總體經濟領先指標對電信類股價長短期之影響效果–時間數列轉換函數模型之應用

The Long-term and Short-term Effects of Macroeconomic Leading Indicators on Prices of Telecommunication Stocks：Application of Time Series Transfer Function Model

林伯修(Po-Hsiu Lin)

指導教授 ： 林泉源 ； 呂姿瑩

國立臺北大學/商學院/企業管理學系/碩士(2013年)

摘要

影響股價變動的原因可說是相當錯綜複雜，其中，總體經濟因素經常被投資者引用來預測股價走勢，但其實際的預測準確性為何？總體經濟與股價兩者間又存在何種關係？因影響股價之總體經濟因素參雜了相當多不易量化的變數，使得相關研究較難以進行，故本研究擬從已量化且較具邏輯推導的景氣領先指標著手，主要以時間數列之轉換函數模型來探討三家電信公司（中華電信、台灣大哥大及遠傳電信)之股價與經建會所發布之國內景氣領先指標綜合指數及其構成項目間長期與短期之因果關係。研究期間為自2001年12月至2012年9月，共130筆月資料。資料分析過程利用統計分析軟體 SAS 進行時間數列恆定、建立轉換函數模型、以最大概似法進行模型參數估計、運用自我相關檢定與交互相關檢定作模型診斷、利用 AIC 法則與 SBC 法則選取最適模型及以 MAPE 檢視模型預測準確度。
研究實證結果發現：
一、國內景氣領先指標綜合指數及其構成項目對三家電信公司股價長期與短期之影響效果：
1.景氣領先指標綜合指數及製造業存貨指數之變動率對遠傳電信股價報酬率有顯著的長期影響。
2.外銷訂單指數與股價指數之變動率對三家公司之股價報酬率皆有顯著的長期影響，股價指數變動率對台灣大哥大與遠傳電信之股價報酬率皆有顯著的短期影響。
3.貨幣總計數與工業及服務業每人每月加班工時之變動率對中華電信與遠傳電信之股價報酬率有顯著的長期影響。
4.核發建照面積變動率對中華電信及台灣大哥大之股價報酬率有顯著的長期影響，且對台灣大哥大及遠傳電信之股價報酬率有顯著的短期影響。
5.SEMI半導體接單出貨比變動率對中華電信及台灣大哥大之股價報酬率有顯著的長期影響，對遠傳電信之股價報酬率有顯著的短期影響。
二、時間數列轉換函數模型預測三家電信公司股價之 MAPE 值介於2.65%與3.43%之間，模型預測能力皆屬高精確度。

關鍵字

景氣領先指標綜合指數 ； 電信類股價 ； 時間數列 ； 轉換函數

並列摘要

The factors affecting stock price are quite complex, wherein the macroeconomic is usually referenced by investors to forecast the movement of stock price. But how accuracy is the actual forecast? What relationship is existed between the macroeconomic and the stock price? Since the factors affecting stock price include many non-quantifiable variables, it’s hard for the related researches to carry out. So this study tried to use the macroeconomic leading indicator variables with quantization and logic derivation and discussed the short-term and long-term causality between the stock prices of the three telecommunication service companies (including Chunghwa Telecom Co., Ltd. (CHT), Taiwan Cellular Corporation Ltd. (TCC) and Far EasTone Telecommunications Co., Ltd. (FET)) and eight macroeconomic leading indicator variables issued by Council for Economic Planning and Development (CEPD) by utilizing Time Series Transfer Function Model. The study period starts from December 2001 to September 2012 with a total of 130 monthly samples. This study utilized the statistical software SAS to deal with stationarity for building up the Transfer Function Model, estimating by MLE, diagnosing by ACF and CCF, choosing the most suitable model by AIC and SBC rules, and evaluating model by MAPE.
The results can be shown as follows:
1.The short-term and long-term effects between the macroeconomic leading indicator variables and the stock prices of the three telecommunication service companies：
a.The change rates of composite leading index (LI) and index of producer's inventory (PSI) are significant long-term effective to the return rate of stock price of FET.
b.The change rates of index of export orders (EOI) and stock prices index (PI) are significant long-term effective to the return rates of stock price of CHT, TCC and FET, the change rate of stock prices index is significant short-term effective to the return rates of stock price of TCC and FET.
c.The change rates of monetary aggregates (M1B) and average monthly overtime in industry & services (ISH) are significant long-term effective to the return rates of stock price of CHT and FET.
d.The change rate of building permits (IA) is significant long-term effective to the return rates of stock price of CHT and TCC and is significant short-term effective to the return rates of stock price of TCC and FET.
e.The change rate of SEMI book-to-bill ratio (SEMI) is significant long-term effective to the return rates of stock price of CHT and TCC and is significant short-term effective to the return rate of stock price of FET.
2.The MAPE value of forecasting ability of Transfer Function Model to the stock prices of three telecommunication service companies is between 2.65% and 3.43%, the forecasting abilities are all high precise.

並列關鍵字

Composite leading index ； Prices of telecommunication stocks ； Time series ； Transfer Function Model

參考文獻

李佳璜（2008）。總體經濟與財務因素對股票報酬之影響-以台灣上市電子股為例。新北市：淡江大學會計學系碩士在職專班未出版碩士論文。

黃淑慧（2004）。股價的評估及其動態調整─以我國電信產業為例。桃園縣：中原大學國際貿易學系未出版碩士論文。

石世杰（2010）。總體經濟變數對類股報酬率預測模型之研究—以汽車、鋼鐵、塑膠產業為例。新北市：國立臺北大學企業管理學系碩士在職專班未出版碩士論文。

林威凱（2010）。台股指數與總體經濟變數相關性之探討。台北市：國立政治大學金融研究所未出版碩士論文。

胡天佑（2011）。台灣航運股股價與總體經濟的關聯性-類神經網路模型之應用。新北市：國立臺北大學國際財務金融碩士在職專班未出版論文。