Do Odometer Readings Accurately Predict Annual Mileage?
By Raj Bhat
Multiple industry studies show a direct correlation between miles driven and auto claims. Consequently, it's extremely important for insurers to correctly estimate annual miles driven in order to price auto exposures accurately. Unfortunately, most estimates are very inaccurate. A 2006 California Department of Insurance study found frequent inaccuracies regarding mileage estimates: 11 percent were understated by more than 10,000 miles; 15 percent by 6,000 to 10,000 miles; and 30 percent by less than 5,000 miles.[1]1. Don McNeil, Policy Research Division, California DOI, Errors in Self Reported Mileage for California Vehicles, September 21, 2006.
Some carriers — and even some regulators — believe that odometer readings are a good predictor of future annual mileage. These companies collect self-reported odometer readings or tap third-party data sources such as service records to estimate annual miles. But just because the methodology is based on measurable data, is it an accurate way to calculate premium?
Quality Planning (QPC) decided to test this premise.
Study Methodology
The California Bureau of Automotive Repair (BAR) records odometer readings from smog checks, which are required biannually for vehicles six years and older. Most of the odometer readings in the database are verified and accurate.
QPC sampled more than 4.5 million vehicles with three BAR odometer readings from 2004 to 2009. Odometer readings number 1 and number 2 were used to calculate annual mileage in the "prior period." Readings number 2 and number 3 were used to calculate the "subsequent period." Finally, we calculated the percentage change in the subsequent period from the prior period.
Using odometer readings to predict annual mileage assumes
that there is consistency between prior and subsequent periods and that the prior period will accurately predict miles driven in the subsequent period. If so, the percent change should be close to zero.
Study Results
Table 1 shows the change in average mileage from the prior to subsequent period. The table also gives the percentage of vehicles with mileage increases and mileage decreases along with the average increase or decrease. The overall decrease of 1,097 miles hides the real difference because the positive and negative variances cancel each other. Disregarding the sign, the absolute delta is 3,864 miles. Figure 1 and Table 2 further illustrate that more than half the vehicles would have significant mileage rating error of 25 percent or more when using prior odometer readings to estimate future annual mileage.
Table 1:
Change in Annual Miles by Model Year
Click to enlarge
Table 1 shows the change in average mileage from the prior to subsequent period. The table also gives the percentage of vehicles with mileage increases and mileage decreases along with the average increase or decrease. |
Annual Mileage Factors
Numerous factors affect driving behavior and total annual miles driven. Foremost among these are driver characteristics (demographic profiles, public transit availability), lifestyle changes (unemployment, marriage), and household vehicle inventory. It's essential to include these factors when predicting annual mileage.
Figure 1:
Distribution of Mileage Changes
Click to enlarge
Figure 1 shows the percent change in annual mileage from prior to subsequent period. |
Table 2:
Percent of Vehicles Exceeding a Given Mileage Change
Click to enlarge
Table 2 shows the percentage of vehicles that exceed (either positive or negative) a specific mileage delta. |
Conclusion: More than half the vehicles would have significant mileage rating error of 25 percent or more when using prior odometer readings to estimate future annual mileage. |
Summary
- When odometer readings are used to estimate annual mileage, a key assumption is that future use of the vehicle will be similar to its usage in the past.
- QPC's analysis indicates this process is flawed. More than half the vehicles would have an error of more than 25 percent in estimated annual miles using this process.
- Lifestyle and vehicle changes are the main factors that cause a change in miles driven. The odometer-based methodology fails to account for these changes.
- The odometer method will result in rating error and significant premium undercharges or overcharges.
- Premium misallocation creates long-term rating integrity issues. Insufficient premium puts profitability at risk and attracts less desirable drivers. Overcharging makes insurers uncompetitive and vulnerable to losing their best customers.
Conclusion
Pure odometer-based mileage estimates are error-prone and should be supplemented with additional data that accounts for changes in lifestyle and household characteristics. 
Raj Bhat, Ph.D., is president of Quality Planning.
To review the complete study, visit www.qualityplanning.com.
1. Don McNeil, Policy Research Division, California DOI, Errors in Self Reported Mileage for California Vehicles, September 21, 2006.
|
