Minjie Guo

Date of Award

Spring 2019

Document Type

Open Access Dissertation


Moore School of Business

First Advisor

John McDermott


This dissertation investigates the association between public debt, high inflation and economics depression by using the survival analysis approach.

In the first chapter, I use non-parametric survival analysis to check the association between high debt episodes and high inflation episodes. Both the Kaplan-Meier estimate and the Nelson-Aelon analysis indicate that the existence of an overlapping high inflation episode is correlated with longer duration of high debt episodes. The reverse is also true: the existence of a high debt episode overlap is associated with longer duration of high inflation episodes. Whether or not the country is a member of the OECD does not matter for the survival experience.

In the second chapter, I use the Cox PH model, the exponential model and the Weibull model to analyze the survival experience of high debt episodes and high inflation episodes. I first consider the duration of high debt episodes (HDEs). Across all specifications, HDE duration depends on two variables: (1) the existence of an overlap with a high inflation episode; (2) the growth rate of GDP. These effects are not independent. At high GDP growth, an HIF overlap is associated with a shorter HDE. At low GDP growth, however, an HIF overlap is associated with a longer HDE. I also examine the duration of HIFs, using HDE overlap and GDP growth as regressors. Interestingly, the results are mixed and there is no evidence of interaction between HDE overlap and GDP growth.

Debt of the public sector has been associated with low, or even negative, growth. This position is most closely associated with Reinhart and Rogoff (2010). In the third chapter, we analyze the duration of economic crises, in the form of depressions, to see if there is an association with consecutive years of high public-debt-to-GDP ratios. We find that high debt is positively correlated with the duration of depressions while inflation is negatively correlated with such duration.

The last chapter uses parametric survival analysis to analyze the duration of eco- nomic crises to see if they have an association with the occurrence of consecutive years of high public-debt-to-GDP ratios. By comparing the results from five para- metric models, including the exponential, the Weibull, the log-normal, the log-logistic, and the generalized gamma regression, all of which have the accelerated failure time (AFT) metric, I find that there is a positive association between high debt episodes and the length of economic crises. All of the results are consistent with each other throughout the models. The main results were not refuted by adding covariates including economic factors, political factors, cultural factors and financial crises as controls.

Included in

Economics Commons