|dc.description.abstract||This dissertation is comprised of two essays focused on the central theme of sovereign default. In the first essay, I detail a method to improve forecasting models of sovereign default by evaluating the impact of a country's composition of debt. For my second essay, I assess the long-horizon impact of a sovereign default episode on a country's economic volatility.
A country's external debt relative to gross domestic product is positively associated with the probability of being in default. Aggregate measures of external debt are commonly used for this type of risk analysis. However, based on the details of each loan, there are numerous subsets of external debt. Using a dataset of 32 developing countries from 1971-2010, I find that short-term debt, commercial bank loans, and concessional debt owed to other countries are the categories responsible for the positive relationship between the aggregate and the probability of being in default. In addition to isolating the categories driving the relationship between total external debt and sovereign default, I distinguish between the associated impact of each debt category on the probability of entering default and the probability of prolonging default. This is an important distinction as some subsets, such as bonds and multilateral concessional debt, are only related to the probability of entering default, while others, such as the use of IMF credit, are only significant when a country is already in default. Additionally, I find that short-term debt, commercial bank loans, and concessional bilateral debt positively affect both the probability of entering default as well as the probability of remaining in default. Focusing on the composition of external debt provides a more complete picture of the effect of debt on the probability of default, allowing for more precise forecasts of default probabilities.
In my second chapter, I evaluate the impact of sovereign default on the volatility of gross domestic product, a relationship related to two strands of literature: one focused on the impact of sovereign default on output and another that evaluates the impact of economic volatility on growth in output. In addition to bridging the gap between the existing strands of literature, the dataset and techniques employed in this analysis offer advantages over those currently in use. First, the use of the Beveridge-Nelson decomposition addresses issues encountered by other, common trend-cycle decomposition methods. Second, this dataset, which spans 1870-2008, includes more sovereign default episodes per country, allowing for a richer region-specific evaluation. Using a dataset of 7 South American countries, I find that the volatility of output decreases following a sovereign default episode. Taking advantage of the considerable longitudinal dimension, I am also able to assess the impact of volatility on economic growth by looking at different sub-periods. I show evidence that from 1870-1959, economic volatility is positively related to growth while from 1960-2008, a commonly used time period in the literature, there is no effect. These findings suggest that volatility's influence on economic growth may change over time.||