!!Экзамен зачет 2023 год / Tort liability versus other approaches

(Cite as: 14 Int'l Rev. L. & Econ. 53)

International Review of Law and Economics

March, 1994

*53 TORT LIABILITY VERSUS OTHER APPROACHES FOR DETERRING CARELESS DRIVING

Frank A. Sloan

Copyright © 1994 by Butterworth-Heinemann; Frank A. Sloan

J. Alexander McMahon Professor of Health Policy and Management and Professor of Economics, Duke University, Durham, North Carolina 27708, USA

BRIDGET A. REILLY

Former Research Associate, Institute for Public Policy Studies, Vanderbilt University, Nashville, TN 37203 USA

AND

Christoph M. Schenzler [FNa1]

Graduate Student in Economics, Vanderbilt University, Nashville, TN 37235 USA

I. Introduction

A main justification for tort liability is that it deters injuries. [FN1] Yet empirical evidence on deterrence is quite limited. [FN2] If tort liability does not achieve appropriate levels of deterrence, its rationale is largely undercut since there are clearly less expensive ways to compensate injury victims, such as first-party insurance.

For several reasons, motor vehicle fatalities represent an interesting outcome measure for analysis of effects of tort liability versus other methods for decreasing accident costs. First, a very large percentage of tort cases involve automobile accidents. Thus, a strong test of the impact of tort law is to determine whether such law reduces automobile accidents. Second, there is evidence that drivers' precaution levels are at least as important as road conditions and other external factors in the determination of traffic accident rates (Evans, 1991; Zlatoper, 1989). Third, there are sufficient differences across states and over time with respect to rules affecting tort liability that statistically reliable results can be obtained. Fourth, a question that is often ignored *54 in the economics literature, the effect of the intercession of insurance between the tort system and the tortfeasor, can readily be analyzed in the context of automobile accidents. Finally, one can make comparisons between effects of tort sanctions and other policies to reduce accident rates, such as criminal fines, policies affecting the price of alcohol, and restrictions on alcohol availability. [FN3]

Since data concerning automobile fatalities are readily available and because automobile injury data are neither reliable nor reported systematically, this study tests for the effect of various policies on mortality using data from the Fatal Accident Reporting System (FARS). Section II provides a conceptual framework for analyses of factors likely to affect fatality rates. Particular attention is given there to effects of tort and insurance regimes. Section III describes the data used and the empirical specifications. Section IV presents the results of the empirical analysis. Section V presents further discussion and conclusions.

II. Conceptual Framework

The purpose of this section is to summarize the theory underlying our analysis of determinants of motor vehicle fatality rates. We begin with a discussion of tort and insurance rules. We then discuss the roles of other explanatory variables.

Tort and Insurance Rules

In the bilateral version of the model of accidents, both injurers and victims can take precautions and thereby

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

lower the probability of an accident. Thus, each potential party to an accident considers the precaution levels of others in making his or her own decision about care. In equilibrium, no party has an incentive to alter his/her behavior (Shavell, 1987, Chapter 2). In this study we specify and estimate a reduced form equation, relating traffic fatalities to driver characteristics and various exogenous factors, including liability rules and insurance laws and practices. Observed fatality rates reflect equilibrium care levels in response to these various factors.

Dram shop liability. Research suggests that a high percentage of drunk drivers obtain their alcohol from establishments that sell alcohol for on-premise consumption. [FN4] Most states have imposed tort liability on commercial servers of alcoholic beverages ("dram shop" laws) either by statute or case law. [FN5] A typical dram shop law provides that a commercial server of alcoholic beverages may be found liable for accident costs *55 from injuries caused by their patrons. Often separate laws apply to minors. The efficacy of such laws depends in part on costs servers incur in monitoring their customers. One commentator was pessimistic about dram shop laws' deterrent potential because, for one, the due-care standard for serving an obviously intoxicated person tends to be too low. [FN6] However, some empirical evidence suggests that such laws may be effective in reducing alcohol-related motor vehicle deaths (Chaloupka et al., 1993).

Demand for liability insurance and deterrence. Conceptually, the deterrent effect of tort liability depends on the extent to which potential tortfeasors are made to bear the cost of the injury they cause. A defendant is only potentially at risk up to the value of his assets. The cost of many motor vehicle-caused injuries far exceeds the wealth of the majority of drivers (Miller et al., 1989). The fact that potential tortfeasors have limited liability has led to speculation that the rich take more care than the poor (Shavell, 1986, 1987). [FN7]

The fact that persons may be liable for amounts up to their wealth provides the motive for purchasing liability insurance. [FN8] With liability insurance, there is likely to be moral hazard. The cost of an accident to a negligent injurer no longer equals the victim's loss, but rather the present value of the increase in premiums resulting from "chargeable accidents." Even if premiums are increased using standard actuarial principles, the increase in premiums that arises when a driver has been found liable for an accident may be appreciably less than the damages that are levied as a result of that accident. A careless driver may also be surcharged for traffic violations. In such cases, surcharges may exceed the fines imposed. Because of a preference for "flat" rate structures in some states (Rottenberg, 1989), automobile liability insurance premiums often do not accurately reflect expected accident costs of individual drivers (Bruce, 1984; Sloan and Githens, forthcoming). This makes drivers less careful than they would be if premiums were fully experience rated.

The possibility that negligent drivers would not have sufficient assets to compensate victims for the cost of their injuries led many states to implement compulsory liability insurance statutes during the 1980s, requiring persons registering a motor vehicle to show a certificate stating coverage of liability insurance in a specified minimum amount. [FN9] These statutes replaced financial responsibility laws. The latter are weaker because they only require that drivers demonstrate they have sufficient ability to compensate victims after an accident occurs or being convicted of a traffic law violation (Commerce Clearing House, 1989).

Making drivers buy liability insurance plausibly makes them less careful. The extent of moral hazard depends on the amount of required coverage (minimum liability limits imposed) and on the extent to which premiums are experience rated. [FN10] *56 Mandatory insurance with a flat premium structure should reduce drivers' incentive to be careful since there is little or no offset to the reduction in exposure to damages. But, with compulsory coverage and hefty premium surcharges, drivers who have little wealth and might otherwise not purchase automobile liability insurance are made to pay for being careless through higher premiums. [FN11]

No-fault insurance. Several states and other countries have adopted no-fault insurance. Under pure no-fault, the right to sue is eliminated and is replaced by compulsory first-party insurance, termed personal injury protection (PIP). The effect of this change on deterrence depends on (1) the fraction of claims barred from tort liability and (2) the extent of sensitivity of drivers' premiums to the driver's precaution level. The argument made about experience-rated premiums and liability insurance also applies to first-party insurance.

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

In the United States, states adopting automobile no-fault laws have retained victims' right to sue for most serious injuries. However, the share of injuries barred from tort liability varies among the states; dollar thresholds above which victims can sue tend to be fixed in nominal terms, and the thresholds have not kept up with inflation (AIRAC, 1989). Some states do not exclude any injuries from tort recovery but require that drivers purchase minimum levels of PIP.

Unfortunately, there are no data on premium-setting practices for such insurance for the observational period we studied. In 1993, we surveyed insurance departments in each state with any form of no-fault insurance. Over half the states permitted a surcharge on PIP by itself if the driver was determined to have been at fault for the accident. In virtually all the remaining states, a surcharge on PIP was allowed by statute or regulation if another coverage, such as liability or collision, was surcharged. Generally, respondents believed that insurers did in fact impose surcharges on PIP.

The empirical evidence on deterrent effects of no-fault laws is mixed. Landes (1982) found that implementation of no-fault laws led to increased highway fatalities in the United States and, she argued, an erosion of incentives to take care. One shortcoming of the study is that she failed to control for other policies and factors that plausibly affect fatality rates, such as premium surcharges and criminal penalties for careless driving. Subsequent analyses with U.S. data have not confirmed her finding. [FN12] Evidence that no-fault insurance leads to more accidents does exist for Quebec and New Zealand, where no-fault has been adopted in purer forms. [FN13]

Liability rules. Traditionally, the liability rule has been negligence with a defense of contributory negligence. Under this rule, an injurer escapes liability if it can be proven that, at the time of the accident, the victim's precaution level fell below the due care standard.

*57 The pure comparative negligence rule is identical to negligence with a defense of contributory negligence when only the injurer is at fault, but when both injurer and victim fail to satisfy the due care standard, accident costs are dividend according to each party's relative contribution to the accident. A second form of comparative negligence, now much more common than the pure form, is modified comparative negligence. Under some states' comparative negligence rules, a plaintiff is barred from recovery if his degree of negligence is greater than or equal to the defendant's. In others, the plaintiff is not compensated if his negligence is greater than the defendant's. A final form of comparative negligence, rarely applied, is the "slight-gross" rule. Under this rule, the plaintiff's contributory negligence bars recovery if his negligence is more than slight and the defendant's less than gross. A recent analysis of compensation patterns using a national database on closed automobile claims revealed that compensation levels under the slight-gross rule were similar to those in states that retained the contributory negligence rule (Sloan and Schenzler, 1992). [FN14]

Although the motivation for the shift away from the contributory negligence rule has probably been to achieve a fairer sharing of losses among the parties to an accident, the effects of various rules on deterrence are of interest here. The usual theoretical result is that choice of contributory versus comparative negligence does not affect care levels. Potential injurers and victims have incentives to set care at socially optimal care levels under both rules, and each party acting in a socially optimal way yields a unique equilibrium. [FN15]

This conceptual analysis did not, however, consider how changing from contributory to comparative negligence affects drivers' incentives in the presence of insurance. To incorporate effects of insurance, we consider the effects of the rules on payments that drivers make to others as injurers and the compensation that they receive as victims. To the extent that drivers carry liability insurance, a major effect on drivers as injurers is felt on the premium surcharges imposed on them under alternative liability rules. [FN16] Under contributory negligence, an individual could look forward to escaping liability and hence a premium surcharge if the other driver were partially at fault. Under comparative negligence, the driver would be charged with an accident under such circumstances, but this assumes that the driver can in fact escape liability if the other driver is partially at fault under contributory negligence. Empirical evidence indicates that payment is frequently made to automobile ac-

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

cident victims who are partly at fault under contributory negligence (Sloan and Schenzler, 1992). Virtually all motor vehicle liability claims are dropped or settled. Even under contributory negligence, it may be more profitable to settle than to prove that the victim was partly at fault for the accident.

The effect of the change from contributory to comparative negligence on victims' compensation is clearer. Under comparative negligence, a partially at fault victim is slightly more likely to receive compensation, and more importantly, the rate at which the victim's percentage at fault reduces his/her compensation is less under contributory negligence (White, 1989; Sloan and Schenzler, 1992). Thus, except for greater delays in receiving compensation under comparative negligence (more cases go to trial), victims should be better off under comparative negligence. Since automobile *58 liability insurers levy surcharges based on claim frequency, not claim size (Lemaire, 1985), drivers as injurers are likely to be unaffected by the lower discount on degree of victim fault under comparative negligence.

With the impact on injurers' incentives unclear and with victims better off under comparative negligence, it would appear that drivers' incentives to be careful are greater under contributory negligence. [FN17] Nevertheless, choice of a comparative versus a contributory liability rule may have little effect on the decision to deviate substantially from the due care standard, such as to drink and drive, because in such cases one party's actions are likely to have been almost the entire cause of the accident.

Other Factors

Fatality rates are plausibly positively related to levels of consumption of alcoholic beverages, which in turn are negatively related to price and availability, including the minimum drinking age. [FN18] Penalties and enforcement of criminal laws, such as those pertaining to driving under the influence of alcohol, like tort liability, are designed to make persons more careful. The difference is that, unlike tort liability, there is no insurance market to cover losses from criminal liability. Levels of driving activity and road conditions also are likely to affect fatalities. [FN19]

III. Data and Empirical Specification

Data

Data on motor vehicle accidents by state and year, 1982-90, came from the Fatal Accident Reporting System (FARS) operated by the National Highway Traffic Safety Administration. Data on individuals were grouped into state-year-age aggregates. The age groups were 18-20, 21-24, and 25-64. Automobile liability insurers typically charge higher premiums to persons under age 25. Special dram shop laws apply to minors. Also, splitting the under age 25 category into 18-20 and 21-24 allowed us to measure the impact of minimum drinking age laws. [FN20] The District of Columbia, Alaska, and Hawaii [FN21] were dropped entirely and those state-years for which we could

TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE

*59 obtain no data on alcohol prices, [FN22] leaving 391 observations in each regression. During 1982-90, fatalities were highest for persons 18-20, followed closely by the 21-24 group; rates were far lower for adults 25 and over (Figure 1). There is no discernible trend in such rates.

Equation Specification

Dependent variables. The dependent variables are motor vehicle fatalities per 1000 population for the three age groups. [FN23] The explanatory variables fall into three categories: tort liability and insurance, other policies, and other factors.

Tort liability and insurance. To measure the effects of variations in states' tort liability regimes, variables were included for dram shop laws, compulsory liability insurance, no-fault insurance, and contributory versus comparative negligence. A binary variable identified states with dram shop laws, implemented by statute or by case law, operating statewide in the year. [FN24] In the analysis of fatalities involving persons *60 18-20, the dram shop variable referred to liability imposed on commercial servers for selling alcoholic beverages to

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

minors. For the other age groups, the variable identified states with dram shop liability for serving obviously intoxicated persons.

We included two variables for compulsory liability insurance, a binary variable identifying states with a compulsory liability insurance law [FN25] and a variable interacting this binary with the surcharge on liability premiums for, alternatively, (1) a DUI conviction or (2) a chargeable accident, for married males over age 25. [FN26] The surcharge variable (measured as a fraction) came from a national survey of insurers conducted in 1991 (Sloan and Githens, forthcoming). Since no national data on surcharges were available for earlier years, the surcharges were specified to be time invariant. However, judging from the information available for some states for earlier years (Bruce, 1984), relative surcharges among states appear to be stable.

No-fault insurance was measured by two variables. First, we included a variable for the fraction of accidents involving bodily injury for which a tort claim could not be filed. In states without a no-fault law, this fraction was zero. [FN27] The second no-fault variable is a binary variable for states that did not bar tort claims for bodily injuries caused by a motor vehicle, but did require that drivers purchase minimum amounts of PIP. As discussed earlier, the effect of mandatory purchases of insurance coverage depends in part on the extent to which a surcharge follows from chargeable accidents and traffic convictions. From our survey we found that surcharges on PIP are typically possible. However, we could not obtain information on actual surcharges from the insurance departments analogous to the information obtained on liability insurance directly from insurers (Sloan and Githens, forthcoming). Thus, our variable for compulsory first-party insurance was only specified as a binary.

We included explanatory variables for pure comparative negligence and modified comparative negligence, with contributory negligence being the omitted reference group. [FN28] Because of the relationship between tort liability, wealth, and bankruptcy, and lacking a time-series cross section of wealth data, we included per capita income *61 as a "tort liability" variable. [FN29] However, the net effect of income on motor vehicle fatalities is unclear since higher income is related to greater demand for alcohol and liability coverage above any staterequired minimum, among other factors. These other influences could counteract any greater incentive to take care among those with higher incomes.

National trends in the tort and criminal sanctions and insurance variables included in this study are presented in Table 1. The fraction of the U.S. population aged 18-24 covered by dram shop liability rose from 0.45 to 0.64 between 1982 and 1990. Nearly two thirds of this population was covered by state laws requiring compulsory liability insurance in 1982; by 1990 over four fifths was. There was an erosion in the fraction of claims barred from tort liability by no-fault laws. Fewer persons were compelled to buy first-party insurance coverage for personal injuries (PIP) in states that did not exclude some claims from tort liability before 1984 and after 1988 than in either 198283 or 1989-90. The pure comparative negligence rule grew slightly less common and modified comparative negligence grew more common during 1982-90.

Other policies. We constructed a state and year-specific index of alcohol prices from data published quarterly by the American Chamber of Commerce Researchers Association (ACCRA). ACCRA provided separate estimates of prices per bottle of beer, wine, and liquor. To obtain an index of overall alcohol prices for this analysis, various prices were weighted by share of beer, wine, and liquor in total consumption measured in gallons. However, the ACCRA price data were only for consumption in the home. During the 1980s, the relative price of alcohol consumed away from home increased appreciably relative to the price of alcohol consumed at home. To account for this, each state-year's value was calibrated so that the national year-to-year change in our alcohol price measure relative to the Consumer Price Index (CPI) for all goods and services was the same as the change in the CPI component for alcohol relative to the change in the overall CPI, with California in 1990 as the base state and year. An alcohol price index was used rather than measures of state alcoholic beverage taxes since tax incidence is likely to differ between states that operate their own retail outlets and those that do not.

In the analysis of fatalities occurring to persons 18-20, we included variables for minimum drinking age and

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

the interaction of the binary variable for minimum drinking age and alcohol price. Although drunk minors probably also affect fatalities of persons in the other age groups, the effect has only been documented for persons under age 25. [FN30] Variables for two types of criminal sanctions, both defined for the first drinking and driving (DUI) conviction, were included: mandatory minimum jail term (in days) and mandatory minimum length of license revocation (in days). [FN31] A variable measured the fine for failing to wear a seatbelt in 1990 dollars. [FN32] The extent

TABLE 1. Tort and criminal sanctions and insurance: Variable means [FN1]

-

-

-

-

---------------------------------------------------------------------------

-

-

-

-

---------------------------------------------------------------------------

fraction

of

claims barred

from tort

liability

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

0.03

compulsory

seatbelt (1990

$)

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

population

-

-

-

-

---------------------------------------------------------------------------

FN1 Means are weighted by the number of persons aged 25-64 in the state-

year.

Means exclude District of Columbia, Alaska, and

Hawaii.

*63 of law enforcement was measured by the number of full-time equivalent police per 1000 population. [FN33]

Other factors. Other exogenous variables included were the number of miles driven per driver and the fraction of miles driven that were in rural areas. Neither the miles driven nor the rural miles fraction were defined for specific age groups. [FN34]

To account for omitted time-invariant and time-varying influences, binary variables were included for state and year. Including the state variables was essential since certain policies, such as those affecting the price of alcoholic beverages in the state, are plausibly correlated with the public's attitude toward consumption of such beverages. Automobile no-fault insurance may be adopted in jurisdictions with more liberal political climates. Although proxies for the public's attitude, such as religious preferences, could have been included as explanatory variables, it is difficult to account for such influences satisfactorily in this way. [FN35]

Inclusion of binary variables for individual years is more controversial. Underlying the inclusion of such variables is the assumption that there have been important attitudinal and behavioral changes, such as an increase in health awareness, not adequately accounted for by the other explanatory variables. We found the year variables to be highly correlated with some of the policy variables, and it was sometimes difficult to distinguish effects of policy changes from trends. Because some of our results are sensitive to this presence/absence, regressions with and without the year binaries are presented below.

Functional form. Following Saffer and Grossman (1987a, 1987b), a logistic transformation of the dependent variable resulting in the following general functional form was used:

TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE

where xkjt is the kth independent variable in the jth state in year t and ujt is the disturbance term. The logit coefficient (beta)k is interpretable as the percentage change in the odds of motor vehicle mortality for a one-unit change in the value of an explanatory variable. Weighted least squares regression with the population in the age group in the state-year as the weights was employed. [FN36]

IV. Results

Tort Liability and Insurance

The first eight variables pertain to tort law and insurance (Table 2). Most of the variables have statistically significant impacts on fatality rates in the variants for the age 21-24 and 25-64 cohorts with the time variables included. With time excluded,

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

TABLE 2. Motor vehicle fatality regressions

[FN1]

-

-

-

-

---------------------------------------------------------------------------

-

-

-

-

---------------------------------------------------------------------------

Tort liability and

1.00-1.03

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.

(Cite as: 14 Int'l Rev. L. & Econ. 53)

© 2007 Thomson/West. No Claim to Orig. U.S. Govt. Works.