many LMI households typically purchase only the minimum liability coverage with the highest possible deductible. This coverage, especially in states with $10,000/$20,000 bodily injury limits, often does not pay for medical expenses in accidents with serious injuries. Nor does it, especially in states with property damage limits of $10,000, cover the cost of replacing most relatively new cars. Uninsured motorist coverage purchased by many insured can make up much or all of any additional expenses. But uninsured motorists who are hit by underinsured motorists must pay these expenses themselves.
Just as importantly, at fault motorists who purchase only liability coverage are themselves liable for all expenses they incur in accidents. These costs include both medical expenses and the expense of repairing or replacing their own car. Even for those who drive very old vehicles, these expenses may well total or exceed $2000, which is a substantial sum for most LMI households to pay given the fact that most appear to have less than $500 in emergency savings.
Practical Policy Solutions
As suggested above, a large majority of LMI households need a car to take full advantage of economic opportunities and personal business necessities, such as medical treatment and food shopping. And almost all those with an automobile are required to purchase liability insurance which rarely costs less than $400 and sometimes costs more than $1000. Moreover, liability coverage does not pay for any medical costs, lost wages, or property damage suffered by insureds themselves. These realities alone make it important to devise ways that LMI households will not be unduly burdened by auto insurance premiums or receive poor claims service. This section discusses what we believe to be some of the more politically feasible approaches to meeting this challenge.
Curtailed Liability Coverage Requirements
Given liability coverage costs and limits, and the significant minority of LMI motorists who choose to violate the law by driving without it, it is worth asking the question, should LMI households be given the option not to purchase this coverage? As noted earlier, liability insurance provides no direct benefits to many LMI motorists other than compliance with the law. And the minimum coverage required by most states is not sufficient to fully compensate other motorists involved in serious crashes who are not deemed at fault. Moreover, those LMI households who are judgment-proof, have Medicare or Medicaid, and drive old cars derive relatively little benefit from any insurance coverage.
As compelling as these reasons are, they seem to be offset by marketplace and political realities. Without minimum coverage requirements, insurers would feel less obligated to provide affordable coverage to LMI households, and regulators would probably feel less need to require insurers to make this coverage available. As suggested earlier, even in today’s marketplace, insurers evidence much disinterest in supplying it. Furthermore, most consumers support mandatory liability laws because they believe that those who cause accidents should accept responsibility and pay for them. This belief has been an important reason that no-fault insurance laws no longer are supported by any stakeholder group. Drivers do not think it is fair for their premiums to increase when they are involved in accidents which are not their fault.
A more practical alternative, however, might be the reduction of required coverages to allow reduced premiums. In October 2010, a large majority of states had minimum liability limits that were no higher than 25/50/25, with sixteen having lower limits than this level, most notably, Florida at 10/20/10 and the California low-income program at 10/20/3.Keeping in mind that all motorists are free to purchase increased coverage, LMI households in Arkansas, Maine, Minnesota, North Carolina, and Wisconsin would probably benefit from reduced minimums to at least the 25/50/25 level. Some research suggests that these lower levels might even reduce costs for drivers with standard coverage if they persuaded more of the uninsured to purchase insurance. And even these relatively low levels would cover out-of-pocket expenses for not-at-fault drivers in a large majority of accidents.
What should certainly be viewed skeptically are efforts to increase these minimums. That occurred recently in Wisconsin, and similar proposals have been introduced in other states including Maryland and Nebraska.
LMI Insurance Programs
Required liability coverage could also be lowered as part of an insurance plan offered to certain LMI motorists. As noted earlier, California offers limited liability coverage to most LMI drivers with a clean driving record and a car worth less than $20,000. In almost all areas of the state, this coverage costs less than $400 annually, and in some areas it costs less than $300. In Hawaii, those receiving public assistance qualify for free no-fault coverage. And in New Jersey, for a dollar-a-day, $365 annually, those enrolled in Medicaid with hospitalization receive coverage of accident-related medical treatment up to $250,000 and a $10,000 death benefit.
These programs have their limitations. Hawaii’s program is practically invisible; it is not even described by the state insurance department’s website. New Jersey’s program does not provide liability coverage and has enrolled only around 20,000 participants. California’s program provides only liability protection and has even fewer participants even though millions of drivers are uninsured.
Nevertheless, if they fully understood how important automotive transport was to LMI households and how difficult it was for these households to afford even minimal insurance coverage, many citizens would probably accept some subsidization of a low-cost plan, especially since it might well end up reducing their own accident losses. They would probably be even more supportive if the plan, as does California’s, prohibits cross-subsidization by non-participating insureds. Moreover, when such plans exist, it is easier to morally justify rigorous enforcement of mandatory liability laws.
More Effective Regulation
Creating low-income insurance programs, however, would in no way obviate the need for more effective regulation if for no other reason than only a relatively small number of drivers participate in these programs. Most importantly, regulators should curb any disparate treatment of, and disparate impact on, LMI policyholders including:
- Ending discriminatory pricing such as charging higher premiums for less coverage.
- Insisting that insurers prove that all rating factors are related to risk and don’t discriminate against members of rated classes.
- Estimating the cumulative effect of auto insurance classification systems on availability and affordability.
- Increasing the importance in rate-making of factors over which motorists have some control including type of vehicle, and how far and how safely the vehicle is driven. Assigning much greater importance to miles driven, as suggested below, offers especially promising opportunities to reduce costs for most LMI households.
- Given the insurer emphasis on reducing claims costs, examining claims settlement more carefully, especially how insurers treat vulnerable claimants.
- Scrutinizing past areas of market conduct abuse such as unfairly priced force placed coverage. Kickbacks to lenders should receive special attention.
At present and into the foreseeable future, this regulation is likely to be mainly the responsibility of state insurance commissioners and their departments. In our view, the concern of insurance commissioners for LMI car owners would increase if research were presented to them, and to the public, showing discriminatory treatment of and impact on these owners. They would be much more likely to research industry practices themselves and, upon finding evidence of discriminatory treatment, would be more likely to mitigate it.
Aggressive Development of Pay-Per-Mile Programs
There is a very strong relationship between income and miles driven. According to the 2010 Consumer Expenditure Survey, the five income quintiles, lowest to highest, spent the following amounts on gasoline — $1009, $1598, $2180, $2634, and $3240. When differences in car ownership and vehicle characteristics are taken into account, it is evident that LMI car owners still drive about half as many miles annually as do high-income households.
Because these mileage differences have not been important rating factors, there is great potential to reduce LMI insurance costs by adequately considering mileage differences in insurance pricing. More effective treatment of these differences would also confer other benefits, including reduced driving that lessened traffic congestion, increased road safety, reduced pollution, and curbed oil consumption. The attractiveness of pay-as-you drive pricing is enhanced by the fact that there is a broad consensus among regulators, insurers, consumer advocates, safety advocates, and environmentalists that it should be developed.
Until recently, a major barrier to implementation of pay-per-mile programs was the difficulty insurers experienced efficiently obtaining adequate mileage readings. This factor is important because some research shows that motorists, in estimates given to insurers, tend to underestimate mileage driven by around 40 percent. Yet insurers were unwilling either to ask agents to check odometer readings annually or to insist that policyholders obtain independent verification of these readings.
Experimentation in the area, aided by new devices, is now well under way. GMAC Insurance offers Florida customers the option of using OnStar tracking devices to accurately measure mileage and potentially reduce premiums. State Farm has begun a Drive Safe and Save discount in California and Ohio based on actual miles driven.
A couple insurers use devices to record not only mileage but also driving habits such as time of day, hard breaking, and maximum speeds. In its Illinois Drive Wise program, Allstate offers a plug-in device to assess these habits. So, in its Florida Snapshot program, does Progressive.
Some critics complain that the devices which measure driving habits as well as mileage are too intrusive. That concern will be eased if insurers do not pressure policyholders to participate in related programs but rather offer participation as an option. Others note that questions have been raised as to whether pre-1996 car models, which are disproportionately driven by lower income consumers, can use the new technology. And consumer advocates believe it important for regulators to make certain that insurers share any efficiency gains with their policyholders.
More Effective Information and Advice
As surveys by the Consumer Federation of America have indicated, some insurance departments provide much more useful information to consumers than do other departments. This information, however helpful, is limited in value to many LMI households who have difficulty understanding the complexities of insurance coverages and dealing with insurance agents. There should be discussion of the value of insurance departments providing individualized advice to vulnerable consumers shopping for auto insurance. This advice could explain the pros and cons of different coverages, suggest ways to comparison shop, and note any special state programs. Given less than universal Internet access and literacy levels of many LMI persons, it is important for this advice to be made available, not just by email, but also by phone and in person in the languages frequently spoken in the state.
Summary and Conclusions
This paper has discussed participation of LMI households in the auto insurance marketplace in terms of need, access, and equity. Its first major conclusion is that most LMI families need affordable insurance coverage. In the first place, they benefit greatly from ready access to a car. Researchers agree that, for most of these families, having this easy access greatly increases economic opportunities related to work and consumption. As one study concluded, “the importance of the automobile in providing employment access to lower-skilled, low-waged labor can hardly be overstated.” This importance is reflected by a U.S. Department of Transportation survey showing that lower-income households take 75 percent of their trips by car and only 5 percent by public transit.
However, if LMI households own a car, they are required by law, and sometimes by lenders, to purchase auto insurance. All states except New Hampshire require car owners to purchase liability coverage that, in fault states, pays expenses suffered by other parties in accidents for which you are at fault and, in no-fault states, pays for your own Personal Injury Protection (PIP). Further, auto lenders require car owners they are financing to pay for adequate collision and comprehensive coverage to protect the lender security interest in the car. Apart from these mandatory coverages, many LMI car owners feel the need to purchase liability protection beyond required state minimums, while many owners without car financing still desire collision and comprehensive protection.
Thus, it is not surprising that compared to spending on other financial services, LMI households spend a great deal on auto insurance — $30 billion a year according to Consumer Expenditure Survey data. These data, together with Survey of Consumer Finance data on car ownership, suggest that low-income car owning households have recently paid an average of about $750 in annual premiums while moderate-income car owning households have paid about $1150 in annual premiums. These premiums vary considerably, however, from household to household and are especially high in many lower-income urban communities. To cite only one of numerous examples, according to data collected by the California Department of Insurance, a single male from Compton — who is under 30 years of age, has been licensed 6-8 years, drives 7,600-10,000 miles per year, and has had one traffic ticket and one-at-fault accident — will be charged between $1628 to $2353 for basic liability coverage and between $5670 and $7511 for standard coverage including collision and comp. These high costs help explain why so many LMI car owners nationwide, probably more than one-fifth and perhaps as many as one-third, drive without any insurance coverage. In California where this issue has been studied most carefully, more than three-fifths of drivers from many lower-income communities are uninsured.
Not only are many LMI car owners charged high premiums relative to their incomes, but these premiums often reflect disparate treatment and/or disparate impacts.
- Less access to insurance offices: Research suggests that those in LMI urban communities have much less access to auto insurance offices than do those in higher-income areas. For example, in the District of Columbia, of 80 insurance offices identified, only three were located in the two wards with the lowest incomes while 45 were located in the two wards with the highest incomes.
- Inability to purchase insurance from some major insurers for reasonable prices: Some major insurers will not even sell auto insurance to certain types of car owners, including the hypothetical man from Compton, California discussed above. Other insurers, according to state insurance department surveys, charge very high rates to these owners that are well above the rates charged by other insurers.
- Being charged higher premiums for less coverage: According to Texas, Arizona, and Arkansas insurance department data, holding all other factors constant, some major insurers charge lower premiums for standard than for minimum liability coverage. It appears that these insurers are discriminating against purchasers of the minimum coverage, who are disproportionately LMI car owners.
- Being charged higher premiums because of rating factors beyond their control: In general, LMI car owners are disadvantaged by rate classification systems used by insurers. They pay higher premiums because insurers use rating factors, such as residence, occupation, education, and credit rating, which are often correlated with risk. But insurers often have not adequately demonstrated to regulators that these correlations exist or that they actually reflect risk and are not surrogates for income.
- Being charged higher premiums because key rating factors are largely ignored: One important factor being ignored in risk-based rating systems is miles driven annually by car owners. LMI car owners drive far fewer miles annually than do higher-income owners — about half the miles of those in the top income quintile — but the lower risks associated with fewer miles driven are not adequately recognized by rating systems.
- Being charged very high premiums for forced place coverage: Collision and comprehensive coverage purchased by auto lenders for borrowers without this coverage is relatively expensive because, as they do for most types of credit insurance with reverse competition, lenders can and do charge insurers large commissions. These commissions are the main reason that, according to one study, loss ratios on forced place coverage averaged 25 percent, well below the industry average of more than 60 percent.
- Being treated unfairly in the claims process: To quote one plaintiff’s attorney who used to work for insurers, “it’s easier for insurers to pick on the sick, the weak, and the poor than someone who is big and tough.”
In trying to explain this evidence of disparate treatment of LMI households, it is difficult to avoid the conclusion that major insurers are far more interested in selling auto insurance to higher-income families. These insurers are well aware that upper-income families are much more likely to own two or three expensive cars, with comprehensive coverages, than are LMI households who often purchase just minimum liability coverage on an old car. Even if they earned a higher profit rate on LMI policies, insurers would earn far more dollars per policy on upper-income policies. Insurers also value the opportunity to sell other types of insurance, such as homeowners, to upper-income customers.
There is much that can be done to meet LMI household auto insurance needs, increase LMI access to fairly priced insurance, and reduce related disparities. The paper discusses several public policy approaches that we believe could “make a difference” in meeting these goals.
- Work to lower minimum liability coverage requirements: These state liability requirements do not directly benefit the many LMI drivers who are effectively judgment-proof. They protect only other drivers — many of whom carry uninsured motorist coverage required by many states — who suffer damages caused by the LMI drivers. Lowering these limits to those in Florida or California, for example, would lower premiums and allow more LMI households to purchase and obey the law. Efforts to raise these limits, as have occurred in several states recently, should be questioned.
- Create special low-income programs: The only serious program of this kind of which we are aware exists in California, which offers low-cost liability coverage to LMI drivers. The premiums are relatively low because the program offers very low liability coverage only to “good drivers.” The program is not subsidized by taxpayers or other ratepayers. However, we believe a strong case could be made for some subsidization of these types of programs.
- Work to eliminate disparate treatment and reduce disparate impacts: Consumer and community groups could productively work together on this issue as they have on related consumer and mortgage lending issues. There is already sufficient evidence of these types of discrimination to raise the public profile of this issue and communicate with state regulators.
While some of these issues can be debated, what is undeniable is that high auto insurance costs for LMI households either impose a substantial financial burden or greatly limit economic opportunity, especially access to jobs. Only state regulators can take the lead in mitigating these problems.
 Brobeck is Executive Director and Hunter is Director of Insurance of the Consumer Federation of America. They wish to especially thank the Ford Foundation for supporting this research, Remy Aronoff and Professor Catherine Montalto for providing essential research assistance, and Frank DeGiovanni, Amy Brown, Travis Plunkett, Doug Heller, Mark Savage, and Birny Birnbaum for contributing helpful comments.
 Because of the nature of the data available, this paper will usually treat LMI households as those in the two lowest income quintiles. Recently the income breaks have been at about $20,000 and $40,000.
 U.S. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey for 2010 (www.bis.gov/cex/).
 Stephens Inc., Payday Loan Industry Report, April 17, 2008, p. 34.
 Insurance Information Institute, The Insurance Fact Book 2009, p. 54.
 This underreporting is acknowledged by many researchers. See Jennifer Shields and Nhien To, Learning to Say No: Conditioned Underreporting in an Expenditure Survey, AAPOR-ASA Section on Survey Research Methods, 2005.
 National Association of Insurance Commissioners, 2008/2009 Automobile Insurance Database Report (2012), p. 27.
 Brian Bucks, Arthur B. Kennickell, Traci L. Mach, Kevin B. Moore, “Changes in U.S. Family Finances from 2004 to 2007: Evidence From the Survey of Consumer Finances,” Federal Reserve Bulletin (Feb. 2009), p. A31.
 NAIC (2012), loc. cit., p. 14.
 Ibid, pp. 19, 22, 25.
 Conversations with Martin Schwartz, Vehicles for Change, March 9, 2011, and with Eric Po, CURE Auto Insurance, Jan. 28, 2011.
 Insurance Information Institute, loc. cit., p. 55.
 PIP (Personal Injury Protection) pays for first-party medical expenses, lost wages, and other benefits in states with no-fault laws.
 National Association of Insurance Commissioners, 2009 Market Share Reports for Property/Casualty Insurance Groups and Companies: Private Passenger Auto (2010).
 Laureen Regan and Sharon Tennyson, “Insurance Distribution Systems,” in Georges Dionne, ed., The Handbook of Insurance (Huwer Academic Publishers, 2000).
 For a useful overview, see Lisa A. Gardner, David C. Marlett, “The State of Personal Auto Insurance Rate Regulation,” Journal of Insurance Regulation (Winter 2007), p. 39ff.
 National Association of Insurance Commissioners, Property and Casualty Insurance Industry 2010 Top 25 Groups and Companies By Countrywide Premium, By Line of Business, Total Private Passenger Auto (March 28, 2011).
 Robert W. Klein, “The Insurance Industry and Its Regulation: An Overview,” in Martin F. Grace and Robert W. Klein, ed., The Future of Insurance Regulation in the United States (Brookings Institution Press, 2009), pp. 13-51. See also Gardner, loc. cit.
 The need for regulation is accepted even by those arguing for rate deregulation. See Sharon Tennyson, Efficiency Consequences of Rate Regulation in Insurance Markets, Policy Brief (Networks Financial Institute, 2007).
 Gardner, loc. cit., pp. 10-12 of 19.
 See R.E. Cheit and J.D. Youngwood, “How Not to Reform Auto Insurance,” Public Interest (Summer 1991), pp. 67-79. For a different view see J. Robert Hunter, State Automobile Insurance Regulation: A National Quality Assessment (Consumer Federation of America, 2008).
 Gardner, loc. cit. 12-14 of 19. See also Scott E. Harrington and Helen I. Doerpinglaus, “The Economics and Politics of Automobile Rate Classification,” The Journal of Risk and Insurance, v. 60, n. 1. (1993), pp. 59-84.
 Gardner, loc. cit., 13 of 19.
 Insurance Information Institute, Auto Insurance Facts and Stats (2007).
 New York State Insurance Department, 2010 Consumer Guide to Auto Insurance (www.ins.state.ny.us/auto/2010/auto10pdf).
 “Annual Auto Insurance Rates,” Washington Consumers Checkbook, v. 15, n. 1 (Winter-Spring 2009), p. 12.
 John Pucher and John L. Renne, “Socioeconomics of Urban Travel: Evidence from the 2001 NHTS,” Transportation Quarterly, v. 57, n. 3 (Summer 2003), p. 56.
 Bucks, loc. cit., p. A31.
 Anam Aizcorbe, Arthur B. Kennickell, Kevin B. Moore, “Recent Changes in U.S. Family Finances: Evidence from the 1998 and 2001 Survey of Consumer Finances,” Federal Reserve Bulletin, v. 83 (January 2003), p. 19.
 Mobility and the Melting Pot, NPTS Brief, U.S. Department of Transportation Federal Highway Administration (Jan. 2006), p. 3.
 Steven Garasky, Cynthia Needles Fletcher, Helen H. Jensen, “Transiting to Work: The Role of Private Transportation for Low-Income Households,” Journal of Consumer Affairs, v. 40, n. 1 (Summer 2006), p. 8 of 13.
 B.D. Taylor and P.M. Ong, “Spatial Mismatch or Automobile Mismatch? An Examination of Race, Residence and Commuting in U.S. Metropolitan Areas,” Urban Studies, v. 32 (1995), p. 1471. See also Donald S. Houston, “Methods to Test the Spatial Hypothesis,” Economic Geography, v. 81, n. 4 (Oct. 2005), pp. 422-423.
 Alice E. Kidder, “Passenger Transportation Problems in Rural Areas,” in William R. Gillis, ed., Profitability and Mobility in Rural America (Pennsylvania State University Press, 1989), pp. 132ff.
 Charles L. Wright, Fast Wheels, Slow Traffic: Urban Transport Choices (Temple University Press, 1992), p. 126.
 Pucher, loc. cit., p. 59.
 Technically, in states such as California only proof of financial responsibility is required of drivers, and this proof can be by bond, but practically, this proof can be shown only by carrying auto insurance.
 Insurance Information Institute, Insurance Fact Book, loc. cit., pp. 66-67.
 Insurance Information Institute, Compulsory Auto/Uninsured Motorists (Oct. 2010) (www.iii.org/issues_updated compulsory-auto/uinsured-motorists.html).
 Gardner, loc. cit., p. 2 of 19.
 Bucks, loc. cit., pp. A40, and A45.
 Thomas J. Methrin, “The Poverty Industry-Con,” The Alabama Lawyer, v. 58 (1997), pp. 234ff. Interview with Aaron Bresko, Boeing Federal Credit Union (March 9, 2011).
 Bucks, loc. cit., p. A46.
 Julie Patel, “Buying Auto Insurance? Some Things You Should Know,” Tribune Business News (April 12, 2010).
 Bucks, loc. cit. U.S. Bureau of the Census, loc. cit.
 See discussion in Karen Secombe, “Families in Poverty in the 1990s: Trends, Causes, Consequences, and Lessons Learned,” Journal of Marriage and the Family (Nov. 2000), v. 62, n. 4, pp. 1094-1099.
 Sylvia Allegretto, Basic Family Budgets: Working Families’ Incomes Often Fail to Meet Living Expenses Around the U.S., Economic Policy Institute Briefing Paper #165 (Sept. 2005).
 Many Low-Income People Struggle to Pay Necessities, Pew Research Center, December 15, 2010 (http://people-press.org/2010/12/15/section-3-affording-the-necessities/).
 U.S. Department of Labor, loc. cit.
 Bucks, loc. cit., p. A31.
 Matt Fellowes, From Poverty Opportunity: Putting the Market to Work for Lower Income Families, The Brookings Institution (2006), pp. 37-39.
 California Automobile Assigned Risk Plan (CAARP), Status of the California Low Cost Automobile Insurance Program: Annual Report to the California Legislature (Jan. 2011).
 Sari Martin, “Tips to Save Money and Keep Your Teen Safely Insured,” Insurance.com RateWatch (May 5, 2009).
 CAARP, loc. cit.
 Texas Department of Insurance, HelpInsure.com, June 2011.
 California Department of Insurance, 2011 Auto Insurance Rate Comparisons, June 2011.
 Jongho Lee, Celina Torres, Yin Wang, Living in the Present, Hoping for the Future: Latinos and Insurance, a Los Angeles Case Study (Tomas Rivera Policy Institute, August 2005).
 California Department of Insurance, 1996 Commissioner’s Report on Underserved Communities, Tables C and E.
 California Department of Insurance, 2008 Commissioners’ Report on Underserved Communities, pp. 22 and 28.
 Research using Washington DC Car Insurance List of Agents and DC ward median incomes, June 2011.
 Research using Chicago Car Insurance List of Agents and Chicago zip code median incomes, June 2011.
 California Department of Insurance, 2011 Auto Insurance Rate Comparisons, June 2011.
 Texas Department of Insurance, loc. cit. In both Arizona and Arkansas, several companies, especially Allstate, sometimes charged higher rates for minimum than for standard liability coverage. In New York and California, however, we found no instances of an inverse relation between the extent of liability coverage and rates. Research in Allstate and Nationwide websites for nine cities also found no instances of an inverse relationship, but also discovered substantial variation in this relation, including differences as little as two percent and as large as 35 percent.
 Fellowes, loc. cit.
 Consumer Federation of America news release (March 20, 2006).
 Federal Trade Commission, Credit-Based Insurance Scores: Impacts on Consumers of Automobile Insurance (July 2007). Research by Scott E. Harrington and Greg Niehaus (“Race, Redlining and Automobile Insurance Prices,” Journal of Business, v. 71, n. 3, 1998, pp. 439-469) using loss ratios in Missouri found that “auto-insurance-affordability problems urban areas with large minority populations reflect higher costs of providing coverage rather than discrimination.” Similar research by George Galster and Jason Booza (“Are Home and Auto Insurance Policies Excessively Priced in Cities? Recent Evidence From Michigan,” Journal of Urban Affairs, v. 30, n. 5, 2008, pp. 507-527) found similar loss ratios for liability and collision coverages in urban and suburban areas, but lower loss ratios for comprehensive coverage in urban areas.
 Gardner, loc. cit., and Harrington, loc. cit.
 Brian T. Glenn, “The Shifting Rhetoric of Insurance Denial,” Law and Society Review, v. 34, n. 3 (2000), pp. 779-808.
 See discussion by Wanda A. Wiegers in “The Use of Age, Sex, and Marital Status as Rating Variables in Automobile Insurance,” The University of Toronto Law Journal, v. 39, n. 2 (Spring 1989), pp. 149-210.
 Max C. Tang, Auto Insurance In California: Differentials in Industrywide Pure Premiums and Company Territory Relativities Between Adjacent Zip codes, California Department of Insurance (March 25, 2005).
 See Martin J. Katz, “Insurance and the Limits of Rational Discrimination,” Yale Law and Society Review, v. 8, n. 2 (1990), pp. 436-458. See also: William R. Keeton and Ewen Kwerel, “Externalities in Auto Insurance and the Underinsured Driver Problem,” Journal of Law and Economics, v. 27, n. 1 (April 1984), pp. 149-179. Suzanne Yelen, “Withdrawal Restrictions in the Automobile Insurance Market,” The Yale Law Journal, v. 102, n. 6 (April 1993), pp. 1431-1455.
 See Kenneth Abraham, The Liability Century: Insurance and Tort Law from the Progressive Era to 9/11 (Harvard University Press, 2008), ch. 3. Also see Harrington, loc. cit.
 Articles in Tribune Business Service (May 1, 2008; Jan. 20, 2010; May 16, 2010). Consumer Watchdog website. See also: .
 National Association of Insurance Commissioners, Aggregate Consumer Complaints Report (updated March 28, 2011).
 California Department of Insurance, Consumers Auto Complaint Composite Page (www.insurance.ca.gov/0100-consumers/0010-studies-reports/0020-complaint-study/AutomobileComposite.ctra).
 Washington Consumers Checkbook, loc cit., p. 14.
 Consumer Federation of America, Consumer Alert: Some Auto Insurers May Be Underpaying Bodily Injury Claims (December 9, 2010).
Interview with Reggie Whitten, Jan. 24, 2011.
 Data supplied by Birny Birnbaum from NAIC sources.
 Letter from Jonathan Sheldon, National Consumer Law Center, and Mary Griffin, Consumers Union, to National Association of Insurance Commissioners (Sept. 17, 1996). See also Methrin, loc. cit, pp. 7-8 of 16, and Bill Streeter, “New Faith in Collateral Protection,” American Banking Journal, v. 91 (1999).
 Letter from the two groups to congressional financial services leaders sent on July 29, 2009, pp. 5-6.
 Interview with Aaron Bresko, Boeing Federal Credit Union, March 9, 2011.
 State of New York Insurance Department, Report on Examination of the Balboa Life Insurance Company of New York (May 4, 2005).
 See numerous complaints on Complaints Board website. Most target Wells Fargo, which is alleged to have sold redundant force placed coverage on vehicles with sufficient collision and comp coverage from existing policies.
 Interview with Eric Po, loc. cit. National Association of Insurance Commissioners, Report on Profitability by Line by State in 2009 (2010) pp. 100-101.
 Abraham, loc. cit., p. 73.
 Two especially informative papers on uninsured motorists are: J. Daniel Khazzoom, What We Know About Uninsured Motorists and How Well We Know What We Know, Resources for the Future (Dec. 1997, revised April 2000). Lyn Hunstad, Characteristics of Uninsured Motorists, California Department of Insurance (Feb. 1999).
 Tom Elliott, “State Insurance Companies Dispute Uninsured Motorist Projection,” Medill Reports (Chicago, Jan. 21, 2009).
 Stephanie K. Jones, “Uninsured Drivers Travel Under the Radar,” Insurance Journal (Aug. 18, 2003), p. 1.
 Insurance Research Council, news release on Jan. 21, 2009.
 Khazzoom, loc. cit., p. 25. Hunstad, loc. cit., pp. 5-6.
 California Department of Insurance, Commissioner’s Report on Underserved Communities for 1996, 2006, and 2008, loc. cit.
 See discussion in Katz, loc. cit.
 Stephen Brobeck, Understanding the Emergency Savings Needs of Low- and Moderate-Income Households, Consumer Federation of America (2008).
 Keeton, loc. cit., pp. 149-152.
 No-fault auto insurance proposals were overwhelmingly rejected by Arizona voters in 1990 and by California voters in 1995. Abraham, loc. cit., p. 101. And no-fault laws were repealed by Georgia in 1991, by Connecticut in 1993, and by Colorado in 2003. Gardner, loc. cit., p. 3 of 19. An increase in automobile fraud rates and losses following implementation of no-fault laws was also a factor in these policy decisions. See Brent Kabler, “The Case Against Auto Choice,” Journal of Insurance Regulation, v. 18, n. 1 (Fall 1999), pp. 53-79.
 National Association of Insurance Commissioners, Auto Insurance Database Report, loc. cit., p. 236.
 Keeton, loc. cit. Gardner, loc. cit., p. 12 of 19.
 L.H. Ross, the Social Process of Insurance Claims Adjustment (Aldine Publishing Co., 1980).
 Liam Marlaire, “Auto Insurance Law Takes Effect Today,” Tribune Business News (June 1, 2010). Matthew Glass, Research & Commentary: Wisconsin Auto Insurance Update (Jan. 25, 2011).
 Hawaii Administrative Rule, Title 17, Subtitle 6, Chapter 654.
 State of New Jersey Department of Banking and Insurance, Special Automobile Insurance Policy (SAIP) (www.state.nu.us/dobi/division-consumers/insurance/saip.html).
 This program also allows purchase of a $1000 medical payments coverage.
 A study done by Consumer Watchdog (CW) looked at an individual purchasing insurance in the St. Louis area. The individual they chose was a single man, age 30, driving since age 16, owning a Ford Taurus, having a perfect driving record, commuting round-trip 20 miles a day buying a basic limits policy plus Comprehensive and Collision, both with a $500 deductible. CW asked the question, what would happen to this man’s rate if certain factors were varied. The started the test assuming the man was an executive with an MBA living in Richmond Heights, an affluent suburb (Zip Code 63117). His rate was $558. If He was only a high school graduate, his rate rose $71. If he became unemployed, his rate rose another $84. If he moved into the city to Zip Code 63115, his rate rose $347. If he chose to pay on an installment basis, his rate rose $60. If he had a period where he was uninsured, his rate rose $638. If he didn’t have a car, his rate rose another $337. These changes drove his total rate to $2,095, $1,537 or 3.75 times the $558 originally charged the MBA exec in the suburbs.
 Todd Litman, “Pay-As-You-Drive Pricing and Insurance Regulatory Objectives,” Journal of Insurance Regulation, v. 23, n. 3 (Spring 2005), p. 5 of 11. Also, Randall Guensler et al, “Current State Regulatory Support for Pay-As-You-Drive Automobile Insurance Options,” Journal of Insurance Regulation, v. 21, n. 3 (Spring 2003), pp. 31-51.
 Survey data collected by The Roper Center for Public Opinion Research, University of Connecticut.
 Litman, loc. cit., pp. 1-11.
 Litman, loc. cit., p. 3 of 11.
 Edmonds.com business wire, More Consumers Jumping on Pay As You Drive Car Insurance Plans (Feb. 9, 2011). “Pay-As-You-Drive Discounts Let You Save As You Go On Auto Insurance,” Tribune Business News (Feb 20, 2011).
 Consumer Federation of America, State Insurance Department Websites: A Consumer Assessment (November 2008).