總體因素與現金流量風險值 —台灣半導體產業之實證分析

Macroeconomic Factors and CFaR: Evidence in Taiwan Semiconductor Industry

黃珮純(Pei-Chun Huang)

指導教授 ： 吳博欽

中原大學/商學院/國際貿易研究所/碩士(2007年)

https://doi.org/10.6840/cycu200700448

摘要

企業較過去面臨更多的不確定因素，且面對的風險種類亦增加，本文將現金流量風險值模型應用於非金融企業的經營上，採用現金流量風險值進行風險管理，由現金流量的角度來定義企業的整體風險，選取 4 個總體經濟指標以解釋現金流量，包括新台幣兌美元匯率、一銀三個月期定期存款利率、領先指標綜合指數、費城半導體指數，衡量台灣半導體產業中的個別公司現金流量風險值，以 2000 年第 4 季至 2006 年第 2 季台灣半導體產業的 37 家廠商為研究對象，針對固定效果模型－共同斜率及隨機效果模型－共同斜率與固定效果模型－差異性斜率估計現金流量方程式。
實證結果顯示，利用逐步迴歸分析與混合估計法，從選取的 4 個總體經濟指標檢測對台灣半導體產業的 EBITDA 之關係，無論在固定效果模型－共同斜率及隨機效果模型－共同斜率與固定效果模型－差異性斜率下，新台幣兌美元匯率、一銀三個月期定期存款利率、領先指標綜合指數與費城半導體指數對上市 ( 櫃 ) 半導體產業的 EBITDA 具有顯著解釋能力。
由 RMSE 、 MAE 、及 Theil’s U 三種預測能力評估指標得知，台灣半導體產業中的IC設計大多數公司均採取隨機效果模型－共同斜率及蒙地卡羅模擬法時的預測績效為佳；台灣半導體產業中的IC製造公司除漢磊適合採用隨機效果模型－共同斜率及蒙地卡羅模擬法進行預估，其餘大多數公司均採用隨機效果模型－共同斜率及歷史模擬法時的預測績效為佳；台灣半導體產業中的IC封測公司除日月光、京元電、矽品及泰林適用於採用隨機效果模型－共同斜率及歷史模擬法進行預估，其餘公司皆以採取隨機效果模型－共同斜率及蒙地卡羅模擬法時的預測績效為佳。

關鍵字

混合估計法 ； 蒙地卡羅模擬法 ； 歷史模擬法 ； 現金流量風險值

並列摘要

Enterprises face more uncertain factors than before to investigate in risk management; firms have to develop a model to measure their faced risks. This thesis hope to apply the concept of “Cash-Flow-at-Risk, CFaR” to nonfinancial firms, construct and estimate their operating risk.
We fit Cash Flow equation by using four kinds of Macroeconomic indicators, including TWD/USD Exchange Rate, Deposit Rate-3 Month, Philadelphia Semiconductor Index and Composite Leading Index, and further measure individual firms CFaR of Taiwan Semiconductor Industry. We make use of firms of Taiwan Semiconductor Industry as object of study from 4th quarter of 2000 to 2nd quarter of 2006, aim at Fix Effect Model-Common Slopes, Random Effect Model-Common Slopes and Fix Effect Model-Differential Slopes to estimate Cash Flow equation.
In empirical study, we make use of stepwise regression and pooled estimation to understand the relation among four kinds of Macroeconomic indicators and EBITDA of Taiwan Semiconductor Industry. Regardless of in the Fix Effect Model-Common Slopes, Random Effect Model-Common Slopes and Fix Effect Model-Differential Slopes, TWD/USD Exchange Rate, Deposit Rate-3 Month, Philadelphia Semiconductor Index and Composite Leading Index have significantly effects on cash flow.
Through RMSE, MAE, and Theil’s U three kinds of predictperformance indices we know that, most of Taiwan Semiconductor IC Design firms are suitable to adopt Random Effect Model-Common Slopes and Monte Calro Simulation to fit cash flow equation and evaluate CFaR; most of Taiwan Semiconductor IC Manufacturing companies are more suitable to adopt Random Effect Model-Common Slopes and Historical Simulation; most of Taiwan Semiconductor IC Packaging and Testing firms are more suitable to adopt Random Effect Model-Common Slopes and Monte Calro Simulation.

並列關鍵字

Pooled Estimation ； Historical Simulation ； Cash Flow at Risk ； Monte Calro Simulation

參考文獻

陳宏裕 (2004)，「估計現金流量風險值之比較模型－台灣資料實證」，國立成功大學財務金融研究所碩士論文。

Andrén, Niclas, Jankensgård, Hakan, Oxelheim (2005), “Exposure-based Cash-Flow-at-Risk: An Alternative to VaR for Industrial Companies,” Journal of Applied Corporate Finance, 17(3): 76-86.

Dickey, D.A., and W. A. Fuller, (1979), “Distribution of The Estimators for Autoregressive Time-series With a Unit Root,” Journal of the American Statistical Association, 74, 427-431.

Granger, C. W. J. and R. Newbold, (1974), “Spurious Regressions in Econometrics,” Journal of Economics, 111-120.

Hausman, J.A., (1978) “Specification Tests in Econometrics,” Econometrica, 46, 1251-1271.