Data-Driven Analyses & Policy Implications in the Transportation World: A Focus on Pennsylvania Inspection & Registration Data and Nationwide Fatal Crash Data

Vehicles are an integral part of today’s society and there is a large dependence on them for both business and pleasure. With the increasing concerns in vehicle congestion, air pollution, costly crashes, and highway funding sources, technological advances in data collection and analysis are now necessary in the transportation sector and must be taken to the next level. This thesis focuses on various analyses that can be performed pertaining to lightduty vehicles by primarily using one, low-cost data source from the annual vehicle safety inspection records in Pennsylvania. It also exemplifies how adding additional, low-cost, data sources can add even more depth and breadth to data-driven transportation analyses. By using these combined data resources, various research and policy questions can be answered through data-driven analyses. Data analyses in transportation studies do not always take into account factors such as urbanity and vehicle age, yet as shown in this thesis, these factors are necessary to make effective policy recommendations. Chapter II of this dissertation assesses various vehicle safety inspection failure rates using the VSIR data by using the data fields resulting from the safety inspections. This information provides both technical measurements and general pass/fail metrics in order to determine if any maintenance was necessary during the inspection. The data show clearly that vehicles require more maintenance for each of the following: the higher the odometer reading, the older the vehicle, and the more rural the registration zip code. Additionally, the claimed 2% failure rate only applied to vehicles within their first year. Failure rates were found to be much higher for all other vehicles with the average found to be around 12%-18%. Chapter III examines whether the vehicles safety inspection program saves lives. It can be concluded safety inspections are statistically effective in reducing fatality rates by approximately 1-2 fatal crashes per billion VMT in a given year. Additionally, urbanity was always found to be significant, which confirms the need for robust VMT estimates. Results show there are approximately 1,540 fatalities avoided in current safety states. On the other hand, in states with no safety program about 2,600 fatalities could be avoided if a program similar to PA were implemented. Chapter IV increases the breadth of analysis with this inspection data by using it to analyze travel patterns for individual vehicles and households with multiple vehicles. The vi primary contribution of this chapter is to provide data-driven insight to annual travel patterns based on age, urbanity, and time, in order to eventually be able to make informed policy decisions in order to distribute funding most efficiently. While average VMT data is publically available as averages for each urban and rural area by state, it is not available on a zip code level nor does it contain ranges or other characteristics of vehicles, such as age. Results from this thesis show that while VMT is generally decreasing in recent years, when observing average VMT by vehicle age, VMT is increasing in recent years. This leads to the conclusion that owners are keeping their older vehicles longer and driving them more than the average. Differences in travel and vehicle ownership at a home zip code level are observed and therefore variations within counties and overall urbanity in the state are also seen. Additionally, we observe that while average annual VMT over time is relatively consistent over many of the years observed and much higher in rural areas, vehicle ages consistently increase each year and are approximately the same in comparing urban versus rural areas. Finally, calculations are made in order to assign vehicles to households. This limited the analyses largely due to low sample sizes and the inability to check for representativeness, but loose conclusions could be drawn between households based on vehicle counts and align with a similar study using NHTS data. Chapter V provides the summary, policy implications of the research, and final conclusions. The vehicle safety inspection program has long been debated within states over the past ten years. This state-driven policy must be analyzed, using a data-driven approach, on a zip code or county level, for the entire U.S. There are questions as to whether this program is effective in keeping roads safe and worth the money being spent. It is calculated that for states without a current safety program, the cost effectiveness (defined in terms of $/life saved) of implementing a safety inspection program similar to PA is about $6.8M ($1.9M - $180M), which falls entirely around the U.S. DOT’s value of a statistical life of $5.2 million to $13 million. It is noted that this calculated cost per life saved is likely an upper bound since the estimate does not include benefits of non-fatal crashes avoided and assumes that every vehicle has some repair performed (versus paying only the inspection cost or a zero repair cost). A bigger question is if these state-mandated vehicle policies make sense, especially in areas where there is a lot of cross-border traveling. The cost-effectiveness in this sense may not be accurate.