考量經濟指標領先及落後效應之違約率預測模型

Abstract

近幾年來全球面臨金融風暴，台灣的銀行也受到影響到，遇到許多企業借款卻違約不還的問題，因此，能夠準確預測違約率對銀行與金融機構就能降低金融風險。現有中外文獻已提出利用經濟指標(economic indicator)來建構違約率預測模型方法，然而有些經濟指標有領先及落後的效應，會影響到違約率預測模型的準確度。因此本研究之主要目的是考慮經濟指標之領先及落後效應，利用相關分析及向量自我迴歸(Vector Autoregression, VAR)方法來分析經濟指標之領先及落後效應，再使用逐步迴歸以及自組性演算法(Group Method of Data Handling, GMDH)建構違約率預測模型，最後並驗證本研究方法確實較未考慮經濟指標領先及落後效應之違約率預測模型準確。

In recent years, facing the world financial crisis, Taiwan's banks are also affected. Banks face many default corporate borrowers. Therefore, accurately forecasting the default rate of banks and financial institutions will reduce financial risk. The existing literature has proposed using economic indicators to construct the default rate forecasting model. However, some economic indicators have leading or lagging effect, and they will affect the accuracy of the forecasting model of default rate. Therefore, the main purpose of this study is to consider the leading and lagging effects of the economic indicators, using correlation analysis and Vector Autoregression(VAR) method to analyze the leading and lagging effects of economic indicators, and then using the stepwise regression, and Group method of the Data Handling (GMDH) to construct the default rate forecasting model. Finally, this study utilizes a real case to verify that the proposed method has higher accuracy rate than that of a prediction model without not considering the leading and lagging effect of economic indicators .