This paper proposes an alternative methodology to evaluate the service quality of urban taxis and develops a level-of-service (LOS) standard for taxi customers to monitor performance. A customer satisfaction survey was conducted from January to March 2013 in Hong Kong, with the respondents invited to give specific satisfaction ratings for ten service aspects individually and a global satisfaction rating for the overall taxi service quality, as well as to rank the important aspects that influence the given global rating. The ten service aspects related to the amount of time consumed when taking taxis, services and facilities provided for finding taxis, internal environments of taxis, and personal services provided by taxi drivers. An enhanced linear regression model was developed to identify the priority areas for improvement of urban taxi service quality. Based on the numerical score of overall taxi service quality, a six-level LOS standard, similar to academic grading, is introduced accordingly to improve the general public’s understanding of the current service level. This paper discusses the potential policy implications to enhance the taxi service quality in Hong Kong, which can be applied to other metropolitan cities that provide similar urban taxi services.

Introduction

Taxis continuously circulate in search of customers, which consumes a great deal of road space and worsens traffic congestion and air pollution. To tackle these problems, taxi regulation policies have often been studied and implemented in the form of price controls and entry restrictions (e.g., Arnott, 1996; Douglas, 1972; Moore and Balaker, 2006), but such studies have ignored the spatial structure of the taxi market. To address this issue, Yang and Wong (1998) first introduced a model to determine the taxi movements on a given road network. Later, the model was improved by other researchers to capture congestion effects, multiple user classes and taxi modes, day-to-day learning processes, stochastic travel time, temporal variation in passenger demand, time-dependent customer-search behavior, probability of meeting customers, and smartphone-based e-hailing applications (e.g., He and Shen, 2015; Leng et al., 2016; Long et al., 2017; Rose and Hensher, 2014; Wong and Yang, 1998; Wong et al., 2001, 2002; 2008, 2014a; 2014c, 2015a; Yang et al., 2005, 2012; 2014; Zhang et al., 2017). However, these models mainly investigate the vacant taxi drivers’ decisions in search of customers while ignoring the customers’ preference of finding vacant taxis on streets. To study the bilateral taxi-customer meeting and searching, Wong et al. (2005) adopted absorbing Markov chain approach to modeling micro-searching behavior of both customers and taxi drivers in a network. Yang et al. (2010) further extended this study by incorporating a meeting function between customers and taxis to capture different meeting natures at taxi stands and on streets, and the searching decisions of customers and taxi drivers. Yang and Yang (2011) studied the bilateral searching and meeting function that characterizes the search frictions between vacant taxis and taxi customers. Wong et al. (2014b; 2015b) conducted stated-preference surveys to taxi drivers and customers, respectively, to investigate their search preferences to validate the modeling concepts. Apart from the above surveys, annual taxi service surveys have been conducted in Hong Kong. The surveys have commenced since 1986 to gather quantitative measurements of customer and taxi waiting times, taxi utilization, and taxi availability at sampled taxi stands and roadside observation points (Transport Department, 1986–۲۰۰۹). These survey data have been used in numerous studies (e.g., Loo et al., 2007; Xu et al., 1999; Yang et al., 2000, 2001) to analyze how taxi fleet sizes influence customer waiting time (which was considered a unique measure of taxi service quality), vacant taxi headway, taxi occupancy, taxi passenger demand, and taxi idling time. They formulated the relationship between taxi fleet size and service quality, and arrived at a similar conclusion that taxi service quality increases as the taxi fleet size increases.