以結構模型衡量國家信用風險

Abstract

本論文考量在國家的資產負債表下，以三種不同的結構模型探討國家信用風險。其模型的參數係以資料轉換最大概似估計法（transformed-data maximum likelihood estimation method）及雙重極大演算法（maximum-maximum algorithm）進行估計。在樣本期間內，我們發現當重大的經濟事件發生時，其模型在四個受檢驗的樣本國家都可得到相對高違約機率。實證的結果顯示，結構模型可以高度的被用來衡量國家的信用是否已到了岌岌可危的地步。本文亦說明結構模型對於國家信用風險的監控是有效的工具。

In this paper, we use three structural models to investigate a country’s credit risk by applying it to a sovereign balance sheet. The transformed-data maximum likelihood estimation method and the maximization-maximization algorithm are adopted for model calibration. The derived probability of default over time for four sample countries matched well with the events and economic conditions that occurred during the sample period. Our empirical analyses show that structural models can be used to determine with high accuracy whether the credit of a sovereign country is in a precarious situation. We then illustrate how the structural approach can be an effective tool to monitor the sovereign credit risk.