Theory versus implementation in congestion-priced parking: An evaluation of SFpark, 2011–2012

Abstract

Shortages of street parking can cause cruising, a major source of urban congestion. We used SFpark, a federally funded experiment in market-priced parking in San Francisco, to study how changes in meter prices influenced on-street parking availability. We supervised observations of more than 13,400 vehicles parked on a subset of dynamically priced and control blocks at three points in time during 2011 and 2012. Repeated-observation, change-on-change regressions show that when prices rose, the block-level occupancy of parking fell, suggesting that SFpark worked as intended. But blocks where prices rose showed no discernible improvement in parking availability—the share of time at least one space on a block face was vacant. Price increases also had no association with other factors we would expect to be influenced by price, including parking duration, vehicle turnover, and carpooling. These relationships were robust to controlling for the parking zone, the previous price level, nearby employment, and the weather. A price system designed to improve average occupancy may not improve parking availability, and thus may not reduce cruising. Cities trying to reduce cruising may need to adjust prices based on minimum vacancy, and price changes may need to be larger in many cases.