Abstract
The US federal fuel tax is not adjusted for inflation, and thus has been effectively reduced by 70% in the last 25 years. This puts significant pressure on budgets seeking to fund highways and public transit systems.
Various entities have proposed alternative fees and programs. We initially perform analysis on the fees (cents/mile) needed for equivalent gas tax revenue in Pennsylvania. We use odometer data from inspections to track VMT, and consider various demographic factors such as types of vehicles, income, county of residence, etc., to suggest an equitable mileage-based fee. We will partner with a connected vehicle company to demonstrate how VMT data could be robustly collected and provided.
Description
Fuel taxes are the primary means of generating transportation revenue. The US federal fuel tax (18.5 cents/gallon) was last raised in 1993, and is not adjusted for inflation, and thus has been effectively reduced by 70% on a per-gallon basis in the last 25 years. Compounding that problem, increasing fuel economy (and also electric vehicles) has led to many vehicles providing little or no contribution in terms of fuel taxes. These factors put significant pressure on transportation budgets seeking to fund construction and maintenance of highways and public transit systems.
Various entities have proposed alternative fees, such as higher vehicle registration fees or mileage-based fees (where drivers pay per mile driven instead of per gallon of fuel consumed). Zhang et al. (2009) proposed a vehicle mileage fee be placed based on income and spatial equity, based on households. The General Accountability Office (2012) similarly recommended that Congress further explore mileage fees and consider implementing a pilot program. In order to create mileage-based fees, more robust data sources of miles driven are necessary to find the most effective policy to implement. In this project, we focus on mileage-based fees.
Such fees allow variable cost basis but also ensure that all vehicles operating in a jurisdiction are able to contribute to transportation funds. The same entities that have proposed alternative fees have also called for pilot programs to demonstrate how these alternatives could be implemented. Various pilot projects have been done in the past decade (such as those in the Pacific Northwest), mostly with smart tags on windshields and other transponder-like devices. While they were proven to work, they were relatively cumbersome and expensive to create and maintain on a per-vehicle basis given added technology needed to be installed. However, the connected vehicle (CV) and smartphone era has ushered in game-changing technology opportunities to make alternative transportation fees practical and feasible in a very short time horizon.
Our proposed work in this project has two parts. First, we will perform the analysis needed to assess the average level of mileage-based fees (i.e., X cents per mile) necessary to maintain the equivalent gas tax revenue generated in Pennsylvania (which was $4 billion in 2018) from the current 56 cents per gallon state tax on gasoline. We will then create a finance/policy modeling framework to consider how various implementations of the mileage-based fee could lead to equity problems across the State. To do this, we will use odometer data from inspection records in Pennsylvania over the past 15 years to track miles traveled at the vehicle level and see how this average VMT varies by county, but also by various other demographic factors such as types of vehicles, income, etc., to suggest a robust mileage-based fee framework for the State. The total mileage-based fees paid can be compared to the expected total gas taxes paid across all of these geographic and demographic groups as well. This equity analysis is necessary because it will be critical to know in advance whether "marketing" the mileage-based fee to state legislators would be problematic if rural drivers would be paying significantly more (or less) in mileage-based fees than they are already paying from gas taxes. In terms of a modeling framework, we also intend to assess whether or how differential fees could be created for urban versus rural areas (for example, to compensate for higher local use of public transit), arterial or interstate highways, or to balance socio-demographic concerns. In short, we will create various prospective "fee structures" to be used in the mileage-based fee system. We envision a spreadsheet dashboard-like deliverable that shows comparative transportation fees paid by drivers in each zip code or county in Pennsylvania to help the discussion of fee levels and structure.
The inspection-odometer data is historical in nature and provides representations of miles driven per vehicle across the state. However, it does not provide any details into where those miles might have been driven (e.g., were they on Interstate highways, or even within Pennsylvania?). In the second (parallel) part of the project, we will partner with a connected vehicle (CV) company - Automatic Labs, a SiriusXM company - to demonstrate how data on vehicle operation and use could be robustly collected and aggregated to a third party provider of information on behalf of Pennsylvania in support of a mileage-based fee system. While all CV companies collect and organize data differently, inevitably they all access the same general information from vehicle on-board systems. As a third party provider (i.e., not a manufacturer) Automatic has implemented their platform by creating bluetooth and 3G devices that plug into the vehicle DLC port under the dashboard (the same connection used for OBD scans by service technicians and inspectors). Automatic's bluetooth dongle also connects to smartphones to leverage location and Internet communications. This CV data, which includes many fields but inevitably is most useful at the trip level, is collected in real-time, but does not need to be reported at such a high frequency. Instead it could be organized by day or week and reported as needed. But depending on the eventual technology implementation, short turnaround information (such as miles driven in the past week) could be useful. For example, CV devices could note the miles driven in a trip, as well as the specific roadway segments followed (using the same segment IDs as defined by the google maps API). If reported to a third party aggregator that calculates the mileage-based fee, they could assess a constant or differential fee per mile for specific segments in Pennsylvania and present it as a bill to the driver at a certain frequency. Our intention in this second part of the project is to show how the underlying "mileage data" already exists to be combined with the "fee structure" created in the first part of the project.
The combination of a fee structure and proof of existence of mileage data will demonstrate the feasibility of mileage-based fees. We expect this demonstration to lead to a leapfrog awareness of how soon mileage-based fees could be a reality after years of having expensive transponder-like visions for the technology. We look forward to working with PennDOT or other jurisdictions interested in experimenting with our vision.
Prior Work Done In Support of This Project
We have already done substantial analysis of odometer data and created detailed estimates of VMT by Pennsylvania county and zip code (Peck et al 2018). While the data used and results are several years old, the code and methods can be quickly and easily leveraged for this project. We will use updated data from PennDOT and Partner CompuSpections for this purpose. CompuSpections is a private company that creates computer systems to collect, track, and manage data from passenger vehicle inspections in Pennsylvania. They have roughly 10 million safety inspection records from the state in the past 15 years and will provide new data not previously provided to other projects from 2017 and 2018 for this project (in-kind contribution). Likewise data from PennDOT provides registration and emissions inspection data for vehicles. With all of this data, we can create very robust estimates of VMT for the first part of the project.
DataDrivenIM is a new company founded by PI Scott Matthews and CompuSpections CEO Bernie Elder and provides analytic solutions for organizations in the vehicle safety or emissions inspection-maintenance (IM) industry. DataDrivenIM and CompuSpections will provide in-kind support of algorithms, code, and other insights that will help understand data from vehicle on-board diagnostic (OBD) systems as well as computational analytics support of the approximately 100 million inspection records to be used in the dataset.
Note: We are still negotiating support letter language from two of the project partners (CompuSpections and Automatic). We will upload these before the specified deadline. Note that we have worked with both organization in PITA and/or UTC projects in the past and will continue to work with the same project contacts as before, thus we do not expect significant problems in finalizing these letters.
References
United States Government Accountability Office (2012) Pilot Program Could Help Determine the Viability of Mileage Fees for Certain Vehicles.
Peck, Dana E., and H. Scott Matthews, "SCAVENGING DATA FROM ODOMETER READINGS IN SAFETY AND EMISSIONS INSPECTION RECORDS FOR MORE ROBUST ESTIMATES OF VEHICLE MILES TRAVELED", Proceedings of the 2018 Hong Kong Society for Transportation Studies (HKSTS) Conference, Hong Kong, Dec 8-10, 2018.
Zhang, L., McMullen, B. S., and Valluri, D. (2009) Vehicle mileage fee on income and spatial equity. Transportation Research Record, 2115, pp. 110-118.
Timeline
Summer (July-Sept) 2019: Inspection Data Collection, Extraction and Analysis
Fall (Sept-Nov) 2019: Creation of Equivalent Revenue Financial / Policy Models for Mileage-Based Fees and Gasoline Taxes
Winter (Nov 2019 - Mar 2020): Demographic/Equity Analysis of Mileage-Based Fees Compared to Gasoline Taxes
Mar-June 2020: Project Documentation and Report Writing
In parallel, we will work with Deployment Partner Automatic (a Connected Vehicle Company) to Demonstrate a Proof of Concept of Reporting Vehicle Miles Traveled (from trip level data and inclusive of only Pennsylvania roadways) to a third party in support of mileage-based fees. The schedule for that work is as follows.
Fall 2019: Exchange of Data Fields and Protocols
Winter 2019: Data Collection and Sampling, Anonymization
Strategic Description / RD&T
Deployment Plan
The primary work in this project will be to develop the insights needed to create the system in the future. Thus much of the work will involve non-deployment activities, such as data analytics and construction of a finance/policy model to create equivalent revenue from a mileage-based fee and the existing gasoline tax system in Pennsylvania.
However, we do expect to have a "proof of concept" demonstration activity in the final 4-5 months of the project as noted in the project timeline. In this part of the project, we will work with our partner (Automatic Labs, a SiriusXM company) to show to a State how connected vehicle data on passenger vehicle operation and movements could be the basis for mileage-based fees.
In those 4-5 months, we will:
* Work with the partner to compare available and needed data fields (e.g., miles per trip, odometer before and after trip, vehicle location, time of day, etc)
* Query/extract a sample of anonymized vehicle records for vehicles registered in Pennsylvania containing these data fields
* Create a hypothetical report/invoicing system and consider appropriate reporting/billing frequencies given current ways of collecting fuel tax revenue
We note that the next phase of the work (which may or may not be submitted to Mobility21 in the future) would constitute an actual field deployment of the system, which might involve scanners or technology at fuel pumps, toll booths, etc, as well as a reporting system to a third party.
Expected Outcomes/Impacts
Key accomplishments and deliverables for this project include:
* Completion of Inspection Data Collection and Analysis
* Creation of Equivalent Revenue Financial / Policy Models for Mileage-Based Fees and Gasoline Taxes (Spreadsheet Model)
* Completion of Demographic/Equity Analysis of Mileage-Based Fees Compared to Gasoline Taxes
* Project Documentation and Report Writing
* Proof of Concept Technology for Connected Vehicle Reporting of Mileage
Expected Outputs
TRID
Individuals Involved
| Email |
Name |
Affiliation |
Role |
Position |
| pacharya@andrew.cmu.edu |
Acharya, Prithvi |
EPP |
Other |
Student - PhD |
| pf12@andrew.cmu.edu |
Fischbeck, Paul |
SDS/EPP |
Co-PI |
Faculty - Tenured |
| hsm@cmu.edu |
Matthews, H. Scott |
CEE/EPP |
PI |
Faculty - Tenured |
Budget
Amount of UTC Funds Awarded
$73132.00
Total Project Budget (from all funding sources)
$73132.00
Documents
| Type |
Name |
Uploaded |
| Data Management Plan |
Data_management_Plan_for_Mobility21_VMT_2019.docx |
March 19, 2019, 12:51 p.m. |
| Presentation |
2019_Mobility_21_VMT_Study_HSM_PF.pptx |
March 19, 2019, 1 p.m. |
| Progress Report |
297_Progress_Report_2020-03-30 |
April 6, 2020, 8:36 p.m. |
| Final Report |
Final_Report_-_297.pdf |
March 22, 2021, 6:59 a.m. |
| Publication |
TRBAM-S-20-01947_xklwtJV.pdf |
March 22, 2021, 5:07 p.m. |
| Publication |
TRB_2021_MBUF_VMT_Comparison_PA_Dec15_Final.pdf |
March 22, 2021, 5:07 p.m. |
| Presentation |
2020_Deployment_Partner_Poster.pptx |
March 22, 2021, 5:07 p.m. |
| Presentation |
cmu-smc-connected-data-analytics-2019.pptx |
March 22, 2021, 5:07 p.m. |
| Presentation |
iup-connected-data-analytics-2019.pptx |
March 22, 2021, 5:07 p.m. |
| Progress Report |
297_Progress_Report_2021-03-31 |
March 22, 2021, 5:08 p.m. |
Match Sources
No match sources!
Partners
| Name |
Type |
| Automatic Labs (A SiriusXM company) |
Deployment Partner Deployment Partner |
