`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`LIBERTY MUTUAL INSURANCE CO.
`
`Petitioner
`
`V.
`
`PROGRESS-IVE CASUALTY INSURANCE CO.
`
`Patent Owner
`
`Case CBM2012-00002 (IL)
`Patent 6,064,970
`
`Declaration of Michael J. Miller
`
`PI!-126G629v2
`
`Progressive Exhibit 2010
`
`Liberty Mutual V. Progressive
`CBM2012-00002
`
`
`
`Declaration of Michael J. Miller
`
`1, Michael J. Miller, hereby declare under penalty of perjury:
`
`Scope of Assignment
`
`1.
`
`I was retained by the law firm of Jones Day, on behalf of the Progressive
`
`Casualty Insurance Company (“Progressive”), to render opinions regarding the
`
`determination of vehicle insurance premiums.
`
`2.
`
`All of my statements and opinions herein are based on my training and
`
`education as an actuary, and on my forty-four years of work experience with the
`
`pricing, underwriting, and marketing of private passenger auto and commercial
`
`Vehicle insurance. Unless noted otherwise, my statements and opinions reflect the
`
`understanding as of January 1996 of a person of ordinary skill in the art of pricing
`
`and underwriting of motor Vehicle insurance.
`
`3.
`
`My retention agreement with Jones Day calls for me to be compensated
`
`at my normal rate of $450 per hour, plus out-of—pocket travel expenses.
`
`Qualifications
`
`4.
`
`I am the owner of EPIC Consulting, LLC and am currently the sole
`
`consulting actuary at EPIC. EPIC is an independent consulting firm that provides
`
`consulting services to the insurance industry related to the pricing, marketing, and
`
`underwriting of property/casualty insurance.
`
`PII-1266629v2
`
`2
`
`
`
`5.
`
`I am a Fellow of the Casualty Actuarial Society (“CAS”), having first
`
`qualified for membership in the CAS i11 1971. I have been elected to two terms on
`
`the CAS Board of Directors. I have also served the CAS as Vice President for
`
`Research/Development, as the Chair of the CAS Risk Classification Committee, and
`
`as the Chair of the CAS Committee on Principles of Ratemaking. As Chair of the
`
`Ratemaking Committee, I was the principal drafter of the Statement of Ratemaking
`
`Principles.
`
`6.
`
`In the past, I have served the Actuarial Standards Board as the Chair of
`
`the Property/ Casualty Operating Committee. In that capacity I was responsible for
`
`drafting several Actuarial Standards of Practice applicable to all property/casualty
`
`lines of insurance.
`
`7.
`
`I began my actuarial career at State Farm Insurance in November 1967.
`
`My entire career at State Farm was in the Auto Actuarial Department where I was
`
`directly involved in the determination of insurance premiums for both private
`
`passenger autos and commercial motor vehicles. My work at State Farm started as an
`
`actuary trainee, doing the arithmetical calculations necessary to determine auto
`
`insurance premiums. In approximately four years I was promoted to a position of
`
`manager where I was responsible for pricing State Farm’s auto insurance coverages
`
`in the eastern one-third of the United States and the five Canadian provinces in which
`
`State Farm operated.
`
`Pll-1266529V2
`
`3
`
`
`
`8.
`
`During the last four or five years of my State Farm career I was
`
`responsible for State Farm’s auto insurance rate filings throughout the United States
`
`and Canada. I was also a member of a management team responsible for the pricing,
`
`underwriting, marketing, and servicing of auto insurance.
`
`9.
`
`During my entire career at State Farm, I was directly involved in the
`
`design, implementation, and administration of private passenger auto and commercial
`
`vehicle insurance rating plans. My work involved those instances when State Farm
`
`introduced new risk characteristics into its auto insurance risk classification plans.
`
`10.
`
`I left State Farm in 1984 and became an actuarial and management
`
`consultant. As a consultant I have provided a wide variety of actuarial and
`
`management consulting services involving many different property and casualty lines
`
`of business. Over my 29-year career as a consultant, I estimate that the greatest
`
`emphasis of my practice has been related to personal auto and residential property
`
`insurance. At one time or the other, I have provided consulting services to most of
`
`the major insurers in the United States and to several state insurance regulators.
`
`11. During my career as a consultant I have been involved with the
`
`formation of a new insurance company that tested new and innovative risk
`
`characteristics and introduced new risk characteristics into its risk classification plan.
`
`I have also consulted many times with established insurers regarding significant
`
`revisions to their existing rating plans, including the use of new risk characteristics.
`
`Pll-1266629v2
`
`4
`
`
`
`12.
`
`Exhibit 2015 is a copy of my curriculum vitae, which sets forth my
`
`experience, qualifications, and publications.
`
`Person of Ordinary Skill in the Art
`
`13.
`
`I understand that the Board in this matter has stated that a person of
`
`ordinary skill in the field of determining vehicle insurance premiums as of 1996
`
`would have a B.S. in mathematics (or equivalent degree); would have at least five
`
`years of actual work experience in the insurance industry determining premiums for
`
`auto insurance; and would be at least an Associate member of the CAS. I will apply
`
`this level of skill in my analysis in this matter and will assume that this level of skill
`
`applies to a person of ordinary skill in the art (sometimes abbreviated below as
`
`“POSITA”).
`
`14.
`
`It is unlikely that a person of ordinary skill as of 1996 would have had
`
`the opportunity to obtain actual work experience introducing a new risk characteristic
`
`into a risk classification system. As of 1996, making improvements in the accuracy
`
`of assessing automobile risk, including the introduction of new risk characteristics for
`
`determining auto insurance premiums, was a slow and deliberate process throughout
`
`the auto insurance industry. The person of ordinary skill would not have had
`
`experience using or applying fuzzy logic to the determination of insurance premiums,
`
`but would have been relatively sophisticated in the use of n1u1ti~variant statistical
`
`PII-1266629v2
`
`5
`
`
`
`analysis of risk classification data. The person of ordinary skill in the art would not
`
`have had experience using telematics data to determine the cost of insurance.
`
`Materials Reviewed
`
`15.
`
`- In preparing this Declaration, I have considered the following materials
`
`listed:
`
`a.
`
`b.
`
`c.
`
`d.
`
`e.
`
`The ‘970 patent (Ex. 1001).
`
`Kosaka (Ex. 1004).
`
`Herrod (Ex. 1007).
`
`The Board’s Institution Decision (Paper 10).
`
`Other materials cited in this Declaration.
`
`Definition of Terms
`
`16.
`
`In the field of motor vehicle insurance as of 1996, a person or ordinary
`
`skill in the art would have understood that “actuarial class” had the same meaning as
`
`risk class. An actuarial class, or risk class, is a grouping of risks (i.e., insureds) with
`
`similar risk characteristics and expected insurance claims loss (or insurance costs). A
`
`person of ordinary skill in the art would have applied actuarial standards in creating
`
`and evaluating a risk characteristic’s eligibility to be the basis for the establishment of
`
`an actuarial class. Actuarial class claims data is used to determine expected insurance
`
`claims loss. This definition is consistent with the definition in the Risk Classification
`
`PII-126Ei629v2
`
`6
`
`
`
`Statement of Principles of the American Academy of Actuaries. A person of ordinary
`
`skill in the art in 1996 would have adhered to this Statement of Principles.
`
`17. A risk characteristic is a measurable or observable factor or
`
`characteristic that has been found to be predictive of future insurance losses.
`
`18.
`
`The future insurance loss (i.e., risk of loss) being estimated is the
`
`product of the probability of an occurrence of an insured claim times the likely cost
`
`of the claim. Because the probability of an insurance claim occurring is a different
`
`value than the probability of an auto accident occurring, auto insurance rates are
`
`typically calculated based on the likelihood of claim occurrence, not the likelihood of
`
`accident occurrence.
`
`19. An actuarial class for a particular risk characteristic has a risk factor
`
`associated with the risk characteristic. A risk factor is a numerical value for that
`
`actuarial class and is used to calculate the expected loss for an insured. The
`
`numerical value is a ratio of the expected loss of one actuarial class to another. The
`
`expected loss for an insured is sometimes called the “pure premium.”
`
`20. An actuarial class also has a rate factor associated with it. A rate factor
`
`is a numerical value for the actuarial class that is used to calculate the premiums for
`
`an insured. It relates to the difference in premiums charged to insureds. The rate
`
`factor reflects not only the differences in the expected losses (i.e., the risk factor), but
`
`Pl]-126662912
`
`7
`
`
`
`also the differences in ‘expected expenses and all other components of the insurance
`
`rate.
`
`21. An insurance exposure base reflects the risk being insured and is the unit
`
`of coverage upon which the insurance premium is calculated. For each of the private
`
`passenger auto insurance coverages, and most commercial vehicle insurance
`
`coverages, the exposure base is per insured vehicle. Premiums for auto insurance are
`
`typically not calculated or quoted on a per person or per driver basis.
`
`22.
`
`The “rated-driver” is the specific driver among multiple drivers in the
`
`household, whose risk characteristics impact on the premium calculation for a
`
`specific insured auto.
`
`Background Regarding Determination of Auto Insurance Premiums
`
`23.
`
`Insurance is generally described as the transfer of risk of financial loss
`
`arising fi'orn the accidental events described in the insurance policy. In the case of
`
`auto insurance the risk being transferred to the insurer is generally described as the
`
`risk of a financial loss arising from the ownership and operation of the insured
`
`vehicle. The premium is calculated to reflect the total insured risk associated with the
`
`ownership and operation of a vehicle that potentially has multiple operators.
`
`24.
`
`The insureds under a personal auto insurance policy are the named
`
`insured(s) listed on the policy’s declaration page, as well as all relatives resident in
`
`the household. The auto insurance policy is written so as to cover the risk associated
`
`PIE-1266629v2
`
`8
`
`
`
`with the ownership and operation of a specific car, rather than written on a per
`
`insured—person basis.
`
`25.
`
`The insurance premium charged for the transfer of risk is determined so
`
`as to reasonably reflect both the degree of risk being transferred to the insurer and the
`
`expected operational expenses of the insurer. Generally speaking, the greater the risk
`
`being transferred, the higher the premium. Due to the influence of the expense
`
`provisions in the premium calculations, the total premium may not be directly
`
`proportional to the degree of risk.
`
`26. An insurance premium reflects more than the degree of risk being
`
`transferred. In addition to provisions for expected fiiture claim costs and claim
`
`settlement expenses, an insurance premium also includes provisions for expected
`
`operational and administrative expenses, and the insurer’s cost of capital. The
`
`difference between any two insurance premiums will not be identical to the difference
`
`in the degree of risk being transferred because the premium must also include
`
`provisions for operating expenses and profit.
`
`27. Auto insurance rates and premiums are regulated in the United States on
`
`a state-by—state basis. While the rate regulatory laws do vary from state to state, there
`
`is much in common. Nearly all states require auto insurance rates to be adequate, not
`
`excessive, and not unfairly discriminatory. These three rate standards are applied to
`
`P[l—1266629v2
`
`9
`
`
`
`the total premium for each auto insurance coverage and are not applied separately to
`
`each component of the premium.
`
`28.
`
`Insurance rates and premiums are generally considered to be fairly
`
`discriminatory if, allowing for practical limitations, any premium differences
`
`reasonably reflect differences in expected losses and expenses. The burden is
`
`typically on the insurer to demonstrate to the regulator that its insurance rates are
`
`fairly discriminatory, and neither excessive nor inadequate.
`
`29. Demonstration that an insurer’s rates comply with the statutory rate
`
`standards typically involves consideration of past and prospective loss experience of
`
`the insurer, the experience of other rate filers, business judgment, and all other
`
`relevant information and data within and outside the state. Any claims data gathered
`
`by the insurer via its risk classification plan, when combined with the insurer’s
`
`expense data, can be important in convincing the insurance regulator that the
`
`insurance premiums in question are fairly discriminatory. However, there are
`
`instances when the premiums must be justified on some basis other than risk
`
`classification data because credible claims data by risk class are either not. available,
`
`or an alternative approach is preferable.
`
`30.
`
`In the case of auto insurance, it is not an insured person that is being
`
`“assigned” to an actuarial class of similar risks. It is the premium charged, and any
`
`future claim losses, associated with the insured car that are coded to an actuarial class
`
`PII-1266629\f2
`
`1 0
`
`
`
`for each of the risk characteristics used in determining the premium for the insured
`
`car.
`
`31.
`
`For example, assume the premium for an insured car is determined based
`
`on only three risk characteristics: the rated-driver of the insured car is an adult driver,
`
`the coverage is subject to a $500 deductible, and the insured is eligible for a claims-
`
`free discount. The insurer’s policyholder records for this insured car will reflect a
`
`separate code for each of the three risk characteristics (i.e., adult driver, $500
`
`deductible, and claims—free). When analyzing the difference in risk between adult
`
`drivers and youthful drivers, the premium and claims data of our hypothetical insured
`
`car will be included with the adult risk class. When analyzing the difference in risk
`
`between a $500 deductible and a $250 deductible, the premium and claims data of
`
`our hypothetical insured car will be included with the $500 deductible risk class.
`
`When analyzing the difference in risk between insureds that are claims-free and those
`
`that are not, the premium and claims data of our hypothetical insured car will be
`
`included with the claims-free risk class.
`
`32.
`
`The premium and loss data from each actuarial class are. the basis for
`
`determining risk factors (and rate factors), which I mentioned above. The risk factors
`
`derived from the actuarial class data, in conjunction with an insurer’s operating
`
`expenses, become the basis for determining rate factors that are associated with each
`
`risk characteristic. An insurer’s base rate or base premium, after adjustment using all
`
`PII-1266629v2
`
`l l
`
`
`
`the rate factors applicable to a specific insured, results in the actual premium for each
`
`auto insurance coverage, for each specific insured auto.
`
`33.
`
`Returning again to our hypothetical example, assume the insurer utilizes
`
`a base rate, or base premium, of $400 which applies to an adult-rated auto with a
`
`$250 deductible and no claim-free credit. Further, assume the rate factor is .85 for a
`
`$500 deductible coverage and a rate factor of .90 is applied if the insured qualifies for
`
`a claim-free credit. Under this scenario the insurance premium will be $306 (i.e.,
`
`$400 base rate x 1.00 adult factor x .85 deductible factor x .90 clairn-free credit).
`
`34. When selecting risk characteristics to use in a risk classification system,
`
`the POSITA would have adhered to certain actuarial standards. These standards
`
`include statistical considerations such as the homogeneity, credibility, and predictive
`
`reliability of the claims data that will be gathered for each actuarial class. There are
`
`also some operational standards for actuarial classes including: the practicality and
`
`cost effectiveness of administering the actuarial class; the degree of objectivity in
`
`defining the risk characteristic; and the inability of the insured to manipulate or
`
`misrepresent the risk characteristic. Exhibit 2012 is a copy of standards in effect as
`
`of 1996.
`
`35.
`
`I have read the Declaration of Mary L. O’Neil (Ex. 1009). I disagree
`
`with certain of her opinions expressed in paragraph 21 of her declaration. Ms. O’Neil
`
`states that “generat[ing] actuarial classes of insurance that group operators or vehicles
`
`P|l—1266629v2
`
`1 2
`
`
`
`having a similar risk characteristic and bas[ing] insurance rates on these classes .
`
`.
`
`.
`
`is required in order to produce insurance rates that meet the statutory standard of not
`
`being unfairly discriminating,” citing the New York Guide (page 14). Ms. O’Neil is
`
`correct that New York regulations (and those in other states) require that vehicle
`
`insurance rates not be unfairly discriminatory, but they do not require that claims data
`
`generated by an actuarial class be usedto avoid unfair discrimination in rates. Ms.
`
`O’Neil also states that “[t]his practice of generating actuarial classes arose out of the
`
`rating law standards adopted by most states beginning in 1946 based on the NAIC
`
`Model Law from 1946 .
`
`. .,” citing an article that I authored. This is incorrect. The
`
`practice of generating actuarial classes predated 1946 by many years, and my article
`
`does not indicate otherwise.
`
`The Kosaka Reference
`
`36. Kosaka discloses a risk evaluation device and an insurance premium
`
`determination device. The risk evaluation device detects the speed relative to a
`
`preceding vehicle. I understand that the speed and wave length data are among the
`
`input values used to determine a range of filzzy risk values via fuzzy logic.
`
`37.
`
`Based on the state of the art in 1996, Kosaka would have taught the
`
`person of ordinary skill nothing with respect to the determination of insurance
`
`premiums using actuarial classes. There are two primary reasons for this.
`
`P||—1266629\r2
`
`13
`
`
`
`38.
`
`The person of ordinary skill in the art of determining insurance
`
`premiums would have had no experience with fuzzy logic. The person of ordinary
`
`skill would have had no understanding of how to determine insurance premiums
`
`using filzzy logic, or an understanding of how to apply fuzzy logic- to data used to
`
`establish actuarial classes. The person of ordinary skill in the art would not have
`
`understood how to process the data in Kosaka using fuzzy logic.
`
`39.
`
`I understand that fuzzy logic relies on a filzzy-set mathematical theory
`
`that results in data sets that are not mutually exclusive. The person of ordinary skill
`
`could not be certain there was any true difference in risk between two Kosaka risk
`
`values produced via fuzzy logic, especially if those risk values were near the
`
`intersection of two sets. Two operators with different Kosaka risk values would not
`
`necessarily have objectively different driving patterns or objectively different
`
`insurance risks. The ordinary skilled artisan would not use such data for purposes of
`
`establishing actuarial classes.
`
`40.
`
`Even if the Kosaka risk values were to be based on objective measures
`
`of relative speed and distance, these data would be incomplete for determining
`
`insurance premiums and generating actuarial classes. To be useful for determining
`
`premiums using actuarial classes, measures of relative speed and distances between
`
`vehicles must be made relative to some safety or risk standard, or to some relevant
`
`circumstances (e.g., speed limits or traffic congestion). For example, assume that two
`
`PIE-1266629v2
`
`14
`
`
`
`cars being op.erated at exactly the same speed relative to the lead vehicle and at
`
`exactly the same distance generate the same Kosaka risk value for both vehicles.
`
`However, the same relative speed and distance from the preceding vehicle likely
`
`represents significantly different insurance risk if the driving pattern occurs on a
`
`congested city street versus a four-lane expressway. Kosaka would assign the same
`
`risk values to those two cars, even though they would comprise a heterogeneous
`
`grouping of insurance risk, not an actuarial class.
`
`41.
`
`The actuarial standards provide for homogeneity in actuarial class data:
`
`1.
`
`Homogeneity
`
`The expected costs for each of the individual risks in a class
`should be reasonably similar. In a given class, there should be no
`clearly identifiable subclasses with significantly different potential
`for losses. Significantly dissimilar risks should be assigned to
`different classes.
`
`The concept of homogeneity is based upon expected costs as
`viewed when the risk is originally classified. It does not suggest
`the system can or should precisely anticipate the subsequent actual
`claim experience of a given insured risk. The occurrence, timing
`and magnitude of an unforeseen event for a specific risk cannot be
`predicted in advance. Thus, it is inevitable that not all risks in a
`class will have identical actuarial claim experience. Instead, the
`individual risks’ claim experience will be statistically distributed
`around the average experience for the class. The concept of
`homogeneity in no way is comprised by this inevitable outcome.
`
`By the same token, differences in expected costs between classes
`do not preclude the actual claim experience of risks in one class
`from being the same as the actual claim experience of risks in
`another class. This overlap phenomenon is both an anticipated
`
`Pl|—1266629v2
`
`15
`
`
`
`and, indeed, statistically inevitable ramification of any sound risk
`classification system.
`
`42.
`
`The Kosaka risk values, standing alone, teach us nothing about
`
`differences in insurance risk (expected losses) b.etween any two insured cars. The
`
`person of ordinary skill in the art would not have generated or applied actuarial
`
`classes, and their associated risk factors and rate factors, to the data processed in the
`
`Kosaka risk evaluation device.
`
`43.
`
`I have read pages 124-127 and 170-172 of the transcript of the O’Neil
`
`deposition taken on April 13, 2013.
`
`I disagree with her opinions expressed in
`
`paragraph 26 of her declaration (Ex. 1009) and in the cited pages of her deposition
`
`transcript regarding the use of actuarial classes in Kosaka. A person of ordinary skill
`
`would have understood that Kosaka’s insurance premium change determination was
`
`not derived from the application of actuarial classes. Kosaka does not mention
`
`actuarial classes. Kosaka does not describe the classification of any risk
`
`characteristics into actuarial classes at any point in the operation of his risk evaluation
`
`device, or in the operation of his insurance premium determination device. Kosaka
`
`instead discloses calculating “risk evaluation values” using fuzzy logic.
`
`In contrast to
`
`the case with the crisp logic applied in classifying risk characteristics into actuarial
`
`classes, I understand that output data values processed using fuzzy logic could
`
`potentially qualify for assignment to more than one group at a time. In contrast, an
`
`PII—1266629v2
`
`1 6
`
`
`
`insured will be classified in only one actuarial class for a given risk classification. As
`
`such, I find no explicit or inherent disclosure of the use of actuarial classes in Kosaka.
`
`44.
`
`I disagree with Ms. O"Neil’s opinions, found at pages 124—27 and 170-
`
`72 of her deposition transcript, that the passages from Kosaka she cites indicate the
`
`use or application of actuarial classes. Kosaka criticizes prior art insurance
`
`arrangements as “unfair” because they would charge the same premium to both
`
`“operators who always operate safely” and “operators who occasionally take risks,”
`
`Kosaka is actually criticizing the prior art system which determines premiums based
`
`on actuarial classes. In addition, Kosaka is not referring to actuarial classes in
`
`disclosing that his risk evaluation values “can be obtained that are matched to
`
`empirical evaluation of an individual.” An empirical evaluation of an individual’s
`
`risk of accident is fundamentally different than the determination of insurance risk
`
`based on claims data generated via actuarial classes. Finally, with respect to the
`
`description in Kosaka that the fiizzy memory stores risk evaluation values determined
`
`“when fuzzy logic has been carried out in advance offline”: There is no disclosure
`
`that any actuarial class data (i.e., premium and claim loss data) are ever input into the
`
`fuzzy logic of Kosaka, whether offline or not. This further confirms that actuarial
`
`classes are neither explicitly nor inherently disclosed in Kosaka.
`
`45.
`
`I disagree with Ms. O’Neil’s opinion in paragraph 24 of her declaration
`
`(Ex. 1009) that the “prepayment amount” disclosed in Kosaka would necessarily be a
`
`PI|—1266629v2
`
`I 7
`
`
`
`“base insurance premium .
`
`.
`
`. determined by the insurer based on information
`
`collected about the policyholder, including coverage limits, deductibles and other
`
`information such as location.” In my opinion, a POSITA reading Kosaka would not
`
`have associated it with any traditional means for determining automobile insurance
`
`premiums. Furthermore, since it is a Japanese patent application by a Japanese
`
`inventor and assigned to a Japanese company, a POSITA would not likely have
`
`understood Kosaka to teach anything about pricing automobile insurance in the
`
`United States under each state’s rate regulatory statutes.
`
`46.
`
`In my opinion, the explanation of tlie “prepayment amoun ” i11 paragraph
`
`24 of Ms. O’Neil’s declaration is incorrect. Contrary to Ms. O’Neil’s understanding,
`
`a POSITA would have understood that the prepayment amount of Kosaka was simply
`
`a deposit from which Kosaka"s charges could be drawn. Kosaka discloses nothing
`
`that suggests the prepayment amount would necessarily be based on information
`
`collected regarding the insured vehicle prior to any monitoring.
`
`The Herrod Reference
`
`47.
`
`The Herrod reference involves electronic equipment for measuring and
`
`recording certain driving behaviors of specific drivers. Certain acceleration data
`
`obtained from each operator’s driving behavior is apparently stored on a removable
`
`card or disk.
`
`Pl!-1266629v2
`
`1 8
`
`
`
`48. Herrod states that the data pertaining to each operator’s driving habits
`
`could be_ used by “safe drivers” to “demonstrate their competence to insurance
`
`companies.” Neither this statement nor any other part of Herrod’s disclosure would
`
`have suggested to a POSITA that the data generated by the monitoring device of this
`
`patent application had any relevance to, or application in, the creation of actuarial
`
`classes or the use in the determination of auto insurance premiums. A POSITA
`
`would have had little or no interest in Herrod, since it was directed principally to the
`
`electronic equipment. A POSITA would have been aware that in 1996 insurance
`
`companies did not require drivers to demonstrate their “driving competence” and, as
`
`such, a POSITA would likely not have considered Herrod to have any meaningful
`
`application or value to the field of determining insurance premiums. At most, a
`
`POSITA may have understood that the reference in Herrod to a “demonstration of
`
`competence” meant that the data could have been used by an insurer to determine a
`
`driver’s eligibility to be offered insurance coverage. However, a POSITA would also
`
`have recognized that Herrod’s data was not suitable for that purpose since, as
`
`explained below, the data is incomplete and unreliable for insurance purposes of
`
`determining insurance premiums.
`
`49.
`
`One reason why a.person of ordinary skill in the art would not have
`
`understood that the data generated by Herrod’s monitoring equipment could be. used
`
`to determine auto insurance premiums is that the data is driver-specific. In other
`
`P||—1266629v2
`
`I 9
`
`
`
`words, the Herrod device separately records the acceleration data for a specific driver
`
`onto a card or disk. (Ex. 1007 at 000003.) However, a POSITA would have
`
`understood that to accurately determine an auto insurance premium, the risk
`
`characteristics of all drivers resident in the household are needed. These- data are
`
`necessary so that the insurer can determine which of the operators in the household
`
`should be the “rated-driver” on each insured car in the household. Herrod does not
`
`disclose that all of the drivers in the household would have cards, and there is no
`
`suggestion that they would.
`
`50. A POSITA would have understood that the “behavioral groups”
`
`disclosed in Herrod could not be actuarial classes. First, the Herrod behavioral
`
`groups data is incomplete and would fail the actuarial standard for homogeneity.
`
`Two persons using the Herrod device may have identical acceleration/retardation
`
`data, yet have significantly different insurance risk (e. g. , one person driving in a
`
`congested urban area, another driving in a rural area). Because of this, a POSITA
`
`would understand.that two drivers could be slotted in the same behavioral group by
`
`Herrod even though they would likely represent significantly different degrees of
`
`insurance risk (i.e., risk of loss). As such, the Herrod data would have been of no use
`
`to a POSITA in the establishment of new actuarial classes.
`
`5 1.
`
`Further, a POSITA would have recognized that there was no way to
`
`determine whether the acceleration data associated with a particular Herrod
`
`PIE-1Z6E629v2
`
`
`
`behavioral group was actually representative of similar risk characteristics because
`
`Herrod discloses no reliable means tor collecting the data needed to make that
`
`determination. Rather, Herrod discloses the use of “accident statistics .
`
`.
`
`. obtained
`
`from a national survey of drivers using the device" for this purpose. (Ex. 1007 at
`
`000002.) In determining actuarial classes and auto insurance premiums, it is claims
`
`data that are used, not accident statistics.
`
`It is my opinion that data “obtained from a
`
`national survey of drivers using the device" would be unreliable for purposes of
`
`establishing an actuarial class, such that no POSITA would have created an actuarial
`
`class that depended on survey data. There would be no way for the insurer to identify
`
`all the drivers using the Herrod equipment. And even if there were a way to identify
`
`a population of drivers to survey, a POSITA would have no confidence in the
`
`accuracy of the survey results, "since it is likely that those with recent accidents would
`
`be under-represented in the people who responded.
`
`Date: May 1, 2013
`
`Signature:
`
`Pll-1265629v2
`
`El