| dc.description.abstract | Econometric choice models have been widely used in travel behaviour research to understand human activities, time-use, mobility choices, consumption, and related preferences. Most research in this area had focused on analysing consumer choice of a single discrete alternative from a set of alternatives that are perfect substitutes of each other. In the past two decades, however, a stream of research has emerged to analyse consumer choice of potentially multiple discrete alternatives from a set of alternatives that are imperfect substitutes, along with the continuous choice(s) of “how much to consume?” of the chosen alternative(s). Such choice situations, referred to as multiple discrete-continuous (MDC) choices, are pervasive in travel behaviour research. For example, the most widely analysed MDC choices arise in the context of individuals’ daily time-use, where an individual can potentially participate in multiple activities in a day and allocate the fixed time available in the day to perform those activities. Among the methods used to analyse such MDC choices, the random utility maximization (RUM)-based models have gained traction and resulted in numerous empirical applications in the context of time-use, vehicle ownership and usage, and recreational travel.
This dissertation advances the fields of RUM-based MDC choice modelling and travel behaviour research in the following directions: (a) formulation of new models to analyse and forecast MDC choices by introducing greater flexibility in the constraints faced by consumers and flexible stochastic specifications to represent consumers’ utility functions, (b) enhancing current understanding of the properties of state-of-the-art MDC choice models with flexible utility forms, and (c) application of the newly formulated models to understand time-use patterns of non-working adults in Los Angeles region of California, time-use patterns of commuters in major metropolitan cities of India, and tourism travel expenditures of domestic vacation travellers in India.
The specific methodological contributions of the dissertation are as follows:
(A) A new model formulation to analyse MDC choices at a disaggregate-level, including the number of instances (aka, episodes) different alternatives are chosen and the amount of consumption at each instance of choice, while also accommodating logical constraints across different instances of consumption of an alternative;
(B) A new model formulation to accommodate alternative-specific upper (and lower) bounds on consumptions, and the extension of this formulation to the above-mentioned analysis of disaggregate, episode-level consumption;
(C) Enhanced understanding of the properties of MDC choice models with alternative utility functional forms, which led to: (a) analytic derivations of the distributions of demand functions arising from a specific class of MDC choice models with linear utility functions, and (b) guidelines on what type of MDC choice formulations to use for modelling different types of consumption patterns; and
(D) A new model formulation to accommodate non-IID (not-independent and not-identically distributed) stochastic specifications in MDC choice models with flexible utility functional forms.
The substantive contributions of the dissertation are as follows:
(A) Application of the newly proposed MDC choice formulations to analyze individuals’ daily activity participation and time allocation decisions at an episode level, while considering episode-level upper and lower bounds on time allocation to different activities – for an empirical analysis of non-working individuals’ time-use in Los Angeles, California;
(B) Application of the proposed MDC choice formulations to understand the determinants of expenditure allocations of Indian domestic tourists on their leisure trips – toward identifying strategies for enhancing domestic tourism revenue in India; and
(C) Application of MDC choice models to understand the differences in time-use patterns between commuting women and men in major metropolitan cities of India, with a focus on gender differences in the impact of commute duration on their time-use patterns.
The above-mentioned empirical applications augmented the extensive simulation experiments conducted in the dissertation to evaluate the efficacy of the proposed MDC choice formulations. Specifically, the analysis of Californian non-workers’ time-use helped demonstrate the benefits of episode-level time allocation models and those that accommodate bounds on time allocation – in terms of improved understanding of time-use, statistical fit, and
xii
prediction performance – over the traditional MDC choice models. The analysis of Indian tourists’ expenditures (on their leisure trips) and Indian commuters’ daily time-use helped verify the properties of MDC choice models with linear utility functions.
The empirical analysis of Indian tourists’ expenditure patterns offered insights that can potentially be used to device strategies toward increasing revenue for the tourism and hospitality industry. The empirical analysis of Indian commuters’ time-use patterns brought to light notable differences in the time-use patterns of working women and men in India. Importantly, this analysis highlighted the need for policies aimed at addressing working women’s time poverty issues exacerbated by their commute. | en_US |
