UC Berkeley

This dissertation studies three different dynamic environments and trade-offs that arise from the dynamic structure of the problem.

The first chapter considers a dynamic pricing problem. The Internet allows sellers to track "window shoppers," consumers who look but do not buy, and to lure them back later by targeting them with an advertised sale. This new technology thus facilitates intertemporal

price discrimination, but simultaneously makes it too easy for a seller to undercut her regular price. Because buyers know they could be lured back, the seller is forced to set a lower regular price. Advertising costs can, therefore, serve as a form of commitment: a seller can actually benefit from higher costs of advertising. Based on this framework, the impact of advertising costs on prices, prots, and welfare are analyzed using a dynamic pricing model. Furthermore, it is demonstrated how buyers' time preferences give rise to price fluctuation or an everyday-low-price in equilibrium.

The second chapter originates from a joint work with William Fuchs and Andrzej Skrypacz. We analyze a dynamic market with adverse selection and correlated values. Uninformed buyers compete inter- and intra-temporarily for a good that is sold by an informed seller who is suffering a liquidity shock. We contrast a transparent (public price offers) with an opaque (private price offers) information structure. For infrequent trading, all equilibria with private offers are in pure strategies and coincide with equilibria with public offers. If the frequency of trade exceeds a certain threshold, then pure strategy equilibria cannot be sustained with private offers. We characterize the set of mixed-strategy equilibria with private offers if the seller has two opportunities to trade before the deadline and show that any equilibrium with private offers Pareto-dominates the unique pure strategy equilibrium with public offers. We provide a strong intuition for why these results extend to more general settings.

The third chapter is based on joint work with Michèle M�üller-Itten and it is concerned with dynamic affirmative action policies. We provide a dynamic overlapping generations framework to analyze the costs and benefits of affirmative action policies if mentoring complementarities are present. The old population boosts the young population's productivity through mentoring, assuming the mentoring boost is increasing in the fraction of mentors from the same group. In such a framework, the main trade-off is between using the most able workers and making the optimal use of the mentoring complementaries. A balanced labor force is always a steady state. We establish conditions under which this steady state is welfare optimal and stable. Affirmative action is more likely to be welfare-enhancing in markets with an under-supply of workers because it can increase the total labor force. This effect is the largest with a subsidy in favor of minority workers because no majority workers will be crowded out. We also establish conditions under which drastic affirmative action policies should be implemented initially and terminated after one period. Finally, we present some simulations of the effect of different policies for a continuous time version of the model.