Cruising for parking around a circle

There has recently been considerable interest in cruising for curbside parking as a major contributor to traffic congestion in the downtown areas of major cities. The density of cars cruising for parking in the downtown area is related to the rate at which cars in transit in the downtown area start cruising for parking and the expected search time of a car that starts cruising for parking. This paper focuses on this expected search time. The literature has employed three different approaches to estimate expected cruising-for- parking time: direct measurement, inference based on the equilibrium condition that (for the marginal parker) the expected cost of curbside parking equals the expected cost of garage parking, and inference based on the observed occupancy rate of curbside parking and an assumed statistical relationship between expected cruising-for-parking time and the curbside parking occupancy rate. The last approach typically obtains estimates of expected cruising-for-parking times that are lower, and with high occupancy rates much lower, than those estimated using the other two approaches. This paper takes a step towards resolving this inconsistency by demonstrating, through computer simulation of cars cruising for parking around a circle in stochastic steady state, that an approximating assumption in the derived statistical relationship between expected cruising-for-parking time and the curbside parking occupancy rate leads to underestimation of average cruising-for-parking time, and at high occupancy rates very considerable underestimation.