Probability of default, private and public information.

Summary The first chapter is motivated by three elements of bank panics, which in combination are still very poorly understood. The first element is the fact that information about the quality of banks' long-term investments is not perfect. The second is the fact that the quality of banks' investments can be correlated. The third element is the fact that depositors of one bank can observe when there is a run on another bank. This chapter extends the application of "global games" but to examine how these elements affect the deposit contract that banks offer to depositors and the ex ante probability of sequential bank panics. While noting the attention paid to the facts that the quality of banks' investments is not perfect and that this quality may be correlated, the second chapter focuses on the interbank market. This chapter also extends the application of "global games" to examine, this time, how these three new elements affect the deposit contract that banks offer and the ex ante probability of bank panics. The third chapter presents a dynamic rational expectations model to explain the volatility observed in emerging market risk premia. This model is based on the assumption that agents regularly receive new information about the country's ability to repay its sovereign debt. We then empirically study the original predictions of the theoretical model, using monthly data for emerging markets over the period 1994-2006.