An analysis of parking behaviour using discrete choice models calibrated on SP datasets

Abstract

Parking policy is an important component of contemporary travel demand management policies. The effectiveness of many parking policy measures depends on influencing parking type choice, so that understanding the factors affecting these choices is of considerable practical importance. Yet, academic interest in this issue has been, at best, intermittent. This paper reports the results of an analysis of parking choice behaviour, based on a stated preference (SP) dataset, collected in various city centre locations in the UK. The analysis advances the state of the art in the analysis of parking choice behaviour by using a mixed multinomial logit (MMNL) model, capable of accommodating random heterogeneity in travellers’ tastes and potential correlation structure induced by repeated observations being made of the same individuals. The results of the analysis indicate that taste heterogeneity is a major factor in parking type choice. Accommodating this heterogeneity leads to significantly different conclusions regarding the influence of substantive factors such as access, search and egress time and on the treatment of potential fines for illegal parking. It also has important effects on the implied willingness to pay for timesavings and on the distribution of this willingness in the population. Our analysis also reveals important differences in parking behaviour across different journey purposes, and the models reveal an important locational effect, in such that the results of the analysis vary substantively across the three locations used in the SP surveys. Finally, the paper also discusses a number of technical issues related to the specification of taste heterogeneity that are of wider significance in the application of the MMNL model.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 18 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.