Measuring fare non-interactions and revenue loss in bus transit using automated data
Abstract
Fare evasion remains a persistent challenge for public transit systems, yet traditional methods of estimating its impact such as manual inspections and surveys are limited in coverage and subject to sampling bias. This study develops a scalable methodology to quantify fare non-interactions and estimate associated revenue loss using automatically collected data from the Massachusetts Bay Transportation Authority (MBTA) bus network. By integrating Automated Fare Collection (AFC) and Automated Passenger Counter (APC) data, we calculate fare non-interaction rates at the stop-event level, identifying when passengers board without a recorded fare transaction. After data cleaning and quality control to remove systematic APC errors, the analysis reveals a systemwide non-interaction rate of 32%, with clear spatial and temporal patterns. Non-interactions were lower during morning peak hours and higher during midday, afternoon, and evening periods. Estimated revenue loss, calculated using observed fare payment distributions, reached approximately $12.5 million in 2019. Maps and route-level analyses highlight locations and corridors with the highest concentrations of lost revenue. The findings underscore the limitations of relying solely on inspection data and demonstrate the potential of automated data to support targeted, equitable, and cost-effective fare enforcement strategies.