Emerging market economies have witnessed recurrent large-scale sovereign debt crises. Many of these crises shared two features: a significant shift towards short-term debt before a crisis, and a prolonged debt renegotiation afterwards. This dissertation studies each phenomenon in a chapter.

The first chapter constructs a dynamic model of sovereign debt default and renegotiation in which the shift towards short-term debt before a default results from two inefficiencies in the sovereign debt market: no legal enforcement of debt repayment and no explicit seniority structure. These inefficiencies give rise to "default risk" and "debt dilution risk". The mechanism by which default risk favors short-term debt is well understood: long-term bonds incorporate additional default risk, and hence bear higher risk premia than short-term bonds, giving the government the incentive to rely more on short-term debt before a crisis. "Debt dilution" occurs because the absence of an explicit seniority structure implies that new short-term debt issuances can reduce the amount recoverable by existing long-term debt-holders, and thus "dilute" existing long-term debt. This effect is more severe as default and debt restructuring become more likely, resulting in higher risk premia on long-term bonds as a crisis approaches. Quantitative analysis shows that the model generates a large shift to short-term debt before a default. The fraction of this shift due to debt dilution is significantly larger than the fraction due to default risk.

The second chapter provides an explanation to the observed length of delays in debt restructuring negotiations. Contrary to the common wisdom that delays are costly and inefficient, this chapter argues that delays can be beneficial in that they allow the economy to recover from a crisis and make more resources available to settle the defaulted debt. As a result, the negotiating parties can be better off by waiting and then dividing a larger "cake". By introducing a stochastic bargaining game, based on Merlo and Wilson's (1995) framework, into a sovereign default model, this chapter shows quantitatively that our argument can generate an average delay length comparable to that experienced by Argentina in its most recent debt restructuring.