Thesis on approval behaviour

This thesis answers the research questions in the field of decision maker (DM)’s ‘approval’ behaviour. A DM makes approval decisions by putting items into binary classes (authorization or rejection, approval or denial, inclusion or exclusion). This thesis introduces novel models which can be used to study approval behaviour. This thesis also draws on recent empirical frameworks that allow us to use an online commerce dataset to analyse how consumers include or exclude listings into their consideration sets (i.e., the sets of alternatives DMs actually choose between).

In History Dependent Approval chapter, we provide two models to study both deterministic and stochastic sequential approval behaviours that exhibit history dependency. The deterministic model is fully characterized from an Approval Consistency axiom, which states that if x is approved and y is disapproved given a history, then there does not exist any history given which x is disapproved and y is approved. The stochastic model is also characterized from a single axiom that can be seen as a stochastic version of Approval Consistency axiom.

Logit Function in Stochastic Categorization chapter develops a logit categorization function that can be applied to approval data. We characterize the model from a simple condition on approval data. We show that the logit categorization function can be derived from adaptations of the multinomial logit model. The logit categorization function can be useful in future empirical studies on approval behaviour.

In Consideration Set Formation in Online Commerce chapter, we estimate the factors that affect the probability of considering a listing on the web page using a detailed browsing dataset from eBay. We find strong evidence that consumers form their consideration sets depending on characteristics like shipping cost, total price, seller type, etc. We analyse the change of preference and consideration formation after a major platform redesign on eBay deployed in 2011.