%0 Work
%T Essays on Industrial Organization
%A Yongbae Lee
%D 2008-11-10
%8 2018-09-20
%R http://localhost/files/x920fw991
%X In the first part of the dissertation, I investigate the nature of retail coupons, a popular tool for non-price competition. The widely expressed view that coupons are primarily a tool to allow price discrimination has received mixed empirical supports. I depart from the static framework of the price discrimination theory to explore what alternative roles the coupons may play in an environment where demand is dynamic. In Chapter 1, I examine the consumer-level panel data. I show that, while coupons themselves do not have any lasting effect on consumers' brand choice, they induce different responses from consumers with varying degree of consumption experience. The evidence implies that coupons may have promotional effects that reinforce consumers' decaying consumption experience. In Chapter 2, I examine the retailer-level sales data and investigate whether coupon availability constitutes a state variable for the retailers' pricing decision. I estimate a linear probability model to show that coupon availability does have an influence over retailers' sale decision even after accounting for accumulation of latent demand over time.
In the second part, I conduct an econometric exercise using the dynamic discrete choice model. The stage utility functions in dynamic discrete choice models are, in general, not nonparametrically identified even when the discount factor and the distribution of the unobservable state vector are known to the researcher. Aguirregabiria (2002) demonstrated that it is feasible to identify the counterfactual choice probabilities without evaluating the stage utility function, when the policy in question linearly modifies the stage utility function. I study a different type of policy implementation that results in a shift in transition probabilities, with which Aguirregabiria's results are not replicated. It is shown, however, that with a sufficient variation in transition probabilities, we can point identify the stage utility function when we are given the opportunity to observe the change in the agent's behavior following such a policy implementation.
%[ 2018-09-20
%9 Dissertation
%~ Arch : Northwestern University Institutional Repository
%W Northwestern