考慮會計、產業與總體經濟因素之水泥業違約預測模型

Abstract

本研究自水泥產業、上游、下游與總體經濟構面中挑選出若干變數，用以預測水泥產業違約率。本文第三章將會採用簡單線性迴歸與層級迴歸兩種模式，檢測解釋變數與產業違約率之間的關係。結果顯示，水泥進口量佔總需求量之比率、水泥廠商之全體銷貨毛利率、營建廠商銷貨毛利率、核發樓地板面積、人口增加率、房價指數等以上變數，產業內企業違約率有顯著的相關性。而樣本外的測試結果發現，相比於傳統模型，新模型的預測誤差非常微小，因此同時採用會計、產業、總體基礎的變數去預測產業違約率，有益於提升預測效率。

This thesis considers several variables whick are picked from upstream , downstream , cement industry and macroeconimics dimension to predict the default rate of Taiwan cement industry . In chapter 3 , I’ll examine the correlation between dependent and independent variables by using linear regression model and piecewise regression model. The result reveals that cement quantity imported divided by total cement quantity demanded , cement firms’ gross margin , construction firms’ gross margin , population increasing rate , house price index , and approbation of the building floor area have significant relevance with default rate of Taiwan Cement industry. After conducting out-sample testing , compare with trational prediction model, it shows that prediction error is tiny .Therefore , default rate prediction model which is built by account-based, industrial , and macroeconomics variables is beneficial to enhance prediction efficiency.