The Effects of Proposition 213 on the Costs of Auto Insurance in California

by Stephen J. Carroll, Allan Abrahamse

Download Free Electronic Document

FormatFile SizeNotes
PDF file 0.7 MB

Use Adobe Acrobat Reader version 10 or higher for the best experience.

What Is Proposition 213?

The cost of automobile insurance has been a major public policy issue in California, as in many other states, for more than a decade. A variety of public and private organizations and individuals have proposed alternative, purportedly less expensive, automobile insurance plans. But, to obtain the savings from such plans, the state would have to limit the compensation traditionally provided to people injured in auto accidents. To date, policymakers confronted with this trade-off have not been willing to limit accident victims' compensation rights.

Proposition 213, the Personal Responsibility Act of 1996, has qualified for the November 1996 California ballot. The proposition differs from previous proposals in that it would limit the compensation rights of only those people who were breaking certain laws when they were injured. If approved, the proposition would bar drunk drivers and uninsured motorists from compensation for any non-economic losses[1] resulting from auto accident injuries. It would also bar compensation for any loss incurred by felons who were involved in auto accidents while committing crimes or fleeing from them.[2]

We estimated the likely effects of the proposition's provisions regarding uninsured or drunk drivers on the costs of private-passenger auto insurance. Because of data limitations, we did not consider the provisions' effects regarding felons or its effects on the costs of commercial auto insurance.

Key Findings

Our analyses suggest that the proposition would reduce auto insurance costs. If current claiming, negotiating, and insurance-purchase patterns persist, the proposition would reduce auto insurers' compensation costs for personal injuries by about 10 percent relative to the costs under California's current auto insurance rules. Given the past relationship between compensation costs and auto insurance premiums in California, this difference would translate into a reduction of about 5 percent in the average California driver's auto insurance premiums.

To put this estimate in perspective, we estimated what auto insurance premiums would have been in 1994, the most recent year for which we have data on total auto insurance premiums, if the proposition had been in force then. Statewide, California drivers' auto insurance premiums would have been about $550 million lower if the proposition had been in force in 1994, a reduction of roughly $40 to the average California insured driver.

Our results address relative costs; they show the difference between what will happen if the current system is retained and what would occur if the proposal is adopted. We do not suggest that auto insurance costs will necessarily fall if California adopts the proposal. Proposition 213 may not reverse the long-term trend toward higher auto insurance costs. Rather, we suggest that Proposition 213 will slow the rate of growth in premiums so that future premiums would be roughly 5 percent less, on average, than they would be if the current system is not modified.

Our results address the effects of the proposition on the average California driver. Both the expected costs of insuring a driver under the current auto insurance system and the likely effects of the proposition vary from one part of the state to another. For example, the uninsured-motorist rate is much higher in urban areas than in rural areas. Consequently, the savings that would result from limiting compensation to uninsured drivers injured in auto accidents would be greater in urban areas.

Because approval of the proposition could engender changes in behavior, we recalculated our estimates under different sets of assumptions incorporating such changes. We also explored the sensitivity of these results to sampling error. Although the precise estimates vary from one set of behavioral assumptions to another, the results generally suggest that the proposition would cut the costs of compensating auto accident victims by 10 to 13 percent. Thus, our basic conclusion—that Proposition 213 would result in savings of about 5 percent on the average driver's auto insurance premiums—holds for all the alternatives we considered.

The Probable Effects of Proposition 213: Detailed Discussion

The traditional rules of the tort system govern recovery for auto accident injuries in California. An accident victim may seek compensation for all economic and non-economic losses from the driver who caused the accident. However, the victim is entitled to compensation only to the degree that the other driver is responsible for the accident.

Proposition 213 would eliminate compensation for non-economic losses to uninsured motorists and to drunk drivers injured in auto accidents. The proposition would not affect the rights of these drivers to compensation for economic losses. Nor would it affect the compensation rights of any other person injured in an auto accident—insured, sober drivers; passengers; pedestrians; bicyclists; etc.—including passengers injured while riding in cars operated by uninsured or drunk drivers.

In sum, the only accident victims that would be affected by the proposition are

  • uninsured or drunk drivers injured by an insured driver
  • drunk drivers covered by uninsured-motorist insurance injured by a negligent, uninsured motorist.

The savings achieved by the proposition would be the amount of compensation for non-economic loss that would be paid to these victims under the current law, plus the transaction costs—claims handling and defense costs—that insurers would have incurred in providing that compensation.

To estimate the effects of Proposition 213, we used data derived from a representative sample of California auto-accident injury claims closed with payment during 1992.[3] For purposes of the analysis, we assume that the distributions of accidents, losses, and claimants reported in those data are representative of the corresponding future distributions. We also assume that, in the future, 30 percent of California drivers will be uninsured and that 90 percent of insured drivers will purchase uninsured-motorist coverage.

Given these assumptions, we estimate that about 11 percent of future California auto accident victims will be uninsured drivers injured by an insured driver. Another 2 percent of future victims will be insured drunk drivers who are either injured by another insured driver or are injured by an uninsured motorist and have uninsured-motorist coverage. In all, the proposition would bar compensation for non-economic loss to about 13 percent of auto accident victims. If the costs of compensating uninsured or drunk drivers hurt in auto accidents are reduced by the average compensation for non-economic loss paid California drivers hurt in auto accidents, plus the associated transaction costs, the total costs of compensating auto accident victims would fall about 10 percent.

Personal-injury coverages account for about half of auto-injury premiums; property-damage coverages account for the other half. Thus, a 10 percent reduction in the costs of compensating auto accident victims translates into a 5 percent reduction in total auto insurance premiums. In 1994, total auto insurance premiums in California added up to more than $11 billion. Accordingly, if Proposition 213 had been in force then, the costs of auto insurance in 1994 would have been about $550 million lower:

  • Drivers denied compensation for non-economic losses because they were drunk or uninsured when they were injured would have lost about $360 million. (Because some accident victims are represented by attorneys paid on a contingency fee basis, a cut of $360 million in accident victims' gross compensation would be divided between the victims—in the form of lower net compensation—and attorneys—in the form of lower fees.)
  • Because insurance companies would have faced smaller claims from drunk and insured drivers injured in accidents, they would have had to pay about $70 million less in claims handling and defense costs.
  • Finally, if insurance companies' other costs (overhead expenses, selling expenses, taxes, and license fees) vary in proportion to premiums, the companies would be able to cut premiums $118 million and still earn the same rate of profit.

Possible Behavioral Responses to Proposition 213

The estimates described above assume that past behaviors persist. But it is possible that people will change their behavior if Proposition 213 is adopted. We speculated about what some of these possible behavioral changes might be, modified our model to reflect alternative behavioral assumptions, and re-estimated the proposition's effects. We emphasize that we have no evidence that any of these behavioral changes will occur if the proposition is approved. Our purpose is to identify the extent to which our estimates are sensitive to the behavioral assumptions that underlie the calculations.

It is possible that the claiming behavior of uninsured or drunk drivers might change if they could no longer obtain compensation for non-economic loss. We have found evidence of extensive excess claiming for medical costs in auto personal-injury cases across the United States, and particularly in California.[4] California's current system encourages excess claiming as a means for leveraging greater compensation for non-economic loss; by eliminating that incentive to affected drivers, Proposition 213 would discourage fraudulent or excessive claims. At the same time, many accident victims rely on compensation for non-economic loss for the funds needed to pay their attorney; eliminating this source of funds to affected drivers may reduce their ability to obtain an attorney and, consequently, may discourage some legitimate claims.

The civil justice policy implications of reducing the frequency of excessive claims are very different from the policy implications of reducing the frequency of legitimate claims. But from a cost perspective, the two look the same: Fewer claims imply lower costs.

To estimate how reducing the frequency of claims—excessive claims, legitimate claims, or some combination—would affect costs, we assumed that adoption of Proposition 213 would result in either a 25 percent or a 50 percent reduction in the frequency of claims by drunk and uninsured drivers, and estimated what the savings would be in each case.

The negotiating behavior of accident victims, of their attorneys, or of claims adjusters might change if the proposition is adopted. In many cases, objective measures of an accident victim's non-economic losses or of the insured's negligence or both are not available and are determined in the course of negotiations. Our data reflect the results of those negotiations as they are conducted in the context of California's current insurance system. It is possible that under the proposition those involved in resolving a claim by an uninsured or drunk driver will agree on a compensation figure that is less than what would have been paid under the current system, but not by the full amount that our data suggest is being paid for non-economic loss.

To estimate how a partial, rather than full, elimination of compensation for non-economic loss to uninsured or drunk drivers would affect our estimates, we assumed that, despite the formal provisions of Proposition 213, uninsured or drunk drivers injured in auto accidents would be compensated for either 25 percent or 50 percent of their non-economic loss and estimated what the savings would be in each case.

Adoption of the proposition could also change some drivers' insurance-purchase behavior. The potential costs of going uninsured would be increased—uninsured drivers would not only be in violation of the law, they would not have access to compensation for non-economic loss if they were injured in an auto accident. At the same time, the proposition would reduce the costs of purchasing auto insurance relative to the current system. It is possible that some drivers who would go uninsured under the current system will choose to purchase insurance under the proposition.

To estimate how an increase in the fraction of drivers who purchase insurance would affect our estimates, we assumed that either 25 percent or 50 percent of the uninsured-motorist population chooses to purchase insurance and estimated what the savings would be in each case.

Our estimates are based on data obtained in a sample of claims; they are subject to sampling error. Some of these claims were high-dollar claims, and it is possible that these high-dollar claims had an undue influence on our results. However, high-dollar claims are a fact of life, and although they are relatively rare, they might indeed have a real influence on savings under the proposition.

To examine the possible effect of sampling error on our results, we estimated the proposition's cost effects under three very different assumptions about the sample: (1) We used all the cases in our sample. (2) We dropped the 10 percent of all cases with the greatest economic loss. (3) We doubled the economic loss of cases in the top 10 percent of all cases. It is unlikely that the effect of sampling error in a file of 6,000 cases would be as great as the effect of discarding or doubling the top 10 percent of the sample.

In sum, we considered the sensitivity of our results to three alternative assumptions about the values of each of four factors: claim frequency, the fraction of non-economic loss compensated, the percentage of uninsured drivers induced to purchase insurance, and the frequency of very large claims. We calculated relative savings under Proposition 213 under all 81 combinations of the four factors over the three levels discussed above.

Table 1 shows the results of these calculations.

Table 1. Relative Proposition 213 Savings (%) Under Alternative Assumptions

Claiming Rate Proportion of Non-Economic Loss Compensated Percentage of Uninsured Drivers Getting Insurance Cost Savings Estimates
Nominal Drop Top 10% Double Top 10%
No claims reduction No non-economic loss 0% 10.0 9.0 10.8
25% 11.2 10.4 12.2
50% 12.4 11.9 13.6
25% of non-economic loss 0% 7.5 6.7 8.1
25% 9.2 8.7 10.1
50% 11.0 10.7 12.1
50% of non-economic loss 0% 4.9 4.5 5.3
25% 7.3 6.9 7.9
50% 9.6 9.4 10.5
25% claims reduction No non-economic loss 0% 11.4 10.5 12.2
25% 12.3 11.7 13.3
50% 13.2 12.8 14.4
25% of non-economic loss 0% 9.5 8.8 10.1
25% 10.8 10.3 11.7
50% 12.2 11.9 13.3
50% of non-economic loss 0% 7.6 7.1 8.1
25% 9.4 9.0 10.1
50% 11.1 10.9 12.1
50% claims reduction No non-economic loss 0% 12.8 12.0 13.6
25% 13.4 12.9 14.4
50% 14.1 13.7 15.2
25% of non-economic loss 0% 11.5 10.9 12.2
25% 12.4 12.0 13.3
50% 13.3 13.1 14.5
50% of non-economic loss 0% 10.3 9.8 10.9
25% 11.4 11.1 12.3
50% 12.6 12.4 13.7
Maximum 14.1 13.7 15.2
Top quartile 12.4 12.0 13.5
Median 11.2 10.7 12.2
Bottom quartile 9.5 9.0 10.3
Minimum 4.9 4.5 5.3

The results in Table 1 support three important conclusions.

First, relative savings always exceed about 5 percent, regardless of how we combine the various factors. It seems quite likely that Proposition 213 will reduce compensation costs.

Second, relative savings generally exceed about 10 percent. Savings drop below 10 percent only in the relatively few cases in which drivers negotiate high compensation for non-economic losses. If the terms of the proposition are really implemented, it seems unlikely that such negotiations will occur frequently. Thus, it seems quite likely that Proposition 213 will modestly reduce compensation costs.

Finally, relative savings rarely exceed about 13 percent. They do so only when many currently uninsured drivers decide to purchase insurance after Proposition 213 goes into effect and/or if we assume that our data file under-represented high-dollar claims.

In light of these findings, we believe that the relative savings under Proposition 213 will fall somewhere between 10 and 13 percent.

Data and Methods

We obtained data from closed-claim surveys conducted by the Insurance Research Council.[5] These surveys obtained detailed information on a random sample of about 6,000 California auto-accident injury claims closed with payment during 1992 under the principal auto-injury coverages. The data describe each victim's accident, resulting injuries and losses, and the compensation obtained from auto insurance. We combined data from several sources to estimate insurers' transaction costs,[6] including both allocated loss-adjustment expenses—costs, primarily legal fees and related expenses, incurred on behalf of and directly attributed to a specific claim—and unallocated, or general claim-processing costs, for each line of private-passenger auto insurance.[7]

We estimated the effects of Proposition 213 on insurance costs by comparing the costs of compensating the accident victims in the sample under the current insurance system with the costs of compensating the same victims, for the same injuries and losses, under Proposition 213. We included all accident victims—insured and uninsured drivers, passengers, pedestrians, bicyclists, people injured in single-car accidents, etc.—in these calculations.

We assumed that the proportion of drivers who will purchase each available type of auto insurance personal-injury coverage and, by implication, the proportion of drivers who will go uninsured under California's current system will be about the same as those today. Given these assumptions, we computed the probability that an accident victim will have access to compensation under each coverage, multiplied by the average compensation paid California accident victims under that coverage, and summed over all coverages to estimate insurers' expected compensation costs under the current system. We then estimated a break-even premium for the current system—the amount insurers would have to charge the average insured driver to just recover what they paid out in compensating victims and the transaction costs they incurred in providing that compensation.

We assumed that drivers would make the same insurance-purchase decisions under the proposition and, by implication, that the same proportion of drivers would go uninsured. We computed insurers' expected compensation costs under these assumptions, and estimated the break-even premium under the proposition—the amount insurers would have to charge insured drivers to just recover compensation costs.

Finally, we calculated relative savings under the proposition as the percentage difference between the break-even premium under the current system and the one under the proposition.

We focused on the effects of the proposed plan on auto insurers' compensation costs, including both the amounts they pay out in compensation and the transaction costs they incur in providing that compensation. We neglected the many other factors (e.g., insurers' overhead, profit margins, and investment income) that also affect insurance premiums.

We focused on the relative costs of the two insurance systems. Because any factors that proportionately affect costs under both the current system and the proposed plan net out in the comparison, the results are insensitive to changes in such factors over time.


Abrahamse, Allan F., and Stephen J. Carroll, The Effects of a Choice Auto Insurance Plan on Insurance Costs, Santa Monica, Calif.: RAND, MR-540-ICJ, 1995.

Carroll, Stephen, Allan Abrahamse, and Mary Vaiana, The Costs of Excess Medical Claims for Automobile Personal Injuries, Santa Monica, Calif.: RAND, DB-139-ICJ, 1995.

Carroll, Stephen, et al., No-Fault Approaches to Compensating People Injured in Automobile Accidents, Santa Monica, Calif.: RAND, R-4019-ICJ, 1991.

Insurance Research Council, Auto Injuries: Claiming Behavior and Its Impact on Insurance Costs, Oak Brook, Ill., September 1994.

National Association of Insurance Commissioners, State Average Expenditures & Premiums for Personal Automobile Insurance in 1994, Kansas City, Mo., January 1996.


  • [1] Economic losses include an accident victim's medical costs, lost wages, burial expenses, replacement service losses, and other pecuniary expenditures. Non-economic losses include physical and emotional pain, physical impairment, mental anguish, disfigurement, loss of enjoyment, and other nonpecuniary losses.
  • [2] Felons would be allowed to collect damages for intentional acts of harm against them.
  • [3] The data were collected by the Insurance Research Council (1994) from 61 insurance companies that together accounted for about 77 percent of California's private-passenger automobile insurance (by premium volume) in 1992.
  • [4] See Carroll, Abrahamse, and Vaiana (1995).
  • [5] Insurance Research Council (1994) provides a detailed description of the data.
  • [6] Carroll et al. (1991), Appendix D, describe the data and methods used to estimate insurers' transaction costs.
  • [7] We do not include claimants' legal costs, the value of claimants' time, or the costs the courts incur in handling litigated claims. Those costs do not affect insurers' costs and, hence, do not affect auto-insurance premiums.

This report is part of the RAND Corporation issue paper series. The issue paper was a product of the RAND Corporation from 1993 to 2003 that contained early data analysis, an informed perspective on a topic, or a discussion of research directions, not necessarily based on published research. The issue paper was meant to be a vehicle for quick dissemination intended to stimulate discussion in a policy community.

Permission is given to duplicate this electronic document for personal use only, as long as it is unaltered and complete. Copies may not be duplicated for commercial purposes. Unauthorized posting of RAND PDFs to a non-RAND Web site is prohibited. RAND PDFs are protected under copyright law. For information on reprint and linking permissions, please visit the RAND Permissions page.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.