時間序列是用來預測未來趨勢的一種重要技術，然而在實務上建構時間序列模型時，參數很難有效估計。原因可能來自於時間序列本身的模糊性質，而導致參數的不確定性使得預測結果產生極大誤差。如果將參數模糊化引進時間序列的模型中，往往過於複雜。本論文提出相對移動率為新的模糊時間序列建構方法，讓原本具有模糊性質的時間序列經由反模糊化(defuzzification)後，以點估計的方式估計起始中心點，經由適當的修正調整為較佳的中心點以及半徑，建立有效的區間時間序列。並將相對移動率引進門檻自廻規模型中，取代原有之門檻值設定，並建立區間時間序列。最後，我們使用台灣加權股價指數為例，以本論文所提出之方法進行區間預測及效率評估。 / The time series is an important technology that is used to predict future trends, however in the real world, parameter is difficult to estimate effectively when we construct a time series model due to the of the fuzzy property of the times series data. The estimated parameters in the time series will cause a big error due to the uncertainty of fuzzy data. It is too complex to introduce the fuzzy parameters into the time series model. In this thesis, we propose relative moving ratio as a new criteria in constructing procedure of an interval time series. We defuzzify a fuzzy data and use point estimation to obtain an initial center, then we adjust the center and radius making it more appropriately. The resulting center and radius is then become an interval time series that can be use to forecast an interval data. We also apply relative moving ratio in threshold autoregressive models by replacing the threshold in constructing interval time series. Finally, in empirical studies chapter, we use Taiwan weighted Stock Index as examples to evaluate the performance of the proposed two methods in building the interval time series.

銅價在近年來不斷的創下歷史新高，由於台灣蓬勃的電子、半導體、工具機產業皆需要銅，因此銅進口量位居全球第五(ICSG,2009)，使得台灣企業的生產成本受國際銅價的波動影響甚鉅，全球有70%的銅價是按照英國倫敦金屬交易所(London Metal Exchange, LME)的牌價進行貿易，因此本研究欲建置預測模式以預測銅價未來趨勢。 　　本研究之資料來源為2003年1月2日至2011年7月14日的LME三月期銅價，並依文獻探討選取LME的銅庫存、三月期鋁價、三月期鉛價、三月期鎳價、三月期鋅價、三月期錫價，以及金價、銀價、石油價格、美國生產者物價指數、美國消費者物價指數、聯邦資金利率作為影響因素的分析資料。時間序列分析、類神經網路已被廣泛的用於預測股市及期貨，本研究先藉由向量自我迴歸模型篩選出有影響力的變數，同時建置GARCH時間序列預測模型與具有遞迴的Elman類神經網路預測模型，再整合兩者建置GARCH-Elman類神經網路預測模型。 本研究之向量自我迴歸模型顯示銅價與金、鋁、銅庫存前第1期；自身前第2期；鎳、錫前第3期；鋅前第4期的變動有負向的影響；受到石油前第2期的變動有正向的影響，這其中以銅的自我解釋變異最高，銅庫存最低，推測其影響已有效率地反映到銅價上。也驗證預測模型必須考量總體經濟變數，且變數先經向量自我迴歸模型的篩選能因減少雜訊而提升類神經網路的預測能力。依此建置的GARCH模型有33.81%的累積報酬率、Elman類神經網路38.11%、整合兩者的GARCH-Elman類神經網路56.46%，皆優於實際銅價指數的累積報酬率。對銅有需求的企業者，能更為準確的預測漲跌趨勢，依此判斷如何跟原物料供應商簽訂合約的價格與期間，使其免於價格趨勢的誤判而提高生產成本，並提出五點建議供未來研究者參考。 / The recent copper price in London Metal Exchange (LME) has breaking the historical high. Taiwan’s booming electronics, semiconductor and machine tool industry causing copper import volume ranked fifth in the world (ICSG, 2009). Because of 70% of copper worldwide trade in accordance with the price of the London Metal Exchange, this study using time series and neural networks to build the LME copper price forecast model. 　　This study considering copper, copper stocks, aluminum, lead, nickel, zinc, tin, gold, silver, oil ,federal funds rate, CPI and PPI during the period of 2003/1/2 to 2011/7/14. Time series model and neural networks have been widely used for forecasting the stock market and futures. In this study, using Vector Autoregressive (VAR) model screened influential variables, building GARCH model and Elman neural network to forecast the LME copper price; and further, integrating this two models to build GARCH-Elman neural network prediction model. 　　This study’s VAR models show that the copper has negative effect with gold, aluminum, copper stocks, nickel, tin, zinc and itself. And has positive impact with oil prices. The highest of explained variance is copper. Copper stocks are lowest, speculating that its impact has been efficiently reflecting on the price of copper. Verifying the prediction model must consider the macroeconomics variables. Using VAR model screened influential variables can reduce noise to enhance the predictive ability of the neural network. This study’s GARCH model has 33.81% of the cumulative rate of return, Elman neural network has 38.11% and the GARCH-Elman neural network has 56.46%. All of them are better than the actual price of copper.

無 / Discussions of the South China Sea maritime territorial disputes are rife with assertions that certain state actors escalate regional tensions and that it is only a matter of time before provocations trigger armed conflict. However, these claims are based primarily on incomplete evidence, inaccurate comparisons with historical conflicts, and country or individual biases. This dissertation questions these common assertions and uses empirical evidence to assess their validity. Using time-series event data from the Global Database of Events, Language, and Tone (GDELT), it analyzes (1) the relationship between state involvement and South China Sea tensions and (2) which forecast models can most accurately predict South China Sea tensions based on data from earlier time periods. For RQ1, the analyses reveal that the involvement of certain countries corresponds with significantly higher tensions in the South China Sea, that state involvement and tensions are correlated at various positive and negative lags of interest, and that these correlations go in both directions. These findings have important implications for policymakers and researchers in that they offer empirical evidence that confirms or refutes assertions suggesting that certain countries’ actions lead to escalation or deescalation. They also provide a solid foundation for future research, which could take specific countries as individual case studies to further investigate the relationships between state involvement and South China Sea tensions. Moreover, the results indicate that there may be even more interesting phenomena at play that merit attention in future research: evidence suggesting that certain countries may either contribute to lower tensions or avoid becoming involved when there are heightened tensions, and evidence that some countries may not be contributing to but instead reacting to tensions and volatility in the South China Sea. For RQ2, two of the four forecast models perform better than the four benchmark models using both datasets. These findings also have important implications for policy and research. As governments become increasingly interested in using continuously updated global databases to facilitate policy-making, the results suggest that empirical data can help to inform conclusions about trends of escalation and deescalation in the South China Sea and be used to make relevant predictions. As a first cut at the data and a pioneering approach to analyzing South China Sea tensions, the analyses and findings of this dissertation represent a significant contribution to knowledge and a foundation for future research using time-series event data to understand the relationship between state involvement and tensions and the extent to which tensions can be forecasted in the South China Sea and around the world.

本文考量波動度不對稱效果(Volatility Asymmetric Effect)與均數回歸(Mean Reverting)兩個特性，並考量台股市場特性，嘗試建立一個適合台灣市場的波動度交易策略。利用GARCH(1,1)波動度與VIX指標建構第一個交易訊號，並建立當日沖銷部位。以賺取日內行情為出發點，利用時間序列模型捕捉波動度的高估或低估且搭配純跨式(Pure Straddle)策略或根據Delta調整後的跨式(Adjusted Straddle)策略。第二個交易訊號則是利用市場敏感指標，觀察外資與自營商在交易部位與未平倉部位的變化，找出對於波動度的影響。建立由選擇權與期貨組成的Delta-Hedged部位，藉由觀察市場上主力籌碼的變化，動態調整部位契約，尋找波段之間的獲利機會。 實証部分以期交所公布的每日交易資料與VIX日資料，利用2007至2008兩年的歷史資料，估計參數與測試交易訊號。樣本外期間為2009年1月開始至3月結束共55個交易日。考量交易成本後，兩個不同型態的交易訊號，仍然能夠藉由本研究的策略，獲得正的報酬。本文認為台灣為一個淺碟市場，過度反應資訊的特性，讓波動度策略出現獲利的機會。藉由這個波動度交易系統的研究，除了讓資金豐沛的機構投資人使用外，也能夠讓一般投資大眾建立自己的波動度交易策略 關鍵字:波動度交易，選擇權交易策略，GARCH(1,1)，VIX，市場情緒指標 / Trying to apply a preliminary study of volatility trading strategies in Taiwan derivative market is the topic of this dissertation. Capturing the market movement or even the dynamic of underlying asset is a Pandora’s Box for academic researchers and industry participants. Mean-reverting and asymmetrical effects are the two special characteristics of volatility for us to design our trading system according to the previous empirical studies. In our study, we use different type of volatility signal to capture the trading opportunities. Use the new released information form TAIFEX including VIX and Position Structure of Institutional Traders to design our signal. We apply the idea to use pure option position and delta-hedged position as our trading tools in this volatility trading system and look for the opportunities between realized volatility and implied volatility. An over-reaction may rises the uncertainty and also lead the market volatility change coherently. We use history data from 2007 to 2008 test our trading signal and parameters. The out sample period is from 2009 January to 2009 March which has 55 trading days to simulate our strategies. In the end, we see a positive result in both trading signals which earns positive return after considering the trading cost. Key words: Volatility Trading, Market Sentiment Indices, Option Strategies, VIX, GARCH(1,1)

在這篇論文中，我們討論的是一階非線性衝擊動態方程的週期邊界值問題。利用Schaefer定理及Banach固定點定理，我們得到一些解的存在性結果。 / In this thesis, we are concernd with nonlinear first-order periodic boundary value problems of impulsive dynamic equations on time scales. By using Schaefer’s theorem and Banach’s fixed point theorem we acquire some new existence results.

近年來國際金融局勢詭譎多變，金融危機層出不窮，無論外資或有意投入國外股市的投資人都勢必更加小心。本研究乃針對亞太盆地國家的股票報酬率與國家信用水準變動做分析，並依開發中國家與已開發國家之別，嘗試探究其差異。本文的研究方法採用時間序列的模型(干預模型與衝擊反應分析)，檢定亞太盆地國家股票報酬率的時間序列型態，及國家信用變動對股票報酬率及其波動性的影響。實證結果顯示： 1.無論開發中國家或已開發國家，其股票報酬率並不受自身股票報酬率的波動度影響，此結論恰與Baillie and Degennaro(1990)吻合。 2.國家信用水準變動對股票報酬率的影響未落後達一個月之久。 3.國家信用變動對已開發國家的股票報酬率及其波動度不具顯著的解釋力。 4.國家信用改變對開發中國家股票報酬率及其波動度較具解釋力，但影響方向不一致，原因可能是開發中國家的投資障礙較多。 5.衝擊反應分析受限於開發中國家股市資料較少及國家信用分數型態，而未有一致而明顯的結論。 第一章、 緒論••••••••••••••••••••••••1 第一節 研究背景與動機••••••••••••••••••••1 第二節 研究目的•••••••••••••••••••••••4 第三節 研究限制•••••••••••••••••••••••4 第四節 研究內容與研究架構••••••••••••••••••5 第二章、 文獻探討••••••••••••••••••••••7 第一節 股票波動度的相關文獻•••••••••••••••••7 第二節 國家風險的相關文獻••••••••••••••••••8 第三節 干預分析的相關文獻••••••••••••••••••10 第四節 衝擊反應分析的相關文獻••••••••••••••••12 第三章、 資料來源與說明•••••••••••••••••••13 第一節 太平洋盆地國家的股市資料來源與類型••••••••••13 第二節 敘述統計•••••••••••••••••••••••15 第三節 使用Institutional Investor的國家信用評等指標為解釋變數之因•22 第四節 國家信用風險評比決定因子與過程••••••••••••24 第五節 國家風險指標的有效性•••••••••••••••••25 第四章、 實證結果••••••••••••••••••••••26 第一節 理論架構•••••••••••••••••••••••26 第二節 檢驗各國股價報酬的時間序列型態••••••••••••29 第三節 變異數自身相關檢定(ARCH、GARCH檢定)••••••••38 第四節 波動度解釋力檢定•••••••••••••••••••45 第五節 國家信用評等解釋力檢定••••••••••••••••48 第六節 衝擊反應分析•••••••••••••••••••••59 第五章、 結論與後續研究建議•••••••••••••••••74 第一節 結論•••••••••••••••••••••••••74 第二節 後續研究建議•••••••••••••••••••••76 參考文獻•••••••••••••••••••••••••••77 一、 中文部份••••••••••••••••••••••••77 二、 英文部份••••••••••••••••••••••••77 附錄(外國專業投資機構直接投資國內狀況分析表) •••••••••80 / For the recent years, the global financial environment has been changing rapidly, which reminds qualified foreign institutional investors of more caution. This survey focuses on the relationship between stock returns, volatility and country credit rating changes among countries in the Asia-Pacific Basin. This research further divides the 12 sample countries into two categories, developed markets and emerging ones, and finds out the differences between both groups. The empirical methods used here are intervention analysis and impulse response analysis. The empirical results are as follows: 1. The stock return and its volatility do not have statistically significant relation in both developed markets and emerging ones, which coincides with the conclusion by Baillie and Degennaro(1990). 2. The impact of changes in country credit level on stock returns will work within one month; that is to say, changes in country credit level this month will affect stock returns of the current month. 3. Country credit change has no impact on stock returns of the developed markets. 4. Changes in country credit levels of the emerging markets have apparent influence on their stock returns but no identical signs. This may result from the common barriers existed in the emerging markets for foreign investors. 5. The impulse response analysis doesn't have an apparent and agreeable result owing to the constraint of rare data.

學界間廣泛地認為一般金融資產報酬具有的特性是:線型不可預測性,條件 異質變異數,非條件尖峰態 ... 等特性o 固然金融資產報酬具有線型不可 預測之特性,可是並不能否決其間可能有非線型依存關係的存在o目前大部 份經濟計量分析方法中的模式建構問題均是在假設模式的結構訊息已知的 條件下求解,然若真實體系的結構訊息未知或不明朗時,貿然地假設為某種 特定的模式結構,則可能又難於避免模式設定錯誤的困擾,因而對於真實體 系行為的描述亦將可能是誤導且不合理的,這意味著:除非該特定的模式結 構正是真實體系的表徵, 否則無論該特定模式的結構特性多完美,均難以 建構一令人信服的數理化模式來表徵真實體系之行為o 不幸地,此一問題 在高度非線型的動態隨機體系中尤其嚴重, 甚至是否存在一 ``真實'' 模式來據以表徵體系之行為,亦是相當值得懷疑, 故考慮一種無需特定結 構訊息假設的無母數方法或函數逼近法實屬必要o 類神經網路中的倒傳遞 網路模式即是符合此種特性的方法之一o然而學界間仍無法確定的是金融 資產報酬序列資料所產生的 ARCH 效果本身是否為真實序列資料產生機制 特性之顯現, 還是應歸咎於被忽略掉條件均數方面之非線性所衍生模式設 定錯誤情況下的代用模式, 並不得而知;另一方面, ARCH 模式的顯著成就 及其價值亦不能予以輕易地漠視, 因此, 試圖將 ARCH 模式所能提供的攸 關訊息納入倒傳遞網路模式的考量之中而形成倒傳遞網路-自迴歸條件異 質變異數 (BPN-ARCH) 模式以增進樣本外預測能力的精度便是本論文最 主要的嘗試重點與目的o

在這份研究中，我們提出一個類分散式神經網路框架，此框架為高頻交易系統研究下之子專案。在系統中，我們透過資料探勘程序發掘財務時間序列中的模式，其中所採用的資料探勘演算法之一即為類神經網路。我們實作一個在分散式平台上訓練類神經網路的框架。我們採用Apache Spark來建立底層的運算叢集，因為它提供高效能的記憶體內運算（in-memory computing）。我們分析一些分散式後向傳導演算法（特別是用來預測財務時間序列的），加以調整，並將其用於我們的框架。我們提供了許多細部的選項，讓使用者在進行類神經網路建模時有很高的彈性。 / In this research, we introduce a distributed framework of artificial neural network (ANN) as a subproject under the research of a high-frequency trading (HFT) system. In the system, ANNs are used in the data mining process for identifying patterns in financial time series. We implement a framework for training ANNs on a distributed computing platform. We adopt Apache Spark to build the base computing cluster because it is capable of high performance in-memory computing. We investigate a number of distributed backpropagation algorithms and techniques, especially ones for time series prediction, and incorporate them into our framework with some modifications. With various options for the details, we provide the user with flexibility in neural network modeling.

不景氣的年代，民間企業裁員、減薪或強迫員工休無薪假的事件層出不窮，襯托出公職相對起薪高、福利制度健全，任職免職程序有政府法令保障。在公職逐漸被當前的社會氛圍視為是兼具地位及幸福的工作時，愈來愈多的民眾競相投入公務人員的考試，而競相爭捧鐵飯碗的現象，也成為近期媒體報導的新聞焦點。 惟前述種種的論述都仍停留在主觀的聯想及推論上，國內鮮少針對經濟變數與公務人員考試錄取率間之關係，建立統計實證模型進行客觀量化分析。基於這樣的時空背景及社會氛圍，本研究遂以客觀的高普考錄取率表示公務人員考試競爭程度，觀察經濟環境變數對其造成的影響，是否誠如媒體所言，當景氣愈差時，公職考試就愈熱門的現象。 經過實證模型分析後，發現影響經濟變數對高考錄取率較普考錄取率變動數的影響較為顯著，包括：當期或前期的高考薪資占民間薪資比、當期或前期的失業率、前期臺股指數變動數、當期或前期臺股指數標準離差率以及時間趨勢等解釋變數，並且各自存在不同程度的影響及合理的正負關係。而普考錄取率變動數部分，僅受當期普考薪資占民間薪資比、前期失業率及時間趨勢等變數所影響。 本文最後，提出針對可能影響民眾報考公務人員的重要因素，提出相應政策建議，以期抒緩公職考試過熱的現象並精進政府政策。 / Recession-era, layoffs, pay cuting, and forcing employees to take unpaid leave are more and more in private sector, highlight the work of public sector is high starting salary, benefits sound system, and having protection by law in appointment and dismissal. More people want to participate in civil service examination, then civil service examination craze has become the focus of recent news. Provided the foregoing various opinions are still subjective conjecture, almost no study about relationship between economic variables and the qualification rates of civil service examination in domestic studies. In this context, this study used a senior and junior civil service examination qualification rates to represent the competitive of civil service examination, and to observe the effects of economic variables on the qualification rates of civil service examination, if consistent with the media reports, the worse economy is, the less qualification rates of civil service examination will be. After empirical model analysis, we found that the effects of economic variables on the qualification rates of senior civil service examination are more significant than the changes of the qualification rates of junior civil service examination. Finally, make recommendations to relief civil service examination craze.

資本資產定價模型已被廣泛使用於分析股票風險與要求報酬率之間的關係。然而，個別股票風險Beta是否足以解釋其報酬，也受到愈來愈多的質疑。Fama和French在1993年提出額外兩個因子來解釋股票報酬。我們將應用資本資產定價模型和三因子模型來分析1963年7月至2002年12月之美國的三大股票交易所上市公司。藉由一次改變分析過程中的一部分，以觀察參數估計值是否穩定。結果發現Beta_HML總是顯著且最為穩定，而Beta_SMB並不顯著。Beta經常顯著，但變動情況較大。另外，我們將考慮個別股票本身的變異，亦即將隨機效果納入考量。 / The Capital Asset Pricing Model (CAPM) has been widely used to analyze the relationship between risk and required rate of return on a stock, while it is doubted that individual stock's risk Beta has enough explanatory power for it's returns. Fama and French (1993) proposed two more factors to help explaining stock returns. We use the CAPM and the three-factor model to analyze listed companys in American stock exchanges, during the period from July 1963 to December 2002. We change part of the analyzing process a time to see if the estimates of the parameters are stable. The risk-premium Beta_HML is always significant and it performs most stable, while another risk-premium Beta_SMB is never significant. Beta is usually significant but it varies. Furthermore, we take within-stock variation into account, so random effects are considered.