Gregory D Squires & Charis E Kubrin. Journal of Insurance Regulation. Volume 24, Issue 4. Summer 2006.
“Very honestly, I think you write too many blacks … you got to sell good, solid premium paying white people … the white works. ”
— Sales manager for American Family Mutual Insurance Company, 1988 (from case files in NAACP v. American Family Mutual Insurance Company, 1992)
It is unclear whether the mortgage lending or property insurance industry “pioneered” the use of neighborhood, and particularly the racial composition of neighborhood, in evaluating applicants for housing related financial services. But such redlining practices have a long history. And if racial profiling is most closely associated with the behavior of certain police officers and other security officials, it has a “rich” tradition in insurance.
Racial profiling has emerged as a leading civil rights issue in social science research and policy circles today. Profiling refers to practices through which individuals are classified, at least in part, on the basis of thenrace or the racial composition of their neighborhoods and treated differently as a result. Although the debate over racial profiling most frequently focuses on policing and administration of justice issues, such practices are not restricted to this arena. In fact, at least financially and economically, far more damage is done by racial profiling in other areas of public and private life. One of those is the property insurance industry. The costs include not just diminished opportunities for racial minorities, but the exacerbation of uneven development of metropolitan areas, and the many costs associated with that pattern. This article examines the historical and ongoing practices of racial profiling and related discriminatory actions on the part of the property insurance industry in the United States—actions that helped to create and reinforce the linkages among place, race, and privilege. These practices are hardly unique to any particular industry. In fact, they reflect longstanding racial stereotypes that have stigmatized racial minorities throughout much of American society, and continue to do so at great expense to minority communities and metropolitan areas generally. Remedies are available and directions for future policy initiatives are explored.
Insurance, Home Ownership, and Urban Development
The property insurance industry has a long and continuing tradition of racial profiling. If such practices were once considered sound professional business practices and explicitly endorsed by the industry, few publicly defend them today. Yet they persist. While redlining and racial discrimination by mortgage lenders and banking institutions generally have long been subject to research and public policy initiatives (Goering and Wienk, 1996; Haag, 2000; Munnell et al., 1996; Ross and Yinger, 2002; Stuart, 2003), equally pernicious, but less scrutinized, has been the behavior of the property insurance industry (Badain, 1980; Galster et al., 2001; Squires, 1997). Yet insurance is critical, or, in the industry’s term, “essential.” If a potential homebuyer cannot obtain a property insurance policy, no lender can provide a mortgage. The risk of financial loss to the mortgage lender would simply be too great if the property is not insured. Should the home be damaged, the lender needs to know that its investment is secured and that the loan will be repaid. Property insurance on the home, along with the value of the land which could be sold in case the home were totally destroyed, provide that security. Without a mortgage, the vast majority of homeowners would not have been able to purchase their homes. In light of the essential nature of home insurance, lenders often refer their customers to local insurance agents, or increasingly now offer insurance services themselves. So, as the Seventh Circuit Court of Appeals stated in the 1992 case of NAACP v. American Family Mutual Insurance Co., “No insurance, no loan; no loan, no house; lack of insurance thus makes housing unavailable.”
Households experiencing what the industry refers to as a problem of “insurance availability” are not randomly scattered throughout metropolitan areas. They tend to be located within central city neighborhoods, usually with high concentrations of non-white residents. While some rural communities experience availability problems due primarily to limited fire protection, for a variety of reasons this has been a particularly urban problem. For example, in Milwaukee 72.4 percent of homes in white areas compared to 61.6 percent of homes in black areas were covered by insurers required to comply with that state’s disclosure requirements in 1999. (Very few states have such requirements as will be discussed below.) Remaining homes either have no coverage or are protected by smaller insurers or socalled “surplus lines” or “off shore” or “non-admitted” insurers. These insurers are not regulated by the states and, therefore, are not included in state guaranty funds which means consumers are not protected if the companies should go bankrupt (Squires, O’Connor, and Silver, 2001).
Urban communities tend to have older homes with electrical, heating, and other major systems that have not been updated in recent years. Older wood frame homes, generally concentrated in cities, pose a greater fire risk than newer suburban brick homes. The dense nature of housing patterns means that a fire on one property may damage a nearby property, leading insurers to avoid high concentrations of policies in a particular neighborhood. Theft rates are higher in many urban neighborhoods than in most suburban communities. Relatively lower valued dwellings in cities also make urban properties less profitable to insure. A recent insurance industry study of loss costs in eight major metropolitan areas between 1989 and 1994 found that the frequency of claims was 18 percent higher in cities than in the neighboring communities within five miles of the city boundaries; there were 124 claims per 1,000 insured homes in the cities compared to 105 claims in the surrounding communities. And the average claim was 20 percent higher in cities. Consequently, industry costs per insured home were 42 percent greater for urban than suburban policyholders (Insurance Research Council, 1997).
But in addition to risk factors that may differ between some cities and suburbs in general, a host of other practices (discussed below) that are not based on risk adversely affect urban communities. Redlining of older urban neighborhoods, including practices of racial profiling and discrimination, exacerbate urban insurance availability and affordability problems. Compounding the racial effect is the fact that racial minorities tend to have lower incomes, live in lower-valued homes, and reside in cities (U.S. Bureau of the Census, 2002). The connection between property insurance practices and the fate of cities was captured by a federal advisory committee in 1968 which observed:
Insurance is essential to revitalize our cities. It is a cornerstone of credit. Without insurance, banks and other financial institutions will not—and cannot make loans. New housing cannot be constructed, and existing housing cannot be repaired. New businesses cannot expand, or even survive.
Without insurance, buildings are left to deteriorate; services, goods and jobs diminish. Efforts to rebuild our nation’s inner cities cannot move forward. Communities without insurance are communities without hope (President’s National Advisory Panel, 1968: 1).
There is a direct line between the actions of the property insurance industry and the critical problems facing the nation’s most distressed urban communities that was captured in the title of one law review article, “Property Insurance and the American Ghetto: A Study in Social Irresponsibility” (Yaspan, 1970). If progress has been made since the federal advisory report in 1968, the problems of urban insurance availability and affordability, including its racial dimensions, retain their largely urban character. As another legal scholar concluded:
Hardest hit by unavailability and unaffordability difficulties are transitional neighborhoods in older cities and members of minority groups. So long as unavailability and unaffordability problems remain, communities without affordable insurance become communities with diminishing hope (Badain, 1980: 76).
A related reason for the significance of property insurance for cities and the economy in general is the sheer size of the industry. In 2001 total assets of the insurance industry reached $4.2 trillion with the assets of those other than life insurers totaling $881 billion. For all financial services sectors, including banks and securities, total assets were $37.6 trillion. So insurers accounted for just over 11 percent and non life insurers accounted for just over two percent of total assets in financial services (Insurance Information Institute, 2003: 4). In 2000 the property insurance industry (which includes automobile, commercial, marine as well as homeowners insurance) received $299.6 billion in premiums. (Premiums are the dollars collected for policies that are sold.) Homeowners premiums reached $32.4 billion. In 1999 the property insurance industry’s net after-tax income was $20.6 billion. And the insurance industry generally, including life and health insurers as well as property insurers, employed 2.3 million people in the U.S. (Insurance Information Institute, 2002: vii, 60).
Not surprisingly, in 1999 the states that generated the most premiums were primarily large urban states with California ranking first followed by New York, Texas, Florida, and Illinois. Average premiums ranged from a low of $266 in Wisconsin to a high of $861 in Texas. Insurers generally pay out more in losses and loss cost expenses than they collect in premiums. In 2000 property insurers paid out approximately 10 percent more than they received from their underwriting activities. But in most years earnings from invested funds along with money set aside as loss reserves compensate for underwriting losses and enable insurers to generate a profit (Insurance Information Institute, 2002: 14,64).
The property insurance industry, therefore, constitutes an important actor, economic and otherwise, in urban and metropolitan areas. It is also an integral piece of the institutional infrastructure of inequality in urban and metropolitan areas. The industry reflects and reinforces the role of race and place in framing the opportunity structure confronting residents of the nation’s cities and surrounding communities. The restructuring of American cities in recent decades has been accompanied by growing inequality and concentration of poverty along with a range of social problems associated with those developments (Goldsmith, 2002; Harrison and Bluestone, 1988; Jargowsky, 1996, 2003; Massey and Denton, 1993; Wilson, 1996; Wolff, 1995). If the overt expression of racist sentiments has been subdued, the continuing reality of racial profiling, grounded in longstanding and persisting racial stereotypes, reveals the ongoing centrality of racism in the political economy of urban communities. Again, the devil is in the details. One of those critical details is the property insurance industry.
From documents describing industry underwriting guidelines and marketing strategies, court documents, as well as research by government agencies, industry and community groups, and academics, the following pages document historical and contemporary practices of racial profiling and related forms of redlining and racial discrimination on the part of the property insurance industry. Such practices incorporate elements of both disparate treatment and disparate impact discrimination that are unlawful under the federal Fair Housing Act and many state statutes. Under the disparate treatment standard plaintiffs must establish that the respondent intentionally discriminated on the basis of a protected class membership (e.g., race, ethnicity, gender). Under the disparate impact standard intent is not necessary. The universal application of an apparently neutral policy or practice that excludes a disproportionate share of protected class members (e.g., racial minorities) would violate the act unless the respondent could establish a legitimate business purpose for that policy or practice and that no lesser discriminatory alternative is available to accomplish that objective (Crowell, Johnson, and Trost, 1994: 158-162). No parallel legal definition of racial profiling has emerged from the legislative debates and court cases in the fair housing area, but given the legal standards that have emerged it clearly incorporates many of the elements of unlawful practices that have been identified.
Following the discussion of past and present profiling and discrimination, successful efforts to combat these practices are reviewed and policy recommendations are offered to further reduce the role of race in the delivery of property insurance products and services. Racial profiling may not be as visible within the property insurance industry as it is in law enforcement but insurers are equally proficient, perhaps because they have had so much practice.
The Insurance Industry’s Character Problem: Moral, Morale, and Other Hazards
Insurers generate their revenue from the sale of insurance policies. In so doing they incur a range of costs. In 2000 for each dollar collected in premiums insurers paid 79.6 cents for claims, 25 cents for sales and administrative expenses, 2.5 cents in taxes and 1.3 cents in dividends. These costs come to more than 100 percent, which is normal for most insurers’ underwriting activities. Investment income compensates for these losses and permits insurers to generate a profit and continue making insurance available (Insurance Information Institute, 2002: 14). But in order to stay competitive and maximize their returns, insurers need to determine whether a given applicant is eligible for a policy, and, if so, how much to charge.
The insurance industry has one major problem. It does not know the actual cost of its product (an insurance policy) when the product is sold. This makes the decision to sell a policy, the price at which it should be sold, and other terms and conditions most problematic. Property insurance policies that cover homes are generally sold on an annual basis. A premium or price is charged and is often paid in full at the beginning of the policy period. But the cost to the insurer will not be known until the end of that time period. In most cases there is no measurable damage to the home so no claims are filed and the insurer incurs few expenses other than transaction costs involved in processing the application and premium payments. In other cases the property that is insured is damaged and, on occasion, totally destroyed. These costs are generally far higher than the annual premium that is charged.
So the industry tries to determine in advance who is likely to experience a loss and how large those losses will be, and because it is too expensive to collect information on the unique characteristics of each applicant, the industry categorizes applicants into groups based on expected losses. The industry attempts to identify those attributes that account for losses and which people share those attributes. Actuaries develop risk classifications and underwriters determine in which class a given applicant belongs. Two sets of considerations generally enter into this process: 1) characteristics of the property to be insured and the neighborhood in which it is located and; 2) characteristics of the people to be insured. Compounding the complexities is the fact that decisions by insurers can affect the behavior of insureds. Once a homeowner is insured against a particular risk or event that could cause a loss, the household has less of an incentive to avoid such situations and may take fewer precautions to reduce the chances of such an event occurring.
Several property-related factors affect whether or not an applicant is eligible, and if so under what terms. These include the construction of the dwelling which involves the type and age of materials, the condition of the building, and adequacy of maintenance. For example, a wood frame building is more susceptible to fire than a brick structure so all else equal, a wood home is more expensive to insure. Occupancy, or the purpose for which the home is used, is another factor. If the home is also used for certain types of business use, it might be ineligible for home insurance and the owner would have to seek out a commercial policy. Protection is a third consideration. Presence (or absence) of smoke alarms, security systems and other protective devices can affect eligibility for coverage. Proximity to fire hydrants and the quality of local fire protection services are additional factors. Exposure is another property-related consideration. This refers to hazards or risks in neighboring properties such as certain types of industrial concerns, abandoned lots, or other environmental hazards (Wissoker et al., 1998).
Characteristics of people are also important. The industry identifies two general types of hazards that relate to the character and behavior of applicants and insureds: “moral hazards” and “morale hazards.” The former refers to any condition that increases the likelihood of fraud. Someone who is intent on fraud can pose challenges to an insurer. The industry argues, for example, that someone in financial trouble may be more likely to submit fraudulent claims. Requiring credit reports as part of the underwriting process is justified as part of an effort to learn if the applicant poses such a risk. Some companies will not provide a full replacement cost policy (i.e., a policy that will pay the full cost for repairing or replacing damage resulting from a loss) if the market value of the home is substantially less than its replacement cost. A market value policy might be offered but such a policy covers only the current market value (purchase price less depreciation) of items, which is often insufficient to replace them. The fear is that such an insured has an incentive to burn their house down for the insurance money.
A morale hazard refers to a situation where an insured simply becomes less careful once their property is covered. Though no fraud is intended, knowing that an insurance policy is in force may cause some to be less careful in preventing loss than would otherwise be the case. This problem can be dealt with, at least in part, by offering incentives to take preventive action. For example, discounts can be offered for the installation of smoke alarms or security systems. Deductibles are often included whereby the insured is responsible for at least the first few hundred or thousand dollars of any loss (Heimer, 1982, 1985).
So the challenge for the insurance industry is to identify those characteristics of individual properties and people that are conducive to loss and either avoid them or charge higher premiums. The overriding problem confronting insurers remains the fact that they still do not know the cost of its product when it is sold to the consumer. Race has been used as part of the effort to solve that problem. That is, in addition to the tools noted above, a longstanding practice of the industry has been to use race—both the race of individual applicants and the racial composition of neighborhoods—in efforts to classify and price risks. Where race is associated with loss, insurers may have a financial incentive to engage in “statistical discrimination,” but these practices are illegal nevertheless. It is unlawful to use average characteristics of a racial group to determine whether housing related services will be provided to any particular individual (Yinger, 1995: 67-69). Where race is used but is demonstrably not predictive of loss, there is virtually no justification for such practices. Yet drawing on traditional stereotypes that persist throughout the United States (e.g., racial minorities and particularly blacks are still viewed as less motivated to work, more likely to be engaged in crime—Feagin 2000; Bobo and Massagli 2001; Schuman et al., 1997) racial profiling in the insurance industry has been a fact of life that undercuts economic development opportunities for stigmatized groups and hinders urban redevelopment in general (Badain, 1980; Metzger, 2001; Powers, 1997; Smith and Cloud, 1997; Yaspan, 1970). Consequently, minorities, particularly those who live in distressed neighborhoods, face the double whammy of race and place.
The Role of Race in Evaluating Risk and Marketing Products
The property insurance industry has long asserted that risk drives underwriting and pricing activity and that race has virtually nothing to do with these practices. Urban insurance availability and affordability, from this perspective, simply reflect the higher losses in those neighborhoods. As indicated above, one study of loss costs in eight major metropolitan areas found that as a result of greater frequency and higher costs of claims in urban communities than in surrounding neighborhoods, urban policyholders cost insurers 42 percent more per policy than did policyholders in nearby neighborhoods (American Insurance Association, 1993; Insurance Research Council, 1997; National Association of Independent Insurers, 1994). Yet race is a factor that has long been explicitly taken into consideration in evaluating risk (Heimer, 1982; Yaspan, 1970) and many industry practices have an adverse disparate impact on minority communities (i.e., result in a higher share of residents in these communities compared to those in white communities that is denied a policy, charged higher prices, or otherwise offered less advantageous terms and conditions) even though no intentional racial considerations may be present (Kincaid, 1994; Powers, 1997).
The following statement by one marketing consultant illustrates the importance of race, and the link between character and race that was widely and openly expressed at least through the 1950s:
It is difficult to draw a definite line between the acceptable and the undesirable colored or cheap mixed white areas; the near west side (Madison Street) and near north side (Clark Street) still attract the derelict or floating elements with “honky tonk,” mercantiles and flop houses. Any liability in the areas described should be carefully scrutinized and, in case of Negro dwellings, usually only the better maintained, owner occupied risks are considered acceptable for profitable underwriting (National Inspection Company, 195 8).
This statement makes it clear that one of the keys to profitable underwriting was racial discrimination. Apparently, where there are colored or mixed areas it is difficult to determine acceptable from unacceptable areas. And it is the racial composition of such neighborhoods that raises the initial question about acceptability. What is it about race that matters? Apparently, it is the association with derelict behavior. If there is profitable business to be written for “Negroes” (but apparently not for whites) only well maintained properties in which the owner resides are acceptable.
More recently, through the early 1990s, at least one major insurer used explicit racial stereotypes to identify neighborhoods in Richmond, Virginia, where it avoided writing insurance. Among the neighborhood descriptions found in that company’s marketing guidelines were the following:
Difficult Times—Black Urbanite households with many children … they do watch situation comedies and read T.V. guide. Metro Minority Families … mostly black families with school children … they enjoy listening to news/talk radio, and watching prime time soap operas (Housing Opportunities Made Equal, 1998).
In part because of such marketing practices, a jury ruled that Nationwide Insurance Company violated the Virginia fair housing act (Housing Opportunities Made Equal, Inc. v. Nationwide Mutual Insurance Company et al. No. 2B-2704 (Circuit Court, Richmond, VA, Feb. 2,1999)).
Other labels recently employed by various consultants to characterize different types of neighborhoods that have guided insurers and other financial service providers in their marketing include “Low Income Southern Blacks,” “Middle Class Black Families,” and “Urban Hispanics.” At least one of these firms has dropped the race and ethnic labels, but in ways that reflect a downgrading of those neighborhood clusters. “Middle Class Black Families” was changed to “Working Class Families,” and “Low Income Southern Blacks” was replaced with “Hard Times” (Metzger, 2001). The primary result is that many residents of such areas are offered less attractive products than are available in other communities, in part for reasons that are unrelated to the actual risk they pose. The American Family sales manager quoted at the opening makes it clear that race is important and why; whites work. Again, the role of race in identifying underlying character traits is indicated.
In a 1995 survey of insurance agents in the Lehigh Valley in southeastern Pennsylvania three percent stated that an applicant’s race was a factor in then decision to insure a home. When asked to agree or disagree with the statement “The race of a homeowner is never a factor when deciding whether or not to insure a home,” 94 percent said they “Completely Agree.” When asked about “the racial mix of a neighborhood” 88 percent “Completely Agree” it is never a factor. The vast majority, in other words, state that race or racial composition is never a factor (Community Action Committee of the Lehigh Valley Inc., 1995: 5, 7). Yet more than 25 years after the Fair Housing Act was passed, at least some agents continue to openly endorse the use of race in the underwriting of insurance policies. This finding may well understate the number of agents who explicitly take race into account. Survey respondents often give what they perceive to be socially acceptable responses to interviewers that may differ from their true beliefs. When questions are related to race, this generally means providing answers that reflect a more liberal or tolerant attitude than some respondents actually hold (Schuman et al., 1997: 92-98).
In a confidential conversation in 2002 an insurance broker said he was often asked the following two questions, in what he referred to as “verbal underwriting” for multi-family dwellings: 1) is there any section 8 at these properties; and 2) are the kids in this neighborhood more likely to play hockey or basketball. Both of these questions were understood by this broker, and by others, to be subtle code words to elicit information on the race of the tenants (Luquetta, 2002).
Much of this evidence is anecdotal but there is also quantitative evidence of the systematic use of race, and of practices that have a disparate impact on racial minorities. The National Association of Insurance Commissioners examined the distribution and costs of homeowners insurance policies across 33 metropolitan areas in 25 states in the mid 1990s. Researchers found that racial composition of neighborhood remained statistically significantly associated with the number and cost of policies even after controlling on loss experience and other demographic factors (Klein, 1995, 1997). (For contradictory findings in Texas where the effect of race was not significant, see Grace and Klein, 1999).
Some of the reasons for these disparities have been uncovered by fair housing organizations in audit or paired-testing studies. In these experiments, white and non-white “mystery shoppers” (or shoppers from white and non-white neighborhoods) are assigned the same relevant individual, home, and neighborhood characteristics, and then they contact various insurance agents in their communities posing as householders that are interested in purchasing a policy for their homes. The only difference in each pair is their race or the racial composition of the neighborhood of the home they want to insure. Because each pair is matched on the relevant criteria (e.g., income and occupation of householder, age and construction of home, fire protection ratings of residential neighborhoods) any differences in treatment are assumed to constitute racial discrimination.
Tests of major insurers conducted by several fair housing organizations around the country have routinely found disparities in the way white and non-white testers and neighborhoods have been treated. Where white testers and testers from predominantly white neighborhoods generally have been aggressively pursued as customers, blacks and Hispanics as well as testers from black and Hispanic neighborhoods have confronted many barriers.
- The willingness to provide a policy for whites but denying or referring minority applicants elsewhere;
- Not returning calls from minority testers while promptly responding to whites;
- Offering policies with different terms and conditions (e.g., full replacement cost policies for whites, market value policies for non-whites);
- Charging different prices for the same policy;
- Requiring inspections in non-white but not white areas;
- Requiring non-whites to supply social security numbers (so credit checks could be run) but not soliciting such information from whites.
Between 1992 and 1994 the National Fan- Housing Alliance tested major insurers in nine cities and found evidence of unlawful discrimination in the following percentages of tests in the respective cities: Chicago 83 percent, Atlanta 67 percent, Toledo 62 percent, Milwaukee 58 percent, Louisville 56 percent, Cincinnati 44 percent, Los Angeles 44 percent, Akron 37 percent, and Memphis 32 percent (Smith and Cloud, 1997: 108-109).
Similar disparate treatment has been found hi approximately half the tests conducted of major insurers by several fair housing organizations (National Fair Housing Alliance, 2000; Smith and Cloud, 1997; Toledo Fair Housing Center, 1999). The one study that attempted to assess the extent of racial discrimination market-wide (rather than among particular insurers as has been the case with most of the insurance testing that has occurred) did not find differences in terms of access to insurance. Researchers with the Urban Institute examined the Phoenix and New York City markets and found that quotes were offered to the vast majority of white, black and Hispanic testers. But in Phoenix Hispanics were slightly less likely to be offered full replacement coverage on the contents of then homes than were whites (92 percent and 95 percent) and were more likely to be told the quote would not be guaranteed without an inspection of the home (3 percent compared to 0.4 percent among testers who contacted the same agents). Quotes were also 12 percent higher for Hispanics, though in line with rates filed with the state insurance commissioner for different rating territories, which raises questions about the validity of those state-approved delineations. And in New York white testers were slightly more likely to receive both a written and verbal quote (18.1 percent) compared to 11.8 percent for blacks who were more likely to receive just a verbal quote. Though not large, these differences were statistically significant (Galster et al., 2001; Wissoker et al, 1998: 3).
Many insurers market their products in ways that, by intent or effect, favor white neighborhoods. The location of agents is one key indicator of where an insurer intends to do business. A study of agent location and underwriting activity in the Milwaukee metropolitan area found that twothirds of all policies these agents sold covered homes within the zip code or one which bordered the zip code in which their office was located. Coupled with the fact that the proportion of insurance agents in metropolitan areas located in central cities has consistently declined as their numbers have increased in suburban communities, the location (and relocation) of agents has an adverse disparate impact on the service available in minority communities. In Milwaukee, for example, the number of suburban agents increased from 32 to 297 between 1960 and 1980 while the number in the city initially grew from 113 to 157 during the 1960s but then dropped to 125 by 1980. The ratio of agents per 1000 owner-occupied dwellings remained virtually constant in the city (1.01 and 1.09) while increasing from .34 to 1.25 in the suburbs (Squires, Velez, and Taeuber, 1991). A study of two major insurers within the city of Chicago also revealed a concentration of agents in predominantly white neighborhoods, and an avoidance of nonwhite neighborhoods, within the city limits (Illinois Public Action, 1993). Housing values, loss experience, and other economic and demographic changes might account for some of this movement. But studies of agent location in the St. Louis and Milwaukee metropolitan areas found that racial composition of neighborhood was associated with the number of agents and agencies even after controlling for various socio-economic characteristics including loss experience, income, housing value, and age of housing (Schultz, 1995, 1997; Squires, Velez, and Taeuber, 1991).
Underwriting guidelines utilized by many insurers have an adverse disparate impact on non-white communities. Restrictions associated with credit history, lifestyle (e.g., prohibitions against more than one family in a dwelling; references to morality, stability), employment history, and marital status are frequently utilized though no “business necessity” has been demonstrated (Powers, 1997). Two commonly utilized underwriting guidelines are maximum age and minimum value requirements. For example, insurers often reject or limit coverage for homes that were built prior to 1950 or are valued at less than $100,000. The disparate impact of maximum age and minimum value guidelines is most evident. In 1999 23.6 percent of owner-occupied housing units nationwide were built prior to 1950. But 30.6 percent of black owner-occupied housing units and 41.7 percent of Hispanic units were built before 1950. And while 46.0 percent of all owner-occupied housing units were valued at less than $100,000, for blacks the figure was 65.5 percent and for Hispanics it was 50.8 percent. Clearly, these two underwriting guidelines exclude a larger share of black and Hispanic households than whites (U.S. Bureau of the Census, 2002.) Practices that exclude a disproportionate share of a protected group may constitute unlawful, disparate impact discrimination even in the absence of evidence of intent to discriminate. These underwriting guidelines may fall in this category and, arguably, would not constitute racial profiling. But the impact of these underwriting guidelines is foreseeable and, therefore, perhaps the racial effect is not unintentional. Consequently, they comprise part of the complex web of practices that constitutes racial profiling in the property insurance industry.
A related problematic underwriting rule is the moral hazard, noted above, that many insurers assume exists when a property’s replacement value (what it would cost to repair or rebuild a home) exceeds the market value (what it would sell for). For example, if a home would cost $100,000 to rebuild but would sell for only $50,000, the fear is that a homeowner would intentionally burn the home in order to collect the insurance proceeds. Others contend that, despite the apparent incentive, owneroccupants have many social and psychological, as well as financial, investments in their homes and do not present such a risk. The industry itself is split on the question of whether or not homeowners are engaged in any significant arson for profit schemes. Arson has long been a problem in urban communities but it is primarily a problem with commercial rather than personal property. In 1998 arson was reported to be a cause of fires in 10.8 percent of residential and 20.4 percent of non-residential fires. Property damage from arson grew from $1.5 billion in 1991 to $2.4 billion in 1992 and then declined to $1.3 billion in 2000 (Insurance Information Institute, 2002: 92, 96). Arson occurs primarily where property owners have encountered financial difficulties. They may owe back taxes, payments on loans that are overdue, or other debts they are unable to meet. They may have encountered an immediate emergency such as a medical crisis for a family member. But no empirical evidence has been presented to establish that homeowners residing in properties where replacement value exceeds market value are indeed “selling their homes to the insurance industry” (Brady, 1984). Given the neighborhoods where replacement value most often exceeds market value, such an underwriting rule excludes a substantially higher share of homes in non-white than in white neighborhoods (Powers, 1997).
Though clearly an under-researched issue, the claims process is also affected by racial and ethnic stereotypes held by many adjusters. (Adjusters are insurance professionals who evaluate losses and settle claims filed by policyholders (Brenner, 1993: 4)). According to one former adjuster for a major insurer, “black claimants routinely received smaller settlements than white claimants” and her company “routinely set lower reserve amounts for Hispanics than for any other type of claimant” (Saadi, 1987: 55, 58). Her company questioned claims filed by blacks and Hispanics more than those filed by whites, in part because of beliefs that racial and ethnic minorities did not occupy the same occupational status and, therefore, might falsify a claim to get more money or could simply be fooled into accepting less. Lower claims settlements were also justified on the grounds that the medical profession would not provide the same level of care for minorities and, therefore, such claimants could not utilize the funds to the same extent as whites (Saadi, 1987: 55-62).
An examination of claims settled following Hurricane Andrew in south Florida in 1992 concluded that Hispanic claimants were 60 percent less likely than whites to be paid within 60 days of filing after controlling for income and education of claimants and level of damage to homes. A law professor and a sociologist at the University of Miami observed insurance claims mediations and interviewed claimants, adjusters, and mediators. They noted the strong subjective dimension of the claims settlement process and the types of indicators adjusters looked for to identify the likelihood of fraud. Types of neighborhoods people lived in, the cars they drove, their business or professional background, immigrant status, and other social attributes were openly acknowledged by adjusters as factors they take into consideration. Stereotypes adjusters held about immigrants generally and Hispanics in particular led them to be more suspicious of claims from these groups. There was no difference in the claims ultimately paid, just the length of time in paying them, which reinforced the conclusion that untrustworthiness was a major factor underlying the claims adjustment process (Baker and McElrath, 1996, 1997).
There is a contradictory element to these stereotypes. If racial minorities are easier to exploit in the claims process, arguably they would be more profitable (and desirable) customers. But there is no evidence that the industry favors minority applicants on any systematic basis, and it appears just the opposite is the case. Again, limitations in data availability (discussed below) hinder efforts to precisely quantify the role of race in the sale and service of insurance products.
The insurance industry is primarily concerned with risk exposure when it writes policies. But perceptions of race and the places that minorities occupy have long influenced the industry’s methods for assessing and responding to the ambiguous liabilities it assumes when it issues a policy. While debates over redlining and racial discrimination in the property insurance industry have raged for decades, in recent years more aggressive responses have been proposed and in some cases implemented by community organizations, law enforcement officials, and the industry itself.
From Redlining to Reinvestment?
Responses to urban insurance availability problems, or redlining and racial discrimination by property insurers, have taken several forms. The NAIC has issued model laws prohibiting what is referred to as “unfair discrimination” and several states have implemented those statutes. But there has been little enforcement. State insurance commissioners basically have been missing in action in the insurance redlining debate. Their activities focus on rate regulation, establishing licensing procedures, reviewing financial statements and determining solvency standards. Their primary concern is to assure that companies remain solvent (Brenner, 1993: 90). Several insurers have launched a range of voluntary initiatives including educational programs, mentoring initiatives, and related outreach efforts. The most effective responses have come from fair housing organizations that have filed a series of lawsuits and administrative complaints resulting in substantial institutional changes on the part of the nation’s largest insurers. But given the absence of publicly available data on underwriting and marketing activities, it remains unclear how much progress has been made in eradicating the role of race and ameliorating urban insurance availability problems.
Two basic problems have undercut the effectiveness of state insurance commissioners; the absence of political will and the limitation of resources. Those who enforce the law frequently are closely connected to the industry they are charged with regulating. A study of state legislators who are members of insurance committees in ten large states found that almost onefifth either own or are agents for an insurance business or are attorneys with law firms that have large insurance practices (Hunter and Sissons, 1995). Many state insurance commissioners came from and went to the industry prior to and after then public service as then state’s chief law enforcement officer (Paltrow, 1998). And the resources available at the state level to regulate what are increasingly global corporations are insufficient. To illustrate, as of 1998 13 state insurance commissioners offices employed no actuaries to examine the fairness of rates that companies charged and the states approved. Indiana received 5,278 consumer complaints in 1997 bringing the total for the previous four years to more than 21,000. Disciplinary action was taken against 11 insurers. With a limited staff, most complaints were simply forwarded to the companies against whom the complaints were filed (Paltrow, 1998).
Some states are engaged in a range of educational and outreach activities, often in conjunction with insurers and trade associations. The Neighborhood Reinvestment Corporation created a National Insurance Task Force consisting of several leading insurance companies, state insurance commissioners and trade associations to conduct a range of educational initiatives. Homeowners are advised on loss prevention programs including fire safety, crime prevention, and home maintenance efforts in order to reduce their risk potential and increase their eligibility for insurance. Insurance companies and agents are educated on how to identify good business in urban areas and to market their products in previously underserved communities (Neighborhood Reinvestment Corporation, 1995, 1997).
The Cincinnati based National African American Insurance Association is working with Howard University and the District of Columbia Insurance Commissioner to tram minority students for careers in insurance (Mazier, 2001a). The Independent Insurers Association of America and several insurers including Chubb, Safeco, and Travelers have joined in an effort to provide additional support for, and to mentor, minority agents (Mazier, 2001b; Thomas, 1999). These same insurers, along with others, also have launched formal diversity training to assist their agents to serve and work with minority communities (Ruquet, 2001). Some insurers are simply finding profitable business in neighborhoods they had ignored in the past (Bowers, 1999).
Fair housing organizations have been the most effective vehicle for changing the way property insurers serve urban communities, and minority markets in particular. Since 1995 evidence produced primarily from paired testing audits conducted by non-profit fair housing organizations has led to settlements of administrative complaints and lawsuits, and one jury verdict involving several leading insurers including Allstate, State Farm, Fanners, American Family, Nationwide, Liberty and others. This group represents the four largest, and six of the ten largest, homeowners insurers accounting for half the premiums written in the U.S. market in 2000 (Insurance Information Institute, 2002: 61). As a result of these actions, these insurers have provided financial compensation to plaintiffs, eliminated maximum age and minimum value underwriting guidelines, opened agencies in previously underserved urban neighborhoods, developed educational and marketing campaigns in these communities, and financed future testing as part of an effort to evaluate the effectiveness of these reinvestment efforts. In some of these cases funds have been made available to assist homeownership in urban communities, and in one case an affirmative action plan was implemented to increase employment opportunities for minorities at all levels within the company. Examples include a $17 million commitment by Nationwide for damages and various reinvestment efforts in Richmond, Virginia (Housing Opportunities Made Equal Inc. v. Nationwide Mutual Insurance Company et al. No. 2B-2704 (Circuit Court, Richmond, Va. 1999; Millen and Chamberlain, 2001). American Family negotiated a $14.5 million agreement that included $5 million for plaintiffs and $9.5 million to subsidize loans and grants for home purchase and repair (United States v. American Family Mutual Insurance Company (C.A. No. 95-C-0327 [E.D. Wise. 1995]), NAACP v. American Family Mutual Insurance Company (978 F.2d 287 [7th Cir. 1992]). Discussions are currently underway with insurers in several cities and more settlements are likely.
An emerging point of contention is the industry’s use of mathematical formulas in which credit scores are systematically used in determining eligibility for, and the price of, insurance policies. While credit information has been used by some insurers for selected applications in the past, now approximately 90 percent of property insurers use credit scores systematically in their underwriting or pricing activities (Ford, 2003). Insurers claim that people with better credit scores are less likely to file claims. It is argued that those who are more careful in the management of their financial assets also will be more careful in their handling of other assets including their homes and automobiles. Because credit scoring leads to more accurate pricing of insurance policies, according to this perspective, the market is more competitive with more companies offering policies resulting in greater choices for consumers (American Insurance Association, undated; Snyder, 2003; Texas Department of Insurance 2004). Critics contend that, due to racial disparities in income, debt ratios, bankruptcies, inaccurate credit reports, and other financial matters, the use of credit reports exerts an adverse disparate impact on minority communities and, therefore, constitutes a new form of redlining. One problem is that the data the industry rely on in drawing its conclusions are not available for public scrutiny, making independent verification of its claims difficult (Birnbaum, 2003, 2004; Willis, 2003). One outcome of this debate was the introduction of the 2003 “Insurance Credit Score Disclosure and Reporting Act” in the 107th Congress by Rep. Luis Gutierrez (D-IL). This bill would require insurers to disclose the use of credit scoring to all applicants along with the impact of the credit score on the price of all policies, it would prohibit insurers from taking any adverse action regarding insurance coverage based solely on credit history, it would require insurers to refund premiums calculated on the basis of inaccurate credit information and it provided for additional protections for consumers in the use of credit information.
Despite the wide range and large number of new initiatives, it remains unclear just how differently property insurers are serving older urban communities and racial minorities in particular. A critical piece of a future agenda is the documentation of precisely how effectively various communities are being served.
Beyond Racial Profiling: Future Research and Policy Implications
Thirty-five years ago when financial institutions were widely accused of redlining and racial discrimination, Congress stepped in and enacted three critical pieces of legislation. The Civil Rights Act of 1968 (the Federal Fair Housing Act) banned racial discrimination in mortgage lending. In 1975 the Home Mortgage Disclosure Act (HMDA) required most mortgage lenders to annually disclose the number, type, and dollar amount of loans they made by census tract in all metropolitan areas. The Act has been modified several times and now requires lenders to report the race, gender, and income of all applicants, the disposition of applications (e.g., whether they were approved or denied), and pricing information on selected high-cost loans. In 1977 the Community Reinvestment Act (CRA) provided a federal prohibition against redlining. It places on all federal depository institutions (e.g., banks and savings and loans) an affirmative obligation to ascertain and be responsive to the credit needs of the communities they serve, including low- and moderate-income neighborhoods.
These statutes are widely credited for increasing lending activity in low- and moderate-income communities and for racial minorities in particular (Avery et al., 2005a; Gramlich, 1998, 2002; Joint Center for Housing Studies, 2002; Meyer, 1998). Disclosure, coupled with federal prohibitions, appear to have had the intended effect. No comparable requirements exist for property insurers. The limited disclosure data available have had some salutary effects. A broader, nationwide proposal might do for insurance what HMDA has done for mortgage lending.
A recent survey of all state insurance commissioners solicited information on HMDA-like disclosure requirements that were currently in place. Just eight states had some geographic disclosure requirements, all at the zip code level. Data on individual insurance companies were available in just four of these states. Loss experience and cost information was available at the aggregate level in three states. No state made loss or cost data and pricing information available for individual insurers (Squires, O’Connor, and Silver, 2001).
Despite the limitations of available data, they have proven useful in some instances. Plaintiffs in the American Family case noted above utilized the Wisconsin disclosure data in negotiations that resulted in the $14.5 million settlement including commitments to write at least 1200 new policies and open new offices in Milwaukee’s black community, elimination of maximum age and minimum value underwriting guidelines, $9.5 million in subsidized loans to support home ownership, and other reinvestment initiatives (Lynch, 1997; Ritter, 1997).
In an analysis of 1999 Wisconsin data researchers found that six insurers had a market share in white zip codes that was at least 50 percent larger than their share in black areas. In regressing the percentage of owner-occupied dwellings covered on racial composition, race was negatively and statistically significantly associated with coverage for each insurer. Controlling on income resulted in a statistically significant finding for two insurers, Prudential and Integrity Mutual (Squires, O’Connor, and Silver, 2001). Data on loss experience were not available. Research reported above by Klein (1997) and Schultz (1995, 1997) did control on loss experience because in their capacity as employees of the National Association of Insurance Commissioners and the Missouri Department of Insurance they had access to information not available to the general public. Independent investigation of Prudential by fair housing organizations found that this insurer utilized maximum age and minimum value underwriting rules that adversely affected minority neighborhoods, placed relatively few agents in minority communities, refused to provide African American and Hispanic callers with the same level of information provided to white callers, and took other actions that made insurance less available in minority neighborhoods in Milwaukee, Philadelphia, Richmond, and Washington, D.C. A formal fair housing complaint was filed and is currently pending (National Fair Housing Alliance v. Prudential, Case Number 1:01CV02199, U.S. District Court for the District of Columbia, 2001).
From a public policy perspective, the logical next step is to enact a federal disclosure requirement for property insurers modeled on HMDA. Such a requirement would call for insurers to publicly report, on an annual basis, Information on applicants, properties, and neighborhoods including: the race, gender, and income of applicants; type of policy and amount of coverage applied for; replacement value of home; disposition of those applications; price of policy; census tract in which the property is located; structure (e.g., brick or frame) and age of home; number of rooms and square feet of home; number and severity of claims; and distance to nearest fire hydrant.
Such disclosure would allow for far more comprehensive understanding of which, if any, markets were underserved and would facilitate, in particular, understanding the extent to which race remains a factor. This information could assist insurers in their marketing strategies. It would help state insurance commissioners target scarce enforcement resources. And it would help community organizations identify potential partners for reinvestment initiatives. As John Taylor, Executive Director of the National Community Reinvestment Coalition, observed regarding disclosure in mortgage lending, “The mere act of data disclosure motivated partnerships among lending institutions, community organizations, and government agencies for designing new loan products and embarking on aggressive marketing campaigns for reaching those left out of wealth building and homeownership opportunities” (National Community Reinvestment Coalition, 2001b).
Many fair housing and community development advocates, along with some policymakers, also have endorsed CRA-like requirements for the insurance industry. The proposed Community Reinvestment Modernization Act of 2001 (H.R. 4893, 106th Congress, 2nd Session, which has been introduced in subsequent years but not enacted), would establish an affirmative obligation for insurers to provide insurance products and investment activity in low- and moderate-income neighborhoods, along with comprehensive disclosure of where such services were being offered. Massachusetts requires insurers to invest in low-income communities in exchange for tax relief offered by that state. California has created a voluntary program in which community groups bring investment opportunities to the insurance commissioner who attempts to attract commitments from insurers in that state to finance those projects (Luquetta and Goldberg, 2001).
These are baby steps, however, relative to what lenders have been doing for decades and what appears to be the needs of many low-income and particularly minority neighborhoods. Again, absent systematic disclosure, it is difficult to identify areas of greatest need or appropriate intervention strategies. State regulators currently have the necessary data or the authority to collect them but few have demonstrated a desire to do so. Social reform frequently bubbles up from the local level to states and the federal government. In light of the history of racial profiling and redlining in the property insurance industry, the contentious nature of responses, and the questions that persist, the time appears ripe for a federal insurance disclosure requirement.
Despite the limitations of current data availability, there is substantial anecdotal and quantitative evidence that indicates the persistence of racial profiling, discrimination, and redlining on the part of property insurers. But the fundamental causes of these problems extend far beyond the insurance industry. The specific policies and practices that have been identified are firmly grounded in stereotypes that continue to permeate the United States. A number of regulatory, legislative, and voluntary industry initiatives could ameliorate racial profiling and discrimination within the property insurance industry. But more meaningful progress in combating these industry-related problems may await more progress in addressing the problems of stereotyping and discrimination in American society generally.
Research on racial attitudes demonstrates that white Americans continue to view blacks as being less intelligent, less hardworking, and more prone to criminal behavior than whites (Feagin, 2000: 109-140). When asked to account for racial disparities, lack of motivation on the part of blacks is the argument with the greatest appeal among whites. Their problems would be largely solved if they worked harder, according to this dominant perspective. Whites exhibit little recognition of past or present discrimination as a factor blocking black progress (Schuman et al., 1997: 155-170). Such beliefs reflect and reinforce patterns of inequality leading to structured or institutionalized racial inequalities that often appear to be inevitable if not natural outcomes of intrinsic cultural characteristics (Bobo and Massagli, 2001). Concerns with work and morality on the part of insurance agents, underwriters, and others simply reflect stereotypical attitudes that transcend any one industry.
Once formed, stereotypes, and the structured inequalities they generate, change slowly. If there is a kernel of truth to stereotypes (e.g., black unemployment is higher than white unemployment) there is a tendency to paint everyone in the group with the same broad brush. People respond to labels, and their stereotypical images of those to whom the label has been attached, rather than to individuals in those groups. This results in sweeping misjudgments that have critical racial and spatial consequences (Bobo and Massagli, 2001). Racial segregation, the uneven development of metropolitan areas characterized by urban sprawl and concentrated poverty, and the associated social costs are just some of those consequences (Orfield, 1997, 2002; Rusk, 1999). For an industry like insurance that depends on risk classifications and the categorization (influenced by stereotypes) that this entails, the negative consequences are magnified.
One kernel of truth may well be that some urban neighborhoods pose greater risks to insurers than other neighborhoods that are not underserved. Insurers may be responding to signals of the marketplace in their underwriting and pricing decisions. But to the extent that objective measures of risk explain the industry’s behavior, a key question is why various neighborhoods pose different risk levels. To the industry, such uneven development is largely a reflection of the culture, morality, and behavior of residents with race being a major determinant. Rarely does the industry point to disinvestment by private industry, fiscal crises of municipalities, public policy decisions that have long favored suburban over urban communities (e.g., federal highway construction, exclusionary zoning laws, mortgage deductions and other subsidies for home ownership), steering by real estate agents, subjective and discriminatory property appraisals and many others (Gotham, 2002; Jackson, 1985, 2000; Massey and Denton, 1993). Given these structural realities and subjective stereotypes of the industry, eventually the prophecy becomes self-fulfilling. So it becomes “rational” to avoid some minority communities. But this reflects the “crackpot realism” Mills wrote about more than forty years ago (Mills, 1958: 185-6). Such behavior is rational, only given the larger irrationality of private practices and public policies that have nurtured uneven development (Dreier, Mollenkopf, and Swanstrom 2001). As the evidence cited earlier indicates, however, the industry is not responding just to risk. Race appears to have an independent and adverse impact even after loss experience, risk, and other objective measures are taken into account (Klein, 1997; Schultz, 1995, 1997).
Racial profiling persists in the insurance industry and it leads to unlawful disparate treatment and disparate impact discrimination and exacerbates uneven development and racial inequality generally. This dynamic is grounded in unflattering racial stereotypes that reinforce these structural dimensions of inequality. Profiling and discrimination may be less pervasive today than in previous decades, or these practices may simply be more subtle. Progress appears to have been made in recent years in part from “universalistic” approaches like loss mitigation and other educational efforts directed at urban consumers and insurers generally. But racial disparities resulting from both objective economic factors and subjective discriminatory practices continue and, to be effective, proposed remedies should be mindful of the overt and subtle racial dynamics. The necessary data do not exist to draw precise conclusions regarding the extent to which objective and subjective considerations drive these decisions. Insurers will always face the problem of not knowing the actual costs of its product when that product is sold but steps can be taken to maximize the extent to which such decision-making is predicated on actual risk and minimize the role of race.
Insurance unavailability or unequal terms and conditions on which insurance is available, can deny homeownership before a family even has applied for a mortgage loan to buy a home. Insurance redlining denies homeownership particularly in better neighborhoods, forcing families to seek housing in less desirable communities characterized by inferior education, limited opportunities for employment, exposure to crime and other indicators of distress. These patterns reinforce the links among race, place, and privilege. While insurance has been subjected to far less inquiry and policy debate than other housing and financial services issues, insurance plays a critical gatekeeping function in the distribution of privilege.