At-Fault Claims: The Good, The Bad And The Ugly

Two mechanisms exist for compensating not-at-fault automobile property damage (PD) claims in Canada. Many provinces rely on a traditional third-party model, while Ontario, New Brunswick and Quebec use a first-party model, according to which insurers compensate their own insureds for not-at-fault vehicular damage. Insurers have advocated the use of first-party recovery for the following reasons. First, pricing is improved because insurance reflects the cost to repair the insured’s vehicle rather than the cost of repairing the average vehicle. Both administrative costs and time to settlement is reduced because there is less subrogation between insurers. Furthermore, from a business perspective, insurers prefer to “look after” their own policyholders. We examine these and other implications of first-party recovery below.

THE GOOD

To examine whether claims settle faster and with lower adjustment expenses in a first-party model, we compared one year of both at-fault and not-at- fault physical damage claims for a large insurer writing in Alberta and Ontario. For Ontario, we collected claims arising from direct compensation/ property damage and collision coverages. For Alberta, we used claims arising from third-party PD liability and collision coverages. We find that it costs less to settle claims in Ontario. The average loss adjustment expense associated with a property damage claim was $82.32 in Alberta and $55.26 in Ontario. Additionally, claims settled faster in Ontario. In Alberta, the average number of days from the time the claim was reported until the first payment was 47 days, while in Ontario it was 31 days. In Alberta, it took 20 days from the date of the last payment until the claim closed and only six days in Ontario. These differences in settlement times are statistically significant. From an operational perspective, claimants benefit from a first-party recovery mechanism as they receive payments faster and insurers benefit if settlement costs are proportional to the length of time it takes to close a claim.

We then examined the average payment made to claimants, adjusting for subrogation and salvage and difference in the cost of living between the two provinces. The average payment made to a claimant in Ontario was $3,531 and in Alberta, $3,464.The difference between these two amounts is not statistically significant, but there are significant differences in payments made to not-at-fault claimants. Not-at-fault claimants were compensated at a higher rate in Ontario. The average payment made in Ontario was $3,143 versus $2,903 in Alberta.

Next we examined whether differences in compensation arose because accidents in Ontario were different than accidents in Alberta. We compared average payments to all claimants — both atfault and not-at-fault — in Ontario and Alberta for the seven most-frequently-observed accident types and found no obvious pattern. For some accidents, payments were statistically higher in Ontario; for others, payments were statistically higher in Alberta. In some instances, there was no difference. However, for each accident type, not-at- fault claimants in Ontario received higher payments than not-at-fault claimants in Alberta.

It seems that this difference in payments to not-at-fault claimants exists because of the nature of first-party versus third-party relationships. Requesting anonymity, one claims manager noted: “Honestly, we may be a little more liberal in Ontario dealing with our own customer on a not-at-fault loss than with a third-party in Alberta paying them for the not-at-fault loss.”

THE BAD

In theory, fault assignment should be driven only by fault determination rules and not the recovery mechanism. In Ontario, the degree of fault is established by the Fault Determination Rules. In Alberta, most insurers are signatories to the Insurance Bureau of Canada (IBC)’s Agreement Respecting Standardization of Claim Forms and Practices, and Guidelines for the Settlement of Claims. The two documents are very similar, leading us to expect no systematic differences in fault assignment between the provinces. However, we find insurers in Ontario seem to assign fault less frequently.

Strong incentives exist in a third-party recovery mechanism to assign fault accurately, since fault assignment affects the indemnification paid by each insurer. In addition, a built-in control exists because fault assignment requires the agreement of all the adjusters involved. Under a first-party recovery mechanism, however, if collision coverage has been purchased, the indemnification paid is the same (except for the collision deductible) for at-fault and not-at-

fault claims. Moral hazard arises because the adjuster has less of an incentive to assign fault correctly when the size of claim paid does not depend on correct fault assignment. When settling a claim for his own insured, it may be expedient for the adjuster to assume his insured is not-at-fault. Additionally, the adjuster might not assign fault correctly because settling in favour of one’s insured may reduce settlement costs and increase goodwill.

There are many ways to test the impact of this moral hazard on risk classification. First, if this moral hazard exists, more drivers will be classified as lowrisk in first-party recovery than in third-party recovery jurisdictions. An analysis of the distribution of insureds by driving record class reveals that more drivers are classified as low-risk drivers in Ontario than in Alberta. However, road safety statistics also show that Alberta has more reported collisions per registered vehicle than Ontario. Even if fault is assigned correctly in Ontario, there should be a greater proportion of lowrisk drivers. Thus this statistic cannot be used to provide conclusive evidence on fault assignment.

Another way to examine fault assignment is to compare the proportions of PD claims classified as at-fault and not-at- fault in both provinces. We received statistics from three large insurers. In Alberta, the percentage of PD claims coded as at-fault ranged from 54% to 62%. But in Ontario, the percentage of claims coded as at-fault ranged from 37% to 44%. This raises an interesting question: How can the frequency of atfault claims be less than 50%? Fault determination guidelines dictate that fault should be assigned in every auto accident. Drivers in a single-vehicle accident should be deemed at-fault. In multiple- vehicle accidents, fault assignment rules allow for at least one and up to all drivers to be assigned some degree of fault. And in both provinces, any level of fault assignment is coded as an at-fault accident. Thus, the ratio of at-fault claims to all claims should be greater than 50%. This provides compelling evidence that fault is not assigned properly in Ontario.

THE UGLY

Assigning fault correctly in Ontario may not affect the cost of claims paid by insurers, but it has an impact on risk classification and pricing. When adjusters do not assign fault correctly, the classification that occurs via the experience rating component of the insurance premium becomes less effective. Some high-risk drivers will be misclassified as low-risk and overall there will be more pooling of risks within each driving record class.

Because auto insurance coverage is not comparable between the two provinces, we cannot use price differences between provinces to measure the impact of misclassification. Instead, we have developed a stochastic model of experience rating. In both provinces, drivers are typically assigned to a driving record class based on the number of years of at-fault-claims-free driving. Those in Class 0 have zero years of at-fault-claims-free driving; those in Class 1 have one year without an at-fault claim; and so on up to Class 6, which represents at least six consecutive years of no at-fault claims. In addition, we allow for Class 5*, the modified accident forgiveness state for Cl ass 6 drivers who have an at-fault claim. All drivers in the model are assigned an accident rate drawn from an underlying statistical distribution. The parameters for this distribution are chosen to resemble observed accident frequencies in each driving record class. Therefore the overall frequency of at-fault claims is 3.5% — with Class 6 drivers having a frequency of less than 3% and Class 0 drivers having a much-higher frequency. The model is run twice. The first time, it is assumed that at-fault accidents are properly recorded. The second time, it is assumed that the adjuster misclassifies a claim 30% of the time. Thirty percent roughly matches the difference in at-fault claim rates between Ontario and Alberta.

As a result of misclassification, drivers look better to the insurance company than they really are, since there are more Class 6 drivers than there should be. Driving record classification loses some predictive power: there is more heterogeneity within classes (there is greater dispersion of underlying accident probabilities) and the average accident frequencies in adjacent driving record classes are not as different as they should be. Misclassification results in more pooling of risk types, and rate class differentials are closer to one than if at-fault accidents were correctly classified.

Insureds are more interested in the cost of insurance and not in rate class differentials. Insurers, on the other hand, are more interested in collecting enough premiums to cover losses and not in what a single insured pays. Therefore, we model the relationship between pure premiums required to cover losses and the level of misclassification. We use the same distribution of accident frequencies as above and set the severity per claim to $5,000. Also, we assume that all insureds carry collision coverage with a zero deductible; hence, under first-party recovery, fault assignment does not affect total claims paid. At-fault PD claims incorrectly coded as not-at fault are paid from first-party PD coverage and at-fault claims are paid from collision coverage. In the third-party recovery mechanism, all accidents are coded as at-fault and are paid under collision coverage. The total dollars paid by insurers are the same under both recovery mechanisms.

By simulating two portfolios, one with at-fault accidents correctly recorded, and the other with at-fault accidents coded not-at-fault 30% of the time, we obtain the actuarially fair premiums to cover the PD claims (shown in Table 1).

When claims are misclassified, all drivers pay more for insurance for two reasons. First, as a result of misclassification, fewer drivers are in the high-risk classes. Second, the difference in differentials across driving classes under misclassification is smaller than if fault were assigned correctly. Premiums collected from the high-risk classes are less than if adjusters correctly assigned fault. Because the premiums are higher than they should be, all drivers who have not been misclassified lose: they subsidize drivers who were misclassified.

Lower settlement costs and maintaining goodwill provides justification for misclassification of at-fault claims, but not without cost. Although the total premium collected by insurers is the same, who contributes these dollars differ in the two scenarios. When fault is assigned correctly, premiums increase for drivers who have at-fault accidents. When fault is not assigned correctly, insurers must use overall rate increases to cover losses. This should provide incentives for correct fault assignment.

OPPORTUNITIES FOR INSURERS

Real benefits arise from first-party recovery for not-at-fault property damage. First party recovery is associated with lower settlement expenses and quicker settlement times. As such, first-party recovery could provide noticeable savings if adopted in other jurisdictions. Not expected was the higher compensation for not-at-fault claimants under first-party recovery, suggesting that one’s own insurer is more generous in claims payment than a third-party insurer.

Fault may not be correctly assigned in the first-party recovery mechanism: insureds are less likely to be found at fault. Although this may lead to happier insureds, it degrades the experience-rating component of the insurance premium. Experience rating becomes less effective in separating high and low risks. Moreover, our model shows this results in premiums that are marginally higher for all insureds.

The mispricing result has interesting implications from a competitive standpoint. If an insurer were to become aware of an incentive to not assign fault, it could gain a competitive advantage by reducing misclassification. Although settlement costs may increase, the experience rating component would classify insureds correctly. This might lead to a reduction in premiums charged to properly classified drivers in all classes. In addition, economic theory suggests that correctly classifying insureds reduces moral hazard because the cost of insurance is aligned with each insureds’ true risk. On the other hand, misclassification is a relatively inexpensive way to “grease the squeaky wheel,” if indeed those insureds that are misclassified would otherwise give rise to the largest settlement costs.

Overall, in practice, an examination of the good, bad and the ugly is necessary to ensure that insurers are using the best compensation mechanism for not-at-fault property damage.

———

In theory, fault assignment should be driven only by fault determination rules and not the recovery mechanism for compensating auto claims.