Using Vehicle Inpection Data to Understand Personal Car Mileage

Using Vehicle Inpection Data to Understand Personal Car Mileage

Nominated for The Planning for Sustainable Land Use and Transport Award


Sally Cairns, Transport Research Laboratory, Paul Emmerson, Transport Research Laboratory, Simon Ball, Transport Research Laboratory


This paper reports on new insights into personal car ownership and use in Britain, derived from a research dataset generated from vehicle test and vehicle stock records.


In Britain, vehicles of three years or more are subject to an annual roadworthiness inspection, during which a vehicle odometer reading is taken. This information has been comprehensively recorded on a centralised computer system since 2006, and was first made publicly available in 2010. A major research project, running from 2012-2016, has been undertaken to understand the potential of this data to explore a range of issues, including vehicle ownership and use, vehicle-related energy use and vehicle-related emissions that affect air quality impacts and climate change. As part of this work, the project team has also had confidential access to information about vehicle location, enabling relatively fine-scale geographical analysis. This paper focuses on one aspect of the project, namely the insights generated on vehicle mileages. As well as briefly explaining the techniques used to convert the randomly timed odometer readings into a set of usable mileage information, the paper demonstrates:

• At both large and small geographical scales, the distribution of private car mileages follows a reasonably similar pattern, with the best fit probably being a gamma distribution. One implication is that mean mileages are a relatively meaningful way of comparing areas, although examination of the distributions also shows that other distributional properties vary by a greater amount, and may also be helpful for differentiating areas that mean mileages would otherwise suggest were similar.

• Looking at the variation in both mileages and vehicle characteristics between areas, there is considerably greater differences in the average amount that people drive compared with the average engine size or CO2 emissions of the average cars owned. This highlights the importance of policies aimed at changing travel behaviour not simply the types of vehicles owned for achieving energy efficiency, air quality and climate change objectives.

• Multiple regression models can predict car ownership and use to a relatively high degree of accuracy when underlying socio-geodemographic information is included. Population density and land use patterns emerge as a key explanatory factor. Inclusion of measures of accessibility and public transport provision considerably improve the power of the models and can individually be used to predict a non-trivial amount of the observed variation in car ownership and use.


Association for European Transport