Calculating Auto Insurance Refunds for COVID-19 isn’t Easy

Several personal auto insurance companies have announced refunds/credits recognizing the unexpected reduction in accidents resulting from COVID-19. The responses have included flat per vehicle credits, percentage credits for a specified period and credits applied to the next policy term. The Insurance Journal reported the following actions by the top 5 auto insurers by premium volume (listed alphabetically):

Most have applauded the actions, but some have questioned the adequacy of response as miles driven are reportedly down 50-60%. Insurance Business cited a JD Power study that found 57% of consumers who were aware of the discounts thought they were insufficient. This post highlights some of the challenges insurers face determining actuarially sound refunds for their traditional (non-telematics) personal auto portfolio.

First, there is a fundamental difference between personal auto insurance and other consumer products. Most product companies know the cost of goods sold when the product is sold which makes it relatively easy to quantify the impact of a change and adjust prices accordingly. In contrast, auto insurance is a promise to indemnify the customer for claims that may happen on or after the policy effective date. Since some claims take years to settle, insurers may not know the full cost of goods sold until years later. Actuaries rely on historical data to estimate the future costs; unfortunately, the historical data is less relevant when conditions have changed so dramatically. Tier 1 insurers have an advantage as they can quickly generate new data, but most insurers must make decisions with limited company data to guide them.

Second, the reduction in accidents is caused mainly by a reduction in miles driven which has led to the exposure being much lower than expected. Insurers do not collect actual miles driven daily for traditional policies. Consequently, insurers cannot easily calculate the reduction in miles driven for their traditional portfolio and are forced to rely on extrapolations from their telematics programs and/or data about the general population. Even if insurers had the data, there is not a one-to-one relationship between miles driven and insurance losses. In other words, a 50% reduction in miles does not result in a 50% reduction in crash-related insurance losses (all else being equal). To further complicate matters, “everything else is not equal” during this time. For example, there is less traffic congestion and fewer miles driven during high-risk times which should further decrease insurance losses. There are also factors that could raise insurance costs with the most obvious being supply chain issues or increased fraud during the economic crisis.

Third, insurance companies--like all businesses--incur expenses in running the business. For auto insurers these expenses (employee salaries, office building, equipment, acquisition costs to name a few) typically represent 30%+ of each premium dollar. Some of these costs will automatically reduce as losses and premium reduce during the crisis, but many will not. Additionally, auto insurance policies indemnify losses not related to crashes (e.g., weather and theft). If accidents are down 40% and $0.35 of every premium dollar covers non-accident costs, then the indicated discount would only be 26%. This is easy to deal with mathematically but probably results in an answer that is lower than the expectation of a customer who feels they are driving significantly less than normal.

A fourth complication is that insurance is regulated, and the regulation happens on a state-by-state basis. Typical product companies can change prices at their discretion at any time. In contrast, auto insurers must provide justification for price changes in each state impacted. State regulators are undoubtedly working diligently with the insurance companies to approve justified actions quickly; even so, that process does introduce some extra time. Additionally, if the insurance company overestimates the reduction and gives too much money back, they will not be able to raise their prices to recoup any excess refunds.

Finally, no good deed goes unpunished. If an unforeseen event results in more losses than expected, the insurance companies are not able to recoup the losses. Fortunately for auto insurers, this crisis is resulting in far fewer losses than expected. Insurers could have done nothing, but many chose to proactively return money to their customers at a time when people are hurting economically. Unfortunately, the J. D. Power study found that these refunds may have raised awareness and kicked off some unintended consequences. Many customers wanting further relief plan to reduce coverage, increase deductibles or shop around for a better deal. This points out that good intentions and the best “actuarial” answer may not be the best long-term answer for the insurer. The marketing departments have a big role here.

There is clearly a lot of pressure on auto insurance companies to respond. Hopefully, this post illuminates some of the challenges facing insurers as they do so. I am certain that all insurers are monitoring the situation carefully and may respond with further actions if this continues for longer than expected.

Here is hoping that all the readers are safe from COVID-19 AND accidents.

SIDE NOTES:

This post is dealing with traditional auto insurance policies only. If you want to read thoughts about UBI, I’d suggest you read my post “Is UBI the COVID-19 Economic Auto Insurance Cure: https://www.werneradvisory.com/post/is-ubi-the-covid-19-economic-auto-insurance-cure.

Actions of more companies can be found here: https://www.insurancejournal.com/news/national/2020/04/13/564510.htm

If you want to learn more about the actuarial techniques use to predict the future, please refer to the “Basic Ratemaking” text I co-authored with Claudine Modlin:https://www.casact.org/library/studynotes/Werner_Modlin_Ratemaking.pdf