PAYDAYS: dynamic nature of driving behavior.

Traditional insurance models are based on a combination of static factors such as the driver’s socio-demographic profile and vehicle information in conjunction with driving history. With the advent of new mobile technologies, some insurance companies have begun exploring a new auto insurance model known as Usage-Based-Insurance [UBI] or Pay-As-You-Drive-And-You-Save [PAYDAYS], which aims to incorporate individualized, real-world dynamic driving patterns into actuarial pricing. While most existing UBI or PAYDAYS efforts can be considered a major leap forward, the majority of researchers and insurers currently rely solely on user GPS trajectories, which only measure the specific driver’s performance--such as the number of miles driven, travelling speeds, and hardness of braking--without considering other critical risk factors in the surrounding environment that may also contribute to crash risk, i.e. the contextual-sensitive risk factors.

Metropia proposed that the consideration of these contextual risk factors would offset risk for automobile insurers providing PAYDAYS coverage and lead to greater savings for low-risk drivers operating under such a policy. In September of 2013, The Federal Highway Association contracted Metropia to prove their PAYDAYS theory and study the implications of such an approach to insurance companies' premium calculation model.

Metropia collected individualized driving behavior data from the smartphone GPS module, combined with geographical network information and dynamic traffic conditions, to identify driving risk factors and evaluate driving behaviors under various contexts. The main contribution and research findings from this research project include:

  • Through its app, Metropia collected a total of 131,537 trips taken by 503 panel members over an 18-month period, which corresponds to 1,090,136 miles traveled.
  • Analysis definitively proved that driving behavior is context-sensitive, particularly with regard to the traffic conditions and the roadway geometry surrounding the vehicle of interest.
  • Through comparison to prior studies without context information, Metropia’s analysis demonstrated the benefits of utilizing context-relevant information in the driving behavior assessment process.
  • The developed model found that the rates insurers charge PAYDAYS customers to be off by as much as 25%-31% in either direction--undercharging or overcharging drivers. By applying a driver’s individual risk assessment, insurers are able to offer safer drivers a more competitive rate while charging less-safe drivers a rate more appropriate for their level of risk.

The findings of our study can further existing knowledge about driving exposure factors that are closely linked to crash risk, help insurers restructure their existing pricing models to allow for variation in premiums based on individualized driving characteristics, and provide the actuarial foundation for advanced forms of PAYDAYS insurance pricing.

  1. Identifying Driving Risk Factors to Support Usage-Based Insurance using Smartphone Personalized Driving Data
  2. Studying Driving Risk Factors using Multi-Source Mobile Computing Data
  3. Contextual Driving Risks Analysis using Dynamic Smartphone-Based Data: The Potential for Usage Based Insurance
  4. Analyzing Driving Behavior Using Context-Sensitive Individualized Smartphone-Based Trajectory Data (under review by Accident analysis and Prevention)
  5. A Bayesian Network Model for Contextual versus Non-Contextual Driving Behavior Assessment (under review by Transportation Research Part C: Emerging Technologies)
  6. The Use of Insurance Telematics Data in Auto Insurance Rate Making (under review by the Journal of Risk and Insurance)
  7. Federal Highway Administration Pay-As-You-Drive-And-You-Save (PAYDAYS) Insurance Actuarial Study Final Report (to be online soon)
Federal Highway Administration

Allen Greenberg

Project Manager

Allen Greenberg

Project Manager
Overall project oversight and guidance for project sponsor.
Metropia

Yi-Chang Chiu, PhD

Principal Investigator

Yi-Chang Chiu, PhD

Principal Investigator
Overall project management and visionary.

Xianbiao Hu, PhD

Co-Principal Investigator, Project Manager

Xianbiao Hu, PhD

Co-Principal Investigator, Project Manager
Subject-matter expert on travel behavior and modeling; Led project management, methodology development, and execution.

Xiaoyu Zhu, PhD

Metropia Research Lead

Xiaoyu Zhu, PhD

Metropia Research Lead
Subject-matter expert on statistical analysis and modeling.

Yifei Yuan

Research Team Member

Yifei Yuan

Research Team Member
Provided various support for this research project.

Zheng Li

Research Team Member

Zheng Li

Research Team Member
Provided various support for this research project.

PingChang Chen

Research Team Member

PingChang Chen

Research Team Member
Provided various support for this research project.
Illinois State University

Yu-Luen Ma, PhD

Co-Principal Investigator

Yu-Luen Ma, PhD

Co-Principal Investigator
Subject-matter expert on insurance and actuarial science.

PAYDAYS DATA DESCRIPTION

Abstract

PAYDAYS dataset is the dataset using for the PAYDAYS (Pay-As-You-Drive-As-You-Save) project. The dataset mainly contains three parts: survey data, trajectory data, and dynamic traffic data. Survey data contains the collected survey about the demographic profile and accident history of volunteers. Trajectory data contains the GPS trajectory of volunteers when they navigating with the smartphone app. Dynamic traffic data contains the road network information and estimated travel time of the link every 15 minutes.

Sample Dataset and Full Dataset

To separate different usage, we provide two types of datasets: sample dataset and full dataset. Both of them show the data used in the PAYDAYS project. The difference is the full dataset contains all the data used in the project but the sample dataset only contains a small part. Both datasets have three kinds of data: survey, trajectory, and dynamic traffic. Note that in the sample dataset, trajectory data, and dynamic traffic data are merged into the file. Please do not share the dataset without the permission from Metropia, Inc. The trajectory data contained in the full dataset is over 1.5 GB after compression. So the access to the full dataset is upon request through contact.


PAYDAYS: Executive Summary