現金流量風險值之估計

Abstract

本文分別建立線性的AR與多元迴歸模型，以及非線性的STAR與STARX等四種模型，以求得公司的最適現金流量預測模型，並透過蒙地卡羅模擬估計公司的CFaR，以期更精確地評估公司營運的風險衝擊。研究對象爲台灣50指數中，上市長達十年以上的21家非金融公司，研究期間爲1996年第4季至2007年第3季。實證結果顯示，無論就時間序列或多元迴歸模型而言，絕大多數公司的現金流量變化爲一非線性過程，且與景氣領先指標綜合指數、消費者物價指數、新台幣兌美元匯率與一銀三個月期定期存款利率等總體經濟變數落後期間存在非線性關係。在樣本內估計結果上，STAR模型與STARX模型較線性的AR模型與多元迴歸模型皆提供較佳的配適度；惟樣本外預測結果則顯示，超過70%的樣本公司適合採用線性模型估計CFaR。

This paper first sets up linear AR and multiple regression, nonlinear STAR and STARX models to find out the optimal cash flow forecasting model, then utilizes Monte Carlo simulation to evaluate company's CFaR. The sample objects are 21 non-financial component companies in TSEC Taiwan 50 index. Sample period spans from the 1996. 4Q to 2007. 3Q. Empirical result shows that for time series or multiple regression models almost all companies’ cash flow are fitted by nonlinear process and the relationship between cash flow and macroeconomic variables (Composite Leading Index, Consumer Price Index, TWD/USD Exchange Rate, and 3-month Deposit Rate) is also nonlinear. STAR and STARX models provide better goodness of fit than linear models. But for over 70% companies are suitable to adopt linear model to forecast their CFaRs.