National Academies Press: OpenBook

A Look at the Legal Environment for Driverless Vehicles (2016)

Chapter: VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES

« Previous: V. CRIMINAL LAW AND PROCEDURE
Page 47
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 47
Page 48
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 48
Page 49
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 49
Page 50
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 50
Page 51
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 51
Page 52
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 52
Page 53
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 53
Page 54
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 54
Page 55
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 55
Page 56
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 56
Page 57
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 57
Page 58
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 58
Page 59
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 59
Page 60
Suggested Citation:"VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES." National Academies of Sciences, Engineering, and Medicine. 2016. A Look at the Legal Environment for Driverless Vehicles. Washington, DC: The National Academies Press. doi: 10.17226/23453.
×
Page 60

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

47 the U.S. Constitution, most insurance regulation has been delegated by the federal government to the states. Consequently, state regulation varies and sometimes presents challenges to accommodate a system designed for cars with “drivers” (who are often “at fault”) to the new paradigm. B. Regulation of Automobile Insurance The business of insurance moves in interstate commerce.379 Insurance may, therefore, be regulated by the federal government. In 1945, however, Con- gress ceded the regulation of insurance to the states in the McCarran-Ferguson Act.380 With rare excep- tion, Congress has been content to leave regulation of insurance to state regulators. When uniformity is desirable or state regulated markets do not work well, Congress sometimes exercises its power to intervene. Examples include: the Employee Retire- ment Income Security Act of 1975 (ERISA),381 the National Flood Insurance Program (NFIP),382 the Product Liability Risk Retention Act of 1981383 and 1986,384 and the Patient Protection and Affordable Care Act (ACA) (although the latter’s mandate was upheld as a “tax”).385 After the recent failure of a number of financial institutions, the federal regulation of insurance was again mooted. For insurers whose size may cause exis- tential threats should they become insolvent, the Dodd- Frank Wall Street Reform and Consumer Protection Act imposed capitalization and regulatory require- ments beyond those imposed by state regulators.386 Dodd-Frank also established the Federal Insurance Office (FIO) attached to the Treasury Department. The FIO’s role to date has been largely to monitor insurance issues. Mandatory insurance rates for interstate truck- ing also fall within the federal purview. become common, the law likely will evolve to respond to the capabilities of vehicles that incorporate increas- ingly sophisticated autonomous functionalities. In the future, as now, most of the crimes pertinent to the operation of driverless vehicles likely will take the form of “rules of the road” and equipment violations. These crimes lie on the border of admin- istrative (civil) and criminal offenses, carry modest penalties, and require limited or no proof of culpable intent. Some of these offenses will be applied to the users of driverless vehicles in much the same way that they presently pertain to users of conventional vehicles. Others will require adjustments in their vocabulary or substantive provisions to make sense as applied to driverless vehicles. Furthermore, new offenses likely will appear to govern how driverless vehicles should be operated. Nevertheless, while driverless vehicles therefore may occasion an increase in the number of regula- tory offenses that are recognized, if they become common, they likely will lead to a reduction in the number of offenses committed. Other, more serious crimes specific to driverless vehicles probably will take longer to appear. Some of these crimes will evolve gradually while others will fall in place during opportune policy windows. If the history of computer crimes provides any indication, legislators and prosecutors likely will rely on exist- ing offenses to address most misbehavior involving driverless vehicles for a while before sufficient inter- est develops for the enactment of offenses that explicitly address criminal behavior that involves these devices. Even a single real or hypothetical crime may prove sufficient to spark such a move- ment, however. VI. THE EVOLVING INSURANCE MATRIX FOR DRIVERLESS VEHICLES A. Introduction Insurance issues arising from the relationship among driverless vehicles, insurers, manufacturers, and policymakers will present a host of interesting challenges. The reduction in injuries and the reduced responsibility and liability on the part of operators will benefit the public and the operators, but it will also present challenges to the business plans of many insurers. Although the cost of injuries will fall, there will be some need to insure the increased legal responsibility on the part of those in the commercial chain. This is a very different model of insurance. As vehicles become more connected, cyber risks will present a new set of insurance challenges. In addi- tion, while the federal government is empowered to regulate insurance under the commerce clause of 379 United States v. South-Eastern Underwriters Ass’n, 322 U.S. 533, 64 S. Ct. 1162, 88 L. Ed. 1440 (1944) (holding that federal antitrust laws apply to insurance). 380 15 U.S.C. §§ 1011–1015 (2012) (federal antitrust laws apply to insurance to the extent insurance is not regulated by the states). 381 29 U.S.C. §§ 1001–1461 (2012) (especially with respect to self-funded employer health plans). 382 42 U.S.C. § 4104c (2012). 383 15 U.S.C. §§ 3901 et seq. (2012). 384 15 U.S.C. §§ 3901–3906 (2012). 385 National Federation of Independent Business v. Sebelius, 567 U.S. ______, 132 S. Ct. 2566, 183 L. Ed. 2d 450 (2012). 386 See Ian Katz, Republicans Order Records in New Attack on FSOC Transparency, bloomberg bUs. (May 29, 2015) http://www.bloomberg.com/news/articles/2015-05-29/ republicans-demand-records-in-new-attack-on-fsoc- transparency. The Financial Stability Oversight Council (FSOC) is the federal entity charged with deciding whether a financial institution is “systemically important.”

48 mandatory automobile insurance (usually referred to as Financial Responsibility Laws). These laws are designed to insure that a person injured by an auto- mobile has some recourse against a financially respon- sible party. The minimum amount of insurance required by states, however, is not consistent. For example, California requires drivers to carry a policy with minimum personal injury limits of $15,000 per person, $30,000 per accident, and $5,000 for property damage (usually referred to as a 15/30/5 policy). This limit has not changed since 1967. Minimum limits in other states vary from 15/30 (AZ, CA, DE, LA, NV) to 50/100 (AK, ME).390 Congress, however, exercised its power under the interstate commerce clause to set a minimum limit of $750,000 (or more if carrying radio- active or hazardous material) for large trucks through the Federal Motor Carrier Safety Administration (FMCSA).391 In May 2015, Nevada became the first state to permit the testing of large (18-wheeler) self- driving trucks on its public highways.392 C. The Standard Automobile Policy The standard automobile policy contains a bun- dle of coverages. While one may think of a “vehicle” as being covered, the insurance actually insures a constellation of people who have some relationship with the vehicle. For example, those covered for lia- bility may include the named insured, any family member residing with the insured, an unemanci- pated child away at school, and any person driving the car with the permission of an insured. The cover- ages most pertinent to driverless cars are coverages for liability, physical damage, uninsured/underin- sured, collision, and MedPay.393 D. Automobile Safety Although automobile insurance is regulated by each state, minimum levels of safety for automo- bile design are set at the federal level. NHTSA sets standards for the minimum performance of automobiles. NHTSA prescribes standards for, The regulation of automobile insurance, there- fore, falls within the purview of the individual states. Each state has a commissioner or similar official who oversees insurance regulation. Some are elected (as in California), but most are appointed (as in Nevada).387 These officials belong to the National Association of Insurance Commissioners (NAIC). The NAIC meets regularly to consider issues of broad significance to insurance and to develop and propose model laws and regulations to promote con- sistency among states. The NAIC, however, does not have the power to impose its model rules on states. All states endorse the rule that insurance rates may not be excessive, inadequate, or unfairly dis- criminatory.388 One of the main purposes of insur- ance regulation is to assure that insurers remain solvent enough to pay claims. This purpose is served by the requirement that rates not be “inadequate.” Regulators regularly review the financial condition of insurers to assure their solvency. In the context of insurance, “unfairly discriminatory” means that, so far as practical, rates should reflect risk, expenses, and reasonable profit. “Excessive” means that insureds are being overcharged because the rate exceeds these parameters. In addition, all states have some form of insurance guarantee fund (funded by assessments on insurers) to pay claims should an insurer become insolvent. In the context of automobile insurance, generally speaking states attempt to achieve these goals using four different models: No-File, File-and-Use, Use- and-File, or Prior-Approval. No-File, File-and-Use, and Use-and-File states rely primarily on competi- tion to set rates. In these states insurers may either file, then use their rates, or use their rates provided they, within a specified amount of time, file them. Prior-Approval states, on the other hand, require that rates be approved before they can be used. Cali- fornia and New York are examples of Prior-Approval states. Even in states other than Prior-Approval states, insurance regulators have the power to dis- approve rates under appropriate circumstances.389 In addition to regulating insurance in different ways, states also differ with respect to other details governing automobile insurance. For example, every state except New Hampshire requires some form of 387 A list of states with elected or appointed commission- ers is available at State Commissioners-2014, nAT’l Ass’n of ins. Comm’rs, http://www.naic.org/documents/members_ state_commissioners_elected_appointed.pdf. 388 See, e.g., CAl. ins. Code § 1861.05(a) (West 2013) (“No rate shall be approved or remain in effect which is excessive, inadequate, unfairly discriminatory or other- wise in violation of this chapter.”). 389 See generally The insTiTUTes, insUrAnCe regUlATion § 5.4 (Karen Porter, ed. (2010)). 390 State financial responsibility limits (as of February 2015) are listed at Compulsory Auto/Uninsured Motorists, ins. info. insT., http://www.iii.org/issue-update/compulsory- auto-uninsured-motorists. PIP (no-fault) jurisdictions include AR, DE, DC, FL, HI, KS, KY, MD, MA, MI, MN, NJ, NY, ND, OR. PA, and UT. 391 Information on this topic is available at DMV.Org, http://www.dmv.org/insurance/federal-motor-carrier- insurance.php. 392 Peter Valdes-Dapena, Self-driving Semi Hits the Road, CNN money (May 6, 2015), http://money.cnn. com/2015/05/06/autos/self-driving-truck/index. html?iid=ob_homepage_tech_pool&iid=obnetwork. 393 See generally Automobile Insurance—The Personal Auto Policy (PAP), ThismATTer.Com, http://thismatter.com/ money/insurance/types/auto-insurance.htm.

49 of legislation,398 so the effect of legislation is often to narrow the circumstances under which testing can take place. The Uniform Law Commission is studying whether to adopt a proposed model law for the states, including insurance requirements for drivers. California Department of Motor Vehicles testing regulations went into effect on September 16, 2014. As mandated by California Vehicle Code Section 38750,399 regulations for the operational stage were due to be completed by January 1, 2015. At the date of this writing, however, the Department has not yet published final, or even proposed, regulations. E. Insuring Driverless Vehicles 1. How Insurers Create Automobile Insurance Rates Most insurers create automobile rates in a two- step process. First they create a “rate plan” to calcu- late a “base rate” (a “base rate” is also called an “indication” in the insurance industry). The insurer looks at its book of the relevant business and asks the question: How much must the insurer collect in premium from each insured to service this book of business over the next rating period, including over- head and profit? For example, if there were 100 insureds all purchasing the same coverage, and it would cost $100 to service this book of business over the next policy period, then the base rate would be $1. The insurer must collect, on average, $1 from each policyholder.400 Insurers, however, do not charge all policyholders the same amount. This is because policyholders, even if purchasing the same coverage, do not pres- ent the same risk of loss. If the insurer charged the same amount, than those who present lower risk (perhaps $.80) would pay too much. Not only might this be “unfairly discriminatory,” but they would, unless mandated to carry this coverage, likely drop out of the pool because they are not receiving appro- priate value for their premium. Even if coverage is mandated, as with automobile insurance, insureds would over time migrate to a different insurer who differentiated rates to more closely approximate the insured’s risk. This would leave the insurer with a group of policyholders who present a greater risk than the $1 base rate. If they paid only $1 for their policies, but on average cost $1.20 in losses, the insurer would eventually be out of business. To avoid this outcome (known in the industry as “adverse inter alia, windshield wipers, trunk releases, seat- belts, air bags, anti-brake lock (ABS) braking, crashworthiness, electronic stability control (ESC), gasoline mileage, and many other aspects of car construction and design. As with many federal agencies, NHTSA’s regulation of automobile design is governed by rigorous cost/benefit analysis.394 NHTSA’s requirements preempt any less exacting state standards.395 NHTSA has not adopted any standards for the design or safety of AVs. Indeed, NHTSA has stated that “in light of the rapid evolution and wide varia- tions in self-driving technologies, we do not believe that detailed regulation of these technologies is fea- sible at this time at the federal or state level.”396 In this regulatory vacuum, states are moving for- ward with their own regulations for testing and driv- ing AVs. Thus far, four states (Nevada, California, Michigan, and Florida) and the District of Columbia have adopted laws or regulations allowing testing of AVs. Virginia has also announced that it will open cer- tain highways for testing. Bills are pending in 11 states, and bills have failed in 7 states.397 It is likely that testing is legal on public highways in the absence 394 See generally Robert W. Hahn and Cass R. Sunstein, A New Executive Order for Improving Federal Regulation? Deeper and Wider Cost-Benefit Analysis, 150 U. PA. l. rev. 1489 (2002). 395 49 U.S.C. § 30103(b)(1) (2012) (state standards valid “only if the standard is identical to the standard prescribed under this chapter”). 396 NHTSA PreliminAry sTATemenT, supra note 184, at 12–13. For a criticism of NHTSA with respect to overseeing safety critical computer programing, see David Benjamin, Toyota Underestimated “Deadly” Risks, EE Live! keynoter says, EDN neTWork (April 02, 2014), http://www.edn.com/ electronics-blogs/now-hear-this/4429744/Toyota-Under estimated—Deadly—Risks--EE-Live—keynoter-says. The Uniform Law Commission is presently studying the advis- ability of proposing uniform state laws for driverless vehi- cles. See State Regulation of Driverless Cars, Unif. l. Comm’n (2015), http://www.uniformlawcommission.com/ Committee.aspx?title=State%20Regulation%20of%20 Driverless%20Cars. 397 The American Association of Motor Vehicle Administrators (AAMVA) maintains a site tracking state legislation, http://www.aamva.org/WorkArea/linkit.aspx? LinkIdentifier=id&ItemID=5826&libID=5802 (herein- after “AAMVA Autonomous Vehicle Legislation Chart”). Other related information is collected by the AAMVA at their Autonomous Vehicle Information Library, http:// www.aamva.org/Autonomous-Vehicle-Information- Library/ (last visited Sept. 20, 2015). Meanwhile, legis- lative developments for automated cars are tracked at Gabriel Weiner & Bryant Walker Smith, Automated Driving: Legislative and Regulatory Action, CTr. for inTerneT & soC’y (Sept. 2, 2015), http://cyberlaw. stanford.edu/wiki/index.php/Automated_Driving:_ Legislative_and_Regulatory_Action#State_Bills. 398 Smith, supra note 180. 399 CAl. veh. Code § 38750 (West 2014). 400 For a more detailed look at how actuaries actually model automobile insurance rates, see CAsUAlTy ACTUAriAl soCieTy, bAsiC rATemAking (2010), http://www.casact.org/ library/studynotes/werner_modlin_ratemaking.pdf.

50 one jurisdiction, territory (where the automobile is garaged), credit score, or gender may be permitted, while not permitted in another.403 Calculating appropriate relativities is a very sophisticated undertaking. It involves analyzing large amounts of data collected by insurers from their own loss experiences and from other sources. This information is then utilized by actuaries to predict the frequency and severity of future losses over the policy period for each of the characteristics evidenced by the insured. This includes predicting future trends in losses. It is possible, for example, that medical or repair costs are predicted to rise over the policy period. It is also possible that gasoline prices may spike or plummet, causing less or more driving and fewer or more accidents over the policy period. Trend is one of the most difficult factors to predict and, where insurance rates are subject to prior approval (as in California), disputes over the future trend for losses occupy much of the regulator’s attention. 2. Emerging Issues for Insuring Driverless Vehicles “The advent of autonomous cars could revolution- ize the world of motor insurance.”404 NHTSA estimates that approximately 93 percent of accidents are caused, at least in part, by human error.405 NHTSA estimates that the economic cost alone amounts to over $900 per person per year in the United States. There are approximately $871 billion in economic and societal losses per year.406 The U.S. Centers for Disease Control and Preven- tion (CDC) reports that in 2012 over 2.5 million Americans were sent to the emergency room because of accidents (approximately 7,000 per day). Nearly 200,000 were hospitalized.407 These figures suggest that reducing human error can substantially reduce injuries, deaths, and other costs caused by automo- bile accidents. It is likely that self-driving cars will not be totally autonomous for a number of years. They will operate selection”), the insurer creates a “class plan.”401 Fail- ure to have a class plan that reasonably discrimi- nates among risks can, then, result in a slow “death spiral” for the insurer. The class plan applies “rating factors” to adjust the base rate depending on the risk presented by the policyholder. With respect to automobile insurance, many of the rating factors are familiar: age, gender, driving record, miles driven, location, accident his- tory, vehicle class, and credit score (where permit- ted). California’s regulations, for example, permit the use of 19 rating factors for automobile insur- ance. Many of these rating factors include a welter of subdivisions. Applying some rating factors to the above exam- ple, A’s rate calculation might look something like this. Ms. A is a female, and females as a group have fewer accidents than males. A neutral rating factor has a relativity of 1, but as a female, she may have a relativity of .90. Thus, her rate would be $.90. If the average person drives 12,000 miles per year, then 12,000 miles per year would have a relativity of 1. Ms. A, however, drives 20,000 miles per year. This increases the likelihood of an accident and may yield a relativity of 1.10. This would likely offset the ben- efit Ms. A received because of her gender and move her back to a rate of $1. Sadly, Ms. A also has a poor driving record, having been convicted of three mov- ing violations in the past 3 years. This driving record may yield a relativity of 1.20. Now Ms. A’s premium would move from $1 to $1.20. Ms. A may, however, enjoy some downward adjustments if the automo- bile is garaged in an area with fewer claims or (if her state permits its use) she has a good credit score. The combination of all of the rating factors and their relativities yields the ultimate rate Ms. A must pay for her policy. The actual process is not as simple as this may appear. It often includes multiplicative algorithms, sequential analysis, and “pumping” and “tempering.”402 Formulating a class plan is largely a zero-sum game. If a relativity lowers a rate for some policy- holders, that lower rate must be balanced by a higher rate for some others in the class plan. Thus, a lower rate for a low mileage driver would likely be balanced by higher rates for higher mileage drivers. When done correctly, the net rate over the book of policies will equal the base rate ($1 in the above example). Politics and social policy play a role in deciding which rate classifications are acceptable. In 401 See, e.g., CAl. Code regs. tit. 10, §§ 2632.7, 2632.8 (2015) (California auto insurers must file a “class plan” with the Department of Insurance). 402 See Spanish Speaking Citizens’ Foundation, Inc. v. Low, 85 Cal. App. 4th 1179, 1186 (2000); CAsUAlTy ACTUAriAl soCieTy, supra note 400. 403 See generally insUrAnCe regUlATion, supra note 389, § 8.10; Laura Adams, How Age, Gender and Marital Sta- tus Affect Your Auto Insurance (Mar. 31, 2015), http://www. huffingtonpost.com/laura-adams/how-age-gender-and- marital-status-affect-your-car-insurance_b_6973360.html. 404 gilliAn yeomAns, AUTonomoUs vehiCles: hAnding over ConTrol: oPPorTUniTies And risks for insUrAnCe 18, lloyds, (2014) (hereinafter “lloyd’s rePorT”). 405 NATionAl highWAy TrAffiC sAfeTy AdminisTrATion, nATionAl moTor vehiCle CrAsh CAUsATion sUrvey, DOT HS 811 059 (July 2008). 406 Larry Blincoe, National Highway Traffic Safety Administration, The Economic and Societal Impact of Motor Vehicle Crashes DOT HS 812.013, 2 (2015). 407 Centers for Disease Control and Prevention, Motor Crash Injuries, viTAl signs (Oct. 2014), http://www.cdc.gov/ vitalsigns/crash-injuries/index.html.

51 California, to make every attempt to pass these sav- ings on to policyholders. Passing these savings to the policyholder has two principal benefits. It helps consumers with lower rates, and lower rates encourage drivers to purchase these safer vehicles. There are differing estimates as to the cost of equipping an automobile to drive itself. Some are concerned that the higher price will deter many from purchasing the automobile, or the higher price will create a class difference between those who can afford a self-driving vehicle and those who cannot. If, however, insurance savings over the life of the vehicle offset the additional cost of equipping the vehicle, cost should neither deter acceptance nor create two classes of drivers. As with other technol- ogy, as driverless cars mature, one would expect dra- matic drops in price. Creating an awareness in consumers of the inter- play between the added vehicle cost and the lower insurance costs presents a marketing issue. One analogous marketing model is that used to inform consumers of the energy efficiency of some appli- ances, such as water heaters and refrigerators. In that context, consumers are now accustomed to pric- ing an appliance with its future costs in mind. The same could be applied to vehicles. Since rates must be neither “excessive, inade- quate, nor unfairly discriminatory,”410 regulators and insurers will face some new challenges as these vehi- cles are introduced. Automobile insurance rates are based on extensive data bases. Apart from testing data, which may or may not be available to insurers or regulators (OEMs and others may treat this infor- mation as proprietary or trade secrets), there will be a considerable amount of guesswork going into ini- tial rate making. At present, insurers have little or no data on which to base a rate, even assuming that it was clear where liability would lie.411 In addition, the added technology may raise repair costs. Perhaps they will be offset by lower fre- quency, perhaps not. Insurers would be inclined to play it safe by estimating higher rates, while regula- tors may be inclined to protect consumers and encourage adoption of this safer technology by esti- mating lower rates. Similar challenges have been confronted when new safety measures were intro- duced, such as rear window mounted taillights and electronic stability control. Some insurers are in a shared driving mode (NHTSA Level 3) in which the driver may trust the driving to the car, but must be available on adequate notice to take over the driv- ing. In addition, there will be circumstances, such as the present inability to drive in snow, where the vehi- cle must be driven by the operator. Thus, there are parts of the country where the benefits of self-driving vehicles will not be as great as in others. It is unlikely, therefore, that the savings from self-driving vehicles will match, or come close to matching, NHTSA’s esti- mates of the current costs of accidents. For that portion of the shared driving experience in which the operator manually drives the automo- bile, insurance and the calculation of insurance rates should be fairly straightforward. There will be some adjustments that will account for the fact that the automobile, when in manual mode, will more frequently be driven in more dangerous circum- stances. Urban driving, driving in construction zones, or driving in snow or inclement weather are examples. The Casualty Actuarial Society is pres- ently studying how much potential savings may be expected as self-driving cars are introduced.408 To the extent that these losses are presently insured (some, such as air pollution, the cost of con- gestion, and lost productivity, are not), and the cars are moving in self-driving mode, the cars should enjoy lower insurance rates. One study suggested an average saving of $475/driver/year.409 These savings will either flow to the insurer’s bottom line, or, either because of competition or rate regulation, will be savings to the policyholder. One would expect regu- lators in Prior-Approval states, such as in 408 For their first report, see CAsUAlTy ACTUAriAl soCieTy AUTomATed vehiCles TAsk forCe, resTATing The nATionAl highWAy TrAnsPorTATion sAfeTy AdminisTrATion’s nATionAl moTor vehiCle CrAsh CAUsATion sUrvey for AUTomATed vehiCles (2014), http://www.casact.org/pubs/forum/14fforum/ CAS%20AVTF_Restated_NMVCCS.pdf. See also If Cars Drive Themselves, How Will Actuaries Make Sense of Data to Price the Risks?, CArrier mgmT. (May 20, 2014), http://www. carriermanagement.com/news/2014/05/20/123377.htm. 409 Alain L. Kornhauser, Smart Driving Cars: History and Evolution of Automated Vehicles 49 (Sept. 20, 2013), http://orfe.princeton.edu/~alaink/SmartDrivingCars/ Presentations/Florida_Seminar_Nov2013_URLsV2.pdf. Other predictions are similar. One recent report advised: The personal lines sector could fall to 40% of cur- rent size. …The personal lines automobiles sector will likely bear the brunt of the transformation, as it will hold a smaller share of a smaller market. Currently, the personal auto sector accounts for almost $125 bil- lion in loss costs. By 2040, we believe this sector could cover less than $50 billion in loss costs. KPMG, AUTomobile insUrAnCe in The erA of AUTonomoUs vehiCles, sUrvey resUlTs 9 (2015), https:// www.kpmg.com/US/en/IssuesAndInsights/Articles Publications/Documents/automobile-insurance-in-the- era-of-autonomous-vehicles-survey-results-june-2015.pdf. 410 See, e.g., CAl. ins. Code § 1861.05(a) (West 2013) (“No rate shall be approved or remain in effect which is excessive, inadequate, unfairly discriminatory or otherwise in violation of this chapter.”) 411 Alexander C. Kaufman, Tesla’s Self-Driving Feature Leaves Insurers Idling as States Scramble, hUffingTon PosT (Mar. 28, 2015), http://www.huffingtonpost.com/2015/03/28/ tesla-self-driving-cars_n_6961922.html.

52 policies is to insure against liability of the insured or (in the case of uninsured/underinsured motorist coverage [UM/UIM]) the liability of another. Put another way, most of what insurers insure against today is human error. A typical insuring clause in the liability portion of a policy insures the “insured” for any amounts up to the policy limits for which the insured may become “legally liable.” UM/UIM coverage compensates those injured by UM/UIM drivers up to the policy limits of the UM/UIM cov- erage only if the insured is “legally entitled to recover” from the uninsured or underinsured driver. Thus, legal responsibility is the lynchpin of both of these important coverages. To the extent human error is removed, to a similar extent the insurers’ business model changes. Driverless vehicles will challenge these tradi- tional insurance models for several reasons. As vehi- cles move towards driverless capabilities, there will be a transitional period of shared driving experi- ence—some driving in manual mode and some driv- ing in self-driving mode. To the extent vehicles are driven in manual mode, one would expect liability and insurance to look much as it does at present. In order to accurately rate this portion of the driving, however, insurers will need to know how many miles the car is driven in manual mode and what the nature of this driving is (snow, urban, construction, etc.). Risks presented by these miles may differ sub- stantially from the risks presented by average miles in general. In addition, gathering this information may present some privacy issues (discussed else- where in this report). If the vehicle causes an accident while operat- ing in driverless mode, under present products liability law the responsibility would be allocated to those in the commercial chain (dealer, OEM, possibly the programming entity if different, etc.) rather than the driver.417 If the operator were sued, the automobile insurer would have a duty to defend under the policy. However, once it is shown that the operator was not “legally liable” for the accident, the traditional policy should not pay for the damages. Autonomous cars could potentially lead to a substantial reduction in motor insurance claims if accidents signifi- cantly reduce in frequency. Lower claims would be expected to result in lower premiums, and tighter profit margins. Some might argue that if cars really do become crashless, there may not even be a need for motor insurance….Dam- age or theft can still occur when a car is parked in a drive- way, and for the present at least, cars with semi-autono- mous capabilities are more expensive than their traditional working with OEMs to insure that they understand the new technology well enough to propose realistic insurance products and rates.412 In time, there will be a growing database, however it will likely be much less static than databases used to rate current vehicles. OEMs will continually improve and update their algorithms, and these will be downloaded to all of the vehicles in the fleet. This could happen daily, or perhaps on a virtually contin- uous basis. As a consequence, yesterday’s self-driving car will not be the same as today’s or tomorrow’s. In March 2015, Tesla downloaded over the web a revised algorithm to one of its models that increased its acceleration.413 In the same month Tesla announced that it would, in 3 months, download a hands-free autopilot feature that would allow freeway driving without the driver’s 100 percent attention.414 Given the rapid advances of technology in other areas, one can expect the safety of these vehicles to rapidly improve. An insurance system that can respond to these improvements with similar alacrity will deliver numerous benefits. Many regulatory sys- tems in prior approval states, however, are not pres- ently equipped to quickly adjust rates. In California, for example, an insurer may not change a rate either up or down without filing a complete rate applica- tion.415 Many other states allow some degree of flexi- bility without prior approval.416 This regulatory chal- lenge is only just beginning to be confronted. 3. Liability-Related Insurance Issues Liability issues drive automobile insurance issues because a major function of automobile 412 See Cathy Schwamberger, Counsel for State Farm, Testimony before the California Department of Insur- ance (Sept. 15, 2014), https://www.insurance.ca.gov/0400- news/multimedia/0030VideoHearings/upload/AVHearing- SchwambergerWrittenComments.pdf. 413 Jon Fingas, Tesla Model S is getting even quicker through a software update, engAdgeT (Jan. 29, 2015), http:// www.engadget.com/2015/01/29/tesla-model-s-acceleration- update/. 414 Although TESLA has since made it clear that driving is still the operator’s responsibility. Evan Ackerman, Tesla Model S: Summer Software Update Will Enable Autonomous Driving, IEEE sPeCTrUm (Mar. 23, 2015), http://spectrum. ieee.org/cars-that-think/transportation/self-driving/tesla- model-s-to-combine-safety-sensors-to-go-autonomous. 415 CAl. ins. Code § 1861.05(b) (West 2013) (“Every insurer which desires to change any rate shall file a complete rate application with the commissioner.”). 416 States allowing flex rating permit insurers to use new rates without prior approval if they do not exceed a cer- tain percentage of the previously filed rate. See insUrAnCe regUlATion, supra note 389, § 8.15. For a list of states and their regulatory frameworks, see Regulation Modernization, ins. info. insT. (Apr. 2015), http://www.iii.org/issue-update/ regulation-modernization. 417 See Fluor Co. v. Jeppesen & Co., 170 Cal. App. 3d 468, 216 Cal. Rptr. 68 (1985) (air chart treated as “product” when error in map contributed to crash).

53 or available. Although “New GM” established a fund and process to compensate those injured by its defec- tively designed ignition switches, its official position is that it is not legally responsible for “Old GM’s” defective products. Although still subject to appellate review, a federal bankruptcy judge has ruled that those liabilities were discharged in bankruptcy.422 Every state has a guarantee fund to compensate, up to a limit, those who would be entitled to compen- sation should an insurance company become insol- vent. These funds are funded by assessments on insurance companies admitted to sell insurance within the state. Unlike insurance companies, at present there is no guarantee fund to insure that those injured by insolvent OEMs are compensated.423 In addition, OEMs may choose to insure (if at all) with non-admitted, surplus lines insurers, risk reten- tion groups, captives, or through other methods. Typi- cally, guarantee funds do not back these liabilities.424 Given the shift in responsibility to the commer- cial marketers of driverless vehicles, one would also expect that the insurance burden (to the extent they choose to insure) would also shift to commercial poli- cies covering dealers, OEMs, and others. Although physically injured parties may have claims against a number of parties in the commercial enterprise, the commercial parties may bargain to distribute any losses (which, with respect to them, are purely economic), among themselves as suits their com- mercial interests.425 4. Imposing Some or All of the Initial Liability on the Operator To give an injured person a local and marginally solvent responsible party, it may be possible to make the vehicle’s performance a “non-delegable duty” for which the operator is responsible regardless of the lack of fault. There is some precedent for this counterparts. It is possible that this risk could become part of a household contents policy coverage.418 Others believe that any significant impact on the insurance industry is far in the future.419 Similarly, if a person were hit by a UM/UIM vehicle, there would be coverage if the uninsured or underin- sured vehicle were driven in manual mode. If in driv- erless mode, however, then the injured party would not be “legally entitled to recover” from the operator of the uninsured or underinsured vehicle.420 Therefore, under present policies, the UM/UIM coverage would not apply. Rather than a UM/UIM claim against the insured’s own insurance company, the insured’s claim would be a products liability claim against the OEM and/or those in the commercial chain. A product’s liability claim has the potential of being much more complex. As accidents increasingly become the responsibility of the commercial sup- plier, legislators and other policymakers may not be content to make every fender-bender a products lia- bility case. Defense and cost containment expenses for different classes of claims differ dramatically. In 2013, insurers’ cost containment expenses for pri- vate passenger auto liability was 6.8 percent of incurred losses, while for products liability it was 75.1 percent.421 One would expect similar expenses on the part of the injured party. It may also present special challenges to the injured party if the manufacturer is no longer solvent 418 lloyd’s rePorT, supra note 404, at 18. Beginning in March 2015, some insurance companies began citing self- driving cars as threats to their business model in SEC fil- ings. See Benjamin Preston, Insurers Worry Self-Driving Cars Could Put a Dent in their Business, The gUArdiAn (Mar. 8, 2015), http://www.theguardian.com/technology/2015/mar/ 08/insurers-worry-self-driving-cars-could-put-a-dent-in- their-business. See also Kristen V. Brown, Self-Driving Cars: Bumpy Ride for Insurance Industry, S.F. Chron. (Apr. 13, 2015) (citing four insurance companies that noted the potential threat of self-driving cars to their businesses), http://www.sfgate.com/business/article/Self-driving-cars- bumpy-ride-for-insurance-6195316.php. 419 Mark Hollmer, Progressive at PCI: Telematics will Become Ubiquitous Underwriting Tool, CArrier mgmT. (Oct. 28, 2014), http://www.carriermanagement.com/news/ 2014/10/28/131022.htm (quoting Tom Hollyer of Progressive Insurance as stating, “I don’t think there’s any kind of [auto- mobile insurance industry] cliff in the foreseeable future.”). 420 A recent Farmers Insurance TV ad promoting UM/ UIM coverage shows a robot driving a car into the rear of the insured’s parked car. The robot then runs away. Query as to whether the insured would have a claim against a robot driv- en car under UM/UIM insurance. Against whom would the insured be “legally entitled to recover?” The manufacturer of the robot? The car owner who, apparently, was a passenger in the car? The ad is available at http://www.ispot.tv/ad/7fjV/ farmers-insurance-robo-driver. 421 Products Liability, ins. info. insT. (2015), http://www. iii.org/fact-statistic/products-liability. 422 Hilary Stout and Dannielle Ivory, Ruling Shields G.M. From Ignition Suits, N.Y. Times (Apr. 15, 2015), http://www. nytimes.com/2015/04/16/business/general-motors-wins- ruling-shielding-it-from-most-claims-over-ignition-flaw. html?_r=0. For a colorful commentary on this issue, see Further on the Effect of the GM Bankruptcy on New GM’s Exposure to Ignition Defect Litigation (June 5, 2014), http:// thenecessaryandproperblog.blogspot.com/2014/06/further- on-effect-of-gm-bankruptcy-on.html. See also Linda Sandler and Patrick G. Lee, Bankruptcy Order Could Shield GM from Ignition Switch Claims, ins. J. (Mar. 20, 2014), http:// www.insurancejournal.com/news/national/2014/03/ 20/323778.htm. 423 See insUrAnCe regUlATion, supra note 389, § 12.12. 424 See id. § 12.13. 425 Philippine Airlines, Inc. v. McDonnell Douglas Corpo- ration, 189 Cal. App. 3d 236, 234, 234 Call. Rptr. 423 (1987) (Airline and manufacturer allowed to allocate among them- selves losses, including personal injuries, caused by crash).

54 responsible driver carries a minimal policy (e.g., 15/30/5 policy in California, another policy at a different state’s minimum, or no insurance at all), a seriously injured party is likely to settle with the driver for far less than the party’s actual inju- ries. Sadly, adequacy of compensation for serious injuries depends largely on the financial suffi- ciency of the injurer. If the injured party carries uninsured/underinsured motorist coverage (UM/ UIM), there may be an additional source for com- pensation. UM/UIM coverage, however, is often modest in amount and is subject to numerous lim- itations. Many losses in serious cases, therefore, fall on the individual or on the public through such programs as Medicaid and Medicare. In addition, there are exclusions in standard automobile policies which remove some otherwise insurable injuries from the insurance pool. For example, if a careless driver suffers injuries and also injures other family members in the vehicle, none of these injuries is covered. They fall within the “fam- ily” or “insured” (all relatives living in the home are “insureds”) exclusion. This dynamic changes dramatically if other sources of coverage or assets become available. As responsibility shifts from drivers to commercial suppliers, more injuries will be compensated at rates closer to their true value because commercial suppliers will have adequate assets or insurance. For example, if the driver and other family mem- bers were injured due to a defect in the automo- bile’s ignition switch, all would have claims against the OEM. These injury costs will be passed to vehi- cle owners in the cost of the cars. Passing the true cost of a product, including injury costs, to those who use the product is one of the aims of “strict liability” under products liability tort law.430 It is also fairer to innocent injured parties if they must bear fewer of their injuries. It does mean, however, that the reduction in the frequency of accidents with self-driving cars may not net a linear savings to car owners. This is because injuries that would be under- or uncompensated when responsibility stops with the driver now will be compensated at closer to their actual value. Claims costs may also rise because self-driving cars will likely be more expensive to repair. At the same time, however, costs attributed to assigning responsibility should decrease. Both California and Nevada require the event data recorder (“black box,” or EDR) in self-driving cars to preserve all of the approach with respect to brake failure.426 In the United Kingdom, liability falls on the driver/user even if not at fault. The driver’s or insurer’s remedy lies in subrogation against the manufacturer.427 If, however, unlimited liability remains on the operator for accidents caused by a failure of the product, many may be deterred from purchasing these safer vehicles. One possible compromise may be to fasten the initial responsibility on the operator up to the minimum financial responsibility limits set by the state. At present, drivers are required to carry insurance up to these modest limits. Fault on the part of the driver would be required for any claim beyond the minimum. This would give an injured party a convenient source for compensation for the majority of claims that fall within these lim- its without adding an additional burden on present drivers. Indeed, given the lower frequency of acci- dents, the premium may be substantially lower than at present. If an insurer paid a claim for which its insured was not at fault, the insurer would have the right to pass the loss to the commercial suppliers through subrogation. Unlike the insured, it may be possible for insurers to consolidate similar claims against a manufacturer, thus making the processing of the claims much more efficient. Insurers and manufac- turers may even find it in their interest to agree to arbitrate disputed claims. There is an existing model for mutual arbitration agreements. When an insurer pays a collision claim for which another insured may be responsible, insurers have agreed to arbitrate the claims with the responsible party’s insurer.428 Like- wise, uninsured/underinsured motorist claims are subject to arbitration.429 These claims are processed very quickly and at minimal expense. A similar model may emerge for dealing with subrogation claims. Insurers’ rates should be net of any subroga- tion recovery (less expenses), so rates should still be lower than at present. 5. Product Liability and Higher Per Claim Costs Under the present system, insurance payouts do not accurately reflect insurable losses. It is not uncommon for more serious injuries to go under- compensated because insurance coverage is inad- equate to compensate for serious injuries. Setting aside possible coverage for health care costs from health insurance or public sources, if a 426 Maloney v. Rath, 69 Cal. 2d 442, 445 P. 2d 513 (1968) (faultless driver responsible for injury caused due to brakes negligently repaired by brake shop). 427 lloyd’s rePorT, supra note 404, at 18–19. 428 See insUrAnCe regUlATion, supra note 389, § 9.18. 429 See id. § 12.12. 430 Greenman v. Yuba Power Products, Inc., 59 Cal. 2d 57, 377 P. 2d 899 (1963); Phipps v. General Motors Corp., 278 Md. 337, 363 A. 2d 955 (Ct. App. 1976) (discussing the ratio- nales for strict liability).

55 should be a substantial incentive for many to adopt self-driving cars. This will especially be so if the insurance savings substantially offset the added cost of the self-driving components. In a recent study by the Boston Consulting Group, their survey showed that, despite the added expense for the technology, 44 percent to 55 percent of those polled would buy either a partially or fully autono- mous car. For partially autonomous cars, lower insurance costs were the top reason for making the purchase (safety was second). For fully autonomous cars, lower insurance was the second ranking rea- son, with safety in the first spot.435 Thus, properly rating these cars and passing the insurance savings on to the consumer will be critical to their rapid acceptance. Since there are also benefits to the public in gen- eral (e.g., less congestion, fewer accidents, better fuel economy, etc.), it may be reasonable to view their adoption as a public good. As such, there will be sound reasons for public policymakers to offer incentives to adopt self-driving automobiles. Tax credits, as applied to electric vehicles, is one approach. Offering money to retire older cars (“Cash for Clunk- ers”) is already an incentive in place in some states. Air Quality Control Districts in California offer $1,000 to retire older cars simply because they pol- lute more than newer ones. Driving in the carpool lane might also be offered as an incentive, along with, perhaps, a higher legal speed limit. Where trucks are currently limited to 55 mph and cars are limited to 65 mph, perhaps driverless cars, because of their enhanced safety features and better reac- tions, could be permitted an official 75 mph limit. There may be other incentives to more quickly intro- duce self-driving cars. 7. New Models of Insurance As mentioned above, under the current legal regime, the role of traditional automobile liability insurance will decrease as the number and severity of accidents decreases and the legal responsibility for accidents shifts away from drivers or operators. Commercial insurance for those in the commercial chain will increase in importance. One would not expect this to present any special challenges, as commercial insurance has been available for thou- sands of products in the market place. It is also possible that, at Level 4, many OEMs will not sell cars to individuals, but rather will sell them to operators of fleets of cars. Users will sub- scribe to use the vehicles as needed. One would expect the insurance burden, then, to fall on the com- mercial insurers of the OEM and/or the fleet owner. data for the last 30 seconds prior to an accident.431 With some retraining, adjusters and lawyers should be able to assign responsibility among the driver, the vehicle, or others with relative ease. Whether the higher claims value and higher repair costs will be offset by the lower claims frequency and lower adjustment costs remains to be seen. 6. Adoption of Self-Driving and Driverless Cars For reasons stated above, the adoption of self-driv- ing cars may present serious challenges to companies writing traditional personal automobile policies. The threat to their premium base and business model has been noted. At the same time, the public has much to gain by the adoption of self-driving cars. These chal- lenges and benefits depend to a large extent on the rate at which self-driving cars are adopted. The average life of a car in the current fleet is between 11 and 12 years (up from 9.8 years in 2002).432 Thus, if all new cars were required to be self-driving cars, one would expect one-half of the fleet to be self-driving in approximately 11 years. Electronic Stability Control (ESC) has been required on all light vehicles since 2011, yet the Insurance Institute for Highway Safety (IIHS) and Highway Loss Data Institute (HLDI) estimate that there will not be 95 percent penetration of ESC until 2030.433 Since self-driving cars are not mandated and will not be available for several years, one might expect the penetration of self-driving cars to take even lon- ger than ESC.434 There are some good reasons to believe that sig- nificant penetration may arrive sooner than these estimates. Safety features short of Level 3, such as ESC, do not drive the car to the extent that the driver may put driving time to other productive uses, whether that be texting, reading, or consulting with clients. Adding productive value to driving time 431 CAl. veh. Code § 38750(c)(1)(G) (West 2014), nev. Admin. Code § 482A.110(2)(b) (2014). See also Majorie A. Shields, Annotation, Admissibility of Evidence Taken from Vehicular Event Data Recorders (EDR), Sensing Diagnostic Modules, or “Black Boxes,” 40 A.L.R. 6th 595 (2008). Privacy issues are discussed in Glancy, supra note 210, at 1175–76, 1202–03. 432 Average Age of Vehicles on the Road Remains Steady at 11.4 years, According to IHS Automotive, IHS (June 9, 2014), http://press.ihs.com/press-release/automotive/average-age- vehicles-road-remains-steady-114-years-according-ihs- automotive. 433 See Adrian Lund, Advanced Safety Technologies and Other Guideposts on the Road to Vision Zero (June 5, 2014), http://www.iihs.org/iihs/topics/presentations. (Scroll down to June 5, 2014, and link to article.) 434 Hollmer, supra note 419 (It will be many years “before the new technology gets to even a 50 percent penetration of the install base.”) 435 bosTon ConsUlTing groUP rePorT, supra note 188.

56 easily adapt to self-driving cars. When a car is being driven in self-driving mode, the person who would ordinarily have been the “driver” is really no more than a passenger. The car is driven by the algorithm built into the car’s computer system and by any updates. This is analogous to riding in a taxi or limou- sine driven by someone else. Imagine a robot sitting in the driver’s seat. Most, and perhaps all, states regu- late the insurance requirements for taxis, limousines, and (now) online ride services (also known as Trans- portation Network Companies [TNCs] such as Uber and Lyft). In April 2015, approximately 35 states had TNC legislation either enacted or pending.440 California enacted its own statute.441 This legisla- tion submits regulation of TNCs to the California Public Utilities Commission (CPUC), but also sets minimum insurance requirements below which the CPUC may not go. The statute requires a minimum of $1 million in liability and uninsured motorist cov- erage when a passenger is in the car; it requires $1 million in liability coverage from the time the driver accepts a passenger; and it requires $250,000 in cov- erage ($50,000 primary and $200,000 “excess”) from the time the driver turns on the app which allows potential passengers to solicit rides from the driver. To the extent driverless cars are deployed in fleets, either by the OEM or others, this model may commend itself. California’s current TNC statute applies only to businesses connecting passengers with drivers using their “personal vehicles.”442 Therefore, it may not be broad enough to cover driv- erless cars deployed on a fleet basis. A fleet may, however, fall within the jurisdiction of the CPUC as a “charter party carrier.” A charter party carrier includes “…every person engaged in the transporta- tion of persons by motor vehicle for compensation, whether in common or contract carriage, over any public highway in this state.”443 To the extent the cars are privately owned, simi- lar insurance might be offered on a group basis.444 The group would be those who own the vehicles and those related to the owner in much the same way Depending on their business relationship, they could allocate this insurance burden among themselves. If traditional products liability insurance is too expensive or unavailable, there is presently federal legislation in place to allow commercial entities to form associations to pool the risk themselves. In the 1970s, products liability insurance became very dif- ficult to obtain. Congress responded by adopting the Product Liability Risk Retention Act of 1981436 and 1986.437 These acts allow OEMs, wholesalers, dis- tributors, and retailers to form their own risk reten- tion groups to spread and assume all of, or a portion of, their products liability exposure.438 This shift may create an opening for some new and innovative insurance products. Rather than pur- suing a products liability claim against a commercial supplier, people may prefer to insure themselves against injury from driverless cars. Such a policy might resemble UM/UIM insurance (first-party insurance) or health insurance. The insured would have a claim against the insured’s own insurance company. The difference would be that the ability to recover would not turn on whether the other driver was liable to the injured person (this will seldom be the case). A first-party claim may be far more conve- nient and efficient than pursuing a claim directly against the commercial suppliers of the vehicle. Health insurance, which may soon be ubiquitous, is of that kind. Even in the event of an automobile acci- dent, the insured’s health costs, less any deductibles or co-pays, are covered by the insured’s own health insurer without regard to the legal responsibility of any other party. Indeed, health costs, which are often a significant part of an automobile accident injury claim, are covered regardless of fault. The health insurer, however, may or may not have a subrogation claim against the injuring party’s recovery.439 If insurers were to employ the UM/UIM or health insurance model to offer a policy covering, for exam- ple, pain and suffering, it might be offered as a stand-alone policy or as an endorsement to some other existing policy (e.g., auto policy, homeowners policy, or rental policy). 8. Insure Like Online Ride Services? There is an insurance model evolving in related vehicle transportation areas that lawmakers may 436 15 U.S.C. §§ 3901 et seq. (2012). 437 15 U.S.C. §§ 3901–3906 (2012). 438 See insUrAnCe regUlATion, supra note 389, § 3.8. 439 For a recent decision discussing the controversial area of subrogation by a health insurer in the context of ERISA, see Wurtz v. The Rawlings Co., 761 F.3d 232 (2d Cir. 2014). See also U.S. Airways v. McCutchen, __ U.S. __, 133 S. Ct. 1537, 185 L. Ed. 2d 654 (2013) (self-funded ERISA plan entitled to recover health expenditures from injured party, including injured party’s UM/UIM coverage). 440 ISO Introduces Endorsements to Address Rideshar- ing Policy Gaps, ins. J. (Apr. 2, 2015), http://www.insurance journal.com/news/national/2015/04/02/363254.htm. 441 CAl. PUb. UTil. Code §§ 5430–5443 (West 2010 & Supp. 2015). 442 Id. 443 Id. § 5360. 444 The California Insurance Code states that “Any insurer may issue any insurance coverage on a group plan, without restriction as to the purpose of the group, occupation or type of group.” CAl. ins. Code § 1861.12 (West 2013). At this writ- ing, the California Commissioner of Insurance is studying the possibility of narrowing the definition of “group” under this provision. Whether owners of self-driving or driverless cars may qualify as a “group” is at present an open question.

57 10. Telematics-Based Policies As OEMs become increasingly responsible for driving cars, they will need to gather information about how and where the car is driven. This infor- mation may be used to improve programing, avoid misuse, and achieve overall safety for those in the vehicle and others. This information may also be useful to improve insurance products. Ideally, an insurer would rate a policy based on real knowledge about the driver, the vehicle, and other hazards that it might present. It is impracti- cal, however, to put an observer in every vehicle, so insurers instead rely on “proxies”—e.g., driving record, age, gender, location, vehicle type, etc. These are the familiar rating factors used to price policies. With the new information flowing back and forth between the driver and the OEM, it may be possible to price policies so that the premiums better match the actual exposure. Proxies are, after all, only very rough approximations of risk. OEMs may find it in their interest to offer their own policies, or they may share this information with affiliated insurers. Use of this information for other than improving the safe driving of the vehicle may cause serious pri- vacy concerns. At present a number of insurers offer telematics-based policies on an optional (i.e., opt-in) basis.450 Even with an opt-in, however, California’s commissioner of insurance has not approved collec- tion of any data by insurers beyond mileage.451 In a November 12, 2014, letter to the Federal Trade Com- mission, a number of OEMs pledged not to pass on information to insurers and others without the own- er’s permission.452 11. The Future of Mandatory Automobile Insurance As the dangers of automobiles became apparent, many states adopted some form of mandatory insur- ance (usually known as Financial Responsibility Laws, or FR). Having in mind that at one time automobiles were causing approximately 55,000 fatalities per year in the United States, some form of mandatory cover- age seemed imperative. Eventually every state except New Hampshire adopted some form of mandatory automobile insurance, along with some form of either mandatory or optional UM/UIM insurance. that private auto insurance covers family members, permissive users, and others. 9. Cyber Insurance Cyber risks from hackers have become all too famil- iar as the personal and commercial world becomes ever more connected. Driverless cars may raise the impor- tance of cybersecurity because, unlike most hacking today, malicious cyber interference with an automobile may cause serious personal injury and property dam- age.445 At present there is little financial motive to hack into cars, but this may change.446 As a consequence, NHTSA is doing focused research on hacking and hack- ing defenses at its Transportation Research Center.447 In addition, the vehicle, or its manufacturer, may acquire data of a personal privacy nature. These dangers would suggest that there may be an evolv- ing market, at least at the commercial level, for cyber insurance policies to cover these enhanced risks.448 There is considerable doubt whether standard Commercial General Liability (CGL) policies cover cyber risk. The issue may turn on whether there was “property damage” or merely damage to electronic media and records. In any event, insurers are begin- ning to add cyber exclusions to the policies to avoid any ambiguity with respect to the issue.449 445 Three recent “friendly” hacking experiences have been widely reported. In one case the hacker hacked into a com- monly used dongle of the kind supplied by insurance compa- nies and others to monitor driving. Andy Greenberg, Hackers Cut a Corvette’s Brakes Via a Common Car Gadget, Wired (Aug. 11, 2015), http://www.wired.com/2015/08/hackers-cut- corvettes-brakes-via-common-car-gadget/. In another, hack- ers took over the steering of a jeep. Greenberg, supra note 194. The third incidence was the hacking into a TESLA. Kim Zetter, Researchers Hacked a Model S, But Tesla’s Already Released a Patch (Aug. 6, 2015), http://www.wired.com/2015/ 08/researchers-hacked-model-s-teslas-already/ Tesla claims to have fixed the vulnerability with a download. These all were friendly hacks, in the sense that those who did them were merely demonstrating to the industry the vulnerability without malicious intent. 446 One study reported that a disgruntled employee used a Web-based system to immobilize approximately 100 cars and leave them with their horns honking. Dowling & Part- ners Securities, LLC, Property and Casualty Research, It’s Been A Great Ride…, (Sept. 27, 2013), at 13. 447 Jim Travers, Inside the Government Lab Hacking Into Cars, ConsUmer rePorTs (May 7, 2015), https://www. yahoo.com/autos/inside-the-government-lab-hacking-into- cars-118366695712.html. See also Can Your Car Get Hacked? Your Driving Data Is At Risk. Someday, Your Car’s Controls Could Be As Well, ConsUmer rePorTs (Apr. 30, 2015), http:// www.consumerreports.org/cro/magazine/2015/06/can-your- car-get-hacked/index.htm. 448 lloyd’s rePorT, supra note 404, at 16–20. 449 Judy Greenwald, Insurers fight to bar cyber coverage under commercial general liability policies, bUs. ins. (Oct. 26, 2014), http://www.businessinsurance.com/article/20141026/ NEWS07/141029850. 450 Mark Hollmer, Progressive at PCI: Telematics will Become Ubiquitous Underwriting Tool (Oct. 28, 2014), http://www.carriermanagement.com/news/2014/10/28/ 131022.htm. 451 CAl. Code regs. tit. 10, § 2632.5(i)(5)(a) (2015) (“An insurer shall only use a technological device to collect information for determining actual miles driven….”). See also lloyd’s rePorT, supra note 404, at 18. 452 Joan Lowy, Automakers Vow to Protect Motorists’ Pri- vacy, sAn Jose merCUry neWs (Nov. 13, 2014), http://www. mercurynews.com/drive/ci_26929248/automakers-vow- protect-motorists-privacy.

58 other than cause. The program is administered by the Federal Court of Claims.456 G. Some Different Models for Compensating Those Injured by Driverless Vehicles Automobile insurance is not the only way to pro- tect the public with respect to accidents. Automo- biles are “Goods” under the Uniform Commercial Code and they are “Products” for products liability purposes. Apart from any express warranties, as goods they come to the consumer with implied war- ranties of merchantability and fitness for purpose. As products, they must be free of defects in design and manufacture (including, in California, satisfy- ing the reasonable expectations of a consumer).457 In addition, they must be accompanied by adequate warnings. Moreover, claims under the UCC and products liability may be asserted against the OEM and all in the commercial chain of distribution.458 This may include the entity programming the “map” for the vehicle.459 These rules were developed for the purpose of protecting consumers in much the way insurance protects the consumer. They also fold the costs of injuries into the cost of the goods, thus encouraging the development of safer goods and influencing rational consumer choices by reflecting injury costs in the price. Since those in the commercial chain are likely to be responsible for injuries caused by driverless cars, one would expect funding this liability will shift also to the business judgment of those in the chain. OEMs, for example, may self-fund, retain some of the risk, insure, insure through a captive, or adopt some other model. Since some surveys suggest that drivers believe that they are safer than a self-driv- ing car, marketing self-driving cars with an express warranty of their safety may be an effective market- ing tool. If the OEM “owns” the responsibility any- way, there would be little additional cost to them in making their existing responsibility express. In addition, it is unlikely that purchasers of auto- mobiles will own the software that drives the car. Like most other computer programs, ownership will Much has changed since those days. With far greater population, far more cars, and far more miles driven, the rate of fatalities has declined to approximately 33,000 per year.453 This is still a significant number. Over 10 years this is 40,000 more deaths than the population of Saint Paul, Minnesota. Nevertheless, the introduction of self- driving cars offers the prospect of dramatically reducing this toll, along with the related injury rate. This raises the question whether it will be neces- sary to continue mandatory automobile insurance requirements.454 Although many choose to insure against liability for non-auto related injuries, usu- ally through endorsements on their homeowners or renters policy, there is no requirement that they do so even though they may engage in any number of dangerous activities. Accidents arising from boating, ATVs, firearms, scalding water, power tools, play equipment, swimming pools, lawnmowers, and many other hazards are but a few examples. F. Connected Vehicle Communications Issues Looking further into the future, there will likely be a time when connected vehicle communications (perhaps V2V or V2I) play a very important role in transportation and road safety. Once most or all cars are communicating with one another, it may be almost impossible to assign responsibility for any particular accident. This will especially be so if they communicate in an anony- mous manner to protect the privacy of their passen- gers. Moreover, even if assigning responsibility were possible, it may not be worth the effort. This would suggest that it may be appropriate to consider an entirely different compensation and/or insurance system for those, hopefully rare, accidents. Since the many benefits flowing from such an integrated sys- tem accrue not only to the individual driver, but the public in general, policymakers may explore more publicly oriented compensation methods. As dis- cussed in Section IV, infra, one possible model is that designed for the rare adverse side effects that flow from vaccines. A $0.75 tax on each dose of the vac- cine funds the National Vaccine Injury Compensa- tion Program.455 Within some limits, those injured by vaccines may recover for their injuries without showing fault, defect, or any other responsibility 453 See Adrian Lund, Advanced Safety Technologies and Other Guideposts on the Road to Vision Zero (June 5, 2014), http://www.iihs.org/iihs/topics/presentations (Scroll down). 454 lloyd’s rePorT, supra note 404, at 18; Stephanie K. Jones, Future Visions: Will Driving Become Too Safe to Insure, ins. J. (May 8, 2012), http://www.insurancejournal. com/news/national/2012/05/08/246831.htm. 455 National Childhood Vaccine Injury Act of 1986, 42 U.S.C. §§ 300aa–1 to 300aa–34 (2012). 456 See Kevin Funkhouser, Paving the Road Ahead: Autonomous Vehicles, Products Liability, and the Need for a New Approach, 2013 UTAh l. rev. 437 (2013) (arguing for a no-fault scheme modeled after the federal National Childhood Vaccination Injury Act.) 457 Campbell v. General Motors Corp., 32 Cal. 3d 112, 124– 27, 649 P. 2d 224, 231 (1982). 458 Henningsen v. Bloomfield Motors, Inc., 32 N.J. 358, 161 A. 2d 69 (1960). 459 See Fluor Co. v. Jeppesen & Co., 170 Cal. App. 3d 468, 216 Cl. Rptr. 68 (1985) (treating an air chart as a “product” when an error in the map contributed to a crash).

59 This raises the question whether it would be in the public interest to look into ways to guarantee some protection when injured parties may no longer look to the commercial chain for compensation. Much like the National Vaccine Injury Compensa- tion Program, it might be funded with an assess- ment on sales or licenses. Regulations for the testing of self-driving cars in California, Florida, and Nevada have taken a small step in that direction. Testers must maintain $5 million in insurance, bonds, or audited net worth in order to test self-driv- ing cars on California’s public roads.461 I. Example of Awkward State Insurance Regulation—California’s Proposition 103 California is the largest insurance market in the United States and also has the largest number of cars on the road of any state. It also presents an interesting case study illustrating how state regula- tion of insurance may have unintended conse- quences for insuring driverless cars.462 In 1988, when driverless cars existed only in sci- ence fiction, California voters adopted Proposition 103. Unlike ordinary legislation, the proposition may be amended only by a 2/3 vote of the legislature, and then only if the changes “further” the purposes of the proposition. Otherwise, it may only be amended by another proposition adopted by the voters. Proposition 103 changed the regulation of insur- ance in a number of significant ways. Like a number of states, the proposition made automobile rates subject to “prior approval.” The proposition also mandated that three rating factors must be weighted higher than any others (including the capabilities of the vehicle) and in the following order: 1. Driver’s driving record (e.g., accidents and convictions for moving violations); 2. The number of miles driven per year; and 3. The number of years of driving experience.463 In addition to mandating that a driver’s driving record and years of driving experience be weighted more than any other rating factor, Proposition 103 also requires all auto insurers to offer a “Good Driver remain in the OEM, and the program will be licensed to the operator.460 In order to keep mapping and algorithms up to date, there will be a constant flow of information between the supplier and the self- driving car. Although there will be some privacy con- cerns, some flow in information will doubtless be necessary in order to keep driverless cars as safe as they may reasonably be made. Licensing, rather than selling, the programs also helps address a separate issue—how to address aging technology. Computers, including personal computers, age and become outdated even with the benefit of updates. Although many cars sold today will run for 20 years, it is doubtful that the technol- ogy behind a self-driving car will last that long. Ownership of the program by the OEM will allow the OEM to “retire” dated technology. The OEM could disable technology that is not updated or has become inadequate. This may compel the owner to return to the dealer for the installation of a new pro- cessor or program, or return the car to manual mode, or, perhaps, force retirement of the vehicle (with some attendant marketing issues presented by forced retirement). To the extent that driverless cars are marketed on a fleet basis with subscriptions, their constant use will make earlier retirement more economical. H. Guarantee Funds for OEMs Assuring the solvency of insurers is a primary function of state insurance regulators. When an insurer becomes insolvent, all states also have guar- antee funds covering some of the liability for insol- vent insurers (see discussion above). Guarantee funds usually have caps on coverage—e.g., $500,000 in California, although the guarantee is unlimited for workers’ compensation claims. There is no similar guarantee fund for suppliers of automobiles. As responsibility for injuries shifts from drivers to OEMs and others in the commercial chain, injured parties must look there for compensa- tion. The financial condition of those in the commer- cial chain is not regulated or vetted like insurance companies. There are any number of auto manufac- turers who have disappeared, and even the well- known brand, General Motors, is no longer the same company that bore that name a few years ago. There are also mergers and acquisitions that will raise questions of responsibility for injuries. 460 Kyle Wiens, We Can’t Let John Deere Destroy the Very Idea of Ownership, Wired (Apr. 21, 2015), http://www.wired. com/2015/04/dmca-ownership-john-deere/ (noting that with the exception of Tesla, automakers and even manufactur- ers of farm machinery assert ownership over the software in their products). 461 CAl. veh. Code § 38750(b)(3) (2014); flA. sTAT. § 316.86 (2015); NRS § 482A.060 (2015). 462 Robert W. Peterson, New Technology—Old Law: Autonomous Vehicles and California’s Insurance Framework, 52 sAnTA ClArA l. rev. 1341 (2012); Hilary Rowan, “Out-of Date Rating Factors May be Impediment to Self-Driving Cars,” dAily JoUrnAl (San Francisco, CA), Nov. 12, 2014, at 1. 463 CAl. ins. Code § 1861.02(a) (West 2013). In contrast to personal automobiles, the mandatory rating factors would not apply to automobiles deployed on a fleet basis because they do not apply to any policy insuring more than four vehicles. Id., § 660(a)(2).

60 developers to develop a sufficient understanding of the technology and the risks to make an educated guess at appropriate rates. It might be helpful to insurers, and regulators who must assure that rates are not excessive, inadequate, or unfairly discrimina- tory, if the insurance industry were more closely inte- grated in the development and approval process. As driverless cars move into the market place, they will begin to generate frequency and severity data. Unlike data generated from manually driven vehicles, the credibility of this data may rapidly change. The programs, algorithms, and maps driving the automobiles are likely to be updated frequently, or, perhaps, continuously. Thus, yesterday’s rates may no longer be appropriate for tomorrow’s vehicle. Assuming that driverless cars will prove much safer than manually driven cars, reducing the insurance burden on owners should increase acceptance of the vehicles. Unfortunately, the regulatory systems of many states, including California, are not geared to nimble rate adjustments. Some states do, however, allow insurers to flex within a range without approval. While lower frequency and ease of assigning responsibility because of information stored in the event data recorder should push rates lower, two factors push in the opposite direction. As responsi- bility moves from individuals (who may be unin- sured or underinsured) to the commercial side, the more serious injuries are likely to be adjusted at closer to their actual value. In addition, driverless cars may be more expensive to repair. How these two vectors will interact remains to be seen. California is in a unique position because of Propo- sition 103. Proposition 103 is driver-centric, not vehicle-centric. Two of the three mandatory rating factors (driving record and years of driving experi- ence) assume that there is a driver who is legally responsible for operating the vehicle. Likewise, the Good Driver Discount assumes that there is a driver who, if good, deserves the discount, and if “not good,” then not. These rating factors and the Good Driver Discount would do little mischief if they were not mandatory. How this regulatory system will accom- modate driverless cars is an open question. In the more distant future, as self-driving cars begin to dominate the market, the public may prefer to insure itself against injuries caused by faulty cars and faultless drivers. There may be a market for first party insurance (something like UM/UIM) to compensate for these kinds of claims. Health care costs, which are a large part of claims for more seri- ous injuries, are already of this kind. As we move even further into the future, it is likely that automobiles will both communicate with each other (V2V) and communicate with the Discount.” Insurers are to offer those who qualify a discount of “at least 20% below the rate the insured would otherwise have been charged for the same coverage.”464 Proposition 103 also does not allow insurers to change a rate (either up or down) without filing a “complete rate application.”465 Driverless cars, which are really computers with wheels, are likely to improve in safety at a rate more consistent with computer development than with Detroit design. Accidents, and perhaps near accidents, can be ana- lyzed, the algorithm can be adjusted, and it can be downloaded to every vehicle in the fleet.466 Although insurance regulators attempt to insure that rates are not excessive, inadequate, or unfairly discrimi- natory, unlike California, a number of states allow some degree of flexing up or flexing down without prior approval. This is usually within a range of 5 percent, 7 percent, or 15 percent.467 On September 15, 2014, the California Depart- ment of Insurance held its first hearing to begin to address some of these issues.468 J. Conclusions Whether rating driverless cars under a personal liability regime or a products liability regime, insur- ers will be challenged by lack of data. Testing data and simulations are helpful, but they are a poor sub- stitute for actual data generated by the driving of these vehicles in the hands of the public. Much of this data, such as it is, may be reported (as in California) to the DMV under its testing regula- tions, but it may not be available to insurers or others because it is considered proprietary by those report- ing. Some insurers are working closely with product 464 Id., § 1861.02(b)(2). 465 Id., § 1861.05(b) (“Every insurer which desires to change any rate shall file a complete rate application with the commissioner.”). 466 Evan Ackerman, Why You Shouldn’t Worry About Self-Driving Car Accidents, IEEE sPeCTrUm (May 12, 2015), http://spectrum.ieee.org/cars-that-think/transportation/self- driving/why-you-shouldnt-worry-about-googles-selfdriving- car-accidents (“The specific cause of the accident could then be identified, and then, more than likely, engineers could develop a way of making sure that the car would never, ever have that accident again. Furthermore, the update could be instantly propagated to every other autonomous car, mak- ing them all that much safer. Needless to say, humans don’t work this way, and we just keep having the same sorts of accidents over and over again.”). 467 A list of states and their regulatory framework appears at Regulation Modernization, ins. info. insT., http://www.iii. org/issue-update/regulation-modernization. 468 For a report of the hearing, see Don Jergler, Prop. 103 vs. Self-Driving Cars Revving Up in California, ins. J. (Sept. 18, 2014), http://www.insurancejournal.com/news/ west/2014/09/17/340898.htm.

Next: VII. PRIVACY AND SECURITY LAWS »
A Look at the Legal Environment for Driverless Vehicles Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB's National Cooperative Highway Research Program (NCHRP) Legal Research Digest 69: A Look at the Legal Environment for Driverless Vehicles explores legal policy issues that may be associated with driverless vehicles. It provides an introduction to how civil and criminal liability may adhere to driverless vehicles, the implications of these vehicles for privacy and security, how these vehicles are likely to become subject to and potentially alter prevailing automobile insurance regimes, and other related topics.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!