This dissertation includes three chapters. The first chapter studies the question of whether countries with different fiscal capacity should optimally have different ex-ante minimum bank capital requirements. In an environment with endogenously incomplete markets and overinvestment because of moral hazard and pecuniary externalities, I show that countries with larger fiscal capacity should have lower minimum ex-ante bank capital requirements. I also show that, in addition to the minimum capital requirement, regulators in countries with a concentrated financial sector and large fiscal capacity (which are also countries with strong moral hazard) should impose a limit on the amount of liquidity pledged by financial institutions in a crisis state (for example, restrict the amount of put options/CDS contracts sold by financial institutions). The second chapter studies the welfare implications of a concentrated, imperfectly competitive banking sector, which faces a bank net worth constraint in a small open economy (SOE) environment. There are two standard sources of inefficiency --- pecuniary externalities, which lead to overinvestment, and a standard monopolistic underinvestment force. I show that the optimal policy instruments include subsidies on firm borrowing costs in certain periods and capital account controls in others, which is a good proxy for the behavior of emerging markets. For every country, there exists a financial sector with a particular banking sector concentration, for which the inefficiencies offset each other and no government intervention is required in some periods. Furthermore, this paper documents a novel theoretical result --- the interaction between future binding bank net worth constraints and dynamic (future) underinvestment could lead to ex-ante overinvestment even in economies with a single monopolistic bank where there are no pecuniary externalities. The last third chapter, which is coauthored with Kenneth Rogoff, evaluates a new class of exchange rate forecasting studies, which claim that structural models are getting closer to being able to forecast exchange rates at short horizons. We argue that misinterpretation of some new out-of-sample tests for nested models, over-reliance on asymptotic test statistics, and failure to sufficiently check robustness to alternative time windows have led many studies to overstate even the relatively thin positive results that have been found. / Economics