This thesis is devoted to a novel approach to intertemporal decision-making. Specifically, it
applies the reference-dependent model to intertemporal choices and shows that one can model
intertemporal choice even without the assumption of time discounting.
In Chapter I, I present a brief review of the background literature, which could be of help for
our understanding of intertemporal decision-making. Specifically, it focuses on three prominent
economic theories, which are the most relevant for my work on intertemporal decision-making,
namely, the exponential discounted utility theory, the hyperbolic discounting theory, the theory of
reference-dependent preferences, and the similarity-based procedure.
In Chapter II, I present the intertemporal reference-dependent model composed of an intrinsic
consumption utility and a reference-dependent gain-loss utility which are additively separable
over time. Three simple ways of reference point determination - choice-set dependent, choice-set
independent, and hybrid reference points are also explored. In addition, I also argue that the
intertemporal reference-dependent model may offer an alternative explanation of the
experimental observations by Rubinstein (2003). Furthermore, in Chapter III, I extend the
intertemporal reference-dependent model, where an individual is uncertain about her future
circumstances, and her reference levels follow a random walk process. While the model does not
explicitly include time discounting and return on saving, it nevertheless may offer an alternative
explanation of present bias and negative time preference (future bias), and may also offer an
alternative explanation of dynamically inconsistent preferences over time.
Finally, in Chapter IV, using a survey-based within-subject choice experiment, I explore whether
subjects' behaviour is consistent with a number of existing and emerging theories, namely
hyperbolic discounting theory, Rubinstein's similarity procedure, and the novel intertemporal
reference-dependent model. The main result of the survey-based within-subject choice
experiment is that none of these three theories explain subjects' behaviour. However, I do find
that, among these three theories, the novel intertemporal reference-dependent model performs no
worse than Rubinstein's similarity procedure, and hyperbolic discounting theory performs the
worst with subjects' behaviour on subset of questions. Moreover, I could reject the hypothesis
that the subjects made their choices at random. Finally, I also found that their behaviour is not
consistent with the Independence of Irrelevant Alternatives.
