Disparate Impact and Unfairly Discriminatory Insurance
`Rates
`Michael J. Miller, FCAS, MAAA
`
`__________________________________________________________________________________________
`Abstract.
`Historic actuarial literature, general insurance literature, and legislative histories reveal “unfairly
`discriminatory rates” to be a cost-based concept. A rate structure is unfairly discriminatory if the
`insurance premium differences between insureds do not reasonably correspond to differences in expected
`insurance costs. More recently a new rate concept has arisen in some court cases which is referred to as
`“disparate impact” (or “adverse impact”). Disparate impact has nothing to do with underlying insurance
`costs and is solely based on the disproportionate impact of the insurance rate structure on the insurance
`premiums paid by protected minority classes defined by race, color, religion, sex, or national origin. It
`would likely be a rare instance where the rate standard of unfairly discriminatory and the concept of
`disparate impact could be applied simultaneously to a risk classification plan without conflict. It is the
`author’s opinion that if the standard of disparate impact eventually prevails over the historical rate
`standard of unfairly discriminatory, then accurate risk assessment will be destroyed, adverse selection
`will be widespread in the insurance marketplace, and coverage availability will suffer.
`Keywords. Risk classification plans; risk assessment; credit scoring; insurance law; rate regulation;
`adverse selection; disparate impact; adverse impact.
`__________________________________________________________________________________________
`
`1. INTRODUCTION
`
`In today’s society, the terms discrimination and disparate impact connote unfairness. Without
`any historical context as background, it would not be surprising for the average person to
`mistakenly conclude that the term unfairly discriminatory is redundant, and that the term
`disparate impact is just another form of unfair rate discrimination. However, a review of
`insurance literature, legislative histories, and court cases reveal that the terms disparate impact
`and unfair rate discrimination are fundamentally different. In insurance ratemaking there has
`always existed a form of rate discrimination which is considered to be fair if the rates are based
`on underlying insurance costs. On the other hand, disparate impact is defined without any
`reference to underlying insurance costs.
`
`The origins of the common rate standards applied by actuaries (i.e., reasonable, adequate, not
`excessive, and not unfairly discriminatory) are discussed in this paper, with special emphasis on
`the rate standard of unfairly discriminatory. The insurance literature and legislative histories
`show the four common rate standards to have meanings based entirely on the underlying
`anticipated insurance costs. It is precisely because these rate standards are cost-based that
`actuaries have adopted these standards as terms of art, as set forth in Principle 4 of the Casualty
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`Liberty Mutual
`Exhibit 1011
`
`276
`
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`Actuarial Society’s Statement of Principles Regarding Property and Casualty Insurance
`Ratemaking (i.e., CAS Statement of Ratemaking Principles).
`
`More recently, some courts have considered the application of a new standard of disparate
`impact (or adverse impact) to insurance rate structures. Thus far no court has actually applied
`the disparate impact standard to insurance rates, but it is only a matter of time before some court
`does so. The standard of disparate impact has its origins in federal civil rights laws and has been
`applied by the courts in a range of issues including employment, educational testing, housing, and
`age discrimination. Unlike unfairly discriminatory rates, disparate impact is not a cost-based
`concept. If applied to insurance, a risk/rate factor will potentially be said to have a disparate
`impact if it more adversely impacts a protected minority class than it does the majority class,
`regardless of its relationship to underlying costs.
`
`It is reasonable to assume a priori that no protected minority class (i.e., race, religion, sex, etc.)
`will be uniformly distributed throughout any given insurance risk classification plan. This
`assumption implies that all risk factors used to measure and assess risk are potentially in violation
`of a disparate impact rate standard, even though each risk factor accurately reflects expected
`losses and expenses.
`
`If a risk classification plan were changed to eliminate one or more risk factors found to have a
`disparate impact, the resulting rates would likely be unfairly discriminatory because the rate
`differences would no longer be based on the underlying insurance costs. Therein lies the
`inevitable and irreconcilable conflict between the two standards.
`
`This paper concludes with a brief discussion of the potential role of an actuary with the
`various issues related to disparate impact. Even though disparate impact is not cost-based, and
`therefore not an actuarial term of art, actuaries do have expertise in measuring the statistical
`significance of any differences in rate impact between the majority class and a protected minority
`class. Actuaries could also provide expertise in defining the data needed to measure disparate
`impact and in establishing the business necessity of any risk factor in question.
`
`2. THE DAWNING OF U.S. RATE REGULATION AND RATE
`STANDARDS
`
`The origin of property/casualty insurance rate regulation in the U.S. is rooted primarily in the
`history of fire insurance. It was solvency concerns and destructive price competition in the fire
`insurance business in the 1800’s that spurred the need for cost-based actuarial ratemaking
`procedures and the need for rate regulation by the states.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`277
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`In the early to mid-1800’s local boards (i.e., voluntary associations of insurers) were organized
`to provide a means of sharing loss data and to enforce uniform rates among the insurers.
`Uniform rates were desired so that rates were adequate to protect against insolvencies and were
`not unfairly discriminatory. The primary concern with unfairly discriminatory rates, often stated
`at the time, was that rich and powerful insureds could unfairly negotiate lower rates than were
`being charged to less influential insureds, even though their degree of risk and underlying
`insurance costs did not warrant a lower rate.
`
`In 1866 a national association of insurers, the National Board of Fire Underwriters (i.e.,
`NBFU), was formed to gather industrywide data and to develop a uniform rate schedule. The
`NBFU decreased the need for local boards. During the ensuing profitable years the insurers
`regularly violated their NBFU membership agreements by engaging in destructive rate-cutting.
`On the verge of disbanding just prior to the 1871 Chicago fire, the insurer insolvencies which
`followed the Chicago fire gave new life to the need for rate discipline and new life to the NBFU.
`But profitability soon returned to fire insurers and destructive rate-cutting returned to the market.
`Rampant rate-cutting caused the NBFU to finally disband in 1887, thereby shifting “control” of
`fire insurance rates back to local boards and associations.
`
`Federal legislation in the 1880’s, which outlawed combinations of insurers in restraint of trade,
`led about half the states to adopt anti-compact laws between 1885 and 1907. The anti-compact
`laws sharply reduced the ability of local boards to maintain uniform, adequate, and fairly
`discriminatory rates. The pressing need for insurers to associate so as to create a combined,
`credible fire insurance database and the existing lack of discipline in fire insurance rating practices
`in the late 1800’s led to many proposals for state regulation of rates.
`
`3. UNFAIRLY DISCRIMINATORY RATES
`
`3.1 Early Rate Regulatory Laws
`The first modern-style rate regulation statute was enacted in Kansas in 1909. The Kansas law
`required fire insurance rates to be filed with the Insurance Commissioner and required the rates
`to be reasonable, not excessive, adequate to the safety and soundness of the insurer, and not
`unjustly discriminatory. Unjust discrimination was defined as charging different rates to persons
`with “risks of a like kind and hazard”.
`
`Soon after enactment of the Kansas law, although largely as the result of the insolvencies and
`the subsequent sharp fire insurance rate increases ensuing from the fires following the great San
`Francisco Earthquake of 1906, the New York legislature appointed the Merritt Committee and
`launched an investigation of fire insurance rates. The Merritt Committee Report led to New
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`278
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`York’s first rate regulatory law in 1911. This law permitted insurers to gather data and act in
`concert to set rates through rate bureaus. The New York law also required fire insurance rates to
`be filed with the Superintendent of Insurance and prohibited unfairly discriminatory rates. The
`law and the Merritt Committee Report made it clear that rates were considered to be unfairly
`discriminatory if different rates were charged to risks in the same class or of essentially the same
`hazard. Class rate differentials based on differences in risk and loss experience were expressly
`permitted by the New York legislation.
`
`New York, working through the National Convention of Insurance Commissioners (i.e.,
`NCIC), offered its new fire insurance rate law as a prototype for other states. Many states (e.g.,
`New Jersey in 1913) did adopt similar rate regulatory laws which permitted collusive rate setting
`through rate bureaus and prohibited unfairly discriminatory rates. Consistently, the clear
`purposes of these early laws were to permit collusion in regard to data gathering and rate setting,
`and to ensure that rates were established commensurate with the degree of risk and hazard being
`insured. In a speech before the NCIC in 1915, the New Jersey Insurance Commissioner spoke
`about the need to base insurance rates on the degree of risk being insured and the unfair
`discrimination that arose when “some people were getting insurance for less than it was worth
`and others were paying for it.”
`
`3.2 McCarran-Ferguson and Modern Rate Regulation
`The enactment of Public Law No. 15 (i.e., McCarran-Ferguson) on March 9, 1945 reaffirmed
`the right of the states to regulate insurance by providing an antitrust exemption for insurance to
`the extent that insurance was regulated by state laws. McCarran-Ferguson spurred a new and
`modern round of state rate regulatory laws throughout the United States. As a result of
`McCarran-Ferguson, the National Association of Insurance Commissioners (i.e., NAIC)
`immediately turned its attention to drafting model rate regulatory laws that could be considered
`for adoption by the majority of state legislatures which were scheduled to begin to meet next in
`1947. The 1945 NAIC proceedings indicate that the model laws and the rate standards were
`based largely on existing state rate regulatory statutes, as witnessed by the following quote from
`the May 12, 1945 Report of the Subcommittee on Federal Legislation:
`
`“On the subject of rate regulation the Committee felt that there were well-defined
`patterns available based upon the actual experience of a number of states which
`could be used as a foundation for the drafting of rate regulatory statutes at this
`time. This fact was recognized by certain segments of the insurance industry
`which prepared so-called model rating bills based largely upon existing statutes
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`279
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`and which were used as guides for the enactment of rate regulatory laws recently
`in several states.”
`
`The NAIC’s model fire/marine and casualty/surety rate regulatory bills of 1946 utilized the
`rate standards of not excessive, inadequate or unfairly discriminatory and required that rates be
`based on consideration of past and prospective loss and expense experience. These model bills
`specifically allowed for the grouping of risks by classifications for the establishment of rates.
`Classification rates could be modified for individual risks if, and only if, the modification was
`based on “variations in hazards or expense provisions, or both.”
`
`The NAIC model bills were a pervasive influence on individual state legislatures. It is not at
`all surprising that the rate regulatory laws throughout the U.S. today contain similar, if not the
`same, language as the 1946 NAIC model bills. As an example, the influence of the 1946 NAIC
`model bills on individual state rate regulatory laws can be found in the California McBride-
`Grunsky Act of 1947 (S.B.1572). This California statute prohibited rates that were unfairly
`discriminatory and specifically allowed for differences in rates between risk classifications, if the
`rate differences were based on the differences in the underlying hazard or expenses.
`
`A new rate regulatory statute was established in California in 1988 with the passage of
`Proposition 103. Proposition 103 reestablished the unfairly discriminatory rate standard, as well
`as placed certain restrictions on some rate factors used in rating personal auto insurance.
`Subsequent to the passage of Proposition 103 new rate regulations were adopted and some lower
`courts addressed the definition of unfairly discriminatory rates in California. In this author’s
`opinion thus far there have been no changes in California to the traditional concept that rates
`should be based on expected costs and not be arbitrary.
`
`4. DISPARATE IMPACT ON INSURANCE RATES
`
`4.1 History
`The concept of disparate impact1 has its roots in certain federal civil rights laws, including the
`Civil Rights Acts of 1866, 1964, and 1991 and the Fair Housing Act (42 U.S.C. Sec. 3604) (i.e.,
`FHA). Broadly speaking, this category of federal laws prohibits discrimination based on race,
`
`1 Note: As in this paper, the terms disparate impact and adverse impact are generally used interchangeably to mean
`that a protected minority class is being adversely and disproportionately impacted as compared to the impact on the
`majority class. Disparate impact and adverse impact are both distinguished from disparate treatment, which involves
`intent to discriminate in a way that is prohibited by federal civil rights law.
`In this paper the terms disparate impact and adverse impact are used with the recognition that the impact may occur
`in neutral processes without the specific intent to violate any civil rights prohibitions. Disparate treatment, based on
`the intent to violate discrimination prohibitions, is not related to actuarial considerations, is a mutually exclusive
`theory from disparate impact, and is not addressed in this paper.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`280
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`color, religion, sex, or national origin. The seminal disparate impact case was decided by the U.S.
`Supreme Court in Griggs v. Duke Power 401 U.S. 424, 430-32; 1971. The Civil Rights Act of
`1991 codified the disparate impact findings in Griggs.2
`
`Disparate impact has been defined by various courts as an unintentional discrimination against
`the protected minority class and its existence is not necessarily illegal. If a plaintiff is able to
`establish that a specific practice leads to a significantly higher adverse impact on the protected
`minority class than on the majority class, the defendant then has the burden and opportunity to
`prove that the practice in question has “legitimate business reasons” or “business necessity” (see
`Watson v. Forth Worth Bank & Trust, 487 U.S. 977, 978; 1988). Even if the defendant is
`successful in showing the practice in question is of a business necessity, the plaintiff still has the
`opportunity to show that other practices would serve the defendant’s business purposes without
`disparate impact against the protected minority class (see Albermarle Paper Co. v. Moody, 422
`U.S. 405, 425; 1975).
`
`In summary, past court decisions seem to suggest that a business practice with disparate
`impact on a protected minority class will be considered illegal by the courts if:
`
`a.
`
`there is a significantly higher adverse impact on a protected minority class than on the
`majority class, and
`
`b. either the practice in question cannot be shown to have a legitimate business necessity,
`or an alternate practice is shown to achieve the business purpose without the
`disproportionate adverse impact on the protected minority class.
`
`4.2 Measurement of Significance
` The Uniform Guidelines on Employee Selection Procedures (adopted in 1978 by the EEOC,
`U.S. Civil Service Commission, Department of Labor, and the Department of Justice) provided
`the so-called “4/5’s Rule” as a guideline for employment selection practices. This guideline
`allows for some disproportionate adverse impact against the protected minority class as long as
`the impact is not considered to be significant by the court. The adverse impact is considered to
`be significant only when the “4/5’s Rule” is failed. For example, if 60% of the job applicants in
`the majority class are hired and only 50% of the job applicants in the minority class are hired, the
`difference in impact is considered not significant and not discriminatory. This is because the
`hiring rate of the minority class is more than 80% of that of the majority class.
`
`
`2 Employment Discrimination Law, American Bar Association, Barbara Lindemann and Paul Grossman, Volume I;
`Chapter 3, 2007.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`281
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`The “4/5’s Rule” to determine significance is not the only test of significance that has been
`used by the courts. In some cases, statistical tests of significance or a showing of a disparity of
`two or more standard deviations have also been applied to determine if the adverse impact is
`significant enough to be a problem. To guard against a relatively small difference being
`considered statistically significant because of a large sample size, some courts have required that a
`statistically significant disparity also have a practical significance.
`
`Although anything is possible in terms of future lawsuits, it is the author’s opinion that the
`“4/5’s Rule” may not be accepted as a test of significance for insurance ratemaking. It is more
`likely that the determination of significance of any disparate impact of insurance rates will be
`based on statistical tests of significance.
`
`4.3 Application to Insurance Practices
`There is a strong legal argument that federal civil rights laws, including the FHA, should not
`be applied to the pricing and underwriting of insurance because of the McCarran-Ferguson
`exemption. Thus far the courts have rejected this McCarran-Ferguson argument. However,
`most of the insurance cases in which the courts have rejected the McCarran-Ferguson defense
`have involved claims of either fraud or intentional discrimination (i.e., disparate treatment).
`
`One such “disparate treatment” case was NAACP v. American Family Mutual Insurance
`Company (978 F.2d 287, Seventh Circuit, 1992). The complaint in the American Family case
`involved an alleged violation of the FHA due to charging higher rates for residential property
`insurance in racial minority neighborhoods. The Court observed that there was an important
`distinction between disparate treatment and disparate impact because the nature of insurance
`inherently requires risk classification and discrimination by degree of risk. As the Court said in
`the American Family case, “risk discrimination is not race discrimination.”
`
`In a more recent insurance case (DeHoyos, et al. v. Allstate, et al., 345 F.3d 290, Fifth Circuit,
`2003), it was charged that Allstate’s residential property insurance rates had a racially disparate
`impact because of the use of credit-based insurance scores in its rate structure. This was a true
`disparate impact case, rather than a disparate treatment case, because intent to racially
`discriminate was not at issue. The Fifth Circuit ruled that McCarran-Ferguson did not preempt
`the application of the FHA in this case. Allstate appealed the preemption decision to the U.S.
`Supreme Court, which refused to take the case. After the Supreme Court declined to review the
`preemption decision of the Fifth Circuit, Allstate settled the case. Even though the Court never
`had the opportunity to address the issues of disparate impact (i.e., the existence and significance
`of the difference, the business purpose, or the potential substitutes for credit data) in the Allstate
`case, it is only a matter of time before some court does.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`282
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`5. DEFINITIONS
`
`5.1 Unfairly Discriminatory
`As previously discussed, the definition of unfairly discriminatory insurance rates has
`historically and consistently been related to the underlying costs of providing insurance. Prior to
`the first rate regulatory law in Kansas, insurance literature consistently refers to the unfairness of
`charging different rates to risks with similar risks of loss and similar hazards. The literature
`surrounding the adoption of the first rate regulatory laws in Kansas, New York, New Jersey and
`the 1946 NAIC model rate regulatory bills are consistent on this point.
`
`Professor C. Arthur Williams, Jr.3 has put forward what is probably the most commonly used,
`and the most succinct, definition of unfairly discriminatory insurance rates as follows:
`
`“An insurance rate structure will be considered to be unfairly discriminatory. . . .,
`if allowing for practical limitations, there are premium differences that do not
`correspond to expected losses and average expenses or if there are expected
`average cost differences that are not reflected in premium differences”
`
`5.2 Actuarial Term of Art
`It is precisely because the concept of unfairly discriminatory insurance rates has historically
`been a cost-based concept, that actuaries adopted that rate standard as a term of art. Although
`this term of art was embodied in much of the early actuarial literature, it was not until 1988 that
`the CAS Statement of Ratemaking Principles was formally adopted, which declared in Principle 4:
`
`“A rate is reasonable and not excessive, inadequate, or unfairly discriminatory if it
`is an actuarially sound estimate of the expected value of all future costs associated
`with an individual risk transfer.”
`
`5.3 Disparate Impact
`Court cases reveal that the term disparate impact is not a cost-based concept and, therefore, it
`is not currently considered to be an actuarial term of art. Disparate impact is strictly a standard
`based on a significantly disproportionate and adverse impact on a protected minority class
`defined by race, color, religion, sex, or national origin. In an insurance context disparate impact
`has nothing to do with the underlying costs of providing insurance.
`
`5.4 Conflict in Definitions
`
`
`3 Insurance, Government, and Social Policy, The S.S. Huebner Foundation for Insurance Education, C. Arthur Williams,
`Jr., Chapter 11, Price Discrimination in Property and Liability Insurance, 209-242.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`283
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`It is likely that the rate standard of unfairly discriminatory will be in direct conflict with the
`application of a disparate impact standard to insurance rates. This conflict will potentially exist
`for nearly every risk factor used to develop property/casualty insurance rates because protected
`classes, most if not all of the time, will not be evenly distributed throughout the various risk
`classifications. If a court or legislature were to order that all disparate impacts be eliminated from
`insurance premiums, it is likely that accurate risk assessment would be destroyed, resulting in
`unfairly discriminatory rates. Paraphrasing a 1915 NAIC speech by the New Jersey Insurance
`Commissioner, unfairly discriminatory rates mean that some people would pay less than the
`insurance was worth, at the expense of other people who would be required to pay more than the
`insurance is worth in order to subsidize the under-payers. It is possible that the only rate
`structure which could survive a strict disparate impact standard is “one-rate-for-all.” If such a
`scenario materializes, adverse selection would be rampant in the insurance market and coverage
`availability would suffer.
`
`6. ROLE OF THE ACTUARY
`
`6.1 Determination of Unfair Discrimination
`The role of the actuary in determining underlying insurance costs and verifying that the rate
`structure is not unfairly discriminatory is well-established and uniquely actuarial in nature. The
`costs which an actuary considers in a review of any rate structure are prospective losses,
`prospective expenses, and an appropriate provision for risk commensurate with the cost of
`capital necessary to support the insurance mechanism.
`
`6.2 Determination of Disparate Impact
`The role of the actuary with disparate impact issues has not yet been fully established.
`Certainly actuaries are not trained to opine on social policies or to determine which minority
`classes deserve the protection of the law. Society’s definition of overall fairness needs to be left
`to the legislatures and courts.
`
`However, actuaries do possess the unique expertise to measure the impact on insurance rates
`of any risk factor and to determine the degree and statistical significance of any apparent
`disparate impact on any protected minority group defined by law. If a court were to find that a
`particular risk factor had a disparate impact on the insurance premiums of a protected minority
`group and the disparate impact was statistically significant enough to be of concern to the court,
`actuaries would be uniquely qualified to opine on the predictive power and business necessity of
`the risk factor in question, as well as opine on any risk factors that might replace the risk factor in
`question.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`284
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`6.2.1 An example
`When disparate impact arises in the context of insurance rates, it will likely be an issue with
`personal auto or residential dwelling insurance. The risk factor in question could be territory rate
`factors, because racial groups likely differ in their geographical distributions. Or in the case of
`auto insurance, the risk factor in question could be age of driver, gender of driver, credit-based
`insurance scores, or etc. In the case of homeowners insurance, rates based on the age of the
`home have already been challenged as having a disparate impact. Since the distribution across
`the various rate classes of racial groups is likely to vary somewhat for every risk factor, there is a
`potential for “disparate impact” with every risk factor.
`
`For purposes of this example, assume the risk factor in question is credit-based insurance
`scores as applied to personal auto insurance. In its July 2007 report in the U.S. Congress, the
`Federal Trade Commission (i.e., “FTC Study”) found that credit-based insurance scores “are
`effective predictors of risk” for auto insurance. The FTC also found that credit-based insurance
`scores “are distributed differently among racial and ethnic groups, and this difference is likely to
`have an effect on the insurance premiums that these groups pay, on average”. While the FTC did
`not attempt to actually measure the effect on auto insurance premiums, or opine on the statistical
`significance of any premium impact, the mere suggestion of a “likely” unequal impact on average
`premiums raises the spectre of disparate impact for this risk factor.
`
`The following sections discuss the role of an actuary in a hypothetical lawsuit where the
`charge is that credit-based insurance scores have a disparate impact on the auto insurance
`premiums for a protected racial minority.
`
`6.2.2 Data to determine disparate impact
`However a court or legislature might define disparate impact as applied to insurance practices,
`it is highly likely that the determination of its existence will involve sophisticated analyses of data.
`Unlike employment/hiring cases, it will be difficult, if not impossible, to accurately apply any
`racial disparate impact definition to insurance rates in an objective, statistical way because the
`racial data needed are simply not available.
`
`The FTC Study was based on racial information for each policyholder obtained from the
`Social Security Administration. Due to limitations in this data prior to 1981, the FTC also relied
`on a Hispanic surname match and Census tract data to identify some Hispanics, Asians, and
`Native Americans. The reliance on a surname match and Census tract data to identify Hispanics,
`Asians, and Native Americans for policyholder records prior to 1981 raises concerns about the
`accuracy of those racial identifications. Plaintiffs in disparate impact cases will likely have access
`to databases that are even less perfect than the database available to the FTC.
`
`Casualty Actuarial Society E-Forum, Winter 2009
`
`285
`3DJH 
`
`

`

`If Rates Have a Disparate Impact Are They Unfairly Discriminatory
`
`In our hypothetical lawsuit, there will be no racial information in the insurer’s policyholder
`records that can be produced through the discovery process. Neither the plaintiff nor the
`defendant will have access to the Social Security Administration’s database, as did the FTC. In
`order to carry the burden of showing disparate impact on any racial group, the plaintiff will
`necessarily be restricted to a conjecture and inference of each policyholder’s race based on
`surname matches, Census tract data, or other potentially inaccurate indicia of race.
`
`Since actuaries routinely use data to analyze insurance rates, actuaries will be able to offer this
`hypothetical court a great deal of expertise with regard to the reliability and credibility of any
`demographic data used to measure the extent of disparate impact on insurance rates.
`
`6.2.3 Statistical significance of disparate impact
`Assume the plaintiff is able to convince the court that its data are of sufficient accuracy and
`that some adverse disparate impact actually exists on the average premiums paid by a protected
`minority group. The next question before this hypothetical court is whether the disparate impact
`is significant enough to be of concern.
`
`Since historically the “4/5’s Rule” relied on by some courts in employment/hiring cases has
`been applied to binary decisions, (i.e., the decision to hire or not hire), it is not obvious how it
`would be applied to insurance rates. Perhaps as long as the impact on the average insurance
`premiums for a protected minority group is no greater than 20% of the impact on the premiums
`for the majority, then the disparate impact is deemed acceptable. However, this is only one of
`many possible tests that might be applied in disparate impact litigation. It is likely the plaintiff in
`this hypothetical lawsuit will argue for a narrower range of acceptability.
`
`Actuaries are well-qualified to opine on the statistical and practical significance of any
`disparate impact found by the court, whether the degree of significance is based on some
`variation of the “4/5’s Rule” or on the application of other common statistical tests of
`significance.
`
`6.2.4 Business necessity and potential replacements
`If our hypothetical court finds that credit-based insurance scores disparately, and significantly,
`impact the insurance premiums of a protected minor