Mobile device location data: Can they capture the relationship between VMT and gasoline prices?

Historically, personal travel behavior has been tracked through national and local travel surveys where respondents report their travel on a given day. These surveys suffer from a number of limitations. Mobile Device Location Data (MDLD) make it possible to observe a large number of devices over multiple days, and to monitor the mobility of the population virtually on a continuous basis. Because these data are anonymized, trip origins, destinations, modes, and purposes must be inferred relying on certain rules and assumptions. Most current research applications and commercial use of the MDLD aggregate individual VMT to produce totals for certain geographical areas. We use MDLD data from the Washington, DC, metro area in 2021–2022, to form longitudinal (panel) datasets that follow the same device over time, or the same census area (tract or block group) over time. We examine whether the MDLD data confirms expectations in terms of VMT patterns over time, and whether it is related to characteristics of the residents of the area, land use and transportation infrastructure, and the local price of gasoline. We find that these expectations are borne out in the data. Higher gasoline prices will reduce VMT, and the price elasticities are very reasonable, whether we use individual-level data or block-group or census-tract averages. Finally, we find that total VMT in an area capture the demand for gasoline and help explain the price of gasoline charged at each gas station in the area, but this effect disappears when we include population, land use and infrastructure in the area.