III. Insurance Data That Could Support Defects Investigations
IV. Data Collected By Insurance Companies
V. Legal Issues
The Transportation Recall Enhancement, Accountability, and Documentation (TREAD) Act (Pub. L. 106-414; November 1, 2000) directs the U. S. Department of Transportation to undertake certain activities to improve highway safety. The National Highway Traffic Safety Administration (NHTSA) is responsible for implementing these provisions. Section 3(d) of the TREAD Act directs NHTSA to study the feasibility and utility of obtaining aggregate information from insurance companies regarding claims made for private passenger automobile accidents, as follows:
NHTSA investigates potential safety defects in motor vehicles and motor vehicle equipment under 49 U.S.C. § 30166. The purpose of these investigations is to determine whether safety-related defects exist in motor vehicles or motor vehicle equipment and whether manufacturers should conduct notification and remedy campaigns, known as recalls, thereby reducing the potential for crashes, injuries, fires, and deaths. NHTSA bases its decisions on whether to open defect investigations on the information that is available to it. If the agency obtains more comprehensive information about crashes and injuries that may have been caused by defects in motor vehicles, it will be better able to focus its investigative resources on potential defects that have led to real-world safety problems. Thus, there may be information in the hands of automobile insurance companies that could help the agency significantly in identifying potential defects.
NHTSA has obtained some claims information on an informal basis from a limited number of insurance companies in the past. However, those insurance companies have supplied most of that information in response to specific requests from the agency, as opposed to providing it on a regular basis as an "early warning" of possible safety problems. In the past, some insurance companies have refused to provide information requested by NHTSA during its investigations.
This document summarizes the agency's defects investigation data needs, the data collected by the insurance industry, and the potential utility of the insurance industry's data.
In order to gather the necessary information to conduct this study, the agency used a two-pronged approach. First, the agency sent identical Special Orders to the eight largest automobile insurance companies, requiring each of them to respond to six questions. Second, NHTSA published a Federal Register notice seeking input from the public. Each of these approaches is described below.
Because NHTSA had a limited amount of time to conduct the study, we concluded that the most efficient and effective way to obtain information from the insurance industry was by issuing Special Orders. The agency sent the Special Order on November 29 and November 30, 2000, with all responses due by January 5, 2001. All eight companies submitted responses. However, some responses were not complete. NHTSA intends to follow up with those companies. However, given the time constraints, any supplemental responses are not included in this study. A copy of the text of the Special Orders is attached in Appendix 1.
Some of the companies submitted information about their databases but requested confidential treatment for that information. NHTSA has granted confidential treatment for some of the submitted information, and this information must be withheld from public disclosure. NHTSA has denied confidential treatment for other portions of the submitted information. However, the companies that submitted these portions have the right to request reconsideration of the agency's decision. Accordingly, until the agency's decision not to afford confidential treatment for these portions becomes final, this information also must be withheld from public disclosure.
Federal Register Notice
On December 11, 2000, NHTSA published in the Federal Register a request for comments notice (65 FR 77339). The notice stated that NHTSA was seeking input from the public to assist the agency with the study. The comment period closed on January 5, 2001. Two groups submitted comments. Four insurance associations (Alliance of American Insurers, American Insurance Association, National Association of Independent Insurers, and National Association of Mutual Insurance Companies) that "collectively represent virtually the entire U.S. private passenger automobile insurance industry," joined together to send consolidated comments outlining their concerns about privacy and the possible burden that a data collection requirement might present to small companies. Also, the Insurance Institute for Highway Safety's (IIHS) Highway Loss Data Institute (HLDI) provided information about the data that HLDI receives from insurance companies.
III. Insurance Data That Could Help to Identify Potential Safety Defects and Support Defect Investigations
Safety-related defects investigations, conducted by NHTSA's Office of Defects Investigation (ODI), are initiated based on the information available to the agency. In most cases, ODI opens its investigations on the basis of a trend in reports of incidents. Vehicle owners, consumers, fleet managers, etc., file such reports with NHTSA in various ways. Approximately 70 percent of the investigations that ODI opens are based on these reports. To file a report with NHTSA, the person experiencing the incident must be aware of NHTSA and its mission; know how to contact NHTSA; and be willing to expend the effort necessary to file the report. ODI conducts an ongoing outreach effort to make the public aware of NHTSA and its mission and to encourage consumers to report defects to the agency. However, ODI estimates that only a small portion of the incidents that occur that might be related to a potential safety defect are reported to the agency.
Insurance companies regularly receive reports of crashes or fires involving insured vehicles that, at a minimum, involve property damage. Also, the first report filed by an owner is typically made to the insurer since the owner's first need is to repair or replace the vehicle or to cover medical bills. Because of this, obtaining access to insurance company's claims data could provide NHTSA with a more timely identification of emerging trends in crashes or fires that may be caused by safety-related defects in motor vehicles or equipment.
In order to fully utilize a report about a potential defect, NHTSA needs the following information:
|Vehicle information||Personal Identifiers||Incident||Claims|
|full VIN (or tire identification number)||full name||date||failed part or subsystem|
|make||address||time of day||possible defect|
(also fax number or email address if possible)
|location||vehicle contributing factor|
|model year||type of crash|
|mileage at incident||identity of investigating police jurisdiction|
|identification of other vehicle(s) and people involved|
|number of persons injured or killed|
Currently, only one insurance company (State Farm) provides a significant amount of information to NHTSA about potential defects. Moreover, State Farm generally provides such information only in response to specific requests, although on rare occasions, State Farm has provided information without such a request. However, State Farm has not routinely provided the full VIN, personal identifiers, or specific crash information. Without that information, it is impossible for NHTSA investigators to follow-up on the information to see if it identifies or supports a defect trend. State Farm has agreed in the past to provide personal information to the agency only if it could obtain permission from the policyholders to release their names to the agency.
It should be noted that State Farm is providing similar types of personal information for other purposes. The company is participating in a study with the Children's Hospital of
Philadelphia /University of Pennsylvania (CHOP/Penn) in which it provides the hospital with detailed information about incidents involving children. State Farm has trained 7,750 claims representatives from 365 field offices to obtain the data and get consent from the insured to be included in the study. These claims representatives, on a daily basis, identify qualifying claims and electronically transfer to State Farm headquarters information, including such personal information as driver names and the ages and genders of the children in the vehicle at the time of the incident. State Farm claims representatives then seek permission from policyholders to release this personal information. State Farm headquarters conducts several quality assurance checks and forwards the data daily, via email, to CHOP/Penn. CHOP/Penn conducts several more quality assurance checks and selects a representative sample to be included in its surveillance system by subjecting the data to an automated sampling algorithm. CHOP/Penn sends the contact information to Response Analysis Corporation (RAC) in New Jersey to conduct a telephone interview with the insured. In the telephone interview, RAC obtains in-depth information about the crash and any injuries to the child passenger(s).
The CHOP/Penn surveillance system provides a valuable service to the traffic safety community through its participation in this system. The consent rate for claimants was 85 percent. This indicates that it is possible to obtain consent from policyholders in order to foster advances in traffic safety.
Early Warning Requirements
In Section 3(b) of the TREAD Act, NHTSA is required to issue a final rule requiring manufacturers of motor vehicles and motor vehicle equipment to report certain warranty, claims, and incident data that could indicate the existence of safety defects. NHTSA published an Advanced Notice of Proposed Rulemaking (ANPRM) on this subject in the Federal Register on January 22, 2001 (66 FR 6532). Comments are due by March 23, 2001. The ANPRM refers to warranty data, serious injury or death claims, property damage claims, field reports, consumer complaints, changes to parts, fuel leaks, fires, and rollovers. This information should help NHTSA learn about possible safety-related defects that warrant investigation sooner than under its current method, which relies primarily on consumer complaints received by the agency. Motor vehicle and equipment manufacturers would be affected by these changes, but not the insurance industry.
The ANPRM indicated that the agency is considering requiring motor vehicle and equipment manufacturers to report incidents of which they receive actual notice which involve deaths or serious injuries which are alleged or proven to have been caused by a defect, regardless of whether there is a "claim." We believe that to achieve the goals of the TREAD Act, "claim" must be construed broadly. For example, we have tentatively concluded that it includes subrogation claims filed by an insurer against a manufacturer. Subrogation information from the manufacturer would not be as timely as similar information from insurance companies, since the insurance companies have knowledge of the matters leading to the subrogation claim prior to filing it with the manufacturer.
IV. Data collected by insurance companies
The primary business of the insurance industry is to issue and underwrite policies and to pay claims. Insurance companies generally have claims and subrogation databases. These data bases vary in levels of sophistication. The majority of insurance companies are small and operate in limited geographical areas. Some of the smaller companies do not have computer databases, or if they do, they are rudimentary systems. We analyzed the database information provided by the eight insurance companies in response to the Special Orders to determine whether the data within their databases would be useful to the agency.
Electronic Claims Data
It appears that only State Farm and USAA currently have electronic databases that can be used to point to possible safety defects. State Farm's Claims Research Auto Support Hotline (CRASH) database is most likely to be useful. The "type of loss" and "facts of a loss" information in State Farm's CRASH database appear to provide information as to whether the loss was caused by a fire or collision and about the type of collision, which could identify potential safety defects. The CRASH database includes fields for make, model, model year, and Vehicle Identification Number (VIN) so specific vehicles can be identified. Suspected defects are also reported to CRASH by claims representatives. State Farm warns that "This process involves judgment, which suggests that merely obtaining raw data will not be helpful. A report on a single case could be serious enough or dispositive enough to warrant reporting on the case alone. Other cases would require multiple reports to convince the reviewer that [further action] is warranted."
The CRASH database is unique to State Farm. It is the largest automobile insurance company in the United States, handling over 10 million claims annually. It may not be feasible for other, smaller insurance companies to create similar databases. For example, GEICO General notes that it might be possible to search for possible defects, but expressed concerns related to the resources that would be required to perform this type of analysis. GEICO General stated that, in order to do a search, "a stand-alone database would have to be created for this specific purpose." Smaller insurance companies likely could not afford the time or expense of setting up this type of system.
USAA uses two databases to store information on automobile claims: a Loss Report Database and a Claims Database. If these two databases could be merged together, the data they contain might be useful in identifying potential safety defects. The Loss Report database has a product defect flag that allows claims representatives to indicate yes/no as to whether the loss is alleged to involve a product defect. It also has product name and product defect description fields. Data elements that can specifically identify the vehicle are also available (make, model, model year, and VIN). A fire investigation field is available on the Loss Report database, but collision information and collision type (accident description code) is only available on the Claims database. The Claims database does not have any vehicle information or defect information.
USAA thinks the vehicle defect information is not reliably and consistently collected: "USAA does not track vehicle defects systematically." The information collected "is based on the judgment of the claims representative and is used only to refer the matter to the subrogation unit." The subrogation unit investigates each case separately and does not aggregate claims information. Therefore, USAA concludes that "it is not feasible to provide a report containing aggregate claims information on a regular basis that may serve as a pointer to possible safety defects that could be investigated by the agency."
Table 2 shows the data elements available on the claims databases of six insurance companies. Because some of the information provided has been classified as confidential, we do not identify the companies by name. Data elements were not provided by two of the insurance companies and their responses to the Special Order provided no useable information on data element availability. Therefore, these companies are not included in Table 2.
Four of the six insurance companies in Table 2 have other vehicle information available in their databases. This information is not useful to identify possible safety defects, because the data indicates the presence of a part but does not indicate if the part or system failed. Most of this information is contained in flag fields. For example, the power steering field appears to indicate if the vehicle did or did not have power steering. Other vehicle flag data fields available are: power brakes, vinyl top, stereo, type of radio, air conditioning, and passive restraint system. A few of the databases also contain information available on tire size and engine type. None of these variables store information that indicates if a particular part or system failed or contributed to the crash or fire.
|Data Elements available on the Claims data|
|Company 1||Company 2||Company 3||Company 4||Company 5||Company 6|
|Make, Model, and Model Year||X||X||X||X||X||X|
|Type of Crash||X||X|
|Vehicle Subsystem Involved|
Three insurance companies, GEICO General, Progressive, and Allstate, mentioned that they use a software product sold by CCC Information Services, Inc. (CCC) to estimate repair costs. The information provided by CCC to the insurance companies does not include specific parts listings. Only information on broad categories of repair are available such as: sheet metal replacement or repair, mechanical/electric replacement or repair, frame replacement or repair, and paint material. GEICO General says, "The data stored in the CCC database provides detailed estimates information to include individual parts listing, parts number and parts prices. This detailed information is retained in the CCC database. Only summary estimates and payment data is transmitted to GEICO General for storage on GEICO General internal databases." Allstate and Progressive also do not have detailed repair information in their databases.
The insurance associations state that there are approximately 300,000 subrogation cases arbitrated yearly, but most of these are claims against other insurers or self-insured entities. In their response to the Federal Register notice, they explain that "virtually all property/casualty insurers are signatories to national agreements that mandate binding arbitration to resolve auto subrogation disputes between insurers." Also, "nearly all automobile subrogation claims are between insurers and self-insureds that are party to the national arbitration agreements." According to the American Insurance Association, there are no published numbers of subrogation cases against motor vehicle or parts manufacturers.
The industry stated that subrogation claims against vehicle or component manufacturers by insurance companies are rare and they are only pursued if financially viable. Allstate says, "Subrogation against vehicle or component manufacturers by our company is a relatively infrequent occurrence." USAA states that "as a practical matter, the costs of hiring experts and pursuing litigation, as well as the difficulty in obtaining sufficient evidence, may outweigh any potential recovery in certain cases."
Subrogation files (whether leading to a formal claim or not) contain important information that may be of use to the agency. Liberty Mutual states, "On receipt of the referral in Subrogation [field office's electronic referral for subrogation], the claim is thoroughly reviewed to confirm the facts of the loss, place the at fault party on notice of our claim and follow up with supporting documentation (estimates, police reports, expert reports pictures, statements, etc.)." Thus, the information in the insurer's file often goes beyond information in the hands of the vehicle or equipment manufacturer.
Allstate has a "subrogation unit specializing in subrogation recoveries relating to defects. By centralizing the handling of these cases we have been able to further identify specific manufacturers and/or trends in products that cause losses across the country." Allstate explains, "This specialty unit has been able to learn of recalls and has utilized the expertise of attorneys to help direct further investigations needed on these more complex loss types."
Most of the insurance companies have databases to track their subrogation cases. Generally, information about the claim, type of claim (automobile fire, automobile collision, or collision and fire), type of third party (uninsured person, other insurance company, or collection agency or attorney), amount of damages, recovered damages, and tracking information are available on subrogation databases' claims file. However, some important information that may be in the claims file, such as vehicle information, or information about the alleged defect or involved vehicle components are not typically contained in the electronic databases.
NHTSA believes that information on subrogation investigations and subsequent claims against a vehicle or component manufacturer may point to possible safety defects. The information NHTSA would find useful is the manufacturer involved, the vehicle involved, the suspected or alleged defective part or system, whether a fire or collision occurred, and, if a collision, the details about the crash. Insurance companies' claims representative have this information, either electronically, on paper, or a combination of both. Since subrogation investigations are rare, it should not be burdensome for the insurance companies to report this information to NHTSA on a periodic basis when they consider subrogation against a vehicle or component manufacturer.
The insurance companies contend that such information would not be useful. State Farm says, "Obtaining such information on subrogation claims directly from manufacturers is more useful because it will capture claims from all sources including insurers." USAA says, "Because of the amount of investigation and requisite period of time required to determine whether a product defect was the cause of the accident, this method is not feasible for serving as a pointer to possible safety defects. Any information that is derived from this investigation and determination would be several months old." However, subrogation information from the insurance companies would be more timely, since information would be available to the insurance company's subrogation unit before the vehicle or equipment manufacturer is notified, and more detailed than similar information from the manufacturers or any other source.
The insurance companies also argued that the information contained in their subrogation files is not "aggregate information" as specified in the TREAD Act. Since claims of defective equipment are rare, a very small number of such claims would be considered "aggregate information" in this context. In those rare cases where companies consider pursuing subrogation for a potential defect, it is most often because they have received several similar claims that indicate the existence of a vehicle or equipment problem that may have caused the loss.
Appendix 2 contains summaries of the electronic data available for each insurance company included in this study.
Unusual Claims Rates
In the Special Order, we asked the insurance companies about how they analyze claims rates that are higher or lower than the norm. We found that most of the companies do perform an annual actuarial review of the loss experience by make and model year. However, they cannot analyze the information for causation of the claim, e.g. potential defect.
For example, USAA stated, "We do not generate reports concerning the percentage of a particular make/model/model year that are involved in accidents in certain time period that could be used to compare to a vehicle class or comparable design, even if we aggregated vehicles by class or comparable design....We do not generate any reports that could be used in determining whether the claims rate for specific vehicles is higher or lower than the claims rate for that vehicle class or the claims experience for comparable designs. As a result, we are not able to determine whether the claims rate for certain types of crashes or fires for specific vehicles was higher or lower than for the vehicle class or comparable designs."
Data Clearing Houses
We also asked the insurance companies about data they send to data clearing houses. All eight insurance companies reported that they send information to at least one data clearing house. There are three main clearing houses: Insurance Services Office, Inc. (ISO); ChoicePoint Incorporated's Comprehensive Loss Underwriting Exchange (CLUE); and Highway Loss Data Institute (HLDI). Most reported that they also send data to a fourth data clearing house, National Insurance Crime Bureau (NICB), but it recently merged with ISO so we are treating it as a part of ISO for this study. Each of the data clearing houses is described below.
CLUE: All of the companies reported that they send data on a monthly basis to ChoicePoint, Inc. for input into its CLUE database. The information provided includes: personal identifiers (name, address, social security number); date of loss; license number and state; sex; policy type; coverage type; payment information; and claim disposition (open, closed, subrogation).
ISO: The insurance companies submit paid loss statistical reports to ISO. USAA stated, "ISO collects statistical data for all stateside locations except Massachusetts and Texas, which have their own, similar statistical plans. All data is submitted on a unit transaction basis. We submit paid losses (indemnity payment), ALAE (allocated loss adjusting expenses), salvage and subrogation recoveries and expenses." The companies also provide ISO information for its Damage Notification Report, Theft Notification Report, and Total Loss Report.
The companies also report information to NICB for theft and suspicious claims. Liberty Mutual stated, "The primary focus of these data submissions is to identify reoccurring claimants and/or frequent provider payments. Generally this data lacks specific accident related information and focuses on who is filing claims, who is being paid and for what services."
HLDI: HLDI collects information on make/model and VIN along with policy information such as policy identification number, type of coverage, collision deductible, and date of loss. HLDI's database contains information on over 100 million vehicles. HLDI has information on claims and dates but does not know who was driving, road conditions, etc. It does not have personal information such as age, gender, names, and addresses.
HLDI data on noncrash fires will be useful to the agency. HLDI codes noncrash fires as fire losses under Comprehensive Coverage. Currently, HLDI prepares an annual report for its members on noncrash fires, but HLDI agreed in its comments to begin to provide this information to the agency quarterly.
Because the data provided to the data clearing houses do not contain causation or subsystem failure information, the data clearing houses do not appear to be useful sources for identifying potential defects, except for information on noncrash fires from HLDI.
V. Legal Issues
There are a number of legal issues that must be considered by the agency. The issue of privacy for policyholders was brought up by the insurance companies' and the insurance associations' responses. The insurance associations also discussed the possibility of liability regarding their exposure to lawsuits based on reporting the possible existence of defective equipment.
The agency recognizes the privacy rights of individuals and has the ability to protect from public disclosure any information that would constitute a clearly unwarranted invasion of personal privacy if released, pursuant to exemption 6 of the Freedom of Information Act, 5 U.S.C. § 552(b)(6). For example, the agency can (and does) protect personal information about policyholders (such as the name, address, telephone number and policy number for an individual) that has been submitted to the agency by insurance companies. Moreover, the agency has an affirmative duty under the Privacy Act of 1974, 5 U.S.C. § 552a, to protect any information that is maintained by the agency in a Privacy Act "system of records," such as NHTSA's database of consumer complaints.
As noted earlier, insurance companies have regularly stated that they will not provide the names of policyholders involved in crashes or other losses to NHTSA without the consent of the policyholder. In the Special Orders, NHTSA asked each of the companies to identify the particular legal provisions that required or supported this practice. We have reviewed the responses and do not feel the information submitted supports this practice. Therefore, in view of our obligation under the Privacy Act to keep this information confidential, NHTSA does not believe that there is any valid privacy interest that would preclude the companies from providing this information.
If, at any time, an insurance company is submitting information that it wishes to keep confidential, the company can submit the information in accordance with NHTSA's confidentiality regulations, 49 CFR Part 512. Specifically, the portions claimed to be subject to protection must be clearly marked, and the submission must be accompanied by a request for protection.
The insurance associations stated, "When creating a new duty to report sensitive information, especially if the duty is imposed on an industry, which is not the manufacturer of any possibly defective product, it is both commonplace and fair to provide the reporting entity with immunity from liability when they comply. This is strongly justified here, where insurers did not produce the products in question, do not make defect determinations, are not the best source of product information and do not now collect and report most of the requested data. The absence of any reporting immunity language would be a fatal flaw, needlessly undercutting the very safety purposes for which the legislation is intended."
The insurance associations also noted, "Among the many examples of reasonable immunity is 49 U.S.C. 33109(d), which in connection with auto theft reporting, provides that: 'Any person performing any activity under section 33110 or 33111 in good faith and with a reasonable belief that such activity was in accordance with such section shall be immune from any civil action respecting such activity which is seeking money damages or equitable relief in any court of the United States or a State.' Similar immunity provisions exist in many State insurance fraud-reporting statutes, as well. The absence of even this minimal degree of protection may discourage thorough reporting in some cases, thereby obstructing the main goal of the legislation, which ought to be accident prevention rather than creating costly litigation over peripheral matters."
The associations did not identify any scenario in which insurance companies might incur any liability associated with the provision of claims or subrogation information to the agency, and NHTSA is not aware of any such potential liability. Moreover, the immunity for auto theft reporting, referenced above, was established by Congress in statute. Congress did not include a comparable provision in TREAD and, therefore, NHTSA was not authorized by the TREAD Act to provide immunity. Moreover, NHTSA has reviewed the information provided to the agency by the insurance companies in their responses, and based on that review, NHTSA has no reason to believe that liability would be created by the actions contemplated in this instance. Accordingly, the agency concludes that immunity would not be required.
After reviewing the information from the insurance companies, the insurance associations, and IIHS, we have come to the following conclusions:
Because large insurance companies all pursue subrogation against vehicle and equipment manufacturers on at least some occasions when they believe that the manufacturer was responsible for the loss, they have some data on possible safety defects. We understand that subrogation investigations and subsequent claims against a motor vehicle manufacturer or parts manufacturer are relatively rare. This leads us to conclude that it would not be unduly burdensome for the insurance companies to provide data that they have gathered in subrogation investigations to NHTSA. It would be useful for the insurance industry to supply this information to NHTSA on a regular basis.
Although most of the companies do perform annual analyses of claim experience for actuarial purposes, the information from those reports is not useful to NHTSA in conducting its defects investigations.
The information collected by insurance data clearing houses does not contain causation information; therefore, it will not help NHTSA conduct defect investigations, with the exception of noncrash fires. HLDI collects noncrash fire data and in its response to the Federal Register Notice, it has agreed to send this information voluntarily to NHTSA each quarter.
State Farm established its CRASH unit and database to address the issue of potential vehicle safety defects. That company is able to review the data for possible trends and report this information to NHTSA. We understand that the CRASH data base is unique to State Farm and that it may not be feasible for all companies to set up similar systems, particularly smaller companies. Costs associated with developing data sets, such as CRASH, would vary significantly from company to company. USAA also collects such data in separate electronic databases. If it is possible to link the reports, USAA could also report information that could identify potential safety defects to NHTSA.
400 Seventh Street, SW
Washington, DC 20590
In re: Insurance Claims Information Study)
Pursuant to 49 U.S.C. 30166(g) and 49 CFR Part 510, the National Highway Traffic Safety Administration ("NHTSA" or "agency") issues this Special Order relating to the agency's Congressionally-mandated study, under Section 3(d) of the Transportation Recall Enhancement, Accountability, and Documentation (TREAD) Act (Pub. L. 106-414, signed November 1, 2000), to determine the feasibility and utility of NHTSA's obtaining aggregate information on a regular basis regarding claims made for private passenger automobile accidents from persons in the business of providing private passenger automobile insurance or of adjusting insurance claims for such automobiles. [Company] is hereby required to provide information as specified below.
A. [Company]'s response to this Special Order must be accompanied by an affidavit(s), signed by one or more authorized corporate representatives, stating that the answers to the inquiries in Part III below are accurate and complete.
B. The information provided in response to this Special Order must include information, known to, or in the possession or control of, [Company]. This includes but is not limited to review of [Company] data bases and computer search practices and capabilities.
C. Each inquiry in this Special Order must be answered separately and fully, in writing. Above each response, the applicable question, or portion of such request, must appear.
D. If [Company] claims that any of the information provided in response to this Special Order constitutes confidential commercial material within the meaning of 5 U.S.C. § 552(b)(4), or is protected from disclosure pursuant to 18 U.S.C. § 1905, [Company] must submit supporting information together with the materials that are the subject of the confidentiality request, in accordance with 49 CFR 512, to the Office of Chief Counsel (NCC-30), National Highway Traffic Safety Administration, Room 5219, 400 Seventh Street, S.W., Washington, D.C. 20590.
E. [Company]'s response to this Special Order, together with a copy of any confidentiality request, must be submitted so that it is received no later than January 5, 2001 to the Office of Plans and Policy, Attention Jane Dion (NPP-01), National Highway Traffic Safety Administration, Room 5208R, 400 Seventh Street, S.W., Washington, D.C. 20590. If you have any questions, Ms. Dion may be reached at (202) 366-6779.
For the purpose of this Special Order, the following definitions apply (regardless of capitalization of the first letter of a word):
A. "[Company]" means [Company], including its divisions and subsidiaries that are not independent entities (e.g., that are not separately incorporated), including all [Company's] employees who design, administer, or implement [Company's] systems or data bases relating to claims information, including preparing analyses thereof or using such information.
B. "Claims information" means information or data concerning claims, including claims by or against one or more policy holders and/or insureds and subrogation claims by [Company], regarding automobile crashes or fires, which information or data includes in whole or in part identification of the involved vehicle by make, model, model year; the date of the crash or fire; any alleged defect in the motor vehicle, motor vehicle equipment or component; information concerning causation of or contributing factors in the crash or fire; and/or information relating to the severity of the crash or fire, including fatalities. This relates to motor vehicle model years 1991-2001.
1. Describe in detail each electronic or computerized data base in which [Company] enters and stores automobile claims information (including a description of all data fields contained therein, with definitions). Describe in detail the feasibility of providing to NHTSA, on a regular basis, a report containing aggregate claims information developed from each such data base, that may serve as a pointer to possible safety defects that could be investigated by the agency.
2. Describe in detail [Company's] system for entering, updating and storing subrogation claims information. Describe in detail the methods for providing to NHTSA, on a regular basis, a report containing aggregate subrogation claims information derived from this system, that may serve as a pointer to possible safety defects that could be investigated by the agency, and comment on the feasibility of the methods. Describe [Company's] subrogation process (from determining whether the company will pursue subrogation, to identification of any alleged defect, to case resolution/closure).
3. Describe in detail how [Company] could determine whether the claims rate for certain types of crashes/fires for specific vehicles, with and without subclassification by alleged defect, by make/model/model year, is higher or lower than the claims rate for that vehicle class or the claims experience of comparable make/model/model year designs. State whether [Company] has conducted this type of analysis (including but not limited to for actuarial purposes) or a variation thereof within the last three years and if so, how often. Also, describe in detail the methods for providing to NHTSA, on a regular basis, a report containing aggregate information developed from any such analyses, and comment on the feasibility of the methods.
4. State whether [Company] has provided or provides vehicle claims data to one or more industry-wide vehicle claims information clearing houses or associations, e.g., the Comprehensive Loss Underwriting Exchange (CLUE). If so, describe in detail each type of information provided over the last three years; state the frequency with which [Company] provides the information; and identify, by organization name and address (and name and telephone number of contact person, if known), each entity to which [Company] provides the information.
5. State [Company's] views and suggestions regarding each method by which NHTSA could regularly consider insurance claims information from insurance companies of relevance to NHTSA's vehicle defects investigation program. State what information would be provided and the source of that information and comment on the feasibility and utility of each method.
6. Identify and describe any limitations, under state law or otherwise, that would affect the ability of [Company] to provide specific information about individual claims to NHTSA related to its defects investigation program.
Allstate keeps its claims data in three different systems: Loss Reporting System (LRS), Auto Dispatch System (ADS), and Claim File and Claim Portfolio databases. Allstate requested confidential treatment for most of the data the company submitted. The sections that were not considered confidential are described here.
Allstate also uses two vendor-provided databases (ADP and CCC) to estimate repair costs (Exhibit 5). Vehicle subsystems are not identified even for repairs.
Allstate provides information to ChoicePoint for CLUE and ISO/AISG (Exhibit 7). The only relevant information provided is basic vehicle information (VIN, car year, make). Allstate also provides information to NICB (Exhibit 8) and HLDI (Exhibit 9). The only relevant information provided is basic vehicle information.
Farmers' considered its entire submission confidential.
GEICO General did not provide a description of all data fields contained in its automobile claims database, so it is impossible to say if the company's claims data could point to possible safety defects.
Liberty Mutual did not provide a description of all variables on its claim database. Liberty Mutual does not have a separate electronic system for subrogation claims, instead, the claims system is used. Thus, it is impossible to tell if the claims data/subrogation data has information on vehicle subsystems that could point to possible safety defect. Liberty Mutual says "none of this information is specific enough to serve as a pointer to possible safety defects."
Nationwide requested confidential treatment for all of its database information.
Progressive's claims data is stored on a mainframe system called PACMan. Progressive contends that PACMan has no information on possible safety defects, but did not include a description of all the variables on the database. Thus, it is impossible to tell if Progressive's claims data has information on vehicle subsystems that could point to possible safety defects. Progressive also uses software to estimate cost of repair. Variable descriptions for this software along with the standard reports generated was included, but collision information about the crash is not included.
Progressive's subrogation files have facts of loss, claimant information, investigation, amount of damages, and activity to date.
State Farm has three claims databases that store information on automobile claims: Claim Service Record, Claims Research Auto Support Hotline (CRASH), and Estimating database. State Farm did not included a description of the data fields contained in any of these files. It is impossible to know if the claim service database and/or the estimating database can point to possible safety defects, without a description of the data fields contained on these files. The little information State Farm provided about the CRASH database suggests that it could be used to identify possible safety defects. The database has make, model, model year, and VIN of the vehicle and suspected product defect. Claims representatives, based on their judgment, submit information on suspected defects to the database. The database can be searched for similar defects by make, model and model year. The frequency of similar defect cases over a short time frame or a relatively high frequency of defect cases for one make/model compared to similar but different make/model might indicate a NHTSA investigation is warranted.
USAA uses two databases to store information on automobile claims: Loss Report Database and Claims Database. Although the Loss Report database contains some information on product defects, it do not have any information on type of crash. It has a product defect flag that allows claims representatives to indicate (yes/no) whether the loss is alleged to involve a product defect. It also has product name and product defect descriptions. Data element that can specifically identify the vehicle are available (make, model, model year, and VIN). The Loss database also has other vehicle data element such as: tire size, vinyl top, power brakes, power steering, stereo, type of radio, air conditioning, and vehicle color. A fire investigation flag is available on the Loss Report database. Crash type (accident description code) is only available on the Claims database which does not have any vehicle information or defect information. It is unclear if the two databases could be merged together to get all the needed information in one place.