I study three questions related to competition and market design.In the first chapter, I study whether a peer-to-peer platform should set prices directly, or should it let sellers set prices while providing price recommendations. A platform can centralize prices and use exclusively available demand information, while price recommendations let sellers compete using their private information. On sharing economy platforms, for example, we observe a myriad of such pricing regimes. We investigate the implications of each pricing regime for the profits of platforms, buyers and sellers. When a platform recommends prices, it effectively plays the role of a sender in a multi-receiver cheap-talk game. In the second chapter, I study a particular instance of the trade-off between fairness and efficiency - a topic of increasing interest for marketers. After recreational cannabis legalization in 2012, Washington state policymakers had to choose whether to allocate retail licenses by lottery (a fair mechanism) or by auction (an efficient mechanism). They chose a lottery. Using transaction data from the Washington State Liquor and Cannabis Board, I estimate an equilibrium model of competition in the recreational cannabis market and use it to simulate the counterfactual auction allocation of licenses. I find that an auction would have increased total sales by 5% and reduced prices by 3%. As a result, the state lost $137M over ten years in tax revenue, which amounts to 0.39% of the state's annual budget. From the perspective of fairness, I find that under an auction, Black applicants are on average 21% less likely to receive a retail license and majority-White areas of the state reap disproportionately larger consumer benets from the auction (20% increase vs 3% increase in consumer surplus for majority-Black and -Asian areas). In the third chapter, I document the effect vertical integration between broadcast television networks and major movie studios has on the studios' advertising strategies. I use a matching procedure together with a Difference-In-Differences estimator to separate the effect of vertical integration from the "content match". I find significant affiliation effect for ABC and Walt Disney, and FOX and 20th Century, but not NBC and Universal.