Theoretical and Econometric Models of Behavioural Decision-making: Limited Attention, Framing effects and Online Search

This thesis comprises three papers. The first paper (Chapter 2) presents a theoretical model of choice under limited attention. In this model, the decision-maker (DM) forms a consideration set, from which she chooses her most preferred alternative. Both preferences and consideration sets are stochastic. We present axiomatisations for this model. Our focus is on the following identification question: to what extent can an observer retrieve probabilities of preferences and consideration sets from observed choices? Our first conclusion is a negative one: if the observed data is choice probabilities, the model is not identified - probabilities of preferences and consideration sets cannot be retrieved from choice probabilities. To solve the identification problem, we assume that an "enriched" dataset is observed. This dataset includes choice probabilities under two frames. Given this enriched dataset, the identification problem can be solved. The intuition is that the observed data reveals how a change in attention causes a change in choice, holding preferences constant. This allows us to isolate the influence of attention on choice. The second paper (Chapter 3) develops a theoretical model for a broad class of framing effects, which we call "resolution framing effects". In this model, a frame influences a DM's choice when she is indecisive. A DM is said to be indecisive between two alternatives when she does not have a strict preference for one over the other. A frame helps to "resolve" her indecision. Our model explains frame-dependent choices associated with a range of framing effects, including the salience effect, the endowment effect, the compromise effect and the priming effect. Axiomatisations and revealed preference relations are obtained for this broad class of framing effects. Our paper provides a response to commentators who have argued that framing effects are (i) evidence of irrationality (ii) inconsistent with the existence of a stable preference relation that is revealed in choice. In our model, frame-dependent choices are generated by a rational decision-process that is governed by a stable preference relation, which is revealed in observed choices. The third paper (Chapter 4) presents a structural econometric model of ordered online search. Search is said to be "ordered" if alternatives are presented in a particular order. The data is a set of searches for hotels on Expedia.com, which includes searches where orderings were random rather than generated by Expedia's ranking algorithm. Even for searches with random orderings, there is a "falling purchase rate" - that is, the propensity to purchase is low at low positions in the ordering. To explain the falling purchase rate, we assume that consumers have "declining expectations": consumers expect that, for the alternatives not yet browsed, alternatives higher in the ordering are better. Declining expectations explain the falling purchase rate, because they ensure that consumers may abandon a search part-way through. The declining expectation model generates a fall in the purchase rate similar to that in the data. Our model is an adaption of McCall's job search model. At least for modelling ordered consumer search, McCall's model is more appropriate than the commonly-used Weitzman search model